Using a Screening Toolkit to Determine the Prevalence of Non-Communicable Diseases in an Urban Slum of Mumbai, India: A Cross-Sectional Study

Background: Higher than national rates of non-communicable diseases have been found in some urban slums of India. This has been attributed to potentially lower levels of education and decreased access to preventative care. We sought to assess the prevalence of NCDs in a Mumbai slum compared to national averages and understand the association with social determinants of health. Methods: We used a screening toolkit called THULSI (Toolkit for Healthy Urban Life in Slums Initiative) in a community health-camp setting to screen 266 slum dwellers for obesity (BMI above 25 kg/m 2 ), elevated blood pressure (SBP > 120 mm of Hg and DBP > 80 mm of Hg), and elevated blood glucose (RPG > 200 mg/dL). A health survey was administered to understand the demographic data and information about health-seeking behavior of the slum dwellers. The collected data was analyzed to determine the prevalence of each condition and associations with different social determinants of health such as literacy and education level, diagnosis and marital status, age, gender, place of care, employment, availability of toilet, and languages spoken. Results: Of the screened population, 72.6% had elevated blood pressure, 9.0% had elevated blood glucose, 63.2% were obese, and 12.4% were overweight. These rates were 2.81, 1.04, and 3.84 times higher than the national averages, respectively. Of the study population, 26.7% had one condition, 53.4% had two conditions, and 7.9% had all the three conditions we screened for. Male gender (OR: 3.33, 95% CI: 1.37, 8.07) and older age (35-49 years OR: 3.03, 95% CI: 1.52, 6.05, 49-63 years OR: 7.22, 95% CI: 3.06, 17.05, >= 63 years OR: 6.82, 95% CI: 2.12, 22.00) were associated with increased odds of elevated blood pressure, not being previous diagnosed (OR: 0.04, 95% CI: 0.01, 0.13) was associated with lower odds of elevated blood glucose, and older age (35-49 years OR: 2.51, 95% CI:

Nationally, roughly 19.7% are overweight or obese, 25.8% have raised blood pressure, and 8.7% have raised blood glucose in India. 9,10,11 Therefore, previous research has highlighted two important NCD characteristics for the Indian slum population: higher than national rates of NCDS are found in slums and residents are often unaware of their condition.
This lack of awareness in slums could be attributed to factors such as poverty, low levels of education, and the high opportunity cost (the time and money spent in services could potentially be better utilized elsewhere) of pursuing preventative and wellness services. 12 For instance, research done in a Bangalore slum found that the median monthly income of slum dwellers was around $47 and although education was an important factor for dwellers, only the top 10% highest earning households could afford a school education for their children. 13 Low levels of education can further exacerbate unawareness of the importance of preventative health care. 14 There is also ineffective health outreach in the slums due to social exclusion. 14 Lack of early intervention programming contributes to the higher rates of NCDs and complications from these NCDs including strokes, myocardial infarctions, and kidney failure. 15 In this study, we sought to further our understanding of the rates of NCDs among slum-dwellers in India and the potential association with social determinants of health. Understanding these associations are critical to determining avenues for intervention and improved healthcare. Our speci c objective is to determine the prevalence of obesity, elevated blood pressure, and elevated blood glucose in an urban slum of Mumbai with a chronic disease screening toolkit previously used in Bangalore. We then sought to compare the prevalence to the national average and determine the presence of any associations with critical social determinants of health.

Study Setting
Mumbai is located in southwestern India and is considered to be the country's nancial and commercial centre. 16 Approximately 18.41 million live in Mumbai and roughly 40% of these individuals reside in slums. 17 This study was conducted in slums located in P-South Ward of Mumbai. The total population of the slums located in this region of Mumbai is 210,591. 18 Before the study was initiated, extensive working relationships were formed with the community leaders of the local area who are familiar with the residents of the slums located in P-South Ward. These were members of a local nonpro t organization called Shivraj Prathisthan who regularly organize events and charity drives in the slum. The project was explained in detail to the leaders and their feedback was incorporated into the development of the study. This study was approved by the Institutional Review Board at Yale University (IRB Submission ID: 2000027023) and the Ethics Committee for Biomedical and Health Research at BhaktiVedanta Hospital in District Thane.

Study Design
This was a community-based cross-sectional study. Data on socio-economic characteristics and health-seeking behaviors was collected by self-report via a health survey that was distributed at the time of the study.

The Screening Toolkit
In this study, screening was conducted for three non-communicable diseases: obesity, elevated blood glucose, and elevated blood pressure. The toolkit used for screening was called THULSI (Toolkit for Healthy Urban Life in Slums Initiative). 19 This toolkit was developed as a cross-collaborative global effort between the University of She eld, Bangalore Baptist Hospital, ZUYD University of Applied Sciences, e-Health enablers, and Icarus Nova. THULSI has already been used for screening purposes in a Bangalore slum. 8 THULSI consists of various screening tools and devices such as glucometers, weighing scales, tape measure, BP monitor, and a cuff. It also has an electronic component in the form of an android-based application that was used to collect the health survey on an electronic tablet. The health survey was tailored to collect demographic information and health-seeking behaviors of the residents. The tablet was connected to a thermal printer that was used to immediately print out the results of the screening. Trained paid workers from the community used the toolkit to screen residents. These were college-level educated women who resided in the slums and certi ed medical technicians who were uent in the local languages.
These workers were provided extensive training over a period of two days which involved practice sessions on using the toolkit. The training was overseen by Digi Health Platform Pvt Ltd, the company that manufactured THULSI.

Recruitment and Data Collection
The study was conducted over a period of four days. The study team, which included the trained workers, medical technicians, the principal investigator, and members from Shivraj Prathisthan, set up camp in the community ground of the slum for a minimum of four hours each day. This community ground was accessible to all members of the P-South Ward slums. The camp was publicized in the slums with the use of a displayed banner and word-of-mouth advertising by the non-pro t and workers. Since the camp was set up in the central community ground, residents of the slum saw it when they left their homes and they could walk-in and get screened on any of the four days. When a resident approached the study, they were rst asked for their written consent in Hindi, Marathi, or English. They then completed the health survey and measurements for blood pressure, blood glucose, height, and weight were taken by the trained medical technicians using the toolkit. Results were provided to the slum resident which could be used as a referral for future treatment at the local government healthcare provider. This helped connect residents to further care and promoted awareness of local healthcare services. Upon screening, each resident was offered a health literacy brochure containing information and preventative measures for the condition they screened positive for and a small household item that cost 100 rupees as remuneration. This promoted awareness of the NCDs and importance of preventative measures.

Standard De nitions
The standard cutoffs used for the determination of each non-communicable disease are categorized as follows:  20 Those with a BMI that fell in the ranges of overweight and obese were categorized as overweight or obese.
Elevated blood pressure: Blood pressure was recorded for all participants in the sitting position in the right arm. Those with a blood pressure that fell in the prehypertension, Stage 1 hypertension, Stage 2 hypertension, and critical ranges were categorized as having elevated blood pressure.
Elevated blood glucose: Blood glucose was measured using the glucometer for all participants. Random plasma glucose was used for the determination of elevated blood glucose. Those with a measurement greater than 200 mg/dL were categorized as having elevated blood glucose.
Sex: The two categories used in this study were male and female.
Literacy level: Participants were categorized in three groups, illiterate, able to read, and able to read and write.
Education level: Participants were categorized in three groups, those with less than high school level of education, those with a high school degree, and those with some college or a college degree.
Employment status: Participants were categorized in four groups, those who were unemployed, employed in the service or labor industry, self-employed, and the other group. The other group comprised of those who were employed in the government or private sector.
Languages spoken: Since Hindi and Marathi are the primary languages spoken in the slum, the two categories used were those who spoke Hindi or Marathi and those who spoke only another language, such as English, Bengali, Gujarati, Kannada, or Telugu.
Marital status: The two categories used were married and unmarried, separated, or widowed.
Primary place of care: The two categories used were government and private clinic or hospital.
Availability of toilet: Participants were categorized in two groups, those with a toilet in their home and those who used a community toilet. This was used as a marker for socio-economic status. This approach of using availability of toilet to assess socio-economic status is often taken in low to middle income countries since it can indicate a household's "material circumstance." 23 Diagnosis status: Participants were categorized in two groups based on whether they had been previously diagnosed or not.
Comorbidity: Participants were categorized based on whether they had one, two, or all three of the NCDs screened for in the study.

Statistical Analysis
All data analysis was done using SAS software. 24 Descriptive analysis was performed to understand the demographical characteristics of the screened population. The prevalence of each condition was analyzed. The three variables of BMI, blood pressure, and blood glucose were considered to be dependent variables and the other study variables were independent variables. Bivariate analysis in the form of chi-squared or Fisher test was performed at the 0.05 level to determine signi cant associations between each dependent and independent variable. Those independent variables that were deemed to be signi cant in the bivariate analysis were entered into a logistic regression model (both unadjusted and adjusted) to estimate the odds of having each condition.

Power and Sample Size
Minimum sample size was estimated using the formula n = [Np( where Z is the Zscore, Np is the population size, d is the margin of error, and p is the estimated prevalence of the condition based on previous studies. Sample size was determined to have 80% power with a type I error rate of 5%. A detailed sample size calculation is shown in an additional le (Additional File 1). The estimated minimum sample size is 260 among the three outcomes to be included. In our study, we screened a total of 294 individuals with 266 individuals included in this analysis due to missing data.
Of the screened population, 72.6% had elevated blood pressure, 9.0% had elevated blood glucose, 63.2% were obese, and 12.4% were overweight ( Table 2). These rates of overweight and obesity, elevated blood pressure, and elevated blood glucose were 3.84, 2.81, and 1.04 times higher than the national average, respectively. 25,11,10 Of the screened population, 26.7% only had one of the three conditions screened for, 53.4% had two, and 7.9% had all three. .2% of those with elevated blood pressure had never been diagnosed for this condition before (Table 3).  (Table 4). Upon adjustment, males had 3.33 (95% CI 1.37-8.07) times higher odds of having elevated blood pressure than females (Table 4). Without adjusting for any covariates, those who used the community toilet had 0.42 (95% CI 0.24-0.74) times lower odds of having elevated blood pressure than those with a toilet in their homes (Table 4). However, upon adjustment, this signi cant association disappeared.  (Table 3).
Without adjusting for any covariates, those between 49 to 63 years of age had 4.19 (95% CI 1.67-10.50) times higher odds of having elevated blood glucose than those in the 21-49 age category (Table 4). However, upon adjusting for other covariates, this association disappeared. Those who used the community toilet had 0.09 (95% CI 0.01-0.64) times lower odds of having elevated blood glucose than those who had a toilet in their home (Table 4). This association also disappeared upon adjusting for other covariates. Upon adjustment, those who had not been previously diagnosed had 0.04 (95% CI 0.01-0.13) times lower odds of having elevated blood glucose than those who had been previously diagnosed (Table 4).

Obesity
Age was statistically signi cantly associated with being over-weight or obese, where 67.2% of those between the ages of 35-63 were overweight or obese (Table 3).
Those between the ages of 35 and 49 and those between the ages of 49 and 63 had 2.51 times and 4.95 (95% CI 1.21-5.21, 95% CI 2.06-11.92) times higher odds of being overweight or obese than those between the ages of 21 and 35 (Table 4), respectively. This model was not adjusted for any covariates since the bivariate analysis did not show any other statistically signi cant variables.
(Insert Tables 3 and 4 here) Comorbidity There was a statistically signi cant association between age and availability of toilet and comorbidity. A majority of those with all three conditions were between 49-63 years of age (52.4%, P < 0.001) and had a toilet in their home (95.2%, P 0.033). 38.0% of those with one condition were between 21-35 years of age ( Table 5). None of the other variables were signi cantly associated with comorbidity.  understand what factors are associated with NCDs in these settings as a means of curbing the NCD epidemic in the country. The study done in Bangalore using THULSI too found that the rates of all three conditions were higher than the national average. 8 This highlights the fact that slum residents are more often vulnerable to non-communicable diseases than the general population in India.
We found that a vast majority of those with identi ed NCDs had not been previously diagnosed and had more than one condition. This signi es the need for raising awareness of the importance of screening services and providing access to adequate services. A vast majority of the screened residents were also found to be illiterate, unemployed, and had less than high school level of education. These socio-economic characteristics can play a role in the poor health-seeking behavior of the slum community due to factors such as ineffective health outreach, high opportunity cost of pursuing preventative care, and lack of awareness. 12 The study also found that a majority of participants with elevated blood pressure, elevated blood glucose, and two or more conditions had a toilet in their home. As mentioned earlier, this variable was used as a marker for socioeconomic status, and availability of toilet in the home is indicative of a higher status than those who used a community toilet. 23 A systematic review conducted by Allen et al. has suggested that a higher socio-economic status is associated with greater physical inactivity, and consumption of processed foods, salts, and fats in low-income and lower-middle-income countries. 26 This association could explain the link between presence of toilet in the home and greater prevalence of NCDs and comorbidity.
THULSI not only allowed us to bring screening to a traditionally underserved population, but also created part-time work opportunities for residents of the slum to work as study staff members. The extensive training imparted to the staff members is a step in the direction to create a self-sustaining and local model of screening for diseases. This community-based partnership approach is especially important for the success of global healthcare ventures and to allow the screening tool to reach the residents who need it the most. 27 It has been found that unawareness of the health condition and distance to the healthcare facility are often common barriers to linkage to care. 28 Residents who screened positive were provided with a health literacy brochure and linked to primary care by provision of receipts. This was a crucial part of the study to reduce barriers to linkage, ensure continuity of care, and make the residents aware of their condition and healthcare services in their area. Linkage to care is especially important to ensure long-term access to care and reduce the possibility of manifestations such as strokes, myocardial infarctions, and kidney failure. 15 There were some limitations to our study that are worth noting. For instance, the methodology of setting up camp in the community area, inviting slum residents for screening, and offering remuneration could have increased the possibility of selection bias and affected the generalizability of the study since the residents who took part in the study may not be an accurate re ection of the general slum population. We measured blood pressure and blood glucose at one time point in this population, but these measurements are not perfect markers for the presence of hypertension and diabetes since more than one abnormal value separated in time is often required to con rm the diagnosis of the two conditions. The health-survey was conducted by staff members, introducing the possibility of social desirability bias --where the screened population answers questions in a manner that will be viewed as favorable.

Conclusions
This study found high rates of non-communicable diseases in the slum population in Mumbai and provided screening services to several individuals who had never been screened before. The high percentage of unemployment, lack of education, and illiteracy signify the importance of focusing on these socio-economic characteristics for the development of this population. The results of this study highlight the excessive burden of non-communicable diseases in the urban slum in Mumbai and the need to develop preventative interventions for this population. Availability of data and materials

Abbreviations
The datasets generated and analyzed during the current study are not publicly available due to a non-disclosure agreement made between Yale University and RX Digi Health Platform Private Limited, the data owner. The data can be made available by the corresponding author on reasonable request and prior approval of the data owner and the author.