<%BANNER%>

Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2015-05-31.

DARK ITEM
Permanent Link: http://ufdc.ufl.edu/UFE0045630/00001

Material Information

Title: Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2015-05-31.
Physical Description: Book
Language: english
Creator: Patel, Nirav N
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2013

Subjects

Subjects / Keywords: Geography -- Dissertations, Academic -- UF
Genre: Geography thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Statement of Responsibility: by Nirav N Patel.
Thesis: Thesis (M.S.)--University of Florida, 2013.
Local: Adviser: Mao, Liang.
Electronic Access: INACCESSIBLE UNTIL 2015-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2013
System ID: UFE0045630:00001

Permanent Link: http://ufdc.ufl.edu/UFE0045630/00001

Material Information

Title: Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2015-05-31.
Physical Description: Book
Language: english
Creator: Patel, Nirav N
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2013

Subjects

Subjects / Keywords: Geography -- Dissertations, Academic -- UF
Genre: Geography thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Statement of Responsibility: by Nirav N Patel.
Thesis: Thesis (M.S.)--University of Florida, 2013.
Local: Adviser: Mao, Liang.
Electronic Access: INACCESSIBLE UNTIL 2015-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2013
System ID: UFE0045630:00001


This item has the following downloads:


Full Text

PAGE 1

1 MEASURING SPATIAL ACCESSIBILITY TO HIV TB TREATMENTS IN AHMEDABAD CITY, INDIA A GIS BASED APPROACH By NIRAV NIKUNJ PATEL A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN COMPLETE FULFILLMENT OF THE REQUIREMENTS FOR TH E DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2013

PAGE 2

2 2013 Nirav Nikunj Patel

PAGE 3

3 To my grandpar ents, for the inspiration through their selfless determination to their community and t o my colleagues and faculty for providing me support throughout this entire process, especially to Dr. Liang Mao

PAGE 4

4 ACKNOWLEDGEMENTS I thank my family, my advisor (Dr.Mao), my colleagues and my past teachers. The input from Dr.Fik and Dr. Waylen was also i nvaluable for the generation of this thesis. I also thank the Government of Gujarat, India as well as the Akhand Jyot Foundation in the assistance of my work. The AsiaPop project was extremely helpful as well as the SANET program developed by Atsu Okabe a nd his team.

PAGE 5

5 TABLE OF CONTENTS Page ACKNOWLEDGEMENTS ................................ ................................ ............................... 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 LIST OF ABBREVIATIONS ................................ ................................ ........................... 10 ABSTRACT ................................ ................................ ................................ ................... 11 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 12 2 LITERATURE REVIEW ................................ ................................ .......................... 14 2.1 Natural History of TB and HIV ................................ ................................ ........... 14 2.2 Treatments for TB and HIV in India ................................ ................................ ... 16 2.3 Control Programs for TB and HIV in India and in Ahmedabad City .................. 17 2.4 Accessibility Measures ................................ ................................ ...................... 19 2.4.1 Supply/ Demand Based Methods ................................ ........................... 19 2.4.2 Fixed Ca tchment Area Based Methods ................................ ................. 21 2.4.3 Floating Catchment Area (FAC) Based Methods ................................ .. 26 3 STUDY AREA AND DATA COLLECTION ................................ .............................. 31 3.1 Study Area ................................ ................................ ................................ ........ 31 3.2 Data Collection and Pre processing ................................ ................................ .. 32 3.2. 1 Data Collection ................................ ................................ ...................... 32 3.2.2 Data Pre Processing ................................ ................................ ............... 32 3.2.2.2 Georeferencing Facility Locations/Extents with Attributes .................... 33 3.2.1.1 Population data ................................ ................................ ......... 34 3.2.1.2 Road network ................................ ................................ ............ 36 4 METHODOLOGY ................................ ................................ ................................ ... 40 4.1 Estimation of Accessibility to TB and HIV Treatments in Ahmedabad City in 2010 ................................ ................................ ................................ ................. 40 4.1.1 Delineating Catchment Areas of Fa cilities based on Road Network ...... 41 4.1.2 Estimation of TB and HIV Cases Per Cell ................................ .............. 42 4.1.3 Calculation of Accessibility ................................ ................................ .... 43 4.2 Assessing Accessibility for Scenarios ................................ ............................... 44

PAGE 6

6 5 RESULTS AND DISCUSSION ................................ ................................ ............... 52 5.1 Delineation of Network Based Catchment Areas ................................ .............. 52 5.2 Estimation of Possible TB and HIV Case Numbers by Catchment Area ........... 52 5.3 Accessibility to TB and HIV Treatments in Ahmedabad City in 2010 ................ 53 5.4 TB and HIV Scenario Accessibility ................................ ................................ .... 55 5.3.1 TB Minimum Capacity Scenarios ................................ ........................... 57 5.3.2 TB Maximum Capacity Scenarios ................................ .......................... 57 5.3.3 HIV Minimum Capacity Scenarios ................................ ......................... 58 5.3.4 HIV Maximum Capacity Scenarios ................................ ........................ 59 6 CONCLUSION S ................................ ................................ ................................ ..... 70 7 FUTURE D IRECTIONS ................................ ................................ .......................... 74 APPENDIX: PERMISSION TO CONDUCT RESEARCH ................................ ............. 76 LIST OF REFERENCES ................................ ................................ ............................... 77 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 81

PAGE 7

7 LIST OF TABLES Table Page 3 1 Characteristics of Datasets involved in this Research ................................ ........ 33 3 2 ART Center Geographic Coordinates and Attending/DOTS Adjusted Attending Rates for Year 2010 ................................ ................................ ........... 34 3 3 ICTC Center Geographic Co ordinates and Attending/DOTS Adjusted Attending Rates for Year 2010 ................................ ................................ ........... 35 5 1 Estimated TB Cases within 2010 ICTC Catchment Areas ................................ .. 53 5 2 Estimated HIV Cases within 2010 ART Catchment Areas ................................ .. 53 5 3 Statistical Summary of Accessibility Rates to TB Treatment in 2010 .................. 55 5 4 Statistical Summary of Accessibility Rates to HIV Treatment in 2010 ................ 55 5 5 Total Population Values, HIV Cases and Area of HIV Catchment Areas in 2 010 ................................ ................................ ................................ ................... 55 5 6 Total Population Values, TB Cases and Area of TB Catchment Areas in 2010 .. 56 5 7 Statistical Summary of Minimum Capacity Accessibility Rates to TB ................. 57 5 8 Statistical Summary of Maximum Capacity Accessibility Rates to TB ................ 58 5 9 Statistical Summary of Minimum Capacity Accessibility Rates to HIV Treatment ................................ ................................ ................................ ........... 59 5 10 Statistical Summary of Maximum Capacity Accessibility Rates to HIV Treatment ................................ ................................ ................................ ........... 60

PAGE 8

8 LIST OF FIGURES Figure Page 2 1 Variables of Voronoi Diagram along with Graphical Expression, obtained from [28] ................................ ................................ ................................ ............. 29 2 2 Driving Time [31] ................................ ................................ ................................ 30 2 3 Example of Floating Catchment Area Method [33] ................................ ............. 30 3 1 Stu dy Area Map of Ahmedabad, Gujarat, India and Road Network .................... 36 3 2 Ahmedabad AIDS Control Society HIV Zones and ART Facility Locations ........ 37 3 3 Ahmedabad TB Control Society Zones and ICTC Facility Locations .................. 37 3 4 Map of Ahmedabad City Extents with AsiaPop 2010 Population Grid of India ... 38 3 5 Map of OpenStreetMap Road Polyline (displaying different road types) within Ahmedabad City Extents ................................ ................................ .................... 39 4 1 Complete Met hodological Flow Chart of Analysis Processes (subdivided into Sections (a), (b), (c). Blue Ovals indicate Data Sources, White Ovals/Boxes Indicate Intermediary Processes, and Green Ovals Indicate Final Products ...... 46 4 2 Detail from Section (a) from Figure 4 1 showing creation of Population Grid Cell Values with Attributes ................................ ................................ .................. 47 4 3 Detail from Section (b) from Figure 4 1 showing processes leading to the creation of Accessibility Rates for HIV/AIDS Treatment per Facility ................... 48 4 4 Detail from Section (c) from Figure 4 1 showing processes leading to the cr eation of Accessibility Rates for TB Treatment per Facility .............................. 49 4 5 Cell Calculations: (a) Zone (b) Disaggregation of Zonal Attributes into cells (c) Re aggregation of cellular Attribut e into four different zones ......................... 50 4 6 ICTC Facility Locations with Potential Facility Locations (TBLOC1, TBLOC2, and TBLOC3) ................................ ................................ ................................ ..... 50 4 7 ART Facility Locations with Potential Facility Locations (HIVLOC1, HIVLOC2, and HIVLOC3 ................................ ................................ ................................ ..... 51 5 1 ICTC Catchment Area Creation based on SANET Voronoi Diagrams ................ 61 5 2 ART Catchment Area Creation based on SANET Voronoi Diagram ................... 61

PAGE 9

9 5 3 Distribution of TB Cases in 2010 within ICTC Facil ity Catchment Areas ............ 62 5 4 Distribution of HIV Cases in 2010 within ART Facility Catchment Areas ............ 62 5 5 Accessib ility Rate to TB Treatment in 2010 within ICTC Facility Catchment Areas ................................ ................................ ................................ .................. 63 5 6 Accessibility Rate to HIV Treatment in 2010 within ART Facility Catchment Areas ................................ ................................ ................................ .................. 63 5 7 TB Scenarios (a), (b), (c) Catchment Area Creation based on SANET Voronoi Diagram ................................ ................................ ................................ 64 5 8 HIV Scenarios (a), (b), (c) Catchment Are a Creation based on SANET Voronoi Diagram ................................ ................................ ................................ 65 5 9 Minimum Capacity Accessibility Rates to TB Treatment within all TB Scenarios (a), (b), and (c) ................................ ................................ ................... 66 5 10 Maximum Capacity Accessibility Rates to TB Treatment within all TB Scenarios (a), (b), and (c) ................................ ................................ ................... 67 5 11 Minimum Capacity Accessibility Rates to H IV Treatment within all HIV Scenarios (a), (b), and (c) ................................ ................................ ................... 68 5 12 Maximum Capacity Accessibility Rates to HIV Treatment within all HIV Scenarios (a), (b), and (c) ................................ ................................ ................... 69

PAGE 10

10 L IST O F ABBREVIATIONS T ERM : Definition ART: Anti Control Society) course Treatment Regimen GIS: Geographic Information System HIV/AIDS : Human Immunodeficiency Virus/ Autoimmune Deficiency Syndrome ICTC: Integrated Counseling and Testing Centers (operated by the city of TB: Tuberculosis

PAGE 11

11 Abstract of Thesis Presented to the Graduate School of the University of Florida in Complete Fulfillment of the Requi rements for the Degree of Master of Science Measuring Spatial A ccessibility to HIV TB T reatments in A hmedabad C ity, India A GIS B ased Approach By Nirav Nikunj Patel May 2013 Chair: Liang Mao Major: Geography This study examines how accessible treatment facilities are to the populations suffering from HIV and TB within the city of Ahmedabad, and explore s potential solutions to improve overall accessibility Field collected data, high resolution population data and road network data are employed to evaluate accessibility to TB and HIV treatments based on a GIS approach. S cenario analysis identif ies optima l plans for health agencies and policymakers to invest in new treatment facilities. The analysis results show that a ccess ibility to treatment facilities for individuals that suffer from TB and HIV varies significantly with underserved areas concentrating in peripheral regions of the city The Ahmedabad Municipal Corporation (city government) can improve access ibility to TB treatment by adding a treatment facili ty in the southwestern margin of the city. For HIV treatment, t he government can improve its acce ss ibility by placing two new facilities in the northern and southern outskirts This study adds to the research literature by examining accessibility for HIV and TB on the metropolitan scale. By utilizing the catchment area generation method and the access ibility rate calculation, relates to the HIV/AIDS and TB co epidemic within metropolitan regions in India

PAGE 12

12 CHAPTER 1 INTRODUCTION AIDS (Acquired Immuno D eficiency Syndrome) is a chro nic disease caused by the Human Immunodeficiency Virus (HIV). In 2010, there has been estimated more than 34 million cases of HIV/AIDS with 1.7 million associated deaths [ 1 ] Tuberculosis (TB) is an infectious disease caus ed by strains of mycobacteria, leading to a wide range of symptoms, such as chronic cough with blood tinged sputum, fever, night sweats, and weight loss In 2011 there were an estimated 8.7 million new cases and 1.4 million associated deaths [ 2 ] HIV/AIDS often coexists with TB because HIV decreases the itself from diseases and hence increases the probability of acquir ing TB infection s [ 3 ] It has also been f ound that HIV infection promotes the progression of latent TB infection to an active disease and the relapse of the disease in patients that have been previously treated [ 4 ] These two diseases are so closely connected that they are considered to be a co epidemic [ 1 ] HIV/AIDS and TB ha ve posed tre mendous socio economic burdens o n India. In 2009, India reported 2 million new cases of TB, the hig hest in the world [ 5 ] That same year it was estimated that 2.4 million people were also living with HIV in the country, with a prevalence rate of 0.3% ( i.e., 3 HIV cases per 1000 people ). India has the third largest population of patients living with HIV/AIDS in the world [ 6 ] To combat HIV/AIDS and TB, the Indian government had initiated several programs, including the Revised National Tu berculosis Control Program (RNTCP) as well as the National AIDS Control Program (NACP). In recent years, controlling HIV/TB dual epidemic s have been a pressing issue in the city of Ahmedabad, within the state of Gujarat in India. It is estimated that there

PAGE 13

13 were approximately 8 855 cases of TB and 1 910 cases of HIV within the city in 2010 ( [ 7 ] [ 8 ] ) Currently, t he Indian Government offers free therapies for HIV/AIDS patients and several different drug regimens for TB. However, it has been reported that 14% of TB patients in the city discontinued their treatment befor e they are fully cured [ 9 ] The city TB Control Society estimate d that each of these patients can infect as many as 15 other people [ 9 ] P ossible reason s for patients discontinu ing their treatment include lack of government support, escalating traveling costs and the long treatment regimen. To provide better access for people t o obtain TB and HIV treatment, t he State of Gujar by allowing their public policymakers that work with the HIV TB dual epidemic to use innovative research methods. Against this background, thi s thesis research aims to build a GIS based health accessibility model for the city of Ahmedabad and answer s the following two research questions: 1. How accessible are the treatment facilities to population s with HIV and TB, respectively? 2. How can city gov ernment improve overall accessibility to these treatment facilities for populations at risk ? The remainder of t his thesis is organized in six further chapters. The next chapter discusses the previous health accessibility studies and their relevance to the HIV/TB dual epidemic. An introduction to the study area as well as explanation of data collection is provided in Chapter 3. Chapter 4 includes a full description of the methodology behin d the study. Chapter 5 presents the results and their discussion Chap ter 6 explains the results of the study and includes a thorough examination of its implications Chapter 7 is the final chapter of the study and examines future directions for the resea rch considering its limitations

PAGE 14

14 CHAPTER 2 LITERATURE REVIEW This chapter provide s a detailed review of Tuberculosis and Human Immunodeficiency Virus (and the treatment s associated with the diseases ), explains the control programs for TB and HIV in India as well as the city of Ahmedabad, and most importantly, reviews accessibility meas ures that can be applied for the TB/HIV studies 2.1 Natural H istory of TB and HIV Tuberculosis is an infectious disease caused by strains of mycobacteria, the most common of these, mycobacterium tuberc ulosis. The disease can be spread by people who have an ac tive TB infection, when they cough, sneeze or transmit their saliva through the air. When people are infected, often times they do not report any symptoms or have a latent infection [ 10 ] One out of ten that have the latent infection also has the chance of the disease becoming a full version of the disease, and patients must follow a strict treatment regimen T he symptoms of the disease are characteristic of respiratory illness (chronic cough, fever, night sweats, weight loss, blood tinged sputum) [ 11 ] TB can develop t hrough progression of recently acquired infection (primary disease), reactivation of latent infection, or exogenous reinfection. Infection with Mycobacterium tube rculosis (or M tuberculosis) may occu r when a person inhales particles (less than 5 m in size ) containing the tubercle bacilli from an infected person If these bacilli travel to the pulmonary alveoli, they can be ingested by alveolar macrophages. These macrophages are the first line of defense against M tuberculosis. Within the macrophage, surviv ing tubercle bacilli multiply and spr ead to other areas of the body [ 11 ]

PAGE 15

15 HIV infection or AIDS are often an independent risk factor for recent acquisition of infection and rapid progression to disease. People with HIV, when infected with TB; rapidly develo infections. TB infection is becoming more and more common with individuals that have uncompromised immunity systems and because of this increased virulence in ulosis can occur early in the course of HIV infection [ 12 ] One way scientists have been measuring the strength of immunity is through the density measurement of CD4 T cells. These T cells are instrumental in combating basic infections, and the lack of T cells within HIV/AIDS patients is among the primary reasons as to why such patients are susceptible to disease [ 12 ] I n o ne particular study o f HIV infected patients with pulmonary TB, the median CD4 T cell count was greater than 300 cells per cubic millimeter. In contrast, within patients that have extrapulmonary or disseminated disease, the median CD4 T cell count was much lower, in some cases measuring less than 80 cells per cubic millimeter [ 12 ] The same study found that the median CD4 T cell count to be 144 cells per cubic millimeter in HIV infected pat ients with all forms of TB, and as mentioned before, as the CD4 T cell count decreases in CD4 T cell the risk of developing TB becomes much greater [ 12 ] In addition, the TB also has a la rge role in the how HIV spreads in its disease history. Immune response is extremely essential to controlling TB, but immune activation is just as important. HIV has been found to accelerate the clinical course of HIV infection. Recently it has been found that high levels of tumor necrosis factor (TNF) have been contributing to the capability of increasing HIV replication in T cells when TB is also affecting the immune system [ 13 ]

PAGE 16

16 Ultimately, there are two pathways that are maintaining the HIV TB dual epidemic. The first pathway is that TB is being reactivated by HIV; scientists often cite the reason for this being the d epletion of CD4 T cells and more recently is has been found that CD8 T cells that have been affected by HIV also have a large role in controlling latent TB [ 14 ] The second and often sim ultaneously occurring pathway is the exacerbation of HIV infection by M tuberculosis infection, as mentioned before the TNF interactions play an important role in this other than the known effects that HIV infection has on the immune system [ 14 ] 2.2 Treatments for TB and HIV in India Treatment for HIV and TB in India has been greatly assisted by the free treatment initiatives that have been implemented all over the world due to the assistanc e and coordination by the World Health Organization (WHO). For HIV infected TB patients the treatment is the cotrimoxazole prophylaxis therapy (CPT) treatment. The same therapy is given to MDR TB (Multi Drug Resistant TB) infected HIV patients. This CPT t that patients with HIV can be protected from other infectious diseases that can progress their HIV [ 15 ] For patients with HIV, the same mandated antiretroviral therapy (ART) tre atment must be taken along with CPT. This ART treatment is a combination of medications that combats the HIV virus from progressing and greatly prolongs the life of an HIV patient. TB treatments include Isoniazid, Rifampicin, Pyrazinamide, Ethambutol, and Streptomycin, and must be taken three times weekly. The regimens for TB are adjusted based on what type of TB the patient has, if they are also taking ART treatment, and for other potential scenarios [ 16 ]

PAGE 17

17 The regimen for the ART and CPT treatments are similar, but all of these (Directly Observed Therapy Short course) program require patients to check into a local clinic twic e within the intensive phase two month period, followed by once every four month periodic check ins [ 16 ] 2.3 Control Programs for TB and HIV in India and i n Ahmedabad City In the 1950s, high prevalence of TB was confirmed by surveys in India In the 1960s, the Indian government initialized the National Tuberculosis Control Programme (NTCP) for infant vaccination and treatment of pulmonary TB. The NTCP was de emed to be unsuccessful between 1990 and 1992, and hence the RNTCP (Revised National Tuberculosis Control Programme) program was implemented in 1993. HIV, which was detected in India in 1986, complicated the control of TB, and led to the creation of the Na tional AIDS Control Programme (NACP) [ 17 ] The RNTCP operates in all states and districts, uses WHO guidelines to estimate and treat TB. The WHO benchmarks of 70% case detection and 85% microbiological cure have been achieved, increasing the rates of cure, decreasing case fatality, and preventing the emergence of drug resistance. Despite the successes, new cases continue to be detected without dec line, and the yearly rate of tuberculosis in children has remained at 1 2% per year since the 1970s, showing no reduction in three decades [ 18 ] India found its need for the RNTCP as the transparency of the previou s NTCP was called into question when HIV/AIDS was first discovered in the country in 1986. HIV/AIDS control has been most successful in the state of Tamil Nadu, where infection was first detected in female sex workers in 1986, the program was very successful because it worked on the epidemiology of HIV in female sex workers, patients with

PAGE 18

18 STIs, pregnant women, and blood donors [ 19 ] These factors were subsequently adopted by the NACP. Of the HIV burden, women and girls account for 40% and children account for 4%.TB, which has the highest disease burden in India, has been worsened by HIV. A high rate of resistance to sever al anti tuberculosis drugs has been reported in patients with HIV and TB in India. These problems have prompted the government to create HIV/TB collaboration divisions within control societies all across the country. A great deal of progress has been made by the AIDS control program to deliver first line antiretroviral treatment (ART) to 300,000 people by 2010, a target that was set to be achieved by March 2012 [ 20 ] Given the national history of the HIV and TB control divisions, the state of Gujarat untry, but the problems that are faced within the control societies in Gujarat are faced all across the country. In examining the Ahmedabad AIDS Control Society and TB Control Society, one of the main problems is to ensure patients have proper access to al l the treatments necessary for HIV and TB. As mentioned before, the long regimen for treatment for TB has led to individuals tapering off the treatment and creating a greater public health threat in fostering MDR TB and XDR TB (Extremely Drug Resistant TB) [ 9 ] The AIDS Control n the city where ART, CPT and TB treatments are available and managed. The TB Control Society of the city zones, where TB and CPT treatments are available as well as t he capability of

PAGE 19

19 2.4 Accessibility Measures Health accessibilit y refers to the relative ease with which locations of health care can be reached from a given location [ 21 ] A ccess to healthcare varies across space due to uneven distribution of healthcare providers and consumers, which are considered to be spatial factors. Nonspatial factors refer to the different distributions of population groups based on demographic and socioeconomic characteristics. When considering both sets of factors, it is also important to consider spatial access and aspatial access. Spatial access refers to the importance of geographic barriers like dista nce or time between the user and provider of medical services. Aspatial access refers to the non geographical barriers or facilitators such as various demographic c haracteristics like caste income, ethnicity, age, sex, etc. The interplay between these fa ct ors is useful in understand ing the dynamics of accessibility [ 22 ] A ccess to health care varies geographically because health p roviders and the general population are not distributed equally. This means that as long as the population is distributed across an area unevenly, and health providers are situated randomly across an area, access to health care will inevitably vary. In the case of providing treatment to patients with infectious diseas e in India, this is an especially pressing problem [ 23 ] 2.4.1 Supply/ Demand Based Methods The most basic accessibility measures compare the supply of facilities wi th the potential demand for their services, based on aggregates of population w ithin a defined area [ 24 ] as shown in Equation 2 1 : ( 2 1 )

PAGE 20

20 upply could be characterized as the number of physician s in a given area or the number of facilities, etc. The selection of the upply variable directly conditions the nature of th ma ny accessibility studies as how many individuals are perceived to use a p articular service or treatment. The higher the accessibility rate, the more eas ily the population can access the supply ; the lower the accessibility rate, the tougher it is for the population to access it This accessibility rate makes sense only if both variables r eflect values that are collected over a specific period of time and are consistent with each other. Joseph and Phillips incorporated this Supply Demand based method (a practitioners population ratio) used to examine spatial patterns of health services in relation to demand for a particular geographic unit of analysis [ 24 ] The advantage of this method is the relative ease of use for non specialists, which facilitates reasonably informed decisions on the ground. This most basic equation has still been incorporat ed in more sophisticated accessibility measures Such measures are limited because they do not assume the cross boundary flow of people accessing facilities in adjoining areas. The spatial resolution of the census unit under consideration affects the accur acy of the accessibility exercise, i.e., the higher th e spatial resolution, the better the ability to understand where populations are exactly located [ 25 ] Another limitation to this method is the assumption of equal access to facilities for all consumers independent o f where they live in the census tract or their personal circumstances. Some s tudies have taken this into account by using probabilistic techniques that consider overlapping areal units in order to allocate the

PAGE 21

21 supply of health services in relation to the time spent traveling to access such services [ 26 ] 2.4.2 Fixed Catchment Area Based Methods With the availability of higher spatial resolution data, more studies have created of facilities for the purpose of measuring accessibility A catchment area refers to a specific geographic region that an institution exclusively provides service for the population living within it When considering medical catchment areas, medical facilities are placed in locations that would allow for the maximum amount of indivi duals li ving within a particular geographic area to utilize the facility [ 27 ] Thus, catchment areas are used to define the popu d emand s the population ( Equation 2 1 mentioned in Section 2.4. 1). There are two different methods to generate catchment areas: the Voronoi diagram and the service area method. The concept of the Voronoi diagram is widely employed in creating catchment areas of facilities. In general, the Voronoi d iagram consider s that there is a distinct set of P i points on a plane. The plane is divided into cells so that each cell will contain only one P i point. As it can be observed from Figure 2 1 the Voronoi vertices ( v) and edges ( e ) surround the Site Points (P i ). In a typical Voronoi diagram for any spatial location ( l) within the cell, th e Euclidean distance of that spatial location to the site point within the cell has to be shorter than the distance of that spatial location to any other site point in the plane [ 28 ] To create such a diagram Voronoi edges ( e ) are used to divi de site points (P i ) that are equidistant from each other. The point where these boundaries meet is called a Voronoi vertex ( v ). In a simple method of construction, Voronoi diagrams can

PAGE 22

22 be created using perpendicular bisectors. More advanced methods like th e Fortunes algorithm use parabolic components [ 28 ] A review of data structures and algorithms that use Voronoi diagrams attest to their usefulness [ 29 ] The strength of the Voronoi diagram method is that it integrate s topology (in the form of spatial adjacency of shared vertices) and also provides an approach that fills geographic space. Through the imple mentation of a Voronoi diagram, space is fully occupied and fragmented into tiles around each geographic object, to create different zones of influence. In this way, Voronoi diagrams are created based on the spatial distribution of the geographic objects, and way the zones of influence can be explained is contingent upon the definition of the geographic object. In medical geography applications, if the geographic objects are different hospital facilities, the Voronoi diagrams that are created based on these hospital locations represent the zones of influence or catchment areas of the hospital [ 29 ] The service area me thod uses a series of considerations and weights to generate catchment areas. The critical components necessary to establish a service area are as follows: a network (in measure) a set of point locations along that network (the se rvice providers) and a variable that relates the point locations to the network (mechanism to generate service area) [ 30 ] First, the network must be bro ken down into different segment lengths by the speed limit or similar characteristics to estimate travel time through the network. Second to generate the service area, the break value in term s of travel time needs to be specified These break values relat e the point locations to the netwo rk For example, the service area that

PAGE 23

23 minute time span would be drawn to the geographic extent of how far the provider can travel on t he network within five minutes through the road network The accuracy of this service area creation is contingent upon the specificity of the data for the network and the capacity of the service provider [ 30 ] One particular study by Bullen et al. uses GIS (Geographic Information System) to define service areas (using data from British Road Networks generating travel time and District Health Authority locat ions ) for health planning in West Sussex, United Kingdom. Three main considerations were taken, first the authors focused on the need to identify meaningful areas where existing health service data could be mapped for the purpose of equitable allocation of health care resources. Second, the authors wanted to identify how to make health care delivery more effective especially to areas where the authors knew a great deal about demographics and other information. For example, within the study, West Sussex was characterized as a large rural area where access to care is very difficult for the elderly population. One fifth of the population was over age 65 at the time of the study. Third, the authors wanted to facilitate collaboration with the authorities that the y were working with, in this case between the District Health Authorities within the County of West Sussex and the West Sussex Social Services department [ 3 0 ] Given these parameters, the authors came to the conclusion that the District Health Authorities must be identifiable within any system of localities. The authors delineated their catchment areas by breaking West Sussex into 12 local areas, that are nested within each District Health Authority (which there are 36 localities in total), al so ensuring that each locality had approximately equivalent population size s [ 30 ]

PAGE 24

24 Current GIS software has been updated to include creating catchment areas based on the methodology outlined in Bullen et al In Figure 2 2 a series of fire stations was generated using their own service areas using ArcGIS. The service ar ea has been d riv ing time. Different polygon break values can be assigned for each service area facility in the can be observe d in Figure 2 2 is tw o fire stations, one with 5 minute, 10 minute, and 15 minute service area polygons (dark green point), and the other with 3 minute, 6 minute, 9 minute, 12 minute, and 15 minute service area polygons (light turquoise point) [ 31 ] Schuurman et al. developed travel time based catchment areas for British Columbia hospitals in Canada. The research used a vector based network analysis to model catchments that better represented access to health care service in British define the travel time bas ed hospital catchments based on the most accurate road network that was available. Census blocks were chosen to be the finest units of analysis, with population and dwelling counts. To model the access to hospital based ser vice using travel time catchments four health care service scenarios were modeled using The Interior Health Authority is responsible for healthcare planning and service delivery for approximately 655,000 individuals that are distributed geogr aphically in a very uneven manner. There are 22 hospitals that range in size and capability within this area. The authors of this study modeled four scenarios: 1. Population within one hour travel time of the 22 hospitals.

PAGE 25

25 2. Population within one hour travel ti me of hospitals providing critical and surgical care (10 hospitals) 3. Population within one hour travel time of any hospital with an obstetrician resident for maternal care (8 hospitals). 4. The removal of the obstetrical service from one hospital and the calc ulation of the percentage of population effected by this removal. To create travel time based hospital catchments the first task was to build the road network into a net work dataset within ArcCatalog in ArcGIS. Before establishing network dataset, travel c ost attributes are created using the segment distance, speed assigned the same cost in minutes based on a set of attributes. The new service area tool within ArcGIS wa s used to create the catchment a reas, by creating layers that used the road network line segments that are within one hour of the hospital. Each line hospital, which cou ld be used to aggregate individual segments into catchment areas. To link population to the one hour travel time catchments, 2500 meter buffer was created and the Census Blocks were linked using a spatial query that selected all block centroids within the buffer. The Census Blocks that were selected were then placed in a new spatial layer file and their populations were summed to provide total population within th e one hour travel time catchments [ 32 ] The Voronoi diagrams or the service area method have been criticized for their inability to incorporate facility si ze, facility capacity or account for chang ing elevation and variable roa d conditions. For example, both methods treat each facility the same, regardless of their capacity of offering supplies. This is not realistic, particularly in health care, because large hospitals usually have wider catchment area than small hospitals.

PAGE 26

26 By assuming the facilities are accessed in the same way, these methods also assume that the behavior of individuals in a particular catchment area are all the same, in that all individuals will solely access the hospital that is the closest to their location Both Voronoi diagrams and service based method are considered to be vector based approaches to create catchments [ 32 ] In addition, b oth of these methods assume that the all persons within delineated polygons will all equally use the facility within the catchment. In other word s people in the catchment have same accessibility to the facility, no matter how far away or how close they li ve with respect to the facility. One study advocates the use of raster based methods to weight catchment areas (for example through the creation of cost surfaces) to indicate different gradations of use and avoid these issues [ 32 ] 2.4.3 Floating Catchment Area (FAC) Based Methods The FAC method addresses the problems behind the Fixed Catchment Area methods detailed i n Sections 2. 4.2. The FAC methods can better characterize how populations access services by using circles of specified radii calculated through a geographic information system (GIS) to buffer a specified distance or travel time based on assumed utilization behavior. As shown i n Figure 2 3 centroids, catchment areas are generated A small circle is drawn around each tract centroid and is used as the basic unit to calculate the physician to population ratio. The radius of the ci travel to see a physician. When the circle is moved tract to tract, the shortage variation in different locations can be explained ( [ 21 ] [ 33 ] )

PAGE 27

27 To estimate the accessibility to health resources, a two step FCA approach has been proposed to take into account the interaction between patients and physicians across administrative boundaries The first step is to evaluate the accessibility as the ratio between supply and demand For example, this step could correspond to assigning physi cian to population ratio to service areas at physician locations Equation 2 2 demonstrates the c alculations behind the first step in the two step FCA approach. In this equation, for each physician location j gather all population locations ( k) that are within a threshold travel time ( d 0 ) from location j (referring to catchment area j ), and then comp ute the physician to population ratio R j within the catchment area. P k refers to the population of tract k whose centroid falls within the catchment ( d k less than or equal to d o ). S j is the number of physicians at location j and d kj is the travel time b etween k and j [ 21 ] (2 2) The second step is to calculate travel time cat chment around each supply point and to sum up the supply/demand ratio s at these locations. Using the example from the first step this step would correspond to summing up the initial ratios in overlapped service areas where residents have access to multiple physician locations [ 3 3 ]. As demonstrated in Equ ation 2 3, for each population location i all physician locations ( j ) that are within the threshold travel time (d o ) from location i (or catchment area i ), and

PAGE 28

28 sum up the physician to population ratios, R j at these locations to calculate the second step. (2 3) A F i i based on the two step FCA method, and R j is the physician to population ratio at physician location j whose centroid falls within the catchment centered at i (d ij less than or equal to d o ) and d ij is the travel time between i and j. The larger the value for A F i the better the accessibility [ 21 ] Th e floating catchment area (FCA) method handle s the problem of cross bo undary flows because it allows a moving catchment area [ 26 ] This is a distinct advantage over the fixed catchment area based methods, as those methods assume static and unchanging catchment areas The FCA method is useful especially if there is detailed background know ledge on a fine spatial resolution about the region where the catchment areas are being generated. For example, if extremely socioeconomically disparate populations are living in cl ose proximity to each other, the overlapping catchment areas can help identify how both populations compare to each other in terms of access to the same physician or facility [ 34 ] There are also a number of limitations of the FCA method. First, this method is still limited by assuming equal access within the catchment areas Second the FCA method assumes that all individuals within that zone live at a specific centroid of census unit and this greatly simplifying the geographic distribution of population [ 33 ] Some FCA methods assume equal distribution of population withi n a geographical unit, but this is limited as in reality populations even

PAGE 29

29 within very small census block populate areas irregularly [ 25 ] The limitations often involve researchers not having access to finer resolution data that are required for more accurate assessments of spatial distribution of individuals living within census tracts, and henc e the researche r s have to assume that services will be equally available to all the people living in the catchment area [ 25 ] There are also questions regarding the sensitivity of the supply/demand ratios and the size of the radius of the circle using the floating catchment methodology [ 35 ] Figure 2 1 Variables of Voronoi Diagram along with Graphical Expression, obtained from [ 28 ]

PAGE 30

30 Figure 2 2 Driving Time [ 31 ] Catchment Area Colors: Light Pink (3 minutes), Green (5 minutes), Orange (6 minutes),Dark Pink (9 minutes), Dark Red (10 minutes), Yellow (12 minutes), Light Red (15 Minutes) Figure 2 3. Example of Float ing Catchment Area Method [ 33 ]

PAGE 31

31 CHAPTER 3 STUDY AREA AND DATA COLLECTION 3.1 Study Area This study was conducted in the city of Ahmedabad within the Indian State of Gujarat, as shown in Figure 3 1 It is located on the banks of th e River Sabarmati and As it is located in the state of Gujarat, it is only about 20 miles away from the state capital of Gandhinagar. The city has a current population of 5.5 million with an e xtended population of 6.3 million in the peripheral areas [ 36 ] In 2011 the city was recently ranked by the Times of [ 37 ] The health department of Gujarat currently uses the Worl model for administration of treatments for TB and HIV. The city of Ahmedabad has had particular problems getting assistance and treatment to the populations on the outskirts of the city, away from the city center. The main problem with getting individuals to treatment is the cost of travel for many patients that have these diseases. The go vernment is trying to determine the best way to get fund ing to these at risk and underserved popula tions, so that they can successfully maintain the DOTS course of treatment which requires patients to check in four times during one year [ 38 ] The city is governed by the Ahmedabad Municipal Corporation. Within the city extent (redistricted in 2010), the area is divided into 6 zones and 65 wards. Each z one includes a number of wards. The AIDS Control Society of the city collects its prevalence rates of HIV/AIDS (Figure 3 2). There are two Anti Retroviral Therapy (ART) centers within Ahmedab [ 39 ]

PAGE 32

32 The RNTCP program within the city has t he city subdivided into 10 separate zones ( Figure 3 3 ). The TB Control Society collects its prevalence rates based on these 10 zone boundaries. The RNTCP program operates 17 Integrated Counseling and Testing Centers (ICTC) centers dispersed around the cit y These facilities diagnose and treat all forms of TB and provide free treatment. Also these ICTC centers diagnose HIV and if patients are diagnosed with HIV they are referred to a neighboring ART Center for treatment [ 40 ] 3.2 Data Collection and Pre p rocessing 3.2.1 Data C ollection To analyze the were collected, namely, HIV/TB related data (prevalence and treatment), population data road network data and geographic boundary data. Details of these datasets are given in Table 3 1 Particularly, to estimate the supply of HIV/TB treatments, about a month was spent during late June of 2012 in Ahmedabad to collec t TB/HIV treatment data on the ground by visiting a majority of the facilities included in this study (Appro val is included in Appendix A). 3.2.2 Data Pre Processing To integrate all datasets into ArcGIS for analysis, pre processing work was applied to each dataset, as described below. G eographic b oundary d ata ained from CEPT University in Ahmedabad as an AutoCAD file, courtesy of Dr. Anjana Vyas. These boundaries were originally in an AutoCAD drawing file, and were converted to ArcGIS shapefile format for subsequent analysis. The boundaries for the AIDS Control and TB Control Societies Zones were obtained from

PAGE 33

33 their respective organizations as pictures, and then digitized into shapefile format by hand using ArcMap (Figures 3 2 and 3 3). After conversion, the associated TB and HIV prevalence data was integrated w ith corresponding boundaries, as TB zones and HIV zones. Table 3 1 Characteristics of D ata sets involved in this R esearch Data Source Description Population Grid AsiaPop Project Spatial Resolution of 100 meters by 100 meters Ahmedabad City Extent for ye ar 2010 Ahmedabad Municipal Corporation, CEPT University AutoCAD Drawing File of City and its Extents TB Clinic Statistics for year 2010 Ahmedabad TB Control Society Monthly Records (Attending Rates for all 17 ICTC Centers) HIV Clinic Statistics for year 2010 Ahmedabad AIDS Control Society Monthly Records (Attending Rates for the 2 ART Centers) TB Detection Zone Statistics and Extents for year 2010 Ahmedabad TB Control Society Incidence data for each of 10 Zones, JPEG Image of Extent, 10 Zones HIV Detec tion Zone Statistics and Extent for year 2010 Ahmedabad AIDS Control Society Incidence data for each of 6 Zones, JPEG Image of Extent, 6 Zones Ahmedabad Road Network OpenStreetMap Polyline GIS shapefile, India Specific 3.2.2.2 Georeferencing F acility Lo cations/Extents with Attributes The 17 ICTC and 2 ART Locations were georeferenced given their geographic coordinate s and the attributes for these clinics given by the Ahmedabad AIDS Control and TB Control Societies were merged into these locations. The mo st important attribute was the during the year 2010. This attribute linic for the entire

PAGE 34

34 2010 year. Attending values refer to the total am ount of patients t hat attend a facility each month for treatment to an ICTC Center means that the patient has been diagnosed and treated for Tuberculosis, and a visit to the ART Center means the patient has bee n diagnosed and tre ated for HIV. The attending values were a djusted to the DOTS regimen by dividing the yearly attending rates by 4 as these clinics count patients monthly and patients mus t visit 4 times within one year [ 38 ] The data on Tables 3 2 and 3 3 reflect the amount of DOTS Adjusted Attending Values for ART and ICTC facilities, as well a s geographic coordinate location for each facility. Table 3 2. ART Center Geographic Coordinates and Attending/DOTS Adjusted Attending Rates for Year 2010 ART Facility Name Latitude Longitude Attending DOTS Adjusted Attending ART BJMC, Ahmedabad 23.05163 89 72.6040611 30591 7648 ART VS Gen. Hos. Ahmedabad 23.0206722 72.571 3816 954 3.2.1.1 P opulation d ata The population data was obtained from the AsiaPop project. Th is project generates high resolution population distribution data for public use by utilizing la nd cover classifications as well as census data to generate 100 meter by 100 meter population grids for many countries [ 41 ] Once the populati o n grid for India was downloaded, it was cut down to the extent of Ahmedabad City, and then transformed to the projected coordinate system of WGS 1984 UTM Zone 43 N (Figure 3 4).

PAGE 35

35 Table 3 3. ICTC Center Geographic Coordinates and Attending/DOTS Adjusted Attending Rates for Year 2010 ICTC Facility Name Latitude Longitude Attending DOTS Adjusted Attending AMC B.J.MEDICAL COLLEGE ICTC (GENERAL) 23.0516389 72.60406 11 7375 1844 AMC BEHRAMPURA REFERRAL HOSPITAL ICTC (GENERAL) 23.0043917 72.5869667 2687 672 AMC CHANDLODIA URBAN HEALTH CENTRE 23.0764056 72.5493528 4726 1182 AMC GOMTIPUR REFERRAL HOSPITAL ICTC (GENERAL) 23.0187194 72.6125167 3479 870 AMC KESAR SAL HOSPITAL 23.0836611 72.4957222 4162 1041 AMC L.G. HOSPITAL, AHMEDABAD ICTC (GENERAL) 22.9994222 72.605 3021 755 AMC NAGAR AROGYA SEVA KENDRA, RAIPUR 23.0197 72.59365 3433 858 AMC NARODA URBAN HEALTH CENTRE 23.0678361 72.6680417 2249 562 AMC ODHAV URBAN HEALTH CENTRE 23.0288833 72.667 4340 1085 AMC RUKSHMANIBEN HOSPITAL 23.0058917 72.622475 2435 609 AMC SABARMATI MATERNITY HOME 23.0804139 72.5903806 2397 599 AMC SHARDABEN HOSPITAL, SARASPUR, AHMEDABAD ICTC (GENERAL) 23.032819 7 2.6105278 4856 1214 AMC V.S. HOSPITAL, AHMEDABAD ICTC (GENERAL) 23.0206722 72.571 4573 1143 AMC VATVA URBAN HEALTH CENTRE 22.9544028 72.6184111 2360 590 AMC VEJALPUR URBAN HEALTH CENTRE 23.0071889 72.5179194 2801 700 AMC CHC, CHANDKHEDA 23.1133 25 72.57635 553 138 AMC CHC, SARKHEJ 22.97935 72.4950056 270 68

PAGE 36

36 3.2.1.2 Road n etwork The road network data was obtained from the OpenStreetMap organization, a free editable map of the world ( [ 42 ] [ 43 ] ) Th is org anization allows free access to the full map dataset and allows for great accuracy by allowing users to map and make improvements. Similar to the AsiaPop data adjustment, once this data for India was downloaded, the data was cut to the extent of Ahmedabad City and transformed to the projected coordinate system of WGS 1984 UTM Zone 43N ( Figure 3 5 ). Figure 3 1. Study Area Map of Ahmedabad, Gujarat, India and Road Networ k

PAGE 37

37 Figure 3 2. Ahmedabad AIDS Control Society HIV Zones and ART Facility Locations Figure 3 3. Ahmedabad TB Control Society Zones and ICTC Facility Locations

PAGE 38

38 Figure 3 4. Map of Ahmedabad City Extents with AsiaPop 2010 Population Grid of India

PAGE 39

39 Figure 3 5. Map of OpenStreet Map Road Polyline (displaying different road types) within Ahmedabad City Exten ts

PAGE 40

40 CHAPTER 4 METHODOLOGY This chapter explains the methods designed to estimate spatial accessibility to TB and HIV treatment in Ahmedabad City, as summarized in Figure 4 1 to Figure 4 4 4.1 Estimation of Accessibility to TB and HIV Treatments in Ahmedabad City in 2010 According to the catchment area based approach, accessibility is a ratio between the amount of supplies by a facility and the population living in the catchment area of this f acility with a demand of such supplies. In order to calculate the Accessibility Rate patients that attend the facility over a given year (in this case 2010). The subscript i represents the different catchment areas, denoting that the accessibility calculation is for a particular catchment area i. This value is placed in the numerator in Equation 1, ent area, as a variable di in Equation 4 1, is defined as the number of HIV or TB cases living in the catchment area, assuming that the cases would only seek out treatment from the facility within the catchment area. Then, the accessibility rate was calcul ated by Equation 4 1: (4 1) The information about the numerator can be directly obtained from the TB and HIV Cl inic Statistics for the Year 2010, as provided in Table 3 2 and 3 3. The challenges are the delineation of catchment area of each facility and the estimation of potential TB and HIV populatio n within each catchment area

PAGE 41

41 4.1.1 Delineating Catchment Areas of Facil ities based on Road Network To create more realistic catchment areas, this study used network distance instead of Euclidean distance, and generated network based Voronoi diagram A network based Voronoi diagram is created by converting a defi ned network (in the case of this study, a road network) into Voronoi edges that surround site points (in the case of the study, facility locations) [ 44 ] In a netwo rk based V oronoi diagram, site points must be isolated. In the case of this study, the site points are equivalent to the facility locations and t he road network acts as the pre drawn Voronoi edges. First, the polyline network must be broken down into indiv idual line segments or links. Second, a continuous graph is generated by eliminating all of the isolated polylines that are not connected to the main network. Third, once the network polylines have been properly processed, new points are generated to the n earest poi nt on network on the polylines. Within this step, polyli nes are cut at the points on the network Now that the netw ork is completely prepared, a Voronoi diagram can be created by specifying the site point locations, once the points are placed aro und the network, the Voronoi diagram splits the network around the site points, using the same methodology used to create the Voronoi diagram in Figure 2 1 [ 44 ] A n advantage of using the network based Voronoi diagrams in an Indian metropolitan area such as Ahmedabad is that since there is no small scale data on income levels and general demographic knowledge over the geography of the city, the utilization of the me thods lends health situation without relying on unreliable census data. In this circumstance these two methods are still appropriate, as the more advanced methods require more reliable and advanced dat a that are not available in most countries [ 32 ]

PAGE 42

42 T he SANET program was used to generate network based Voronoi diagrams, and delineate d the catchmen t areas for each facility. Catchment areas were generated for 17 ICTC facilities for TB treatment and 2 ART centers for HIV t reatment. T his study assume s that the population living within a catchment area is more likely to travel (through th e road network) to the facility within that particular catchment area. 4.1.2 Estimation of TB and HIV C ases P er C ell After the delineation of catchment areas, the next step is to estimate how many TB and HIV cases are living in each catchment area, as the dema nd of TB/HIV treatment. The problem is that the city government reports the TB and HIV prevalence rates by zone, not by catchment area and there is a mismatch among different sets of geographic boundaries To address this issue, this research first estima tes the TB and HIV cases per population grid cell, using the finest population unit available for the study area, and then aggregates the case numbers of all cells falling within each catchment area as illustrated in Figure 4 5 The first step wa s to use zonal TB/HIV incidence rates (Figure 4 5 a) to estimate potential TB/HIV case n umbers per population grid cell (Figure 4 5 b) First, the raster population grid cells were converted to feature points with each point associated with a population size wit hin the corresponding cell location T hese feature points were then spatially joined with attributes from the HIV and TB Zones (Figures 3 2 and 3 3 ) As a result, each cell obtains the TB/HI V incidence from the zone that contains it The estimated TB and HIV case number in a population grid cell can be estimated by Equation 4 2: (4 2)

PAGE 43

43 The ArcGIS raster calculation tool was employed to implement the Equation 4 2 on each population grid cell, producing two grid maps: the TB case number grid an d the HIV case number grid. The second step (f rom Figure 4 5 b to c ) was to reaggregate cell information by catchment areas so that the potential TB and HIV case number per catchment area can be estimated. T he data in feature point form were into catchment areas using a spatial join, and the feature points were zonally summed to calculate amounts of HIV and TB cases within each catchment area (Equation 4 3): ( 4 3) T he variable g represent all numerical attributes of the populati on grid cells, including the estimated TB or HIV count per grid cell that fall within a particular catchment area. The ArcGIS Spatial Join f unction was employed to implement Equation 4 3 which also allows the user to sum the values of the features that fa ll within a particular zone 4.1.3 C alculation of Accessibility After computing the numerator and denominator, this research employed the ArcGIS Field Ca lculator to implement Equation 4 1 and estimate d the accessibility for TB and HIV treatment by catchment ar ea, respectively. Statistical analysis was then applied to the results to summarize centrality and distribution of accessibility rates in the study area. The measurements include the mean accessibility, median, standard deviation, the frequency distributio n, and the over served and under served population. Further, the results were mapped by catchment area to identify over served and under served areas. These analyses were then served as a basis for designing scenarios to improve the overall accessibility o f the entire city.

PAGE 44

44 4.2 Assessing Accessibility for Scenarios This research tested the 6 TB Scenarios and the 6 HIV Scenarios to investigate how they can improve the overall accessibility to TB and HIV treatment. The 6 TB Scenarios were designed by adding 1 ex tra ICTC facility into the city, but at three different geographic locations within the city (Figure 4 6 ) The three location scenarios were designed as follows : TBLOC1 was placed at the northeast end of the city, as there is an at risk TB population in th at ward. TBLOC2 and TBLOC3 were placed southwest and southeast respectively, as the southern wards of Ahmedabad tend to have the poorest and most disease susceptible populations [ 45 ] In addition to the three location scenario s, two capacity scenarios were proposed to take into account the supply capacity of the new facili ty. O ne scenario assumed that the added facility would take in as many patients as the highest attending facility and the other assumed that the added facility would take as many patients as the least attending facility. Thus a total of 6 TB Attending Rat es were gener ated for 6 possible scenarios (3 Possible Locations x 2 Capacity Sc enarios (Min & Max) = 6 total TB S cenarios). In order to calculate 6 TB a ccessibility r ates new facility catchment areas were generated for each scenario and the TB case number was re calculated for each catchment area These a ccessibility rates were calculated by using Equation 4 1. Similar to Section 4.1.2, statistical analysis was applied to summarize the results of each scenario. A comparison analysis was then conducted to i dentify the best scenario that achieve d high mean accessibility and low standard deviation, i.e ., all people in the city have high level and equal accessibility to TB treatment. Similar to the TB scenario analysis, t he 6 HIV Scenarios were designed by add ing 2 extra H IV facilities into the city, but at three different geographic locations

PAGE 45

45 within the city (Figure 4 7 ) The three location scenarios for the new facilit ies were designed as follows, HIVLOC1 was placed at the north and south ends of the city, in order to test access in the periphery. HIVLOC2 was placed northwest and centrally southeast to test changes in the accessibility when adding clinics closer to the center of hery. HIVLOC3 was placed at the most at risk areas of the periphery in the southwest and northeast [ 45 ] In addition to the three location scenarios, two capacity scenarios were proposed to take into account the supply capacity of the new facility. One scenario assumed that the added facility would take in as many patients as the highest attending facility and the other assumed that the added facility would take as many patients as the least attending facility. Thus a total of 6 HIV Attending Rates were generated for 6 possible scenarios (3 Possible Locations x 2 Capacity Scenarios (Min&Max) = 6 total HIV Scenarios). In order to calculate 6 HIV Accessibility Rates new facility catchment areas were generated for each scenario, and the HIV case number was re calculated for each catchment area These accessibility rates were calculate d by using Equation 4 1. Similar to Section 4.1.2, statistical analysis was applied to summarize the results of each scenario. A comparison analysis was then conducted to identify the best scenario that achieved high mean accessibility and low sta ndard deviation, i.e ., all people in the city have high level and equal accessibility to HIV treatment.

PAGE 46

4 6 Figure 4 1. Complete Methodological Flow Chart of Analysis Processes (subdivided into Sections (a), (b), (c). Blue Ovals indicate Data Sources, Wh ite Ovals/Boxes Indicate Intermediary Processes, and Green Ovals Indicate Final Products

PAGE 47

47 Figure 4 2. Detail from Section (a) from Figure 4 1 showing creation of Population Grid Cell Values with Attributes

PAGE 48

48 Figure 4 3. Detail from Section (b) from Figure 4 1 showing processes leading to the creation of Accessibility Rates for HIV/AIDS Treatment per Facility

PAGE 49

49 Figure 4 4 Detail from Section (c) from Figure 4 1 showing processes leading to the creation of Accessibility Rates for TB Treatment per Facility

PAGE 50

50 Figure 4 5 Cell Calculations: (a) Zone (b) Disaggregation of Zonal Attributes into cells (c) Re aggregation of cellular Attribute into four different zones Figure 4 6 ICTC Facility Locations with Potential Facility Locations (TBLOC1, TBLOC2, and TBL OC3)

PAGE 51

51 Figure 4 7 ART Facility Locations with Potential Facility Locations (HIVLOC1, HIVLOC2, and HIVLOC3

PAGE 52

52 CHAPTER 5 RESULTS AND DISCUSSION This chapter shows the results of the analysis by using the methodology described in Chapter 4. 5.1 Delineation of N etwork B ased C atchment A reas Catchment areas were created for ICTC Clinics in 2010, ART Clinics in 2010 as well as the TB and HIV Scenarios using the methodology outlined in Section 4.1. Figure 5 1 shows 17 catchment areas for each ICTC facility in 2010. There is a clu ster of facilities in close proximity in the center of the city, with less concentration of facilities in the periphery of the city. These catchment areas are used to calculate acce ss to TB treatment in 2010 ( Section 5.2). Figure 5 2 show s the two catchmen t areas for the two ART facilities in 2010. There are only two facilities within the city, splitting the city diagonally into two catchment areas. These catchment areas are used to calculate acce ss to HIV treatment in 2010( Section 5.2). 5.2 Estimation of P ossi ble TB and HIV C ase N umbers by C atchment A rea Using the procedures outlined in Sections 4.1.2 and 4.1.3, the amount of TB and facilities (Table 5 1) There is a great amount of variation in the amount of TB cases wit hin each ICTC catchment area. 15 out of 17 catchment areas within the city have at least over 100 TB cases with the exception of two catchment areas in the northern part of the city ( Figure 5 3) The data from Figure 5 4 and Table 5 catchment area has 1190 possible

PAGE 53

53 cases. Both numbers are sizeable populations, and HIV cases are especi ally at risk if other cases are not consistent with their treatment regimen. Table 5 2. Estimated HIV Cases within 2010 ART Catchment Areas Facility Catchment Area Estimated HIV Cases ART BJMC, Ahmedabad 721 ART VS Gen. Hos. Ahmedabad 1190 Total 1910 5.3 Accessibility to TB and HIV Treatments in Ahmedabad City in 2010 For access ibility to TB treatment in 2010 (Figure 5 5), it is apparent that the northern areas of the city are b eing over served while the east, south and west peripheries of the city are relatively underserved. The east, south, and west peripheries Table 5 1. Estimated TB Cases within 2010 ICTC Catchment Areas Facility Catchment Area Estimated TB Cases AMC B.J.MEDICAL COLLEGE ICTC (GENERAL) 1043 AMC BEHRAMPURA REFERR AL HOSPITAL ICTC (GENERAL) 724 AMC CHANDLODIA URBAN HEALTH CENTRE 647 AMC GOMTIPUR REFERRAL HOSPITAL ICTC (GENERAL) 304 AMC KESAR SAL HOSPITAL 41 AMC L.G. HOSPITAL, AHMEDABAD ICTC (GENERAL) 441 AMC NAGAR AROGYA SEVA KENDRA, RAIPUR 351 A MC NARODA URBAN HEALTH CENTRE 830 AMC ODHAV URBAN HEALTH CENTRE 642 AMC RUKSHMANIBEN HOSPITAL 660 AMC SABARMATI MATERNITY HOME 266 AMC SHARDABEN HOSPITAL, SARASPUR, AHMEDABAD ICTC (GENERAL) 841 AMC V.S. HOSPITAL, AHMEDABAD ICTC (GENERAL) 6 74 AMC VATVA URBAN HEALTH CENTRE 658 AMC VEJALPUR URBAN HEALTH CENTRE 520 AMC CHC, CHANDKHEDA 99 AMC CHC, SARKHEJ 114 Total 8855

PAGE 54

54 are underserved, the huge standard deviations indicates great differences in how clinics are operating ( Table 5 3 ) It is important to note that the value for the IQR (Inter Quartile Range) is 90% which means that there is great variation within the data, and not that much consistency with how cases access treatment within Ahmedabad City. The same results were apparent wh en examining accessibility to HIV treatments. Although one ART catchment area has significantly more cases of HIV and is underserving, the other ART catchment area is over serving its con stituents by a great margin ( Figure 5 6). A part the reason why the n umbers are skewed is that the BJMC facility taking in patients from all over the state of Gujarat. This is why the facility has such a high attending rate, and this is why Table 5 5 shows a very large IQR value. When comparing the results from Tables 5 3 a nd 5 5, it is interesting to note that the 2 ART Centers actually have a lower standard deviation then the 17 ICTC Centers. When looking at the data from these two tables, it could be noted that both the average accessibility rates are well over 100%. Howe ver, when considering the great degree of variance of facility performance in being able to provide treatment for disease, one serve cannot counteract the negative effects of another ses and hence further perpetuates the HIV TB co epidemic even further. This danger is very clear when observing the results from Table 5 4 and Table 5 6. The differences in population serviceability based on location are apparent when observing the total p opulation values within each catchment area. By placing facility locations in more informed ways around the road network, the possibility to even out the

PAGE 55

55 population counts that fall under each catchment area can help dr astically reduce facility load. Table 5 3. Statistical Summary of Accessibility Rates to TB Treatment in 2010 Mean Standard Deviations Median Q1 Q3 IQR TB Access Rate 2010 293 563 169 93 183 90 Table 5 4 Statistical Summary of Accessibility Rates to HIV Treatment in 2010 Mean Standard Deviations Median Q1 Q3 IQR HIV Access Rate 2010 571 490 571 325 816 490 Table 5 5 Total Population Values, HIV Cases and Area of HIV Catchment Areas in 2010 Facility Catchment Area Total Population HIV Cases Area (km 2 ) Area (mi 2 ) ART BJMC, Ahmedabad 1 397 795 721 134 52 ART VS Gen. Hos. Ahmedabad 2 403 731 1190 309 119 5.4 TB and HIV Scenario Accessibility TB and HIV Scenario Catchment areas were created based on the methodology outlined in Section 4.2. For TB Scenarios, one extra facil ity was added and in HIV Scenarios, two extra facilities were added. Thus TB Scenarios have a total of 18 catchment areas per scenario (Figure 5 7) and HIV Scenarios have a total of four catchment areas per scenario (Figure 5 8)

PAGE 56

56 Table 5 6. Total Popul ation Values, TB Cases and Area of TB Catchment Areas in 2010 Facility Catchment Area Total Population TB Cases Area (km 2 ) Area (mi 2 ) AMC B.J.MEDICAL COLLEGE ICTC (GENERAL) 278753 1043 21 8 AMC BEHRAMPURA REFERRAL HOSPITAL ICTC (GENERAL) 204871 72 4 53 20 AMC CHANDLODIA URBAN HEALTH CENTRE 428042 647 37 14 AMC GOMTIPUR REFERRAL HOSPITAL ICTC (GENERAL) 89600 304 5 2 AMC KESAR SAL HOSPITAL 29995 41 36 14 AMC L.G. HOSPITAL, AHMEDABAD ICTC (GENERAL) 178763 441 10 4 AMC NAGAR AROGYA SEVA KENDRA, RAIPUR 80115 351 4 2 AMC NARODA URBAN HEALTH CENTRE 293140 830 30 12 AMC ODHAV URBAN HEALTH CENTRE 190566 642 24 9 AMC RUKSHMANIBEN HOSPITAL 228274 660 16 6 AMC SABARMATI MATERNITY HOME 208398 266 24 9 AMC SHARDABEN HOSPITAL, SARASP UR, AHMEDABAD ICTC (GENERAL) 232757 841 13 5 AMC V.S. HOSPITAL, AHMEDABAD ICTC (GENERAL) 415044 674 25 10 AMC VATVA URBAN HEALTH CENTRE 355421 658 66 25 AMC VEJALPUR URBAN HEALTH CENTRE 416444 520 40 15 AMC CHC, CHANDKHEDA 81855 99 23 9 AMC CHC, SARKHEJ 93161 114 17 6

PAGE 57

57 5.3.1 TB Minimum C apacity Scenarios For the TB Access Scenarios the same principles are applied as mentioned in 4.2. There are 17 ICTC facilities as opposed to the 2 ART facilities for HIV treatment. In the TB scenarios, one facil ity is added in a designated location. For th ese minimum capacity scenario s as observed in Figure 5 9 it is hard to discern differences in any of the scenarios and it seems like the same areas around the cit y are left underserved when the ext ra TB facilit y only has minimum capacity. The data in Table 5 7 shows that all of the Scenarios underperform regarding mean accessibility rates, but are all slightly better in terms of lowering the standard deviation and IQR value, which indicates more equal accessibil ity. Scenario 2 has a slightly lower IQR than Scenarios 1 and 3 due to its geographic location in a very high case area (Figure 5 9), the location was very impo rtant in assisting to allow f or nearby c linics to provide equal access. Table 5 7. Statistical S ummary of Minimum Capacity Accessibility Rates to TB Mean Standard Deviations Median Q1 Q3 IQR 2010 TB Access 293 563 169 93 183 90 TBLOC1 Minimum Scenario 287 549 170 118 212 94 TBLOC2 Minimum Scenario 282 550 159 110 191 81 TBLOC3 Minimum Scenario 282 549 159 97 191 94 5.3.2 TB Maximum Capacity Scenarios In these maximum capacity scenarios, as predicted, the added TB facility over served the geographic area that it was placed in (Figure 5 10). All facilities averaged a very high mean, but what was very

PAGE 58

58 location enabling a low standard deviation and low IQR relative to the other scenarios, as observed in Table 5 8. Scenarios 2 and 3 performed very well in maximizing mean accessibility values, but it is impo rtant to note that these locations were placed in close proximity to multiple already existing ICTC clinics whereas Scenario 1 is extremely proximate to one very high risk area for TB. Although the mean accessibility rate for Scenario 1 was only slightly h igher than the mean accessibility rate for all ICTC centers in 2010, the location of the facility is very relevant. Selection to place a new facility extremely proximate to another might see m counterintuitive, but using network analysis it can be observed that the positioning of the facility in Scenar io 1 can help improve accessibility Table 5 8. Statistical Summary of Maximum Capacity Accessibility Rates to TB Mean Standard Deviations Median Q1 Q3 IQR 2010 TB Access 293 563 169 93 183 90 TBLOC1 Max imum Scenario 302 544 173 138 224 85 TBLOC2 Maximum Scenario 349 590 170 118 218 100 TBLOC3 Maximum Scenario 367 623 170 118 218 100 5.3.3 HIV Minimum C apacity Scenarios In Section 4.2 a rationale was outlined as to how HIV Minimum Capacity Scenarios are to be generated. By applying the lowest attending rate to the two new HIV treatment facilities within each scenario, changes in accessibility can be ob served when looking the data. Observing the data from Figure 5 11, it can be observed that HIV Scenario 1 mi ght have the best facility locations in this circumstance, as the accessibility of the scenario

PAGE 59

59 extends to a very large geographic area. This observation is confirmed by Table 5 9, where HIV Scenario 1 has the lowest standard deviation value and the second lowest IQR value. It averaged only 30% under from the ART facilities in 2010, but gives a drastic improvement in creating optimal accessibility. HIV Scenario 2 produced very distr ibution in HIV Scenario 1. HIV Scenario 3 did not perform as well, most notably having an IQR that was approximately 300 more than the other two scenarios. The only metric where HIV Scenario 3 is better than the others is with its mean value for accessibil ity. However, with such great variation with its standard deviation and IQR values, the scenario is not optimally accessible. Table 5 9. Statistical Summary of Minimum C apacity Accessibility Rates to HIV Treatment 5.3.4 HIV Maximum C apacity Scenarios In th ese maximum capacit y scenario s it is assumed that the two new added facilities maintain the maximum facility attending rate (rationale mentioned in Section 4.2). Observing Figure 5 12, HIV Scenario 1 and HIV Scenario 2 are both equivalently very good. HIV Scenario 3 seems t o be plagued by the huge service catchment area that is created by adding facilities in line from west to east. Mean Standard Deviations Median Q1 Q3 IQ R HIV Access Rate 2010 571 490 571 325 816 490 HIVLOC1 Minimum Scenario 545 397 452 198 799 602 HIVLOC2 Minimum Scenario 593 539 375 161 806 645 HIVLOC3 Minimum Scenario 792 641 677 266 1203 937

PAGE 60

60 Table 5 10 indicates that HIV Scenario 2 has the lowest standard deviation and IQR values, indicating a higher degree of consistency. Scenarios 2 and 3 have very high mean values as does Scenario 1, but their other statistical values are equally skewed. The relative success of Scenario 2 in this circumstance is due to the distribution of the clinics on parts of the road ne twork that directly conne cts patients on the right corridors. In contrast, HIV Scenario 1 has facilities located more distant from the city center and its facility locations are very north and south, these partitions the catchment areas in a very peculiar way that is not as equal as HIV Scenario 2 ( Figure 5 12) Table 5 10. Statistical Summary of Maximum C apacity Access ibility Rates to HIV Treatment Mean Standard Deviations Median Q1 Q3 IQR HIV Access Rate 2010 571 490 571 325 816 490 HIVLOC1 Maximum Scenario 2130 2039 1424 8 83 2670 1787 HIVLOC2 Maximum Scenario 1871 1690 1321 916 2276 1360 HIVLOC3 Maximum Scenario 3167 3068 2167 1318 4016 2699

PAGE 61

61 Figure 5 1. ICTC Catchment Area Creation based on SANET Voronoi Diagrams Figure 5 2. ART Catchment Area Creation based on SA NET Voronoi Diagram

PAGE 62

62 Figure 5 3. Distribution of TB Cases in 2010 within ICTC Facility Catchment Areas Figure 5 4. Distribution of HIV Cases in 2010 within ART Facility Catchment Areas

PAGE 63

63 Figure 5 5. Accessibility Rate to T B Treatment in 2010 within ICTC Facility Catchment Areas Figure 5 6. Accessibility Rate to HIV Treatment in 2010 within ART Facility Catchment Areas

PAGE 64

64 a) TBLOC1 b) TBLOC2 c) TBL OC3 Figure 5 7. TB Scenarios (a), (b), (c) Catchment Area Creation based on SANET Voronoi Diagram

PAGE 65

65 a) HIVLOC1 b) HIVLOC2 c) HIVLOC3 Figure 5 8. HIV Scenarios (a), (b), (c) Catchment Area Creation based o n SANET Voronoi Diagram

PAGE 66

66 a) TBLOC1 b) TBLOC2 c) TBLOC3 Figure 5 9. Minimum Capacity Accessibility Rates to TB Treatment within all TB Scenarios (a), (b), and (c)

PAGE 67

67 a) TBLOC1 b) TBLOC2 c) TBLOC3 Figure 5 10. Maximum Capacity Accessibility Rates to TB Treatment within all TB Scenarios (a), (b), and (c)

PAGE 68

68 a) HIVLOC1 b) HIVLOC2 c) HIVLOC3 Figure 5 11. Minimum Capacity Accessibility Rates to HIV Treatment within all HIV Scenarios (a), (b), and (c)

PAGE 69

69 a) HIVLOC1 b) HIVLOC2 c) HIVLOC3 Figure 5 12. Maximum C apacity Access ibility Rates to HIV Treatment within all HIV Scenarios (a), (b), and (c)

PAGE 70

70 CHAPTER 6 CONCLUSION S This the sis examines how accessible treatment facilities are to the populations with HIV and TB within t he city of Ahmedabad and explores potential solutions to improve overall accessibility to these treatment facilities. Field investigation data, h igh resolution population data and road network data were employed to evaluate accessibility to TB and HIV tre atments based on a GIS approach Optimal plans were suggested from scenario analysis to guide health agencies and policy makers to invest new treatment facilities There are four major findings from the research: 1. Access to treatment facilities for individu als that suffer from TB varies significantly in Ahmedabad. Large standard deviations as a well as a large inter quartile range indicates great differences in how TB clinics are operating, showing that the 17 facilities do not provide uniformly equal access T he northern areas of the city are over served while the east, south and west peripheries of the city are underserved. The underserved areas correlate with the distribution of most at risk populations in Ahmedabad. 2. Access ibility to treatment facilities for individuals that suffer with HIV in Ahmedabad also varies significantly. T here is a big difference in access between those two facilities, with one facility taking in ten times as many patients. The B.J. Medical College Hospital takes in ten times as many patients because it has been taking patients from the entire state of Gujarat. This leaves the V.S. Hospital as the only dedicated city 3. The Ahmedabad Municipal Corporation (city government) can improve accessibility to TB treatm ent by placing a new facility in the southwestern periphery of the city. This scenario performed the best in comparison to its counterparts, when assuming both the minimum and maximum capacity performance of the facil ities The resultant overall accessibility of the population is higher than any other scenario, while the variance of accessibility among facilities is the minimum, thus achieving the highest degree of health equality. 4. The Ahmedabad Municipal Corporation (city government) can improve accessibility to HIV treatment by adding facilities in the northern and southern peripheries of the city. This scenario performed the best in comparison to its counterparts, when assuming both the minimum and maximum capacity performance of the facilities The resultant overall accessibility of the population is higher than any other scenario, while the variance of accessibility among facilities is the minimum, thus largely reducing the health in equality issue in the city

PAGE 71

71 Thi s study adds to the research literature by examining a ccessibility for HIV and TB on the metropolitan scale. By utilizing the catchment area generation method and the accessibility rate calculation, public policymakers can evaluate the implementation of th e DOTS program as it relates to the HIV/AIDS and TB co epidemic within metropolitan regions in India. This study provides a geographic analysis methodology to analyze accessibility to facilities for HIV and TB treatment through the use of an accessibility rate based on attending rates. Secondly, i n the literature on accessibility research, this study is important in that it shows the value of high resolution population spati al scales. Very few accessibility studies have studied the impact on accessibility of a dual epidemic and viewing the results, it is clear that there is a great value in the way individual clinics can be assessed for performance in handling infectious dise ase. In terms of being implemented within the Indian government, the greatest advantage by having an example to judge metrics for performance within clinics is giving opportunities for the gover nment to identify geographic problem areas within a city and t o rectify issues with in certain areas not getting as much service as they should. The biggest problem is being able to coordinate between the RNTCP and NACO programs, and what GIS research can do for all of India is provide some basis to identify what make s having low accessibility to TB treatments so dangerous to people living with HIV ( [ 46 ] [ 47 ] ) There are several limitations to the study. First, this study assumes that all cases of HIV and TB patients within a catchment area would attend a treatment facility. In India, there is a huge problem with HIV/AIDS and TB patients seeking treatment due to

PAGE 72

72 social stigma, lack of education and lack of access to basic transportation. The results from this study provide a way to study the most at risk regions of the city t o rectify these issues of access. In the future, b y possibly collecting more detailed data within each sub region of the city on population movement as it relates to these barriers this study can be improved greatly by understanding sub regional bias towa rds certain facilities [ 17 ] Second, this study also assumes that the facilities within the city will match their attending capacity year by year. As the data was estimated for 2010, these clinics sometimes have drastic changes in resources and sometimes might not have the same staff for the next year over time would be useful in crea ting the scenarios, as all of the facility attending rates can be adjusted for better accuracy In the case of this study the only year that had the most complete data available was 2010, hopefully in the future more data is collected on an annual basis fo r HIV and TB treatment facility performance. Granted these two limitations, the public policy knowledge that can be derived from this study can be instrumental to solving these two problems [ 17 ] Despite not having a cure for HIV/AIDS, HIV/AIDS does not have to be the accomplice to the death of people that have TB. The current set of treatments that are currently available can prolong lives. The Go vernment of Gujarat is attempting to help provide travel support for patients in areas that are distant from ART centers, ensuring that people who suffer from such diseases can get access to treatment and lower the disease burden. This study can help de vel op a public policy plan for this effort in an efficient manner. Through the utilization of accessibility rates to evaluate potential new

PAGE 73

73 facility locations, this study makes it possible to make informed policy decisions in reducing the burden of the HIV TB dual epidemic and possibly save many lives.

PAGE 74

74 CHAPTER 7 FUTURE DIRECTIONS geographic catchment areas, when in actuality these catchment areas vary based on different characteristics of the environment. The main limitation in creating a dynamic and adaptable model to this study is the limitation of data. The possibility to implement a catchment area creation that reflects travel time or socioeconomics, for example, creates a more accurate id ea of how people are accessing treatment as opposed to using a static model. The Voronoi diagrams within this study were created along a network not taking into account and statistical characteristics related to clustering or using a weighted index. The ad vantages of using a weighted index to generate Voronoi diagrams is that if there is more data known about the geographic region in question and this affects the shape of the Voronoi diagram, then it can be integrated into the creation of the Voronoi diagra ms. In adding these attributes as a component to Voronoi diagram creation, the kernels of the Voronoi segments that are created are oft en clustered to create kernel density indexes so that different statistical assumptions can be tested [ 48 ] One particular indicator that would be very useful in the city of Ahmedabad is the aspect of distance minimization based on affluence. As mentioned before Ahmedabad ha s very distinct and disparate socio economic classes. Distance to kernel clusters where there are very affluent or very poor populations can help create Voronoi diagrams tell more about the constraints of an individual and their ability to access treatment [ 49 ] This use of distance to kernel clusters ties into utilization of the distance decay function as well, the farther away a cluster of poorer people are to a given f acility

PAGE 75

75 location, the less likely it is for that cluster to reach that facility for treatment in comparison to a more affluent population. Gravity models integrate distance by providing a combined indicator of accessibility and availability by integrating Law of Gravitation. These models strive to represent the interaction between a population point and all service points within a reasonable distance [ 50 ] Another limitation is the placement of facility locations within the study The assumptions of facility placement are all assumptions that the author made based on information on where poorer populations of the cit y are likely to reside. Having more fine spatial data on socioeconomic information as well as other indicators would make an analysis of suitable locations to add new facilities less arbitrary In evaluating the performance of these facilities it is also important to consider the underlying objectives and goals of each facility. As the author was not given permission to ask each individual facility of its own needs and problems, it was assumed that new facilities would be added in performing maximally or minimally given the data of the given year. In the case of the study, 2010 was being used as a point of reference [ 50 ] U sing one year (2010) to examine and predict outcomes in a city is also inherently problematic, but it would be better if instead of looking at one particular year things can be potentially expanded for five years in the future if given better quality data. The author attempted to get some more detailed data within the city on socio economic data by collected ration card and transportation card data, which is a system that the government offers to people needing fina ncial assistance

PAGE 76

76 APPENDIX A PERMISSION TO CONDUC T RESEARCH

PAGE 77

77 LIST OF REFERENCES 1. Statistics:Worldwide [ http://www.amfar.org/about_hiv_and_aids/facts_and_stats/ statistics__worldwide/ ] 2. Tuberculosis [ http://www.who.int/mediacentre/factsheets/fs104/en/ ] 3. Frequently Asked Questions about TB and HIV [ http://www.who.int/tb/challenges/hiv/faq/en/ ] 4. Bauer AL, Hogue IB, Marino S, Kirschner DE: The Effects of HIV 1 Infection on the Latent Tuberculosis Mathematical Modeling of Natural Phenomena 2008, 3 (7):229 266. 5. Organization WH O : Gl obal Tuberculosis Control 2010. 6. UNAIDS: Report on the Global AIDS Epidemic 2010 In Geneva: UNAIDS; 2010. 7. Society ATC: Monthly Records for Zonal TB Statistics and Clinic Statistics In ; 2010. 8. Society AAC: Monthly Records for Zonal HIV/AIDS Statis tics and Clinic Statistics In ; 2010. 9. Desai K, Dave J: 14% TB Patients in City Negligent; Each Can Infect 15 People In: Daily Bhaskar. Daily Bhaskar; 2012. 10. Dye C: Global Epidemiology of Tuberculosis The Lancet 2006, 367 (9514):938 940. 11. Barry CE, Boshoff HI, Dartois V, Dick T, Ehrt S, Flynn J, Schnappinger D, Wilkinson RJ, Young D: The Spectrum of Latent Tuberculosis: Rethinking the Biology and Intervention Strategies Nature Reviews Microbiology 2009. 12. Lawn SD, Churchyard G: Epidemiology of HI V associated Tuberculosis Current Opinion in HIV and AIDS 2009, 4 (4):325 333. 13. Kedzierska K, Crowe S, Turville S, Cunningham A: The influence of cytokines, chemokines and their receptors on HIV 1 replication in monocytes and macrophages. Rev Med Viro l 2003, 13 :39 56. 14. Pawlowski A, Jansson M, Skold M, Rottenberg ME, Kallenius G: Tuberculosis and HIV Co Infection Tuberculosis and HIV Co Infection 2012, 8 (2):1 7.

PAGE 78

78 15. Raizada N, Chauhan LS, Sai Babu S, Thakur R, Khera A, Wares DF, Sahu S, Bachani D, Rewari BB, Dewan PK: Linking HIV Infected TB Patients to Cotrimoxazole Prophylaxis and Antiretroviral Treatment in India PLoS ONE 2009, 4 (6). 16. Dye C, Garnett G, Sleeman K, Williams B: Prospects for Worldwide Tuberculosis Control under the WHO DOTS Stra tegy The Lancet 1998, 352 (9144):1886 1891. 17. John JT, Dandona L, Sharma VP, Kakkar M: Continuing Challenge of Infectious Disease in India The Lancet 15 21 2011, 377 (9761):252 269. 18. Chadha V, Agarwal S, Kumar P, al. e: Annual risk of tuberculosis inf ection in four defined zones of India: a comparative picture Int J Tuberc Lung Dis 2005, 9 :569 575. 19. John T, Babu P, Jayakumari H, Simoes E: Prevalence of HIV infection in risk groups in Tamil Nadu, India Lancet 1987, 329 :160 161. 20. National AIDS Co ntrol Organization MoHaFW, Government of India: National AIDS Control Programme, Phase III 2006. 21. Luo W, Wang F: Measures of Spatial Accessibility to Health Care in a GIS Environment: Synthesis and a Case Study in the Chicago Region Environment and Pl anning B: Planning and Design 2003, 30 (6):865 884. 22. Wang F, Luo W: Assessing Spatial and Nonspatial Factors for Healthcare Access: Towards an Integrated Approach to Defining Health Professional Shortage Areas Health & Place 2005, 11 (2):131 146. 23. Pe ters DH, Garg A, Bloom G, Walker DG, Brieger WR, Rahman MH: Poverty and Access to Health Care in Developing Countries 1136 2008:161 171. 24. Joseph AE, Phillips DR: Accessibility and utilization: Geographical perspectives on health care delivery London, UK: Harper & Row; 1984. 25. Makuc DM, Haglund B, Ingram DD, Kleinman JC, Feldman JJ: The use of Health Service Areas for measuring provider availability The Journal of Rural Health 1991, 7 (4):347 356. 26. Wing P, Reynolds C: The availability of physician services: A geographic analysis Health Services Research 1988, 23 (5):649 667. 27. Regier DA, al. e: The NIMH Epidemiologic Catchment Area program: historical context, major objectives, and study population characteristics Archives of general psychiatry 1 984, 41 (10):934.

PAGE 79

79 28. Miu A: Lecture 7: Voronoi Diagrams In: Computational Geometry. Boston: Massachusetts Institute of Technology; 2001. 29. Gahegan M, Lee I: Data structures and algorithms to support interactive spatial analysis using dynamic Voronoi dia grams Computers, environment and urban systems 2000, 24 (6):509 537. 30. Bullen N, Moon G, Jones K: Defining localities for health planning: a GIS approach Social Science & Medicine 1996, 42 (6):801 816. 31. Finding a Service Area In: ArcGIS 92 Desktop Hel p. ESRI. 32. Schuurman N, al. e: Defining rational hospital catchments for non urban areas based on travel time International Journal of Health Geographics 2006, 5 (1):1 11. 33. Luo W: Using a GIS based floating catchment method to assess areas with shorta ge of physicians Health & Place 2004, 10 (1):1 11. 34. McGrail MR, Humphreys JS: Measuring spatial accessibility to primary care in rural areas: improving the effectiveness of the two step floating catchment area method. Applied Geography 2009, 29 (4):533 5 41. 35. Hewko J, Smoyer Tomic KE, Hodgson MJ: Measuring neighbourhood spatial accessibility: Does aggregation error matter Environment and Planning A 2002, 34 :1185 1206. 36. Ahmedabad City Census 2011 Data In: Indian Census 2011. 2011. 37. Ahmedabad Best City to Live In, Pune Close Second In: The Times of India. 2011. 38. Bhagyalaxmi A, Jain S, Kadri AM: Effectiveness of different models of DOTS providers under RNTCP in Ahmedabad City, Gujarat Indian Journal of Community Medicine 2010, 35 (4):495 497. 39 Prevention [ http://www.gsacsonline.org/prevention.php ] 40. Programme RNTC: Technical Guidelines for TB Control In: Central TB Division, Directorate General of Health Services. Nirman Bhavan; 199 7. 41. Project A: India In Online; 2010. 42. OpenStreetMap: OpenStreetMap In Online. 43. Haklay M: How good is volunteered geographical information? A comparative study of OpenStreetMap and Ordnance Survey datasets Environment and Planning B: Planning and Design 2010, 37 (4):682 703.

PAGE 80

80 44. Okabe A, Okunuki K, Shiode S: The SANET Toolbox: New methods for network spatial analysis Transactions in GIS 2006, 10 (4):535 550. 45. Somani A: Study of Slums and Incidence of Diseases in Wards of Ahmedabad In: IDRC TTI. IRMA Anand; 2012: 1 19. 46. Hill RA, Manikal VM, Riska PF: Effectiveness of directly observed therapy (DOT) for tuberculosis: a review of multinational experience reported in 1990 2000 Medicine 2002, 81 (3):179 193. 47. Walley J, Khan M, Newell J, Kha n M: Effectiveness of the direct observation component of DOTS for tuberculosis: A randomized controlled trial in Pakistan Lancet 2001, 357 :664 669. 48. Inaba MK, Naoki; and Imai, Hiroshi: Applications of Weighted Vor onoi Diagrams and Randomization to Var iance Based k Clustering Computational Geometry 1994, 94 6 (94):332 339. 49. Campbell EBPJL: Measuring access to primary medical care: some examples of the use of geographical information systems Health & Place, 1998, 4 (4):183 193. 50. Guagliardo MF: Spat ial accessibility of primary care: concepts, methods andchallenges International Journal of Health Geographics 2004, 4 (4):1 13.

PAGE 81

81 BIOGRAPHICAL SKETCH Nirav Patel was a M.S. student in Geography at the University of Florida. He was born in Tampa, attendin g elementary, middle and high school in the city before coming up to Gainesville for college. His interest in geography first started off with research in neurosurgery at the University of South Florida when he was in high school, progressed into an intere st in public health policy in his senior year of high school and finally culminated with medical geography at the University of Florida. After graduating with B.A. degrees in geography and philosophy from UF in 2011, he took an interest in conducting his r esearch in Ahmedabad, India, and his adviser Dr. Liang Mao guided him to become a public policymaker within the city! His parents grew up in the city, and his grandparents started a social work and philanthropic organization in the city called Akhand Jyot Foundation 32 years ago, giving some much needed support for his research. Nirav worked with the Indian government in policymaking for HIV and Tuberculosis prevention in metropolitan areas. He had been working with Dr. Dixit Kapadia of Civil Hospital in Ah medabad to develop disease prevention strategies in metropolitan areas by identifying areas that lack adequate access to health service AsiaPop Project, aiming to provide high resolution After he graduated in spring 2013, he hoped to continue working with the AsiaPop project and was looking for a full time job in the Geospatial Technology industry.