Chlamydial infection in pregnancy


Material Information

Chlamydial infection in pregnancy
Physical Description:
Schmitt, Karla
Publication Date:

Record Information

Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 21675919
oclc - 48548202
System ID:

Full Text







Copyright 1999


Karla Schmitt

This study is dedicated to my parents John and Ardelle Schmitt. Their life has long
been an example of hard work, love and commitment to each other, their children and
grand children. They have always provided generous support to the dreams we each have
had and more importantly never doubted our ability to achieve them. They will always
remain a powerful influence and beacon light for the path ahead, and for each of
'"my tomorrows."


I would like to thank my committee chair, Dr. Sharleen Simpson, who for the

duration of several years and from great distances has been a mentor and provided

generous guidance. And each of my committee members, Dr. Charles Mahan, Dr. Donna

Treloar, Dr. Lois Malasanos and Dr. James Stansbury, the contribution of their expertise

and, most notably, their consistent availability for consultation.

I would like to acknowledge with appreciation the following individuals whose

contributions have been significant and without whom this study would not have

happened! The relational database was prepared through the assistance and many hours

contributed by Alwin Chang, Ken Kampert, Melodee Dew and Karen Freeman. Dan

Thompson provided generous time and support to all steps of the statistical analyses. Phil

Moncrief provided willing guidance on my interpretation of the case morbidity system and

syphilis findings. The patient teaching of many colleagues strengthened the scope of

content contained in this work. The extensive assistance ofDeberah Keith provided

organization to my wandering path on countless occasions. Jack Wroten encouraged my

professional growth and achievement through his exemplary leadership, and allowed me

the opportunity to explore a topic of enduring interest.

Finally, I would like to thank my family and friends for their loving support words

of encouragement, tolerance, nourishing meals and many spontaneous acts of kindness.


ACKNOW LEDGM ENTS.............................................................................................. iv

LIST OF TABLES........................................................................................................ vii

LIST OFFIGURES........................................................................................................ ix

ABSTRACT............................................................................................................. x


1 INTRODUCTION ..................................................................................................... 1
Problem Statement ............................................................................................... 1
Purpose of the Study............................................................................................ 4
Research Question ............................................................................................... 4
Definition of Terms.......................................................................................... 5

2 REVIEW OF THE LITERATURE............................................................................ 9
Epidemiology and Prevalence of Chlnamydia trachomatis ...................................... 9
Epidemiology and Prevalence of Low Birth Weight............................................ 25
Biology of Chlamydia trachomatis..................................................................... 27
Developmental Cycle ................................................................................... 30
Pathophysiology and Pathogenesis................................................................ 34
Laboratory Diagnosis of Chlamydial Infections................................................... 38
Confounding Factors and Quality of STD Specimens Submitted for Testing....... 43
Standards on Gen-Probe PACE2C Testing Techniques Within Florida.............. 46
Risk Factors for Low Birth W eight................................................................. 47
Contribution of Psychosocial and Behavioral Factors.................................... 48
Contribution of Physiological and Medical Care Factors............................... 50
Contribution of Sexually Transmitted Diseases............................................. 59

3 M ETHODOLOGY .................................................................................................. 64
Research Design ................................................................................................64
Protection of Human Subjects ............................................................................ 65
Confidentiality...................................................................... ........... ............. 65

Data Sources ...................................................................................................... 66
Birth and fetal death records ......................................................................... 66
Healthy Start prenatal screen ........................................................................ 70
Maternal and infant laboratory test report database ....................................... 73
M aternal and infant case morbidity database ................................................. 78
Congenital syphilis database ................................................. ........... ......... 80
Methodological Issues in the Use of Administrative Databases........................... 81
Definition of Study Variables .............................................................................. 83
M methods ofAnalysis .................................................................. ......................... 94
Development of the relational database ......................................................... 94
Descriptive analyses ................................................................ ..................... 97
Logistic regression ........................................................ ........... .................... 98
Inclusion and Exclusion Criteria ......................................................................... 99
Assumptions............................................... ..................................................... 100
limitations of the Study Design and M ethodology ........................................... 100
Strengths of the Study Design.......................................................................... 103

4 RESULTS............................................................................................................. 104
Results of Linking the Respective Databases .................................................... 104
Descriptive Comparison of the Relational Database and Study Sample............. 110
Descriptive Comparison of the Dependent and Independent Variables .............. 120
Bivariate Analysis ..................................................................................... ....... 122
Descriptive Analyis of 1996 Statewide Gen-Probe PACE2C Results ............... 128
Logistic Regression Results ...................... ........................................... ............. 131
Other STD Related Study Findings of Interest .................................................. 153
Outcome of Research Question ........................................................................ 156

5 DISCUSSION AND RECOM MENDATIONS........................ .............................. 157
Analysis and Discussion of Research Findings .................................................. 158
Implications for Clinical Practice...................................................................... 171
Implications for Public Health Policy................................................................ 174
Implications for Future Research ...................................................................... 179

LIST OF REFERENCES.................................................................................. ........... 182

VARIABLE LIST ....................................................... ............................ ......... 213

SYNTAX ......................................................................................................... 218

BIOGRAPHICAL SKETCH ......................................................................................... 235


Table page

1. Comparison of Low Birth Weight Rates: Florida and United
States, 1989- 1997 ...................................................................... 26

2. Historical Milestones in the Recognition and Study of
Chlamydia trachomatis................................................................................. 28

3. List of all Variables Used in Analysis .................................................. 84

4. Total Records Linked to the 1996 Birth and Fetal Death Records ............... 108

5. Study Sample Records Linked to the 1996 Birth and Fetal........................ 109
Death Records

6. Comparison of Select Demographic Variables Between Relational
Database and Study Sample ...................................................................... 112

7. Comparison Between Relational Database and Study Sample of
Select Variables Used in Combination to Create
Independent Indicator Variables for the Logistic Modeling .....................117

8. Comparison Between Relational Database and Study Sample of
Records Identified for Use as the Dependent Indicator Variables...............121

9. Comparison Between Relational Database and Study Sample
of Records Identified for Use as the Independent Indicator Variables....... 123

10. Unadjusted Odds Ratios for Low Birth Weight, Based on Bi-variate
Analysis of Independent Variables in the Study Sample ...................124

11. Unadjusted Odds Ratios for Term Low Birth Weight, Based
on Bi-variate Analysis of Independent Variables in the Study Sample ....... 126

12. Unadjusted Odds Ratios for Pro-Term Low Birth Weight, Based
on Bi-variate Analysis of Independent Variables in
the Study Sample ........................................................................127

13. 1996 Statewide Chlamydia and Gonorrhea Prevalence Estimates based on
Laboratory Test Results ......................................................... 130

14. Adjusted Odds Ratios, Based on Logistic Regression,
for all Variables ........................................................................ 133

15. Adjusted Odds Ratios, Based on Logistic Regression,
without Race Variable .................................................................. 137

16. Adjusted Odds Ratios, Based on Logistic Regression,
for White Race ......................................................................... 140

17. Adjusted Odds Ratios, Based on Logistic Regression,
for Black Race .......................................................................... 143

18. Adjusted Odds Ratios, Based on Logistic Regression,
for Inadequate Weight Gain .......................................................... 147

19. Adjusted Odds Ratios, Based on Logistic Regression,
for Adequate Weight Gain ........................................................... 151

20. Comparison of Sexually Transmitted Disease Findings for
the Study Sample ....................................................................... 155


Figure page

1. Chlamydia Rates Per 100,000 Females, Aged 15-44 Years:
Florida and the United States from 1994 1998 ..............................10

2. Chlamydiales Tree ........................................................................ 29

3. The Developmental Cycle of Chlamydiae ............................................. 32

4. Primary Steps in Research Analysis ................................................. 106

Abstract of Dissertation Presented to the Graduate School
ofthe University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Karla Schmitt

December 1999

Chair Sharleen H. Simpson
Major Department: College of Nursing

The purpose of this study was to determine whether infection with Chiamydia

trachomatis during pregnancy was associated with low birth weight. A retrospective

population based study was conducted on a sample of 14,002 records. The records were

extracted from a large relational database constructed from birth and fetal death records,

prenatal risk screening records, sexually transmitted case reports and laboratory test

reports. Extensive independent indicator variables were created to control for potential

interaction between known risk factors and chlamydial infection. Descriptive, bi-variate

and logistic regression analyses were conducted.

Statistically significant associations were observed among women with inadequate

weight gain, chlamydia infection and low birth weight at 95% confidence interval (OR

1.98, p <0.02). A stronger association was observed with pre-termnn low birth weight (OR

2.34, p <0.01). Other risk factors identified as strongly associated with low birth weight in

this population were mother reporting a history of prior poor pregnancy outcome, alcohol

use, smoking, mother having been low birth weight herself

Among women who had adequate weight gain, gonorrhea infection increased the

likelihood ofhaving a pre-term low birth weight infant by more than five times (OR 5.11,

p<0.003). Women of black race and smoking were also significantly associated with low

birth weight in this group.

This study indicates that chlamydia infection in pregnancy is strongly associated

with low birth weight and that along with other sexually transmitted infections is a

significant public health problem that warrants further investigation.


Problem Statement

Chlamydia trachomatis is the most commonly reported sexually transmissible

disease in the United States since 1995. The national rates for Chlamydia trachomatis are

particularly high among women, ranging from fewer than 150 to more than 300 cases per

100,000 female population during recent years (CDC, 1996a, 1997c, 1998a). Among

populations of childbearing women, case report rates vary by age, race, socioeconomic

status, service setting, and area of the country. Young women are most at risk for

infection, with national case report rates during 1997 as high as 2,044 per 100,000 among

females age 15-19 years and higher in Florida at 2,208 per 100,000 among females age

15-19 years (CDC, 1998a; Florida Department of Health, 1997a).

Trends in prevalence rates for chlamydia are not well established because, while

some states have been reporting the disease since the 1980s, others began as recently as

1995. A national survey of 405 facilities indicated that the number of facilities that

provided testing for chlamydia increased, from 160 in 1993 to 288 facilities in 1994, the

year of the survey (Beck-Sague et aL, 1996). In the population from which this study

sample was drawn, the estimated prevalence of chlamydia among pregnant women

during 1995 was 9.1% in those under 14 years old, 8.8% in women aged 15-19 years, and

5.5% in those aged 20-24 years. The rates were lower among women more than 24 years

old (Schmitt, 1996a).

Close to 80% of infected females and more than 65% of infected males are

asymptomatic (Gaydos et aL, 1998; Quinn et aL, 1996; Schmitt, 1996b, 1999). Hence the

potential exists for undetected and untreated infections, or inadequately treated

chlamydial infections to lead to significant morbidity, with an increased risk of

postpartum endometritis, ectopic pregnancy, pelvic inflammatory disease, salpingitis,

preventable infertility, chronic pain syndromes, septic disseminated infection,

spontaneous abortion, pre-term labor, or even death. In the infected neonate chlamydial

infections are associated with pneumonia, otitis media, and conjunctivitis (Batteiger,

Fraiz, Newhall, Katz, & Jones, 1989; Brunham, Holmes, & Embree, 1990; CDC, 1998b;

Berman et aL, 1987; Askienazy-Elbhar, 1996; Genc & Mardh, 1996; Harrison &

Alexander, 1990; National Academy of Sciences, 1996; Gencay et aL, 1995; Datta et aL,

1988). Chlamydia trachomatis has also been associated with adult and childhood

myocarditis and atherosclerosis (Muhlestein et al., 1996; Grayston, Mordhorst, & Wang,

1981). Pathologic synergism has been identified between chlamydia and cervical

dysplasia (Paavonen, Koutsky, & Kiviat, 1990; Yla-Outinen et aL, 1990). Chlamydia has

also been shown to enhance transmission of human immunodeficiency virus infection by

three to four fold (Laga et aL, 1993; Plummer et al., 1991).

During the last decade there has been an increased awareness of chlamydial

infections and reporting of most identified cases of chlamydia (CDC, 1995, 1996b,

1998b). The epidemiology of Chlamydia trachomatis during pregnancy suggests a range

of prevalence from less than 6% to over 20%, depending on the age, clinic setting and

area ofthe country (Allaire, Huddleston, Graves, & Nathan, 1998; Gittens, Nichols, &

Apuzzio, 1994). Recent research suggests that the bacteria invade human host cells

within minutes, cross the placental barrier to invade amniotic cells, and cause

chorioamnionitis (Patton et aL, 1998; Thomas, Jones, Sbara, Cetrulo, & Reisner, 1990).

During the period from 1981 to 1997, the rates of low birth weight (LBW) have

overall decreased from 8.8% to 7.5% nationally (Ventura, Martin, Curtin, & Mathews,

1999). However, since 1988 there has been a creeping upward trend from 6.9 to 7.5 in

1997, the highest since 1973. During this similar period of time in Florida, there has been

an increase from 7.5% in 1980 to 8.1% in 1998 (Office of Vital Statistics, 1996a, 1999).

Not the entire upward tread is attributable to increases in multiple births among older

women secondary to treatment for infertility. Multiple contributing factors have been

identified. Nationally, low birth weight among singleton births rose from 6.03% in 1996

to 6.08% in 1997, or 4% since 1986 (Ventura, Martin, Curtin, & Mathews, 1999). The

group most affected were black women with an increase of 10.3%. Recent analyses of

Duval County, Florida births suggest the following factors may be involved with the

changing rates: 1) increased numbers of multiple deliveries, by 0.36 for 1998 over the

prior year; 2) increased LBW among multiple births, especially twins; 3) LBW among

singleton deliveries has increased; 4) increased numbers ofbirths to women of black

race/ethnicity, up 3% in 1998; and 5) a downward trend for births reported with

macrosomia (Gest, Thompson, & Hopkins, 1999).

Nationally the numbers of women who initiated prenatal care early and received

at least the recommended number of visits increased during the period from 1981 to 1995

(Kogan et al, 1998). Yet the rates of low birth weight have continued to rise slowly. One

possible factor may be the understudied role of sexually transmitted infections in adverse

pregnancy outcomes. A recent study estimates that 4.8% of LBW is attributable to

infection with Chlamydia trachomatis during pregnancy in populations with positive test

results comparable to that observed in statewide samples (Mittendorfet aL, 1994).

Purpose of the Study

This investigator and colleagues conducted a prior pilot study with 2,885 birth

records and Chlamydia trachomatis test results during 1997. The findings from this

Florida study found LBW rates were slightly lower in the sample population than in the

statewide population. Among the women with positive test results for chlamydia during

pregnancy, the LBW rate exceeded that of women with negative test results. Adjusted

LBW odds ratios for chlamydia, smoking, and black race were significant at the 95%

confidence level Odds ratios were 2.17, 2.49, and 2.09 respectively. The adjusted odds

ratios of chlamydia and smoking were highest for term LBW, 2.68 and 2.93, respectively.

Therefore this larger study was designed to further examine potential associations

between Chlamydia trachomatis and birth outcomes among a population-based sample of

pregnant women and adolescents who initiated prenatal care through county health


Research Question

The following research question was asked in this dissemrtation: What associations)

exists) between low birth weight and Chiamyia trachomatis infection during pregnancy?

Definition of Terms

The following definitions were used in this study.

Asymptomatic refers to an absence of symptoms e.g., discharge from urethra or

vagina, vulvar itching, intermittent pelvic pain, change in menstrual flow or consistency,

burning on urination, or vaginal discharge with an odor. Symptoms are different from

signs that the clinician identifies as indicative of the presence of infectious processes e.g,

"frothy green discharge versus adherent white clumping discharge," painless ulcer

visualized on the surface of the cervix, or palpable mass on the ovary. At times symptoms

may be present but the individual may not recognize them as such due to their frequency

of occurrence, or unawareness of their relevance in regard to their health, e.g., inter-

menstrual spotting.

Chlamydia is the common term used to refer to Chilamydia trachomatis. It is the

most common sexually transmitted infection in the United States, capable of causing

long-term adverse and permanent sequelae.

Chlam a trachmtis is an obligate intracellular parasite that requires a host cell

in order to live and reproduce. In the context of this study those serovars that cause

genital and congenital infections are the reference. Several other serovars are the cause of

lymphogranuloma venereum or "LGV."

DNA hybridization is a laboratory technique used to increase the likelihood of

detecting genetic material specific to chlamydia and gonorrhea present in the test

specimen. This technique is used for Gen-Probe PACE2 testing and was employed by

the laboratories participating in this study.

EIA, enzyme immunoassay, is one of the earlier non-culture tests for chlamydia.

This method detects chlamydial antigens measured by enzyme-linked immunoassay with

polyclonal or monoclonal antibodies.

Fetal growth restriction or retardation is defined as a birth weight and height

below the 10& percentile for a specific gestational age. Another term that is often used

interchangeably is "small for gestational age."

Gonorrhea is the common term used to refer to Neisseria gonorrhoeae, a gram-

negative intracellular diplococcus, predominantly sexually transmitted, and capable of

causing long-term adverse sequelae when untreated.

Inadequate specimen indicates a Gen-Probe test specimen that does not contain

enough cellular material to test for the presence of Chlamydia trachmatis or Neisseria


Indeterminate is a term used to report specimens that do not fall distinctly within

the parameters used to measure relative light units on Gen-Probe testing equipment. This

may indicate a specimen that is a false negative, a false positive, or a specimen that had

inadequate quantity on which to complete the confirmatory test.

Late syphilis is a term used to refer to the period of syphilis infection that

continues after the cessation of clinical manifestations and symptoms, associated with

primary and secondary syphilis infection. The organism, T. Pallidun, is still present,

primarily in the spleen and lymph nodes. Early latent syphilis spans the period of the first

year of infection. Late latent begins one year after infection, and may last a life time. As

during primary and secondary syphilis, a pregnant woman with latent syphilis can

transmit the infection to the fetus in utero. Tertiary syphilis may occur at anytime during

latency and congenital transmission at this stage is rare.

LCR is a brand name for ligase chain reaction, which is based on polymerase

chain reaction, a DNA amplification technology.

Low birth weight (LBW) is the term used to indicate a live born infant whose

weight is less than 2,500 grams.

Pooling is the term used to indicate a testing methodology where a portion of

multiple specimens are combined in a single vial and all are tested simultaneously. If

negative, all are reported out as negative. If the pooled result is positive, then each

specimen must be fixurther tested individually to identify the positive specimen. In

populations with low prevalence, pooling is a cost saving procedure that conserves

reagents. h is not useful in populations with high prevalence of infection.

Pre-term low birth weight (PTLBW) is the term used to indicate a live born infant

whose weight is less than 2,500 grams and less than 37 weeks gestation.

Sensitivity of a test as used in this study, is the probability of the Gen-Probe test

to report a positive test in an individual truly infected with Chlamydia trachomatis or

Neisseria gonorrhoeae.

Serology is the term used to indicate a laboratory test technique using blood

serum for testing. This is also commonly used to indicate a syphilis blood test.

Serovars of ChlwnMydia species appear to be associated with different levels of

immune response and virulence in the human host.

Specificity of a test as used in this study is the probability of the Gen-Probe test to

report a negative test in an individual truly not infected with Chiamydia trachomatis or

Neisseria gonorrhoeae.

STD or STI refers to sexually transmitted diseases or infections. In the past this

group of infections was known as "venereal diseases." These are the "commonly" known

STDs and other STDs less well known to clinicians and the public as sexually

transmittable. The commonly known STDs include syphilis, gonorrhea, trichomonas,

chlamydia, and HIV. Cytomegalovirus, hepatitis B and C, group B Streptococcus, and

bacterial vaginosis are among those infections less well recognized as sexually

transmitted. Those infections for which common and reliable testing is available are

routinely reported to state health agencies as required by law. For other less frequently

reported infections, there are generally no reasonably priced and sensitive tests available

for use by clinicians to assist in diagnosis. As a consequence, reporting is dependent upon

clinician recognition of syndromes and diagnosis.

Term low birth weight (TLBW) is the term used to indicate a live born infant

whose weight is less than 2,500 grams and 37 or more weeks gestation.

Unsatisfactory specimen indicates a Gen-Probe test specimen that contains

excessive amount of blood, mucous, or other material that interferes with the testing

procedure and is reported as an unsatisfactory specimen.

Very low birth weight (VLBW) is the term used to indicate a live born infant

whose weight is less than 1,500 grams and less than 37 weeks gestation.


The specific aim of this chapter is to review the literature relevant to the research

question. This chapter will consist of eight sections. The epidemiology, biology, and

pathogenesis of Chlamydia trachomatis will be reviewed. The epidemiology of low birth

weight and common risk factors for low birth weight will be covered in other sections.

Finally laboratory diagnosis of chlamydial infections, confounding factors related to

specimen collection, testing standards for the specific test used in this study, and the

sensitivity and specificity of tests available for the detection of Chamydia trachomatis will

be discussed.

Epidemiology and Prevalence of Chlamv/a trachomatis

"Infection in the female reproductive tract (especially in the cervix) can cause

premature rupture of membranes and induce premature labor [and] this is responsible for

many preventable infant deaths," 1950 quote attributable to L C. Knox and J. K. Homer

(McGregor & French, 1997).

Chlamyia trachomatis has led the list of nationally reportable diseases since 1995.

Chlamydia and other sexually transmitted diseases such as gonorrhea, AIDS, primary and

secondary syphilis, and hepatitis B accounted for 87% of the top ten most frequently

reported diseases in 1995 (CDC, 1996b). National reporting for chlamydial infections

dates to 1995 (with the exception of the state of New York, as only New York City

reports, and of Georgia, where only cases diagnosed in women are reportable). The

estimated incidence of chlamydia cases is 4 million annually in the United States, mostly

among adolescents and young adults (National Academy of Sciences, 1996). In Florida

during 1996, Chamydia tracmatis accounted for 52.6% of all sexually transmitted

disease case reports, with 81.3% from females and with 78% from women of reproductive

--- 15-34 US -- 15-44 FIl -- 15-44 US

-- 15-34 F
"s 1 -----1

6W 4 A- -- e-"
W at l, -

| 200-
19941 19951 1996 1997 1998
'US data 1994 1995 not available.
Contat from: Florida Departmmt of Health, 1999; CDC, 1998a, 1999a.

Figure 1. Chlamydia Rates per 100,000 Females, aged 15-44 Years: Florida and the
United States, 1994- 1998.

age of 15-44 years. The distribution is markedly more notable among the youngest of

women with about 44% of cases reported from those between the ages of 15 and 19

years. Figure 1 compares the rate of reported cases per 100,000 female population aged

15-34 years and 15-44 years in Florida to the United States.

Widespread screening activities among different populations and clinic settings

have demonstrated rates of infection from 3% to 25%. The positive rates in 1997 among

women 15-24 years of age, screened in family planning clinics nationally ranged from less

than 4% to greater than 11% (CDC, 199T7b). A national population based survey of 1,144

participants aged 12 to 39 years, of whom 54% were female, demonstrated a 10%

chlamydial infection from urine specimens (Mertz et aL, 1998). The specimens were

collected from persons not seeking medical care and contacted in their households as part

of the study to estimate the prevalence of various diseases and conditions in a non-

institutionalized United States population. Brodine et aL (1998) using ligase chain reaction

(LCR) urine technology reported positive rates of 2.7% among female naval personnel

assigned to an anchored ship, compared to 6.9% among those living on the naval base

located on shore. The average ages for these two groups were 25 and 27 years

respectively. Rates drop off significantly in age groups over 24 years, while serologic

evidence rises to about 50% of the population by age 30 years (Stamm, 1988)

Screening for chlamydia conducted in adolescents demonstrates the highest

positivity in the younger ages. Burstein et aL (1998) reported 24.1 % among adolescents

on initial visit and 13.9% on repeat visits. The investigators prospectively examined 3,202

sexually active females, following them for 33 months. Both urine and cervical specimens

were tested by PCR. Rates in female military recruits were also high. Gaydos et aL (1998)

using urine LCR reported chlamydial infection was 9.2% to 12.2% among 17-year-old

recruits, with rates higher for those from five of the southeastern states, Florida not

included. Screening conducted using EIA among U.S. Job Corps females aged 16-24

years during 1990-1994, shows 14.5% for Florida applicants (Shakarishvii, 1995).

Authors reporting on a prospective cohort study among urban adolescents reported initial

chlamydial infection rates of 23.2% and 20.8% at follow-up (Oh et aL, 1996). The study

conducted in the southeastern United States used the tissue-culture method. Specimens

collected for culture ofNeisseria gonorrhoeae were found to be positive for 11.6% of the

study group at their initial examination and again for 17.1 % at the follow-up screening

12-24 months later. This is not an uncommon finding to observe dual infection with

chlamydia and gonorrhea among a certain proportion of any population studied. From 20-

30% of those infected with gonorrhea are co-infected with chlamydia in other studies

(Hook & Hansfield, 1999; CDC, 1998b).

In Florida, during the first half of 1999 the rate ofpositivity for females in family

planning, STD, and prenatal clinics participating in the infertility prevention project was

4.45%, 10.82%, and 6.42% respectively. (Baden, 1999). The positivity for different

groups of young women tested was 5.44% (15-19) and 3.79% (20-24) for young women

in family planning clinics, 12.88% (15-19) and 12.09% (20-24) for young women in STD

clinics, and 9.21% (15-19) and 5.21% (20-24) for young women in prenatal clinics. These

rates are similar to those observed during 1996 and 1998 among these same populations

(Schmitt, 1996b, 1999).

Chamnydia trachomatis is primarily an asymptomatic infection and disease process

known to contribute to widespread community transmission among unsuspecting sexual

partners. Improved testing technologies in recent years have helped elucidate that the

previously believed disparity between asymptomatic infection in males and females does

not really exist. Males and females both have very high rates of chlamydial infections.

Among female populations screened in Florida as many as 80-85% and among males 70-

75% are asymptomatic when infected (Schmitt, 1999). Nationally, the reported

asymptomatic rate ranges from 50-75% (Cates & Wasserheit, 1991). Following exposure

and infection, symptoms may begin within 1 to 2 weeks. Females generally present with

cervicitis however urethritis is also common. Asymptomatic infection of the rectum or

urethra may accompany symptomatic infection of the cervix up to 50% of the time (Cates

& Wassesheit, 1991; Stamm, 1999). Many women will have only mild symptoms of

vaginal discharge, spotting, lower abdominal pain, or dysuria. Infection may also present

as salpingitis, endometritis, peritonitis, Bartholinitis, perihepatitis, pharyngitis, and reactive

arthritis. Adults, like infants, can present with conjunctivitis and cases ofmyocarditis have

been reported (Stamm & Holmes, 1990; Freund, 1992; Bergstrom & Llbombo, 1995;

Berman et at, 1987; Grayston, Mordhorst, & Wang, 1981; Kessler, Pierer, Stuenzaer,

Auer-Grumbach, Hafler, & Marth, 1994).

The natural history of the infection in a nonpregnant woman is one initially of

cervicitis, with ascent to cause salpingitis, sometimes having first caused endometritis en

route. Without treatment, one-fourth to one-half of women with chlamydia will go on to

develop pelvic inflammatory disease (PID), involving inflammation of the endometrium,

fallopian tube(s), and potential involvement of the peritoneunm. Rates of identification of

Chlamydia trachomatis by culture, antigen, or serology in cases of salpingitis and PID

range from 5 to 55% depending on the clinic setting, geographic site, type technology and

country (Cates & Brunham, 1999; Schachter, 1999a). The leading hypothesis for PID

pathogenesis is that endometrial and Chlamydia trachomatis and Neisseria gmonorrhoeae

initiate tubal infection. Then secondary groups of anaerobic and aerobic bacteria may

invade to contribute to the inflammatory disease process (Cates, Rolfs, & Aral, 1990;

Martens, Young, Uribe, Buttram, & Faro, 1993). Reported recovery from women

examined by laparoscopy has ranged from 10% to 80%, with secondary bacteria

recovered much less frequently. More often in clinically milder or "silent PID," chlamydia

is recovered or there is immunologic evidence of recent infection with Chlam)dia

trachomatis (Patton et al., 1994). Tubal scarring and development of tubal infertility

follow the acute or silent PID. This same scarring can set the stage for later life

threatening ectopic pregnancy events. Moore and Cates (1990) suggest that infertility may

follow either acute or clinically detected PID and silent salpingitis. They and others

provide ample evidence to suggest that the majority of tubal factor infertility follows

events of silent salpingitis, in women who report no history of PID but demonstrate

serologic evidence of prior chlamydial infection (Cates & Wasseirheit, 1991; Patton et al,

1989; Westrom, 1975, 1994; WHO, 1995). In contrast the Wolner-Hanssen (1995) in-

depth study using laproscopy and questionnaires suggests that 'silent' PID is secondary to

the failure of the medical community to elicit more complete information from a patient

regarding their menstrual history, abdominal pain, and episodes of infection. The author

does not suggest that chlamydia is not associated with PID, merely that the silent or

atypical status of the PID experienced by women is likely overstated and a result of failure

to elicit complete medical histories.

Studies conducted among pregnant women have identified rates of less than 6% to

close to 33% infected depending on the age, clinic setting, and area of residence. Allaire

and others found a rate of 14.8% among a high-risk indigent obstetric population using

both rNA hybridization and enzyme immunoassay (1998). Nearly 21% prevealce was

reported by researchers who studied an adolescent pregnant group using immunoassay

(Gittens, Nichols & Apuzzio, 1994). In that same study 25% had more than one sexually

transmitted infection. Cohen and colleagues (1990) reported 5.75% prevalence using

direct antigen methods. Another group of researchers who cultured vaginal lavage

specimens after premature rupture of the membranes found 14% positive for chlamydia

(Harger et al, 1991). Nearly 22% prevalence based on culture was reported for initial

prenatal visits among urban lower socioeconomic women (Ryan, Abdella, McNeeley,

Baselski, & Drummond, 1990). Using two antigen detection systems researchers reported

that specimens collected from pregnant women had higher rates of inclusions than those

collected from nonpregnant women. However the difference was not statistically

significant (p < 0.096) in this study conducted with a population whose prevalence was

9.1% among nonpregnant women and 12% among pregnant women (Smith et aL, 1987).

The natural history of the infection in a pregnant woman is less well understood.

There are inconsistent reports and evidence of disease progression from initial chlamydial

infections. Among pregnant women infected with Chlamydia trachomatis, fetal loss has

rarely been reported, premature delivery is experienced by 10-30%, and perinatal infection

by 40-70% (Jones, 1999). There is evidence to support a progression from cervicitis with

ascent to cause intrauterine infection. Greater than 11% of infants born to women with

cervical infection were found to have antichlamydial antibodies in their cord serum (Fejgin

et aL, 1997). Harger and colleagues (1991) reported a different finding ofno chlamydia

positive cultures from amniotic fluid in their study of sub clinical chorioamnionitis among

asymptomatic afebrile women in pre-term labor with intact membranes. Among patients

with premature rupture of membranes in another study, the presence of chlamydial

infection neither increased the incidence of chorioamnionitis nor decreased the latent

period from rupture of membranes to delivery (Ismail, Pridjian, Hibbard, Harth, &

Moawad, 1992). Chlamydia was identified by amniocentesis from a case of induced labor

secondary to suspected chorioanmionitis (Thomas, Jones, Sbarra, Cetrulo, & Reisner,

1990). The cervical specimen was culture positive for Chlamydia trachomatis, CaMndidia

albicans, U urealyticwnum, and group B streptococci. Chlamydial elementary bodies were

identified by fluorescent stain in both amniotic fluid and placental tissue specimens

suggesting that only these organisms ascended from the lower genital tract to cause

infection in the amniotic fluid and fetal membranes.

Stillbirth or neonatal death was reported ten times more often among chlamydia

positive women in a study matched with controls for age, marital status, socioeconomic

conditions, pregnancy order, and race (Martin et al., 1982). This pregnant population had

a cervical infection rate of 6.7%. Gencay et aL reported a stillbirth at 36 weeks gestation

with chlamydia DNA positive placental tissue and histologic evidence of Chlamydia

trachomatis (1995). Thorp and colleagues reported a fetal death at 34 weeks with

histologically confirmed Chlamydia trachomatis pneumonia on autopsy (1989). Fetal loss

perhaps may be more accurately estimated from animal models. Mice models suggest

19.2% intrauterine fetal demise among those chlamydia positive (Oshiro, 1994).

Postpartum endometritis will follow approximately one-third of cases of cervical

infection during pregnancy with development of symptoms at 48 hours after delivery

(Schachter, 1999). As reported by Watts and Brunham (1999), Wager and fellow

researchers in 1980 found 22% of pregnant women with cervical infection during

pregnancy developed late postpartum endometritis (1999). Paavonen et al (1985)

suggested nonpregnancy related chlamydial endometritis is characterized by plasma cell

infiltration of the endometrium. This raises the possibility "that such infections are

associated with failure of implantation or early pregnancy loss due to spontaneous

abortion." However, Sozio and Ness (1998) do not support a relationship between acute

chlamydial infection and the subsequent development of spontaneous abortion.

Bell and others (1994) examined the perinatal transmission in relationship to mode

of delivery. With the use of both culture and serology, they concluded that chlamydia may

be transmitted more often than is suggested by other reports. The transmission rate to the

infant was 60% among infected women delivering vaginally (75 of 125 infants). Those

women who delivered by caesarian section were not significantly less likely to be infected

than those delivered by the vaginal route with cephalic presentation, however the numbers

studied were small (10 infants delivered by caesarian section). Two often infants (20%)

delivered by caesarian section were later found to be infected in the conjunctiva or

nasopharynx. Cord blood was tested for IgM antibody to Chlamydia trachomatis for 26

ofthe infants included in the study. In all cases the cord serology was negative; a positive

IgM which would have indicated prior intrauterine infection.

As with other adverse outcomes of pregnancy, the causal link between known

infection with Chlamywdia trachomatis and LBW has not been conclusively established.

Recent studies have shown an increased risk of low birth weight and premature rupture of

the membranes linked to recent chlamydial infection while others failed to identify any

associations. Gencay et aL (1995) reported that gestational age was longer among IgG

and IgM sero-negative infants. They also found less chorioamnionitis and atelectasis and

pneumothorax among the sero-negative group. In another study designed to examine the

effect of treatment on pregnancy outcome, low birth weight was reported in 19.6% of

those infants with infection who were not treated as compared with 11.0% that were

treated (Ryan, Abdela, McNeeley, Baselski, & Drummond, 1990). This finding was highly

significant, at 95% CI, p<0.0001. Others found an odds ratio of 1.5 for low birth weight

associated with chlamydia positivity (Gravett et al, 1986). Harrison et al (1983) identified

the presence of IgM antibodies among infants born with low birth weight in a population

with 8% prevalence on culture. Martius et at, (1988) reported an even higher odds ratio

of 3.9 for chlamydia positive pregnancies to be associated with premature rupture of

membranes or pre-term labor.

Investigators of the Johns Hopkins Study of Cervicitis and Adverse Pregnancy

Outcome reported an odds ratio of 2.4 for intrauterine growth retardation in a population

with 15.5% positivity (1989). In contrast, Germain and colleagues (1994) found no

association when cultures were taken at 23-26 weeks. Cohen, Veffille, and Calkins (1990)

identified a non-significant reduction in low birth weight and small-for-gestational-age

infants among another treated group when compared to non-treated controls. Hardy et al

(1984) only found association for chlamydial infection and low birth weight if co-infected

with Trichonnas vaginalis. One unique aspect of this study however was the destruction

of the chlamydial McCoy culture cells by the protozoa. This aspect may have confounded

the findings of a lack of an association between low birth weight and chlamydia infection.

In 1991, Much and Yeh observed a significant difference in the incidence of low birth

weight between two groups when one was treated with erythromycin, p 0.05. Clearly

additional data is needed to help clarify the relationship between chlamydial infections

during pregnancy and adverse outcomes.

The difficulty even in the existence of either epidemiologic or statistically

significant evidence for a relationship between chlamydial infection and adverse pregnancy

outcomes is that neither type of study has provided enough information about the

sequence of events. Therefore it is difficult to demonstrate a direct cause and effect

relationship between chlamydial infections and adverse pregnancy outcomes. Additionally,

the variable signs, symptoms, test types, definitions of variables, timing of the specimen

collection during the pregnancy, and inclusion or exclusion of specimen collection to test

for other STDs reduces the value of prospective cohort studies conducted to explore the

associations between Chlamydia trachomatis and pregnancy outcomes.

In the United States Chlamydia trachomatis is primarily a sexually acquired genital

infection. However the infection is also a serious congenitally acquired infection. Risk of

any chlamydial infection among infants born to infected women is estimated at 50% to

75%. Between 35% to 50% of infants born to infected mothers will go on to develop

conjunctivitis and 8% to 22% will develop pneumonia (Harrison & Alexander, 1990;

Crombleholmhne, 1991). "Initial perinatal infection involves mucous membranes of the eye,

oropharynx, urogenital tract, and the rectum" (page 57, CDC, 1998b). The typical course

is inclusion conjunctivitis first noted at 5 to 12 days of age. If left untreated conjunctivitis

may result in corneal scarring and vascularization. It may also be aysmptomatic and self-

limited (Schulz, Schulte, & Berman, 1992).

Among infants born to infected mothers, 20% to 50% wll develop conjunctivitis

and 10% to 20% will develop pneumonia and chlamydial respiratory disease syndrome,

with increased sensitization of the infant to further chlamydial infections (Datta et aL,

1988; Schachter et al., 1986). Schachter et al (1986) reported sub-clinical rectal and

vaginal infections in 14% of infants at risk ,secondary to exposure during birth to maternal

infection. Of interest is the variable temporal delay between sites for identification of

positive isolates. All conjunctival infections were detected in less than three weeks.

Nasopharyngeal infection was detected sporadically during the first three months, often in

the later period with pneumonia. Neonatal rectal isolates overall were detected after two


Datta and colleagues (1988) suggested that appreciable infant morbidity might be

associated with higher rates of chlamydial prevalence in pregnant women. In their cohort

study they compared morbidity in chlamydia exposed infants to non-exposed infants.

Among the exposed infants, 37% developed opthalmia neonatorum, and 12% pneumonia

(with one fatality). The reported rates of sequelae among those neonates exposed to

Chlamydia trachomatis ranged from 12% to 80% for ocular infection, and pneumonia, as

compared to 0% to 60% in the non-exposed group of infants. This study was a sub-sample

of a parallel study comparing the efficacy oftetracycline ointment and silver nitrate

solution for postnatal ocular prophylaxis, (Laga et aL, 1988). Inadequate treatment

response was reported to each of the ocular prophylaxis for both conjunctival

complications and pneumonia, extending the morbidity effect. Overall, the incidence of

opthalmia neonatorum was reduced by 77% with the use of tetracycline ointment and 68%

with the use of silver nitrate solution. Pnuemonia remained the major complication among

infants with perinatally acquired chlamydia. The occurrence of both chlamydia pneumonia

and opthahnia neonatorum was not prevented by ocular prophylaxis (CDC, 1998b).

The epidemiology of opthalmia would suggest that there is a disparate relationship

between the prevalence of chlamydia in the community and the recognition ofnewborn

cases, due to the uncommonly low reporting levels. Authors of one recent survey study

found that providers only reported 42% of gonorrhea, 56% of chlamydia and 58% of

primary and secondary syphilis diagnoses (Hammett, Kaufinan, Faulkner, Hoagin, &

Battaglia, 1996). Between the years 1994 and 1997 the total number of reported

chlamydia opthalmia cases for the United States ranged from 152 to 262, while for the

same period 48 to 1,560 cases of gonorrhea opthalmia were reported (CDC, 1995, 1996b,

1997b, 1998a). These reported cases inversely reflect the level of reported gonorrhea

nationally in women of reproductive age.

Additionally, one is reminded of the inefficacy of drugs currently used for ocular

prophylaxis of chlamydia opthalmia at the time of delivery (CDC, 1998b). Drugs currently

recommended for ocular prophylaxis include 1 percent silver nitrate, 1 percent tetracycline

ointment and 0.5 percent erythromycin (Gutman, 1999; Schachter et al., 1986). While

efficacy of these drugs is high for prevention of gonoccocal opthamhnia, their efficacy for

preventing chalmydia opthalmia is unacceptable. In short, there is no "gold standard" at

present for efficacious and comprehensive ocular prophylaxis. Hammerschlag (1999)

summarizes the numerous studies and concludes that the information is inconclusive. He

notes that none of the drugs studied for ocular prophylaxis is reported to be universally

efficacious. He ultimately recommends that effective screening and treatment during

pregnancy remains the most effective method of control

Diagnostic and political considerations further complicate the situation. Nearly all

cases of opthalmia are diagnosed after discharge from the hospital Such cases are often

sub-acute and diagnosis requires appropriately collected specimens that contain

conjunctival cells, not exudate alone, and the use of sensitive and specific tests. Many

clinicians would prefer not to suggest to parents) that a newborn infant be tested for a

sexually transmitted disease. Accordingly many will empirically treat infants exhibiting

suggestive symptoms with antibiotic ointments, without collecting a specimen.

Consequently there are no positive test results reported to authorities. No comprehensive

review of all causes ofneonatal opthaimia and pneumonia has been conducted nationally

or in Florida in recent years. In one study conducted in Florida the authors identified a

higher association between gonorrheal opthalmia with hospitals using eyrthromycin for

ocular prophylaxis (Desenclos, Garrity, Scaggs, & Wroten, 1992).

Infants may also present with pneumonia, following intra-partum exposure to

Chlamyndia trachomatis (Stamm & Holmes, 1990; Freund, 1992). Chlamydial pneumonia

is a far more serious disease sequelae in this country than is conjunctivitis. Often infants

may continue to have reduced pulmonary capacity well into childhood. This is

demonstrated by abnormal pulmonary function tests and obstructive lung disease as

identified in the study by Weiss et al., 1986 (as cited in Hammerschlag, 1999). Between 8

to 22% of infants born to infected mothers will develop pneumonia (Harrison &

Alexander, 1990; Crombleholme, 1991). Chlamydial pneumonia presents with a repetitive

staccato cough accompanied by tachypnea. Hyperinflation and diffuse infiltrates are noted

on chest x-ray. Fever and wheezing are uncommon. Reliable diagnosis is a more invasive

affair than testing for ocular infection, with tracheal aspirates and tissue culture more

sensitive and specific than nasopharygeal specimens for non-culture testing. Effective

treatment often requires a second course of the recommended antibiotic therapy, due to

only 80% efficacy of the recommended eyrthromycin (CDC, 1998a).

There are less common reports of adult conjunctival infection. Conjunctivitis of

adults results from hand-to-eye self inoculation or partner inoculation during sexual

contact. Worldwide conjunctival Chlamydia trachomatis is primarily an endemic disease,

spread person-to-person or through unsanitary conditions. Affecting several hundred

million persons the disease process is both hyperendemic and holoendemic resulting in

blindness for millions of people. In holoendemic areas young children acquire the

infection, primarily from infected individuals or unsanitary conditions, with many infected

by two years of age. In these communities the rates of blindness are the highest following

a process of chronic follicular keratoconjunctivitis that results in corneal damage and

scarring of the eyelid. Many infected persons in hyperendemic areas sustain less permanent

damage, however blindness or badly scarred conjunctiva are not uncommon (Schachter,

1999a; Schachter, 1999b). Lymnphogranuloma venereum (LGV) is another manifestation

of chlamydial infections, caused by serovars LI, L2, L3 and L4 and affecting lymphatic

tissues. After a process of inflammation there is formation of abscess, fibrosis, and

obstruction of lymphatic pathways and disfigurement.

Only recently, literature has begun to identify cervical cancer as a sexually acquired

condition. Primarily these associations are based on studies examining the role of human

papilloma virus. A few studies have suggested that Chkmwydia trachomatis may contribute

a more direct effect to the development of cervical dysplasia rather than act only as a

synergistic bystander (Paavonen, Koutsky, & Kiviat, 1990; Lindner, Geerling, Nettum,

Miller, & Atman; Yla-Outinen, Lehtinen, Romppanen, Luoto, Rantala, & Paavonen,

1990). Recently Paavonen (1999 as cited by Jancin, 1999) reported that Chlamydia

trachomatis was significantly associated with invasive squamous cell cervical carcinoma.

This association was significant after adjustment and in the presence of IgG antibodies to

the serovars G (OR 6.6), D (OR 3.1), I (OR 4.4) and E (OR 2.3).

Other researchers have identified that cervical epithelial changes are significantly

associated with several reproductive tract infections Singh and colleagues (1995)

examined a population of women attending a maternal health center in New Delhi, India.

A specimen was collected for a Pap smear. Other specimens were collected to screen for

chlamydia, gonorrhea, trichomoniasis, bacterial vaginosis, herpes simplex virus, genital

warts, HIV, syphilis, and candidiasis. If indicated, the woman also received a colposcopic-

directed biopsy for any atypical lesions. After controlling for interaction the adjusted odds

ratio for an association with inflammatory epithelial changes and chlamydia was 21.3, 13.5

for human papillomavirus, and 22.6 for bacterial vaginosis. They observed a significant

additive effect: two infections increased the magnitude of inflammatory changes by

adjusted OR 31.4 and three infections by 72.6 fold. Young age was mildly protective and

parity greater than one increased the risk by OR 1.7

Today in an age of fatal STDs like HIV the rate of chlamydial infection in a

community is epidemiologically significant. A growing body of research provides evidence

that the population attributable risk from chlamydial infections is a predictor of increased

HIV transmission rates among young women and their partners (Laga et aL, 1993;

Plummer et aL, 1991). These researchers studied prostitutes prospectively and reported

adjusted odds ratios associating chlamydia with HIV-1 sero-conversion ranging from 3.2 -

5.7 with a median of OR 4.5. Laga and colleagues also calculated a population attributable

risk in their study population of 22% for increased HIV transmission in the presence of

cervical chlamydia. This was more than five time the attriutable risk from genital ulcer

disease in this group.

In summary Chiamydia trachomatis is associated with a profound scope of disease

manifestation in the reproductive age woman. These processes range from mildly annoying

symptoms to life threatening events. When pregnant women are infected with this bacteria

the sequelae may range from spontaneous abortion, premature rupture of membranes, pre-

term labor to vertical transmission that results in low birth weight, stillbirth, and infantile

pulmonary infections that may cause lifelong damage to the infant's respiratory system.

The fill impact of this sexually transmitted infection remains to be establishedL

Epidemiology and Prevalence of Low Birth Weight

The Florida Office of Vital Statistics reported the rate of low birth weight was

8.5% in 1970 and dropped to 7.5% in 1980 (Office of Vital Statistics, 1996a). Between

the years from 1985 to 1995 the LBW rate hovered between 7.4% and 7.8%. From 1996

there has been a gradual upward shift to 8.1% total LBW for 1998, with an incrementally

upward change also for very low birth weight (VLBW) over the same period (Office of

Vital Statistics, 1999). The observations about rates in Florida mirror the gradual upward

shift in percentage of low birth weight reported nationally. However the national rates for

1998 are not yet available for comparison. Overall the percent of VLBW and total low

birth weight is higher in Florida than for the United States over the period of time

presented in Table 1 below.

Nationally the percentage of low birth weight for singletons among Hispanic

women has changed very little since 1989 at 5.35% to 1997 at 5.43.% The rate among

non-Hispanic black women has declined slightly during this same period from 12.22% to

11.46%. Among white non-Hispanic women, there has been more fluctuation with an

overall trend that has been a slightly upward from 4.60% in 1989 to 4.95% in 1997

(Ventura, Martin, Curtin, & Mathews, 1999), The rate of low birth weight among

adolescents 15 to 19 years old in Florida during 1996 was 1.9% for very low birth weight

and 9.9% for low birth weight. These numbers are comparable to national rates for low

birth weight at 9.5%.

Table 1. Comparison of Percent of Low Birth Weight: Florida and the United States,

Florida United States

Very Low Low Birth Very Low Low Birth
Birth Weight Weight <2,500 Birth Weight Weight
<1,500 Grams Grams <1,500 Grams <2,500 Grams

1988 1.4 7.7 1.2 6.9
1989 1.5 7.7 1.3 7.0
1990 1.5 7.4 1.3 7.0
1991 1.4 7.4 1.3 7.1
1992 1.5 7.4 1.3 7.1
1993 1.4 7.5 1.3 7.2
1994 1.5 7.8 1.3 7.3
1995 1.5 7.7 1.4 7.3
1996 1.5 7.9 1.4 7.4
1997 1.5 8.0 1.4 7.5
1998 1.6 8.1 -

Content adapted from Office of Vital Statistics, 1996a, 1999; Ventura, Martin, Curtin, &
Mathews, 1998.

The societal impact from low birth weight infants is enormous in terms of medical

health care costs and other long-term adverse outcomes. The average cost for the first

year of a low birth weight infants medical care has been estimated to exceed $28,058 (this

estimate is from the Office of Technology Assessment, an 1988 estimate adjusted for 1998

dollars by Department of Health, Office of Health Planning and Evaluation.) Recent

analysis on Florida birth records for 1985 to 1990 suggest a significant association

between low birth weight of 1,500 and 2,499 grams and physical impairment (OR 4.35),

profoundly mentally handicapped (OR 4.75), educable mentally handicapped (OR 2.62),

and academic problems (OR 1.26) (Resnick et aL, 1999). Low birth weight infants are at

increased risk ofneonatal and infant morbidity and mortality; nearly 70% of all infant

mortality, nearly one third of all handicapping conditions (Patient Outcomes Research

Team, 1998).

Biology of Chlamydia trachomatis

Chlamydia trachomatis is an intriguing pathogenic bacteria with a long history of

human contact and a distinctly unique growth cycle. Pathology associated with the

bacteria was first described in Egyptian papyri (Schachter, 1999a). Most likely it predates

humankind as a species by billions ofyears. Halberstaedter and Prowazek first stained

conjunctival scrapings from orangutans infected with human trachomatous matter and

demonstrated the presence of inclusions in 1907. In 1914 Lindner isolated inclusions from

the conjunctival of infants, the genital tracts of their mothers, and the urethras of their

fathers (Schachter, 1999a). T'ang and associates contributed the first isolation of

Chlamydia trachomatis from persons infected with LGV during the 1950s (as cited in

Schachter, 1999a). In 1959 Jones, Collier and Smith recovered the bacteria from the

cervix of a woman whose infant had opthalmhnia neonatorum (as cited in Schachter, 1999a).

Other historical milestones are summarized in Table 2 below.

Table 2. Historical Milestones in the Recognition and Study of Chamydia trachwmatis.


18e century












Egyptian papyri contain description of trachoma.

John Hunter first described Lymphogranuloma Venereum (LGV).

Halberstaedter and Prowazek stained conjunctival scrapings from
Orangutans infected with human trachomatous matter and demonstrated
the presence of inclusions.

Lindner isolated inclusions from the conjunctival of infants, the genital
tracts of their mothers, and the urethras of their fathers.

Chlamydia trachomatis isolated from persons diagnosed with LGV by
Macchiavello. Isolate unfortunately lost before confirmed by others.

Respiratory infection in infants first reported by Botsztejn.

Chlamydia trachomatis isolated from persons diagnosed with LGV by
T'ang and associates and later confirmed by other researchers.

The first isolate of Chlamydia trachomatis from the genital tract (non-
LGV) by Jones, Collier and Smith. Obtained from the cervix of a woman
whose infant had opthalmia neonatorum.

Chlamydia trachomatis recovered from male urethras, in association
with epidemiologic studies of conjunctivitis.

Gordon and Quan developed the first clinically useful laboratory
procedure for diagnosis with a tissue culture isolation technique of
intracytoplasmic inclusions that allowed for results in 48-72 hours.

Dunlap and associates demonstrated that up to one third of men with
nongonococcal urethritis had carriage of Chtamydia trachomatis.

Affordable and sensitive non-culture tests become widespread.

Highly sensitive and specific amplified testing available.

Content adapted from Schachter, 1999b; Hanerschlag, 1999; Chemesky, 1999.

Order: Chamydiales

Family: Chlamydiaceae

Genus: Chlamydiales


Psifta affects cats, birds,
humans, and hoofed animals.
Includes the following human
serovars: D85711, D85712,
Cal-10, and human
m Pneumonititis.

Pecorum: affects cattle,
sheep and koala bears.

Phneumonmwae. affects equines and
humans. Includes N16 the only
equine serovar and the following
human serovars; PI-Parola,
S562B, 10L207, TW183.

TrFchomtis: affects humans and other
mammals. Includes the following
serovars that effect humans unless
indicated in parenthesis.

I MoPn (mouse),
SS45, R22 (pig),
Lymphogranuloma SFPD (hamster)

L1,L2,L2s L3

B, D, Da, D,
J, Ja, K

Content adapted from Tanner, Harris, & Pace, 1999; Schachter, 1999a; Stamm, 1999;
Morre et alt, 1998.

Figure 2. Chlamydiales Tree.

A,B, Ba, C

Two of four species in the genus Chlamydia, Chlamydia trachomatis and

Chlamydia pneumomae, prefer humans as their natural host. Please refer to Figure 2. The

other two species are C. psittaci and C. percorum that prefer birds and lower mammals

respectively. However there is cross-over between human, bird and mammalian strains.

Twenty-three serovars and perhaps eleven variants for Chlamydia trachomatis have been

isolated from human specimen by scientists (Schachter, 1999a; Morre et aL, 1998; Tanner,

Harris, & Pace, 1999; Stamm, 1999).

Developmental Cycle

Due to the unique developmental cycle, chlamydiae merited their own order

Chlamydiales, and a single family, Chlamydiaceae. In their unique growth cycle the

bacteria alternate between two morphologic forms. The elementary body (EB) is adapted

to the extracellular environment, while the reticulate body (RB) is specialized for

intracellular growth processes within the host. The metabolically inactive EB is the

infectious form and the RB the metabolically active or replicating form. There are distinct

steps in the chlamydiae developmental cycle. The first step is attachment of the elementary

body (the infectious particle) to the host cell. Second is entry into the cell. The reticulate

particle then undergoes morphologic changes, with intracellular replication and growth.

Further morphologic changes of the reticulate particle to elementary bodies, is followed in

the final step with release ofthe infectious particles (Schatcher, 1999a; Schatcher, 1995).

The steps are illustrated in Figure 3 below.

Wyrick (1998) provides a concise summary of the current body of information on the cell

biology of the chlamydiae gleaned from the scientific community in recent years. The EB

is metabolically inactive and resistant to most environmental challenges, but capable of

rapid attachment to the host ceil This attachment is generally held to be aided by some

mechanism that triggered adhesion with the assistance of the major outer membrane

protein, and generally was held to have occurred within an hour of infection. Newer

scientific knowledge suggests adhesion may instead be a parasite-specified phagocytosis,

receptor-mediated endocytosis in clarthrin-coated pits, pinocytosis in non-coated pits,

chlamydial heat shock protein70 interaction with ATP hydrolysis or movement assisted by

host microvilli (Raulston, Davis, Paul, & Wyrick, 1998; Patton, Cummings, Cosgrove,

Yvonne & Kuo, 1998; Wentworth, Judson, & Gilchrist, 1991). Most noteworthy is the

sheer speed with which the EB can enter the host. Electron microscope observation of the

'in vivo" uptake process has recently demonstrated that the EB are assisted by the

epithelial cell's mnicrovilli and that internalization is complete within five minutes, and has

reached the Golgi apparatus within ten minutes post infection (Patton et al, 1998). This

group of researchers suggests that the rapid uptake reflects the bacteria's need to reach

the host energy source, as it is incapable ofproducing energy on its own. Once inside the

host cell, a vacuole is formed to envelop the EB, where the EB remain throughout the

growth cycle. After a rapid initial fluctuation in pH, a triggered release oftyrosine

phosphoylation of the epithelial proteins follows and rearrangement of the chlamydiae's

cytoskeleton occurs. There is then an accumulation of F-actin and clathrin which it is

believed, aids to redistribute the EB to the peri-nuclear region of the host cell and the

resultant highly permeable vacuole formation (Wyrick, 1998; Schatcher, 1999a; Schatcher,

1995). The vacuole becomes what microbiologists term the "chlamydial inclusion body."

At this point the chlamydiae "prime" their host for obligate intracelhDular

Release at 48-72 hours of
as many as 100-1,000
elementary bodies.
1. Rupture
2. Fission
3. Disimagration


18-24 hours

Initial host cell


Host invasion at 5 minutes,

ion at

reticulate bodies to
elementary bodies;
some continue
binary fusion; one
can also observe

Loss of rigid cytoskeleton
and formation into the
permeable inclusion body
with rapid relocation to
close proximity of Golgi
apparatus at 10 minutes.

Binary fusion of N
reticulate body at 8-
18 hours.

Maturation of elementary
body to reticulate body at
4-8 hours.

Content adapted from Schachter, 1999a, Wyrick, 1998; Schachter, 1995; Wentwrth, Judson & Gilchrist,
1991; Patton, Cummings, Cosgrove, Yvonne, & Kuo, 1998; Martin, 1990, Neeper, Patton, & Kuo, 1990,
Fedorko & Smith, 1991; Beatty, Morrison, & Byme, 1994.

Figure 3. The Developmental Cycle of Chlamydiae.


growth and trigger maturation of the EB into RB. An obligate parasite, chamydiae utilize

the host cell's glycogen, metabolites and ATP in all steps of the developmental cycle. The

RB is somewhat larger and richer in RNA than the elementary body. The cycle then

progresses to binary fusion of the RB lasting from 20 to 24 hours. During this time

another unique activity can be observed: the fusion of multiple inclusions within the host

cel to create a single inclusion. After a period of reorganization, triggers that are unclear

cause many of the RB to mature into a new group of the smaller EBs in preparation of

extracellular exodus. Some RB will continue to divide. At this stage one can also observe

intermediate forms (IB) of the bacteria as well (Moulder, J. W., 1974, as cited in

Schachter, 1999a). Forty eight to seventy two hours after attachment to the host cell the

mature EB is ready for release to infect other epithelial cells. By this time the EBs will

occupy the entire cytoplasm of the host cell (Neeper, Patton, & Kuo, 1990). A single

inclusion body may release from 100 to more than 1,000 elementary bodies at maturity

(Martin, 1990). This release appears to occur in three modes, a volcanic like eruption,

fusion with plasma cell membrane, and a gradual disintegration of the cytoplasm, allowing

lateral invasion to near by cells. In short, scientists remain challenged by the study of the

chlamydiae's developmental cycle with more questions raised than answered in recent


When protected by the vacuole, chlamydiae are able to resist the usual hostile

onslaughts of the host cell The intracelhilar form of Chamydia trachomatis is protected

from phagolysosomal fusion, by mechanisms not fully understood. The bacteria is sensitive

to iron limitations and shows enhanced uptake of some proteins, e.g., hsp60. The parasite

appears able to accept whatever phospholipids available in the host eukayotic cel. The

extracellular form of Chlamnydia trachomatis is inhibited by penicillins, but is not killed. In

this form it is sensitive to temperature, freezing, and dessication. Unlike C. psittaci that

can live for months in contaminated cat litter, C. trachomatis will die at 56C after 30

minutes. C trachomatis is also sensitive to common household disinfectants (Wyrick,

1998; Schachter, 1995; McCalarty & Hatch, 1998).

Pathophysiology and Pathogenesis

The clinical syndromes associated with Chlamydia trachomatis infection are well

recognized in the public health community. The pathogenesis of any of the infections

attributable to Chiamyda trachomatis is not well understood. Of the twenty-three

identified serovars, nine are predominant in the genital tract and in infantile pneumonia: D,

E, F, G, H, I, J, and K, with D, E, and F found most frequently worldwide. One, B, has

been identified in both ocular and genital sites (please refer to Figure 2.). Numerous

serovars and variants have only been identified in the last few years (Stephens et aL, 1998;

Stammi, 1999). Serovars in the C complex grouping (A, C, H, I, J, K) have been identified

as more likely to be associated with symptomatic rectal infection among homosexual

males (Boisvert, Koutsky, Suchland, & Stammn, 1999). It has been hypothesized that

severity of disease may be linked to the serovar. Some studies have supported this;

however, serovar F has often been identified in both less asymptomatic or less

inflammatory infection as well as in women with upper genital tract infection.

Squamocohnmnar epithelial cells are the primary target for non-LGV infections and

macrophages for LGV serovars. Initially, focal inflammation with infiltration of

polymorphonuclear neutrophils occurs. Mononuclear cell infiltration follows. Abundant

immune response occurs that includes circulating antibodies and cell-mediated responses

with both CD-4 and CD-8 T-helper cells. The CD-4 t-cells potentiate the immune

response. It appears resolution of the infection is supported by CD-8 cells (Brunham,


Three models predominate among the numerous models for pathogenesis under

consideration (Peeling & Brunham, 1996; Chlamydia Genome Project, 1999). There is a

considerable body of work to support consideration of delayed-type hypersensitivity

(Beatty, Morrison, & Bymrne, 1994; Rank, Sanders, & Patton, 1995 Schachter, 1999a). In

this model, re-infection with the same or different serovars leads to inflammation and

scarring of the affected mucous membrane. It is possible that the delayed-type

hypersensitivity (DTH) model may also result from reactivation of latent infection,

possibly triggered by altered levels of steroid hormones. Exogenous progesterone is used

in mouse model studies to enhance uptake of chlamydiae, and one study in ewes suggested

reactivation of latent chlamydial infection after either estrus or progesterone treatment

(Patton & Lichtenwalner, 1998). Others mention that both estrogen and progesterone

enhance the growth, survival, and ascent of Chlamydia trachmatis in female animal

models (Krettek, Ark, Chaisilwattana & Monit 1993). While it is unwise to assume

findings from animal studies are directly applicable to human pathogenesis of chlamydiae,

the role of steroids in DTH merits further investigation.

That latent infection can persist has been a controversial topic among researchers.

Some have provided compelling evidence that suggest persistence of a single strain over a

period of two to five years (Dean, Suchland, & Stamm, 1998). This group of researchers

used serial ompi genotyping on recurrent culture-positive specimens and ligase chain

reaction assays on selected intervening culture-negative specimens. In recent years other

researchers have used either culture or DNA amplification to examine persistent infection

concluding that persistence of infection is not supported (National Chlamydia Committee,

1999a). Hopefully, more conclusive evidence will be forthcoming in the future from


A suggested alternate model is one of an inflammatory response mediated by

human heat-shock proteins and chlamydial heat-shock protein-60 or -70. Chlamydiae

evoke persistent host cell production of these pro-inflammatory cytokines. Strong

serologic levels of chsp-60 in severe pathology were observed by some groups and

reduced serologic response in other groups with milder pathology (Rasmussen et aL,

1997; Morrison, Lyng, & Caldwell, 1989; Patton et al, 1994). Witkin and colleagues

(1997) suggested that this same model may contribute to the reduced success with in vitro

fertilization and embroyo transfer observed among women with high levels of cervical IgA

antibodies to Chlamydia trachomatis. Other researchers question if the chlamydial heat

shock proteins are causally involved in chlamydial immunopathogenesis or merely markers

of persistent infection (Peeling & Brunham, 1996).

The third model for pathogenesis of Chlamydia trachomats is one of, genetic

susceptibility (Brunham, 1999; Peeling & Brunham, 1996). In this model the data

examined supported individual differences in immune responses. Those individuals with

weak cell-mediated and strong antibody response were susceptible to re-infection, slower

to resolve, and demonstrated more inflammation and disease. In contrast, those individuals

with a strong cell-mediated immune response and lower antibody response were less

susceptible to both the infection and disease. While the numerous models continue to be

investigated, the actual mechanism ofpathogenesis in humans has yet to be proven. There

is, however, mounting evidence that the bacteria can and do invade epithelial host cells at

different levels of the female reproductive tract, and at different times during the

reproductive cycle, e.g., pregnant and non-pregnant states. The role of persistent

chlamydial infection is also unclear in the process ofpathogenesis. Beatty and colleagues

(1994) reviewed the conflicting evidence for persistent infection and concluded that

further work following 'infected" culture negative persons should be conducted. Byrnme

(1996) also discussed the work to date on persistence and suggested that certain features

associated with in vivo growth of chlamydiae would further support some form of

persistent infection. These include the presence of abnormally large intracellular forms of

the organism, chlamydial nucleic acid in absence of culturable forms, immunologic

evidence of heightened reactivity to stress response proteins, and continuous presence of

chlamydial antigen. Bragina, Gomberg and Orlova (1998) reported morphological changes

in chlamydial bodies in persons with reported latent chlamydial infection. Others have

recently reported in vitro persistence of chlamydial antigens, and chlamydial particles while

studying antibiotic efficacy (Dreses-Werringloer, Jurgens, Zeidler, & Kohler, 1998).

Neeper, Patton, and Kuo (1990) provide cinematographic in vitro observation of

Chlamydia trachomatis growth cycles in primary cultures of human amniotic cells. Of

significant note, given the difficulty of culturing this bacteriumnn, were the sustained cycles

of infection that occurred, until all amnion epithelial cells in the monolayer had been

infected and destroyed. While in this study the amniotic epithelial cells were isolated from

the in vivo placental structure, the chlamydiae were capable of damaging the placental

tissues. The study does not conclusively demonstrate how the chlamydiae could actually

cross the intact membrane.

Neuer et al (1996) have suggested that the development of heat shock proteins

during mouse embryogenesis is a plausible explanation for differentiated expression of

heat shock proteins in early pregnancy using first-trimester decidua and failure of the

embryo to implant or survive. Gencay et at (1996) isolated Chlamydia trachmatis from

placental tissue. Intra-amniotic chlamydial infection can persist in the absence of clinical

symptoms, in the presence of intact membranes, and following apparent eradication of

chlamydiae (Askienazy-Elbhar, 1996; Morrison, 1996; Ghaem-Maghami, Hay, & Lewis,

1996; Koehler et al, 1996; Gencay et aL, 1997; Neeper, Patton & Kuo, 1990; Brunham,

Holmes, & Embree, 1990; Cmningham, 1995).

Laboratory Diagnosis ofChlamydial Infections

According to Lennette (1995) there are three approaches to laboratory diagnosis

of infections: 1) direct detection of the organism, 2) cultivation of the organism in a

suitable host and 3) use of serology to obtain evidence of recent infection. Direct

microscopic examination of the live organism that causes the infection, direct fluorescent

antibody technique, and detection of nucleic acids through hybridization or amplification

are acceptable technologies for direct detection of the organism. An example of

microscopic examination for a sexually transmitted disease pathogen is darkfield

microscopy of freshly collected specimens from moist or dry lesions and lymph nodes for

Treponemapallmidwn subspecies pallidum or Direct Fluorecent Antibody (DFA-TP)

technique for body fluids or lesion exudate (Larsoun, Hunter, & McGrew, 1991). Other

direct visualization examples include gram stain for Neisseria gonorrhoeae and saline or

potassium hydroxide microscopy for bacteria vaginosis and candidiasis (Lowe & Saxe,

1999). Many commercial enzyme inmnmoassay tests for chlamydial and gonorrheal

antigens, and direct fluorescent antibody for visualization of chlamydia elementary bodies

have been available since the 1980s for wide spread screening programs. (Chemesky,

1999; Ehret & Judson, 1991; Fedorko & Smith, 1991). These tests have a sensitivity of

50% to 75% and specificity of 95% to 100% (Pate, Dixon, Hardy, Crosby, & Hook,


Highly sensitive nucleic acid amplification technology has become available more

recently. This method utilizes either target amplification, probe-amplification, or signal

amplification to detect minute numbers of organisms in a specimen. These tests are vastly

superior overall to culture or any other technology now available with sensitivity ranges of

95% to 100% and 99% to 100% specificity, (Kacena et aL, 1998; Everett, Honmung, &

Anderson, 1999; Quinn et aL, 1996b). Two other studies with LCRL show a similar range

of sensitivity. One conducted in France including women ofboth high (STD clinic) and

low risk (prenatal clinic) for chlamydial infection found an overall sensitivity of 95.2% and

specificity of 99.6% (de Barbeyrac, Rodriquez, Dutilh, le Roux, & Bebear, 1995). The

second study conducted in Florida among subjects from obstetric and gynecological clinics

reported the sensitivity as 97.6% and the specificity of 100% (Davis, Riley, Peters, &

Rand, 1998).

Pooling of chlamydia specimens will reduce the sensitivity slightly from 100% to

98.4%, but not for gonorrhea specimens as reported by researchers (Kacena et al.,1999;

Kacena, Quinn, Hartman, Quinn, & Gaydos, 1999). Pooling can provide even with minor

decreases in sensitivity a significant cost savings of 37% to 46% with pooled sensitivity of

92.8% in pools of four urines and 97.9% in pools of 8 urines (Krepel et aL, 1999). Gaydos

and colleagues (1998) reported lower sensitivity (88.6%) using amplified testing on urine

specimens. Unfortunately, amplified tests remain much more costly than other direct

detection methods. Direct detection offers the advantage of rapid results when compared

to cultivation in culture mediums. Other advantages are generally affordable cost and high

volume capabilities, important considerations for any public health screening program. The

disadvantage of the direct detection approach is variable sensitivity and specificity between

testing technologies and manufactured products, as well as variable sensitivity and

specificity among groups with different prevalence, a phenomenon common to all

screening tests.

Cultivation of the organism in a suitable host is the standard by which all other

detection methods are compared, especially for medical-legal documentation. This is true

for Chlamydia trachomatis, even though cell culture is reported to range from 50% to

90% sensitive (Fedorko & Smith, 1991; Newhall et al., 1999). The range of sensitivity is a

function of prevalence in the population, laboratory expertise, quality of specimen

collection and most often, of specimen transportation. Specimens collected to identify

Chlamydia trachomatis must be inoculated onto cycloheximide-treated McCoy cells. The

specimens should be refrigerated and processed within 48 hours of collection; if the

interval will be longer then the specimens should be frozen at 60 C (Schachter, 1995).

These parameters and the bacteria's sensitivity to heat exposure often adversely effect the

viability of organisms for cell culture during the transportation interval These factors

combined with the cost of culture, have contributed to the widespread popularity of the

direct detection methods commercially available since the 1980's.

Numerous forms of serologic tests have been useful for the detection of sexually

transmitted infections. However, only T. pallidum, C. trachomatis, human

immunodeficiency virus (HIV), hepatitis B virus (HBV), and herpes simplex virus (HSV)

have an adequate immune response in current infection for serologic testing (Chernesky,

1999; Chemesky et al, 1998). With these responses there is also significant variability,

reducing the reliability of this technique for laboratory diagnosis and restricting its

application to specific situations. In the case of Chlamydia trachomatis the antibody

response elicited during infection may be long lived, therefore serology for identification

of lower tract syndromes has not been successful (Black, 1997, 1998). Serology is useful

in diagnosis of infant chlamydial pneumonia and HIV, HBV, and HSV. Serology has been

invaluable in the epidenmiologic study of chlamydial infections in different populations

(Numazaki, 1998; Gencay et aL, 1995; Harrison et aL, 1983; Numazaldki & Chliba, 1996;

Numazaki, Kusaka, & Chiba, 1996; Fejgin et at, 1997; Cohen, Tenenbaum, Michaeli,

Beyth, & Sarov, 1990). First used as a screening test in diagnosis of non-lesion syphilis

before World War 1, serology has proven to be very effective in the control of this STD

(Brandt, 1985; Pynchon, 1964). However, even with positive serology for T pallidum in

an individual for whom a darkfield examination or direct antibody test is not available,

confirmation should be sought with the use of a treponemal test technique to identify false

positives (Chemrnesky, 1999; CDC, 1998b).

In populations with low to moderate prevalence, as observed in the study sample,

the sensitivity of the non-culture tests range from 50% to 96% and the specificity from 93

to 99% (Dinh & Martens, 1993; Stamm, 1999; Newhall et al, 1999). The direct detection,

non-culture test employed to diagnose Chlamydia trachomatis genital infections, Gen-

Probe PACE2C, was used in this study. The test manufacturer's published chlamydia

sensitivity ranges from 92.5% among those with a high (17.8%) positivity rate to 94.3%

among those with a low (4.1%) positivity rate (Gen-Probe Incorporated, 1994a).

Early studies conducted on persons who sought care in Florida county health

departments and elsewhere report a sensitivity of 96.4% for the Gen-Probe PACE2C

combination assay among asymptomatic and symptomatic females and males. The

prevalence ranged from 6.6% to 9.3% for chlamydia and from 6.6% to 56.5% for

gonorrhea, with a combined specificity in this same group of 98.0% (Hale, Melton,

Pawlowicz, Halstead & Wright, 1995; Hale, Melton, Lewis, & Willis, 1993; Schwebke &

Zajackowski, 1996). This technology may be less sensitive with male urethral specimens.

Kluytmans et aL (1991) reported a sensitivity of 70% among a male population with a

culture prevalence of 13.2%. In this same group the female sensitivity was 92.7% and

prevalence 8.6%.

More recent studies conducted with Gen-Probe PACE2C compared with DNA

amplification tests and other non-culture tests suggest that earlier studies may have over

estimated the sensitivity of this nucleic acid hybridization test (Newhail et aL, 1999; Wylie,

1998). Wylie et aL(1998) reported a prevalence of 10.4% and sensitivity of 79.3% among

a population offemales residing in Manitoba Canada. Newhall et al. (1999) found a

prevalence of 3.9% and sensitivity of 75.3% and 75.3% respectively, among populations

of family planning patients from Washington and Oregon.

Gen-Probe PACE2C employs the principle of nucleic acid hybridization method.

The concept behind this technology is based on the re-pairing of specific nucleotide bases

that compose the ribonucleic acid (Foorghani & Erdman, 1995). The double-stranded

target ribosomal ribomnucleic acid (rRNA) of the chlamydia or gonorrhea organisms is

dissociated and re-paired with a chemiluminescent single strand probe which can be read

by the testing equipment. This process is based on a biological in vivo amplification of

rRNA. Many more copies of rRNA (5,000 to 10,000) exist in an individual cell as

compared to a single deoxynucleic acid (DNA) strand. This is a different testing

methodology from DNA amplification utilized in PCR, LCR, TMA, and SDA testing. As

with DNA amplification, rRNA hybridization increases the likelihood that the chlamydia or

gonorrhea organisms will be identified in the specimen as compared to some other

technologies, however the sensitivity does not approach that of amplification.

Confounding Factors and Quality of STD Specimens Submitted for Testing.

Specimens submitted for chlamydia, or other STD testing, may be adversely

affected by any number of clinician-controlled behaviors, laboratory management,

transportation interval, and patient-related issues. Among the many variables controlled by

the level of clinician skills and knowledge are: choice of collection implements, the

presence of excessive blood, mucous, pus or exudate, exfoliated cells, vaginal secretions,

debris, fibers, and lubricants included with the specimenss.

Other factors confounding specimen quality are the availability of supplies in the

clinic, types of instruments and implements as well as their consistent supply. This issue is

one faced by all providers, public health as well as private (Pachciarz et al, 1992; Steiner,

1989). For example, unavailability of'large drum' swabs reduce the ease with which a

clinician can adequately and efficiently clean the cervix of excess exudate or blood. In one

study the use of the cytobrush to collect chlamydial specimens improved the rate of

adequate endocervical specimens but not the sensitivity of the enzyme-linked

immunosorbent assay used (Kellog, Seiple, Klinedinst, & Levisky, 1992). In contrast,

Moncada and colleagues (1989) observed improved sensitivity for both direct flourescent-

antibody and enzyme-linked immunosorbent assays with use of the cytobrush. Another

group of researchers who compared swab type and storage temperature reported that

calcium alginate swabs were toxic to Chlamnydia trachomatis and herpes simplex virus and

that cotton on wood appeared to be inhibitory to chlamydiae (Mahony & Chemesky,

1985). The authors also cite others who reported that wooden shafts were toxic to

Ureaplasma urealyticum and Neisseria gonorrhoeae (Mardh & Zeberg, 1981 as cited in

Mahony & Chemesky, 1985). Their concluding recommendation was to use cotton, rayon

or dacron tips on aluminum or plastic shafts to increase the viability of specimens for


Training and skills of those sampling the cervix is variable and may affect the

quality of the specimen collected. The timing and force with which a swab is rotated

within the cervical os may dictate the likelihood of retrieval of adequate numbers of

columnar cells, and adversely impact on a positive finding for direct detection tests with

lower sensitivity. Because Chlamydiae are obligate intracellular parasites, one must collect

the appropriate host cells, columnar epithelial cells located in the cervical os, or at the

transition zone. Additionally, the clinician needs to apply adequate pressure and vigorous

swabbing to obtain the infected cells (Schachter, 1990).

Contact of the specimen collection swab with the vaginal mucosa on exit from the

vagina may introduce confounding inhibitors that interfere with testing equipment and

create false readings with DNA amplified technologies. Other inhibitors associated with

reduced sensitivity in DNA amplification testing include excessive cervical mucous, talc

from latex gloves, and residual urine remaining after DNA purification (National

Chlamydia Committee, 1999b).

Patient related variables such as age, recent coitus, and topical medications,

lubricants, or spermicides are also potential confounding factors that may adversely affect

the quality of the specimen submitted for testing (Bauman, 1993; U.S. Department of

Health and Human Services, 1989).

After collection, the actual management of the specimen can adversely affect the

quality and the findings. If the recommended temperature is not maintained while in

storage prior to transport or during transport the specimen can deteriorate significantly.

For example, failure to maintain chlamydial specimens for culture under refrigeration or

freezing will reduce viability of the bacteria and the likelihood of a positive culture result

and failure to incubate gonorrhea culture specimens at the correct temperature and period

of time will reduce their vigor. Submission of small amounts of blood with specimens does

not interfere with hybridization test performance; however, grossly bloody specimens may

interfere with assay performance (Gen-Probe Incorporated, 1994a). Vaginal secretions

and excessive mucopus have been found to interfere with different screening tests and

assay performance, including Gen-Probe (Cehim et al, 1994).

Standards on Gen-Probe PACE2C Testing Techniques Within Florida

Testing standards in the Office of Laboratory Services require that all female

specimens are first screened with the PACE2C System, a combination chlamydial and

gonococcal nucleic acid hybridization technique. The combination assay allows rapid dual

screening for the presence of either Chlamydia trachwmatis or Neisseria gonorrhea.

While this first assay does not distinguish between which organisms are present, it is a

cost-effective labor reducing approach in populations with lower prevalence, e.g., prenatal

clinics compared to STD clinics (Hale, Melton, Pawlowicz, Halstead, & Wright, 1995).

All specimens screening positive for the presence of an infection are then re-tested using

both the PACE2 Chiamydia and PACE2 Gonorrhea assays. Additionally, all high negative

specimens and low positive specimens are re-screened (Farthing, Brumback, Morris, &

Wright, 1995). This involves a two step process. The first is a repeat ofthe PACE2 assay

utilizing the chemiluminescent labeled DNA probe specific first for chlamydia, followed by

that for gonorrhea. Next the specimen is tested using the Probe Competition Assay (PCA)

with reagents that winl compete for the target binding sites to form stable DNA-RNA

hybrids. A reduction in the signal generated will indicate the specimen contains Chkamydia

trachomatis or Neisseria gonorrhea rRNA contingent on the assay used (Gen-Probe

Incorporated, 1994a). The PCA fimnctions as a confirmatory test for questionable results as

well as a percentage of all positive findings, an appropriate standard when misdiagnosis of

either sexually transmitted infection could lead to psychological anguish or legal

ramifications for the patient and/or their partner. Those specimens that originally tested as

high negative or low positives but on retesting tested positive or negative respectively, are

reported as 'indeterminate" accompanied by the recommendation that the clinician obtain

a follow-up specimen for re-testing if the patient has not already been treated for infection.

Test results are calculated based on the difference between the response in relative

light units (RLU) recorded from the specimen and the mean of the negative reference

readings (Gen-Probe Incorporated, 1994a). Before testing each day, equipment is re-

calibrated and cut-off ranges for RLU reading are set.

Related urogenital organisms may be present concomitantly with either chlamydia

or gonorrhea. Among these are Chlamydia psittaci, Ureplasma ureatlyticum, Gardenella

vaginalis, and Candida albicans. Analytical specificity of Gen-Probe indicates these

organisms do not cross-react with Chlamydia trachomatis or Neiserria gonorrheae

probes during testing (Gen-Probe Incorporated, 1994b).

Risk Factors for Low Birth Weight

Many variables have been reported in the volumes of literature on risk factors

associated with LBW. Investigators have studied the associations between birth outcomes

and malnutrition, smoking, reduced or absent social support, employment, violence, work,

stress, poverty, age and education, high and chronic stress, low-socioeconomic status,

utilization of and access to prenatal care, member of minority ethnic or racial group, drug

and alcohol use and infectious diseases. Much of this work has been focused on

identification of markers for pre-term birth or low birth weight. Less is known about the

role ofpsychosocial factors and the comparison of these variables for term low birth

weight and pre-term outcomes. Some key findings on a number of the potential

independent variables are summarized below.

Contribution of psycho-social and behavioral factors. Psychosocial stress was

examined prospectively by researchers to identify any associations with LBW in a low-

income urban population (Orr, James, Miller, & Barakat, 1996). The researchers used

logistic regression with low, moderate, and high stress dichotomous dependent variables

controlling for exposure to different stressors. The independent stressors included among

others were chronic financial or marital problems, death, divorce, housing, and

employment. Scores were not associated with demographic variables such as race, marital

status or educational level For all women, exposure to stressors was closely associated

with other clinical and behavioral risks for LBW. There were some different and some

similar associations with low birth weight for blacks and whites. Significant for black

women were smoking, hypertension, hospitalization during pregnancy, low pre-pregnancy

weight, prior pre-term birth, and exposure to stressors. Significant for white women were

smoking, drug use, hypertension, hospitalization during pregnancy, and prior pre-term


In contrast, another group of researchers reported finding no association between

psychological distress and birth weight for gestational age (Heddegaard, Henikson,

Sabroe, & Secher, 1996). This prospective population-based study collected measures of

psychological distress at the 16e and the 30e week of pregnancy by questionnaires among

Danish women with singleton pregnancies. Dunkel-Schetter (1998) reviewed numerous

studies conducted with colleagues to highlight their findings and possible mechanisms that

may contribute to interactive processes between pre-term delivery, anxiety, stress, and

stress hormones.

Copper and colleagues (1996) examined stress through measurement of anxiety,

self-esteem, mastery, depression, and stress. They reported that stress was associated with

PTL, SGA, and LBW after adjustment for maternal behavioral and demographic

characteristics. Among black women, this was even more significantly associated. Other

researchers identified maternal residence in public housing, poverty, and feelings of

helplessness with significant decreases in mean birth weight (Shiono et aL, 1997).

The role of maternal employment on LBW and gestation is also conflicting.

However two studies conducted with women in other countries suggest that intrauterine

growth restriction and PTL may be affected by moderate to heavy physical work effort

(Spinillo et at, 1996; Launer, Viar, Kestler, & Onis, 1990).

Smoking during pregnancy has been found to be significantly associated with

LBW, TLBW, and PTLBW. Increased risks ranges from OR 2.1 to OR 2.8 and appear to

be dose related (Cnattingius & Haglund, 1997; Olsen, 1992; Sexton & Hebel, 1984). One

Swedish study examined data for births from 1983 to 1992 and reported a 'true' decrease

in smoking during pregnancy and reduction in the attributable risk for SGA infants

(Cnattingius & Hagluhmd, 1997). The highest odds ratio, 2.8, was observed each year

among women smoking more than ten cigarettes per day.

Another area of study related to smoking and potential LBW is the differences

between race and ethnic groups and tobacco use. Among those women who quit smoking

sometime during their pregnancy, more women were of white race/ethnicity (12.7%)

compared to 4.3% of Hispanic race/ethnicity (Ruggiero & Groot, 1998). Among those

women identified as never having smoked the percent was inversely related 41.5% for

women of white race/ethnicity and 81.6% of Hispanic race/etmhnicity. Ahijevych and

Gillespie (1997) studied nicotine dependence among black and white women and reported

differences in plasma cotinine to cigarette ratios. Black women scored higher on plasma

cotinine levels; cotinine per cigarette ratio and carbon monoxide boost suggest ethnic

differences in nicotine metabolism. They also refer to information from other studies that

indicated black women sustained "greater lunhmg damage and less lung function recovery

following cessation than their white counterparts." These potential effects of smoking

among black women may potentiate the adverse impact on pregnancy and birth weight and

merit additional study to better understand the relationship of this risk factor.

Contribution of physiological and medical care factors. Weight gain has been

associated with fetal growth and consequent birth weight ranges (Abrams & Laros, 1986;

Seidman, Ever-Hadani, & Gale, 1989). Schieve, CogswelL and Scanlon (1999) examined

associations between weight gain per week of pregnancy and net weight gain per week of

pregnancy in a low-income urban population. Their findings suggest an association

between both low and high weight gain and PTL. They concluded that women with a

weekly weight gain at or near the Institute of Medicine guidelines for their respective body

mass index had the lowest risk of pre-term delivery. Abrams and Parker (1990) who

examined only term pregnancy outcomes reported that a wider range of maternal weight

gain than recommended in earlier published guidelines, was associated with good

outcomes. They also found no significance between maternal weight gain and SGA


Lumey (1998) examined the effects of under nutrition in pregnancy among infants

born during 1944-1946 in the Netherlands and hypothesized that potential biological

compensatory mechanisms increase placental growth in conditions of under nutrition. The

placental weight was compared to birth weight, and the average national maternal food

ration during the war years was used to estimate under nutrition. Among infants exposed

to the risk factor in the third trimester, mean birthweight decreased. No change in birth

weight was observed among infants exposed in the first trimester, but there was an

increase in placental weight and index to the birth weight. The author also references other

studies that have reported an increase in placental index in the presence of anemia and

maternal smoking. Another study that examined Women Infant and Children (WIC)

program enrollment identified a small but significant protective odds ratio for small for

gestational age the longer that a women was enrolled in the program (Ahluiwalia, Hogan,

Grummer-Strawn, Colville, & Peterson, 1998).

Hickey, Cliver, Goldenberg, McNeal, and Hoffman (1997) examined risk factors

that might contribute to low prenatal weight gain. They studied non-obese low-income

women and controlled for socio-demographic lifestyle and reproductive characteristics.

Three characteristics were associated with significant odds of low prenatal weight gain

among women of black race: a mistimed or unwanted pregnancy (OR 2.0), more than one

preschool child at home (OR 2.0), and not using her own car for errands (OR 2.1).

Among women of white race only, working more than 40 hours per week was associated

with low prenatal weight gain (OR 9.1).

Many different measures have appeared in the literature to define and assess the

adequacy of prenatal care. Adequacy of care or utilization is then examined along with

other variables for its association with LBW. Overall most of these measures recommend

initiation into care during the first trimester and then some pro-set number of visits with

consideration for length of gestation (Alexander & Kotelchuck, 1996). These measures

overall do not measure content, capture gaps in visits, or adequately consider high-risk

pregnancy visit schedules (Stringer, 1998). The inadequacy ofthe different indices to

accurately quantify adequacy is highlighted in the numerous studies that have compared

the different measures to the same sample population. Alexander and Kotelchuck applied

five measures to data files containing 169,082 singleton births. The proportion of cases

assigned to each utilization category (adequate, inadequate, etc.) ranged from 34% to

58% for adequate care, 9% to 20% for inadequate care, and 7% to 27% for intensive


The Kessner index was applied in a North Carolina study that compared infant

birth weight for women receiving their care at the health departments or from other

providers who accepted Medicaid (Buescher, Smith, Holliday, & Levine, 1987). Medicaid

women who received care from other providers were at twice the risk of having a LBW

infant compared to those who received their care at the health department. Augustyn and

Maiman (1994) utilized the 1988 Institute of Medicine indices to examine psychological

and sociological barriers to prenatal care from the reported literature. Another research

group examined the national changing pattern of prenatal care utilization with four of the

published indices (Kogan, Martin, Alexander, Kotelchuck, Ventura, & Frigoletto, 1998).

Markedly different trends were produced with the different indices. Virtually no change in

utilization for adequate or intensive care was reported using the IOM index, while an

increasing trend was noted for more adequate and intensive prenatal care utilization was

reported with the R-GINDEX and APNCU index. Differing patterns of utilization and

trimester of entry for by teenagers ranged from 2.9% to 16.3% and 12.6%respectively.

Marked differences were also noted for utilization with multiple gestations as reported by

the different indices ranging from a -13% to 23%.

Perloff and Jaffee (1997) compared two measures to examine the utilization of

prenatal care in New York City. The most interesting finding they reported was the

marked differences in sample characteristics that the two indices produced. The magnitude

of risk for inadequate care among blacks, teens, those women not completing high school,

and unmarried women is significantly increased with the use of the APNCUI indices.

Kogan, Alexander, Kotelchuck and Nagey (1994) looked at the content of prenatal

care and its association with LBW. They utilized the Kessner Index to measure adequacy

of prenatal care utilization and controlled for other risk factors such as age, education,

employment status, smoking, etc, and examined interaction terms in their logistic

regression models. Women in this study who did not report receiving all types of advice

recommended by the Expert Panel on the Content of Prenatal Care were more likely to

have a LBW infant (OR 1.38). The authors found no differences between women who

reported that they received all the recommended initial prenatal care procedures, and those

who reported not to have received all prenatal care procedures With the discrepancies

observed in the above reviews of the different indices the risk associated between LBW

and '"inadequate care" in this and other studies must be interpreted with caution.

Very young age appears to be associated in some studies with LBW and also late

initiation into prenatal care. Higher rates of LBW have also been observed for women

over forty. For adolescents aged 15-19 the rate of LBW in 1997 nationally was 13.6%,

and for women over forty, the rate was 10%, while the overall rate was 7.5% (Ventura,

Martin, Curtin, & Mathews, 1998). Among all women nationally, the percent that entered

prenatal care in the first trimester was 82.5% while only 68.1% ofwomen aged 15-19

entered care in the first trimester.

Adolescents are two times as likely to deliver a LBW infant than are adults

(Ventura, Martin, Curtin, & Mathews, 1998). According to Helerstedt, Pirie and

Alexander (1995) parity of the adolescent may also contribute to this observed difference.

They reported a difference in LBW rates between primapara (6.5%) and multipara (7.6%)

adolescent mothers in their study of adolescent parity and infant mortality. Roth and

colleagues (1998) reviewed the literature on young maternal age and the incidence of

LBW infants. They noted that the published studies have examined this association from

numerous perspectives. One association is young gynecological age, and possible

restricted blood supply to the cervix, that in turn may predispose the young female to

infections that contribute to preterm delivery. A second theory put forth is nutritional

competition for nutrients between the mother and developing fetus. A third theory

suggests that combined psychosocial behaviors and conditions like concealment, and

accompanying reduced food intake, delayed entry into prenatal care and poverty are the

real risk factors, not age. The authors also cite work done by Geronimus to explore the

role of race in adolescent LBW rates and noted that the ratio of black LBW to white LBW

for 15-19 year olds was 1.8. Geronimus (as cited in Roth, 1998) theorized that it is a

lifetime of exposure to racial stress that contributes to the development of "hypetension, a

precipitating factor" in pre-term labor and delivery, resulting in LBW.

Swedish researchers examined the births of younger and older women and

reported that birth to young women was a social problem, not associated with LBW. The

adolescent rate of LBW at 5.4% was better than the overall for women between the ages

of 35 and 49 whose rates were from 9.6% to 8.9% (Hemminki & Gissler, 1996). Their

findings in a different population are consistent with much of the epidemiologic data in this

country and suggest that the resulting social stigmata and subsequent increased risk for

lifetime poverty are greater risk factors for the adolescent compared to the older women

for whom the low birth weight event is a greater risk. Lee and colleagues (1988) examined

birth records from Illinois for the years between 1980 and 1984. After controlling for race,

education, parity, prenatal care, and marital status they concluded that the adjusted risk for

low birth weight at term is lowest among teens and increases with advancing maternal age.

Childbearing at an older age in this country is associated with higher rates of low

birth weight. For 1996, 8.1% ofbirths to women aged 35-39 were less than 2,500 grams,

9.5% to women aged 40-44 years, and 14.9% to women 45-49 years of age (Ventura,

Martin, Curtin, & Mathews, 1998). The rate for women of black race/ethnicity is even

more pronounced at advanced childbearing ages: 16% for women 35-39 years, 18.4% for

women aged 40-44, and 18.2% for those 45-49 years. Lee (1988, as cited by Committee

on Unintended Pregnancy, 1995) suggests that this association may be linked to biologic

aging of maternal tissues and systems, or the accumulative effects of diseases e.g.,

hypertension, diabetes.

The epidemiology ofLBW and the literature suggests that black race/ethnicity is

disparately associated with LBW, as well as PTLBW and PROM. Vrji and Cottington

(1991) observed that black women experienced a significant risk for PTL, (adjusted OR

1.56). Others reported an odds ratio of 2.1 for PTL among both black and HiEspanic

women (Berkowitz, Blackmore-Prince, Lapinski, & Savitz, 1998). Collins and David

(1990) examined race and the differential effect of income, education, marital status, and

age on LBW. The risk of LBW remained twice that of white women across all age groups,

education, or income strata.

Others have reported that black race/ethnicity is not a risk for LBW when the

mother is foreign born and in fact is protective when compared with black United States

born counterparts. Cabral and colleagues (1990) observed that foreign-bomrn black women

were more likely to be older, married, better educated, have better pre-pregnancy weight

for height ratios and adequate prenatal care. Their reported adjusted odds ratio for having

a low birth weight infant was 0.81; however, the confidence interval included 1.0 negating

the association. Others also suggested that there is a protective effect from foreign birth

status with an odds ratio of 0.88 and 0.77 among Caribbean and African-born black

women, with the confidence intervals all below 1.0 (Fang, Madhavan, & Alderman, 1999).

Collins and David (1993) examined biracial infants to determine the role of black

to white disparity in birthweight. When all variables were entered into logistic regression

to control for income, education, marital status, etc, the adjusted odds ratio of LBW for

biracial infants born to black mothers and white fathers was 1.4 compared to the biracial

infants born to white mothers and black fathers. Biracial infants born to black women also

had an increased likelihood of prematurity and SGA, OR 1.6 and 1.7 respectively.

Rural residence has been suggested by some as an adverse situation for access into

prenatal care and needed services, with subsequent poor outcomes. Researchers in the

northwest examined linked birth records and hospital discharge abstracts stratified by

insurance source to examine the effects of poor local access on the risk of having a 'non-

normal' infant (Nesbitt, Larson, Rosenblatt, & Hart, 1997). They reported women with

poor access expereinced an increased risk of delivering a non-normal neonate (LBW,

PTLBW and increased associated costs). They also noted that women with private

insurance were more likely to have higher costs and longer stays overall

In contrast, Larson, Hart and Rosenblatt (1997) found no association with

residence in a non-metropolitan area for LBW or VLBW. They did observe an adverse

effect for neonatal mortality and post-neonatal mortality. However, on logistic regression

non-metropolitan residence was not associated with either LBW or neonatal mortality.

Alexy, Nichols, Heverly, and Garzon (1997) reported that rural or urban residence did not

predict LBW. They found race, weight gain, number of total prenatal care visits, and

adequacy of diet resulted in stronger associations to predict LBW. 0' Campo and

colleagues (1997) conducted muti-level modeling to examine macro (census tract) and

individual risk factors. They noted all individual level risk factors for LBW had a different

effect dependent on the neighborhood of residence. They suggested that housing, crime

and unemployment modify the relationship of urban or rural residence to LBW.

Researchers examined the inter-pregnancy intervals (IPI) of small for gestational

age (SGA) and pre-term births among a North Carolina population of blacks and whites

(Shultz, Amrndt, Olshan, Martin, & Royce, 1998). Three categories of IPI were created: 0-

3 months, 4-12 months, and 13-24 months. Those with IPI of greater than 24 months

were excluded, due to the possibility of sub-fecundity and potential increased risk for low

birth weight. Race specific logistic models were used and population attributable risks

calculated. Evaluation of single month intervals indicated that the odds ratio for SGA

infants was elevated for each until 10-12 months, and decreased with increasing IPI. There

was no consistent pattern for pre-term birth associations using the single month

incremental analysis. Overall they found a moderate association (OR 1.6) between SGA

and IPI of 0-3 months. No significant association was observed for the IPI of 4 to 24


Prior LBW events and prior pre-term deliveries have been reported as significantly

associated with an increased risk for subsequent LBW and PTL events. The birth record

and the Healthy Start prenatal screen both contain fields to capture information related to

prior poor birth outcomes. Hulsey and others (1998) reported a 2.8 times increased risk

for a subsequent pregnancy with PTL, following an initial pre-term delivery. The authors

calculated a population attributable risk that attributes 22.5% of second pre-term

deliveries to having had one previously. Black women experienced an overall greater rate

of pro-term deliveries, but if a white woman delivered prematurely on the first pregnancy

she was at a 4.5 times increased risk compared to black women with 2.5 increased risk.

Another risk for poor outcomes may be the role of the father. One study, reported

that among women who changed partners after their first delivery before 34 weeks, a 33%

reduction in the risk of a subsequent early pre-term delivery was observed compared with

those who did not change partners (Li, 1999). Among women who had initially delivered

at gestation over 36 weeks, changing partners increased their risks of a subsequent pre-

term by 16%. Among women between 34 and 36 weeks no effect was observed with the

changing of partners. The weight of the father may also contribute to increased risk of low

birth weight delivery for his partner. Klebanoffand others (1998) studied the offspring of

an historic Danish cohort combining data from birth registries, military records and

midwifery records. They found a significant association between paternal birth weight and

the subsequent birth weight of their offspring, independent of the maternal birth weight.

Other researchers have demonstrated an association between a woman's birth weight and

the development ofpre-eclampsia during their own pregnancies as teenagers or young

adults (limnnes, Marshall, Byers, & Calonge, 1999).

Contribution of sexually transmitted infections. Dunkel-Schetter (1998) has

suggested the possibility that stress increases risky sexual behavior during pregnancy, with

reduced prenatal care utilization for the detection of infection. However, she offers no

evidence to support this concept. Other literature reports ample associations between intra

and inter-uterine infection, low birth weight, spontaneous pre-term labor/delivery and the

more commonly known sexually transmitted infections like gonorrhea, syphilis, herpes,

trichomonas, and more recently, chlamydia and human immunodeficiency virus. Less

commonly known are other diseases often associated with sexually active women, but

whose modes of transmission are even less well understood than the historical STDs.

These include group B Streptococcus, hepatitis B and C, cytomegalovirus, and the many

organisms associated with bacterial vaginosis. The more well known adverse pregnancy

events associated with STDs include stillbirth, perinatal death, and mental retardation,

attributable to early syphilis, herpes, cystomegalovirus, and group B Streptococcus,

(Goldenberg, Andrews, Yuan, Mackay, & St, Louis, 1999).

Eschenbach (1998) suggests that any amniotic infection is a fetal infection, and

consequently may have a role in causality of pre-term delivery before 28 weeks. Other

recent studies with chlamydia that have examined pathogenesis of the fetal, and maternal

tissues would support such a hypothesis and deserve more consideration.

Not much research has been conducted to examine the direct association between

low birth weight and each sexually transmitted infection while controlling for other

confounding variables. Researchers have prospectively evaluated the role of Trichomonas

vaginalis in low birth weight in a large multi-center study among an ethnically diverse

population (Cotch et al., 1991, 1997). They reported at a 95% confidence interval that

pregnant women infected with the protozoan were significantly more likely to deliver a

low birth weight infant, and to have a pre-term birth weight infant (OR 1.3, and 1.4, p <

.01). Also, infection with Trichomonas vaginalis accounted for a disproportionate share

of low birth weight deliveries among blacks as compared to whites or HIispanics in this

study. Earlier studies found no association between the pathogen and low birth weight

(Mason & Brown, 1980; Ross & Middlekoop, 1983, as cited in Wolner-Hanssen, 1999).

Trichomonas vaginalis is more likely to be found in the squamous epithelium, but

evidence has been reported of infection in desquamated cells from human amniotic

membranes (Wolner-Hanssen, 1999).

Syphilis was the earliest recognized maternal sexually transmitted infection to be

associated with adverse outcomes. One study published in 1917 and another in 1951

identified a transmission rate during pregnancy of 60% to 70% (Schultz, Schulte, &

Berman, 1992). While sexual transmission is believed to cease after four years of infection,

the possibility of transmission of the spirochete to the fetus by blood-borne transplacental

infection is more enduring as demonstrated with a 25% risk of fetal infection during early

and 12% during late syphilis (Ingraham, 1951, as cited in Radolf, Sanchez, Schulz, &

Murphy, 1999). Many studies have identified the strong association between stillbirth and

spontaneous abortionthe most common sequelae, that occurs most often in the 2wd and

early 3"' trimesters (Radolf Sanchez, Schulz, & Murphy, 1999). No literature is available

that suggest an association between syphilis and low birth weight while controlling for

other confounding variables.

Herpes simplex virus is not associated with low birth weight, and generally is

believed to be transmitted intra-partum (Stagno & Whitley, 1999). It is associated with

pre-term labor and spontaneous abortion and both primary and recurrent infection can

cause fetal infection in utero. The sequelae that can accompany infection range from skin

vesicles, blindness, disseminated infection, and long term neurological impairment and

most often are lethal The authors ofthe NIAID Collaborative Antiviral Study (as cited in

Stagno & Whitley, 1999) reported in their data that 33% of those with disseminated

infection, 23% of those with central nervous system infection, and 24% of those with skin,

eye or mouth infection were delivered before 36 weeks.

Intra pregnancy infection with Neisseria gonorrhoeae has been studied by several

groups as reported in Gutman (1999) and Morse and Beck-Saque (1999). The rates for

premature delivery were between 13% and 67%; however, no mention is made in CGutman

of the infants' birth weights or association with term low birth weight, or small for

gestational age infants. Other adverse outcomes reported, include spontaneous abortion,

perinatal death, perinatal distress, stillbirth, chorioamnionitis, premature rupture of the

membranes, skin and joint lesions, and meningoencephalitis (Watts & Brunhamni, 1999).

The reported association of gonorrhea with pre-maturity has been conflicting. Amstey

(1976, as cited in Gutman, 1999) reported the rates of poor outcomes in his study were

the same, regardless of treatment during pregnancy. Amstey's high rates ofpre-maturity

were not found by Charles et al (1970 as cited in Watts & Brunham, 1999). One serious

sequelae is an association between pregnancy and disseminated gonoccocal infection, a

condition that usually requires hospitalization to adequately treat.

Bacterial vaginosis has been identified in a significant number of studies as strongly

associated with pre-term low birth weight. The Vaginal Infections and Prematurity Study

Group reported an odds ratio of 1.4 for bacteria vaginosis and preterm delivery of a low

birth weight infant (Hiflier et al, 1995). The Patient Outcomes Research Team (1998)

reported that bacterial vaginosis in women of black race/ethnicity accounts for 40% of

excess pre-term births. They also demonstrated that antibiotic treatment for bacterial

vaginosis in pregnancy reduced pre-term deliveries. Eschenbach (1999) and Hilier and

Holmes (1999) provide comprehensive reviews of bacterial vaginosis. The authors note

the strong associations between this infection and adverse pregnancy outcomes, and the

positive impact (albeit conflicting impact) that treatment during pregnancy has in reducing

incidence of prematurity. Eschenbach also provides a discussion on the potential

association between fetal and maternal immune response in the presence of infection with

bacterial vaginosis.

The actual impact on low birth weight from bacterial vaginosis is probably

underestimated since this condition is roughly twice as common as other reportable STDs

and also asymptomatic. However, a recently completed national muti-centered study

failed to provide the needed evidence that treatment for bacterial vaginosis during

pregnancy reduces the likelihood ofprematurity or low birth weight (Hifflier, 1998). Two

other areas of concern with bacterial vaginosis (BV) are the non-sexual acquisition of the

syndrome and the role of BV in HIV acquisition. Black women are three times more likely

to have BV regardless of the number of sexual partners, and in the absence of sexual

partners. In contrast white women are more likely to develop or acquire the condition with

increasing number of partners (Hilier, 1998). Numerous studies suggest that BV enhance


the acquisition of HIV infection during and after pregnancy with adjusted odds ratios of

1.5 to 3.7 (Hillier, 1999).

Nair and others (1993) reported that HIV vertical transmission was increased with

the presence of clinical chorioaminionitis, any sexually transmitted infection during

pregnancy, and associated with LBW (p < .05). Those infants born before 35 weeks were

significantly smaller than infants non-infected with HIV.


In this chapter the research design, protection of human subjects, confidentiality,

data sources, development of the relational database, methodological issues in the use of

administrative databases, definition of study variables, data analysis, limitations, and

assumptions are addressed.

Research Design

This was a retrospective epidemiological study of a population-based sample of

pregnant women and adolescents who initiated prenatal care through county health

departments. A relational study database was constructed from linked data files. The files

were extracted from numerous administrative databases. The records of infants born

during 1996 and their mothers were linked to laboratory tests, disease reports, and

prenatal screening information through the use of probabilistic matching algorithms.

Descriptive analysis of al files and the extracted study file was conducted. Logistic

regression analysis was conducted to further explore the association between infection

with Chlanydia trachomatis and low birth weight. The data sources, process of matching

records, and analysis that was conducted are described in detail below.

Protection of Human Subjects

"Thiswas a retrospective epidemnriologic analysis of existing data. The University of

Florida, Health Center Institutional Review Board, and the Florida Department of Health,

Review Council for Human Subjects, both approved the study protocol The request to

waive documentation of informed consent was approved by both groups. No direct

contact was made with subjects. No treatment or care was provided or altered for the

subjects included in this study. Since this study retrospectively examined existing

information, no individual informed consents were deemed necessary by either review

body. The reporting ofbirths and fetal deaths and ofpositive tests diagnostic of the

sexually transmitted infections studied and morbidity information is required by law;

Chapters 382 and 384, Florida Statutes. The original consents to examine the data sets as

outlined in the study protocol were obtained from the respective administrative data

managers charged with the confidential maintenance of each data systems that contained

the morbidity or vital records. Oversight review for the study was provided by Department

of Health Review Council for Human Subjects.


During the entire duration of the research period, the confidentiality of all data

systems was maintained in compliance with confidential security management protocols of

the Florida Department of Health. The study data set was stored on a server with access

limited to the investigator and four other departmental staff who assisted with the data

extraction process. The final study data set constructed through the matching process was

password protected and accessible only to the investigator. The researcher was assisted by

departmental staffto extract existing data from multiple data systems maintained for

tracking of client clinic services, recording of client laboratory services, and billing. The

matching of data files was conducted by departmental staff in the routine course of

activities conducted to examine quality of care, evaluate health programs, and manage

data systems. All data systems variables were matched with identifiers of the subject

intact. All hard copy information and computer files were stored for the duration of the

study according to protocol During analysis, the accumulated data were examined,

aggregated, and formatted into a structure useful for analysis without the use of client


Data Sources

Variables for the study data set were extracted from six administrative databases.

These databases were 1) 1996 birth records; 2) 1996 fetal death records; 3) Healthy Start

prenatal screen; 4) Maternal and infant laboratory test results; 5) Maternal and infant case

morbidity; and 6) Congenital syphilis records. Each are described below and information is

provided about the software, hardware, and flow of the data from point of collection to

final repose in the data system from which it was extracted for this study.

Birth and fetal death records. The birth and fetal death records were the first two

databases from which study data was extracted. The first statewide vital statistics law for

Florida was enacted in 1899, establishing a system for physicians to report births and

deaths (Porter, 1901). On January 1, 1917 the first comprehensive registration system

based on the national "Model Vital Statistics Act" became effective in Florida, after earlier

passage in 1915 by the State Legislature (Feamrside, 1916). The base study Sfie, the birth

record file, is an extraction from this vital statistics database. The fetal death records were

also extracted from the vital statistics database. Some of the information regarding birth

and death records is similar, and some aspects are dissimilar. The following discussion will

note where they are similar and where they are different.

The registration of births and fetal deaths is both a local and state function of 67

local registration districts corresponding to the jurisdictional area of the county health

department. (Office of Vital Statistics, 1996b). The county health department director or

administrator serves as local registrar for that county. It is his/her responsibility to oversee

the timely and complete registration of births and deaths occurring in their district. Once a

birth or death event is accepted for registration, the local registrar maintains a copy of the

vital record and forwards the original to the state office.

The information contained in the vital birth and death record is obtained from the

following sources: mother, father, relatives or persons who have knowledge of the facts,

physicians, midwives, fimeral directors, and hospital records. A standard form is used to

record and register the information required by the vital statistics law, Chapter 382,

Florida Statutes. The registration form in use during 1996 reflects the national model birth

record (Office of Vital Statistics, 1996b). The Florida birth registration form supports

collection of all recommended data from the model birth record, with the notable

exception of parental occupation. The hospital administrator (or if the birth is non-

institutional, the birth attendant) is required to file within five days after the birth a

complete and accurate birth certificate with the local registrar. The certificate must be

signed by at least one parent, attesting to the accuracy of the information, and either the

hospital administrator or designee, or birth attendant for non-institutional births. Chapter

64V-1 of the Florida Administrative Code provides clear direction on the preparation of

the certificate regarding paternity, maternity, birth attendance, the child's name and

surname, and residency. Local registrars provide regular training to hospital medical

records staff and the clerical staff responsible for completing the birth certificate (Office of

Vital Statistics, 1996b; R. Shepard, personal communications, June 30, 1999). The birth

record segments reflect the handbook directions and training provided to collect the

information from the numerous sources. The first segment has the legal demographic

information. This detail should be collected from the mother, father or other person with

"knowledge of the facts." The second segment contains the medical and health history

information. The medical history detail should be collected from the medical records, with

both delivery and prenatal supplied to the hospital (or birth attendant, or center). The

details about the condition of the neonate should be collected from the neonatal unit

records, e.g., abnormal conditions of the newborn, anesthetic complications, etc.

Clear directions and regularly conducted training sessions do not necessarily

translate into accurate completion of birth certificates (Sammet, personal communications,

June 30, 1999). During the data entry process at the state registrar's office, a staff member

telephones county health departments for clarification if they are unable to decipher data

on the birth certificate. Routine quality assurance reports are completed monthly to

identify outliers in data field. Habitual problems like late reporting, or high error rates in a

particular data fields are generally identified and addressed at the local registrar's level

though consultation with the hospital administrators. Persistent problems and unusual

outliers in the data are addressed by visits to the hospital from the local registrar or from

state registrar's quality assurance office staff (R Shepard, personal commmications, June


Florida Statute section 382.002(7) requires that fetal death certificates record "the

death of a product of conception prior to the complete expulsion or extraction from its

mother, if the twentieth week of gestation has been reached." The statute further provides

that every fetal death must be registered within five days after the delivery occurred, and

prior burial, disposition, or removal from the state (Office of Vital Statistics, 1996b). The

funimeral director, direct-disposer, or physician/midwife who attended the delivery and is

responsible for registering the record, completes the registration record. The certificate of

fetal death is to be signed by the physician in attendance. The midwife may sign as actually

having attended the delivery but not as to the cause of death. The hospital administrator is

responsible for providing the medical details to the funeral director in charge of the

burial/disposition arrangements. The data entry and quality assurance for the fetal death

registration is the same as for the birth certificate.

The data entry for both files is conducted on personal computers. The two files are

supported by proprietary software and maintained and stored on a mainframe computer

system housed in the Department of Children and Families. Hard copies of the data are

regularly microfilmed. The two files, the birth record and the fetal death record were

extracted in separate programs from their respective databases. Staff in the Department of

Heath then combined these two files after realignment of the fields to create one

continuous birth fetal death record file. In the routine course of operations at the Office of

Vital Statistics these records are maintained separately at all times. Quality assurance

measures are regularly employed to assure that infants initially reported as a birth are not

later reported also as a fetal death. Monitoring steps are designed to identify inappropriate

registration combinations of certificates for the individual live born infant or stillborn

infant. This combined data set provided key variables for the final study database, and are

presented in Appendix A. It also served as the base file against which all other files were

matched and linked to in the final relational study database.

Healthy Start prenatal screen database. The Healthy Start prenatal screen is the

third database from which study files were extracted. In 1991, the Florida Legislature

enacted the Healthy Start program with the creation of Chapter 383.2161 ofthe Florida

Statutes and development of Chapter 64C-7, Florida Administrative Code. Implemented in

April 1992, this program was modeled on prenatal case management programs in other

states like South Carolina and international studies that demonstrated the value of non-

medical interventions for women and infants at risk of low birth weight, infant death, and

developmental delay (Thompson, 1993; Florida Department of Health, 1997b). The core

components of Healthy Start include: 1) universal screening of all pregnant women and

newborns, 2) professional assessment of health, social and environmental risks, and 3)

targeted case management and risk appropriate care. Expanded Medicaid eligibility to

185% of the federal poverty level, and provider reimbursement to providers to administer

the prenatal screens further support the Healthy Start program goals of improved

pregnancy outcomes (Florida Department of Health and Rehabilitative Services, 1995).

First trimester screens are reimbursed at a higher rate in support of timely intervention

(Florida Department of Health, 1998).

The Healthy Start prenatal screen data set was extracted from an existing system

database housed in the Florida Department of Health, Office of Vital Statistics,

Jacksonville, Florida. The source document for this data set is the Healthy Start prenatal

screen. All prenatal care providers are required by law to administer the screening

instrument to pregnant women, preferably at their initial prenatal visit. The screening

instrument collects demographic information and responses to a series of "risk" questions.

The questions are based on medical, psychosocial, and environmental factors associated

with increased risk of poor pregnancy outcomes. A score is calculated based on the risk

factor, and the strength of the risk as a predictor of poor outcomes (Serow, Jones, &

Luke, 1996). In addition to the demographic, medical and provider information collected,

the psychosocial and environmental variables capture risk information about educational

level, access to care, housing, and food, drug, alcohol and tobacco use, risk of domestic

violence, and perceived level of personal stress.

The particular "risk" factors included were initially based on extensive review of

the literature and an evaluation of birth and fetal death records for associations commonly

observed in the Florida population among low birth weight infants and infant deaths. Two

different years of birth records were evaluated in development of the prenatal screening

instrument utilized since 1994. The 1989 birth and fetal death records were used to

calculate risk ratios on commonly held risk associations, e.g., age less than 18 years, age

over 39 years, unmarried, etc. If a factor occurred in more than 1,000 births and had an

associated risk ratio of greater than twice the average risk, then it was considered a

potential "risk factor" and included in the initial screening instrument (Thompson,

Hopkins, & Watkins, 1993).

After the first two years of use, a second evaluation utilized the prenatal screens

and the birth records for the period between April 1, 1992 and April 30, 1993 (Thompson,

Hopkins, & Watdkins, 1993). The adverse pregnancy outcomes were birthweight under

2,000 grams and/or birth date of less than 34 weeks from the date of last menses. The

same adverse outcomes were used to develop the initial screening instrument after

examination of neonatal and postnatal death rates from 1991. It was noted that Florida

death rates were high at birth weights below 2,000, and dropped off above this cut-off

point as compared to the more universally used 2,500 grams. Likewise the death rates

level off above 34 weeks between last menses and birth date. The positive predictive value

of each screening tool risk factor was analyzed from the linked birth records and prenatal

screens. Additionally, besides effectiveness of the screening tool to identify women at risk

of an adverse outcome as defined in the analysis, the related workload to the county health

department was considered. At the recommendation of the Healthy Start Advisory

Committee, a few subtle changes were subsequently made to the scoring weights and risk

factors in the revised 1994 form to achieve a predicted positive screening rate of 39.98%

and sensitivity of 60.64%.

One copy of the screening instrument is forwarded to the Office of Vital Statistics

for data entry. During 1995 and 1996, the data entry system was constructed to require

that all fields were "must enter fields." This data entry system was envisioned as

appropriate to aid evaluation of the instrument and the program in the early phase of

legislative implementation. Any record that was found to have an incomplete data field

was placed in a query status and returned to the local county health department. It was

then the responsibility of the local authorities to contact the appropriate public or private

prenatal care provider and ensure completion of the missing data (K. Freeman, personal

communication, June 30, 1999). During 1997 after the preliminary evaluation, some fields

were changed to reduce the restriction on the data entry process. This change resulted in

higher rates of query flags on many variables.

The primary use of the information collected on the prenatal screening instrument

is to calculate a risk score and provide the clinician with a platform to seek consent from

the woman to assess her need for targeted services. As with other screening instruments,

the intent is a rapid reliable assessment of an at-risk status. The data entry process was not

envisioned or fimded to support a standard consistent with research protocols. During the

data entry process verification of the data fields is restricted to four fields: name, social

security number, score, and date of the screening event (J. Ballard, personal

communication, June 30, 1999). However the Healthy Start coalitions are charged with

the responsibility to designate an agency that will provide training to providers on correct

use of the form, model successful solicitation of the screening process, and identify

patterns of positive risk within the community in order to target resources. Additionally,

the county health departments are charged with the responsibility to monitor the screening

instruments for completeness and coordinate the collection of missing data on queried

forms returned from the Office of Vital Statistics. This data set provided key variables for

the final study database, as are presented in Appendix A.

Maternal and infant laboratory test report database. The maternal and infant

laboratory test report database is the fourth source from which study files were extracted.

The laboratory Gen-Probe PACE2C test report data set was extracted from an existing

system database housed in the Department of Health, Bureau of Laboratories,

Jacksonville, Florida. This system was developed to support reporting compliance with

state and federal laws and regulations regarding biological specimens submitted for

testing, and timely billing of providers. Following is a review of the system specifications

and the flow of data into that system, as regards specimens submitted for testing with the

Gen-Probe PACE2C* nucleic acid hybridization technology (S. Crowe, personal

communication, June 29, 1999).

Beginning in December 1995, the Florida Department of Health, Bureau of

Laboratories commenced implementation of a long-range plan to computerize and store all

test-related data in an electronic system. This represented a transfer from a hardcopy filing

system based on specimen ascension numbers to an electronic data system, allowing recall

of a particular test results) from numerous demographic or biological variables, and the

ascension number. Additionally, the database was structured to support centralized biing

to providers, regardless of the branch laboratory at which the specimen was processed.

Each laboratory was brought onto the new system sequentially, beginning with the main

branch site that processed the highest volume of specimens. By late-year 1996, all branch

laboratory computer systems were functional, as was the central statewide database.

During this same period many test reports from the branch laboratories actually were key

punched into the system at the central office site, to facilitate the billing process.

The system software written in Massachusetts Utility Multi Programming System

(MUMPS) language is a proprietary product. During 1996, it is run on a Unix platform

operating system through centralized servers linking the five branch laboratories. Data are

entered from personal computers located in each respective area of the laboratory. In the

routine course of laboratory operations specimens are received in the delivery/mailroom.

At this initial point specimen labeling is matched against test requisition slip data for

accuracy, completeness in accordance with state and federal laws, and regulations

regarding biological specimens. A numbering machine is then used to stamp the requisition

slips sequentially with coded numbers reflective of the branch laboratory, area of the

laboratory in which the test will be conducted, and by specimen ascension. For example

"JSGxxxxxxxx" is Jacksonville Serology Gen-Probe followed by the accession number.

Preprinted labels are simultaneously numbered with the sam accession number by a

different machine. These preprinted labels are attached to the specimen vial or container

for tracking within the computerized system and through laboratory processing.

Specimens then are transported to the respective area of the laboratory to begin testing.

On arrival into the respective laboratory areas, laboratory technicians enter the ascension

numbers to create the test record in the computer system, along with the date the

specimen was received and the name of the test requisitioned. Additionally, the specimens

are noted for compliance with acceptable collection to testing time frames according to the

individual test type requested. During the period that specimen testing is in progress data

entry operators begin to enter the demographic information into the computer record

created for each specimen. Variables entered at this stage included: last name, first name,

social security number, address, insurance numbers, date of birth, race/ethnicity, sex, date

of specimen collection, provider code, program code, county where specimen originated,

provider address and name, the patient diagnosis, the physiologic source of the specimen

and the laboratory site identifiers.

On completion of the testing process the laboratory technicians check 100% of the

demographics on the specimen labels against the data that has been entered into the

computer tracking system prior to reporting out the test results on each specimen. This

process is aided by a demographic quality assurance report printed out from the system

prior to entry of the test results. During the course of testing the test results are verified

and any applicable qualifiers or specimen rejections are entered into the system, along with

the names of those reporting the results. Data error can and does occur at the initial phase

of specimen and demographic registration into laboratory computer systems. Dauly quality

assurance activities monitor data entry of demographic information and the reporting of

test result&

During 1998, an unfortunate series of events occurred that caused corruption of

segments of the database affecting calendar years 1996 and 1997. While much ofthe

primary database was subsequently repaired from archival tapes, not all of 1996 or 1997

test reports could be restored. Hard copy requisition slips are maintained in archival files

for the seven-year time period as required for medical records by law, Chapter

483.051(7Xf), Florida Statutes. The extent and the scope of the damage to the electronic

operating system and hardware was not identified until many months after this study was

conceptualized, the protocols written and approved, and countless hours expended on

matching of records to the base study file had commenced. Hence it was impractical at

that point to consider re-framing the study around a different birth cohort. Unfortunately,

these records are stored according to specimen ascension number. The estimated cost for

retrieval of missing test results for subsequent re-entry into the data system would serve

no laboratory purpose and was cost prohibitive to this investigator in terms of the study

budget. There is no reason to believe that the individual test reports contained in the Gen-

Probe test report data set are any more or less complete in any given time segment for the

period included in this study. There is reason, however, to believe that the distribution of

the test reports is constrained by both the events associated with the implementation ofthe

computer system and the described corruption event that occurred. The skewed

distribution is discussed below in the results chapter. However, one of the strengths of this

data set is the representiveness of specimens collected from women receiving prenatal care

in county health departments.

In the initial step, an extraction using MUMPS was made from the primary

laboratory data system to select data fields with information pertaining to Gen-Probe

PACE2C (S. Shiver, personal communication, June 30, 1999). This extraction was done

through the use of programming with Visual Basic to create a database flat file. The files

were then transferred onto the Sequel Server subdirectory. (The NT platform operating

system replaced the Unix system in late 1998 following the discovery of the system

failure.) The files were then accessible through the closed-frame relay departmental wide-

area network to the Bureau of STD staff with appropriate authorization and passwords.

At the next step FOX PRO programming software was used to construct an ASCI flat file

extracting only the desired study variables. In this step some variable fields were made to

conform to numeric codes consistent with the base study file, the 1996 Birth Fetal Death

Records, (race, ethnicity, county), and the Year 2000 date format (YYYYMMDD).

Extraction parameters included: all Gen-Probe PACE2C reports with collection date

between March 1995 through June 1997, for females regardless of date of birth, and

reports for males with a date of birth during 1996. The following variables were extracted

to support the matching process and for subsequent descriptive analysis: last name, first

name, social security number, date of birth, race, sex, address, county, specimen number,

date of specimen collection, test type, and results for both Chamydia trachmatis and

Neisseria gonorrheae. The extraction totaled 214,121 records from the primary laboratory

Gen-Probe PACE2C database that met the time, test, and gender study parameters& Each

woman and infant had two records, one for chlamydia results and a second for gonorrhea

results. This data set provided key variables for the final study database, as are presented

in Appendix A

Maternal and infant case morbidity database. The maternal and infant case

morbidity database is the fifth source from which study files were extracted. Sporadic

reporting of "venereal diseases" by providers to the authorities dates to 1918. By 1920 the

numbers of venereal diseases reported exceeded the total of the next three causes of illness

(Hardy & Pychon, 1964). In 1921 the Bureau of Venereal Diseases faded into the Bureau

of Communicable Diseases and it was not until 1938 that finds were appropriated to

support reporting again. Prior to World War II (in 1944) Florida saw the establishment of

a central registry for syphilis (Brink, 1944). That same year the Legislature enacted

prenatal and premarital serology for syphilis. The present centralized computer dates to

1988 and contains records of 458,771 case reports for syphilis, chlamydia, gonorrhea,

chancroid, lymphogranuloma venereum, and since 1997, human immunodeficiency virus

(K. Kampert, personal communications, June 30,1999). The maternal and infant case

morbidity data set was extracted from this existing STD Case Morbidity Reporting

database housed in the Florida Department of Health, Bureau of STD Prevention and

Control, Tallahassee, Florida.

The details of a STD case report travel through numerous hands before actually

entering the system, defined as a "case." At the onset a suitable specimen suspicious of a

STD must be collected and forwarded to a laboratory for testing. Laws and regulations

require that laboratories and physicians both must report all positive tests diagnosing a

reportable sexually transmitted disease per Chapter 384.25, Florida Statutes and Chapter

64D-3, Florida Administrative Code. These reports are made available to the county

health department. During 1995 and 1996 the information from positive laboratory reports

and physician diagnoses was entered into the morbidity system and then field records were

produced to support the activities of disease investigator staff to verify the case details

with the physician or other provider. During this same time period, verification of

treatment was implemented for gonorrhea and chlamydia. Due to the enormous volume of

chlamydia and gonorrhea cases and staffing constraints, this was not previously a

requirement. During this time some counties also began to verify the case data, patient

treatment, and partners) treatment for pregnant women, if the case report indicated the

pregnancy status. All opthalmia neonatorum and neonatal pnuemonia cases were

individually verified for the demographic, treatment, diagnosis and provider data during

this period of time.

Routine quality assurance reports were run at the local district reporting level

These were used to verify that all positive cases had an age, sex, and race identified (P.

Moncreif personal communication, June 30, 1999). These morbidity reports were in turn

then reported to the Bureau of STD Prevention and Control (known as the Office of

AIDS/STD/TB at that time). At the state level, scheduled quality assurance reports are run

to identify outliers, any duplicate records, and contradictory diagnosis code& An example

of an outlier that would be a red flag: a greater number of positive laboratory test reports

from an area that has reported a disparately smaller number of cases. It would be more

usual to see the reverse pattern identified. Annually, morbidity line lists were produced and

then forwarded to the district STD offices for verification of case numbers, age, race,

name, and diagnosis. Generally, the age and sex fields have a 98% completion rate, with

race ranging from 80-85%. This difference is the result of absent race/ethnicity

information from private providers. Each of the quality assurance components is

applicable to the discrete study data set time period. This data set provided key variables

for the final study database, and these are presented in Appendix A.

Congenital syphilis database. The congenital syphilis records are the sixth and final

database from which study files were extracted. This database is managed much as the

maternal and infant morbidity database above. However it was separate proprietary system

created circa 1990 and not linked electronically to the maternal or infant case morbidity in

the Sexually Transmitted Disease Management Information System (STD*MIS). The

movement of information into this system differs from that of the morbidity system in one

distinct way. All data entry for the 1996-birth cohort of congenital syphilis cases was done

in the Bureau of STD, unlike the maternal and infant case morbidity that was entered at

the local STD area offices. Additionally the data on each case was initially submitted on

lengthy case worksheets. Bureau of STD personnel then confirmed the details of the

mother's syphilis laboratory test results, her diagnosis, and treatment history. With this

additional information the case determination was made as to whether or not the

submitted congenital syphilis case report was indeed either a true clinical and laboratory

confirmed congenital syphilis case, or a case that met the national surveillance definition

for '"probable" congenital syphilis (CDC, 1998a). The case determination for congenital

syphilis applied to both live infants born to untreated or inadequately treated infected

women and to syphilitic stillbirths. A positive feature of this database was the close

scrutiny that each case received during the determination process. A limitation of this

database with its stand-alone design was the failure to directly link by computer the infant

case congenital records to the maternal case history, laboratory tests, or treatment records.

This weakness has been resolved in the 1998 STD*MIS system version with the

development of an internal congenital syphilis module.

Methodological issues in the use of administrative databases

Teutsch and Churchill (1994) make recommendations regarding the management

of databases to maintain the integrity and completeness. Among the issues that must be

addressed are the credentials of data entry staff updating of records, back-up of computer

files, automatic data entry fields, e.g., today's date or today's age calculated from date of

birth. Controlled system parameters and "must enter" fields are useful to ensure accurate

spelling of common variables, e.g., county names, and validate data entry for

completeness. No documented independent validation studies have been conducted on the

data contained in any of these data systems and entered during the period spanning late

1995 to the middle of 1997, the timeframes for data-sets included in the study. However,

routine production of quality assurance reports facilitate the identification and frequency

of common errors, identify the appropriateness of the range of values entered, and assess

the completeness of demographic data entry from each of the systems from which these

data set were extracted.

Other published guidelines provide suggestions for the evaluation of surveillance

systems like the STD maternal and infant case morbidity and congenital systems used in

this study. Evaluation of a surveillance systems should include 1) description of the

importance of the health event under surveillance, 2) description of the system, 3)

characterization of the usefulness of the data for decision making processes in public

health, 4) evaluation of the actual system for attributes such as simplicity, sensitivity, and

representativeness, 5) direct costs, and 6) recommendations (Klaucke et. aL, 1998; WHO,

1997). No documented independent evaluation reflective of these guidelines has been

conducted on the data contained in any of these surveillance systems.

Reliability of the information in each data sets utilized from which to extract study

variables is an issue common to all of the systems. Herrmmm (1985) compared self-

administered to interviewer-administered questionnaires in a case controlled study for

agreement and reported that consistently higher agreement levels for medical history with

interviewer-administered questionnaires was observed. Harlow and Linet (1989)

conducted a literature review of studies comparing data from questionnaires with

information derived from medical records. Significantly high accuracy over extended

periods of time was noted for each woman's recall of her pregnancy histories, childbirth

experiences, and events. Less reliability was observed for menarche, menstruation, and

menopause timeframes. All but one of the study data sets utilized interviewer-administered

questionnaires and medical extraction followed by data entry into the data systems from

hard copy records The one exception, Healthy Start prenatal screen, is often completed

by the pregnant woman, but may also frequently be interviewer-administered.

Definition of Study and Indicator Variables

The study variables used by this investigator were obtained from different data sets

within the linked relational database. Dependent variables and independent indicator

variables were created to support the calculation of crude odds ratios and the logistic

regression analyses. Please see Table 3.

The list includes the following: alcohol use in pregnancy, alcohol use in the last

two months, birth interval, body mass index group, chlamydia positive, gonorrhea

positive, gestational age, high school graduate, history of past or current sexually

transmitted disease, inadequate weight gain, low birth weight group by gestation, low

birth weight, marital status, medical history, mistimed pregnancy, mother was low birth

weight at birth, mother is of foreign birth, mother of black race, mother resides in a rural

area, prenatal care indices, prior poor pregnancy outcome, smoked during pregnancy,

smoked in the last two months, and stress identified in mother's life. All indicator variables

used in logistic regression were coded with '0' for absence of and '1' for the presence of

the designated condition, risk or measure. Other variables were coded according to their

use in the study analysis. Supportive syntax utilized for the indicator variables is contained

in Appendix B. The dependent variables included low birth weight and low birth weight

groups by gestational age. All other variables were used as independent variables in the

calculation of crude odds ratios and for the logistic regression models. Due to the

complexity of some variables, and their development from multiple database variables,

more in-depth explanation follows.

Dependent variables: (Please refer to Table 3.).

Low birth weight (LBW) was defined as birth weight greater than or equal to 500

grams and less than or equal to 2499 grams. All live births and fetal death records without

a recorded weight, a weight of less than 500 grams, or more than 5,500 grams (302

births/0.6%) were assigned to the "missing values" in the computations.

Table 3. list of All Variables Used in Analysis.

Dependent Variables

1. Low birth weight
2. Term low birth weight
3. Pre-term low birth weight

Independent Indicator Variables

1. Age< 18, > 40
2. Alcohol use from birth record
3. Alcohol use from Healthy Start screen
4. Birth interval short
5. Chlamydia positive
6. High school -no-
7. Gonorrhea positive
8. Inadequate PNC indices
9. Inadequate weight gain
10. Married not
11. Medical history
12. Mistimed pregnancy
13. Mom foreign born
14. Mom LBW
15. Prior poor pregnancy outcome
16. Race of mother black
17. Rural residence
18. Smoking from birth record
19. Smoking from Healthy Start
20. STDs, past or current
21. Stressful life

Term low birth weight (TLBW) was birth weight defined as greater than or equal

to 1500 grams and less than or equal to 2499 grams and gestational age equal to or

greater than 37 weeks. Pre-term low birth weight (PTLBW) was defined as LBW with

gestational age of equal or greater than 20 weeks and equal or less than 36 (Ventura,

Martin, Mathews, & Clarke, 1996; Graf& Perez-Woods, 1992). LBW, PTLBW and

TLBW were the dependent variables for the logistic regression models

Gestational age was calculated from the date of birth and date of the last menstrual

period reported on the birth certificate (Ventura, Martin, Curtin, & Mathews, 1999).

Where the day of the month was missing the fifteenth was imputed (Buescher, Smith,

Holliday, & Levine, 1987). Reported gestational age in weeks had a slightly narrower

standard deviation (2.56) compared to that of estimated gestation (2.84). In addition, the

skewness for the reported gestation was larger (1.07) compared to the skewness for the

calculated gestation (0.42). When those births with computed gestational age less than 20

weeks and greater than 42 weeks (11.4%) were removed, the skewness was reduced even

further to (0.37), with a histogram that presents a more normal curve. Given this

information, the assumption was made that the calculated gestational age is more reliable

since the calculated gestational age exhibited a more standard bell-shaped curve than

reported gestational age. There may be bias in the reported gestational age data due to the

fact that reported gestation is recorded after the birth has occurred and the size, weight,

and appearance of the infant may influence the clinical estimate of gestational age. Some

researchers have expressed concern that calculated gestation from the LMP date

underestimates in the direction ofpre-term deliveries (Kramer, McLean, Boyd, & Usher,

1988). There may also be recall bias associated with the calculated gestational age. The

last menstrual period date could be entered onto the birth or fetal death record based on

either information contained in the prenatal records collected long before the birth and

available at the time of the delivery, or on maternal recall However calculated gestation is

not as likely to be influenced by the birth outcome.

All live births and fetal death records without a recorded last menstrual period

(LMP) or a LMP month with a value of more than 12 or less than 1, from which to

compute the gestationwere assigned to the missingg values" category (12,557/6.4%).

Additionally, following calculation, any record with a gestational age less than 20 weeks

(79/0.2%) or more than 42 weeks (6,373/3.3%) was also assigned to the "missing values"

in the computations.

Two other methods have been employed elsewhere but not in this study to address

inconsistency in birth weight to gestation ratios. One method is a comparison of the

calculated to the estimated with selection of the one most consistent with the reported

birth weight. Another method has been to redistribute all records with inconsistent outlier

values to the distribution observed in those with valid parameters.

Independent indicator variables: (Please refer to Table 3.).

Age at the time of delivery was calculated from maternal date of birth and

incrementally grouped into < 15, 15-19, 20-24, 25-34, 35-39, 40-44, and > 45 years for

the initial analysis and calculation of crude odds ratios. For the logistic regression

modeling, the maternal age variable was assigned a value of' '1' for women younger than

18 and women 40 years or older, and a value of'0' for age 19 to 39 (Gjerdingen, 1992;

Buescher, Taylor, Davis., & Bowling, 1993; Ketterlinus, Henderson, & Lamb, 1990;

Rosenfeld, 1990).

Alcohol use in pregnancy was obtained from a field on the live birth certificate.

When 'yes' is answered, a secondary part asks for the average number of drinks per week.

This variable does not allow for identification of when, during the pregnancy, alcohol was

consumed, or if the woman drank and then stopped drinking once she realized she was


Alcohol use in the last two months was obtained from a field on the Healthy Start

prenatal screen. This variable does not allow for clarification if the consumption took

place prior to the onset of pregnancy, as might be the situation where a screen was

conducted early in the first trimester. When considered alone, the information from this

variable does not provide clear information on the quantity of alcohol consumed, or

whether the woman drank throughout the pregnancy. Both of these variables are a crude

measure of alcohol use in pregnancy and do not provide an accurate level of risk

assessment or measurement.

Birth interval was created to capture information on the time from any prior event

of pregnancy regardless of outcome, and the present birth. Many authorities have long

espoused that a woman should ideally space births two years apart to allow their bodies to

replace reserves and recover physiologically from the pregnancy and birth processes

(Youngkin & Davis, 1994). Researchers in Utah recently defined a short inter pregnancy

interval (IPI) as one that was less than 12 months based on a parallel study that

demonstrated an increased association for adverse perinatal outcomes with IPIs of less

than 12 months (Duncan, Nagle, Streeter, Bloemaum, & Tingey, 1998). This group

calculated the IPI from the time between delivery dates of consecutive live-born infants

and the gestational age of the most recent child. Another study conducted in Florida

examined the association between low birth weight and the interval between pregnancies,

reporting that the LBW increases markedly at IPI of less than 9 months (Thompson,

1995). In the second study, the inter pregnancy interval was calculated from the date of

the last menses and the date of the last live birth, both of which were taken from the birth

record. This calculation was utilized for the indicator variable with a value of '1' given to

those women with the designated risk of a birth interval less than 9 months.

Body mass index group was calculated in numerous steps based on the following

formula: weight [kilogramsj/height [meters]2 (Abrams & Parker, 1990; Anderson,

Anderson, & Glanze, 1994). Initially, the weight in pounds before pregnancy from the

Healthy Start prenatal screen file was converted to kilograms. The height in feet and

inches was then converted to height in meters. In the next step the weight in kilograms

was divided by the height in meters squared to assign a body mass index value to each

woman. The calculated score placed a woman in an underweight, normal weight,

overweight, or obese category according to published standards for pregnancy BMI

(Florida Department of Health, 1999).

Chlamydia positive and Gonorrhea positive were both taken from the Gen-Probe

PACE2C test results contained in the laboratory data file. Only those test results

designated either positive or negative, were used in the analysis. All results designated

indeterminate or unsatisfactory were assigned to 'missing' for clarity of interpretation.

High school graduate was calculated from information on the mother's highest

level of education completed contained on the birth record, grouped as having not

completed high school, having competed 12 years of education and as having completed

more than 12 years. For the logistic regression analysis, women with an age of equal to or

greater than 18 years at the time of delivery and having less than or equal to eleven years

of education completed were value of '1' for the designated risk of not having completed

high school While an age appropriate score was not assigned to younger adolescents, this

variable does not 'penalize' a young teen for not yet having completed high school at the

time of delivery, or place a younger adolescent within the risk category.

History of past or current STD was created to capture fragmented information

contained in the multiple data sets. Due to distinctly differing rationales to record or not

record any information related to a history of a recent past or present STD, all possible

positive events were captured. However, a woman was only counted once if any field was

positive for any STD event. A report of herpes in this pregnancy from the birth record was

not included. 'Yes' responses were included from the following fields:

1) chlamydia positive or gonorrhea positive on either the mother's laboratory

linked file or the infant's file;

2) a morbidity case report for the mother designating a condition of chlamydia,

gonorrhea, pelvic inflammatory disease, primary, secondary or early syphilis;

3) a morbidity case report for the infant designating a condition ofchlamydial

opthalmia, gonorrhea opthalmia, chlamydial pneumonia, or congenital syphilis;

Inadequate weight gain was computed from the body mass index (BMI) group,

gestational age in weeks and recommended weight gain ranges in pregnancy for each BMI

group. Any woman not achieving the recommended weight gain was assigned a designator

risk measure of'l'. (Florida Department of Health, 1999). See Appendix X for more

detail on the applicable calculations.

Marital status was defined as married or unmarried.