Pakistan Research Repository

A STATISTICAL STUDY OF THE RISK FACTORS FOR URINARY BLADDER CANCER IN PAKISTAN

MUHAMMAD RIAZ AHMAD 09-GCU-Ph.D-STAT-2004

DEPARTMENT OF STATISTICS GC UNIVERSITY, LAHORE (PAKISTAN) iv

A thesis titled

A STATISTICAL STUDY OF THE RISK FACTORS FOR URINARY BLADDER CANCER IN PAKISTAN

Submitted to the GC University, Lahore For the award of degree of

PhD in STATISTICS

MUHAMMAD RIAZ AHMAD 09-GCU-PhD-STAT-2004

DEPARTMENT OF STATISTICS

GC UNIVERSITY, LAHORE (PAKISTAN) i

DECLARATION

I, Muhammad Riaz Ahmad, Registration No. 09-GCU-Ph.D-STAT-2004 student of Ph.D in the subject of Statistics, Session 2008-2011 is hereby declared that the matter printed in the thesis “A Statistical Study of the Risk Factors for

Urinary Bladder Cancer in Pakistan” is my own work and has not been printed, published and submitted as research work, thesis or publication in any form, in any university, research institution etc., in Pakistan or abroad by someone else.

Dated: ------

Muhammad Riaz Ahmad

RESEARCH COMPLETION CERTIFICATE

Certified that the research work contained in this thesis titled “A Statistical Study of the Risk Factors for Urinary Bladder Cancer in Pakistan” has been carried out and completed by Muhammad Riaz Ahmad, Registration No. 09-GCU-Ph.D-STAT-2004 for the completion of Ph.D degree.

Supervisor

Prof. Dr. Muhammad Khalid Pervaiz Dean, Faculty of Arts and Social Sciences Chairperson, Department of Statistics, G.C. University, Lahore.

Dated: ------

Submitted through

Prof. Dr. Muhammad Khalid Pervaiz Dean, Faculty of Arts and Social Sciences Chairperson, Department of Statistics, G.C. University, Lahore.

Controller of Examinations G.C. University, Lahore

Acknowledgement

I am very grateful to Almighty Allah, the most beneficial and merciful, who gave me courage, potential, energy, patience and spirit of hard work to complete this dissertation, which I could not complete successfully without his blessing.

A tribute to His beloved Holy Prophet (Peace Be Upon Him), being a follower to find new things and explore the nature.

It is a great pleasure and honor to express my feelings, gratitude’s and thanks to Prof. Dr. Muhammad Khalid Pervaiz Dean, Faculty of Arts & Social Sciences, Chairperson, Department of Statistics G.C. University, Lahore for his kind and sincere efforts to complete this research work. It would have been difficult for me to complete this research without his guidance, encouragement and sharing of knowledge.

The role of HEC towards higher education in Pakistan is remarkable and can never be ignored. I would never be able to complete my higher education (M.Phil and PhD) without the support of Higher Education Commission. In this connection, the guidance, assistance and services provided by the Miss Madiha Anwar Research Associate HEC during the whole study period of PhD are appreciable.

I am thankful to Principal Rana Abdul Qayyum, Prof. Muzaffer Ali Zia, Prof. Dr. Muhammad Naeem, Prof. Ikhlaq Hussain Shah of Govt. College, Jaranwala, Dr. Zahid Ahmad, Govt. College, Ravi Road, Lahore, Dr. Ahmad Faisal Siddiqi, University of Management &Technology, Lahore, Muhammad Ibrahim, Govt. Dyal Singh College, Lahore and Bashir Ahmad (Principal, Govt. M. D College, 41-JB, Faisalabad) for their co-operation and encouragement during my studies.

Furthermore, I can not ignore the cooperation of my class fellows, Mr. Abbdul Quyyum, Mr. Muhammad Hamid Mehmood, Ms. Asifa Kamal and Ms. Asma Zeb, Muhammad Ghias, Ms. Uzma Hafeez and Ms. Maryam Siddiqa during my studies. The services and cooperation of all the heads and doctors of the hospitals from which I collected data related to my study were also remarkable and I am highly obliged to all.

I am unable to pay thanks for the kindness and loving attitude of my mother who prays day and night for my success. The completion of this research work is not possible without her prayers and encouragement. I do express good feeling and sincere gratitude to my wife whose accommodating behaviour made it possible for me to complete the research and my special thanks to my children for their support, understanding and patience towards my long working hours and for there less demanding behavior during the course of this study. In the last but not the least my acknowledgements are for all my fellows and well-wishers who have encouraged me.

MUHAMMAD RIAZ AHMAD

iii

ABSTRACT

This study investigated the effect of different risk factors in the occurrence of urinary bladder cancer in Pakistan on the basis of a case control study using both descriptive and analytical approaches. A sample of 900 subjects including 300 cases and 600 controls was selected from different areas of Pakistan including headquarter of all four provinces and federal area (Islamabad) through a questionnaire. The requisite information was obtained from all the patients/ controls by the researcher using the direct interview method. From the headquarters of Khyber PukhtoonKhwa, Sindh, Baluchistan and federal area (Islamabad), 150 subjects (including cases and controls) were taken from wards of two selected public hospitals but from the headquarter of the Punjab (Lahore), 300 subjects (including cases and controls) were taken from the wards of four selected public hospitals. Controls are taken by matching the gender, area of residence and age above 40 years. About 22 factors with sub categories were included in the study. For bivariate analysis, the chi-square, phi/v statistics and Kandall’s tau-b are used. For the purpose of multivariate analysis, the binary logistic regression was run by using the SPSS (version-16) to observe the significant risk factors and prediction of the model. In the descriptive analysis, it was observed that risk of bladder cancer increases with an increase in the number of cigarettes smoked per day, years of smoking and risk decreases when the stop smoking period increases. Further more, similar results were observed in the bivariate analysis. In the overall analysis, the six factors including hair dye, chemical exposure, family history, cigarette smoking, fried items and fats items are found to be positively significant with the odds ratios and 95 % confidence intervals of odds ratios (2.96; 1.396-6.279), (2.59; 1.460-4.607), (3.13; 1.325-7.394), (10.6; 7.007-15.941), (2.11; 1.364-3.269) and (2.08; 1.309-3.305), respectively. While the three factors including lifestyle, fluid consumption and use of fruits are found to be negatively significant with odds ratios and 95% confidence intervals for the odds ratios (0.102; 0.056-0.187), (0.268; 0.183-0.392) and (0.292; 0.193-0.440), respectively indicating that these three factors are protective factors against urinary bladder cancer. In area wise study, eight factors age, social status, lifestyle, family history, cigarette smoking, tea, fluid consumptions and

iv fruits in Punjab, three factors cigarette smoking, source of drinking water and fried items in Islamabad, six factors chemical exposure, lifestyle, cigarette smoking, fluid consumption, fried items and fruit in Khyber Pukhtoon Khwa, three factors cigarette smoking, fluid consumption and fruits in Baluchistan and two factors cigarette smoking and fluid consumption in Sindh are found to be significant. In eight factors of Punjab, age, family history, cigarette smoking and tea are found to be positively significant while the other four factors social status, lifestyle, fluid consumptions and fruits are negatively associated with the bladder cancer. In three factors of Islamabad, two factors cigarette smoking and fried items are observed to be positively significant while the source of drinking water (government provided water) is observed to be protective as compared to the tap water. In six factors of Khyber Pukhtoon Khwa, chemical exposure, cigarette smoking and fried items are directly associated with the risk of bladder cancer while the other three factors lifestyle, fluid consumption and fruit are the protective factors for the disease. In three factors of Baluchistan, cigarette smoking is found to be positively significant while the fluid consumption and fruits are inversely associated with risk of bladder cancer. In two factors of Sindh, cigarette smoking is directly associated with disease while the other fluid consumption is found to be negatively significant. Cigarette smoking is the major risk factor and found to be significant in each area of Pakistan. Fluid consumption is also major protective factor and found to be significant in all areas except Islamabad. In studying the occupational risk factors, four categories of the occupations including cooks, drivers, metal workers and textile workers are found to be significant with the odds ratios and the 95% confidence intervals (14.132; 4.068 - 49.088), (7.949; 3.321- 19.025), (7.571; 3.147 - 18.214) and (2.168; 1.136 - 4.138), respectively. While the farmers, painters and leather workers are observed to be insignificant in Pakistan. According to this study, the cooks are at higher risk of bladder cancer as compared to all other occupations.

Key Terms: Bladder cancer, Risk Factors, Logistic Regression, Odds ratio, Controls, Significance, Retrospective

CONTENTS Acknowledgement i Abstract iii Abbreviations ix

List of Tables xi List of Figures xii

CHAPTER 1 INTRODUCTION 1 1.1 Medical Terms 1 1.2 Statistical /General Terminology 5 1.3 Cancer 6 1.3.1 Direct spread 7 1.3.2 Spread through blood 7 1.3.3 Spread through the lymphatic system 8 1.3.4 Epidemiology of Cancer 8

1.4 Urinary Bladder Cancer 9

1.4.1 Main Types of Urinary Bladder Cancer 9 1.4.2 Symptoms of Bladder Cancer 9 1.4.3 Stages of Urinary Bladder Cancer 10 1.4.4 Epidemiology of Urinary Bladder Cancer 11 1.5 Introduction of Pakistan 14 1.5.1 Punjab 16 1.5.2 Khyber Pukhtoon Khwa 17 1.5.3 Baluchistan 18 1.5.4 Sindh 20 1.5.5 Islamabad 21 1.6 Objectives of the Study 22 1.7 Significance of the Study 22

1.8 Study Plan 22

CHAPTER 2 LITERATURE REVIEW 24 2.1 Critical Review of Past Literature 24 2.2 Review of Methodology 45

CHAPTER 3 THEORETICAL FRAME WORK 47 3.1 Study Design 47 3.2 Inclusion Criteria 47 3.3 Sampling Technique 48 3.4 Instrument of the Study 49 3.5 Area Understudy 49 3.6 Pre-Test 50 3.7 Data Collection and Analysis 50 3.8 Data Analysis 51 3.9 Description and Coding for the Factors under Study 51 3.10 Types of variables 54 3.11 Description of Variables 55 3.12 Statistical Tests for Measuring the Association 64

3.12.1 Contingency Table 64

3.12.2 Chi-Square Test for Independence 64 3.12.3 Phi-Coefficient 65 3.12.4 Cramer’s V Statistic 65 3.12.5 Kendall’s Tau b 66 3.13 Odds, Odds Ratio and Confidence Interval 66

3.14 Relative Risk 69

vii

3.15 P-Value 69 3.16 Modeling Strategy 70

3.17 Fitting the Logistic Regression Model 72

3.18 Tests of Significance 74 3.19 Goodness of Fit Measures 76

3.19.1 Pearson test 76

2 3.19.2 Cox & Snell' R cs 77

3.19.3 Nagelkerke’s R2 77

3.20 Verification of the Model Assumptions 77

3.20.1 Detection of Autocorrelation 78

3.20.2 Detection of Multicollinearity 78

3.21 Model Diagnostics 79

3.21.1 Outliers 79

3.21.2 Influential Observations 79

3.21.3 Hosmer-Lemeshow goodness-of-fit Test 80 3.21.4 Omnibus Test 80

CHAPTER 4 RESULTS & DISCUSSION 81

4.1 Descriptive Analysis 81 4.2 Bivariate Analysis 99 4.3 Model Building 102 4.3.1 Detection of Multicollinearity 102 4.3.2 Detection of Autocorrelation 103 4.4 Diagnostics of the Model 103 4.4.1 Influential observations 103 4.4.2 Outliers 104

viii

4.5 Multiple Logistic Regression Model 107 4.6 Interpretations of the model coefficients and odds ratios 108 4.7 Occupational Risk Factors 115 4.7.1 Descriptive Analysis 115 4.7.2 Multiple Logistic Regression Model 117 4.8 Area-wise Descriptive Study 120 4.10 Area-wise Analytical Study 134 4.11 Goodness of Fit of the Models 142 4.12 Summary and Conclusions 146 4.12.1 Summary 146 4.12.2 Conclusions 148 4.13 Recommendations to Society 150 4.14 Recommendations for further Research 151 4.15 Limitations of collecting data 152

References 153 Bibliography 167 Questionnaire 169 Summary Table of Literature Review 171

Abbreviations

Terms Abbreviations Bladder Cancer BC Centre for Nuclear Medicine and Radiotherapy CENAR Chemical Exposure CE Chemotherapy CHEMO Chew Pan CP Chronic Bladder Irritation CBI Cigarette Smoking CS Confidence Interval CI Defect in Bladder by Birth DBB Diabetes DIAB Family History of Cancer FHC Fast Food FF Fats Items FAT Fluid Consumption FC Fried Items FRIED Gender GEND Hair Dye HD Hepatitis HEPA Huqqa Smoking HS Industrial Area IA Institute of Nuclear Medicine & Oncology INMOL Lahore Karachi Cancer Registry KCR Karachi Institute of Radiotherapy and Nuclear KIRAN Medicine Larkana Institute of Nuclear Medicine and LINAR Radiotherapy

Lifestyle LS Marital Status MARI No. of Cigarettes Smoked per day NCSP Nuclear Medicine, Oncology & Radiotherapy NORI Institute Nuclear Institute of Medicine and Radiotherapy NIMRA Number of Coffee Cups CCUP Number of Tea Cups TCUP Number of Years of Cigarette Smoking NYCS Number of Years Stop Smoking NYSS Odds Ratio OR Pakistan Institute of Medical Sciences, Islamabad PIMS Personal History of Cancer PHC Profuse, Periodic and Painless Hematuria PPP Radiation Therapy RADI Relative Risk RR Rural Area RA Sindh Institute of Urology and Transplant SIUT Social Status SS Source of Drinking Water SDW Squamous Cell Carcinoma SCC Sum of Squares of Errors SSE Transitional Cell Carcinoma TCC

Tables Table Page Title No. No. 3.1 Hospitals of Different Headquarters Visited for Data Collection 50

3.2 Coding Scheme for the Factors 51

4.1 City and Gender wise Classification of Data 82

4.2 Classification of Cases/ Controls with different risk factors 94

4.3 Counts and Percentages of the Number of Cigarettes Smoked Per day 97

4.4 Counts and Percentages of Cigarette Smoking Period 98

4.5 Counts and Percentages of Stop Smoking Period 99

4.6 Chi-Square, Phi/ V and Kendall's Tau-b Statistics 101

4.7 Tolerance and Variance Inflation Factors for the Detection of Multicollinearity 102

4.8 Correct and Incorrect Classification of the Data 108

4.9 Model Coefficient, Odds Ratios and 95% CIs of Odds Ratios 114

4.10 Classification of Subjects in Different Occupations with Percentages 116

4.11 Model Coefficients, Odds Ratios and their 95% CIs of Different Occupations 119

4.12 Area Wise Classification of Different Risk Factors with Categories 131

4.13 Omnibus and HL Tests for the Overall Significance of the Model 134

4.14 Area Wise Models Coefficient, Odds Ratios and 95% CIs for Odds Ratios 141

4.15 Logit Models for Different Areas of Pakistan 142

4.16 Cox & Snell R Square and Nagelkerke R Square for all Five Models 143

4.17 Correct Classification for Federal Area and Four Provinces 145

xii

Figures Figure Page Title No. No.

1.1 Different Stages of the Urinary Bladder Cancer 10

3.1 Huqqa which is smoked in Pakistan 59

4.1 Graph of Influential Values in the Model 104

4.2 Graphs of Standardized Residuals against ID 105

4.3 Graphs of Standardized Residuals against Predicted Probabilities 105

4.4 Graphs of Deviance Residuals against ID 106

4.5 Graphs of Deviance Residuals against Predicted Probabilities 106

Chapter No. 1 INTRODUCTION

Cancer is the second cause of death in the developed countries after the heart diseases. In the developing countries, it is the third cause of death after the heart diseases and the diarrhoeal diseases (American Cancer Society, 2007). Therefore, effective planning is required to reduce the threats of this disease. This chapter includes the introduction of cancer, ways of its spread, epidemiology, urinary bladder cancer, its stages, main types, symptoms and epidemiology. Furthermore, the reasons of conducting the study, its objective and significance have also been explained. Before explaining the concept of cancer and urinary bladder cancer, it becomes necessary to explain the important terminologies which are being used in this study. Therefore, the related important terminologies used are presented below.

1.1 Medical Terms (Concise Medical Dictionary, 2002)

 Adenocarcinoma is a malignant epithelial tumour arising from glandular structures, which are constituent parts of most organs of the body. The term is also applied to tumours showing a glandular growth pattern.

 Arsenic is a poisonous greyish metallic element which produces the symptoms of nausea, vomiting, diarrhea, cramps and coma when its large amount is consumed.

 Artery is a blood vessel carrying blood away from the heart. All arteries except the pulmonary artery carry oxygenated blood.

 Benign tumor is a tumour that does not invade and destroy the tissue in which it originates or spread to distant sites in the body.

 Calculosis means the presence of multiple calculi (stones) in the body. The destruction of calculi (stones) by the application of shock waves is called Lithotripsy.

 Calculus (plural of calculi) represents a stone, a hard pebble-like mass formed within the body, particularly in the gall bladder or anywhere in the urinary tract. Calculi in the urinary tract are commonly composed of calcium oxalate and are usually visible on X-ray examination.

 Capillary is an extremely narrow blood vessel, approximately 5-20 µm in diameter.

 Carcinogen means any substance that, when exposed to living tissue, may cause the production of cancer. Known carcinogens include ionizing radiation and many chemicals, e.g. those found in cigarette smoke and those produced in certain industries. Carcinogenesis is a process by which normal cells are transformed into cancer cells.

 Cell is the basic unit of all living organisms, which can reproduce itself exactly. Each cell is bounded by a cell membrane of lipids and protein, which controls the passage of substances into and out of the cell. Cells contain cytoplasm, in which are suspended a nucleus and other structures specialized to carry out particular activities in the cell.

 Chemotherapy means the treatment of disease by the use of chemical substances. The term is increasingly restricted to the treatment of cancer with antimetabolites and similar drugs, but is also still sometimes used for antibiotic and other treatment of infectious diseases.

 Chronic means a disease of long duration involving very slow changes. Such disease is often of gradual onset. Acute means a disease of rapid onset, severe symptoms, and brief duration.

 Communicable, contagious or infectious diseases are those which can be transmitted from one person to another. This may occur by direct physical contact.

 Cyclophosphamide (SY-kloh-FOS-fuh-mide) is a drug that is used to treat many types of cancer and is being studied in the treatment of other types of cancer. It is also used to treat some types of kidney disease in children. Cyclophosphamide attaches to DNA in cells and may kill cancer cells.

 Diagnosis means the process of determining the nature of a disorder by considering the patient's sign and symptom, medical background, and results of laboratory tests and X-ray examinations.

 Endemic means occurring frequently in a particular region or population: applied to diseases that are generally or constantly found among people in a particular area.

 Epidemic means an unexpected occurrence of infectious disease that spreads very fastly and affecting a large percentage of people. The influenza is the very common epidemics now a day.

 Epidemiology is the study of the occurrence, distribution, and control of infectious and noninfectious diseases in populations, which is a basic part of public health medicine. Originally restricted to the study of epidemic infectious diseases, such as smallpox and cholera, it now covers all forms of disease that relate to the environment and ways of life. It thus includes the study of the links between smoking and cancer, diet and coronary disease, as well as communicable disease.

 Epithelium is the tissue that covers the external surface of the body and lines hollow structures (except blood and lymphatic vessels).

 Incidence rate represents a measure of morbidity based on the number of new diagnosed cases of a disease arising in a population over period of time. It is described in terms of patients of some specified disease per 1000 individuals at risk.

 Lymph node is one of a number of small swellings found at intervals along the lymphatic system. Lymph is the fluid present within the vessels of the lymphatic system.

 Lymphatic system is a network of vessels that conveys electrolytes, water, proteins, etc., in the form of lymph, from the tissue fluids to the bloodstream.

 Malignant tumour is a tumour that invades and destroys the tissue in which it originates and can spread to other sites in the body via the blood-stream and lymphatic system.

 Prevalence rate is a measure of morbidity based on current patients in a population, estimated either at a particular time.

 Radiotherapy means a treatment of disease with penetrating radiation, such as X- rays, beta rays, or gamma rays, which may be produced by machines.

 Renal pelvis any structure shaped like a basin, e.g. the expanded part of the ureter in the kidney.

 Risk factor means an attribute, such as a habit (e.g. cigarette smoking) or exposure to some environmental hazard, that leads the individual concerned to have a greater likelihood of developing an illness.

 Sign means an indication of a particular disorder that is observed by a physician but is not apparent to the patient.

 Symptom means an indication of a disease or disorder noticed by the patient himself. A presenting symptom is one that leads a patient to consult a doctor.

 Transitional cell carcinoma is a form of cancer that affects the urothelium, which lines the urinary collecting system of the kidney, ureters, bladder, and most of the urethra. It is the most common type of bladder cancer.

 Tumour is any abnormal swelling in or on a part of the body. The study and practice of treating tumours is called oncology.

 Urethra is the tube that conducts urine from the bladder to the exterior. The female urethra is quite short (about 3.5 cm) and opens just within the vulva, between the clitoris and vagina. The male urethra is longer (about 20 cm) and runs through the penis.

 Urinary catheter is a tube that is passed into the bladder through the urethra to allow drainage of urine in certain disorders and to empty the bladder before abdominal operations.

 Valve is a structure found in some tubular organs or parts that restricts the flow of fluid within them to one direction only. Valves are important structures in the heart, veins, and lymphatic vessels.

 Veins are a blood vessel conveying blood towards the heart. All veins except the pulmonary vein carry deoxygenated blood. Vessel is a tube conveying a body fluid, especially a blood vessel or a lymphatic vessel.

1.2 Statistical /General Terminology

 Black Tobacco (air-cured) is a type of tobacco whose leaves are dried by hanging in sheds which are open from all sides and naturally ventilated. It is mostly used in cigars.

 Blond Tobacco (Flue-cured) is hung up in sheds which have a series of tubes through which hot air or steam is blown. The heat given off by the tubes accelerates the drying of tobacco. This system takes its name from the tubes which are known as flues. Flue-cured tobacco is largely used for cigarettes.

 Case-Control Study is an analytical study which compares individuals who have a specific disease (case) with a group of individuals free from the disease

(control). Case control study is often called a retrospective study that takes a group of individuals and studies the characteristics of the disease backward in time (Dawson and Trapp, 2004). These studies are less costly as compared to the cohort studies and required very small span of time.

 Cohort Study is an observational study, takes a group of individuals and usually follows them forward in time called prospective studies. These studies are very costly and required a long period of time. A researcher selects subjects at the beginning of the study and then determines whether they have the risk factors or have been exposed (Petrie and Sobin, 2005).

 Cross-Sectional Study is a study which is conducted in a single point of time. The study based on a group of patients at single point of time is referred as survey, epidemiological study or prevalence study (Dawson and Trapp, 2004).

 Observational Study is a scientific investigation in which neither the subjects under study nor any of the variables of interest are manipulated in any way (Daniel, 2000).

 Pack year represents a way to measure the amount of cigarettes a person has smoked over a long period of time. It is calculated by multiplying the number of packs of cigarettes smoked per day by the number of years the person has smoked. For example, 1 pack year is equal to smoking 1 pack per day for 1 year, or 2 packs per day for half year, and so on.

Before going to explain the urinary bladder cancer, it is necessary to describe the concept of cancer, its ways of spread in the other organs of the body and epidemiology. Therefore, these concepts are given as below in section 1.3.

1.3 Cancer Cancer is a class of diseases that occurs from uncontrolled division of cells that make up the body's tissues and organs. When growth control is lost and the cells divide

7 large in numbers very fastly, then a cellular mass or "tumor" is formed. ‘A tumor is an abnormal mass of tissues, the growth of which exceeds and is uncoordinated with that of the normal tissues, and persists in the same excessive manner after cessation of stimuli which evoked the change’ (Willis, 1952).

There are two types of tumors, benign and malignant. Malignant tumors are commonly called cancerous. If the tumor has ability to invade the other organs of the body then the tumor is called malignant or cancerous otherwise benign. If the benign tumors are removed, they have no chance of comeback. The cancerous cells are not confined to localized 'overgrowth' and access of surrounding tissue, but can spread to the other parts of the body through the lymphatic system and blood stream, creating secondary deposits is called the metastasis. Walter (1977) and Well (2001)

According to Dollinger and Rosenbaum (1991), the cancer can spread from one organ to the other organs of the human body in three ways (direct spread, through blood and through the lymphatic system) which are discussed in the following sub-sections:

1.3.1 Direct spread As the malignant grows in one organ, it invades surrounding tissues very fastly. For example, in case of man, the bladder cancer in the fourth stage invades the prostate while in case of women, it invades the vigina. Dollinger and Rosenbaum (1991)

1.3.2 Spread through blood Arteries supply blood to the cells and veins take it back. Pieces of tumor can grow through the walls of these vessels and enter in the blood streams. They circulate through the various parts of the body and settle and grow in various organs. Only a few of the circulating cells can find a growing place. The cancerous cells can move through the smaller blood vessels (called the capillaries) and grow in other parts of the body (Walter, 1977). The cancers associated to blood borne spread include melanoma and small cell carcinoma of the lung. Yarbro, Frogge and Goodman (2005)

1.3.3 Spread through the lymphatic system A lymphatic system is a system consisting of minute (very small) lymphatic vessels. Lymph is a liquid which is carried out by the lymphatic vessels. Tumor cells can move from one organ to another through the lymph. In other words, tumor cells can gain access to the lymphatic system and travel along the vessels to the 'regional draining' lymph nodes. Cancerous cells travel from the original tumor, through the lymphatic system, to the next group of lymph nodes and spread the malignancy throughout the body of the patient (Walter, 1977).

1.3.4 Epidemiology of Cancer Like other developing countries, Pakistan is also facing the double burden of the diseases with the significant incidents of cancers and an increasing trend in risk factors’ profile (Ferlay et al, 2004 and Bhurgri et al, 2000). The precise number of new cancer cases and number of deaths annually for Pakistan are unknown. However, WHO estimates provide the figures 61,624 incident cases and 42,684 cancer deaths annually in males and 75,095 incident cases and 43,188 deaths annually in females (Ferlay et al, 2004). Age Standardized Incidence Rate (ASIR) of all cancers combined in Karachi South, a representative population of Pakistan, is 204 per 100,000 among females and 179 per 100,000 among males according to Karachi Cancer Registry (KCR) estimates (Bhurgri, 2004).

In 2007, the estimated cancer cases in all over the world were more than 12 million of which 5.4 million were in economically developed countries and 6.7 million in developing countries. The estimates of total deaths due to cancer in all over the world were 6.7 million (about 20,000 cancer deaths per day) of which in developed countries were 2.9 million and 4.7 million in developing countries (American Cancer Society, 2007).

After explaining the concept of the cancer, different ways of its spreading to the other parts of the human body and its epidemiology, then it becomes easy to understand about the growth of the urinary bladder cancer.

1.4 The Urinary Bladder Cancer Bladder Cancer originates in the inner most layers of the bladder, called mucosa. After invading the outer most layers called serosa, the process of metastasis (a situation in which the cancerous cells break away from the original tumor, travel, and grow within other parts of the body) continues. Tanagho and McAninch (2008)

A bladder is an organ in the lower abdomen that stores urine, the liquid waste produced by the kidneys. Urine passes from each kidney into the bladder through a tube called ureters. Each ureter is a small tube, about 25 cm long that carries urine from the kidneys to the urinary bladder. When the bladder is full, the muscles in the bladder wall can tighten to allow urination. Urine is descended out from the bladder through another tube, the urethra. The bladder is consisted of four layers that are mucosa, sub mucosa, muscles and serosa. The outer layer of the bladder is called peritoneum. Tanagho and McAninch (2008)

1.4.1 Main Types of Urinary Bladder Cancer The most common types of bladder cancer are Transitional Cell Carcinoma (TCC), Squamous Cell Carcinoma (SCC) and Adenocarcinoma. About 90% of all bladder cancer is TCC that generates in the inner lining of the bladder or epithelium layer. About 5% to 10% is the SCC, which is often associated with a history of chronic infection, vesical calculi, or chronic catheter use and adenocarcinoma is about 2% of all bladder cancer. Tanagho and McAninch (2008)

1.4.2 Symptoms of Bladder Cancer The main symptom of bladder cancer is the painless hematuria that occurs in about 85 % of patients (Varkarakis et al, 1974). In smaller percentage of patients, these symptoms are accompanied by the symptoms of vesical irritability, frequency, urgency and dysuria. It is commonly described by three p's called Profuse, Periodic and Painless hematuria (PPP). (Tanagho and McAninch, 2008)

1.4.3 Stages of Urinary Bladder Cancer The different stages of the urinary bladder cancer are shown by the picture given below. Tanagho and McAninch (2008)

Figure 1.1 Different Stages of the Urinary Bladder Cancer

1.4.4 Epidemiology of Urinary Bladder Cancer Bladder Cancer is the nineth most common malignant disease in the world with about 250,000 new cancer cases each year (Parkin, 1999). The highest incidence rates are generally found in industrially developed countries, particularly in North America and Western Europe. Bladder cancer is more common in men as compared to women, with a worldwide male/female ratio of 10:3 (Sylvester, 2004). In USA, the urinary bladder cancer is the fourth most common malignancy in men and 8th in women (Jemal et al, 2003). Males are mostly affected by bladder cancer, with a male/female ratio of 3:1, showing sex-linked etiological reasons (Rabbani and Cardo, 2000).

In 2005, the overall new bladder cancer cases were 330,000 and deaths were 130,000 estimated in the world and only in the United States, the new bladder cancer cases and deaths were over 63,000 and 13,000, respectively (Jemal et al, 2005). In 2009, there were estimated 71,000 new cases of bladder cancer diagnosed in the US (53,000 in men and 18,000 in women), with approximately 14,000 deaths (Jemal et al, 2009). But in developing countries, the research on such fatal diseases is yet not targeted. Every year, a large number of people die due to this terrible and aggressive disease.

Bladder cancer is considered a malignancy with a long latency period, the time between disease initiation and disease detection. This observation arose from that fact that most new cases of bladder cancer are found in those over 65 years of age. The average latent period for the development of bladder cancer has been estimated to be more than 15 years (Meijden, 1998) and has been speculated to range from 18 years to 44 years (Mantanoski and Elliot, 1981; Cohen and Johansson, 1992). Urinary Bladder cancer is a disease of older and rarely diagnosed before the age of 40. The latency period from initial exposure to the development of a urothelial tumor is a median of 18 years. The diagnosis rate of urinary bladder cancer is lower in rural than in urban areas (Devita, Hellman and Rosenberg, 2001).

Some studies investigated that the coffee drinking is an etiology of the urinary bladder cancer (Morrison, 1984; Ciccone and Vineis, 1988; Slattery et al, 1988). A

positive association was found in tea drinkers and urinary bladder cancer (Slattery et al, 1988). Most of the case control studies did not indicate a significant association between alcohol or coffee consumption and the risk of bladder cancer (Viscoli et al, 1993; Risch et al, 1988). High intake of coffee (more than 4 cups per day) was observed to increase bladder cancer risk (Villanueva et al, 2006). Consumption of tea was not established the relationship with the risk of bladder cancer (Zeegers et al, 2004). A poor inverse association was observed between tea consumption and urinary bladder cancer risk (Bianchi et al, 2000; Zeegers et al, 2001).

A high fluid intake is associated with a decreased incidence of bladder cancer in men, and lesser intake of daily fluids proportionally increases the risk of bladder cancer (Claus, 1996). The use of hair dyes was investigated a risk factor for urinary bladder cancer (Gago et al, 2001). But the meta-analysis of 10 published case control studies did not confirm it in which the combined relative risk and 95 % confidence interval for ever users of hair dyes was 1.01 and (0.89-1.14), respectively (Takkouche et al, 2005). High consumption of fats, especially animal fats, can increase the risk of urinary bladder cancer (Riboli et al, 1991; Hebert et al, 1994; Bruemmer et al, 1996; Radosavljevic´ et al, 2005). The risk of urinary bladder cancer was inversely associated with the high consumption of fruits and vegetables (Negri et al, 2001).

Cigarette smoking accounts for 50% and 31% of bladder cancers in men and women respectively (Wynder and Goldsmith, 1971). Cigarette smoking is a major risk factor that generates the urinary bladder cancer about 50% to 65% in males and 20% to 30% in females (Kogevinas and Trichopoulos, 2000; Silverman et al, 1996; Brennan et al, 2001). Cigarette smoking is one of the major reasons of bladder cancer. Cigarette smokers have up to four times higher incidence of urinary bladder cancer as compared to the non smokers (Burch et al, 1989). The risk is associated with the number of cigarettes smoked per day, the duration of smoking and the amount of inhalation of smoke. Contributing agents in the cigarette smoke are considered the alpha and beta naphthylamine, which are coming in urine of smokers (Viscoli et al, 1993). Comparing the number of cigarettes smoked per day, the higher risk of bladder cancer was

investigated in women than men (Castelao et al, 2001). Other forms of tobacco use are associated with only a slightly higher risk of bladder (Burch et al, 1989). In Pakistan 36% of men and 9% women are smokers (Alam, 1998).

Occupational cancer can be defined as cancers that are due to exposure to carcinogens in the workplace. About 50% of all recognized human carcinogens are occupational carcinogens (Siemiatycki et al, 2004) and therefore are modifiable risk factors. Cigarette smoking and certain occupational exposures, such as exposure to aromatic amines, are known modifiable risk factors for bladder cancer (Negri and Vecchia, 2001; Silverman et al, 2006; Meijden, 1998; Tola, 1980; Cohen et al, 2000; Borden et al, 2003; Adami et al, 2002).

Among Canadian males, the age-standardized incidence rate has been decreasing on average by 0.5 percent per year from 1995 to 2004 (Canadian Cancer Society, 2008). In the same time period, there was also a corresponding decrease among males in the age- standardized mortality rate which decreased an average of 0.4 percent. Among Canadian females, however, while the incidence rate has been decreasing over time at an annual average of 0.4 percent, the mortality rate from 1995 to 2004 has been increasing by an average of 0.4 percent annually (Canadian Cancer Society, 2008).

In the light of these facts, it is necessary to take steps for controlling such a fatal, life threatening and aggressive disease. The statisticians, oncologists and medical researchers should try to put their combined efforts for the detection and analyzing the causes of cancer and utilize the results for the betterment of mankind. In Pakistan, no published research is available that can explain the risk factors of the urinary bladder cancer. In prior studies, generally descriptive statistics are used for explaining the risk factors which are not sufficient and able to serve the purpose. The concept of model building must be used for investigating the risk factors of bladder cancer. Therefore, this study includes all possible risk factors about the disease and as well as explores the significant factors by using the bivariate analysis and by an appropriate model.

1.5 Introduction of Pakistan (Country under Study)

Since this study is conducted in the provinces and federal area of Pakistan, it is necessary to introduce the social, cultural and geographical aspects of the each province and federal area of the country. Furthermore, the presences of medical facilities in different areas are also discussed.

Pakistan is a country of south Asia which has got independence from the British India in 1947. India is located in the east, Afghanistan and Iran in the west, China in the far northeast and Tajikistan in the north of Pakistan. There is the Arabian Sea and Gulf of Oman are situated In the south of Pakistan. Pakistan is a federation of four provinces, a capital territory and a group of federally administered tribal areas. The state of Pakistan is comprised of the four provinces Punjab, Sindh, Baluchistan, Khyber Pukhtoon Khwa, a capital territory (Islamabad) and federally administered tribal areas. The total area of this country is 796,096 km2 and the estimated population in 2010 is 170,455,500. Pakistan is a multilingual country with more than sixty languages being spoken. Punjabi is the provincial language of Punjab. Saraiki is also spoken in the larger area of Punjab province. Pashto is the provincial language of Khyber Pukhtoon Khwa. Sindhi is the provincial language of Sindh and Balochi is the provincial language of Baluchistan.

In order to provide the facility of cancer treatment, thirteen cancer hospitals are established by the Atomic Energy Commission in different areas of Pakistan in which two are in Lahore, two Karachi and one at each city Faisalabad, Bahawalpur, Multan, Islamabad, Abbottabad, Larkana, Jamshoro, Quetta and Peshawar. Besides this, Urology departments of different hospitals are also treating the bladder cancer in different cities of the Pakistan. As cancer disease is considered the symbol of death, so the people tried to get the better facilities for their treatments.

For presenting the clear picture of the provinces and federal area (Islamabad Capital territory), the map of the country is provided here. According to map, Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are represented by the colours brown, dark blue, light yellow, light green and light blue, respectively.

Map of Pakistan

1.5.1 Punjab

Punjab is the most populous province of Pakistan having about 56% of the country's total population. The total estimated population in 2010 is 81,330,531. The literacy rate in 2008 is 79.9%. Punjab is known as the "Land of the Five Rivers” referring to the Indus River and its four primary tributaries of Jhelum, Chenab, Ravi, and Sutlej, that flow through Punjab. The capital and largest city of is Lahore and other important cities include Multan, Faisalabad, Jhang, Sialkot, Gujranwala, Rawalpindi and Jhelum. This province is a large multi-ethnic population strongly influence Punjab's outlook on National affairs and induces in Punjab a keen awareness of the problems of the Pakistan’s other important provinces and territories. In the acronym P-A-K-I-S-T-A-N, the P is for PUNJAB.

Punjab is Pakistan's second largest province at 205,344 km2 after Baluchistan. The province is a mainly a fertile region along the river valleys, while sparse desert can be found near the border with Rajasthan and the Sulaiman Range. The region contains the Thar and Cholistan and deserts. The Indus River and its many tributaries traverse the Punjab from north to south. The landscape is amongst the most heavily irrigated on earth and canals can be found throughout the province. Agriculture continues to be the largest sector of Punjab's economy. Weather extremes are notable from the hot and barren south to the cool hills of the north. Since the 1950s, Punjab industrialized rapidly. New factories came up in Lahore, Multan, Sialkot and Wah. In the 1960s the new city of Islamabad was built near Rawalpindi.

Punjab is home to the Punjabis and various other groups. The main languages spoken in Punjab are Urdu, English, Punjabi, Saraiki, Mewati, Pothowari, Hindko, Sindhi, Pashto and Balochi. Punjabis are divided into different Important quoms include the Gondal, Arain, Niazi, Paracha, Aheer, Awan, Dogar, Gakhars, Gujjars, Jat, Kamboh, Khokhar, Khattar, Mughal, Rajputs, Sheikh and Syeds. Other smaller quoms are the Khateek, Maliar, Rawns, Pashtuns, Baloch, Rehmanis and the Maliks. Punjab’s share of Pakistan's GDP has historically ranged from 51.8% to 54.7%. It is especially dominant in the Service & Agriculture sectors of the Pakistan Economy. It is also major manpower

contributor because it has largest pool of professionals and highly skilled (Technically trained) manpower in Pakistan.

Several hospitals are established in tehsil, district and division level in Punjab. But the hospitals having cancer treatments are very rare and not well equipped in the interior areas of the province. So, the people from all areas of Punjab tried to attend the hospitals of Lahore for the treatment of cancer. In Lahore, five public hospitals and one private (Shoukat Khanam) hospital are treating the cancer diseases with latest equipments.

1.5.2 Khyber Pukhtoon Khwa

Khyber Pukhtoon Khwa previously known as the North-West Frontier is one of the four provinces of Pakistan, located in the north west of the country. The province has an area of 74,521 km2. It borders Afghanistan to the north-west, Gilgit-Baltistan to the north-east, Azad Jammu and Kashmir to the east, the Federally Administered Tribal Areas (FATA) to the west and south, Baluchistan to the south and Punjab and the Islamabad Capital Territory to the south-east. The province's main districts are Bannu, Dera Ismail Khan, Kohat, Abbottabad, Haripur and Mansehra. Peshawar and Mardan are the main cities. The hilly terrain of Swat, Kalam, Upper Dir, Naran and Kaghan is renowned for its beauty and attracts a great many tourists from neighbouring regions and from around the world. Swat-Kalam is also termed 'a piece of Switzerland' as there are many landscape similarities between it and the mountainous terrain of Switzerland. The major rivers that criss-cross the province are Kabul River, Swat River, Chitral River, Panjgora River, Bara River, Kurram River, Gomal River and Zhob River. Its snow- capped peaks and lush green valleys of unusual beauty have enormous potential for tourism.

According to the 1998 census, the population of the province was approximately 17 million. Of whom 52% are males and 48% are females. The density of population is 187 per km² and the intercensal change of population is of about 30%. The literacy rate in 2008 is 49.9%. The main ethnic group in the province is the Pashtuns, locally referred to as Pakhtuns, followed by a number of smaller ethnic groups, most notably, the

Hindkowans and Chitralis. The principal language is Pashto, locally referred to as Pukhtoon and the provincial capital is Peshawar.

Khyber Pakhtun khwa's share of Pakistan's GDP has been 10.5%, although the province accounts for 11.9% of Pakistan's total population, rendering it the second poorest province after neighboring Baluchistan. Agriculture remains important and the main cash crops include wheat, maize, Tobacco (in Swabi), rice, sugar beets, as well as various fruits are grown in the province. Numerous workshops throughout the province support the manufacture of small arms and weapons of various types. The province accounts for at least 78% of the marble production in Pakistan.

In this province, the medical facilities of cancerous diseases are not sufficient because in interior areas, there is no cancer hospital. The treatment of cancer diseases is only available at headquarter (Peshawar) of the province. Only three public hospitals Lady Reading, Institute of kidney disease Hayatabad and IRNUM are treating the bladder cancer and other cancerous diseases. Due to this problem, all the patients of cancer diseases including bladder cancer have to visit one of these three stated hospitals.

1.5.3 Baluchistan

Khyber PukhtoonKhwa previously known as the North-West Frontier is one of the four provinces of Pakistan, located in the north west of the country. The province has an area of 74,521 km2. It borders Afghanistan to the north-west, Gilgit-Baltistan to the north-east, Azad Jammu and Kashmir to the east, the Federally Administered Tribal Areas (FATA) to the west and south, Baluchistan to the south and Punjab and the Islamabad Capital Territory to the south-east. The province's main districts are Bannu, Dera Ismail Khan, Kohat, Abbottabad, Haripur and Mansehra. Peshawar and Mardan are the main cities. The hilly terrain of Swat, Kalam, Upper Dir, Naran and Kaghan is renowned for its beauty and attracts a great many tourists from neighbouring regions and from around the world. Swat-Kalam is also termed 'a piece of Switzerland' as there are many landscape similarities between it and the mountainous terrain of Switzerland. The major rivers that criss-cross the province are Kabul River, Swat River, Chitral River, Panjgora River, Bara River, Kurram River, Gomal River and Zhob River. Its snow-

capped peaks and lush green valleys of unusual beauty have enormous potential for tourism.

Khyber Pakhtunkhwa's share of Pakistan's GDP has been 10.5%, although the province accounts for 11.9% of Pakistan's total population, rendering it the second poorest province after neighboring Baluchistan. Agriculture remains important and the main cash crops include wheat, maize, Tobacco (in Swabi), rice, sugar beets, as well as various fruits are grown in the province. Numerous workshops throughout the province support the manufacture of small arms and weapons of various types. The province accounts for at least 78% of the marble production in Pakistan.

1.5.4 Sindh

Sindh is one of the four provinces of Pakistan and historically is home to the Sindh people. It is also locally known as the "Mekran" and "Bab-ul-Islam". The Door to Islam, because Islam in South Asia was first introduced via Sindh. Different cultural and ethnic groups also reside in Sindh including Urdu-speaking Muslim refugees who

migrated to Pakistan from India upon independence as well as the people migrated from other provinces after independence.

The neighbouring regions of Sindh are Baluchistan to the west and north, Punjab to the north, Gujrat and Rajasthan to the southeast and east, and the Arabian Sea to the south. The main language is Sindhi. The name is derived from the Indus River that courses through it. Sindh is located on the western corner of South Asia, bordering the Iranian plateau in the west. Geographically it is the third largest province of Pakistan, with an area of 140,915 km2 of Pakistani territory. The 1998 Census of Pakistan indicated a population of 30.4 million, the current population in 2009 is 51,337,129. The literacy rates in 1972, 1981, 1998 and 2008 are 30.2%, 31.5%, 45.29% and 57.7%, respectively. The literacy rate is low in Sindh as compared to the Punjab.

Sindh is bounded by the Thar Desert to the east, the Kirthar Mountains to the west, and the Arabian Sea in the south. In the centre, fertile plains are around the Indus River. Sindh is situated in a Subtropical region; it is hot in the summer and cold in winter. The main cities of the province are Karachi, Hyderababad, Sukkur, Mirpurkhas, Nawabshah District, Umerkot and Larkana. According to the 2008 Pakistan Statistical Year Book, Sindhi-speaking households make up 65.7% of Sindh's population; Urdu- speaking households make up 20.1%; Punjabi 6.0%; Pashto 3.2%; Balochi 1.1%; Saraiki 1.0% and other languages 2.1%. Other languages include Gujrati, Memoni, Kutchi, Khowar, Thari, Persian and Brahui.

Sindh has the second largest economy in Pakistan. Historically, the Sindh contributes to Pakistan's GDP from 30% to 32.7% of the total GDP. Its share in the service sector has ranged from 21% to 27.8% and in the agriculture sector from 21.4% to 27.7%. Manufacturing includes machine products, cement, plastics, and various other goods. Agriculture is very important in Sindh with cotton, rice, wheat, sugar cane, bananas and mangoes as the most important crops. Sindh is the richest province in natural resources of gas, petrol, and coal.

In Sindh, more than 20 public hospitals are available in different areas of Sindh but only four public hospitals are treating the bladder cancers which are SIUT, Jinnah Hospital, KIRAN and Abbasi Shaheed Hospital at Karachi and two cancer hospitals NIMRA and LINAR are working at Jamshoro and Larkana, respectively. Rare patients visit the Abbasi Shaheed Hospital due to the lack of facility as compared to the Jinnah Postgraduate Medical Centre and KIRAN. From the interior areas of Sindh, mostly cancer patients prefer KIRAN or Jinnah Postgraduate Medical Centre Karachi for their treatment as compared to their local hospitals.

1.5.5 Islamabad

Islamabad is the capital of Pakistan which is located in the surroundings of the Margalla Hills at the northern end of Pothowar Plateau. It is the 10th largest city in Pakistan having population 1.21 million in 2009. The combined population of Rawalpindi and Islamabad is 4.5 million. The city was constructed during the 1960’s by the orders of the President General ‘Ayub Khan’. This city is the cleanest, well organized and divided into sectors and zones. The nickname of Islamabad is ‘The Green City’ because of its greenery. Faisal mosque is the sixth largest mosque in the world which is located in Islamabad. The literacy rate is very high (87.0%) in Islamabad. The total area of the federal city is 910 km2 and the density of population is 880/ km2. Healthy climate, pollution free atmosphere, plenty of water and lush green area are the main features of this city.

The medical facilities are much better in the federal city. This city has four major public hospital including Pakistan Institute of Medical Sciences (PIMS), NORI, Federal Government Services Hospital and Capital Hospital which are treating the bladder cancer and other diseases. In which the frequency of bladder cancer patients is higher only in PIMS and NORI. Most of the people of federal areas visit these hospitals for the treatment of bladder cancer. NORI is the cancer hospital which is one of the thirteen hospitals working under the Pakistan Atomic Energy Commission of Pakistan.

In the capital city, the literacy is much higher as compared to the other areas. Therefore, the higher literacy rate creates the sense of awareness in the public about the

risk factors and treatment of the cancerous diseases. The better lifestyle is not feasible without the better understanding and literacy. It creates the sense of cleanliness that leads to the healthy lifestyle.

1.6 Objectives of the Study The specific objectives of this study are: i. Descriptive and analytical analysis of urinary bladder cancer. ii. Assessing the association of each individual risk factor with the diseases. iii. Development of suitable statistical models to explain the strength (severity) of the risk factors of bladder cancer. iv. The determination of the occupational risk factors. v. Comparison of risk factors of the disease among different provinces in Pakistan.

1.7 SIGNIFICANCE OF THE STUDY This study will be beneficial for the doctors (especially oncologists/ urologists), public and the health policy makers of Pakistan. For the public, it will provide the awareness about the significance of the risk factors. The different policies can be introduced in different provinces keeping in view the risk factors of that area. In this way, this study will provide the basis for controlling or reducing the disease in the country. As a result, money spent on health care can be reduced.

1.8 STUDY PLAN Introduction of the study is presented in Chapter 1, which includes the concept of cancer, ways of its spread, epidemiology, urinary bladder cancer, its symptoms, stages, epidemiology, some important medical and other terminologies related to the study. Furthermore, the objectives of the study and its significance are also explained. The review of the literature is presented in Chapter 2, in which the literature from 1983 to- date, that is latest, has been reviewed by focusing on the risk factors of bladder cancer, not regarding some specific countries. Most of the research related to the bladder cancer has been done in Europe and United States as compared to the other areas of the world.

Theoretical Frame Work is given in Chapter 3, which includes the information about the sampling design, variables used in this study, coding scheme of variables and types of the observational studies. Furthermore, all the methods used in analysis especially; Chi- square, Phi/V Statistic, Kendall’s Tau b and binary logistic regression model with the relevant terminologies are discussed.

The analysis, interpretations, conclusions, discussions, recommendations to the society and to the researchers is made in the Chapter 4. In analysis, both the descriptive and analytical approaches are used to explain the severity of the risk factors. In descriptive approach, counts, percentages and averages are used to explain the characteristics of the risk factors while in analytical section, binary logistic models are run to found out the odds ratio and their confidence intervals to explain the strength of the significant risk factors. Further more, the association between the individual factors and the bladder cancer is also observed by using the Chi-square, Phi/V Statistic and Kendall’s Tau b statistic. At the end, references, bibliography, questionnaire and the summary table of literature review are placed.

Chapter No. 2 LITERATURE REVIEW

Every scientific research is based on some theories, concepts and findings of the other scientists or researchers. These findings may be based on the randomized or observational studies conducted in some specified area. In order to support this study, the risk factors and theories presented by different researchers in different areas including USA, UK, Turkey, France, Taiwan and among many other countries, are reviewed. The latest references starting from 1983 to 2009 are included chronologically in this chapter in order to explore the effect of different factors on the urinary bladder cancer and the methodology used in the binary response variable. The literature before this period has been referred and discussed in the literature which has been presented here in section 2.1. Furthermore, the methodology used in this dissertation has been discussed and reviewed in the section 2.2 of this chapter.

2.1 Critical Review of Past Literature A case control study was conducted by Mommsen et al (1983) in which 212 bladder cancer patients and 259 controls that were matched with respect to age and geographical area. Controls were consisted on 165 males and 94 females while the cases had 165 males and 47 females. Bivariate analysis was run to find the relative risk. This analysis indicated that cigarette smoking, pipe smoking, cigar smoking, chew tobacco, industrial worker, work with petroleum, use of alcohol, working with oil and chemicals materials had a significantly increased relative risk of developing bladder cancer. For multivariate analysis, binary logistic regression model was run. It was observed that cigarette and cigar smoking had the highest significant risks. Both the logistic regression and the bivariate analysis indicated that cigarette smokers and cigar smokers were significantly associated determinants of developing the urinary bladder cancer.

A case-control study was performed in the United States by Burns and Swanson (1991). The authors observed the association between the cigarette smoking, occupation,

industry worker and urinary bladder carcinoma. For this purpose, they had studied the 2,160 bladder cancer patients and 3,979 controls with respect to the history of occupation and use of tobacco. They compared ever cigarette smokers with that of never smokers and found that the ever smokers had significantly 2.4 times higher risk of bladder cancer than the never smokers. The higher risk of urinary bladder cancer was found due to cigarette smoking among Black males and females as compared to the White. The 7 times more risk of bladder cancer was observed in Black males who were mechanic by occupation. A significant excess of bladder cancer was found among armed services personnel; this excess was restricted to White males when the analysis was performed separately by race. This study was also supported by the study of Mommsen (1983) who used the bivariate and multivariate analysis and indicated that cigarette smokers have higher risk of developing the bladder cancer as compared to the non smokers. Furthermore, this analysis is stronger because the sample size is very large and the ratio of cases and control is about 1:2 as compared to the study of Mommsen (1983) who used the case control ratio as about 1:1. When more controls will be used, the discriminating power of the logistic regression will be increased.

Kunze et al (1993) conducted a case-control study taking 531 male and 144 female matched pairs in order to observe the association of occupational and non- occupational risk factors of the urinary bladder cancer. This article was translated from the German Language. Cigarette smoking and heavy pipe smoking were of the significant factors with an increased risk in both males and females like the Burns and Swanson (1991) and Mommsen et al (1983) who had indicated that cigarette smokers have higher risk of developing the bladder cancer as compared to the non smokers. The coffee consumption was observed the significant risk factor in both males and females while the heavy use of beer was also significantly associated with the bladder cancer in males. Family history of bladder cancer and consumption of high fats meals were found to be significant in male. Considering the occupational history, they found that workers in printing, plastics, rubber, mining, dyestuffs industries, oils, petroleum, exposure to paints and pesticides, truck drivers had significantly risks of urinary bladder cancer as compared to the other professionals. Multivariate logistic regression analysis was used to study the risk of these factors like Mommsen et al (1983).

A case-control study was conducted in France by Momas et al (1994). In this region, the incidence and death rates of bladder cancer were very high. This retrospective study of males was consisted of 219 cases and 794 randomly selected population controls from electoral rolls in 1987 to 1989. For this study, the information on demographic, occupation, dietary habits (use of coffee, alcohol, vegetables, spices, etc.), and history of smoking (mean number of cigarettes smoked per day, period of smoking in years, age at starting and cessation smoking, type of tobacco and filter tips), were collected through highly qualified and trained interviewers. Highly significant association was observed in number of cigarettes smoked per day, period of smoking in years and life time smoking with the bladder cancer. The same results were observed by considering these factors either continuous or categorical in the logistic regression. The 88 percent of the smokers were using the black tobacco. The cigarette smokers had 5 times more risk of getting the bladder cancer as compared to the non-smokers like the Burns and Swanson (1991), Mommsen (1983) and Kunze et al (1993) . The comparison was made between the type of tobacco and the age at start smoking. The subjects who began smoking in the age of 13 had 3.42 times more risk than the subjects who began smoking after the age of 21 year. Hence, odd ratio and 95 % CI for the subjects started smoking before age 13 versus age after 21 was 3.42 and (1.07-10.9), respectively. The odd ratio and 95 % CI for the subjects using black tobacco (The tobacco dried by hanging up in the sheds open from all sides called air-cured or black tobacco) versus blond (Tobacco dried by hanging up on a series of tubes and the steam or hot air is passed through the tubes, called flues, is called blond tobacco or flue-cured tobacco) had 1.63 and (0.73-3.64). It was found that the black tobacco is more injurious as compared to the blond (Flue-cured tobacco is hung up in sheds which have a series of tubes through which hot air or steam is blown. The heat given off by the tubes accelerates the drying of tobacco. This system takes its name from the tubes which are known as flues or blond.) tobacco. Multivariate logistic regression analysis was used to study the risk of these factors like Mommsen (1983) and Burns and Swanson (1991). Moreover, in this study the case control ratio was about 1:4 that differentiate it from the Kunze et al (1993) and Mommsen (1983) having case control ratio 1:1. But the study of Burns and Swanson (1991) had much more sample size than

the Kunze et al (1993) and Mommsen (1983) that distinguished it from the two. The statistical analysis was done using BMDP (Biomedical Package) software. The use of this package has become uncommon due the existence of the latest statistical packages like SAS, SPSS, etc.

A study was conducted in Taiwan in order to look at the association between the arsenic water and cancers of the bladder and kidney by Guo et al (1997). For this study, data of patients was obtained from the cancer registry 1980 to 1987 surrounding 243 townships. This study was run to investigate the association for kidney and bladder cancers of various cell types and the association of cigarette sales with that of the urinary bladder cancer. Proportions of wells were used with various specified arsenic level in each township. The high association was observed in both males and females with transitional cell carcinomas of the bladder and kidney. But the adenocarcinoma was linked to the drinking arsenic water in case of males and not in squamous cell carcinomas of the bladder or renal cell carcinomas of the kidney. The multiple linear regression was used for this analysis.

A case-control study of 300 cases having 239 male and 61 female with the equal number of controls was conducted by Pohlabeln et al (1999) in Hessen (a state of Germany), between 1989 and 1992. The controls were derived from the same hospital by matching the age, sex and residential area. The male cigarette smokers had 2.8 times and female smokers 5.33 times more risk of getting the bladder cancer as compared to the non-smokers like the Burns and Swanson (1991), Mommsen (1983), Kunze et al (1993) and Momas et al (1994). The odd ratios and 95% CIs for male and female cigarettes smokers were 2.80 (1.65-4.76) and 5.33 (1.55-18.33), respectively. While the factors number of cigarettes smoked per day, cigarette smoking period, age at beginning of cigarette smoking and the number of years stop smoking were found to be significant with the time and dose response for males. But for females, factors were also significant but not showing the amount of time and dose response. Males using 2 or more cups of coffee and 3 or more bottles of beer per day had significantly higher risk of bladder cancer as compared to non coffee and beer drinkers respectively like the findings of

Kunze et al (1993). No significant association was observed between the fluid consumption and bladder cancer in males but a significant reduction in odds ratio 0.34 with 95% CI (0.11-0.99) was found for females who were consuming more than two liters of fluid. The logistic regression was run by SAS software. But in the study of the Momas et al (1994) had used BMDP software unlike the Pohlabeln et al (1999) who used SAS for analysis.

Mannetje et al (1999) investigated the occupational risk factors of bladder cancer in female by pooling the data of eleven case control studies in the 6 European countries. These 11 studies were conducted from 1976 to 1996 consisting on 700 female cases and 2,425 controls, considering age from 30 to 79. The case control ratios was about 1:1 to 1:8 in the individual studies but after pooling the data the case control ratio was about 1:3.5. Common coding and classification schemes were used for the smoking and occupational history. The average age of cases and controls were 66.2 years and 61.9 years, respectively. Pooling these studies, the sample size had much increased that leads it to the better results. The unconditional logistic regression model was used to compute the odds ratios and the 95% confidence intervals for the risk factors. The odds ratios and 95% CIs were for metal workers (blacksmiths, toolmakers and machine tool operators), tobacco workers, farm worker (crop and vegetable farm workers), tailors and dress makers, sales female, mail sorting clerks were (2.0; 1.1-3.6), (3.1; 1.1-9.3), (1.8; 1.0- 3.1), (1.4; 1.0-2.1), (2.6; 1.0-6.9), and (4.4; 1.0-19.5), respectively. Kunze et al (1993) were also found occupational risk factors like study. They found that workers in printing, plastics, rubber, mining, dyestuffs industries, oils, petroleum, exposure to paints and pesticides, truck drivers had significantly risks of urinary bladder cancer as compared to the other professionals. The risk of bladder cancer was found to be higher in the females having age less than 65 as compared to the older female. The odds ratios and 95% CIs in ex-smokers and current smokers were (2.5; 1.9-3.4) and (3.6; 2.8-4.7), respectively as compared to the never smokers. Pohlabeln et al (1999) observed that male cigarette smokers had 2.8 times and female smokers 5.33 times more risk of getting the bladder cancer as compared to the non-smokers and similar results were observed by Burns and Swanson (1991), Mommsen (1983), Kunze et al (1993) and Momas et al (1994).

Sala et al (2000) conducted a case-control study by pooling the ten European case-control studies of bladder cancer. In this study, the pooled cases and controls were 564 and 2,929, respectively. The information of this study was based on the data taken from the studies conducted in Greece, Germany, Spain, Denmark, Italy and France. Pooling these studies, the sample size had much increased and case control ratio was about 1:5. These two aspects of the study lead it to the better results. The purpose of this study was to observe the risk associated with coffee consumption and of bladder cancer. From these studies the data about the occupation and the coffee consumptions was taken. It was reported that the coffee drinkers in the study population were 79 percent. Furthermore, it was also stated that 2.4 percent coffee drinker were taking 10 or more cups daily (called heavy drinkers). In ever coffee drinkers, no significant risk was observed because the odd ratio and CI for ever coffee drinkers were 1.0 and (0.8-1.3), respectively. But the heavy coffee drinkers had significantly more risk of bladder cancer as compared to the never drinkers with odd ratio 1.8 and CI (1.0-3.3), respectively. Consequently, heavy coffee drinkers had 1.8 times more bladder cancer risk as compared to never coffee drinkers. But the studies of Pohlabeln et al (1999) showed that the subjects who consume 2 or more coffee cups daily had higher risk of bladder cancer as compared to the non coffee drinkers. In this study, the case control ratio was about 1:5. Controls were taken by matching age, gender and occupation. For the purpose of analysis, the SAS and STATA software packages were used to run the unconditional logistic regression model. Mannetje et al (1999) has also used the unconditional logistic regression model like Sala et al (2000).

A case-control study conducted by Brennan et al (2001) was based on the information obtained by pooling the eleven European case-control studies having 685 female cases and 2,416 female controls of bladder cancer. The logistic regression model was used to estimate the odds ratio and CIs. This study was run especially for female to observe the association between the cigarette smoking and bladder cancer. In this study, period of cigarette smoking in years, stop smoking period and the number of cigarettes smoked per day were included as variable. The odds ratios and the 95% CIs were

30 calculated for observing the risk. A significantly increased risk of bladder cancer was found about 2 times in the smokers who were smoking a period of less than 10 years as compared to non smokers having odd ratio 1.9 and CI (1.1-3.1). On the other hand, the odds ratio and CI for the smoking period more than 40 years were 4.1 and (3.0-5.5), respectively. The significant association was also found to the smokers who smoke 15 to 20 cigarettes per day for which the odd ratio and CI were 3.8 and (2.7-5.4), respectively, after which increasing risk of bladder cancer was not found. In case of stop smoking period, a gradually decreased risk of bladder cancer was found. For the stop smoking period of 1 to 4 year, the 30% decrease was observed with odds ratio 0.68. In case of the stop smoking period of more than 25 years, then the odds ratio and CI were 0.27 and (0.21-0.35). These results were similar to those of Burns and Swanson (1991), Mommsen (1983), Kunze et al (1993), Momas et al (1994) and Pohlabeln et al (1999).

A retrospective study was run in order to investigate the possible association between the several factors and the risk of urinary bladder cancer by Radosavljevic et al (2001). The study was based on the 130 cases of urinary bladder cancer and 130 of controls. The case control ratio was 1:1. This information was obtained by matching the cases and controls with respect to age, sex and residential area. The purpose of analysis using the multivariate logistic regression model was used. The factors having increased risk of the urinary bladder cancer were consumption of animal fats, pickled food and habit of smoking while the factors having protection against the urinary bladder cancer were higher educational level and more frequency of urination.

Pitard et al (2001) planned a study to investigate the risk factors of the urinary bladder cancer in male from the six published retrospective studies conducted in Germany, Spain, France and Denmark. The main purpose of pooling the other studies was to observe the association of smoking, cigar and pipe with bladder cancer using large sample size. In this pooled data set, 2,279 cases and 5,268 controls were taken by matching center, age and occupation exposure. The case control ratio was about 1:2. The complete information of smoking pipe, cigarettes and cigars was obtained separately and odds ratio and 95% CIs were calculated. The descriptive results had shown that only cigar or pipe was smoked by 88 cases and 253 controls while 1,420 cases and 2,895

controls were purely cigarette smokers. The odds ratio and CI for only cigarette smokers were 3.5 and (2.9-4.2), respectively which means that the cigarette smokers were 3.5 times more risk of bladder cancer as compared to the non smokers. These results were similar to the Burns and Swanson (1991), Mommsen (1983), Kunze et al (1993), Momas et al (1994) and Pohlabeln et al (1999) and Brennan et al (2001). In the similar way, the odds ratio and CI for only pipe smokers were 1.9 and (1.2-3.1), respectively which means that the pipe smokers were 1.9 times more risk of bladder cancer as compared to the non pipe smokers. Furthermore, the odds ratio and CI for only pipe smokers were 2.3 and (1.6-3.5), respectively. When the smoking duration of cigars and cigarettes was observed, it was found insignificantly less for cigars than for cigarettes. It was suggested from the conclusions that smoking of pipe and cigars was carcinogenic to the urinary bladder. For the purpose of analysis, the unconditional logistic regression model was used. Pelucchi et al (2002) has also used the unconditional logistic regression model for analysis.

A retrospective study was planned by Pelucchi et al (2002) in order to observe the risk factors of the urinary bladder cancer in female from 1985 to 1992 in the two areas of Italy. The study was based on the 110 cases of urinary bladder cancer in female and 298 controls admitted in the same hospital. The case control ratio was about 1:3. For the purpose of analysis, unconditional logistic regression was run and odds ratio were obtained. The odds ratio for the current smokers as compared to the nonsmokers was 2.87 which mean that the current smokers had 2.87 times more risk of bladder cancer as compared to the nonsmokers. In the studies of Brennan et al (2001), an increased risk of bladder cancer was found about 2 times in the female smokers who were smoking a period of less than 10 years as compared to non smokers and 4 times risk in case of smoking more than 40 years. The similar results were also observed by Burns and Swanson (1991), Mommsen (1983), Kunze et al (1993), Momas et al (1994) and Pohlabeln et al (1999). The occupation exposure was also found to be associated with the risk of bladder cancer. Especially the workers in dyes or paints factories, in chemical factories and in pharmaceutical industries had three times more risk of urinary bladder cancer as compared to the workers related to the other occupations like the Mommsen (1983). Similarly, Kunze et al (1993) found that workers in printing, plastics, rubber,

mining, dyestuffs industries, oils, petroleum, exposure to paints and pesticides, truck drivers had significantly risks of urinary bladder cancer as compared to the other professionals. In this study, it was further investigated that the high intake of vegetables had a protection against the disease with odds ratio 0.32.

Zeegers et al (2002) investigated the associations between cigarette, cigar, pipe, Environmaletal Tobacco Smoking (ETS) and bladder cancer by using a prospective cohort study. The basic information was obtained through questionnaire on cancer risk factors from all the subjects included in this study in 1986. This cohort of subjects was followed up for incident bladder cancer through the cancer registries till 1992. The analysis of this study was based on 619 cases and 3,346 controls of considering sub cohort members. For the purpose of analysis, exponentially distributed failure time regression models were used to compute the Incidence Rate Ratios (RR) and corresponding 95% confidence intervals for bladder cancer. The software STATA was used for the purpose of data analysis. The incidence rate ratios and CIs for ex-smokers and current cigarette smokers were (2.1, 1.5-3.0) and (3.3, 2.4-4.6), respectively which mean that the current smokers had 3 times more risk of bladder as compared to the nonsmokers and similarly, ex-smokers had 2 times risk of bladder cancer as compared to the nonsmoker. The similar results were also observed in the studies of Pelucchi et al (2002), Pohlabeln et al (1999), Momas et al (1994), Kunze et al (1993), Burns and Swanson (1991) and Mommsen (1983). The RR for smoking period was 1.03 (1.02-1.04) for one year. This RR will be increased as the number of years increased. Similarly, the RR and CI for the subjects who were smoking 10 cigarettes per days was found to be 1.3 and (1.2- 1.4) as compared to the nonsmokers. Pipe smoking, use of filter-tip, smoking cessation, age at first exposure, cigar and use of ETS were not found to be significantly associated with bladder cancer risk.

Kogevinas et al (2003) were conducted a case control study in order to investigate the risk of bladder cancer in male due to occupation and industries by pooling the information from the eleven case control studies conducted in the 6 European countries (Germany, France, Italy, Spain, Greece and Denmark) from 1976 to 1996. The study was

based on the 3,346 cases of urinary bladder cancer and 6,840 of controls having age between 30 to 79 years. The information about the occupation and smoking attitudes were obtained with the case control ratio was 1: 2. The odds ratios and 95% CIs were computed using the logistic regression model. The software SAS and STATA (version 7.0) were used for data analysis. In this study, the occupations including knitters, automobile painters, machinists, automobile mechanics and textile machinery mechanics had odds ratios 2.56, 1.95, 1.5, 1.38 and 2.86, respectively. In some specific categories of occupations like metal workers, painters, miners, excavating-machine operators, transport operators, textile and electrical workers and in non-industrial workers like janitors and concierges, the higher risks were investigated. Mommsen (1983) and Kunze et al (1993) also found the similar results in their studies.

A retrospective study was planned by Reimar et al (2004) in order to search out the occupational risk factors of the urinary bladder cancer in seven provinces of Canada. The study was based on the 887 cases and 2,847 controls (matching age and gender) of urinary bladder cancer in 1994 to 1997. The case control ratio was about 1:3. The information for this study was obtained through a questionnaire and the response rate was about 60 %. Odds ratios and 95% CIs were computed by using the unconditional logistic regression model in order to assess the association between each occupation and bladder cancer in both males and females. The statistically significant odds ratios and CIs in males for the hairdressers, primary metal workers, minors, and auto mechanics were found to be (3.42; 1.09-10.8), (2.40; 1.29-4.50), (1.94; 1.18-3.17) and (1.69; 1.02-2.82), respectively. Mommsen (1983), Kunze et al (1993) and Kogevinas et al (2003) were also shown similar results in their studies. Similarly, the statistically significant odds ratios and CIs in female for lumber processors, general labourers nurses and general clerks (8.78; 1.28-60.1), (2.18; 1.05-4.52), (1.54; 1.03-2.31) and (1.48; 1.01-2.17), respectively. The risks of bladder cancer were found to be insignificant in case of male general labourers, firefighters, printers, government inspectors, and welders. This study also shown that a subject who smokes 40 years or more in his history had more than 4 times risk of bladder cancer with CI (2.84-6.04) in males and 5.44 times risk of bladder cancer in females as compared to the non smokers. In the studies of Brennan et al (2001), an

increased risk of bladder cancer was found about 2 times in the female smokers who were smoking a period of less than 10 years as compared to non smokers and 4 times risk in case of smoking more than 40 years. Pelucchi et al (2002), Mommsen (1983) and Kunze et al (1993) were also found the similar results in their studies. Furthermore, the 45% reduction in the risk of bladder cancer was observed for the female who were eating fruits more than 15 serving per week as compared to those female who were taking less than 5 servings per week. The odds ratio and CI for this case was 0.55 and (0.37-0.82), respectively. In this study, all the dietary factors were coded into quartiles of serving quantity that is, less than 5, 6-10, 11-15 and more than 15 servings per week and less than 5 servings per week were taken as reference category.

A case-control study was planned by Ugnat et al (2004) in order to search out the occupational and non-occupational risk factors of the urinary bladder cancer in four western provinces of Canada. The study was based on the 549 bladder cancer cases and 1099 controls. The case control ratio was about 1: 2. Unconditional logistic regression model was used for the analysis. The odds ratios and 95% CIs were computed in order to explain the strength of risk factors. An increasing risk of bladder cancer was observed in cigarette smokers with an odds ratio and CI, 3.32 and (2.28-4.82), respectively which indicated that the cigarette smokers had more than 3 times risk of bladder cancer as compared to the nonsmokers. The similar results were also observed in the studies of Maurice et al (2002), Pelucchi et al (2002), Pohlabeln et al (1999), Momas et al (1994), Kunze et al (1993), Burns and Swanson (1991) and Mommsen (1983). An increasing risk was also found in male coffee drinkers. But the coffee drinkers of 4 or more cups per/day had 1.77 times more risk as compared to the nondrinkers with CI (1.11-2.82). The studies of Sala et al (2000) and Pohlabeln et al (1999) also showed the similar results about the coffee drinking.

Another case control study conducted in Serbia to study the non-occupational risk factors for urinary bladder cancer by Radosavljevic et al (2004). The study was based on 130 newly diagnosed cases of bladder cancer and 130 controls. The controls were matched with considering age ( + 2 years) and residential area. The information was obtained from the Clinical Center of Serbia in Belgrade and from Kragujevac in central

Serbia. The case control ratio was about 1:1 and the period of study was June 1997 to March 1999. The multivariate logistic regression model was used for the purpose of analysis. The odds ratio and 95% CIs for the frequency of daily urination, consumption of liver, canned meat, fruit juices, the highest tertile of pork, cabbage and vinegar were (0.18; 0.08-0.39), (13.81; 2.49-76.69), (8.38; 1.74-40.36), (0.08; 0.01-0.56); (4.55; 1.30- 15.93), (0.25; 0.06-1.01) and (4.41; 1.18-16.50), respectively. From these odds ratio and CIs, it was observed that the frequent daily urination, consumption of fruit juices and cabbage were referred as protective factors against the urinary bladder cancer while the consumption of liver, canned meat, pork and vinegar were considered as increased risk factor for bladder cancer.

Radosavljevic et al (2005) also conducted a case control study in Serbia to study the association between diet and risk factors for urinary bladder cancer like Radosavljevic et al (2004). In this study, 130 newly diagnosed cases of urinary bladder cancer and 130 controls were included by matching age, sex and residential area (rural or urban). The case control ratio was 1:1 like Pohlabeln et al (1999). A food frequency questionnaire was used to obtain the relevant information. Multivariate logistic regression model was used to find the odds ratio and 95% CIs for each tertile and the lowest tertile was considered as the reference category. In this analysis, the odds ratios and CIs for consumption of liver, eggs, pork and pickled vegetable were (6.60, 1.89-23.03), (3.12, 1.10-8.80), (2.99, 1.16-7.72) and (3.25, 1.36-7.71), respectively. All the stated factors had the risk of bladder cancer. In the similar way, the odds ratios and CIs for cereals, kale, cabbage, tangerines and carrots were (0.19, 0.06-0.62), (0.21, 0.06-0.73), (0.27, 0.11- 0.68), (0.21, 0.07-0.68) and (0.15, 0.05-0.41), respectively. So, it was observed that the cereals, kale, cabbage, tangerines and carrots had the inverse association with the urinary bladder cancer which means that these were the protective factors. Pelucchi et al (2002) were also investigated that the high intake of vegetables had a protection against the disease with odds ratio 0.32.

Samanic et al (2006) planned this retrospective study to investigate the association between the dose, type of tobacco, inhalation, cessation, and environmental

36 tobacco smoke with bladder cancer risk. For this purpose, the 1,219 cases of bladder cancer and 1,271 controls were selected from eighteen hospitals in Spain. The unconditional logistic regression model was used to assess the odds ratios and 95% CIs for explaining the association for the risk of bladder cancer and the characteristics of cigarette smoking. The ex-smokers male and female were at high risk with odds ratios and CIs (3.8; 2.8-5.3) and (1.8; 0.5-7.2), respectively as compared to the nonsmokers. Similarly, current smokers male and female were at high risk having odds ratios and CIs (7.4; 5.3-10.4) and (5.1; 1.6-16.4), respectively. These results were also supported by the studies of Maurice et al (2002), Pelucchi et al (2002), Pohlabeln et al (1999), Momas et al (1994), Kunze et al (1993), Burns and Swanson (1991) and Mommsen (1983). A significantly increasing trend in risk was found with increasing the duration and amount of smoking. After adjustment for duration, 40% higher risk was assessed in smokers of black tobacco than the smokers of blond tobacco having odds ratio and CI 1.4 and (0.98- 2.0), respectively. For 20 or more years of quitting smoking of blond tobacco, a significant inverse trend was investigated as compared to the current smokers with odds ratio and CI 0.2 and (0.1-0.9), respectively. The increased risk was observed for male who inhaled into the throat and chest with the odds ratios and CIs (1.7; 1.1-2.6) and (1.5; 1.1-2.1), respectively as compared to the male who inhaled into the mouth.

Yaris et al (2006) conducted this study to find out the risk factors of urinary bladder cancer in Turkey to improve the precautionary measures because incidence of the urinary bladder cancer were 2.1 and 13.0 per 1,00,000 in females and males, respectively. For this purpose, a retrospective hospital based study was conducted in the three hospitals of Istanbul. The 290 cases were the patients of the urinary bladder cancer and 580 controls were randomly drawn from the non-urinary patients by matching sex, age and residence. The case control ratio was 1:2. All the cases and controls were interviewed and the variables related to demographic conditions, smoking habits, coffee consumption and occupational history were recorded. The odds ratio and 95% CIs for males and female smokers were (1.15; 1.14-3.04) and (2.80; 0.98-7.72), respectively. From these results it was observed that the female smokers had more risk of bladder cancer than males. The studies of Reimar et al (2004) had also shown the more risk of bladder cancer in female

smokers as compared to the male smokers. The coffee consumption pattern was found to be similar both in cases and controls and insignificant difference was observed in case and control groups. The both groups were using on the average one cup of coffee per day. The studies of Sala et al (2000) and Pohlabeln et al (1999) also showed the similar results about the coffee drinking when less than 2 cups were consumed. In this study, the only odds ratio were calculated for smoking and for all other variables comparison were made on the basis of percentages only.

Kellen et al (2006) conducted study to investigate that the association of vegetables and fruits with the urinary bladder cancer in Belgian province of Limburg. For this purpose, a case-control study was planned by considering 200 cases of urinary bladder cancer and 385 controls. Analysis was performed by using Logistic regression model. The odds ratios and the 95% CIs were calculated to explain the risk of the bladder cancer. The odds ratio and 95% CI for the higher consumptions of vegetables were 1.15 and (0.70-1.88), respectively which means that entire vegetables consumption had insignificant association with the risk of bladder cancer. Pelucchi et al (2002) were also investigated that the high intake of vegetables had a protection against the disease with odds ratio 0.32. It was also observed that the high intake of fruits had inverse association with the bladder cancer having odds ratio and 95% CI 0.61 and (0.37-0.99), respectively. But the low intake of fruits daily had a higher risk of bladder cancer for smokers with odds ratio and 95% CI 4.23 and (1.91-9.40), respectively. The risk of bladder cancer remains significant for ever smokers even the daily fruit consumption was increased having odds ratio and 95% CI 2.15 and (1.15-4.05), respectively. In other words, risk of bladder cancer reduces by high intake of the fruit. Therefore, the damages caused by cigarette smoking may be protected by antioxidants, found in fruit.

Puente et al (2006) run this case control study by pooling the data from 14 case control studies from Europe and North America like Kogevinas et al (2003) who were obtained the data by pooling the information from the eleven case control studies conducted in the 6 European countries (Germany, France, Italy, Spain, Greece and Denmark). The authors conducted this study to investigate the risk of urinary bladder

cancer in female and male when they were smoking the same amount of cigarettes. The combined data set were consisted on 8,316 cases (79 % male) and 17,406 controls (72% male) aged 30–79 years. The case control ratio was about 1:2 and the sample size was very large like Kogevinas et al (2003) and Pitard et al (2001) had large sample sizes of cases and controls (3,346 + 6,840) and (2,279 + 5,268), respectively with same case control ratio 1:2. For the study of odds ratios and 95% confidence intervals, the unconditional logistic regression model was used. The odds ratios and CIs for current smokers in male and female were (3.9; 3.5–4.3) and (3.6; 3.1–4.1), respectively as compared to nonsmokers. Pitard et al (2001) had also shown that the odds ratio and CI cigarette smokers were 3.5 and (2.9-4.2), respectively. It was evident that the risk of bladder cancer was about same for cigarette smokers in both studies. In 11 out of 14 studies, odds ratios were similar for male and female 3.4 and 3.7, respectively in North America. But in case of Europe, male had higher odds ratio for smoking as compared to female which were 5.3 and 3.9, respectively. An increase in odds ratios was observed with increased amount and time duration in male and female. Generally, male were slightly at higher risk than female.

Villanueva et al (2006) conducted a study by pooling the data of 6 case-control studies including 2 studies from the US and one from each Finland, Canada, Italy and France of bladder cancer like Puente et al (2006) who pooled the data from 14 case control studies from Europe and North America and Kogevinas et al (2003) got data by pooling the information from the eleven case control studies conducted in the 6 European countries (Germany, France, Italy, Spain, Greece and Denmark). The sufficient information was retrieved on fluid consumption in order to evaluate the risk of bladder cancer linked with total and specific fluid consumption. The study was based on the 2,729 cases and 5,150 controls like Kogevinas et al (2003) and Pitard et al (2001) had too large sample size. Odds ratios and 95% CIs for fluid consumption were adjusted for gender, age, smoking status, study, occupation and education. The increased risk of bladder cancer was found both in ex- smokers and current smokers having odds ratios and 95% CIs (2.02; 1.77–2.29) and 3.52 (3.09–4.01), respectively as compared to nonsmokers like Pitard et al (2001) and Puente et al (2006) having the odds ratios and CIs

in current cigarette smokers (3.5; 2.9-4.2) and (3.9; 3.5–4.3), respectively. The total fluid consumption was significantly associated with an increased risk of bladder cancer in male with odds ratios and CIs (1.08; 1.03–1.13) and was insignificant in female with odds ratios and CIs (1.06; 0.77–1.46) for per liter increase in daily total fluid intake (taking fluid intake as continuous variable). Using the categories, the odds ratios and CIs in male were (1.33; 1.12–1.58) for the highest category of consumption (more than 3.5 liters per days) as compared to the category consuming less than 2 liters per day. The risk of bladder cancer was found to be increased using the tap water. The tap water consumers of 0.51 liter per day were taken as reference category. The subjects who were drinking 2 liter per day had increased risk of bladder cancer in both female and male with odds ratios and CIs (1.46; 1.20–1.78) and (1.50; 1.21–1.88), respectively as compared to the reference category. It was shown that the male were at higher risk as compared to female. The increased risk of bladder cancer was observed with the consumption of tap water which suggests that the tap water may contain high carcinogenic chemicals. An increased risk of bladder cancer was found both in male and female who were drinking more than 5 cups of coffee per day with odds ratios and CIs (1.23; 1.05–1.44) and (1.31; 0.99–1.74) as compared to the subjects who were drinking less than 5 cups per day. The studies of Sala et al (2000) and Pohlabeln et al (1999) also showed that the coffee drinkers of less than 2 cups per day had not the risk of bladder cancer but the heavy drinkers had significant risk. The coffee consumption pattern was found to be similar both in cases and controls and no significant difference was observed by Yaris et al (2006). Another study showed that the coffee drinkers of 4 or more cups per day had 1.77 times more risk of bladder cancer as compared to the nondrinkers by Ugnat et al (2004). Villanueva et al (2006) tested the statistical significance of the differences among the cases and controls by using Mann-Whitney test (non parametric test) for continuous variables. None of the researcher had used the Mann-Whitney test in their studies. For categorical variables, chi-square test was used. Unconditional logistic regression model was used to compute the odds ratios and confidence intervals.

A Meta analysis was conducted by Jankovic and Radosavljevic (2007) in Serbia to summarize the risk factors of the urinary bladder cancer. MEDLINE database was used for the literature search from1985 to 2006. They reviewed the original references of the

original articles, monographs and reviews for the urinary bladder cancer. From this retrieved information, they found that in the developing countries, the cigarette smoking is an acknowledged major cause of bladder cancer. About 50 percent of all male patients of bladder cancers were caused by the cigarette smoking while in female patients about 25 percent of the disease was accounted for cigarette smoking. They further stated that a strong association between the amount and duration of cigarette smoking was investigated through literature. Arsenic drinking water, especially in some countries of the world like Bangladesh may be considered a risk factor of bladder cancer. High intake of fresh fruits, vegetables and fluid were considered the major protective factors against the bladder cancer. There is a type of infection that was occurred about 15% of all bladder cancer cases were due to the schistosomiasis, but found only in endemic areas. About 5% -10% of the bladder cancer was estimated to be found in industrialized countries due to the occupational exposure. Some drugs used in chemotherapy like cyclophosphamide was also increased the risk of bladder cancer.

This prospective cohort study was conducted by Alberg et al (2007) to investigate the effect of both household exposure to secondhand smoking (also called passive smokers) and the active cigarette smoking on urinary bladder cancer in Washington County, Maryland. This study was based on the persons taken from two cohorts those were established from private censuses in 1963 and 1975. These two cohorts were consisted on the 45,749 and 197,548 members of whom 93 and 172 cases were taken from each cohort, respectively. For the purpose of comparison, those persons were taken as controls that did never smoking nor lived with any smoker. For the purpose of data analysis, the Poisson regression models were used for the both cohorts. The relative risks and the corresponding 95 % confidence intervals for active smokers were (2.7; 1.6-4.7) in the cohort 1963 and (2.6; 1.7-3.9) for the cohort of 1975 after adjusting education, marital status and age. It was observed from these results that they had about 3 times more risk in both cohorts as compared to the controls. Ugnat et al (2004) reported that the current cigarette smokers had more than 3 times risk of bladder cancer as compared to the nonsmokers. The similar results were also observed in the studies of Maurice et al (2002), Pelucchi et al (2002), Pohlabeln et al (1999), Momas et al (1994), Kunze et al (1993),

Burns and Swanson (1991) and Mommsen (1983). The relative risks and the corresponding 95 % CIs for passive smokers (nonsmoker female but passive smokers) were (2.3; 1.0-5.4) in the cohort 1963 and (0.9; 0.4-2.3) for the cohort of 1975. The passive smokers had significant risk of bladder cancer in first cohort but not in second cohort. The authors further stated that more studies are required to observe the association between passive smokers and bladder cancer. The number of passive smokers in this study was 13 in the 1963 cohort and 6 in the 1975 cohort. So, the results on the basis of such a small sample size may not be reliable. Therefore, the further investigations are required to develop the association among the bladder cancer passive smokers.

Stefani et al (2007) conducted this case control study in Uruguay to investigate the role of non alcoholic beverages in the bladder cancer risk. This study was consisted on the 255 cases of bladder cancer and 501 controls by matching sex, age and residence in the time period 1996 to 2000. Face to face interview of the cases and controls were conducted and the history of tobacco smoking, occupation, alcohol, coffee, tea, soft drinks and intake of maté (a bitter infusion from the leaves of a special type) were obtained. The case control ratio was 1:2. The odds ratio and 95% confidence intervals were estimated using the unconditional logistic regression model. The odds ratio and CI for ever maté drinking were 2.2 and (1.2–3.9), respectively as compared to the nondrinkers. The risk of bladder cancer was found to be increased by increasing the time duration and amount of maté drinking. The odds ratios and CIs for coffee and tea drinkers, taking 1-6 cups per week were (1.5; 1.1–2.2) and (2.1; 1.4–3.1), respectively. Similarly, the odds ratios and CIs for coffee and tea drinkers, taking 7 or more cups per week, were (2.1; 1.2–3.6) and (4.1; 1.7–9.9), respectively. Jensen et al (1986) conducted a case control study in Denmark and observed the significant association between bladder cancer and tea drinkers with odds ratio and 95% CI 2.1 and (1.3–3.4), respectively. The results of both studies are similar in tea drinkers. Hence, increased risk was observed in both coffee and tea drinkers as compared to the non drinkers. Villanueva et al(2006) showed an increased risk of bladder cancer both in male and female who were taking more than 5 cups of coffee per day with odds ratios and CIs (1.23; 1.05–1.44) and (1.31;

0.99–1.74), respectively. The studies of Sala et al (2000) and Pohlabeln et al (1999) also showed that the coffee consumers of less than 2 cups per day had not the risk of bladder cancer but the heavy drinkers had significant risk. The coffee consumption pattern was found to be similar both in cases and controls and no significant difference was observed (Yaris et al, 2006). According to Stefani et al (2007), a coffee drinker who drinks 1-6 cups per week had 1.5 times more risk of getting the disease and who drinks 7 or more cups per week had 2.1 times more risk of getting the disease as compared to non drinkers (Stefani et al, 2007). Ugnat et al (2004) showed that the coffee drinkers who were drinking 4 or more cups per day had 1.77 times more risk of bladder cancer as compared to the non-drinkers.

Michaud et al (2007) conducted this large retrospective study to investigate the association between total fluid and water consumption and the risk of bladder cancer risk in Spain. This study was based on the 397 bladder cancer cases and 664 matched controls. The cases and controls were interviewed face to face by the highly trained interviewers who used computer software to record the information. For data collection, a food frequency questionnaire was used that had questions regarding drinking history of beer, liquor, coffee, wine, champagne, soda, juices, tea, milk, and water. Unconditional logistic regression model was used to compute the odds ratios and 95 % confidence intervals. The odds ratios and CIs for males and females were (0.61; 0.38–0.97) and females (0.58; 0.18–1.87), respectively for the comparison of the highest and lowest intake of total fluid. For female, no significant protection was observed from the fluid in take but in case of male, a significantly protection of 31 % was observed. According to Villanueva, et al. (2006), the odds ratios and CIs were (1.33; 1.12–1.58) in male and (1.06; 0.77–1.46) in female for the highest category of fluid consumption (more than 3.5 liters per day) as compared to the lowest category (less than 2 liters per day). Both studies were contradicting each other. This contradiction is possible because two situations rise when the total fluid taken was increased. Firstly, a follow up study of the health professionals provides a strong confirmation of the protective effect of high fluid consumption (Michaud et al, 1999) because high fluid consumption reduces contact time of carcinogen in bladder by frequent urination but contrarily, high fluid consumption

43 increases the risk of bladder cancer if the fluid is full of contaminants that are bladder carcinogens (Cantor et al, 1987). A study conducted by Kadlubar et al (1991) on animal experiments had investigated that the frequent urination is inversely associated to the carcinogens in the bladder. Bitterman et al (1991) had explained that cases having urinary tract cancer were consumed significantly less amount of fluids as compared to the healthy persons. Lesser consumption of fluid resulted in large concentration of carcinogens in the urine and increases the contact time in the bladder because of less frequent urination (Braver et al, 1987 and Kadlubar et al, 1991). For consumption of water, males and females consuming greater than 1,400 ml per day were compared with those consuming less than 400 ml per day. The odds ratios and CIs for males and females were (0.47; 0.33–0.68) and (0.61; 0.23–1.65), respectively. A significant protection was observed in male unlike female.

A hospital-based case-control study was conducted by Samanic et al (2008) in order to observe the association between occupation and bladder cancer in Spain from 1998 to 2000. This study was consisted on 1,219 cases of the urinary bladder cancer and 1,271 controls taken from 18 hospitals in Spain. The information about the occupational history, medical history, smoking habits, and several other factors were collected. The unconditional logistic regression model was used in order to find the odds ratios and 95% confidence intervals (CI). The calculations were made for every occupation and industry, adjusting for age, hospital region, duration of smoking and employment having high-risk for bladder cancer. The increased risk was investigated as significant among male who were working as machine operators in the printing industry and workers in the transportation equipment industry with odds ratios and CIs (5.4; 1.6 - 17.7) and (1.6; 1.1 - 2.6), respectively. The workers in the sanitary / electrical / gas services having a period of more than ten years and the workers of hotels and houses had also increased risk with odds ratios and CIs (3.9; 1.5 - 10.4) and (3.1; 1.3 - 7.3), respectively. Similarly, miscellaneous mechanics and supervisors in production industries were observed to be at high risk with odds ratios and CIs (2.0; 1.1 - 3.6) and (2.1; 1.2 - 3.6), respectively. According to Reimar et al (2004), the statistically significant odds ratios and CIs in males for the hairdressers, primary metal workers, minors, and auto mechanics were found to be

(3.42; 1.09-10.8), (2.40; 1.29-4.50), (1.94; 1.18-3.17) and (1.69; 1.02-2.82), respectively. Mommsen (1983), Kunze et al (1993) and Kogevinas et al (2003) were also indicated the similar results in their studies. An insignificant association was observed between occupation and the risk bladder cancer in female. But according to Reimar et al (2004), the statistically significant odds ratios and CIs in female for lumber processors, general laborers nurses and general clerks (8.78; 1.28-60.1), (2.18; 1.05-4.52), (1.54; 1.03-2.31) and (1.48; 1.01-2.17), respectively.

In Los Angeles, a case-control study was conducted by Jiang et al (2009) in order to observe the association between urinary tract infections and bladder cancer (transitional cell carcinoma). This study was consisted on 1,586 cases and same number of controls by matching gender, age, and race. For this study, the odds ratios and 95% CIs were computed. A reduced risk of bladder cancer was observed with history of bladder infection in female having odds ratio 0.66 and CI (0.46–0.96). In case of male, no effect was observed. Heavy decrease was investigated in risk of bladder cancer among female having multiple infections with odds ratio 0.37 and CI (0.18–0.78). Even Excluding the subjects from the analysis with the history of diabetes, kidney or bladder stones, still the inverse association remained the same. Furthermore, the insignificant association was found in the history of kidney infections and risk of bladder cancer.

Baris et al (2009) conducted a case–control study in Maine (New Hampshire and Vermont) from 2001 to 2004 based on the 1,170 cases and 1,413 controls. The odds ratios and 95% confidence intervals was calculated using binary logistic regression model. The odds ratio and 95% confidence interval for regular and current cigarette smokers were found to be (3.0; 2.4 - 3.6) and (5.2; 4.0 - 6.6), respectively as compared to never smokers. The current smokers had 5.2 times higher risk of bladder cancer as compared to never-smokers. Ugnat et al (2004) reported that the current cigarette smokers had more than 3 times risk of bladder cancer as compared to the nonsmokers. In New Hampshire, statistically significant increasing trend was observed. The odds ratios and 95% confidence intervals of ex-smokers in three consecutive periods including 1994 to 1998, 1998 to 2001 and 2002 to 2004 were (1.4, 1.0 - 2.0), (2.0; 1.4 - 2.9) and ( 2.6; 1.7 - 4.0), respectively). Similarly, the odds ratios and 95% confidence intervals of current smokers

in the same three consecutive periods including 1994 to 1998, 1998 to 2001 and 2002 to 2004 were (2.9; 2.0 - 4.2), (4.2; 2.8 - 6.3) and (5.5; 3.5 - 8.9), respectively. Hence, the higher odds ratios were observed by increasing pack-years smoked. While studying the pack years and intensity, it was found that the smoking of smaller number of cigarettes for a long period of time is more injurious than smoking large amount of cigarettes over a short period of time. Villanueva et al (2006) found that the increased risk of bladder cancer was observed both in ex- smokers and current smokers having odds ratios and CIs (2.02; 1.77–2.29) and 3.52 (3.09–4.01), respectively as compared to nonsmokers like Pitard et al (2001) and Puente et al (2006) having the odds ratios and CIs in current cigarette smokers (3.5; 2.9-4.2) and (3.9; 3.5–4.3), respectively.

In the above described studies, most of these were conducted in Europe and America which indicate the strength of the risk factors according to their own environment. For example, Arsenic drinking water, especially in some countries of the world like Bangladesh and Taiwan, may be considered a risk factor of bladder cancer. Similarly, Bladder cancer cases were also observed only in Egypt due to the schistosomiasis (a type of infection). Alcohol and wine are more common in the non- Muslim countries.

2.2 Review of Methodology In methodology review, it is investigated from the section 2.1 that the binary logistic regression models were run in all retrospective studies e.g., Burns and Swanson (1991) used the logistic regression model while studying the risk of urinary bladder cancer among Blacks and Whites (The Role of Cigarette Use and Occupation). Multivariate logistic regression model was run by Kunze et al (1993) in order to study the Etiology, pathogenesis and epidemiology or urothelial tumors.

Momas et al (1994) like Burns and Swanson (1991) and Kunze et al (1993) also used the binary logistic regression to study the effect of black tobacco cigarette smoking on bladder cancer. Non-occupational risk factors for cancer of the lower urinary tract in

Germany were investigated by Pohlabeln et al (1999) by using the binary logistic regression modeling.

Mannetje et al (1999) worked on the occupation and bladder cancer in European women and used the logistic regression modeling for the binary response variable. Brennan et al (2001) also run logistic regression in order to study the contribution of cigarette smoking to bladder cancer in women (Pooled European Data) like Momas et al (1994), Burns and Swanson (1991), Kunze et al (1993), Pohlabeln et al (1999) and Mannetje et al (1999).

In the similar way, Pelucchi et al (2002), Kogevinas et al (2003), Reimar et al (2004), Samanic et al (2006), Kellen et al (2006), Puente et al (2006), Samanic et al (2006), Kellen et al (2006), Puente et al (2006) and Baris et al (2009) has modeled the binary response variable by using the logistic regression technique in their studies like Momas et al (1994), Burns and Swanson (1991), Kunze et al (1993), Pohlabeln et al (1999) and Mannetje et al (1999). . On the other hand, the prospective study of Zeegers et al (2002) used the exponentially distributed failure time regression models to compute the Incidence Rate Ratios and corresponding confidence intervals. Similarly, Alberg et al (2007) used the Poisson regression models for two cohorts in order to calculate the relative risks and the corresponding confidence intervals for relative risks. Therefore, the most suitable technique for modeling the urinary bladder cancer data having binary response variable is the binary logistic regression model.

Chapter 3 Theoretical Frame Work

Research is a process of investigating new knowledge through variables by the examination of data. This study is designed to investigate the epidemiology and different risk factors for urinary bladder cancer. The purpose of this chapter is to explain the study design and the procedures. Main types of studies, sampling design, instrument of the study, selected areas and hospitals, pre-test, coding and explanations of the variables and the statistical methods used for analysis are discussed. Data collected for this study is analysed and summarized according to the described frame work.

3.1 Study Design It is a hospital-based case-control study which is conducted to investigate the risk factors of the urinary bladder cancer. A retrospective or case-control study is an analytical study which compares individuals who have a specific disease (case) with a group of individuals free from the disease (control). The proportion of each group having a history of particular exposure is then compared. It is advantageous for the control to come from the same population from which the cases are derived to reduce the chance that some other differences between the populations accounted for the difference in the disease that is under study. The overall population of the Pakistan is considered as the target population as the representative sample drawn from the hospitals of the headquarters of each province and federal area is used to generalize the result for the population of Pakistan.

3.2 Inclusion Criteria The inclusion criteria are a criterion which is observed to include the cases or controls in the study. This criterion is given below.

 Patients of all age groups and gender were included in this study.

 For cases, the patients admitted in the urology / cancer wards of the selected hospitals for the treatment of urinary bladder cancer were taken.  For controls, the persons with the patient (admitted in the urology / cancer wards of the selected hospitals for the treatment of urinary bladder cancer) for his care having age 35 or more were taken.  The data collection period was March to December, 2009.

3.3 Sampling Technique It is very important part of the study because it represents the characteristics of the population under study. So, it had been selected very carefully. Pakistan has four provinces Punjab, Khyber Pukhtoon Khwa, Baluchistan and Sindh, the description is given in Section 1.5. The headquarters of each province are Lahore, Peshawar, Quetta and Karachi, respectively. Besides these provinces, there is a federally administered area including Islamabad and some other related areas. Four hospitals having urology / oncology ward were selected at random from Lahore, two from Peshawar, two from Quetta, two from Karachi and two from the federal headquarter (Islamabad). In this way, a sample of 12 hospitals was taken and all the patients of the urinary bladder cancer from each hospital were directly interviewed up till the required/ fixed sample size was reached. Two controls were also interviewed against one case that is the case-control ratio was 1:2. When more controls will be used, the discriminating power of the logistic regression will be increased. To take the control from the same population, it was tried to select the persons as control that were with the case for his care between the age 38 and above. Consecutive sampling technique was used for selecting the samples from different cites and all subjects coming to the selected hospitals for the treatment of urinary bladder cancer, beginning specific dates and terminating the sampling when a predetermined size obtained, were included in this study. This procedure simulated the random selection when the subjects were taken without preference in the study period.

According to this technique, the sample was taken by consecutive visits of the selected hospitals and all the confirmed patients’ of bladder cancer coming to the urology/ oncology wards for the treatment were interviewed. The process was continued

until the required sample size was reached. In this way, all the hospitals were surveyed and requisite informations were obtained through a well defined questionnaire constructed for the purpose. Consequently, a sample of 900 subjects including 300 cases and 600 controls were selected from all the 12 hospitals.

In case of logistic regression, as a rule of thumb, Peduzzi et al. (1996) recommend that the smaller of the classes of the dependent variable have at least 10 events per parameter in the model. Hosmer & Lemeshow (1989) recommend a minimum of 10 cases per independent variable. This study was consisted on the 25 independent variables and subcategories of the variables. Therefore, according to the rule of thumb, at least 250 cases are required to study the matter. Keeping in view the rule of thumb, 300 cases and 600 controls were taken in order to study the risk factors of the bladder cancer.

3.4 Instrument of the Study For the data collection purpose, a questionnaire was developed for both the cases and the controls. A questionnaire is a set of questions which are used for the purpose of data collection. The questionnaire was designed with the help of the supervisor and the medical advisors (urologist and oncologist). Questionnaire of this study, given as Appendix page, includes maximum risk factors of the disease, urinary bladder cancer. The direct personal interviewing method was adopted from the cases and controls in all the hospitals. The prominent risk factors for the study were segregated from the available literature.

3.5 Area under Study It is hospital based cross-sectional study that investigates the risk factors of the occurrence of the urinary bladder cancer in Pakistan. The sample was selected from the cases (patients diagnosed with the urinary bladder cancer) and the controls (free from the urinary bladder cancer) those coming from the different areas of each province and federal city Islamabad. The selected areas and the hospitals were explained in the Table 3.1.

Table 3.1 Hospitals of Different Headquarters Visited for Data Collection Area City Name of Hospital 1. Lahore 1. Services Hospital 2. Sheikh Zayed Hospital 3. Jinnah Hospital 4. Institute of Nuclear Medicine & Oncology (INMOL) 2. Peshawar 1. Institute of kidney diseases Hayatabad medical complex 2. Lady Reading Hospital 3. Islamabad 1. Pakistan Institute of Medical Sciences 2. Nuclear Medicine of Oncology & radiotherapy Institute(NORI) 4. Quetta 1. Provincial Sandeman Hospital 2. Centre for nuclear Medicine & Radiotherapy (CENAR) 5. Karachi 1. Jinnah Postgraduate Medical Centre (JPMC) 2. Karachi Institute of Radiotherapy & Nuclear Medicine (KIRAN)

3.6 Pre-Test Pre testing is a way of assessing the validity of the questionnaire to see whether the questionnaire works properly in the field. In the pre-test, 44 questionnaires were filled by interviewing the patients admitted in the urology wards of the three hospitals i.e., Services Hospital, Lahore, Mayo Hospital, Lahore and Allied Hospital, Faisalabad. The 44 questionnaires were filled in by 19 cases and 25 controls. After the pre-test, minor changes were made in the questionnaire. In this context, age of the controls was considered above 40 because all the patients were above the age of 40 in the pre-test. Income groups were also modified in order to capture the social status. From the pre-test, it is also observed that some of the variables included this study were not reported by the cases or controls e.g., alcohol, coffee, personal history of cancer, chemotherapy, radiotherapy, etc. But those variables were included in study to observe that they may be reported in some other areas.

3.7 Data Collection and Analysis The requisite information was obtained from the patients admitted in the urology / oncology wards in the stated hospitals directly. To obtain the required information, all the

respondents including cases and controls were interviewed. During the data collection, the help of the doctors and paramedics was also obtained in order to excess the relevant patient directly. After the completion of data collection from the relevant department that is urology or oncology, the certificate of the data collection was obtained from the head of the relevant department of the hospital.

3.8 Data Analysis After the data collection, the most difficult task was to analyse the data and sort out the results. All the factors were pre-coded for the computer analysis and the data were entered into the personal computer. The software SPSS version 16.0 for windows was used for data processing and analysis. Descriptive and the analytical results were obtained using the different statistical tools including averages, percentages, χ2, Phi/ V- statistics, Kendall’s Tau-b and binary logistic regression models.

3.9 Description and Coding for the Factors under Study To observe the characteristics of the subjects like age, income, marital status, residential area, lifestyle, profession, education, chemical exposure, family history of cancer, personal history of cancer, smoking cigarettes, No. of cigarettes smoked per day, No. of Years smoking cigarettes, No. of Years stop smoking cigarettes, smoking huqa, chew pan, use of tea, No. of cups of tea used per days, use of coffee, No. of cups of coffee used per day, use of alcohol, chemotherapy, radiation therapy, chronic bladder irritation, fluid consumption, defect in bladder by birth, source of drinking water, Hepatitis, Diabetes and eating habits a questionnaire was used. The coding scheme for all the factors with their variable names was presented in Table 3.2.

Table 3.2 Coding Schemes for the Factors S. No. Factor Coding schemes Variable Reference Name category 1 Gender Female = 0 GEND First Male = 1 2 Age Years Age - 3 Social Status 1=Less than Rs.10,000 SS First

(Income) 2=Rs. 10,000 –Rs.20,0000 3=Rs. 20,000 and above 4 Residential Area1 0=Urban RA Second 1=Rural 5 Residential Area2 0=Non-industrial IA First 1=Industrial 6 Marital status 0=Unmarried MARI First 1=Married 2=widow 3= divorced 7 Lifestyle 0=Sedentary LS First 1= Normal 2= Active 8 Chemical exposure 0=No CE First 1=Yes 9 Education 0=Illiterate LIT First 1=Literate 10 Family history of 0=Negative FHC First cancer 1=Positive 11 Personal history of 0=Negative PHC First cancer 1=Positive 12 Cigarettes 0=No CS First Smoking 1=Yes 13 No. of cigarettes 0=NA NCSP First smoked per day 1=1-10 2=10-20 3= 20 and above

14 No. of years cig. 0=NA NYCS First smoking 1=1-15 2=15-30 3= 30 and above 15 No. of years stop 0=NA NYSS First smoking 1=1-5 2=5-10 3= 10 and above 16 Huqqa smoking 0=No HS First

1=Yes 17 Chew pan 0=No CP First 1=Yes 18 Use of tea 0=No Tea First 1=Yes 19 No. of tea cups 0=NA TCUP First 1=1-4 2=4 and above 20 Use of coffee 0=No Coffee First 1=Yes 21 No. of coffee cups 0=NA CCUP First 1=1-3 2=3 and above 22 Use of alcohol 0=No Alcohol First 1=Yes 23 Chemotherapy 0=No CHEMO First 1=Yes 24 Radiation therapy 0=No RADI First 1=Yes 25 Chronic bladder 0=No CBI First irritation 1=Yes 26 Defect in bladder 0=No DBB First by birth 1=Yes 27 Fluid consumption 0=less than 10 glasses FC First (Glasses) 1= 10 or more glasses 28 Source of 1=Tap SDW First water(drinking) 2=Canal 3=Govt. Provided 4=Mineral water 29 Hepatitis 0=No HEPA First 1=Yes 30 Diabetes 0=No DIAB First 1=Yes 31 Hair dye 0=No HD First 1=Yes 32 Fried items (food) 1=Low FRIED First

2=Normal 3=Excessive 33 Fats items 1=Low FAT First 2=Normal 3=Excessive 34 Fast food 1=Low FF First 2=Normal 3=Excessive 35 Fruits 1=Low FRUIT First 2=Normal 3=Excessive 36 Professions 1= Textile worker Professions Last 2= Cook 3= Dye worker 4= Driver 5= Workshop worker 6= Mining worker 7= Farmer 8= Leather /Rubber worker 9= Diesel related 10= Painters 11=Others, ______

3.10 Types of variables The data obtained through survey by using the questionnaire was of nominal, ordinal and of the quantitative type. For nominal variables, the order of listing the categories is not necessary and statistical analysis should not depend on that ordering. Methods designed for nominal variables give the same results, without mattering the order of the categories listed. In this situation, the categories are exchangeable.

In ordinal variables, categories have natural ordering and there exist a comparison level between the variables. Social status (low, medium and high) is an example of ordinal variable and similarly in dietary habits, use of fats item, fried items, fast food, fruits and age groups etc., is the ordinal variables. According to Hosmer & Lomeshow (1989), categorical variables are also referred to as qualitative to distinguish

55 them from the numerical value or quantitative variables, such as weight, age, and income etc. However, it is, sometimes profitable to treat ordinal data in a quantitative manner, for example, by allotting ordered scores to the categories.

In this study, the response variable was binary and the independent variables were the nominal, ordinal and of quantitative type. So, the description of the variables and the best used methods for the analysis of data in the stated situation are described in the following sections.

3.11 Description of Variables In this section, every factor included in the study is explained with some of its importance and showing its coding structure briefly.

 Gender All males and female admitted in the urology/ oncology wards in the selected hospitals are considered. Generally, the minimum age of the patient is observed 40 years in the pretest. Hence, the controls are selected above 40 years of age. The gender is the nominal variable. Bladder cancer is more common in men than women, with a world wide male/female ratio of 10:3 (Sylvester, 2004). Silverman and Devesa (2006) explain that the urinary Bladder cancer occurs more frequently in males as compared to the females having male / female ratio of 4:1, in most of the western countries. So, the variation in the male to female ratio was found in western countries and world wide. The gender is included in the study to observe the frequency of the disease in male and

females. The urinary bladder cancer was found to be more frequent in male as compared to females having male/female ratio of 3:1 (Rabbani and Cardo, 2000).

 Age The age is taken as continuous variable. Its categories can be used to observe the descriptive results. The risk of bladder cancer goes up with age Urinary Bladder cancer is a disease of older and rarely diagnosed before the age of 40 (Devita, Hellman, Rosenberg, 2001). Different researcher consider the age in different ways in their studies.

By pooling the data from 14 case–control studies on bladder cancer from Europe and North America, the age was taken from 30 year to 79 years in groups categories as 30-39, 40-49, 50-59, 60-69, 70-79 to study the risk of bladder cancer between gender by smoking and the maximum patients was found in the age group 60-69 (Puente et al, 2006). The two third of all bladder caner cases were observed in the subjects having age above 65 years (Ferlay et al, 2004).

 Social Status The urinary bladder cancer is expected to be more frequent in the poor people (having low social status), because they are not able to afford the better diet and other requirements (like fruits) needed by the body. Social status was captured from the income. For this purpose, the income categories were used. Keeping in view the goods and services in Pakistan, following income groups are made. A respondent having income less than Rs.10,000/- was categorized with low social status. While the respondent having income within Rs.10,001/- to Rs. 20,000/- was taken as medium status. In case of income higher than Rs. 20,000/-, the social status was considered high. This variable was taken as ordinal variable. Social deprivation directly affects the Urinary bladder cancer and is a significant protective factor in male and female patients (Shackley et al, 2005).

 Residential Area This variable is used for the comparison between ‘rural and urban’ and ‘industrial and non- industrial’ areas. The urinary bladder cancer is expected to be more frequent in industrial and urban areas as compared to the other areas. The diagnosis rate of urinary bladder cancer is lower in rural than in urban areas (Devita, Hellman and Rosenberg, 2001). These two variables were used as nominal variable in the analysis.

the urinary bladder cancer. The variable marital status was taken as nominal variable using the categories as unmarried = 0, married =1, widow = 2 and divorced = 3.

 Lifestyle Lifestyle is captured from the exercise and taken in three categories sedentary, normal and active and assigned codes as 0, 1 and 2, respectively. Sedentary lifestyle means that the subject has no leisure-time or physical activity (exercises, sports, physically active hobbies). A moderate lifestyle is a lifestyle in which up to 30 minutes are consumed for leisure-time with physical activity and the active lifestyle is a lifestyle in which more than 30 minutes are consumed for leisure-time or physical activity but usually athletes are taken in this category. It is expected that the people having sedentary lifestyle are more likely to have urinary bladder cancer as compared to those with moderate or active lifestyle (Ahmad and Pervaiz, 2009).

 Chemical Exposure The respondents serving as hairdressers, machinists, metal workers, printers, painters, textile workers, farmers using pesticide, cooks, truck drivers, workers in rubber, chemical and leather factories were expected to be considered more at risk of getting bladder cancer because of carcinogens in the workplace. About 20% of the bladder cancer cases were accounted for occupational exposures (Vineis and Simonato, 1991). Higher risks were investigated among painters, machinists, aluminium processors and other metal workers, leather workers, printers, hairdressers, transport workers and workers in the textile industry (Boffetta et al, 1997; Silverman et al, 1996). This variable is taken as nominal variable.

 Education The education creates the sense of awareness about the utility of exercise and better ways of living and about the risk factors of the diseases. Therefore, it becomes necessary to observe the association between the education and urinary bladder cancer. It is taken as nominal variable having two categories as literate or illiterate. The studies of

Stefani et al (2007) had shown that the cases were more educated than controls. Hence, it becomes necessary to observe the impact of education on bladder cancer.

 Family history of cancer Family history was taken as the nominal variable. A person whose first degree relative (father, mother, sister or brother) had bladder cancer is considered at higher risk of the disease. It is also considered as nominal variable. Stefani et al (2007) has showed that 4.3% cases and 1.2% controls having cancer in family history (among first-degree relatives) and a significantly higher percentage of cases as compared to the controls with family history of bladder cancer was observed.

 Personal history of cancer When a tumor in the kidney, the ureters, urethra or bladder is completely removed, then the person will be at higher risk of forming another tumor in the same or some other part of the urothelium. The variable personal history of cancer was considered as nominal variable.

 Smoking cigarettes It is believed that cigarette smokers have the higher risk of bladder cancer. When the tobacco smoke is inhaled, chemicals filter into the urine. Bladder is a container of urine. The chemicals present in the urine can cause cells in the bladder to become cancerous. Smoking cigarettes was considered as nominal variable. The more a person smokes; the risk of bladder cancer may be higher. For this purpose, the variables, the number of cigrettes smoked per days, number of years smoking cigarette and the number of year stop smoking cigarettes were included in this study and used as ordinal after converting them into categories.

 Smoking Huqqa It is also a type of inhaling the smoking of tobacco. So, it also has impact like smoking cigarettes. The huqqa smoking is very common in Pakistan and India both in males and females. In huqqa, smoke is inhaled after passing through the water. In this

study, the variable huqqa smoking was taken as nominal. The huqqa is originated from the sub - continent. Depending on locality, huqqas are well-known with the names like sheesha, nargeela, ghalyan, water pipe and many other. Rafiq (2005) conducted a study in the Nishtar Medical College Hospital Multan based on the 44 females and reported that none of the females were found to be huqqa smokers. It’s model is presented below:

Figure 3.1 Huqqa which is smoked in Pakistan

 Chew Pan Pan is referred to the leaves of the betel vine. The chewing mixture wrapped (with tobacco or without tobacco) in the Betel leaves is commonly referred as Pan. In order to observe the impact of chew pan on bladder cancer, this factor was included in the study. In Pakistan, the pan is being used with tobacco and without tobacco. It is also considered as nominal variable.

 Tea Consumption Wickremasinghe (1978) explains that black tea is the main category of tea which is produced and is most commonly used drink in the world, next to water worldwide. Wickremasinghe (1978) conducted a study and investigated that tea is the second most commonly used drink in the world, next to water. Bokuchava and Skobeleva (1980) state that black tea is major type of tea which is made from the leaves of tobacco those have been wasted before being rolled and dried. To observe the association between the tea and bladder cancer, this variable was included in this study. This variable was directly taken as nominal while the number of tea cups consumed per day was considered as ordinal variable after converting into the categories. Stefani et al (2007) stated that both coffee and tea were strongly associated with bladder cancer risk.

 Coffee Consumption In the similar way, to observe the association between the use of coffee and bladder cancer, this variable was included in this study. Stefani et al (2007) stated that coffee was strongly associated with bladder cancer risk. Sala et al (2000) updated a meta- analysis on coffee and tea consumption showed a small elevated risk of bladder cancer for current coffee drinkers and did not identify an association for tea drinkers as compared with non-drinkers. It was taken as nominal while the number of coffee cups taken daily was considered as ordinal after converting the actual variable into the categories.

 Use of Alcohol Generally, it may be considered that the use of alcohol may play a role in developing the bladder cancer. Case control studies conducted by Viscoli et al (1993) and Risch et al (1988) showed insignificant association between alcohol consumption and the risk of bladder cancer. Zeegers et al (2001) conducted a meta-analysis that had showed the insignificant risk of bladder cancer for alcohol consumption. Alcohol is also considered a variable in this study to observe its impact on the bladder cancer. This variable is taken as nominal.

 Chemotherapy and Radiation therapy The risk of bladder cancer increases with the use of anti-cancer drugs for the treatment of any disease. The treatment with the chemotherapy drug cyclophosphamide also increases the risk. This variable is taken as nominal. High energy X-rays or radiation (called radiotherapy) are used to kill the cancerous cells. The risk of bladder cancer increases when the radiation therapy is used in the pelvic area for the treatment of cervical cancer, prostate or kidney cancer. This variable is taken as nominal.

 Chronic bladder irritation The risk of a squamous cell bladder cancer increases with the long term urinary bladder infection caused by the use of catheter. Tanagho and McAninch (2008) stated that about 5% to 10% is the squamous cell carcinoma which is often associated with a history of chronic infection, vesical calculi, or chronic catheter use. The inclusion of this variable is to observing its percentage of occurrence in this sample. This variable is considered as nominal.

 Defect in bladder by birth The risk of adenocarcinoma of the bladder increases with the defects in bladder by birth. Tanagho and McAninch (2008) explained that the adenocarcinoma is about 2% of all bladder cancer cases. The inclusion of this variable is to observe its percentage of occurrence in this sample. It is also used as nominal variable in analysis.

 Fluid Consumption The risk of bladder cancer is considered to be reduced by high consumption of fluids. Because the consumption of higher amount of water reduces the concentration and contact time of urine in the bladder by frequent urination. Claus (1996) investigated that a high fluid intake was associated with a decreased incidence of bladder cancer in male, and lesser intake of daily fluids proportionally increased the risk of bladder cancer. To observe the association between the fluid consumption and urinary bladder cancer, this variable is included in the study and is taken as ordinal variable.

 Source of Drinking Water The tap water consumption having high chlorination increases the risk of bladder cancer. The four sources of drinking water that is tap, government provided, canal and pure water were included in the study and considered as the nominal variable in analysis. It is tried to observe that which source of the drinking water provides the protection against the urinary bladder cancer as compared to the other three sources. Villanueva et al (2006) conducted a study by pooling the data of 6 case-control studies and found the higher risk of bladder cancer in the users of tap water and further explained that the tap water may contain high carcinogenic chemicals.

 Hepatitis and Diabetes In this study it is tried to observe that the hepatitis has some association with the urinary bladder cancer. Is hepatitis an etiology of bladder cancer? It is taken as nominal variable. In this study it is tried to observe that the diabetes has some association with the urinary bladder cancer. Is diabetes a risk factor of urinary bladder cancer? It is taken as nominal variable.

 Hair Dye The use of hair dye is expected to have a higher risk of the urinary bladder cancer due to the chemical used in the dying colors. The use of hair dyes was investigated a risk factor for urinary bladder cancer by Gago et al (2001). This variable is taken as nominal in this study.

 Fried Items It is believed that the fried items increase the risk of bladder cancer. To observe this association, this variable was used in this study. It is considered as ordinal variable having three categories in this way: Low = taking fried items 2 or less days in a week, Normal = taking fried items 3 or 4 days in a week and Excessive = taking fried items 5 or more days in a week.

The variable ‘fried item’ is taken as the ordinal variable in which three categories low, normal and excessive are taken and assigned codes as 0, 1 and 2, respectively. Reimar et al (2004) investigated the significantly increased risk of bladder cancer only among males who were consuming the higher amount of fried food.

 Fats Items It is believed that the fats items increase the risk of bladder cancer. To observe this association, this variable was used in this study. It is considered as ordinal variable having three categories in this way: Low = taking fats items 2 or less days in a week, Normal = taking fats items 3 or 4 days in a week and Excessive = taking fats items 5 or more days in a week.

The variable ‘fats item’ is taken as the ordinal variable in which three categories low, normal and excessive are taken and assigned codes as 0, 1 and 2, respectively. Kunze et al (1993) conducted a case-control study and found the higher risk of bladder cancer in males who were consuming the higher amount of fats meals.

 Fast Food Excessive use of Fast food items were expected to increase the risk of bladder cancer. To observe this association, this variable was used in this study taking categories as Low = taking fast food 2 or less days in a week, Normal = taking fast food 3 or 4 days in a week and Excessive = taking fast food 5 or more days in a week. The variable ‘fast food’ was taken as the ordinal variable in which three categories low, normal and excessive are taken and assigned codes as 0, 1 and 2, respectively.

 Fruits Negri et al (2001) had shown that higher intake of fruits had been associated with a small but significant reduction in the risk of bladder cancer. This variable is taken in three categories that are low, normal and high. The categories are taken as Low = taking fruit 2 or less days in a week, Normal = taking fruit 3 or 4 days in a week and Excessive = taking fruit 5 or more days in a week.

The variable ‘fruits’ is taken as the ordinal variable in which three categories low, normal and excessive are taken and assigned codes as 0, 1 and 2, respectively. Negri et al (2001) found that the risk of urinary bladder cancer was inversely associated with the high consumption of fruits and vegetables.

 Profession Profession was necessary for capturing the chemical exposure. And validity of the income level was also observed through the profession. The occupations having the chemical exposure were considered and all others are taken in a category that is, others. Siemiatycki et al (2004) investigated that almost half of all recognized human carcinogens were occupational carcinogens. Profession is taken as nominal variable. In the analysis, the ‘others’ category was taken as reference category for the occupations having carcinogen at the work place.

3.13 Statistical Tests for Measuring the Association Chi-square is the most commonly used method to test the independence of the categorical variables. In case of nominal scale variables, the Phi coefficient and Cramer’s V Statistic are used. While in case of ordinal scale variables, Kendall’s Tau b test is used to measure the association between the variables. Further more, logistic regression can be used for odds ratios and their confidence intervals. Agresti (1996)

3.12.1 Contingency Table

Let A and B denote two categorical variables having r and c levels respectively. The r*c possible combinations of outcomes in a 2x2 table can be represented with r rows for the class A and c columns for the class B. The frequencies of the table denote the r*c all possible outcomes. Such a table in which the frequencies consist of aggregate of total numbers is called a contingency table. Agresti (2007)

3.12.2 Chi-Square Test for Independence The most commonly used statistical techniques in experimental work for the analysis of count or frequency data is the chi-square test. This test is mostly used to

measure the strength of association between two variables. Association refers to the degree of dependence between two categorical variables. But the presence of the association does not mean that the relationship is due to the cause and effect. If the variables of interest are declared as independent, it means that the knowledge of one subject does not able to provide the information about the other. On the other hand, if the variables of interest are found to be associated, it means that the knowledge of one variable is beneficial for the prediction of the other variable. The Pearson Chi-Square distribution for testing null hypothesis is calculated as: o  e 2  2   ij ij eij

Karl Pearson developed this technique in 1900. When all oij = eij then the statistic have zero value. It ranges from 0 to ∞. For this distribution, P-value is taken always one sided. Agresti (1996)

3.12.3 Phi-Coefficient The Phi coefficient is used as a measure of the degree of association between two attributes and is mathematically described as (ad  bc) Phi =  = (a  b)(c  d)(a  c)(b  c)

This coefficient ranges from -1 to +1. In case of dichotomous outcomes, the group differences can be compared by using the chi-square test which is equivalent to the Phi coefficient. So, using the chi square test, a rough estimate of the association between the attributes is obtained by using the Phi coefficient also. Hanif and Ahmad (2006)

3.12.5 Cramer’s V Statistic In case of more than (2x2) contingency table like (2x4) or (2x5), association of two attributes can be measured by using the Cramer’s V Statistic. This statistic is preferred over the Phi statistic which is restricted to (2x2) contingency table (Hanif and Ahmad, 2004). The Cramer’s V Statistic can be calculated as

 2 / n V = min(r 1,c 1)

3.12.5 Kendall’s Tau b A contingency table having both the categorical variables at least one on ordinal scale than the association between the variables can be measured by using the Kendall’s Tau b. This is the measure of the coefficient of concordance.

3.13 Odds, Odds Ratio and Confidence Interval The odds of an event can be defined as the probability of the event occurs divided by the probability that the event does not occur. In other words, the odds of the event E is P(E) Odds (E) = , if it is assumed that P(E)  p , then 1 P(E) p Odds (E) = 1 p The odds can take any positive value between zero and infinity. When a success is more likely than a failure, the value of odds will be greater than 1. On the other hand, when a failure is more likely than a success, the value of odds will be less than 1. This is the main reason of using the odds in case of dichotomous data. The probability of the occurrence of the event E can also be expressed in terms of the odds as:

Odds(E) P(E)  1 Odds(E)

For example, if p  0.60 , then the odds of success for the event E will be 0.60 odds(E)=  1.5 , this means that a success is 1.5 times as likely as a failure. 1 0.60 Inversely, the probability of the event E can be expressed in terms of odds 1.5 as P(E)   0.6 . Hence, when odds (E) = 1.5, then the p  0.6 . Kirkwood & 11.5 Sterne (2003) and Agresti (2007)

When two sets of data are compared, a relative measure of the odds of a success in one set of data as compared to the other is called the Odds Ratio (OR). Suppose, p1

and p2 are the success probabilities in two sets of data, so the odds of a success both data

p1 p2 sets are Z1 = and Z2 = , respectively. The ratio of the odds of a success in 1 p1 1 p2 Odds Z one set of data relative to the other is often denoted by , so that  = 1  1 is Odds2 Z 2 called odds ratio.

The value of  is always a positive number. When the two sets of data are independent, the value of  will be 1 and both p1  p2 and hence odds1= odds2. This value is considered as base line or comparison value. When the value of  is greater than one, then odds of success are higher in first set of data than the second set of data. On the other hand, When the value of  is less than one, then odds of success are less likely in first set of data than the second set of data.

If two possible states of variables are labelled exposure (success) and non- exposure (failure), then the odds ratio is a measure of the odds of a success in one group relative to the other. An example is provided that showing the exposure group and non- exposure group as under:

Case Control Total

Exposure a b a  b

Non-exposure c d c  d

Total a  c b  d a+b+c+d

It is observed from the 2 2 table that odds of exposure in cases are a and odds c of exposure in controls are b , then the odds ratio denoted by  is presented as: d

a Odds of exposure in cases ad  = = c = Odds of exposure in controls b bc d

The odds ratio is the product of two pairs of diagonal elements in the above 2 2 table, and for this reason,  is also called the ratio of the two cross- product terms (Yule, 1900, 1912). An odd ratio 1 is implied that the event is equally likely in both groups (cases and controls). An OR >1, implies that the event is more likely in cases than controls. On the other hand, an OR <1, implies that the event is less likely in cases than controls. Agresti (2007)

In order to construct the confidence interval for the true odds ratio, then the logarithm of the estimated odds ratio is better approximated by the normal distribution then the odds ratio itself, especially when the total number of binary observations is not very large. The approximate standard error of the estimated log odds ratio (log  ) is given as

 1 1 1 1  SE (log  )        a b c d 

Thus the 100(1-α)% confidence interval for log  has limits as log  + Z α/2 SE

(log  ), where Z α/2 is the upper of the (100α/2)% points of the standard normal distribution. These confidence limits are then exponentiated to give a corresponding interval for . This method of constructing the confidence interval for  ensures that both limits always are non-negative, since an odds ratio cannot be negative, as a natural requirement. Schlesselman (1982)

Because an odds ratio of unity is obtained, when the success probabilities of two sets of binary data are equal, then the null hypothesis that the true odds ratio is equal to unity, that is, Ho:  = 1 or Ho: log  = 0 can be tested using the confidence interval of odds ratio or log odds ratio, respectively.

3.14 Relative Risk

If p1 and p2 are the success probabilities of two groups in a 2×2 table then the relative risk is defined as the ratio of the success probabilities as:

Relative risk = p1 p2

If p1= p2, then the relative risk will be equal to 1.00. It means that the response group is independent of the other group. The value of relative risk may be any positive real number. The relationship between the odds ratio and relative risk is given as

p1 (1p ) 1 p  Odds ratio = 1 =  2 (Relative risk ) p2    1 p1  (1p2 )

In case of retrospective studies, conditional distributions for the explanatory variable can be constructed, within levels of the fixed response. Usually it becomes impossible to estimate the probability of the response outcome of interest. Its sampling distribution can be highly skewed unless the sample size is very large, so its confidence interval formula is rather complex. Agresti (1996)

The relative risk and odds ratio are about to be similar when the probability of the event of interest is close to zero in both groups. Due to this resemblance, the odds ratios can give the rough indication for the relative risk when it is not directly estimated, like the case control studies.

3.15 P-Value The level of significance is the probability of committing type-I error which is prefixed before the test and denoted by α. The maximum probability of rejecting the true null hypothesis is called as p-value. The P-value for the two-tailed test is taken two times of the p-value of one tailed test. The null hypothesis will be rejected, if the p-value of any test is less than α, prefixed probability of committing type one error. Hanif and Ahmad (2006)

3.16 Modelling Strategy

Several techniques like multiple regression analysis, discriminant analysis and logistic regression analysis are available in literature for modeling the response variable (urinary bladder cancer) which is presented in the form of binary (yes or no). However, some difficulties arise when the dependent variable is binary. In this situation, the necessary assumptions of multiple regression models are violated. For example, when the dependent variable is binary, it becomes awkward to assume that the distribution of errors is normal. In multiple regression analysis, the predicted values of the dependent variable can fall outside the 0 and 1 which means that they can not be interpreted as probabilities. Under the stated constraints, the multiple regression models are not feasible for modeling the binary response variable. The linear discriminant analysis permits the direct prediction of group membership, but it requires the assumption of multivariate normality of the independent variables as well as equal variance-covariance matrices in the two groups. But in case of categorical independent variables, the above stated requirements do not met, so it become difficult to apply this technique. Another multivariate technique in generalized linear models is the logistic regression model that can be used for estimating the probability and odds ratios. This model is much robust than the linear discriminant function and fewer assumptions are needed. The performance of this technique remains well even the all assumptions required for the discriminant analysis are met. Therefore, in the situation of binary response variable and a combination of continuous and categorical independent variables, the best choice is the binary logistic regression models. Hosmer and Lemeshow (2000)

In case of modeling the bladder cancer data in which the response variable is binary, the logistic regression model is considered better as compared to the discriminant analysis. In discriminant analysis, only one subject is assigned in one of two categories either case or control. But in case of logistic regression, the chances of getting a disease of a person at given exposure can be assessed. If the chances of getting the disease are less than 0.5, the event is not likely to occur but if the chances are 0.5 or more, the disease may be expected to occur. The higher the probability, the more will be the chances of the occurrence of a disease. Since this data is obtained from a retrospective

71 study, so odds ratios and confidence intervals of the odds ratios will be easy to obtain in order to explain the risk of the disease involved with the increase or decrease any risk factor.

The main quantity is the mean value of the response variable for a given value of the predictor in all regression problems. This key quantity is known as the conditional mean and expressed as E(Y/X = x) where Y is the response variable and x is the value of the explanatory variable X. In the regression analysis, it can be represented in the form of equation which is linear in X or some transformation of X or Y. Let the model is presented in the form of equation as

Y= X  β+ Є Where

t t X  = [1, x1, x2, …, xn], β = [ β0, β1, β2, …, βn], Є = [ε1,ε2, …,εn],

E(Y/X=x) = X  β and E (Є) =0.

From the behaviour of this equation, it is feasible that X can assume any values X=x). But in case of binary response׀between -∞ to + ∞ in order to find the value of E (Y variable, the conditional mean must lie from 0 to 1. As the conditional mean becomes X = x) due to the per unit change in x becomes׀closer to 0 or 1, then the change in E (Y gradually smaller. Hosmer and Lemeshow (1989)

Cox and Snell (1989) proposed binary logistic model for binary response variable due to two reasons. Firstly, it provides clinically meaningful interpretations and secondly, it is much more flexible and required fewer assumptions than other techniques like discriminant analysis. As the response variable is binary and follow the Bernoulli distribution that has the mean equal to the probability of success which is denoted by X=x) is replaced by  (x ) in logistic׀ (x) . For convenience, the conditional mean E (Y regression. The form of the logistic regression model will be

e x  x  1 e x

A transformation which is known as the central point in the study of logistic regression is called the logit transformation. The logit transformation in terms of π (x) is defined as follows:

  (x)  = x ln  1 (x)

This logit transformation is linear in its parameters and has many desirable properties of a linear regression model. The second major difference between the linear and logistic regression models concerns the conditional distribution of the outcome variable. In case of linear regression model, it is assumed that a value of the outcome variable can be expressed as y = x + є. The quantity є = y – x is called the error that follows a normal distribution with zero mean and constant variance across levels of the independent variable. The є can assume one of two possible values. If y = 0 then є = -π (x) with probability 1-π (x), and If y = 1 then є = 1-π (x) with probability π (x). Therefore, є follows a distribution with mean zero and variance equal to π (x) [1-π (x)]. That is, the conditional distribution of the outcome variable follows a binomial distribution with probability given by the conditional mean, π (x). Hosmer and Lemeshow (1989)

3.17 Fitting the Logistic Regression Model

Generally the method of least square is used to estimate the unknown parameters in the linear regression models. In this method, the sum of squares of residuals (deviation of the observed values from the predicted values) is minimized with respect to β0, β1, …, βk based on the model. The method of least squares generates the estimators under some assumptions and with some attractive statistical properties. In case of dichotomous response variable model, the estimators obtained by the method of least squares do not posses the same properties. The common procedure of estimation, when the error terms are normally distributed, that leads to the least square function under the

73 linear regression model is called Maximum likelihood. The maximum likelihood method provides estimation of parameters by maximizing the probability of getting the observed set of data. The likelihood function is required before applying this method. The likelihood function expresses the probability of the observed data as a function of unknown parameters. So the estimators obtained by this method will be very close with the observed dataset. Cox (1970)

The probability function of Bernoulli distribution is defined as

yi 1 yi f (yi )  [ (xi )] 1  (xi ) , y = 0, 1

The likely hood function for n independent observations is given as

n l( )   f (xi ) i1

n yi 1 yi l( )   [ (xi )] 1  (xi ) i1

n Ln[l( )]  y ln (xi )  (1 y)ln 1  (xi )  3.1 i1

In order to obtain the value of  , differentiate Ln[l( )] with respect to β’s and putting the resulting equations to zero. The resulting equations called likelihood equations are presented below:

Lnl( ) = 0 3.2 

In case of logistic regression model, the equations generated in (3.2) are not linear in parameters. Hence the values of β’s obtained from the solution of equations (3.2)

 are known as the maximum likelihood estimates and denoted by  . McCullagh and Nelder (1989) explained the solution of the equations (3.2) by using an iterative re- weighted least square method. Hosmer and Lemeshow (1989)

3.18 Tests of Significance After the estimation of the parameters, the next step is to test the significance of the variables in the model. This step is consisted on the formulation and testing of the statistical hypothesis in order to investigate whether the predictors in the model are significantly related to the dependent variable. In a logistic regression model, the comparison of observed values to the predicted values is based on the log likelihood function presented in the equation (3.1). This comparison can be expressed in the following way:

(likelihood of the fitted model) D 2 ln 3.3 (likelihood of the saturated model)

A model in which number of parameters is equal to the sample size is called saturated model. The purpose of using -2 times the log of likelihood ratio is to get a quantity of known distribution which can be used for testing of hypothesis. This test is known as likelihood ratio test. Using equation (3.1), the D will be

n  ˆˆii   Dy2 ii ln  1  y ln1    3.4 i1 yyii   Where D is called deviance of the model which plays the same role as Sum of Squares of Errors (SSE) plays in simple linear regression model. Further more, when the response variable is binary having values 0 or 1 then the likelihood of the saturated model will be equal to one. Hence the equation (3.3) reduces as

D =  2 ln(likelihood of the fitted model) The change in D due to introducing the predictors in the model is as follows:

G = D (Model without predictors) D (Model with predictors)

The likelihood of the saturated model is common to both values of the deviance whose difference is taken to calculate the value of G as:

G = 2 ln [(likelihood without predictors) (likelihood with predictors)]

2 G > χ (α , n-p) , Conclude that the fitted model is not adequate.

The null hypothesis is that the slope coefficients in the model are equal to zero and the G statistic follows the chi-square distribution with n-p degrees of freedom. Hosmer and Lemeshow (1989)

Further more, the two other statistically equivalent tests, Wald test and Score test, are used for testing the significance of the variables. These two tests are required the same assumption as the likelihood ratio test. In the logistic regression model, the Wald test is defined by the ratio of the maximum likelihood estimate ˆ (of the slope parameter), to its standard error. The resultant ratio under the null hypothesis β = 0, follows a standard normal distribution. Therefore, the Wald test for the logistic regression model is given as

ˆ W  i ˆ SE ( i )

ˆ Z  ASE

ˆ Where  1 in (3.4) is the maximum likelihood estimate of β1 and ASE is the asymptotic standard error of ˆ . Its calculation is not much difficult in case of single variable. But in case of data set having large number of variables, the iterative procedures are required in order to attain the maximum likelihood estimates. The standard normal table of z can be used to obtain one-sided or two sided p-values. Alternatively, Z2 follows

76 a chi-squared distribution with one degrees of freedom also produce the similar results. Bishop, Fienberg and Holland (1975)

Menard (1995) indicated that the value of the Wald test (chi- square) will be decreased if the coefficients are large and standard error are also large. Agresti (1996) explained that the likelihood ratio test is more appropriate as compared to the Wald test when sample size is small.

3.19 Goodness of Fit Measures

After fitting the model, its appropriateness can be observed. If the observed and the expected values are close enough, the fitted model is considered as a good one, otherwise, further investigation must be made. The adequacy of the fitted model is known as goodness of fit. In linear regression model, R2 is a measure that explains how well the model fit the data. In logistic regression, R2 is an attempt to measure the strength of the association between the response variable and the predictors. But this measure cannot be explained like the R2 in the ordinary regression models. It seems to be alike in range but both are explained differently.

Generally, three types of R2 can be used in the logistic regression models which are Pearson, Cox & Snell and Nagelkerke. These three R2 used to measure the strength of association between the dependent variables and predictors are discussed below.

3.19.1 Pearson test

The Pearson test is used to assess the goodness of fit. This test follows approximately the chi-square distribution. The degrees of freedom for this test are obtained by the number of covariates minus the number of parameters to be estimated. The test is given below:

2 n   2 ( y  yˆ )    i i    Var  i1  i 

Where and are the observed and the predicted response respectively and Var is yi yˆ i i the estimated variance of the response variable. Hosmer and Lemeshow (1989)

2 3.19.2 Cox & Snell' R cs

2 The Cox & Snell' R cs (1989) is based on the log likelihood of the intercept only model and the log likelihood of the full model is presented below.

2 / N 2    1 L(M Intercept ) Rcs    L(M full ) 

This statistic can never attain its maximum value 1. (Hosmer and Lemeshow, 1989)

3.19.3 Nagelkerke’s R2

2 Nagelkerke (1991) has suggested the correction in the Cox & Snell's R cs which is given below.

 2  2    Rcs R  2 / N  1   L(M Intercept ) 

All the stated measures are some what the same and similar to R2 in 2 interpretation. This model can attain its maximum value one but the Cox & Snell's R cs can never attain its maximum value one. Hosmer and Lemeshow (1989)

3.20 Verifications of the Model Assumptions

Before going to build the logistic regression model, the required assumptions including multicollinearity and autocorrelation are needed to be verified. If these

78 assumptions are not met then the model will not be able to test the variables properly due to the rise in the standard errors of the parameters. Similarly, the outliers are also necessary to be determined.

3.20.1 Detection of Autocorrelation

The independence of the errors is judged by using the Durbin-Watson test. The d -statistic is defined as

n 2 (ei  ei1) i2 d  n 2 ei i1

If d  2 (No autocorrelation), If d < 2 (Positive autocorrelation)

If d > 2 (Negative autocorrelation) If 1.5 < d < 2.5, then according to the rule of thumb, the errors are considered independent. Other wise there is some sort of autocorrelation. Chatterjee and Hadi (2006)

3.20.2 Detection of Multicollinearity

For detecting the multicollinearity, the Variance Inflation Factors (VIF) are generally, used. The Variance Inflation Factors measure how much the variances of the estimated regression coefficients are inflated, as compared to when the predictors are not linearly related. The VIF,s are found as

1 VIF  2 for i=1, 2, 3, …, k. 1  R i 2 Where R i is the multiple coefficient of determination found by regressing xi on the th remaining k-1 predictors. If VIFi is greater then 10, then i regressor will be collinear with the remaining explanatory variables. Chatterjee and Hadi (2006)

3.21 Model Diagnostics

The diagnostics about influential observations and the outliers should be employed before fitting the suitable model. These diagnostics are made by using the Cooks’ distance for influential observations, deviance residuals and the standardized residuals for outliers, respectively. The graphically approach for the purpose can also be used. Ghias and Pervaiz (2009) had used the model diagnostics to analyze the data for the identification of epidemiological risk factors for Hepatitis C in Punjab, Pakistan.

3.21.1 Outlier Observations having larger standardized residuals are considered outliers in the response variable. Such observations lie far away from the fitted model. The standardized residuals larger than 2 or 3 standard deviation are called outliers. The outliers are the potentially bad values and these values can also distort the results of the model and can change sign of the model parameters. The standardized residuals are defined as

e d  i i MSresi

Where ei is ith residual and MSresi is the residuals mean square. In order to investigate the outliers, the standardized and the deviance residuals are used. If any absolute value in the standardized or deviance residual is greater than 3, then that value is considered as an outlier in the data. Montgomery et al (2004)

3.21.2 Influential Observation An influential observation is any observation which has large effect on the parameter estimates of the model, if omitted from the regression. Often, influential variables are outliers or leverage points. Influential observations are detected generally by using the cook’s distance. The cook’s d-statistic is defined as

2  i hii Di  . P (1 hii )

1 Where hii = X i(X X ) X i , hii is the diagonal elements of the hat matrix, P is rank

2 of hat matrix and  i is a square of the ith Studentized residual. If any value of the Cooks’ distance is greater than one then that observation will be considered as influential. Montgomery et al (2004)

3.21.3 Hosmer-Lemeshow goodness-of-fit Test Hosmer and Lemeshow (1989) proposed a statistic for the goodness-of-fit used in logistic regression, particularly for models with continuous covariates and studies with small sample sizes. It is based on grouping cases into deciles of risk and comparing the observed probability with the expected probability within each decile. The Hosmer- Lemeshow goodness-of-fit statistic is obtained by calculating the Pearson chi-square statistic from the 2×g table of observed and expected frequencies, where g is the number of groups. The statistic is written n (O  E ) 2 H  g g  g1 N g g (1 g )

Where Og, Eg, Ng, and πg denote the observed events, expected events, observations, predicted risk for the gth risk decile group, and n is the number of groups. Hosmer & Lemeshow (2000)

3.21.4 Omnibus Test It is traditional chi-square test and is an alternative to the HL test. It tests if the model with the predictors is significantly different from the model with only the intercept. If the test is significant, a research conclude that there is adequate fit of the data i.e., at least one of the predictors is significantly contributing to the response variable.

CHAPTER No. 4

RESULTS AND DISCUSSIONS

The study is based on the 900 subjects including 300 cases and 600 controls of age 38 years and above. Both male and female are considered in this sample. About 35 factors are taken. These risk factors are based on the demographic, social, clinical and dietary habits of the subjects. Among these factors only six are taken as quantitative i.e., age, number of cigarettes smoked per day, cigarette smoking period (in years), stop smoking period (in years) and number of tea and coffee cups taken per day. All the other variables are taken as qualitative either nominal or ordinal. For presenting the complete and comprehensive analysis, this section is divided into two major parts, descriptive and analytical.

4.1 Descriptive Analysis The descriptive analysis has been made in order to discuss the frequency of occurrence risk factors of the urinary bladder cancer. The discussion of the results is based on the counts, percentages and averages. The gender wise number of cases and controls taken from hospitals of provincial headquarters and federal city are presented in the Table 4.1. The hospitals from headquarter of each province and federal area are selected due to the fact that the proper facilities of the treatment of urinary bladder cancer are not available in the interior areas of each province and the people of interior areas have to attend the headquarter hospitals for the treatment of this horrible disease which is symbol of death.

From Table 4.1, it is observed that the overall male and female in the sample are 738 (82%) and 162 (18%), respectively. The counts (percentages) of male patients selected from Lahore, Islamabad, Peshawar, Quetta and Karachi are 35%, 17%, 15.9%, 15.9% and 16.2%, respectively. Similarly, the counts (percentages) of female patients selected from Lahore, Islamabad, Peshawar, Quetta and Karachi are 26%, 15%, 20%,

20% and 19%, respectively. The lowest percentage of female patients is found in Islamabad. The two controls are taken against one case i.e., the case control ratio is 1:2.

From Table 4.1, it is also observed that the counts (percentages) of male and female patients (cases) in the sample are 246 (80%) and 54 (20%), respectively. The male /female (cases) ratio in the sample is 4.5:1. In the studies of Puente et al (2003), male/female ratio was 6.4:1. For the comparison purpose, 492 male and 108 female are taken as controls against the 246 male and 54 female cases.

Table 4.1 City and Gender wise Classification of Data

City Bladder Cancer Female Male Bladder Cancer No Yes No Yes Female Male Female Male Lahore 200 100 42 258 28 172 14 86 Islamabad 100 50 24 126 16 84 8 42 Peshawar 100 50 33 117 22 78 11 39 Quetta 100 50 33 117 22 78 11 39 Karachi 100 50 30 120 20 80 10 40 Total 600 300 162 738 108 492 54 246

The counts of each factor are presented in the form of bivariate tables considering cases and controls. The counts and percentages are used to explain the risk factors of the urinary bladder cancer. The individual descriptive results are presented in Table 4.2 on page 94. The findings are discussed as below:

 Gender Among 300 patients of urinary bladder cancer, the counts (percentages) of males and females are 246 (82%) and 54 (18%), respectively. Among 600 controls, males and females are 482 (82%) and 108(18%), respectively. The disease is found more frequent in males than in females. Among 162 female, the counts (percentages) of cases and controls are 54 (33.33%) and 108 (66.67), respectively. Silverman and Devesa (2006) explained that the urinary Bladder cancer occurred more frequently in males as compared to the

83 females with male / female ratio of 4:1, in most of the western countries. The descriptive results are similar to the results of studies of Rabbani and Cardo (2000), Silverman and Devesa (2006) and Sylvester (2004), in which the frequency of the disease is showed higher in males as compared to females.

 Age The age is taken as continuous variable. Devita, Hellman and Rosenberg (2001) explained that the risk of bladder cancer goes up with age and the urinary Bladder cancer is rarely diagnosed before the age of 40 years. The overall study found that the minimum and maximum ages of the patients were 38 and 89 years, respectively. The mean, median and mode age was 57, 56 and 60 years, respectively. The disease was found to be more frequent in the age of 60 years. The number of patients above the age of 65 year was 62 (19.7%).

In case of female patients, the minimum and maximum ages of the patients were 39 and 75 years, respectively. The mean, median and mode ages were 53.2, 50.5 and 45 years, respectively. The disease was found to be more frequent in the age of 45 years. Only 5.6% female patients had age more than 65 years. Similarly, in case of male patients, the minimum and maximum ages of the patients were 38 and 89 years, respectively. The mean, median and mode ages were 58, 58 and 60 years, respectively. The disease was found to be more frequent in the age of 60 years. Only 59(24%) male out of 246 male patients has age more than 65 years. Hence, it is concluded that the higher frequencies of bladder cancer in males and females were found in the ages of 60 and 45 years, respectively.

Different researcher consider the age in different ways in their studies. Puente et al (2006) pooled the data of the 14 case–control studies on bladder cancer from Europe and North America, the age was taken from 30 to 79 years in groups categories as 30-39, 40- 49, 50-59, 60-69, 70-79 to study the risk of bladder cancer between gender by smoking and the maximum patients was found in the age group 60-69. Ferlay et al (2004) observed that the two third of all bladder caner cases had their ages above 65 years.

 Social Status The social status of the cases and controls is directly taken from the income of the individuals. Three categories of the income are used for the social status. The three categories explaining social status are low, medium and high having income less than Rs. 10,000/-, Rs.10,000/- or more but less than Rs.20,000/- and Rs.20,000/- or above, respectively. The numbers (percentages) of the total subjects who are falling in low, medium and high status are 553 (61.44%), 291 (32.33%) and 56 (6.22%), respectively. In case of patients of urinary bladder cancer, the counts (percentages) in low, medium and high status are 209 (69.67%), 78(26%) and 13(4.33%), respectively. The counts (percentages) in low, medium and high status in controls are 344 (57.33%), 213(35.5%) and 43(7.17%), respectively. It is observed from the descriptive results that the income of patients is very low because the income of 70% cases is below Rs.10,000/- which is falling in the low status. The percentage of low income category is much higher than that of controls. Only 4.33% cases and 6.22% controls are falling in the high status. Stefani et al (2007) stated that the cases were observed to be more educated and earned more incomes as compared to controls.

 Residential Area The residential area is studied in two dimensions, Industrial and Non-industrial and Rural and urban. The description of results is given as below.

i. Industrial and Non-Industrial area Among all subjects, 145 (16.11%) belong to the industrial areas while 775 (83.89%) belong to the non-industrial areas. Further more, 53 (17.67%) of all cases and 92 (15.33%) of all controls belong to the industrial areas. While 247 (82.33%) of all cases and 508 (84.67%) of all controls belong to the non-industrial areas. The counts (percentages) of the cases from non-industrial and industrial area are 247 (32.71%) and 53 (36.55%), respectively. Thus, the percentage of the patients from industrial area is higher as compared to non-industrial area. In industrial areas, the chemicals exhausted from the industries generally contaminate the drinking water and environment. The exhausted chemicals from the industries may be carcinogenic.

ii. Rural and urban area The counts (percentages) of the subjects those who are living in urban and rural areas are 474 (52.67%) and 426 (47.33%), respectively. In urban areas, the count (percentage) of cases and controls are 157 (33.12%) and 317 (66.88%), respectively. Similarly from the rural areas, the controls and cases are 283 (66.43%) and 143 (33.56%), respectively. Out of total 300 cases, the number (percentage) belonging to the urban and rural areas are 157 (52.33%) and 143 (47.67%), respectively. The disease is more frequent in urban areas than in rural areas.

Devita, Hellman and Rosenberg (2001) stated that the diagnosis rate of urinary bladder cancer was higher in urban areas as compared to the rural areas. In urban areas, the smoke of all vehicles, pollutant environment and other unhygienic condition may cause to create the carcinogen. While rural areas are free from such problems. In the studies of Momas et al (1994), the percentages of cases belong to the urban and rural areas are 82.5% and 17.5%, respectively and similarly, the percentages of controls belong to the urban and rural area are 77% and 23%, respectively. The percentages of cases belong to the urban area is much higher than the controls.

 Marital status This variable was used to study the association between the marital status and the urinary bladder cancer in the population of Pakistan. It is tried to investigate how the categories of the marital status married, unmarried, widow or divorced were affecting the urinary bladder cancer. In this study, among the whole sample of 900 subjects, only 4 (0.44%) subjects are found to be unmarried and 896 (99.56%) are married. The occurrence of the disease is found in the age of 38 year and above. Therefore, only four cases are unmarried but all the controls are found to be married. None of the subject is reported to be widow or divorcee. A very small percentage of the subjects are found to be unmarried which indicates that the marital status does not play a role in developing the bladder cancer.

 Lifestyle Lifestyle is taken from the exercise and taken in three categories sedentary, normal and active. None of the case or control reported about the third category in the overall sample of 900 subjects. Hence, two categories that is sedentary (no exercise) and normal (about 30 minute exercise daily) are observed and used for analysis. The number (percentage) of the subjects having sedentary and normal lifestyle are 681 (75.67%) and 219 (24.33%), respectively. The number (percentage) of patients having sedentary and normal lifestyle are 266 (88.67%) and 34 (11.33%), respectively. Similarly, in case of controls, the number (percentage) having sedentary and normal are 415 (69.17%) and 185 (30.83%), respectively. The percentage of the patients having normal lifestyle is very low as compared to the controls. A huge percentage (88.67%) of patients is observed without exercise that can lead to the subjects to the diseases.

 Chemical Exposure The chemical exposure is captured from the occupations having carcinogen at their work place. The counts (percentages) of the subjects having and not having the chemical exposure at the work places in the overall sample of 900 subjects are 254 (28.22%) and 646 (71.78%), respectively. In case of patients (cases), the numbers (percentages) of the subjects having and not having the chemical exposure at the work places are 123 (41.0%) and 177 (59.0%), respectively. Out of the 254 subjects having chemical exposure at their work place, 123 (48.4%) belong to cases while 131(51.6%) belong to the controls. On the other hand, the counts (percentages) of the subjects out of 300 cases and 600 controls are 123(41.0%) and 131(21.83%), respectively. The percentage of the cases having chemical exposure at their work place is about double than in controls that showing the higher association among chemical exposure and bladder cancer. Vineis and Simonato (1991) found that about 20% of the bladder cancer cases were accounted for occupational exposures. Higher risks were investigated among painters, machinists, aluminium processors and other metal workers, leather workers, printers, hairdressers, transport workers and workers in the textile industry by Boffetta et al (1997) and Silverman et al (1996).

 Educational status Among the whole sample of 900 subjects, the count (percentage) of the illiterate and the literate are 563 (62.56%) and 337 (31.44%), respectively. The numbers (percentages) of the illiterate and the literate in patients are 197 (65.67%) and 103 (34.33%), respectively. The counts (percentages) of the illiterate and the literate in controls are 366 (61%) and 234 (39%), respectively. The literacy level in patients (cases) is very low as compared to controls. The education creates the sense of awareness about the utility of exercise, better ways of living and about the risk factors of the diseases. The studies of Stefani et al (2007) had shown that the cases were more educated than controls. Momas et al (1994) states that 88.9%, 8.2% and 2.9% of the cases have primary, secondary and higher education, respectively. Similarly, 89%, 7.8% and 3.2% of the controls have the primary, secondary and higher education, respectively. Studies of Stefani et al (2007) contradicts with this study and Momas et al (1994). On the other hand, similarities were found in the results of this study and Momas et al (1994), showing that the controls have better education as compared to the cases.

 Family history of cancer The count (percentage) of the subjects with and without family history of cancer are 857 (95.22%) and 43 (4.78%), respectively. Among the 300 cases, only 22 (7.33%) are found to have cancer in their family history. While Among the 600 controls, only 21 (3.5%) are reported that they have cancer in their family history. The percentage of patients with family history of cancer is higher as compared to the percentage of controls. Stefani et al (2007) has showed that 4.3% cases and 1.2% controls having cancer in family history (among first-degree relatives) and a significantly higher percentage of cases with family history of bladder cancer was observed with odds ratio and 95% confidence interval 4.97 and (1.7– 14.3), respectively. Similarities in the results of Stefani et al (2007) and this study was investigated.

 Personal history of cancer When a tumor in the kidney, the ureters, urethra or bladder is completely removed, then the person will be at higher risk of forming another tumor in the same or

88 some other part of the urothelium. In this sense, personal history of cancer can play in developing the urinary bladder cancer. In the overall sample of 900 subjects, none of the case or control is reported that he has cancer in his personal history of any type. In other words, it is interpreted that none of the subject has been the patient of cancer in his life. Therefore, personal history of cancer is not playing any role in developing the bladder cancer.

 Huqqa Smoking It is also a type of inhaling the smoking of tobacco. The huqqa smoking is very common in Pakistan and India both in males and females. In huqqa, smoke of tobacco is inhaled after passing through the water. In the whole sample of 900, neither of the case nor any control is reported that he is a huqqa smoker. Although huqqa smoking is very common in Pakistan but none of huqqa smoker is observed in this sample. If the smokes of huqqa are playing any role in the occurrence of bladder cancer then there must be some patients like the cigarette smokers. Rafiq (2005) conducted a study in the Nishtar Medical College Hospital Multan based on the 44 females and reported that none of the females were found to be huqqa smokers.

 Chew Pan Only three subjects are found to be the pan chewers in the overall sample of 900 subjects. Among the 300 cases, only one (0.33%) case reported that he is pan chewer and only two (0.33%) controls out of 600 controls reported that they are pan chewers. The counts (percentages) of the pan chewers are the same in cases and controls that is 0.33%, which means that there is no association between chew pan and bladder cancer. Chew pan is considered problematic for mouth cancer both in male and female.

Rafiq (2005) conducted a study in the Nishtar Medical College Hospital Multan based on the 44 females and reported that none of the females were found to be smokers of huqqa , beera and pan but 47% females who were coming from rural areas had reported the history of intake of niswar or beera and pan.

and not taking tea are 821 (91.22%) and 79 (8.78%), respectively. This sample states that more than 90% of the people in Pakistan are the tea drinkers. Wickremasinghe (1978) explains that tea is the second most commonly used drink in the world, next to water.

In case of 300 patients, the counts (percentages) of the tea drinker and non drinker are 277 (92.33%) and 23 (7.67%), respectively. The counts (percentages) of the tea drinker and non drinker in 600 controls are 544 (90.67%) and 56 (9.33%), respectively. The counts (percentages) of the cases and controls in tea consumers are 277(92.33%) and 544 (90.67%), respectively. The percentage of tea consumers in patients is slightly higher as compared to controls. Slattery et al (1988) suggested a positive association in tea drinkers and urinary bladder cancer. Sala et al (2000) conducted a meta-analysis on tea consumption and no association was established in tea drinkers and bladder cancer.

 Use of Coffee All the subjects in the sample are reported that they are not coffee consumers. In other words, none of the case or control of coffee consumer is observed. Its reason may be the hot weather of Pakistan. In Pakistan, weather remains hot maximum time in the year. Therefore, the coffee consumption is not common in this country. Morrison (1984), Ciccone and Vineis (1988) and Slattery et al (1988) studies investigated that the coffee drinking is an etiology of the urinary bladder cancer.

Villanueva et al (2006) stated that high intake of coffee (more than 4 cups per day) was observed to increase the risk of bladder cancer. Sala et al (2000) updated a meta-analysis for coffee and tea consumption showing a small elevated risk of bladder cancer for current coffee drinkers and did not identify an association for tea drinkers as compared with non-drinkers.

 Use of Alcohol The numbers (percentages) of subjects who are the consumer of alcohol are 0 (0%). All the subjects in the sample are reported that they are not alcoholic. Being Islamic country, alcohol is prohibited for Muslims (Constitution of Pakistan 1973, Article

37-h). Case control studies conducted by Viscoli et al (1993) and Risch et al (1988) showed insignificant association between alcohol consumption and the risk of bladder cancer. Zeegers et al (2001) conducted a meta-analysis that had showed the insignificant risk of bladder cancer for alcohol consumption having odds ratio and 95% confidence interval (1.3; 0.9-2.0) for males and (1.0; 0.6-1.7) for females.

 Radiation and Chemotherapy Therapy None of the subject in the sample has taken any radiation therapy prior to this disease. The radiation therapy is not given even to a single subject in his personal history. Similarly, none of the subject in the sample has taken any chemotherapy prior to this disease. The chemotherapy is not given even to a single subject in his personal history.

 Chronic Bladder Irritation The numbers (percentage) of subjects having chronic bladder irritation are 0 (0%). None of the subject in the sample is found a patient of squamous cell carcinoma which is caused by chronic bladder irritation or chronic catheter use. About 5 to 10% is the SCC, which is often associated with a history of chronic infection, vesical calculi, or chronic catheter use (Tanagho and McAninch, 2008).

 Defects in Bladder by Birth The numbers (percentage) of subjects having defects in bladder by birth are 0 (0%). None of the subject in the sample is found with Adenocarcinoma that is caused by defects in bladder by birth. Tanagho and McAninch (2008) stated that adenocarcinoma is about 2% of all bladder cancer cases. But in this sample, no patient of adenocarcinoma is reported from any hospital.

 Fluid Consumption The counts (percentages) of the persons (both cases and controls) who are consuming less than 10 glasses of water and 10 or more glasses of water per day are 345 (38.33%) and 555 (61.67%), respectively. The counts (percentages) of the cases who are consuming less than 10 glasses of water and 10 or more glasses of water per day are 187

(62.33%) and 113 (37.67%), respectively. In the similar way, the counts (percentages) of the controls who are consuming less than 10 glasses of water and 10 or more glasses of water per day are 158 (26.33%) and 442 (73.67%), respectively.

Claus (1996) investigated that a high fluid intake was associated with a decreased incidence of bladder cancer in male, and lesser intake of daily fluids proportionally increased the risk of bladder cancer. Keeping in view the results of these studies, it is suggested that higher amount of fluid intake is a protective measure against the bladder cancer.

 Source of Drinking Water For the purpose of drinking water, four sources are available in Pakistan i.e., tap water, canal water, government provided water and mineral water. In the whole sample of 900 subjects, none of the subject either case or control is using the canal water and mineral water in his daily life. Only the two sources for drinking water are used in this country. Among 900 subjects, 578 (64.22%) are drinking the tap water and other 322 (35.78%) are drinking the government provided water. Among 300 cases, the counts (percentages) of the consumers of tap water and government provided water are 200 (67.67%) and 100 (33.33%), respectively. Among 600 controls, the counts (percentages) of the drinkers of tap water and government provided water are 378 (63%) and 222 (37%), respectively.

Villanueva et al (2006) conducted a study by pooling the data of 6 case-control studies and found the higher risk of bladder cancer in the users of tap water and further explained that the tap water may contain high carcinogenic chemicals.

 Diabetes and Hepatitis The numbers (percentage) of subjects having diabetes are 0 (0%). None of the patient of bladder cancer is found to be the patient of diabetes. While the controls are taken as healthy person or the persons free from the diseases related to the urology and cancer. Hence this situation is an indication of no association between the diabetes and

92 bladder cancer. None of the patient of bladder cancer is found to be the patient of hepatitis in this sample. While the controls are taken as healthy person or the persons free from the diseases related to the urology and cancer. Hence it can be concluded that there is no association between the hepatitis and bladder cancer.

 Hair Dye The counts (percentages) of the subjects those who dye their hair and not dye are 840 (93.33%) and 60 (6.67%), respectively. This sample shows that less than 7% of the people in Pakistan are in the habit of hair dye. In case of 300 patients, the counts (percentages) of the hair dye and non hair dyes are 29 (9.67%) and 271 (90.33%), respectively. In the similar way, out of 600 controls, the counts (percentages) of the hair dye and non hair dyes are 31 (5.17%) and 569 (94.83%), respectively. Percentage of the cases that dye hair is about double than in controls which is showing an association between hair dye and bladder cancer. The use of hair dyes was investigated a risk factor for urinary bladder cancer by Gago et al (2001).

 Fried Items The count (percentage) of the fried items using low and medium categories in the overall sample were 664(73.8%) and 236(26.2%), respectively. The count (percentage) of the fried items using low and medium categories for cases only were 196(65.3%) and 104(34.7%), respectively. In the similar way, the count (percentage) of the fried items using low and medium categories in controls only were 445(74.2%) and 155(25.8%), respectively. None of the subject reported about the use of third category i.e., high (using fried items more than four days within a week). Reimar et al (2004) investigated the significantly increased risk of bladder cancer only among males who were consuming the higher amount of fried food.

 Fats Items The count (percentage) of the fats items using low and medium categories in the overall sample were 641(71.2%) and 259(28.8%), respectively. The count (percentage) of the fats items using low and medium categories for cases only were 187(62.3%) and

113(37.7%), respectively. In the similar way, the count (percentage) of the fats items using low and medium categories in controls only were 477(79.5%) and 123(20.5%), respectively. None of the subject reported about the use of third category i.e., high (using fats items more than four times within a week).

Riboli et al (1991), Hebert et al (1994), Bruemmer et al (1996) and Radosavljevic´ et al (2005) found that the high consumption of fats, especially animal fats, can increase the risk of urinary bladder cancer. Kunze et al (1993) conducted a case- control study and found the higher risk of bladder cancer in males who were consuming the higher amount of fats meals.

 Fast Food In this sample, most of the people/ subjects are very poor and belong to the low status families. So, in this category of the subjects, the use of fast food is not much common. The count (percentage) of the fast food using low and medium categories in the overall sample were 846(94.0%) and 54(6.0%), respectively. There are only 6% subjects who are using the fast food 3 or 4 times within a week. All other were using the fast food 2 times or less within a week. The counts (percentages) of the fast food using low and medium categories in cases only were 283(94.3%) and 17 (5.7%), respectively. In the similar way, the count (percentage) of the fast food using low and medium categories in controls only were 563(93.8%) and 37(6.2%), respectively. None of the subject reported about the use of third category i.e., high (using fast food more than four days within a week).

 Fruits In literature, fruit is a protective factor for the risk of the urinary bladder cancer. The count (percentage) of the fruits using low and medium categories in the overall sample were 449(49.9%) and 451(50.1%), respectively. The counts (percentages) of the fruits using low and medium categories for cases only were 216(72%) and 84 (28%), respectively. In the similar way, the count (percentage) of the fruits using low and medium categories in controls only were 238(38.8%) and 367(61.2%), respectively. None

94 of the case/control reported about the use of third category i.e., high (using fruits more than four days within a week). Negri et al (2001) found that the risk of urinary bladder cancer was inversely associated with the high consumption of fruits and vegetables.

 Cigarette Smoking The counts (percentages) of the smokers and non smokers in the sample of 900 subjects were 305(33.9%) and 595(66.1%), respectively. In 600 controls, the counts (percentages) of smokers and non smokers were 105(17.5%) and 495(82.5%), respectively. The counts (percentages) of smokers and non smokers in cases were 200(66.7%) and 100(33.3%), respectively. In 595 non smokers, only 100 (16.8%) were the bladder cancer cases. While in 305 smokers, the bladder cancer cases were 200(65.6%). This situation is an evident that the cigarette smokers are more at risk than non smokers. According to Wynder and Goldsmith (1971), cigarette smoking accounted for 50% and 31% of bladder cancers in males and females, respectively.

Table 4.2 Classification of Cases/ Controls with different risk factors Factors Categories Bladder Cancer No Yes Total Gender Female 108 54 162 Male 492 246 738 Social Status less than 10000 344 209 553 10000-20000 213 78 291 20000 and above 43 13 56 Residential Area_1 Non-industrial 508 247 755 Industrial 92 53 145 Residential Area_2 urban 317 157 474 Rural 283 143 426 Marital Status Unmarried 0 4 4 Married 600 296 896 Lifestyle Sedentary 415 266 681

Normal 185 34 219 Chemical Exposure No 469 177 646 yes 131 123 254 Education No (Illiterate) 366 197 563 Yes (Literate) 234 103 337 Family history of cancer No 579 278 857 yes 21 22 43 Personal history of cancer No 600 300 900 yes 0 0 0 Huqqa smoking No 600 300 900 yes 0 0 0 Chew pan No 598 299 897 yes 2 1 3 Use of tea No 56 23 79 yes 544 277 821 Use of coffee No 600 300 900 yes 0 0 0 Use of alcohol No 600 300 900 yes 0 0 0 Chemotherapy No 600 300 900 yes 0 0 0 Radiation therapy No 600 300 900 yes 0 0 0 Chronic bladder irritation No 600 300 900 yes 0 0 0 Defect in bladder by birth No 600 300 900 yes 0 0 0 Fluid consumption (Glasses) Less than 10 glasses 158 187 345 10 or More glasses 442 113 555 Source of water(drinking) Tap 378 200 578

canal 0 0 0 Govt. Provided 222 100 322 Mineral 0 0 0 Hepatitis No 600 300 900 yes 0 0 0 Diabetes No 600 300 900 yes 0 0 0 Hair dye No 569 271 840 yes 31 29 60 Fried item Low 477 187 664 Normal 123 113 236 Excessive 0 0 0 Fats item Low 445 196 641 Normal 155 104 259 Excessive 0 0 0 Fast food Low 563 283 846 Normal 37 17 54 Excessive 0 0 0 Fruits Low 233 216 449 Normal 367 84 451 Excessive 0 0 0 Cigarettes Smoking No 495 100 595 yes 105 200 305

 No. of Cigarettes Smoked Per day The count (percentage) of the non smokers, smokers of 1 to 10, 10 to 20 and 20 or more cigarettes per day in the over all sample of 900 respondents were 595(66.1%), 200(22.2%), 65(7.2%) and 40(4.4%), respectively. The count (percentage) of the non smokers, smokers of 1 to 10, 10 to 20 and 20 or more cigarettes per day in the 300 bladder cancer patients were 100(33.3%), 102(34.0%), 62(20.7%) and 36(12.0%), respectively. Similarly, in 600 controls, the count (percentage) of the non smokers,

97 smokers of 1 to 10, 10 to 20 and 20 or more cigarettes per day were 495(82.5%), 98(16.3%), 3(0.5%) and 4(0.7%), respectively. From the Table 4.3, it is observed that as the number of cigarettes smoked per day increase, the percentage of the bladder cancer patients also increases as compared to the controls. Hence an increase in the number of cigarette smoking and percentage increase in urinary bladder cancer cases are directly proportionate.

The studies of the Pohlabeln et al (1999) states that the factors number of cigarettes smoked per day, cigarette smoking period, and the number of years stop smoking were found to be significant with the time and dose response in case of males. The risk of bladder cancer becomes higher when the number of cigarettes smoked per day and cigarette smoking period increases and risk of bladder cancer decreases with the increase in the quitting smoking period.

Table 4.3 Counts and Percentages of the Number of Cigarettes Smoked Per day No. of Cigarettes Bladder Cancer

Smoked Per day No yes Total

0 495(83.2%) 100(16.8%) 595(100.0%)

1-10 98(49.0%) 102(51.0%) 200(100.0%)

10-20 3(4.6%) 62(95.4%) 65(100.0%)

20 and above 4(10.0%) 36(90.0%) 40(100.0%)

 No. of years of Cigarette Smoking The count (percentage) of the non smokers, smokers of the period from 1 to 15, 15 to 30 and 30 or more years in the over all sample of 900 respondents were 595(66.1%), 139(15.4%),116(12.9%) and 50(5.6%), respectively. The count (percentage) of the non smokers, smokers of the period from 1 to 15, 15 to 30 and 30 or more years in the patients of the urinary bladder cancer were 100(33.3%), 55(18.3%),99(33.0%) and 46(15.3%), respectively. In the similar way, The count (percentage) of the non smokers, smokers of the period from 1 to 15, 15 to 30 and 30 or more years in 600 controls were

495(82.5%), 84(14.0%), 17 (2.8%) and 4(0.7%), respectively. From the Table 4.4, it is observed that as the years of cigarette smoking increase, the percentage of the bladder cancer patients also increases as compared to the controls. Hence increase in years of cigarette smoking and percentage increase in urinary bladder cancer cases are directly proportionate.

The studies of the Pohlabeln et al (1999) states that the cigarette smoking period, and the number of years stop smoking were found to be significant with the time and dose response in case of males. The result of Pohlabeln et al (1999) is similar to this study. The risk is associated with the number of cigarettes smoked per day, the duration of smoking and the amount of inhalation of smoke. Contributing agents in the cigarette smoke are considered the alpha and beta naphthylamine, which are coming in urine of smokers (Viscoli et al, 1993).

Table 4.4 Counts and Percentages of Cigarette Smoking Period No. of years of Bladder Cancer

Cigarette Smoking No yes Total

0 495(83.2%) 100(16.8%) 595(100.0%)

1-15 84(60.4%) 55(39.6%) 139(100.0%)

15-30 17(14.7%) 99(85.3%) 116(100.0%)

30 and above 4 (8.0%) 46(92.0%) 50(100.0%)

 No. of years of Stop Smoking In this study, the smokers and non smokers were 305 and 595, respectively. Out of 305 smokers, 185 smokers were never quitted smoking up to the study period. While the counts (percentages) for the period of stop smoking 1 to 5 years, 5 to 10 years and 10 or more years in the overall sample were 50(5.6%), 29(3.2%) and 41(4.6%), respectively. The counts (percentages) for the period of stop smoking 1 to 5 years, 5 to 10 years and 10 or more years for the cases(patients) were 34(11.3%), 13(4.3%) and 15(5.0%), respectively. The counts (percentages) for the period of stop smoking 1 to 5

99 years, 5 to 10 years and 10 or more years for the controls were 16(2.7%), 16(2.7%) and 26(4.3%), respectively. From Table 4.5, it is observed that as the stop smoking period increases, then the percentage of the bladder cancer cases decreases which means that there is inverse relationship between the stop smoking period and the percentage of the bladder cancer cases. On the other hand, the percentage of controls increases as the stop smoking period increases.

A case-control study conducted by Brennan et al (2001) found that stop smoking period gradually decreased the risk of bladder cancer. For the stop smoking period of 1 to 4 year, the 32% decrease was observed with odds ratio 0.68. The results of this study are supported by the studies of Brennan et al (2001). Quitting smoking period reduces the chances of the bladder cancer as compared to the current smokers.

Table 4.5 Counts and Percentages of Stop Smoking Period

No. of years of Stop Bladder Cancer Smoking No yes Total 0 542(69.5%) 238(30.5%) 780(100.0%) 1-5 16(32.0%) 34(68.0%) 50(100.0%) 5-10 16(55.2%) 13(44.8%) 29(100.0%) 10 and above 26(63.4%) 15(36.6%) 41(100.0%)

4.2 BIVARIATE ANALYSIS To observe the association between the individual factors and the bladder cancer, this analysis is being performed. The Chi square, Phi/ V statistics, Kendall's Tau-b are performed to observe the significance of the association. Factor will be considered significant if p-value is less than 0.05. If the factor is found to be significant, it means that the factor is associated with disease. Otherwise, the disease is not affected by the individual factor.

From Table 4.6, it is observed that the risk factors cigarette smoking, family history of cancer, hair dye, source of drinking water and profession are significantly

100 associated with urinary bladder cancer both by chi-square and Phi-statistic. The association between the gender and the bladder cancer is found to be insignificant with p- value equal to one which means that there is no association between the gender and bladder cancer. The factors cup of tea, lifestyle, fluid consumption, fried item, fats item, use of fruits, social status, number of cigarettes to be smoked per day, smoking period and stop smoking period are significant risk factors both by chi-square test and Tau-b.

The negative value of Tau-b explains that inverse association between the factor and bladder cancer risk although this reduction in risk may be small. According to this study, no one was found who use coffee and alcohol. Weather in Pakistan remains hot maximum time in the year. Therefore, coffee is not common in Pakistan due to hot season. Being Islamic country, alcohol is prohibited for Muslims. Case control studies conducted by Viscoli et al (1993) and Risch et al (1988) showed insignificant association between alcohol consumption and the risk of bladder cancer. Zeegers et al (2001) conducted a meta-analysis that had showed the insignificant risk of bladder cancer for alcohol consumption having odds ratio and 95% confidence interval (1.3; 0.9-2.0) for males and (1.0; 0.6-1.7) for females.

The association between social status, consumption of fruits, fluid consumption and lifestyle are found to be negative which means that these factors are the protective factors for the disease. While the factors having positive associations with the disease are the risk factors. The Kendall’s Tau-b for the factor stop smoking period is also negative which indicates that as a cigarette smoker stops further smoking then stop smoking period becomes inversely associated with the urinary bladder cancer. The negative association of the factors explained that as the level of these factors increases, the chance of getting the disease deceases. On the other hand, the nine factors including family history of cancer, cigarette smoking, tea cups to be taken per day, hair dye, use of fried items, use of fats item, and number of cigarettes smoked per day, the smoking period and professions are positively associated with the disease which means that these factors and bladder cancer are directly related.

101

Table 4.6 Chi-Square, Phi/ V and Kendall's Tau-b Statistics S. No. Variables Chi-Square V/Phi- p-value Tau-b p-value Statistic 1. Gender 0.000 0.000 1.00 - -

2. Area industrial 0.806 0.030 0.369 - -

3. Area Rural 0.020 0.005 0.887 - -

4. Education 1.859 -0.045 0.173 - -

5. Family history 6.46 0.085 0.011 - -

6. Cigarette smoking 215.8 0.49 0.000 - -

7. Chew Pan 0.000 0.000 1.00 - -

8. Tea 0.694 .028 0.405 - -

9. Tea cups 24.83 - 0.000 0.135 0.000

10. Hair Dye 6.509 0.085 0.011 - -

11. Lifestyle 41.304 0.000 -0.214 0.000

12. Fried items 30.465 - 0.000 0.184 0.034

13. Fats items 7.614 - 0.006 0.092 0.007

14. Fast-food 0.089 - 0.766 -0.010 0.763

15. Fruits 88.003 0.000 -0.313 0.000

16. Fluid taken 109.6 0.000 -0.349 0.000

17. Source of Drinking 1.17 -0.036 0.279 - - water 18. Social Status 13.114 - 0.001 -0.118 0.000

19. No. of cigarettes 54.977 - 0.000 0.393 0.000 Smoked per day 20. No. of years of 77.195 - 0.000 0.463 0.000 cigarettes Smoking 21. Stop Smoking Period 34.374 0.000 -0.251 0.000

22. Profession 81.559 0.301 0.000

*The bold factors are found to be significant at 5% level.

102

4.3 Model Building Before going to build the logistic regression model, the required assumptions including multicollinearity and autocorrelation are needed to be verified. If these assumptions are not met then the model will not be able to test the variables properly due to the rise in the standard errors of the parameters. Similarly, the outliers are also necessary to be determined.

4.3.1 Detection of Multicollinearity For detecting the multicollinearity, the variance inflation factors (VIF) are used. If the variance inflation factor is greater than 10, then there is a problem of multicollinearity. Otherwise the regressors are independent. From Table 4.7, it is observed that all the VIF are less than 10. So, there is no problem of multicollinearity and all the regressors are independent. Montgomery et al (2004)

Table 4.7 Tolerance and Variance Inflation Factors (VIF) for the Detection of Multicollinearity Factors Tolerance VIF Age 0.915 1.093 Social Status 0.778 1.285 Chemical Exposure 0.928 1.078 Industrial Area 0.539 1.854 Rural Area 0.477 2.095 Lifestyle 0.748 1.337 Education 0.733 1.364 Family history of cancer 0.899 1.112 Cigarettes Smoking 0.814 1.228 Use of tea 0.982 1.018 Hair dye 0.880 1.137 Fluid consumption (Glasses) 0.888 1.126 Source of water(drinking) 0.591 1.691 Fried item 0.871 1.148

103

Fats item 0.715 1.399 Fast food 0.805 1.242 Fruits 0.773 1.293

4.3.2 Detection of Autocorrelation One of the assumptions of binary logistic regression is that the successive error terms should be independent. According to the rule of thumb, if the value of d-statistic lies outside the range 1.5 to 2.5 then there is a problem of autocorrelation that is the successive error terms are dependent. From this study, it is observed that the values of d- statistics is 2.4 and lies in the above range. Hence, there is no problem of the autocorrelation and the residuals are independent. Chatterjee and Hadi (2006)

4.4 Diagnostics of the Model Before fitting the suitable model, the diagnostics about influential observations and the outliers should be employed. These diagnostics are made by using the Cooks’ distance for influential observations, deviance residuals and the standardized residuals for outliers, respectively. The graphically approach for the purpose can also be used. Ghias and Pervaiz (2009) had used the model diagnostics to analyze the data for the identification of epidemiological risk factors for Hepatitis C in Punjab, Pakistan.

4.4.1 Influential observations The influential observations are the potential outliers called bad values and these values can distort the model and can change even the sign of the model parameters. The influential observations have great effect on the model summary and parameters both. In order to investigate the influential observations, the Cook’s distance is used. If any value of the Cooks’ distance is greater than one then that observation will be considered as influential. In this data set, no influential observation is detected because all the values of the Cook’s distance are less than unity. From Figure 4.1, it is observed that there are no influential or bad values in the model because none of the observation is outside the unity. Montgomery et al (2004)

104

Figure 4.1 Graphs of Influential Values in the Model

4.4.2 Outliers The outliers are the potentially bad values and these values can also distort the results of the model and can change sign of the model parameters. In order to investigate the outliers, the standardized and the deviance residuals are used. If any absolute value in the standardized or deviance residual is greater than 3, then that value is considered as an outlier in the data. From Figures 4.2, 4.3, 4.4 and 4.5, it is observed that none of the standardized residuals or the deviance residuals has any absolute value greater than 3, so there is no outlier. After examining the data, the suitable logistic regression model is run for the purpose of analysis. Montgomery et al (2004)

105

Figure 4.2 Graphs of Standardized Residuals against ID

Figure 4.3 Graphs of Standardized Residuals against Predicted Probabilities

106

Figure 4.4 Graphs of Deviance Residuals against ID

Figure 4.5 Graphs of Deviance Residuals against Predicted Probabilities

107

4.5 Multiple Logistic Regression Model Under this model, regression coefficients, odds ratio, p-value and 95% confidence intervals for odds ratios are computed. The p-value is compared with the predefined value of alpha (5%) for the significance of the variables. The interpretations of the significant risk factors are given in term of the odds ratio and 95% confidence interval of the odd ratio. For the adequacy of the model, Omnibus test and Hosmer and Lemeshow (HL) test are used.

The Omnibus test with χ2 =452.42 is significant at P-values =.000 showing that at least one of the factor is significantly affecting the response variable i.e., there is adequate fit of data to the model. This test is the alternate of the HL test. The HL test is much better, when sample size is small and some variables are continuous, than other traditional chi-square tests (Hosmer and Lemeshow, 1989). If HL test is non-significant, then the model is adequately fit the data. In this study, HL test is non-significant with χ2 =7.02 at P-values =0.427 indicates that the model is adequately fitted. For the goodness of fit, the Cox & Snell R Square and Nagelkerke R Square has been used which are 0.40 and 0.549, respectively.

From Table 4.8, it is observed that 536(89.3%) controls and 209(69.7%) cases are correctly predicted. But 91(29.3%) cases and 64 (10.7%) controls are misclassified 91(29.3%) as controls and 64 (10.7%) as cases, respectively. The overall numbers (percentages) of correctly classified and misclassified subjects (including cases and controls) are 745 (82.8%) and 155(17.2%), respectively. The amount of correct classification is much higher, so the fitted model is adequate. From the correct classification, Omnibus test and HL test, it is observed that the model is adequately fit the data and hence, the 95% confidence intervals are also valid for inferences. Higher strength of the association is observed by Cox & Snell R Square and Nagelkerke R square among the predictors and the response variable.

108

Table 4.8 Correct and Incorrect Classification of the Data

observed Predicted Bladder Cancer Percentage No Yes Correct Bladder Cancer No 536 64 89.3 Yes 91 209 69.7 Overall percentage 82.8

From Table 4.9, the significant risk factors and their predictive strengths are observed. Nine risk factors including Hair dye (HD), Chemical Exposure (CE), Lifestyle, Family History (FHC), cigarette smoking (CS), Fluid Consumption (FC), Fried items, Fats items and Fruits are found to be significant. The logit model is given below:

Z = -0.592 + 1.085* HD + 0.953*CE-2.279*(Lifestyle) +1.141* FHC+2.358*CS - 1.318*FC+0.747* Fried +0.732*Fat – 1.232*Fruit

The six factors including Hair dye (HD), Chemical Exposure (CE), Family History (FHC), Cigarette Smoking (CS), Fried items and Fats items are found to be positively significant which means that these risk factors and bladder cancer are directly related. Same results are observed in the bivariate analysis of these factors. On the other hand, the three factors including lifestyle, fluid consumption and use of fruits are found to be negatively significant which means that these factors and bladder cancer are inversely related.

4.6 Interpretations of the model coefficients and odds ratios From Table 4.9, the interpretations of model coefficients, odds ratios and 95% confidence intervals are discussed and compared with the other studies conducted in different parts of the world. It is being observed how the risk factors and the strength of the risk factors vary from different areas.

109

 Hair dye This study investigated that the subjects who dye hair have about three times more risk of urinary bladder cancer as compared to those who never dye hair. The 95% confidence interval for the odds ratio of the hair dye is (1.396, 6.279) which shows that the effect of hair dye is significant as confidence interval does not include one. Gago et al (2001) were investigated that the use of hair dyes is a risk factor for urinary bladder cancer. This study supports the results of the study of the Gago et al (2001). But the meta-analysis of 10 published case control studies by Takkouche et al (2005) did not confirm this relationship which had the relative risk and 95 % confidence interval for these combined studies for ever users of hair dyes 1.01 and (0.89-1.14), respectively. Hence, these combined results of the 10 studies are contradicting the results of this study and the study of Gago et al (2001). So, further studies are required to investigate the relationship between bladder cancer and hair dye.

 Chemical Exposure This study found that the subjects having chemical exposure at their work place have 2.59 times more risk of urinary bladder cancer as compared to other professions. The 95% confidence interval for the odds ratio of the chemical exposure is (1.46, 4.607) which shows that the effect of chemical exposure is significant because its confidence interval does not include one. Kunze et al (1993) found that workers in printing, plastics, rubber, mining, dyestuffs industries, oils, petroleum, exposure to paints and pesticides, truck drivers had significantly risks of urinary bladder cancer as compared to the other professionals.

The studies of Pelucchi et al (2002) found that the occupation exposure was associated with the risk of bladder cancer. Especially, the workers including dyes or paints factories, chemical factories and in pharmaceutical industries had three times more risk of urinary bladder cancer as compared to the workers related to the other occupations. In the similar way, Mommsen (1983) indicated that industrial workers, petroleum workers, working with oil and chemicals materials had a significantly increased relative risk of developing bladder cancer. Hence the results of Mommsen

110

(1983), Kunze et al (1993) and Pelucchi et al (2002) are similar to the results of this study.

Kogevinas et al (2003) were conducted a case control study in 6 European countries Germany, France, Italy, Spain, Greece and Denmark from 1976 to 1996. In their study, it was investigated that the occupations including knitters, automobile painters, machinists, automobile mechanics and textile machinery mechanics had odds ratios 2.56, 1.95, 1.5, 1.38 and 2.86, respectively. It is concluded that the results of this study are supported by Mommsen (1983), Kunze et al (1993), Pelucchi et al (2002) and Kogevinas et al (2003).

 Family History of Cancer The family history of cancer is investigated as the significant risk factor with odds ratio and 95% confidence interval of odds ratio 3.130 and (1.325, 7.394), respectively. This study states that the subjects having cancer in their family members (first degree) have 3.13 times more risk of bladder cancer as compared to those who have not cancer in their family history.

Kunze et al (1993) conducted a case-control study and investigated that the family history of bladder cancer was positively significant in male. Hence the results of this

111 study are supported by the study of Kunze et al (1993). Stefani et al (2007) has shown that 4.3% cases and 1.2% controls having cancer in family history (among first-degree relatives) and a significantly higher percentage of cases with family history of bladder cancer was observed with odds ratio and 95% confidence interval 4.97 and (1.7– 14.3), respectively.

 Cigarette Smoking In this study, it was observed that the cigarette smoking is the highest risk factor of the bladder cancer as compared to all other risk factors. According to this study, out of 305 smokers, 200 (about 66%) smokers are the patients of bladder cancer. Kogevinas and Trichopoulos (2000), Silverman et al (1996) and Brennan et al (2001) found that the cigarette smoking is a major risk factor that generates the urinary bladder cancer about 50 to 65% in males and 20 to 30% in females. The results of this study are supported by the studies of Kogevinas and Trichopoulos (2000), Silverman et al (1996) and Brennan et al 2001. In this study, the odds ratio and 95% confidence interval for the cigarette smokers are 10.569 and (7.007, 15.941), respectively. It shows that the cigarette smokers in Pakistan have 10 times more risk of bladder cancer as compared to the non smokers.

Samanic et al (2006) planned study in Spain to investigate the effect of smoking on bladder cancer and found that the current smoker’s males and females were at high risk. The odds ratio and 95% confidence interval for current smoker in males were 7.4 and (5.3-10.4), respectively. While in female were 5.1and (1.6-16.4), respectively. The results of the Samanic et al (2006) are similar to this study. Puente et al (2006) pooled primary data from 14 case–control studies of bladder cancer from Europe and North America and found that the odds ratios for current smokers compared to nonsmokers were 3.9 for males and 3.6 for females. These studies are also supporting the results of this study.

 Fluid Consumption In this study, the fluid consumption is observed to be negatively significant with odds ratio 0.268 and 95% CI of odds ratio (0.183-0.392). It means that a person who

112 consumes 10 or more glasses of water per day has 0.268 times chance of getting disease (i.e., 73% protection against the disease) as compared to the person who consumes the less than ten glasses of water per day. Hence, consumption of more water is the protection against the disease. Excess of water in the bladder reduces the concentration and stay time of chemicals by frequent urination.

The result of fluid consumption is being supported by the findings of the Hermann et al (1999). They had observed the odds ratio and 95% confidence interval of odds ratio for the females who consumes more than two liters fluid per day were 0.34 and (0.11-0.99), respectively. Claus (1996) stated that high fluid intake was associated with a decreased risk of bladder cancer in males, and lesser intake of daily fluids proportionally increases the risk of bladder cancer.

 Fried Items The use of fried items is investigated as the significant with odds ratio and 95% confidence interval of odds ratio 2.112 and (1.364, 3.269), respectively. A subject who is using fried items more than two days per week have 2.112 times more risk of bladder cancer as compared to those who consume the fried items 2 or less than two days per week. The logit coefficient of fried items (0.747) is positively significant with p-values =0.001 which means that the more use of fried items and bladder cancer are directly associated with the increased risk of bladder cancer. Reimar et al (2004) investigated the significantly increased risk of bladder cancer only among males who were consuming the higher amount of fried food.

 Fats Items The use of fats items is investigated as the significant with odds ratio and 95% confidence interval of odds ratio 2.080 and (1.309, 3.305), respectively. It is observed that the subjects using fats items more than two days per week have 2.08 times more risk of bladder cancer as compared to those who consume the fried items 2 or less than two days per week. The logit coefficient of fried items (0.732) is positively significant with p-

113 values = 0.002 which means that the more use of fried items is directly associated with the increased risk of urinary bladder cancer.

Riboli et al (1991), Hebert et al (1994), Bruemmer et al (1996) and Radosavljevic´ et al (2005) explained that high consumption of fats, especially animal fats, can increase the risk of urinary bladder cancer. Kunze et al (1993) conducted a case- control study and found the higher risk of bladder cancer in males who were consuming the higher amount of fats meals.

Radosavljevic et al (2001) investigated that consumption of animal fats have increased risk of the urinary bladder cancer. These findings are also supported by Riboli et al (1991), Kunze, et al (1993), Hebert et al (1994), Bruemmer et al (1996), Radosavljevic et al (2001) and Radosavljevic´ et al (2005).

 Fruits The use of fruits more than 2 days per week is found to be negatively significant with the odds ratio and 95% confidence interval of odds ratio 0.292 and (0.193, 0.440), respectively. It is observed that the subjects who are using fruits more than two days per week have 0.292 times risk of bladder cancer that is 71% protection against the disease as compared to those who consume the fruits 2 or less than two days per week.

The risk of urinary bladder cancer was inversely associated with the high consumption of fruits and vegetables (Negri et al, 2001). This study is also supported by Negri et al (2001). Pohlabeln et al (1999) conducted a case-control study in West Germany based on the 300 cases of bladder cancer of the lower urinary tract (LUT) which were separately matched to controls from the same hospitals by considering their age, sex and residential area. Dietary routine were evaluated by a food frequency list having six categories of salad, raw carrots, and fruit. The six categories were: (5-7) times per week, (2-4) times per week, (about one) time per week, (2-3) times per month, (Once or less) per month and never. On the basis of the analysis, the authors observed that frequent intake of salads, raw carrots and fruits was established the inverse relationship

114 with the urinary bladder cancer risk. Hence, they investigated that the salads, raw carrots and fruits had the protective effect against the bladder cancer risk. The odds ratio for the users of fruits more than once per week investigated by Pohlabeln et al (1999) was 0.65. It means that the users of the fruits more than once per week have 35% protection against the urinary bladder cancer as compared to those who use fruits once or less per week.

Table 4.9 Model Coefficient, Odds Ratios and 95% CI’s for Odds Ratios Factors β S.E. Wald Sig. Exp(β) 95.0% C.I for EXP(β) Lower Upper HD 1.085 0.3848.006 0.005 2.961 1.396 6.279 CE 0.953 0.29310.571 0.001 2.594 1.460 4.607 LS(Exercise) -2.279 0.309 54.512 0.000 0.102 0.056 0.187 FHC 1.141 0.4396.773 0.009 3.130 1.325 7.394 CS 2.358 0.210126.444 0.000 10.569 7.007 15.941 FC -1.318 0.19545.792 0.000 0.268 0.183 0.392 FRIED 0.747 0.22311.245 0.001 2.112 1.364 3.269 FAT 0.732 0.2369.601 0.002 2.080 1.309 3.305 FRUIT -1.232 0.21034.427 0.000 0.292 0.193 0.440 Constant -0.592 0.184 10.394 0.001 0.553

115

4.7 Occupational Risk Factors Like the factors cigarette smoking, family history, fried and fats items, etc., the occupations adopted by the subjects having chemical exposure at their work place also play a major role in developing the bladder cancer. The studies of Vineis and Simonato (1991) explained that about 20% of the bladder cancer cases were accounted for occupational exposures. Boffetta et al (1997) and Silverman et al (1996) stated that higher risks had been commonly investigated among painters, machinists, aluminum processors and other metal workers, leather workers, printers, hairdressers, transport workers and workers in the textile industry.

In order to investigate the occupational risk factors, both the descriptive and the analytical measures are used. For the analytic purpose, the binary logistic regression model was run in order to find the odds ratios and 95% confidence interval of the odds ratios by using the software SPSS (Version-16).

4.7.1 Descriptive Analysis The mainly considered occupations are cook, driver, farmer, painter, metal worker, textile worker and others (occupations not having the chemical exposure at work place) in order to investigate their effect on bladder cancer. The frequencies of these occupations with percentages are given in Table 4.10.

From Table 4.10, it is observed that the percentage of the leather workers is very small in the sample that is 0.7% only. While the maximum 71.8% of the subjects fall in the category having no chemical exposure at their work place (others). The maximum subjects having chemical exposure at their work place fall in the category of the farmers (14.1%). Being the agricultural country and about 80% of the population is busy in this field. Due to this reason, the percentage of the farmers in this sample is higher as compared to the other occupations. The percentages of cooks, drivers, farmers, painters, leather workers, metal workers, textile workers and others in the cases are 7%, 3.1%, 14%, 1.0%, 1.0%, 6.7%, 6% and 59%, respectively. While the percentages of cooks,

116 drivers, farmers, painters, leather workers, metal workers, textile workers and others in the controls are 3%, 1.2%, 14.2%, 0.7%, 0.5%, 1.2%, 3.7% and 78.2%, respectively.

In the similar way, the overall percentages in the sample for cooks, drivers, farmers, painters, leather workers, metal workers, textile workers and others are 2.1%, 3.1%, 14.1%, 0.8%, 0.7%, 3.0%, 4.4% and 71.8%, respectively. Generally, the numbers of painters and leather workers are very small in the general population as compared to the other occupations. Therefore, the percentage of the painters and the leather workers is also very small in this sample that is 0.8 and 0.7, respectively. Large population of the country is earning their livelihood from the textile industry. So, the percentage of the textile workers is higher in the sample after the farmers.

Table 4.10 Classification of Subjects in Different Occupations with Percentages

Occupations Bladder cancer Percentage of

No yes every Total occupation Cook 3(1.2%) 16(7%) 19 2.1 Driver 7(1.2%) 21(3.1%) 28 3.1 Farmer 85(14.2%) 42(14%) 127 14.1 Painter 4(0.7%) 3(1.0%) 7 0.8 Leather Worker 3(0.5%) 3(1.0%) 6 0.7 Metal Worker 7(1.2%) 20(6.7%) 27 3.0 Textile Worker 22(3.7%) 18(6.0%) 40 4.4 Others 469(78.2%) 177(59.0%) 646 71.8 Total 600(66.7%) 300(33.3%) 900 100

117

4.7.2 Multiple Logistic Regression Model

Under this model, regression coefficients, odds ratio, p-value and 95% confidence intervals for odds ratios are calculated. The p-value is compared with the predefined value alpha (5%) for the significance of the variables. The interpretations of the significant risk factors are given in terms of the odds ratio and 95% confidence interval of the odds ratio. For the over all significance of the model, Omnibus test is used.

The Omnibus test with χ2 =73.97 is significant at P-values =.000 showing that at least one of the category is significantly affecting the response variable i.e., there is adequate fit of data to the model. For the goodness of fit, correct classification is used. The overall correct classified subjects are 71.1%. The amount of correct classification is much more so the fitted model is adequate.

From Table 4.11 (given at the end of this subsection), it is observed that the four categories of the occupations including cooks, drivers, metal workers and textile workers are found to be significant. While the farmers, painters and leather workers are observed to be insignificant in Pakistan. According to this study, the cooks are at higher risk of bladder cancer as compared to all other occupations.

 Cooks The odds ratio and the 95% confidence interval for the odds ratio in cooks are found to be 14.132 and (4.068, 49.088), respectively. According to this study, cooks have 14.13 times higher risk of bladder cancer as compared to the all other professions those do not have chemical exposure at the work place. The confidence interval of odds ratio do not include the value one. Kogevinas et al (2003) conducted a case control study by pooling the primary data of the 11 studies conducted in Germany, France, Italy, Spain, Greece and Denmark and found that the cooks had insignificant risk of bladder cancer with odds ratio and 95 % confidence interval 1.19 and (0.91-1.55), respectively. It is possible that the cooks in the stated countries do not have to face the problem of smoke and excessive heat at the cooking place like the Pakistani cooks.

118

Lund and Borgan (1987) investigated that the standardized mortality ratio values in cooks is much higher for lung cancer as compared to the other working groups in Denmark and Norway and higher relative risk was observed in the insulation workers as compared to others in Sweden.

 Drivers The odds ratio and the 95% confidence interval for the odds ratio in drivers are found to be 7.949 and (3.321, 19.025), respectively. According to this study, drivers have highest risk of bladder cancer. The risk of bladder investigated in drivers is 7.95 times more as compared to the all other professions those do not have chemical exposure at the work place. The reason of developing the bladder cancer in drivers is the maximum stay time of urine (chemicals and carcinogens wasted by the kidneys) in the bladder. Busy in long journey, they stop their urine for long time that becomes the problematic.

Mommsen et al (1983) found that the industrial workers, workers related to petroleum, oil and chemicals materials had a significantly increased relative risk of developing bladder cancer. Kunze et al (1993) found that workers in printing, plastics, rubber, mining, dyestuffs industries, oils, petroleum, truck drivers, exposure to paints and pesticides had significantly higher risk of urinary bladder cancer. The results of Mommsen et al (1983) and Kunze et al (1993) are supporting this study.

 Metal Workers In this study, the metal workers are found to be significant with odds ratio and 95% confidence interval 7.571 and (3.147, 18.214), respectively. This result states that the metal workers have 7.57 times more risk of bladder cancer as compared to the all other professions those do not have chemical exposure at their work place.

Kunze et al (1993) were found that workers in printing, plastics, rubber, mining, dyestuffs industries, oils, petroleum, exposure to paints and pesticides, truck drivers had significantly risks of urinary bladder cancer as compared to the other professionals. Kogevinas et al (2003) has shown the significant risk of bladder cancer in metal workers.

119

Reimar et al (2004) also found the statistically significant odds ratios and 95% CI in males for the primary metal workers and auto mechanics were (2.40; 1.29-4.50) and (1.69; 1.02-2.82), respectively.

 Textile Workers The odds ratio and the 95% confidence interval for the odds ratio in textile workers are found to be 2.168 and (1.136, 4.138), respectively. The risk of bladder cancer in the textile workers is 2.16 times more as compared to the all other professions those do not have chemical exposure at their work place.

Kogevinas et al (2003) conducted a study in the 6 European countries (Germany, France, Italy, Spain, Greece and Denmark) from 1976 to 1996 and observed that the occupations like metal workers, painters, miners, excavating-machine operators, transport operators, textile and electrical workers have the higher risks of bladder cancer as compared to the occupations. Mommsen et al (1983) found that the industrial workers, petroleum workers, oil and chemicals materials workers had a significantly higher relative risk of developing bladder cancer. The results of this study are supported by Mommsen et al (1983) and Kogevinas et al (2003).

Table 4.11 Model Coefficients with Odds Ratios and their 95% CIs for Different Occupations

Occupations B S.E. Wald Sig. Exp(B) 95.0% C.I for EXP(B) Lower Upper Cook 2.648 .635 17.378 .00014.132 4.068 49.088 Driver 2.073 .445 21.677 .0007.949 3.321 19.025 Farmer .269 .208 1.675 .1961.309 .871 1.969 Painter .687 .769 .798 .3721.987 .440 8.968 Leather Worker .974 .821 1.408 .235 2.650 .530 13.251 Metal Worker 2.024 .448 20.423 .000 7.571 3.147 18.214 Textile Worker .774 .330 5.504 .019 2.168 1.136 4.138 Constant -.974 .088 122.021 .000.377

120

4.8 Area- wise Descriptive Analysis In this section, area-wise descriptive analysis approach is accomplished for the purpose of comparisons in different areas of Pakistan. From Table 4.12, area-wise descriptive explanations for the risk factors of the urinary bladder cancer are given as below.

 Gender The percentage of female patients of bladder cancer in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 14%, 16%, 22%, 22% and 20%, respectively. Similarly, for males the percentages are 86%, 84%, 78%, 78% and 80%, respectively. In Punjab, and Islamabad, the percentage of the female patient is less than the Khyber Pukhtoon Khwa, Baluchistan and Sindh. Its reason may be the lack of education, exercise, poor status and low use of fruits especially in females as compared to the Punjab and Islamabad. The percentage of female patients is least in Punjab as compared to Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh. The male / female ratio in the patients of bladder cancer in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 6.1:1, 5.5:1, 3.6:1, 3.6:1 and 4:1, respectively.

 Residential Area The residential area is studied in two dimensions, Industrial and Non-industrial area and Rural and urban area. The description of results is given as below. a) Industrial and non-industrial area The percentage of the patients of bladder cancer in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh coming from the industrial areas are 48%, 04%, 0%, 0% and 6%, respectively. It is indicated that the maximum patients from the industrial areas belong to the Punjab and none of the patient belong the industrial area from Khyber Pukhtoon Khwa and Baluchistan. Its main reason is that the mostly industrial zones are in the Punjab as compared to all other areas. The 45% and 55% subjects in the overall sample (including cases and controls) of Punjab belong to the

121 industrial and non-industrial area, respectively. But in case of Sindh, the 4.67% and 95.33% subjects in the overall sample (including cases and controls) belong to the industrial and non-industrial area, respectively. Industrial areas in Khyber Pukhtoon Khwa and Baluchistan are very rare as compared to the Punjab and Sindh. b) Rural and Urban Areas The counts (percentages) of the subjects who are living in urban and rural areas in Punjab are 136 (45.33%) and 164 (54.67%), respectively. In urban areas, the count (percentage) of cases and controls are 48 (48%) and 88 (44%), respectively. The percentage of the patients of bladder cancer in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh belonging to the urban areas are 48%, 50%, 34%, 54% and 80%, respectively. Similarly from rural areas, the percentages of patients are 52%, 50%, 66%, 46% and 20%, respectively. Only in Sindh and Baluchistan, rate of bladder cancer patient is higher in urban areas than in rural areas. Similar findings were reported by Devita, Hellman and Rosenberg (2001). They reported that the diagnosis rate of urinary bladder cancer is higher in urban areas than in rural areas.

 Marital Status The counts (percentage) of married and unmarried patients in Punjab are 296(98.67%) and 4 (3.33%), respectively. All the controls are found to be married. Out of 100 patients, only 4 (4%) are reported to be married. In the all other areas including Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh, none of the case or control is found to be unmarried. Its main reason is that the minimum age for the disease is found to be 38 years. So, in this age group, generally all cases and controls become married. The unmarried patients present in the sample may have some medical or domestic problem.

 Education The percentages of the illiterate patients of bladder cancer in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 55%, 72%, 82%, 82% and 48%, respectively. The maximum percentage of the illiterate patients is 82% found in the Khyber Pukhtoon Khwa and Baluchistan. In these two areas, 22% female

122 and 78% males (in each area) are found to be the patients of bladder cancer and very rare female are allowed to study. On the other hand, the percentages of the literate patients of bladder cancer in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 45%, 28%, 18%, 18% and 52%, respectively. The highest percentage of the patients (52%) is found to be literate in Sindh. Its reason is that the maximum patients (about 80%) belong to the urban areas (in Karachi and Hyderabad).

 Family History of Cancer The counts (percentage) of the patients who are having and not having cancer in their family history in Punjab are 22 (22%) and 78 (78%), respectively. Out of 100 patients, only 22% are reported that they have cancer in their family history (first degree relatives). From the 200 controls, only 21(10.5%) have reported the cancer in their family history. In the all other areas including Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh, family history of cancer is not reported. In the sample of 300 subjects from Punjab, 14.33% have cancer in their family history.

 Lifestyle The number (percentage) of the subjects in this sample having ‘sedentary’ and ‘normal’ lifestyle belonging to Punjab are 148(49.33%) and 152 (50.67%), respectively. The percentage of patients ‘sedentary and ‘normal’ in Punjab are 80% and 20%, respectively. Similarly, the percentages of controls in this sample ‘sedentary and ‘normal’ are 34% and 64%, respectively. In the sample of Islamabad, 20% patients have normal lifestyle while other 80% sedentary. Hence the tendency of exercise in patients is similar in Punjab and Islamabad. In Sindh, none of the case or control reported that they have normal lifestyle. In the sample of Baluchistan, every patient present in the sample reported that he has sedentary lifestyle but only 2% controls reported the normal lifestyle in their daily life.

 Social Status The social status is categorized as low, medium and high captured from the income groups as ‘less than Rs.10,000/-’, ‘Rs. 10,000/- to Rs.20,000/-’ and ‘Rs.20,000/-

123 and above’, respectively. But in the overall sample of 900 subjects, third category contains only 6% subjects. The mean, median and mode of the income are Rs.11,299/-, Rs.10,000/- and Rs. 6,000/-, respectively. The minimum and maximum income in the sample is Rs. 2,500/- and Rs. 50,000/-. Keeping in view the mean and median of the income, the two groups of income are only used for analysis that is ‘less than Rs. 10,000/-’ and ‘Rs.10,000/- and above’ for low and medium status, respectively.

In the sample of Punjab, the percentages of the patients having low, medium and high status are 77%, 18% and 5%, respectively. It means that very few subjects have high status. The percentages of the patients having low, medium and high status in the sample of Islamabad are 62%, 36% and 2%, respectively. The situation of high status is also same as in Punjab. While the percentages of the patients having low, medium and high status in the sample of Khyber Pukhtoon Khwa are 70%, 26% and 4%, respectively. But the sample of Baluchistan does not contain any patient having high status that is income more that Rs.20,000/-. and the percentages of the patients having low and medium status are 78% and 22%, respectively. Hence the income of patients is very low in Baluchistan as compared to the Punjab, Islamabad and Khyber Pukhtoon Khwa. From the sample of Sindh, it is observed that the percentages of the patients having low, medium and high status in the sample are 54%, 36% and 10%, respectively. This status level of the patients is comparatively better. As the third category contains very few subjects, so it is merged into the second category named as medium.

 Chemical Exposure This variable is captured from the occupations having the chemical exposure at the work place. The percentages of the patients of bladder cancer in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh having chemical exposure are 56%, 44%, 50%, 16% and 24%, respectively.

Out of 300 subjects in Punjab, 106 (35.33%) have chemical exposure at their work place of which counts (percentages) of cases and controls are 56(52.83%) and 50 (47.17%), respectively. Hence the subjects having chemical exposure have high

124 percentage of bladder cancer patients. Similarly, Out of 150 subjects in Islamabad, 44 (29.33%) have chemical exposure at their work place of which counts (percentages) of cases and controls are 22 (50%) and 22 (50%), respectively. Out of 150 subjects belong to Sindh, 29 (19.33%) have chemical exposure at their work place of which counts (percentages) of cases and controls are 17(58.62%) and 12 (41.38%), respectively. It is observed that higher percentage of the subjects becomes patient of bladder cancer among the subjects having chemical exposure at their work place.

 Chew Pan The count (percentages) of the pan chewers in the sample of Punjab in cases and controls are 1% and 1%, respectively. None of the pan chewer is found in all other areas including Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh. This percentage is very small and cannot be considered as contributors as risk factor in the bladder cancer. Even in Karachi (Sindh), no pan chewer was observed as patient or control.

 Tea Consumption Wickremasinghe (1978) explains that tea is the second most commonly used drink in the world, next to water. The use of tea is also much common in Pakistan like the other countries of the world. Bokuchava and Skobeleva (1980) state that black tea is major type of tea which is made from the leaves of tobacco which have been wasted before being rolled and dried. According to Wickremasinghe (1978), black tea is the main category of tea which is produced worldwide.

The percentages of the patients of bladder cancer who are using tea in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 88%, 96%, 88%, 96% and 98%, respectively. Similarly, the percentages of the controls that are using tea in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 85.5%, 95%, 88%, 94% and 96%, respectively. Hence maximum percentage of the subjects is taking tea in the Sindh and minimum in Punjab. The people of Punjab have more awareness about the bad effects of the tea while the respondents of Sindh use excessive tea because they consider it as energizer.

125

Slattery et al (1988) suggested a positive association in tea drinkers and urinary bladder cancer. Sala et al (2000) conducted a meta-analysis on tea consumption and no association was established in tea drinkers and bladder cancer. Miller et al (1983) conducted a case control study in Canada which had showed a slightly higher risk of rectal cancer for men who were taking higher amount of 'beverages' tea, coffee and all colas.

 Hair Dye The percentages of the patients of bladder cancer in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh for the users of hair dyes are 2%, 18%, 26%, 2% and 8%, respectively. While the percentages of the controls in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh for the users of hair dyes are 14.5%, 2%, 0%, 0% and 1%, respectively. The percentage of the cases that dye their hair is much higher as compared to the controls except Punjab. The reason of higher use of hair dyes in patients may be the modernism and following the fashion.

 Fluid Consumption Generally, the excess of water in the bladder reduces the concentration and stay time of chemicals by frequent urination. The percentages of the patients of bladder cancer using less than 10 glasses of water per day in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 60%, 50%, 62%, 78% and 64%, respectively. While the percentages of the controls who are using less than 10 glasses of water per day in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 33.5%, 30%, 11%, 25% and 25%, respectively. It is observed that the percentages of the patients who are using the less than two glasses of water per day are much higher than in controls in all areas.

On the other hand, the percentages of the patients of bladder cancer using 10 or more glasses of water per day in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 40%, 50%, 38%, 22% and 36%, respectively. While the

126 percentages of the controls who are using 10 or more glasses of water per day in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 66.5%, 70%, 89%, 75% and 75%, respectively.

 Source of Drinking Water In Pakistan, generally four sources of drinking water including tap, canal, government provided and mineral water are used. But in the whole sample of 900 subjects from Pakistan report the two sources of drinking water that is tap and government provided. The percentages of the patients of bladder cancer using the tap water in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 76%, 74%, 72%, 74% and 28%, respectively. The percentages of the controls using the tap water in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 77%, 56%, 71%, 72% and 25%, respectively. While the percentages of the patients of bladder cancer using the government provided water in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 24%, 26%, 28%, 26% and 72%, respectively. Similarly, the percentages of the controls using the government provided water in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 23%, 44%, 29%, 28% and 75%, respectively. Hence in Sindh, mostly controls are using the government provided water for drinking purpose and maximum numbers of patients in Punjab are using the tap water. None of the case or control report about the use of the canal or mineral water.

 Fried Items Three categories of the consumption of fried items are taken including low, medium and excessive. None of the subject (case or control) reported the consumption of fried items in third category, excessive. Only first two categories are observed in the whole sample.

127

The percentages of the patients of bladder cancer using the fried items two or less than two days per week (low) in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 55%, 58%, 56%, 76% and 74%, respectively. The percentages of the controls using the fried items two or less than two days per week (low) in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 65%, 96%, 78%, 88% and 85%, respectively. While the percentages of the patients of bladder cancer using the fried items three or four days per week (medium) in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 45%, 42%, 44%, 24% and 26%, respectively.

On the other hand, the percentages of the controls using the fried items three or four days per week (medium) days per week (low) in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 35%, 4%, 22%, 12% and 15%, respectively. The higher percentages of the patients are using the fried items three or four days per week. Inversely, higher percentages of the controls are using the fried items 2 or less than 2 days per week. Hence the use of higher amount of fried items is seemed to be problematic. Reimar et al (2004) investigated the significantly increased risk of bladder cancer only among males who were consuming the higher amount of fried food. While higher fried food consumption was not found to be statistically significant in females.

 Fats Items Like the fried items, three categories of the consumption of fat items are considered as low, medium and excessive. None of the subject (case or control) reported the consumption of fried items in third category. Only first two categories are observed in the whole sample.

The percentages of the patients of bladder cancer using the fats items two or less than two days per week (low) in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 34%, 70%, 70%, 86% and 98%, respectively. The percentages of the controls using the fats items two or less than two days per week (low) in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are

128

41%, 98%, 79%, 87% and 99%, respectively. While the percentages of the patients of bladder cancer using the fats items three or four days per week (medium) in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 66%, 30%, 30%, 14% and 2%, respectively. Consumption of fats items is higher in Punjab than in all other areas and lesser in Sindh than in others. In Punjab, its reason may be the high use of baked and fried items. While in Sindh, majority of people is below the poverty line so they cannot afford expensive meals.

On the other hand, the percentages of the controls using the fats items three or four days per week (medium) in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 59%, 2%, 21%, 13% and 1%, respectively. The higher percentages of the patients are using the fats items three or four days per week as compared to the controls. Inversely, higher percentages of the controls are using the fats items 2 or less than 2 days per week than the cases. Hence the use of higher amount of fats in meals is seemed to be problematic for the bladder cancer.

 Fast Food Like the fried and fats items, three categories of the consumption of fast food are considered as low, medium and excessive. None of the subject (case or control) reported the consumption of fast food in third category. Only first two categories are observed in the whole sample.

The percentages of the patients of bladder cancer using the fast food two or less than two days per week (low) in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 83%, 100%, 100%, 100% and 100%, respectively. The percentages of the controls using the fast food two or less than two days per week (low) in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 81.5%, 100%, 100%, 100% and 100%, respectively. While the percentages of the patients of bladder cancer using the fast food three or four days per week (medium) in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 17%, 0%, 0%, 0% and 0%, respectively. Consumption of fast food is higher in Punjab than in

129 all other areas in this sample because it is ready made, time saving and spices. All the respondents of this sample belonging to Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are using the fast food 2 or less than 2 days per week.

 Fruits Consumption of fruits is also considered in three categories as low, medium and excessive in order to observe its impact on bladder cancer. The percentages of the patients those are consuming fruits in the low category (2 or less than 2 days per week) in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 46%, 80%, 80%, 82% and 98%, respectively. The percentages of the controls using the fruits in low category in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 21.5%, 61%, 35%, 51% and 43%, respectively.

On the other hand, the percentages of the patients those are consuming fruits in the medium category (3 or 4 days per week) in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 54%, 20%, 20%, 18% and 2%, respectively. The percentages of the controls using the fruits in medium category (3 or 4 days per week) in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 78.5%, 39%, 65%, 49% and 57%, respectively. It is observed that the percentages of the fruits users 3 or 4 days per week are higher in controls than in cases. Hence, the higher use of fruits seems more protective against the disease. The purchasing power of people in Punjab is higher than other areas, so they can use fruit 3 or 4 days (medium category) per week. In Sindh, the purchasing power of people is lower than other areas, so they can use the fruit 2 or less than 2 days (low category) per week.

 Cigarette Smoking Out of 300 respondents from the Punjab, 134 (44.67%) are the smokers of which counts (percentages) of cases and controls are 68(50.75%) and 66(49.25%), respectively. The percentages of the smokers in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 44.67%, 32.67%, 25.33%, 25.33% and 30.67%, respectively. The percentages of the patients in smokers in the sample of Punjab,

130

Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 50.75%, 73.47%, 73.68%, 81.58 % and 80.43%, respectively. The percentages of controls in non smokers in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 49.25%, 26.53%, 26.32%, 18.42 % and 19.57%, respectively

The percentages of the non smokers in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 55.33%, 67.33%, 74.67%, 74.67% and 69.33%, respectively. The percentages of the patients in the non smokers in the sample of Punjab, Islamabad, Khyber Pukhtoon Khwa, Baluchistan and Sindh are 19.27%, 13.86%, 19.64%, 16.96% and 12.5%, respectively. The percentages of patients in smokers are very high as compared to the percentage of patients in non smokers.

In the sample of Khyber Pukhtoon Khwa and Baluchistan, the percentages of the smokers are the same and less than other areas. In the similar way, the percentage of patients in smokers is very high in Baluchistan as compared to other areas. In the sample of Sindh, the percentage of the patients in non smokers is very low (12.5%) as compared to Punjab, Islamabad, Khyber Pukhtoon Khwa and Baluchistan. In Punjab, the cigarette smoking is a symbol of fashion especially in the poor families.

131

Table 4.12 Area Wise Classification of Different Risk Factors with Categories

Factors Categories Punjab Islamabad Khyber P. Khwa Baluchistan Sindh No yes Total No yes Total No yes Total No yes Total No yes Total Gender Female 28 14 42 16 8 24 22 11 33 22 11 33 20 10 30 Male 172 86 258 84 42 126 78 39 117 78 39 117 80 40 120 Industrial Area Non-industrial 113 52 165 99 48 147 100 50 150 100 50 150 96 47 143 Industrial 87 48 135 1 2 3 0 0 0 0 0 0 4 3 7 Rural Area Urban 88 48 136 60 25 85 36 17 53 52 27 79 81 40 121 Rural 112 52 164 40 25 65 64 33 97 48 23 71 19 10 29 Marital Status Unmarried 0 4 4 0 0 0 0 0 0 0 0 0 0 0 0 Married 200 96 296 100 50 150 100 50 150 100 50 150 100 50 150 Education No 88 55 143 70 36 106 66 41 107 84 41 125 58 24 82 yes 112 45 157 30 14 44 34 9 43 16 9 25 42 26 68 Family history of No 179 78 257 100 50 150 100 50 150 100 50 150 100 50 150 cancer yes 21 22 43 0 0 0 0 0 0 0 0 0 0 0 0 Lifestyle Sedentary 68 80 148 90 40 130 59 46 105 98 50 148 100 50 150 Normal 132 20 152 10 10 20 41 4 45 2 0 2 0 0 0 Social Status < Rs. 10000 113 77 190 45 31 76 55 35 90 79 39 118 52 27 79 Rs.10000-20000 66 18 84 47 18 65 41 13 54 21 11 32 38 18 56 ≥ Rs.20000 21 5 26 8 1 9 4 2 6 0 0 0 10 5 15 Chemical No 150 44 194 78 28 106 69 25 94 89 42 131 83 38 121 Exposure yes 50 56 106 22 22 44 31 25 56 11 8 19 17 12 29 Chew pan No 198 99 297 100 50 150 100 50 150 100 50 150 100 50 150 yes 2 1 3 0 0 0 0 0 0 0 0 0 0 0 0

132

Factors Categories Punjab Islamabad Khyber P. Khwa Baluchistan Sindh No yes Total No yes Total No yes Total No yes Total No yes Total

Use of tea No 29 12 41 5 2 7 12 6 18 6 2 8 4 1 5 yes 171 88 259 95 48 143 88 44 132 94 48 142 96 49 145 Hair dye No 171 98 269 99 41 140 100 37 137 100 49 149 99 46 145 yes 29 2 31 1 9 10 0 13 13 0 1 1 1 4 5 Fluid taken < 10 glasses 67 60 127 30 25 55 11 31 42 25 39 64 25 32 57 (Glasses) ≥ 10 glasses 133 40 173 70 25 95 89 19 108 75 11 86 75 18 93 Source of water Tap 154 76 230 56 37 93 71 36 107 72 37 109 25 14 39 (drinking) Govt. Provided 46 24 70 44 13 57 29 14 43 28 13 41 75 36 111

Low 130 55 185 96 29 125 78 28 106 88 38 126 85 37 122 Fried item Normal 70 45 115 4 21 25 22 22 44 12 12 24 15 13 28 High 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Low 82 34 116 98 35 133 79 35 114 87 43 130 99 49 148 Fats item Normal 118 66 184 2 15 17 21 15 36 13 7 20 1 1 2 High 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Low 163 83 246 100 50 150 100 50 150 100 50 150 100 50 150 Fast food Normal 37 17 54 0 0 0 0 0 0 0 0 0 0 0 0 High 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Low 43 46 89 61 40 101 35 40 75 93 19 112 91 13 104 Fruits Normal 157 54 211 39 10 49 65 10 75 49 9 58 57 1 58 High 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Cigarette No 134 32 166 87 14 101 90 22 112 93 19 112 91 13 104 Smoking yes 66 68 134 13 36 49 10 28 38 7 31 38 9 37 46

133

Factors Categories Punjab Islamabad Khyber P. Khwa Baluchistan Sindh No yes Total No yes Total No yes Total No yes Total No yes Total Professions Cook 3 10 13 0 1 1 0 2 2 0 1 1 0 2 2 Driver 3 6 9 1 9 10 0 3 3 1 0 1 2 3 5 Farmer 16 9 25 21 9 30 29 16 45 7 4 11 12 4 16 Painter 4 1 5 0 1 1 0 1 1 0 0 0 0 0 0 Leather worker 0 0 0 0 1 1 0 0 0 1 2 3 2 0 2 Metal worker 3 17 20 0 0 0 2 1 3 2 1 3 0 1 1 Textile worker 21 13 34 0 1 1 0 2 2 0 0 0 1 2 3 Others 150 44 194 78 28 106 69 25 94 89 42 131 83 38 121 Total 200 100 300 100 50 150 100 50 150 100 50 150 100 50 150

134

4.10 Area-wise Analytical Study For area-wise inferential study, binary logistic regression models for each area is used for the purpose of comparisons in Pakistan. Therefore, the separate models are run for Punjab, Islamabad, Khyber PukhtoonKhwa, Baluchistan and Sindh. The results of logistic regression including model coefficients, Wald test, P-value, odds ratios and the 95% confidence intervals for the odds ratios are given in Table 4.14.

For testing the overall significance of the model, Omnibus test and Hosmer and Lemeshow (HL) test are used. The significance of the Omnibus test indicates that at least one of the factors is significantly affecting the response variable i.e., there is adequate fit of data to the model. This test is the alternate of the HL test. Hosmer and Lemeshow (1989) indicated that the HL test is much better, when sample size is small and some variables are continuous, than other traditional chi-square tests. If HL test is non- significant, then the model is adequately fit the data. These two tests for the overall significance of the model are given in Table 4.13.

Table 4.13 Omnibus and HL Tests for the Overall Significance of the Model Area Test Chi-square df Sig. Omnibus test 144.639 8 0.000 Punjab HL test 8.229 8 0.411 Omnibus test 78.077 3 0.000 Islamabad HL test 0.628 3 0.890 Omnibus test 116.922 6 0.000 Khyber P. Khwa HL test 3.156 6 0.924 Omnibus test 82.805 3 .000 Baluchistan HL test 6.009 3 .198 Omnibus test 75.085 2 0.000 Sindh HL test 1.259 2 0.533

From Table 4.13, it is observed that all five models are significant by the omnibus test at P-values =0.000 indicating that at least one of the factors is significantly affecting the response variable i.e., there is adequate fit of data to the model. Hosmer and

135

Lemeshow test (HL test) is insignificant with different p-values for different areas indicate that the model is adequately fitted.

From Table 4.14, it is found that eight factors including age, social status, lifestyle, family history, cigarette smoking, tea, fluid consumptions and fruits in Punjab, three factors including cigarette smoking, source of drinking water and fried items in Islamabad, six factors including chemical exposure, lifestyle, cigarette smoking, fluid consumption, fried items and fruit in Khyber PukhtoonKhwa, three factors including cigarette smoking, fluid consumption and fruits in Baluchistan and two factors including cigarette smoking and fluid consumption in Sindh are found to be significant. In the analysis, some variables are affecting negatively to the risk of bladder cancer and some positively. The factors having negatively significant effect on the bladder cancer are the protective factors while the positively significant factors are the risk factors of the disease. The protective factors are inversely associated to the bladder cancer while the risk factors are directly associated to the disease.

 Fluid Consumption The fluid consumption is also found to be significant in all four provinces of Pakistan but insignificant in the federal city Islamabad. In federal area, 50% percent of the patients are consuming the 10 or less than 10 glasses of water and 50% more the 10 glasses of water. In the overall model of the Pakistan, the fluid consumption is also investigated significant factors that provides the protection against the disease. The minimum and maximum odds ratios for the fluid consumption are 0.025 and .327 in the sample of Khyber PukhtoonKhwa and Punjab, respectively. Fluid consumption provides maximum protection that is 97.5% in the sample of Khyber PukhtoonKhwa and minimum 67.3% in Punjab. In Sindh, consumption of more that 10 glasses of water provides 74% protection against the diseases. Its reason may be the different purity level of drinking water in different areas.

136

 Fruits The consumption of fruits is observed to be significantly protective factor in Punjab, Khyber PukhtoonKhwa and Baluchistan with odds ratios 0.483, 0.173 and 0.206, respectively. In other words, fruits consumption in Punjab, Khyber PukhtoonKhwa and Baluchistan in 3 or 4 days per week provides a protection against the diseases of 52%, 83% and 79%, respectively as compared to the subjects using fruit in 2 or less than 2 days per week. The consumption of fruits in 3 or 4 days per week is observed to be significantly protective factor in the overall model for the Pakistan with odds ratio 0.292. Reimar et al (2004) suggested that higher fruit consumption reduced the risk of bladder cancer in females but had little effect on male bladder cancer risk.

 Fried Items The fried items are found to be positively significant in the sample of Islamabad and Khyber PukhtoonKhwa with the odds ratios and 95% confidence intervals (12.206,3.291-45.275) and (5.934, 1.429-24.648), respectively. In other three provinces including Punjab, Baluchistan and Sindh, fried items are found to be insignificant. In federal area, the subjects who are using the fried items three or four days per week have 12.2 times higher risk of getting the bladder cancer as compared to those who use 2 or less than 2 days per week. In Khyber PukhtoonKhwa, the subjects using the fried items three or four days per week have 6 times higher risk of getting the bladder cancer as compared to those who use 2 or less than 2 days per week. In Khyber PukhtoonKhwa, the risk of getting the bladder cancer is half than the federal area (Islamabad). This variable is also significant in the overall sample of the Pakistan.

 Source of Drinking Water In Pakistan, generally four sources of drinking water including tap, canal, government provided and mineral water are used. These four categories are considered in this sample but none of the respondent reported about the use of canal water and mineral water. Hence, only two categories including tap water and government provided water for drinking are reported by the respondents. Government provided water is generally considered better due to the filtration as compared to the tap water which may be

137 contaminated due to the chemicals and other reasons. In the sample of federal area, use of government provided water is found to be negatively significant with the odds ratio and 95% confidence interval 0.192 and (0.061-0.603), respectively as compared to the tap water. It means that the subjects using government provided water has 81% protection against the urinary bladder cancer as compared to the subjects using the tap water because the tap water may contain high carcinogenic chemicals. While the government provided water is filtered and provided to people after the purification. This variable is found significant only in federal area.

Similar findings were also given by Villanueva et al (2006) who conducted a study by pooling the data of 6 case-control studies and found the higher risk of bladder cancer in the users of tap water and suggested that the tap water may contain high carcinogenic chemicals.

 Lifestyle Lifestyle is captured from the exercise and taken in three categories sedentary, normal and active. None of the case or control reported about the third category in the overall sample of 900 subjects. Hence, two categories that is sedentary (no exercise) and normal (about 30 minute exercise daily) are observed and used for analysis. The lifestyle is found to be significant in the overall model with odds ratio 0.102 indicating the 90% protection against the bladder cancer. In the area-wise analysis, the lifestyle is found to be negatively significant in the sample of Punjab and Khyber PukhtoonKhwa with odds ratios and 95% confidence interval (0.077, 0.036-0.165) and (0.171, 0.031- 0.943), respectively. In other three areas including federal area, Baluchistan and Sindh, lifestyle is found to be insignificant. In Punjab, about 92% protection is observed in the subjects having normal lifestyle as compared to the subjects with sedentary lifestyle. Similarly in Khyber PukhtoonKhwa, about 83% protection is observed in the subject having normal lifestyle as compared to the subjects with sedentary lifestyle. The people of Punjab are very conscious about their health so they prefer exercise and lead a healthy life as compared to others.

138

 Chemical Exposure The chemical exposure is captured from the occupations which has carcinogen at their work place. It is observed to be significant in the overall model with the odds ratio and 95% confidence interval 2.594 and (1.460- 4.607), respectively. in the area-wise models, the chemical exposure is found to be significant only in the model of Khyber PukhtoonKhwa with odds ratio and 95% confidence interval 4.637 and (1.022- 21.053), respectively. It indicates that the workers have chemical exposure at their work place has 4.6 times higher risk of developing the bladder cancer as compared to those does not have chemical exposure at their work place. Its reason is the presence of carcinogen at their work place.

 Tea Consumption The use of tea is very common in Pakistan and tea contains much amount of caffeine like the coffee. Wickremasinghe (1978) conducted a study and investigated that tea is the second most commonly used drink in the world, next to water. Bokuchava and Skobeleva (1980) state that black tea is major type of tea which is made from the leaves of tobacco which have been wasted before being rolled and dried. Wickremasinghe (1978) explains that black tea is the main category of tea which is produced worldwide.

In the overall sample of 900 subjects the tea is found to be insignificant but in the sample of Punjab, it is observed to be significant with the odds ratio and 95% confidence interval 3.059 and (1.156-8.095), respectively. It means that the subjects who take tea have about 3 times higher risk of developing the bladder cancer as compared to those who do not take tea.

The studies of Slattery et al (1988) also explain the positive association in tea drinkers and urinary bladder cancer. But Zeegers et al (2004) explained that the consumption of tea did not establish the association with the risk of bladder cancer. Hence, the study Zeegers et al (2004) is contradicting this study and of Slattery et al (1988). Further research is required for observing the association between the use of tea and the risk of bladder cancer.

139

McLaughlin et al (1983) stated that the tea drinking was found to be positively associated with the risk of renal pelvis cancer for females. Similarly, McLaughlin et al (1984) explained that tea drinking was found to be positively associated with the risk of kidney cancer for females and not for males. Kinlen and McPherson (1984) showed in a recent report that the higher risk of pancreatic cancer was observed among heavier tea consumers in both males and females. Miller et al (1983) conducted a case control study in Canada which had showed a slightly higher risk of rectal cancer for men who were taking higher amount of 'beverages' tea, coffee and all colas.

Hartge et al (1983) conducted a very large case-control study and investigated slightly higher risk of bladder cancer among heavier female tea consumers but not in males. Other authors Morgan and Jain (1974), Miller et al (1978), Howe et al (1980) and Sullivan (1982) did not observe any association between the tea consumption and the risk of bladder cancer. Studies conducted by Tajima and Tominaga (1985) and Stocks (1970) had investigated a positive association between the tea consumers and the Colon cancer.

In view of these findings, it is stated that heavy black tea consumers have a higher risk of bladder cancer, renal pelvis cancer, kidney cancer or the colon cancer. Hence, the results of this study are supported by the studies of the Hartge et al (1983) but contradict with the studies of Morgan and Jain (1974), Miller et al (1978), Howe et al (1980) and Sullivan (1982).

 Family History The family history of cancer is found to be significant risk factor in the model of Punjab with odds ratio 3.3130 and 95% confidence interval of odds ratio (1.405-7.812), which means that the subjects with family history of cancer have 3.31 times higher risk of getting the diseases as compared to those without family history of cancer. Its reason may be the lack of information about the disease history of the family members in other areas except Punjab. In all other areas including Islamabad, Khyber PukhtoonKhwa, Baluchistan and Sindh, the family history of cancer is observed as insignificant risk factor. In the whole sample of 900 subjects from the Pakistan, the family history of

140 cancer is also observed as the significant risk factor with odds ratio and 95% confidence interval of odds ratio 3.130 and (1.325, 7.394), respectively which means that the subjects having cancer in their family members (first degree relatives) have 3.13 times more risk of bladder cancer as compared to those who have not cancer in their family history. Kunze et al (1993) also investigated that the family history of bladder cancer was positively significant in males but found to be insignificant in case of females. They considered the family history in their first degree relatives.

 Social Status The social status is observed from the income of the respondents in which three categories of income including low, medium and high are taken. But very few subjects reported about the third category. Therefore, these few subjects of third category are merged in the second category in the analysis. The social status is found to be significant protective factor in the model of Punjab with odds ratio 0.386 and 95% confidence interval of odds ratio (0.183 -0.815), which means that the subjects having medium level of social status have about 61% protection against the diseases as compared to those having low social status. The people with low status do not able to afford proper diet, fruits and other medical facilities. In all other areas including Islamabad, Khyber PukhtoonKhwa, Baluchistan and Sindh, the social status is observed to be insignificant.

 Age The age is also found to be the significant risk factor of the urinary bladder cancer in the sample of Punjab with odds ratio 1.04 and 95% confidence interval (1.011-1.078), which means that the older subjects has 1.04 times higher risk of getting the disease after each year. According to the Devita, Hellman and Rosenberg (2001), the latency period from initial exposure to the development of an urothelial tumor is a median of 18 years. Urinary Bladder cancer is a disease of older and rarely diagnosed before the age of 40. The overall study found that the minimum and maximum ages of the patients were 38 and 89 years, respectively. The mean, median and mode age was 57, 56 and 60 years, respectively. The disease was found to be more frequent in the age of 60 years. The

141 number of patients above the age of 65 year was 62 (19.7%). Ferlay et al (2004) observed that the two third of all bladder caner cases had their ages above 65 years.

Table 4.14 Area-wise Models Coefficient, Odds Ratios and 95% CI’s for Odds Ratios 95% CI .for Area Factors β S.E. Wald Sig. Exp(β) EXP(β) Lower Upper Punjab Age 0.043 0.016 6.963 0.008 1.044 1.011 1.078 SS -0.952 0.382 6.230 0.013 0.386 0.183 0.815 LS -2.566 0.389 43.426 0.000 0.077 0.036 0.165 FHC 1.198 0.438 7.493 0.006 3.313 1.405 7.812 CS 1.905 0.356 28.707 0.000 6.720 3.347 13.491 Tea 1.118 0.497 5.069 0.024 3.059 1.156 8.095 FC -1.117 0.347 10.378 0.001 0.327 0.166 0.646 FRUIT -.727 0.350 4.317 0.038 0.483 0.243 0.960 Constant -2.598 1.089 5.687 0.017 0.074 Islamabad CS 2.842 0.516 30.378 0.000 17.158 6.244 47.147 SW -1.650 0.584 7.983 0.005 0.192 0.061 0.603 FRIED 2.502 0.669 13.995 0.000 12.206 3.291 45.275 Constant -1.750 0.340 26.475 0.000 0.174 Khyber CE 1.534 0.772 3.950 0.047 4.637 1.022 21.053 Pukhtoon LS -1.769 0.873 4.108 0.043 0.171 0.031 0.943 Khwa CS 2.972 0.728 16.666 0.000 19.526 4.688 81.329 FC -3.691 0.781 22.313 0.000 0.025 0.005 0.115 FRIED 1.781 0.727 6.008 0.014 5.934 1.429 24.648 FRUIT -1.753 0.683 6.594 0.010 0.173 0.045 0.660 Constant 1.073 0.618 3.015 0.082 2.925 Baluchistan CS 3.261 0.626 27.150 0.000 26.064 7.645 88.856 FC -1.823 0.512 12.668 0.000 0.161 0.059 0.441 FRUIT -1.579 0.642 6.052 0.014 0.206 0.059 0.725 Constant -.275 0.335 0.675 0.411 0.759 Sindh CS 3.193 0.493 41.973 0.000 24.352 9.270 63.974 FC -1.346 0.481 7.843 0.005 0.260 .101 .668 Constant -1.121 0.388 8.327 0.004 0.326

142

It is observed from Table 4.14, that the cigarette smoking is a risk factor which is found to be significant in all four provinces and federal area with different values of odds ratio. The minimum and maximum odds ratios for the cigarette smoking are 6.72 and 26.064 in the sample of Punjab and Baluchistan, respectively. The effect of cigarette smoking is much higher in Baluchistan than in other provinces and federal area. The higher risk is observed in Sindh after the Baluchistan with odds ratio 24.35 and 95% confidence interval (9.27, 63.97). The cigarette smoking is a universal risk factor of the urinary bladder cancer and found significant in all over Pakistan.

The predicted logit models in different areas including Punjab, Islamabad, Khyber PukhtoonKhwa, Baluchistan and Sindh are presented in Table 4.15 for the comparison purpose.

Table 4.15 Logit Models for Different Areas of Pakistan

AREAS Predicted Models Punjab Z = -2.598 + 0.043* Age-0.952*SS -2.566*(Life style) +1.198* FHC +1.905*CS + 1.118*Tea -1.117 *FC– 0.727*Fruit Islamabad Z= -1.750 + 2.842*CS – 1.650*SW + 2.502*Fried Khyber P. Khwa Z= 1.073+1.534*CE-1.769*Lifestyle+2.972*CS -3.691*FC + 1.781*Fried -1.753*Fruit Baluchistan Z= -0.275+3.261*CS -1.823*FC -1.579*Fruit Sindh Z= -1.121+3.193*CS -1.346*FC

4.11 Goodness of Fit of the Models In order to asses the goodness of fit, correct percentage of classification, the Cox & Snell R Square and Nagelkerke R Square has been used for each model. These measures provide the sufficient evidence about the goodness of fit of the model.

From Table 4.16, it is found that the values of the Cox & Snell and Nagelkerke R Square are higher in the model of Khyber PukhtoonKhwa that is 0.541 and 0.752, respectively. On the other hand, the values of the Cox & Snell and Nagelkerke for the

143 model of Punjab are less than other four models. But these values are not too small so that one can interpret that models are good fitted. Generally, only one independent variable is the continuous that is age and all other variables are dummy variables so the value of the R- square may be small. These values of R-squares are not interpreted like the R-square in the OLS (Ordinary Least Square) but only look like R-squared in the sense that they are on a similar scale, ranging from 0 to 1. Some methods of R-square do not attain the value 0 or 1. Therefore, they do not have the meaning truly like R-Square in OLS.

Table 4.16 Cox & Snell R Square and Nagelkerke R Square for all Five Models R Square Punjab Islamabad Khyber P. Khwa Baluchistan Sindh

Cox & Snell 0.383 0.406 0.541 0.424 0.394 Nagelkerke 0.531 0.564 0.752 0.589 0.547

From Table 4.17, it is observed that in the model of Punjab, out of 200 controls 184 (92.0%) are correctly predicted as controls while out of 100 patients of bladder cancer, 68(68.0%) are correctly predicted as cases (patients). But 32(32.0%) patients and 16 (8.0%) controls are misclassified 32(32.0%) as controls and 16 (8.0%) as patients, respectively. The overall numbers (percentages) of correctly classified and misclassified of subjects in the model of Punjab are 252 (84.0%) and 48(16.0%), respectively.

From the model of Islamabad, it is found that out of 100 controls 91 (91.0%) are correctly predicted as controls while out of 50 patients of bladder cancer, 36(72.0%) are correctly predicted as cases (patients). But 14(28.0%) patients and 9 (9.0%) controls are misclassified 14(28.0%) as controls and 9 (9.0%) as patients, respectively. The overall numbers (percentages) of correctly classified and misclassified of subjects are 127 (84.7%) and 23(16.0%), respectively.

In the similar way, considering the model of Khyber PukhtoonKhwa, it is observed that out of 100 controls 93 (93.0%) are correctly predicted as controls while out of 50 patients of bladder cancer, 42(84.0%) are correctly predicted as cases (patients).

144

But 8(16.0%) patients and 7 (7.0%) controls are misclassified 8(16.0%) as controls and 9 (9.0%) as patients, respectively. The overall numbers (percentages) of correctly classified and misclassified of subjects are 135 (90.0%) and 15(10.0%), respectively. The overall correctly predicted percentage is higher in the model of Khyber PukhtoonKhwa as compared to the all other four models.

Considering the model of Baluchistan, it is assessed that out of 100 controls 97 (97.0%) are correctly predicted as controls while out of 50 patients of bladder cancer, 29(58.0%) are correctly predicted as cases (patients). But 21(42.0%) patients and 3 (3.0%) controls are misclassified 21(42.0%) as controls and 3(3.0%) as patients, respectively. The overall numbers (percentages) of correctly classified and misclassified of subjects in the model of Baluchistan are 126 (84.0%) and 24(16.0%), respectively. Overall correctly predicted percentage is lesser in the models of Baluchistan and Punjab as compared to other three models. The percentage of correctly predicted patients is much lesser in the model of Baluchistan as compared to the others.

In the similar way, considering the model of Sindh, it is observed that out of 100 controls 91 (91.0%) are correctly predicted as controls while out of 50 patients of bladder cancer, 37(74.0%) are correctly predicted as cases (patients). But 13(26.0%) patients and 9 (9.0%) controls are misclassified 13(26.0%) as controls and 9 (9.0%) as patients, respectively. The overall numbers (percentages) of correctly classified and misclassified of subjects are 128 (85.3%) and 22(14.7%), respectively. Overall correctly predicted percentage is lesser in the model of Sindh as compared to Khyber PukhtoonKhwa but higher as compared to Baluchistan, Punjab and Islamabad.

From the correct classification of the fitted models, it is investigated that all these models have too high percentages of correct classification of the subjects and appropriate values of the R-squares which provide evidence that the fitted models are adequate and hence 95% confidence intervals for odds ratios are valid for inferences.

145

Table 4.17 Correct Classification for Federal Area and Four Provinces

Area Bladder Predicted Percentage of Overall Cancer Bladder Cancer correct percentage No Yes classification Punjab No 184 16 92.0 84.0 Yes 32 68 68.0 Islamabad No 91 9 91.0 84.7 Yes 14 36 72.0 Khyber No 93 7 93.0 90.0 PukhtoonKhwa Yes 8 42 84.0 Baluchistan No 97 3 97.0 84.0 Yes 21 29 58.0 Sindh No 91 9 91.0 85.3 Yes 13 37 74.0

146

4.12 Summary and Conclusions

In this section, the overall summary of the results and conclusions drawn are discussed and explained in section 4.12.1 and 4.12.2, respectively.

4.12.1 Summary This study was run in order to investigate the effect of different risk factors of urinary bladder cancer in Pakistan. A sample of 900 subjects including 300 cases and 600 controls was selected from different areas of Pakistan including headquarter of all four provinces and federal area (Islamabad) through a questionnaire. The requisite information was obtained from all the cases and controls using the direct interview method. On the basis of the analysis of the information obtained from the cases and controls, the following important results are obtained.

About 22 factors with sub categories were included in the study. For bivariate analysis, the chi-square, phi/v and Kandall’s tau-b statistics are used. For the purpose of multivariate analysis, the binary logistic regression was run by using the SPSS (Version- 16) to observe the significant risk factors.

The counts (percentages) of male and female patients in the sample are 246 (80%) and 54 (20%), respectively. The male /female (cases) ratio in the sample is 4.5:1. In the studies of Puente et al., (2003), male/female ratio was 6.4:1. Out of total 300 cases, the numbers (percentages) belonging to the urban and rural areas were 157 (52.33%) and 143 (47.67%), respectively. The disease is more frequent in urban areas than in rural areas. Devita, Hellman and Rosenberg (2001) has also shown the higher diagnosis rate of urinary bladder cancer in urban areas than in rural areas.

For analytical study, bivariate analysis was conducted by using the chi-square, phi/v and Kandall’s tau-b statistics and for the multivariate analysis, the binary logistic regression model was run. The software SPSS (Version-16) was used for the analysis of data in order to observe the significant risk factors. In the descriptive analysis, it was observed that the risk of bladder cancer increases with an increase in the number of

147 cigarettes smoked per day, years of smoking and risk decreases when the stop smoking period increases. Further more, similar results were observed in the bivariate analysis for the cigarette smokers.

In the bivariate analysis, the factors including social status, consumption of fruits, lifestyle and fluid consumption are found to be negatively associated with the bladder cancer which means that these factors are the protective factors for the disease. On the other hand, the nine factors including family history of cancer, cigarette smoking, tea cups to be taken per day, hair dye, use of fried items, use of fats item, and number of cigarettes smoked per day, the smoking period and professions are positively associated with the disease which means that these factors and bladder cancer are directly related. The stop smoking period is also found to be inversely related to the bladder cancer. It means that the quitting smoking has reduces the risk of bladder cancer.

By using the logistic regression model, the six factors including hair dye, chemical exposure, family history, cigarette smoking, fried items and fats items are found to be positively significant with the odds ratios and 95 % confidence intervals of odds ratios (2.96; 1.396-6.279), (2.59; 1.460-4.607), (3.13; 1.325-7.394), (10.6; 7.007-15.941), (2.11; 1.364-3.269) and (2.08; 1.309-3.305), respectively. While the three factors including lifestyle, fluid consumption and use of fruits are found to be inversely significant with odds ratios and 95% confidence intervals for the odds ratios (0.102; 0.056-0.187), (0.268; 0.183-0.392) and (0.292; 0.193-0.440), respectively.

Therefore, these three factors are considered the protective factors against urinary bladder cancer. Further more, the occupational risk factors were also investigated using logistic regression model. This analysis suggested that the four categories of the occupations including cooks, drivers, metal workers and textile workers were significant having odds ratios and 95% confidence intervals of the odds ratios (14.132; 4.068 - 49.088), (7.949; 3.321 - 19.025), (7.571; 3.147- 18.214) and (2.168; 1.136 - 4.138), respectively. While the farmers and painters are observed to be insignificant in Pakistan.

148

According to this study, the cooks have higher risk of bladder cancer as compared to all other occupations.

In area-wise study, eight factors age, social status, lifestyle, family history, cigarette smoking, tea, fluid consumptions and fruits in Punjab, three factors cigarette smoking, source of drinking water and fried items in Islamabad, six factors chemical exposure, lifestyle, cigarette smoking, fluid consumption, fried items and fruit in Khyber Pukhtoon Khwa, three factors including cigarette smoking, fluid consumption and fruits in Baluchistan and two factors including cigarette smoking and fluid consumption in Sindh are found to be significant. In these variables stated above, some are affecting the risk of bladder cancer negatively and some positively. Cigarette smoking is the major risk factor and found to be significant in each area of Pakistan. Fluid consumption is also major protective factor and found to be significant in all areas except Islamabad.

4.12.2 Conclusions The first objective of the study was to use the both descriptive and analytical approaches in the analysis to explain the risk factors. This objective of the study was accomplished by explaining the occurrence and non-occurrence of the diseases both in cases and controls. The descriptive part is very important part of the study that explains the risk factors in a scientific manner.

The second objective of this study was to assess the association of each individual risk factor with the diseases. For the achievement of this objective, bivariate analysis was done. The individual association was observed each of the factors with that of the urinary bladder cancer. Actually, this analysis explains the significant direction of the association that is the factor is directly or inversely related to the disease.

In order to accomplish this 4th objective, bivariate analysis was conducted by using the chi-square, phi/v and Kandall’s tau-b statistics. According to this analysis, the factors including social status, consumption of fruits, lifestyle and fluid consumption are found to be negatively associated with the bladder cancer. On the other hand, the nine

149 factors including family history of cancer, cigarette smoking, tea cups to be taken per day, hair dye, use of fried items, use of fats item, and number of cigarettes smoked per day, the smoking period and professions are positively associated with the disease.

A statistical model was developed in order to explain the strength (severity) of the risk factors of bladder cancer in terms of probabilities. This model was used to find the odds ratios and the confidence interval of odds ratios for the factors included in this study. The development of the model was the achievement of third objective of the study. This model suggested that the six factors including hair dye, chemical exposure, family history, cigarette smoking, fried items and fats items were found to be positively significant, while the three factors including lifestyle, fluid consumption and use of fruits were observed to be inversely significant.

The fourth objective of the study was fulfilled by investigating the occupational risk factors of the urinary bladder cancer and the logistic regression model was run. Form this analysis, it was investigated that the four categories of the occupations including cooks, drivers, metal workers and textile workers were significant in Pakistan. As the cooks had higher risk of bladder cancer as compared to the other categories, so the government should take some safety measures at the cooking place. The area must be free from severe heat and smoke.

Comparison of risk factors of the disease was made among different areas in Pakistan by using the binary logistic model for each of the area including Punjab, Islamabad, Khyber PukhtoonKhwa, Baluchistan and Sindh. According to this analysis, eight factors including age, social status, lifestyle, family history, cigarette smoking, tea, fluid consumptions and fruits in Punjab, three factors including cigarette smoking, source of drinking water and fried items in Islamabad, six factors including chemical exposure, lifestyle, cigarette smoking, fluid consumption, fried items and fruit in Khyber PukhtoonKhwa, three factors including cigarette smoking, fluid consumption and fruits in Baluchistan and two factors including cigarette smoking and fluid consumption in Sindh were found to be significant.

150

4.13 Recommendations to Society After studying the problem of bladder cancer with respect its risk factors, the following important suggestions can be given to society in order to minimize the risk of urinary bladder cancer:

1. Create awareness among the people about the aggressiveness of the disease and its risk factors. 2. Special seminars should be conducted to educate the people about the risk factors of the disease. 3. The use of fried and fats items should be reduced in the daily life. 4. More than 30 minutes exercise is necessary for healthy life. 5. As cigarette smoking is the major risk factor in Pakistan, therefore, avoid cigarette smoking. 6. Media (such as newspapers, T.V, Radio, etc.;) can play important role against the smoking and other risk factors of the urinary bladder cancer. 7. People should use the sufficient amount of fruits in their daily diet. 8. The more than ten glasses of total fluid should be consumed daily in order to increase the amount of urination. 9. The chemicals used for hair dye are also problematic and can cause bladder cancer. 10. Some occupations have carcinogen at work place like textile workers, mechanics, minors, painters, leather worker, etc.; Government should provide safety from the carcinogen by consulting the specialists. 11. Tobacco expansion advertisements from the all corners should be banned. 12. Special laws must be framed and strictly implemented to avoid cigarette smoking. 13. The cancer treatment expenditures should be transferred to the cigarette related companies. 14. All the private and public hospitals must be bounded to provide the informations about the cancer incidents to the government.

151

15. The research in the cancer diseases must be funded by the government and other NGO,s.

4.14 Recommendations for further research On the completion of this study, it becomes necessary to give the effective recommendations in order to conduct the further studies that will be beneficial for the society. The important recommendations for further studies are given below:

1. A similar study may be conducted in private hospitals of Pakistan because elite class of people attends the private hospital for the treatment of bladder cancer. 2. A separate study can be conducted in the hospitals of Pakistan Atomic Energy Commission. 3. A combined study may also be conducted in public and private hospitals. 4. Research is required to investigate the effect of passive smoking on bladder cancer. 5. Detailed studies should be conducted to observe the effect of occupations in all areas of Pakistan. 6. Research must be needed to investigate the effect of contaminated water on bladder cancer. 7. Effect of different types of the tobacco may be studied for their injuriousness. 8. As Pakistan is an agricultural country, therefore, the association between pesticide and bladder cancer should be investigated in detail. 9. Effect of different food ingredients on the bladder cancer may be observed. 10. The separate studies are required for gender and area-wise. 11. Cancer patients directories must be maintained in all provinces and for overall the country also.

The policies must be maintained for the treatment of urinary bladder cancer by keeping in view its risk factors. All these efforts will be beneficial for the reduction of the urinary bladder cancer in Pakistan. With out the research, it is not possible to introduce the effective policies in the country.

152

4.15 Limitations of collecting data

During the data collection, the following problems in different cities of Pakistan were faced:

 In some hospitals, heads of the urology wards (doctors) refused to allow for conducting the interview of the patients in their wards.  Keeping in view the problem of terrorism in Peshawar, mostly the heads of the hospitals were not ready to allow data collection.  In Peshawar and Quetta, the language problem was faced where most of the people did not know their national language. Due to this reason, a translator was bonded for the purpose of interview of the patients. Persian and Pushto were spoken in Peshawar but in Quetta, Brahvi, Sindhi, Pushto, Uzbak and Afghani languages were spoken.  In Quetta and Peshawar, most of the females do not know their residential address and address was very necessary for checking the adequacy of data.  In hospitals, the condition of sanitation and cleanliness was very poor and unsatisfactory.

153

REFERENCES

Adami, H.O., Hunter, D. and Trichopoulos, D. (2002). Textbook of Cancer Epidemiology. New York: Oxford University Press.

Ahmad, Z. and Pervaiz, M.K (2009). Incidence of Diabetes and Estimating the Effects of Different Risk Factors on Diabetes in Pakistan. An unpublished PhD thesis at GC University, Lahore (Pakistan).

Alam, S.E. (1998). Prevalence and pattern of smoking in Pakistan. J Pak Med Assoc. Vol. 48, pp. 64-66.

Alberg, A.J., Kouzis, A., Genkinger, J.M., Gallicchio, L., Burke, A.E., Hoffman, S.C., Diener, W.M., Helzlsouer, K.J. and Comstock, G.W. (2007). A prospective cohort study of bladder cancer risk in relation to active cigarette smoking and household exposure to second hand cigarette smoke. Am J Epidemiol. Vol. 165, No. 6, pp. 660-666.

American Cancer Society (2007). Cancer Facts and Figures 2007. Downloaded from cite: www.cancer.org/downloads/STT/CAFF2007PWsecured.pdf

American Cancer Society (2007). Global Cancer Facts & Figures 2007. Downloaded from www.cancer.org/downloads/STT/Global_Cancer_Facts_and_Figures_2007_rev.pdf

Baris, D., Karagas, M. R., Verrill, C., Johnson, A., Andrew, A.S., Marsit, C.J., Schwenn, M., Colt, J.S.,Cherala, S., Samanic, C., Waddell, R., Cantor, K.P., Schned, A., Rothman, A., Lubin, J., Fraumeni, J. F., Hoover, R.N., Kelsey, K. t. and Silverman, D.T. (2009). A Case–Control Study of Smoking and Bladder Cancer Risk: Emergent Patterns Over Time: JNCI J Natl Cancer Inst. Vol. 101, No. 22, pp. 1553-1561.

Bhurgri, Y. (2004). Karachi Cancer Registry Data: Implications for the National Cancer Control Program in Pakistan. Asian Pac J Cancer Prev. Vol. 5, pp.77-82.

154

Bhurgri, Y., Bhurgri, A. and Hassan, S. (2000). Cancer incidence in Karachi, Pakistan: First results from Karachi Cancer Registry. Int J Cancer, Vol. 85, pp.325-9.

Bianchi, G.D., Cerhan, J.R., Parker, A.S., Putnam, S.D., See, W.A., Lynch, C.F. and Cantor, K.P. (2000). Tea consumption and risk of bladder and kidney cancers in a population-based case-control study. Am J Epidemiol. Vol. 151, No.1, pp. 377- 383.

Bitterman, W.A., Farhadian, H., Abu, S.C., Lerner, D., Amoun, H., Krapf, D. and Makov, U.E. (1991). Environmental and nutritional factors significantly associated with cancer of the urinary tract among different ethnic groups. Urologic Clin. N. Am. Vol. 18, No.1, pp. 501–508.

Bokuchava, M.A. and Skobeleva, N.I. (1980). The biochemistry and technology of tea manufacture. CRC Crit. Rev. Food Sci. Nutr. Vol. 12, pp.303.

Borden, L.S., Clark, P.E. & Hall, M.C. (2003). Bladder cancer. Opin Oncol. Vol.15, pp.227-233.

Braver, D.J., Modan, M., Chetrit, A., Lusky, A. & Braf, Z. (1987). Drinking, micturition habits, and urine concentration as potential risk factors in urinary bladder cancer. J. Natl. Cancer Inst. Vol. 78, No.1, pp. 437–440.

Bruemmer, B., White, E., Vaughan, T.L. and Cheney, C.L. (1996). Nutrient intake in relation to bladder cancer among middle-aged men and women. Am J Epidemiol. Vol. 144, No.1, pp. 485-495.

Brennan, P., Bogillot, O., Greiser, E., Claude, J. C., Wahrendorf, J., Cordier, S., Jockel, K.H., Gonzalo, L.A., Tzonou, A., Vineis, P., Donato, F., Hours, M., Serra, C., Audorff, U. B., Schil, W., Kogevinas M. & Boffettal. P. (2001). The

155

Contribution of Cigarette Smoking to Bladder Cancer in Women (Pooled European Data). Cancer Causes & Control. Vol. 12, No. 5, pp. 411-417.

Burch, R.D., Rohan, T.E., Howe, C.R., Risch, H.A., Hill, G.B.; Steele, R. and Miller, A.B. (1989). Risk of bladder cancer by source and type of tobacco exposure. Int J Caner. Vol. 44, pp. 22-628.

Burns, P. B. and Swanson, G. M. (1991). Risk of Urinary Bladder Cancer among Blacks and Whites: The Role of Cigarette Use and Occupation. Cancer Causes & Control, Vol. 2, No. 6, pp. 371-379.

Cantor, K.P., Hoover, R., Hartge, P., Mason, T.J., Silverman, D.T., Altman, R., Austin, D.F., Child, M.A., Key, C.R. & Marrett, L.D. (1987). Bladder cancer, drinking water source, and tap water consumption: a case-control study. J Natl Cancer Inst. Vol. 79, No. 1, pp.1269–79.

Castelao, J.E.; Yuan J.M.; Skipper, P.L.; Tannenbaum, S.R.; Gago, D. M.; Crowder, J.S.; Ross, R.K. and Yu, M.C. (2001). Gender and smoking related bladder cancer risk. J Nat Can Inst. Vol. 93, pp. 538-545.

Chatterjee, S. and Hadi, A.S. (2006). Regression Analysis by Examples ( 4th edition). John Willey and Sons, Inc. Hoboken, New Jersey: pp. 233-239

Ciccone, G. and Vineis, P. (1988). Coffee drinking and bladder cancer. Cancer Lett., No. 41, pp. 45-52.

Claus, E.B.; Schildkraut, J.M.; Thompson, W.E. and Risch, N.J. (1996). The genetic attributable risk of breast and ovarian cancer. Cancer. Vol. 77, pp. 2318.

Cohen, S. and Johansson, S. (1992). Epidemiology and etiology of bladder cancer. Urol Clin North Am; Vol.19, No. 3, pp.421-428.

156

Cohen, S., Shirai, T. & Steineck, G. (2000). Epidemiology and etiology of premalignant and malignant urothelial changes. Scan J Urol Nephrol Supp.; Vol. 205, pp.105- 115.

Devita, V.T., Hellman, S. and Rosenberg, S.A. (2001). Cancer: Principles and Practice of Oncology 6th edition by Lippincott Williams & Wilkins Publishers, pp. 1005.

Dollinger, M., and Rosenbaum, E.H. (1991). Understanding Cancer and Everyone’s Guide to Cancer Therapy. Universal Press Syndicate Company, New York, pp.1- 9.

Ferlay, J., Bray, F., Pisani, P. & Parkin, D.M. (2004). GLOBOCAN 2002: Cancer Incidence, Mortality and Prevalence Worldwide, Version 2.0. IARC Cancer Base No. 5. Lyon, IARC Press. Available at: http://www-dep.iarc.fr/

Gago, D.M., Castelao, J.E., Yuan, J.M., Yu, M.C. and Ross, R.K. (2001). Use of permanent hair dyes and bladder cancer risk. Int J Cancer. Vol. 91, No. 1, pp. 575-579.

Ghias, M. and Pervaiz, M.K. (2009). Identification of Epidemiological Risk Factors for Hepatitis C in Punjab, Pakistan. J Ayub Med Coll Abbottabad. Vol. 21, No. 2, pp. 156-161.

Guo, H.R., Chiang, H.S., Howard, H., Stuart, R.L. and Richard R. M. (1997). Arsenic in Drinking Water and Incidence of Urinary Cancers. Epidemiology, Vol. 8, No. 5, pp. 545-550.

157

Hanif and Ahmad (2006) Biostatistics For Health Students. An Isoss Publication, Islamic Society of Statistical Sciences Lahore-Pakistan.

Hartge, P., Hoover, R., West, D.W. and Lyon, J.L. (1983). Coffee drinking and risk of bladder cancer. J. Natl Cancer Inst.Vol. 70, pp.1021.

Hebert, J.R. and Miller, D.R. (1994). A Cross-national Investigation of diet and bladder cancer. Eur J Cancer. Vol. 30, No.1, pp. 778-784.

Howe, G.R., Burch, J.D., Miller, A.B., Cook, G.M., Esteve, J., Morrison, B., Gordon, P., Chambers, L.W., Fodor, G. and Winsor, G.M. (1980). Tobacco use, occupation, coffee, various nutrients, and bladder cancer. J Natl Cancer Inst. Vol. 64, No. 4, pp.701–713.

Jankovic, S. & Radosavljevic, V. (2007). Risk Factors for Bladder Cancer. Tumori. Vol. 93, No.1, pp. 4-12

Jemal, A., Murray, T., Ward, E., Samuels, A., Tiwari, R.C., Ghafoor, A., Feuer, E.J. and Thun, M. T. (2005). Cancer statistics 2005. CA Cancer J Clin 2005, vol 55, pp. 10.

Jemal, A., Murray, T., Samuels, A., Ghafoor, A., Ward, E. and Thun, M. (2003). Cancer statistics 2003. CA Cancer J Clin. Vol. 53, pp. 5-26.

Jemal, A., Siegel, R., Ward, E., Hao, Y., Xu, J. and Thun MJ. (2009). Cancer statistics, 2009. CA Cancer J Clin. Vol. 59, No. 4, pp. 225-49.

Jensen, O.M., Wahrendorf, J., Knudsen, J.B. and Sorensen, B.L. (1986). The Copenhagen case–control study of bladder cancer. II. Effect of coffee and other beverages. Int. J. Cancer. Vol. 37, No.1, pp. 651–657.

158

Jiang, X., Castelao, J. E., Groshen, S., Cortessis, V. K., Shibata, D., Conti, D. V., Yuan, J.M., Pike, M.C. and Dominguez, M.G. (2009). Urinary tract infections and reduced risk of bladder cancer in Los Angeles. British Journal of Cancer. Vol. 100, No. 1, pp. 834–839. www.bjcancer.com

Kadlubar, F.F., Dooley, K.L., Teitel, C.H., Roberts, D.W., Benson, R.W., Butler, M.A., Bailey, J.R., Young, J.F., Skipper, P.W. and Tannenbaum, S.R. (1991). Frequency of urination and its effects on metabolism, pharmacokinetics, blood hemoglobin adduct formation, liver and urinary bladder DNA adduct levels in beagle dogs given the carcinogen 4-aminobiphenyl. Cancer Res. Vol. 51, No.1, pp. 4371– 4377.

Kellen, E., Zeegers, M., Paulussen, A., Van, D. M. and Buntinx, F. (2006). Fruit consumption reduces the effect of smoking on bladder cancer risk. The Belgian case control study on bladder cancer. Int J Cancer. Vol. 118, No. 10, pp. 2572-8.

Kinlen, L.J. and McPherson, K. (1984). Pancreas cancer and coffee and tea consumption: A case-control study. Br. J. Cancer. Vol. 49, pp. 93.

Kogevinas, M. and Trichopoulos, D. (2000). Urinary bladder cancer. In: Adami HO, Hunter D, Trichopoulos D (eds), Textbook of Cancer Epidemiology. 1st ed. Oxford University Press, New York, pp. 446–466.

Kogevinas, M., Mannetje, A., Cordier, S., Ranft, U., Gonzalez C.A., Vineis P., Claude, J. C., Lynge, E., Wahrendorf, J., Tzonou, A., Jockel, K. H., Serra, C., Porru, S., Hours, M., Greiser, E. and Boffetta, P. (2003). Occupation and bladder cancer among men in Western Europe. Cancer Causes and Control. Vol. 14, No.1, pp. 907-914.

159

Kunze, E., Claude, C. J. and Beyme, F.R. (1993). Etiology, pathogenesis and epidemiology or urothelial tumors. [Article in German] Verh Dtsch Ges Pathol, Vol. 77, No. 1, pp. 147-56.

Kurttio, P., Pukkala, E., Kahelin, H., Auvinen, A. and Pekkanen, J. (1999). Arsenic Concentrations in Well Water and Risk of Bladder and Kidney Cancer in Finland. Environmental Health Perspectives, Vol. 107, No.9, pp. 705-710.

Lund, E. and Borgan, J.K. (1987). Increased lung cancer mortality among Norwegian cooks. Scand J Work Environ Health, Vol.13, pp.156.

Mannetje, A., Kogevinas, M., Claude, J.C., Cordier, S., Gonzalez, C.A., Hours, M., Jockel, K.H, Audorff, U.B, Lynge, E., Porru, S., Donato, F., Ranft, U., Serra, C., Tzonou, A., Vineis, P., Wahrendorf, J. and Boffetta, P. (1999). Occupation and Bladder Cancer in European Women. Cancer causes & control. Vol. 10, No. 3, pp. 209-217.

Mantanoski, G., and Elliott, E. (1981). Bladder cancer epidemiology. Epidemiology Rev: Vol. 3, pp. 203- 229.

Maria, S., Cordier, S., Claude, J. C., Donato, F., Antonio E.P., Fernandez, F., Gonzalez, C. A., Greiser, E., Jockel, K.H., Lynge, E., Mannetje, A., Hermann, P., Porru, S. and Serra, C.(2000). Coffee Consumption and Bladder Cancer in Nonsmokers: A Pooled Analysis of Case-Control Studies in European Countries. Cancer Causes & Control, Vol. 11, No. 10, pp. 925-931.

McLaughlin, J.K., Blot, W.J., Mandel, J.S., Schuman, L.M., Mehl, E.S. and Fraumeni, Jr. J.F. (1983). Etiology of cancer of the renal pelvis. J. Natl Cancer Inst., 71, 287.

McLaughlin, J.K., Mandel, J.S., Blot, W.J., Schuman, L.M., Mehl, E.S. and Fraumeni, Jr. J.F. (1984). A population-based case-control study of renal cell carcinoma. J. Natl Cancer Inst. Vol. 72, pp. 275.

160

Meijden A.P., Mantanoski, G. and Elliott, E. (1998). Fortnightly review: bladder cancer. BMJ; Vol. 17, No.7169, Pp.1366-9: Bladder cancer epidemiology. Epidemiology Rev 1981; Vol. 3, pp.203-229.

Meijden, A.P. (1998). Fortnightly review: bladder cancer. BMJ; Vol. 317, No. 7169, pp.1366-9.

Michaud, D.S., Kogevinas, M., Cantor, K. P., Villanueva, C. M., Closas, M. G., Rothman, N., Malats, N., Real F. X., Serra, C., Closas, R. G., Tardon, A., Carrato, A., Dosemeci, M., and Silverman, D. T. (2007). Total Fluid and Water Consumption and the Joint Effect of Exposure to Disinfection By-Products on Risk of Bladder Cancer. Environ Health Perspect. Vol.115, No.11, pp.1569–1572.

Michaud, D.S., Spiegelman, D., Clinton, S.K., Rimm, E.B., Curhan, G.C., Willett, W.C. and Giovannucci, E.L. (1999). Fluid intake and the risk of bladder cancer in men. N Engl J Med. Vol. 340, No. 1, pp.1390–7.

Miller, A.B., Howe, G.R., Jain, M., Craib, K.J.P. and Harrison, L. (1983). Food items and food groups as risk factors in a case-control study of diet and colorectal cancer. Int J Cancer. Vol. 32, pp.155.

Miller, C.T., Neutel, C.I., Nair, R.C., Marrett, L.D., Last, J.M. and Collins, W.E. (1978). Relative importance of risk factors in bladder carcinogenesis. J. Chronic Dis. Vol. 31, pp.51.

Momas, I., Daures, J. P., Festy, B., Bontoux, J. and Gremy, F. (1994). Bladder Cancer and Black Tobacco Cigarette Smoking: Some Results from a French Case- Control Study. European Journal of Epidemiology, Vol. 10, No. 5, pp. 599-604.

161

Mommsen, S., Aagaard, J. and Sell, A. (1983). An epidemiological study of bladder cancer in a predominantly rural district. Scand J Urol Nephrol. Vol. 17, No. 3. pp. 307-12.

Montgomery, D.C., Peck, E.A and Geoffrey Vining, G. (2004). Introduction to Linear Regression Analysis (3rd Edition), John Wiley and Sons, New York.

Morgan, R.W. and Jain, M.G. (1974). Bladder cancer: Smoking, beverages and artificial sweeteners. Can. Med. Assoc. J. Vol. 111, pp. 1067.

Morrison, A.S. (1984). Advances in the etiology of urothelial cancer. Urol. Clin. North Am. Vol. 11, No.4, pp. 557-566.

Negri, E., and Vecchia, C.L. (2001). Epidemiology and prevention of bladder cancer. Euro J Cancer Prev. Vol.10, pp.7-14.

Parkin, D.m., Pisani, P. and Ferlay, J. (1999). Estimate of worldwide incidence of 25 major cancers in 1990. International Journal of Cancer, Vol. 80, pp.827-841.

Pelucchi, C., Vecchia, L. C., Negri, E., DalMaso, L. and Franceschi, S. (2002). Smoking and other risk factors for bladder cancer in women. Prev Med. Vol. 35, No. 2, pp. 114-20.

Pitard, A., Brennan, P., Clavel, J., Greiser, E., Lopez, A. G. and Claude, C. J. (2001). Cigar, Pipe, and Cigarette Smoking and Bladder Cancer Risk in European Men. Cancer Causes & Control. Vol. 12, No. 6, pp. 551-556.

Pohlabeln, H., Jockel, K.H. and Audorff, U. B. (1999). Non-Occupational Risk Factors for Cancer of the Lower Urinary Tract in Germany: European Journal of Epidemiology. Vol. 15, No. 5, pp. 411-419.

162

Puente, D., Hartge, P., Greiser, E., Cantor, K. P., King, W.D., Gonzalez, C. A., Cordier, S., Vineis, P, Lynge, E., Claude, J. C., Porru, S., Tzonou, A, Jockel, K.H., Serra, C., Hours, M., Lynch, C. F., Ranft, U., Wahrendorf, J., Silverman, D., Fernandez, F., Boffetta, P. and Kogevinas, M. (2006), A pooled analysis of bladder cancer case–control studies evaluating smoking in men and women. Cancer Causes and Control. Vol.17, No. 1, pp. 71–79.

Rabbani, F. and Cordon, C. C. (2000). Mutation of cell cycle regulators and their impact on superficial bladder cancer. Urol Clin North Am, Vol. 27, pp. 83-102.

Radosavljevic, V., Jankovic, S. and Denic, M.L. (2001). Risk factors for urinary bladder cancer, Srp Arh Celok Lek. Vol. 129, No. 7-8, pp. 180-2.

Radosavljević, V., Janković, S., Marinković, J. and Dokić M. (2004). Non-occupational risk factors for bladder cancer: a case-control study. Tumori. Vol. 90, No. 2, pp.175-80.

Radosavljevic´, V., Jankovic´, S., Marinkovic´, J. and Dokic´, M. (2005). Diet and bladder cancer: A case-control study. Int Urol Nephrol. Vol. 37, No.1, pp. 283- 289.

Rafique, M. (2005). Clinico-pathological features of bladder carcinoma in women in Pakistan and smokeless tobacco as a possible risk factor. World Journal of Surgical Oncology. Vol. 3, No. 53. doi: 10.1186/1477-7819-3-53

Reimar, R. W. G., Trpeski, L. and Kenneth C. J. (2004). A Case-Control Study of Occupational Risk Factors for Bladder Cancer in Canada. Cancer Causes and Control. Vol. 15, No.10, pp. 1007-1019.

163

Riboli, E., Gonzales, C.A., Lopez-Abente, G., Errezola, M., Izarzugaza, I., Escolar, A., Nebot, M., Hemon, B. and Agudo, A. (1991). Diet and bladder cancer in Spain: a multi-centre case-control study. Int J Cancer. Vol. 49, No.1, pp. 214-219.

Risch, H.A., Burch, J.D., Miller, A.B., Hill, G.B., Steele, R. and Howe, G.R. (1988). Dietary factors and the incidence of cancer of the urinary bladder. Am J Epidemiol. Vol. 127, pp. 1179-91.

Sala, M., Cordier, S., Claude, J. C., Donato, F., Pujolar, A. E., Fernandez F., Gonzalez, C. A., Greiser, E., Jockel, K. H., Lynge, E., Mannetje, A., Pohlabeln, H., Porru, S., Serra, C., Tzonou, A., Vineis, P., Wahrendorf, J., Boffetta, P. and Kogevinas, M. (2000). Coffee Consumption and Bladder Cancer in Nonsmokers: A Pooled Analysis of Case-Control Studies in European Countries. Cancer Causes and Control. Vol. 11, No. 10, pp. 925-931.

Samanic, C., Kogevinas, M., Dosemeci, M., Malats, N., Real, F.X., Closas, G.M., Serra, C., Carrato, A., Closas, G. R., Sala, M., Lloreta, J., Tardón, A., Rothman, N. and Silverman, D.T. (2006). Smoking and bladder cancer in Spain: effects of tobacco type, timing, environmental tobacco smoke, and gender. Cancer Epidemiol Biomarkers Prev. Vol. 15, No. 7, pp.1348-54 Erratum in: Cancer Epidemiol Biomarkers Prev. Vol. 15, No. 8, pp. 1568.

Samanic, C.M., Kogevinas, M., Silverman, D.T., Tardón, A., Serra, C., Malats, N., Real, F.X., Carrato, A., Closas, G. R., Sala, M., Lloreta, J., Rothman, N. and Dosemeci, M. (2008). Occupation and bladder cancer in a hospital-based case- control study in Spain. Occup Environ Med. Vol. 65, No. 5, pp.347-53.

Siemiatycki, J., Richardson, L., Straif, K. and Latreille, B. (2004). Listing occupational carcinogens. Environ Health Perspec. Vol. 112, No.15, pp.1447-1459.

164

Silverman, D. and Devesa, S. (2006). Bladder Cancer. In Cancer Epidemiology and Prevention: Third Edition. Oxford University Press: Oxford, pp. 1101-1127.

Silverman, D., Morrison, A. and Devesa, S. (1996). Bladder cancer In: Schottenfeld D, Fraumeni J (eds), Cancer Epidemiology and Prevention. 2nd edn. Oxford University Press, New York, pp. 1156–1179.

Shackley, David, C., Clarke and Noel, W. (2005). Impact of socioeconomic status on bladder cancer outcome. Current Opinion in Urology: Vol. 15, No. 5, pp. 328-331

Slattery, M.L., West, D.W. and Robinson, L.M. (1988). Fluid intake and bladder cancer in Utah. Int. J. Cancer. Vol. 42, No.1, pp.17-22.

Stefani, E. D., Boffetta, P., Pellegrini, H. D., Correa, P., Ronco, A. L., Brennan, P., Ferro, G., Acosta, G. and Mendilaharsu, M. (2007). Non-alcoholic beverages and risk of bladder cancer in Uruguay. BMC Cancer. Vol. 7, No. 57, pp.1-8.

Stocks, P. (1970). Cancer mortality in relation to national consumption of cigarettes, solid fuel, tea and coffee. Br. J. Cancer. Vol. 24, pp. 215.

Sullivan, J.W. (1982). Epidemiologic survey of bladder cancer in greater New Orleans. J. Urology. Vol. 128, pp. 281.

Sylvester, R.J., Oosterlinck, W. and Meijden A.P. (2004). A single immediate postoperative instillation of chemotherapy decreases the risk of recurrence in patients with stage Ta, T1 bladder cancer: a meta-analysis of published results of randomized clinical trials. J Urol. Vol. 171, pp.2186–90.

Tajima, K. and Tominaga, S. (1985). Dietary habits and gastrointestinal cancers: A comparative case-control study of stomach and large intestinal cancers in Nagoya, Japan. Jap J Cancer Res. Vol. 76, pp. 705.

165

Takkouche B.; Etminan, M. and Montes, M.A. (2005). Personal use of hair dyes and risk of cancer: a meta-analysis. JAMA. Vol. 293, N0.1, pp. 2516-2525.

Tanagho, E.A. and McAninch, J. W. (2008). Smith’s General Urology: 17th Edition. The McGraw-Hill Companies, Inc. New York, America. pp. 321-22.

Tola, S. (1980). Occupational cancer of the urinary bladder. J Toxicol Environ Health. Vol. 6, No.5, pp.1253-1260.

Ugnat, A.M., Luo, W., Semenci, W.R. and Mao, Y. (2004). Canadian Cancer Registries Epidemiology Research Group : Occupational exposure to chemical and petrochemical industries and bladder cancer risk in four western Canadian provinces. Chronic Dis Can. Vol. 25, No. 2, pp.7-15.

Varkarakis, M.J., Gaeta, J., Moore, R.H., et al. (1974). Superficial bladder tumor: Aspects of clinical progression. Urology. Vol. 4, pp. 414.

Villanueva, C.M., Cantor, K.P., King, W.D., Jaakkola, J.J., Cordier, S., Lynch, C.F., Porru, S. and Kogevinas, M. (2006). Total and specific fluid consumption as determinants of bladder cancer risk. Int J Cancer. Vol. 118, No.1, pp. 2040-2047.

Viscoli, C.M., Lachs, M.S. and Horwitz, R.I. (1993). Bladder cancer and coffee drinking: a summary of case-control research. Lancet 1993, Vol. 341, pp.1432-7.

Walter, J. (1977). Radiation hazards and protection: Cytotoxic chemotherapy. In: Cancer and Radiotherapy: A short guide for nurses and medical students (ed J. Water), Churchill Livingstone, London.

Wells, M. (2001). The impact of cancer. In: Cancer Nursing Care in Context (eds .J Corner and C. Bailey), Blackwell Publishing, Oxford.

166

Wickremasinghe, R.L. (1978). Tea. Adv. Food Res. Vol. 24, pp. 229.

Willis, R.A. (1952). The Spread of Tumors in the Human Body. Butterworth & Co, London.

Wynder, E.L. and Goldsmith, R. (1971). The epidemiology of bladder cancer: a second look. Cancer. Vol. 40, pp. 1246-1268.

Yarbro, C., Frogge, M. and Goodman, M. (2005). Cancer Nursing: Principles and Practices, 6th edn, Jones and Bartlett Publishers, Boston, MA.

Yaris, F., Dikici, M.F., Sabuncu, H.H. and Yaris, E. (2006). A case-control study on the etiology of urinary bladder cancer in Istanbul, Turkey. Asian Pac J Cancer Prev. Vol. 7, No. 4, pp. 591-4.

Zeegers, M.P., Volovics, A., Dorant, E., Goldbohm, R.A. and van den Brandt, P.A. (2001). Alcohol consumption and bladder cancer risk: results from The Netherlands Cohort Study. Am J Epidemiol. Vol. 153, No. 1, pp. 38-41.

Zeegers, M.P., Dorant, E., Goldbohm, R.A. and Van Den Brandt P.A. (2001). Are coffee, tea, and total fluid consumption associated with bladder cancer risk? Results from the Netherlands Cohort Study. Cancer Causes Control. Vol. 12, No.1, pp. 231-238.

Zeegers, M.P., Alexandra G., Piet, A. and Brandt, V.D. (2002). A Prospective Study on Active and Environmental Tobacco Smoking and Bladder Cancer Risk (The Netherlands). Cancer Causes & Control. Vol. 13, No. 1, pp. 83-90.

Zeegers, M.P., Kellen, E., Buntinx, F. and Van Den Brandt, P.A. (2004). The association between smoking, beverage consumption, diet and bladder cancer: a systematic literature review. World J Urol. Vol. 2, No.1, pp. 392-401.

167

Bibliography

Affifi, A.A and Clark, V. (1996). Computer aided multivariate analysis, third edition. Chapman and Hill, Great Britain, pp 21-24.

Agressti, A. (1996). An introduction to categorical data analysis. John Willey and sons INC. Untied states of America.

Bishop, Y.V.V, Fienberg, P.W. (1975). Discrete Multivariate analysis. Cambridge, Ma: MIT press. Holland.

Chatterjee, S. and Hadi, A.S. (2006). Regression Analysis by Examples (4th edition). John Willey and Sons, Inc. Hoboken, New Jersey.

Cox, D.R. (1970). Analysis of binary data London: Chapman and Hill.

Grizzle, J. E.C. F. Starmer and G.G. Koch. (1969). Analysis of Categorical data by linear models. Biometrics.

Hauk, W.W. and Donner, A. (1997). Wald’s test as applied to hypothesis in logit analysis. J. Amer. Statist. Assoc.

Hosmer, D.W. and Lemeshow, S. (1989). Applied Logistic Regression. John willey and sons INC. United State of America. pp. 54-60.

Indrayan, A. (2006). Basic methods of medical research (2nd edition). A.I.T.B.S, India.

Kleinbaumn, D.G. (1992). Logistic regression A self-learning text. Springer - Verlag, New York INC.

Montgomery, D.C., Peck, E.A and Geoffrey Vining, G. (2004). Introduction to Linear Regression Analysis (3rd Edition), John Wiley and Sons, New York.

168

Richard. H, and Roberet, J. (2006). Study guide to epidemiology and biostatistics (6th edition). Jones and Bartlett publishers, London, U.K.

Sundar, R. and Richard, J. (2007). Introduction to biostatistics and research methods. Prentice Hall of India, New Delhi.

Snedecor, G. W. and Coxhran, W. G. (1980). Statistical methods. John Wiley and Sons. New York.

169

A STATISTICAL STUDY OF THE RISK FACTORS FOR URINARY BLADDER CANCER IN PAKISTAN

Name of patient: ...... …………………………… Date: ……………………

Gender: …………… Residential Address:…………………………………………………….

Province: ...... Name of Hospital: ………………………………………………….……

Personal Profile

S. No. Factors Response 1. Bladder cancer Yes No 2. Monthly income(I) (Social Status) Low(I<10000) Medium(10000

Rs.______High (I>20000) 3. Age Age<15, 1560 4. Residential Area a) Industrial Non industrial b) Rural Urban 5. Marital Status Unmarried Married Widow Divorced 6. Life Style Sedentary Normal Active 7. Profession 8. Chemical Exposure Yes (Type______) No 9. Education ______Literate Illiterate 10. Family history of cancer Positive Negative 11. Personal history of cancer Positive Negative 12. Smoking cigarettes Yes No 13. No. of cigarettes smoked per day 14. No. of Years smoking cigarette 15. No. of Years stop smoking Cig. 16. Smoking ( Huqqa, ______) Yes No 17. Chew Pan Yes (With tobacco, Without) No

170

18. Use of tea Yes(No. of cups daily _____ ) No 19. Use of coffee Yes(No. of cups daily _____ ) No 20. Use of Alcohol Yes (Amount ______) No 21. Chemotherapy Yes No 22. Radiation therapy Yes No 23. Chronic bladder irritation Yes No 24. Fluid consumption No. of glasses (daily consumed) ______25. Defect in bladder by birth Yes No

26. Hair dyes Yes No 27. Drinking water (source) Tap Canal Govt. provided Mineral water 28. Hepatitis Yes (Type ____ ) No 29. Diabetes Yes No 30. Eating habits 1.Fried items

Low Normal Excessive 2.Fats items

Low Normal Excessive 3.Fast food Low Normal Excessive 4.Fruits

Low Normal Excessive

171

Summary Table of Literature Review

Authors Sample Size Name of Significant factors Cases Controls Papers Journal Mommsen et al (1983) 212 259 An epidemiological Scand J Urol Cigarette smoking, Pipe smoking, study of bladder cancer Nephrol Cigar smoking, Chew tobacco, in a predominantly Industrial worker, Petroleum rural district workers, Alcohol users, Working with oil and chemicals. Burns and Swanson (1991) 2,160 3,979 Risk of Urinary Cancer Causes cigarette smokers and mechanic by Bladder Cancer among and Control occupation Blacks and Whites: The Role of Cigarette Use and Occupation Kunze et al (1993) 675 675 Etiology, pathogenesis Verh Dtsch Ges Cigarette smoking, Coffee and epidemiology or Pathol consumption, Use of beer, urothelial tumors Consumption of high fats, Printing, Plastics, Rubber, Mining, Dyestuffs Industries, Oils, Petroleum, Exposure to paints and pesticides, Truck drivers. Momas et al (1994) 219 794 Bladder Cancer and European Number of cigarettes smoked per Black Tobacco Journal of day, Period of smoking in years, Cigarette Smoking: Epidemiology Life time smoking, Some Results from a French Case- Control Study. Pohlabeln et al (1999) 300 300 Non-Occupational Risk European Cigarette smokers, Number of Factors for Cancer of Journal of cigarettes smoked per day, the Lower Urinary Epidemiology. Cigarette smoking period, Age at

172

Tract in Germany. beginning of cigarette smoking, Number of years stop smoking, Users of 2 or more cups of coffee, Users of 3 or more bottles of beer per day Mannetje et al (1999) 700 2,425 Occupation and Cancer causes Metal workers (blacksmiths, tool Bladder Cancer in & control. makers and machine tool European Women. operators), Tobacco workers, farm worker (crop and vegetable farm workers), Tailors and dress makers, Sales female, Mail sorting clerks and cigarette smokers Sala et al (2000) 564 2,929 Coffee Consumption Cancer Causes Heavy coffee drinkers and Bladder Cancer in and Control. Nonsmokers: A Pooled Analysis of Case- Control Studies in European Countries. Brennan et al (2001) 685 2,416 The Contribution of Cancer Causes Period of cigarette smoking in Cigarette Smoking to & Control. years, Stop smoking period and the Bladder Cancer in number of cigarettes smoked per Women (Pooled day European Data). Radosavljevic et al (2001) 13o 130 Factors for urinary Srp Arh Celok Animal fats, Pickled food and habit bladder cancer. Lek of smoking were risk factors while higher educational level and more frequency of urination were protective factors. Pitard et al (2001) 2,279 5,268 Cigar, Pipe, and Cancer Causes Cigarette smokers, Pipe smokers, Cigarette Smoking and & Control Cigar smokers Bladder Cancer Risk in

173

European Men. Pelucchi et al (2002) 110 298 Smoking and other risk Prev Med Cigarette smokers, factors for bladder Workers in dyes or paints factories, cancer in women. Workers in chemical and pharmaceutical industries Zeegers et al (2002) 619 3,346 A Prospective Study on Cancer Causes Current smokers, Ex-smokers, Active and & Control. Smoking period, Number of Environmental cigarettes smoked per day Tobacco Smoking and Bladder Cancer Risk (The Netherlands). Kogevinas et al (2003) 3,346 6840 Occupation and bladder Cancer Causes Knitters, Automobile painters, cancer among men in & Control Machinists, Automobile mechanics Western Europe. and textile machinery mechanics, Metal workers, Painters, Miners, Excavating-machine operators, Transport operators, Textile and electrical workers and in non- industrial workers like janitors and concierges. Reimar et al (2004) 887 2,847 A Case-Control Study Cancer Causes Cigarette smoking. Hairdressers, of Occupational Risk & Control primary metal workers, minors, and Factors for Bladder auto mechanics, general labourers, Cancer in Canada. firefighters, printers, government inspectors, and welders while in females, lumber processors, general labourers nurses and general clerks. While fruits are the protective factors. Ugnat et al (2004) 549 1099 Canadian Cancer Chronic Dis Can. Cigarette smokers, Registries Coffee drinkers.

174

Epidemiology Research Group : Occupational exposure to chemical and petrochemical industries and bladder cancer risk in four western Canadian provinces. Radosavljevic et al (2004) 130 130 Non-occupational risk Tumori Frequent daily urination, factors for bladder consumption of fruit juices and cancer: a case-control cabbage were referred as protective study. factors while the consumption of liver, canned meat, pork and vinegar were the risk factors. Radosavljevic et al (2005) 130 130 Diet and bladder Int Urol Consumption of liver, eggs, pork cancer: A case-control Nephrol. and pickled vegetable were the risk study. factors while cereals, kale, cabbage, tangerines and carrots were protective factors. Samanic et al (2006) 1,219 1,271 Smoking and bladder Cancer Current smokers, Ex-smokers, cancer in Spain: effects Epidemiol Smoking period, Stop smoking of tobacco type, timing, Biomarkers period, amount of smoking. environmental tobacco Prev. smoke, and gender. Yaris et al (2006) 290 580 A case-control study on Asian Pac J Smoking in males and females the etiology of urinary Cancer Prev. bladder cancer in Istanbul, Turkey. Kellen et al (2006) 200 385 Fruit consumption Int J Cancer. Fruits as protective and smoking as reduces the effect of risk.

175

smoking on bladder cancer risk. The Belgian case control study on bladder cancer. Puente et al (2006) 8,316 17,406 A pooled analysis of Cancer Causes Cigarettes smoking, amount of bladder cancer case– & Control. smoking and duration of smoking. control studies evaluating smoking in men and women. Villanueva et al (2006) 2,729 5,150 Total and specific fluid Int J Cancer. Current smokers, Ex-smoker, consumption as Total fluid, Users of tap water, determinants of bladder Users of more than 5 cups of cancer risk. coffee, Jankovic and Radosavljevic Meta analysis by Risk Factors for Tumori. cigarette smoking, Arsenic drinking (2007) literature review Bladder Cancer. water, schistosomiasis, Occupational exposure and drugs used in chemotherapy are the risk factors while high intake of fresh fruits, vegetables and fluid are protective factors. Alberg et al (2007) 265 Unspecified A prospective cohort Am J Epidemiol Passive smokers had higher risk of study of bladder cancer bladder cancer. risk in relation to active cigarette smoking and household exposure to second hand cigarette smoke. Stefani et al (2007) 255 501 Non-alcoholic BMC Cancer Increasing the time duration and beverages and risk of amount of maté drinking, Coffee bladder cancer in and tea.

176

Uruguay. Michaud et al (2007) 397 664 Total Fluid and Water Environ Health Intake fluids provide protection for Consumption and the Perspect. male but not in case of female. Joint Effect of Exposure to Disinfection By- Products on Risk of Bladder Cancer. Samanic et al (2008) 1,219 1,271 Occupation and bladder Occup Environ Machine operators in the printing cancer in a hospital- Med. industry and workers in the based case-control transportation equipment industry, study in Spain Workers in the sanitary / electrical / gas services, workers of hotels and houses, mechanics and supervisors in production industries are at high risk. Jiang et al (2009) 1,586 1,586 Urinary tract infections British Journal history of bladder infection in and reduced risk of of Cancer. female reduces the risk of bladder bladder cancer in Los cancer while heavy decrease was Angeles. observed among female having multiple infections.

Baris et al (2009) 1,170 1,413 A Case–Control Study JNCI ( J Natl Current cigarette smokers, of Smoking and Cancer Inst.) Pack years and intensity of Bladder Cancer Risk: smoking. Emergent Patterns Over Time: