Prevalence of Non-Communicable Diseases and Its Associated Factors Among Urban Elderly of Six Indian States

Aims The aim of this study is to investigate the prevalence, impact of health determinants on non-communicable diseases (NCDs), and multimorbidity among urban elderly in India. Methods This is a cross-sectional study involving a total of 1,671 (870 male and 801 female) respondents aged 60-80 years. Multistage sampling was used for the recruitment of the participants. A total of 12 sample areas from 12 cities of six southern states of south India were selected. Through survey form, information regarding demographic characteristics, health-influencing lifestyle factors, and history of nine NCDs was collected. Results The mean age of participants was 68.5 ± 6.01 years.. The prevalence of hypertension was 40.4%, followed by diabetes (31.2%), arthritis (22.1%), sensory impairment (10.1%), heart diseases (7.8%), and dyslipidemia (7.0%). 74.1% of participants had at least one morbidity, and 40.0% of people had multimorbidity. Being overweight is the highest risk health determinant for hypertension, diabetes, heart disease, high cholesterol, stroke, and joint pain. Obese people have 64% more risk of hypertension than people with normal BMI. People with disturbed sleep have increased risk of hypertension, high cholesterol, and joint pain by more than 80% compared to people with proper sleep. Among the modifiable health determinants of obesity, disturbed sleep, constipation, and physical activity up to 30 minutes were positively associated with multimorbidity. Those in the age group of 70 to 80 years have a high risk for NCDs and multimorbidity compared to those in the age group of 60 to 70 years. Conclusions A healthy lifestyle is necessary to reduce the burden of NCDs among the elderly. Developing holistic health policies seems an urgent need.


Introduction
The world population is rapidly aging due to increased longevity, decreased fertility, and mortality rates [1]. As per World Population Prospects, one in 11 individuals was above 65 years in 2019, and by 2050, one in six people will be above 65 years of age [2]. The elderly are at higher risk for multiple health challenges. Noncommunicable diseases (NCDs) are highly prevalent in later life. NCDs have become a global health agenda because they cause global morbidity, disability, and death in later life [3]. Degenerative aging processes, environmental, social, and unhealthy lifestyles are risk factors for diseases in old age.
Along with aging, behavioral factors such as tobacco use, physical inactivity, excess consumption of alcohol, and an unhealthy diet further increase the risk of NCDs and mortality due to NCDs [4,5]. Globally every year, 71% of deaths are due to NCDs. Three-fourths of these deaths occur in low-and middle-income countries [6]. The presence of two or more chronic diseases called multimorbidity is also increasing worldwide [7]. Low-and middle-income countries like India face multiple health challenges arising from NCDs and an increasingly elderly population. World Health Organization gives importance to healthy aging [8].
People living in urban areas are increasing worldwide [9]. It is leading to serious health challenges [10]. Urbanization is directly related to unhealthy behavioral adoption [11,12]. It increases the risk of NCDs [13]. Urban residents are more likely to have multimorbidity [14]. Elderly urban residents are not exceptions to these health challenges [15]. Mortality rates are high among individuals with multimorbidity [16]. Currently, there is a need for more population-based estimates of NCDs among the urban elderly in India. Knowing the current epidemiology of NCDs among the urban elderly will help formulate preventive, promotive, and curative health. With this objective, we undertook a prevalence study among six southern states of India.

Study design and setting
A stratified multistage sampling strategy was chosen for sampling. The study was conducted from October 2018 to September 2019. The study covered urban areas of six states of south India (Kerala, Tamil Nadu, Karnataka, Andhra Pradesh, Telangana, and Maharashtra). From each state, major city names were listed. Out of these, two cities were selected based on a higher proportion of urban population selectively in each state. A total of 12 cities selected were Cochin, Trivandrum, Chennai, Madurai, Bengaluru, Dharwad, Tirupati, Visakhapatnam, Hyderabad, Khammam, Pune, and Thane. The selected city was divided into four geographical zones (East, West, South, North). Furthermore, one geographical zone was selected randomly, and from the selected zone, one residential locality/ward was selected randomly for the survey. Local health workers/influential leaders were contacted for the cooperation and support required during data collection. This helped in decreasing the non-response rate. One household was selected randomly, and after that, subsequent households were interviewed sequentially till approximately 150 elderly were selected from each city based on the overall sample size estimated for the study. In a household where more than one elder person lived, only one was interviewed. Details are shown in Figure 1.

FIGURE 1: Flow chart of the recruitment of the urban elderly residents
Twelve sample areas from 12 cities in six states were selected. The Institutional Ethics Committee of Swami Vivekananda Yoga Anusandhana Samsthana issued approval (RES/IEC-SVYASA/117/2017). Written informed consent was obtained from each participant before interviewing.

Participants
For participants, inclusion criteria were as follows: (1) people aged between 60 and 80 years, (2) living in urban areas for more than 20 years, and (3) willingness to participate and give written informed consent. The elderly living in households and residents of elderly care homes of the selected locality participated in the study. The survey sample size was calculated using Raosoft Inc. (2004). Keeping the confidence level 99%, the margin of error at 3%, and response distribution at 50%, the elderly population in selected six states was 38.7 million. A total of 1,843 was the calculated sample size. After contacting the participants, they were screened at their residence for inclusion. A total of 2,076 elderly were screened, and 1,830 met the inclusion criteria. A total of 1,753 agreed to participate in the survey study, and after the removal of incomplete data, data of 1,671 participants were included in the analysis.

Data collection and measurements
A trained interviewer interviewed all the participants who consented. All the field workers underwent training prior to starting the study. They were trained using a standard protocol, a standard questionnaire, and methods of taking measurements. Participants were interviewed at their residences to reduce the nonresponsive bias. The information was collected from the participants for the demographic characteristics, health-influencing lifestyle factors, and history of medical illness through a standard pre-tested questionnaire. As mentioned earlier in Figure 1, missing or incomplete forms were not taken into the final analysis.
Demographic information included gender (male and female), age, marital status (married, never married, divorced, and widowed), residence (living with family, living separate, and elderly care home), education (total years of education categorized into 1 to 5, 6 to 12, and more than 12 years), employment (selfemployed, private sector, government sector), present work status (retired, full time working, part-time working, homemaker), and economic status (dependent on family, pension, salary, self-sustainable).
Health-influencing lifestyle factors were assessed by collecting information on diet type (vegetarian or nonvegetarian), kind of physical activity (nil or physical activity-walking, aerobics, yoga), duration of physical activity (less than a half-hour or up to one hour), the total period of physical activity in years, substance abuse (nil or substance abuse such as smoking, alcoholic, tobacco chewing, or any other), substance abuse history (current or past), total years of substance abuse, sleep (normal or disturbed), and bowels clearance (normal, constipation, irregular).
History of a prior diagnosis of hypertension, diabetes mellitus, heart diseases, dyslipidemia, sensory impairment (eyes or ear), stroke, psychiatric illness, arthritis/joint pain, and any other disease was recorded. This information was based on a previously diagnosed illness by a physician. The total duration of illness (in years) and treatment for each disease were also recorded. Height was measured manually by measuring tape in centimeters, and weight was measured by a Belita -1101 mechanical personal weighing machine. The body mass index was calculated using a standard formula for each participant. Multimorbidity was defined as having at least two chronic conditions [17].

Data analysis
Statistical analysis was conducted using the open source software R version 4.03. Descriptive analyses were performed to determine the distribution of demographic and health-influencing factors to calculate the prevalence proportions of different NCDs. In addition, we constructed two sets of models. Stepwise logistic regressions were performed for eight NCDs (hypertension, diabetes mellitus, heart disease, high cholesterol, sensory impairment, stroke, psychiatric illness, joint pain) to assess the strength and direction of the association between NCDs and their potential correlates.
We further performed a multivariate logistic regression model adjusting the odds ratio for modifiable and nonmodifiable factors separately. For each regression analysis, we used appropriate model diagnostics, and the models that we used fit well (p<0.01). The adjusted odds ratio of modifiable and nonmodifiable factors was obtained from the stepwise logistic regression model. This model uses the Akaike information criterion (AIC) to iteratively add and remove the predictors to find the subset of variables in the given data set resulting from the best-performing model, which has a lower prediction error.
Furthermore, we created a dependent variable (number of NCDs) with three clearly defined categories (suffering from no NCD/one NCD/two NCDs) with a standard order (zero, one, and two or more). This analysis aimed to determine the association of the independent variables with the odds of having a higher number of NCDs. To increase analysis efficiency, we built three multivariate step logistic regression models: first, no disease versus one disease; second, no disease versus more than one disease; and third, one disease versus more than one disease.

Results
Of 1,671 urban elderly, 870 (52.1%) were male and 801 (47.9%) were female. The details of health determinants are as follows. Overall, 54.3% were vegetarian and 45.7% were non-vegetarian. Also, 19.7% were not doing any physical activity. Most (80.3%) were engaged in physical activity like walking, aerobics, or yoga. Of these, participants doing physical activity for less than half an hour were 35.6% and those for up to one hour were 44.7%. One-fifth (20.7%) used smoking, alcohol, tobacco chewing, or other substance. Of the substance users, 9.0% are still taking these drugs, and 11.7% left the substance use in the past. One-fifth of people reported that they had disturbed sleep (20.8%) and constipation (19.7 %). Nearly half (47.3%) of participants were obese (body mass index ≥ 25), and 2.6% of participants were underweight (body mass index < 18.5). Details are given in Table 1.

Name of the disease Total Duration of illness in years, mean (SD)
Except for psychiatric illness, those aged >70 years have a higher risk for NCDs than those aged <70 years. Widowed people are 1.5 times more at risk than people having a spouse. People whose marital status is "divorced" are the riskiest group. Table 5 shows adjusted odds ratios of modifiable factors on morbidity across no disease versus one disease, no disease versus more than one disease, and one disease versus more than one disease.  Among the correlates, obesity is the common risk factor across all the morbidity segments. Disturbed sleep increases the risk two times for multimorbidity (AOR = 2.12, 95% CI: 1.48-3.06). Constipation is positively associated with multimorbidity compared to one disease and no disease (AOR = 1.34, 95% CI: 1.00-1.80; AOR = 1.92, 95% CI: 1.33-2.79). Substance habit and prolonged usage increase the risk for multimorbidity. Physical activity duration of less than 30-minute duration increased the risk of having more than one disease (AOR = 0.64, 95% CI: 0.44-0.88). As the age increased, the risk of having multiple diseases doubled (AOR = 1.39, 95% CI: 1.07-1.81).

Discussion
This is a cross-sectional survey involving a comprehensive sample of urban Indian elderly residents. The study revealed that about 71% of the survey participants had one chronic NCD, and 40.0% elderly had multimorbidity (≥ two NCDs). These findings are similar to the National Chinese Prevalence study, and the most prevalent NCDs were hypertension followed by diabetes [18]. Compared to an earlier study, BKPAI-2011, in India and South African elderly survey, the self-reported prevalence of at least one chronic NCD and multimorbidity prevalence is higher in the present study [19,20]. In the SAGE study, the prevalence of selfreported NCDs in urban elderly residents was 24.7% for hypertension and 18.1% for arthritis. In contrast, in the present survey, prevalence of hypertension is 40.4% and that of arthritis is 22.1%. This indicates that there are increasing trends in the prevalence of hypertension and arthritis. The BKPAI-2011 study reported that the most chronic NCD experienced by the urban elderly was arthritis. In the present study, hypertension is highly experienced, followed by diabetes, and has a higher prevalence than in the BKPAI-2011 study. This indicates that these NCDs are higher in urban Indian elderly residents. The survey found that the prevalence of NCDs increases with increasing age. It is also corroborating with preliminary evidence [21,22]. NCDs contributed to the disease burden in India, and this transition happened from 1990 to 2010 [21]. Furthermore, the disease burden is increasing due to NCDs. Old-age people had a higher prevalence of these NCDs compared to an earlier decade.