Risk factors of multimorbidity among older adults in India: A systematic review and meta‐analysis

Abstract Background Multimorbidity among older adults is a growing concern in India. Multimorbidity is defined as the coexistence of two or more chronic health conditions in an individual. Primary studies have been conducted on risk factors of multimorbidity in India, but no systematic review has been conducted on this topic. This systematic review aimed to synthesize the existing evidence on risk factors of multimorbidity among older adults in India. Methods The JBI and Preferred Reporting Items for Systematic Reviews and Meta‐Analysis guidelines were followed. Several databases were searched for published and unpublished studies until August 03, 2022. The screening of titles and abstracts and full texts, data extraction, and quality assessment were conducted by two independent reviewers. Any disagreements were resolved through discussion or by involving a third reviewer. Data synthesis was conducted using narrative synthesis and random effects meta‐analysis, where appropriate. Results Out of 8781 records identified from the literature search, 16 and 15 studies were included in the systematic review and meta‐analysis, respectively. All included studies were cross‐sectional, and 10 met a critical appraisal score of more than 70%. Broadly, sociodemographic, lifestyle, and health conditions‐related factors were explored in these studies. The pooled odds of multimorbidity were higher in people aged ≥70 years compared to 60‐69 years (odds ratio (OR) 1.51; 95% confidence interval (CI) 1.20–1.91), females compared to males (1.38; 1.09–1.75), single, divorced, separated, and widowed compared to married (1.29; 1.11–1.49), economically dependent compared to economically independent (1.54; 1.21–1.97), and smokers compared to non‐smokers (1.33; 1.16–1.52) and were lower in working compared to not working (0.51; 0.36–0.72). Conclusion This systematic review and meta‐analysis provided a comprehensive picture of the problem by synthesizing the existing evidence on risk factors of multimorbidity among older adults in India. These synthesized sociodemographic and lifestyle factors should be taken into consideration when developing health interventions for addressing multimorbidity among older adults in India.

multimorbidity among older adults in India.These synthesized sociodemographic and lifestyle factors should be taken into consideration when developing health interventions for addressing multimorbidity among older adults in India.

K E Y W O R D S
chronic diseases, elderly, health conditions-related factors, India, lifestyle factors, meta-analysis, multiple long-term conditions, senior citizen, sociodemographic factors, systematic review

| INTRODUCTION
By the end of the century, one-third of India's population will be aged 60 or more years, 1 increasing the risk of multimorbidity. 2Multimorbidity is the coexistence of two or more chronic health conditions in an individual, each one of which is either a physical noncommunicable disease (NCD) of long duration (e.g., cardiovascular disease), a mental health condition of long duration (e.g., dementia), or an infectious disease of long duration (e.g., hepatitis C), and this is most widely used definition. 3,4In India, the prevalence of multimorbidity among people aged ≥60 years ranges from 24% to 83%. 5 Multimorbidity poses a significant burden on the healthcare system and the economy along with having a negative impact on the patients and their families and carers. 6][9][10] Major health consequences of multimorbidity include negative effects on the physical and mental health of patients such as physical disabilities, psychological distress, cognitive impairment, self-doubts, depression, and anxiety. 7,11[16][17] In India, several primary studies have been carried out to identify risk factors of multimorbidity among older adults, 1,[18][19][20][21][22][23][24][25][26][27][28] but to date, no systematic review has been conducted on this topic that brings together all the available evidence.The findings could be used when developing health interventions for addressing multimorbidity among older adults in India.Therefore, this systematic review aimed to synthesize the existing evidence on risk factors of multimorbidity among older adults in India.

| METHODS
The systematic review was conducted and reported in conformity with JBI systematic reviews of etiology and risk and Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines. 29,30The review protocol was registered with PROSPERO (registration number: CRD42022348425).

| Population
The systematic review included studies conducted among older adults (aged ≥60 years) in India.The phrases "senior citizen" or "elderly" are used in India to refer to those ≥60 years, as per the National Policy on Older Persons and Social Statistics Division of the National Statistics Office. 31,32UN also defines the population aged ≥60 years as being old. 33If a study was conducted among adults, relevant data on older adults were extracted.The study was excluded if it was not possible to extract these data.Any study setting was eligible, for example, community, residential care, primary care, secondary care, and tertiary care.

| Exposure
Studies reporting any risk factor, such as sociodemographic (e.g., age and sex), lifestyle (e.g., smoking and alcohol consumption), and health conditions-related, were included in this systematic review.

| Outcome
Studies on multimorbidity as an outcome were included in this systematic review.The study authors' definition of multimorbidity was used for this purpose.If the term multimorbidity or its definition was not mentioned, an operational concept of multimorbidity was used, that is, "coexistence of two or more chronic health conditions". 36][37][38] and in consultation with a librarian (see Supplementary Materials).No date or language restrictions were applied.The reference list of all included studies and relevant systematic reviews were screened to identify any additional studies.

| Study selection
Following the search, identified citations were collated and exported using a reference manager software, Endnote X9. 39 Once duplicates were removed, records were imported into a web tool for systematic reviews, Rayyan. 40Titles and abstracts were screened against the eligibility criteria by two independent reviewers (NG and IB).Studies identified as potentially eligible or those without an abstract were retained for the full-text screening.If the full text of an article was not available even through the interlibrary loan service at the University of Nottingham (UK), the corresponding author was contacted for the same (at least twice via email).For eligibility assessment, the full text of the articles was screened independently by two reviewers (NG and IB).Any disagreements that arose between them were resolved through discussion.If a consensus was not reached, a third reviewer (KC) was involved.Following the full-text screening, studies not fulfilling the inclusion criteria were excluded, and reasons for the same were recorded (see Supplementary Materials).In the case of multiple publications from the same data set, the article having the most complete data was included.If partial data were reported in articles, then all such articles were included.

| Assessment of methodological quality
2][43] Any disagreements that arose between them were resolved through discussion.If a consensus was not reached, a third reviewer (KC) was involved.No cut-off quality score was applied to exclude studies; therefore, all eligible studies regardless of their methodological quality were included in this review.

| Data extraction
The data were extracted from included studies using a pre-developed and piloted data extraction form.The following details were extracted: publication details (first author and year of publication), Indian state, study design, study year, study setting, population characteristics (study population ≥60 years only; sample size, mean age (in years), and number of females), unadjusted risk factors, adjusted risk factors (as reported by study authors), assessment of risk factors (e.g., self-reported by participants or using medical records), definition of multimorbidity, and assessment of multimorbidity (e.g., self-reported by participants or using medical records).Odds ratios (ORs) and 95% confidence intervals (CIs) were also extracted.Adjusted ORs were preferred over unadjusted ORs.In the absence of adjusted ORs, unadjusted ORs were extracted or calculated (using the available raw data).If a study had multiple categories for a risk factor, two or more categories were merged meaningfully to form a new category for meta-analysis.For example, if general, scheduled castes, scheduled tribes, and other backward classes were the categories available for social caste, then the general category was considered as the reference group, and all the other categories were combined to form a new category.Two reviewers (NG and IB) independently extracted the data.Any disagreements that arose between them were resolved through discussion.If a consensus was not reached, a third reviewer (KC) was consulted.

| Data synthesis
Initially, narrative synthesis was conducted.Where at least two studies reported the same or similar risk factors, a meta-analysis was conducted using Review Manager (RevMan) 5.4 software. 44The ORs with 95% CIs were pooled using the random effects meta-analysis approach and generic inverse variance. 45,46The standard errors were used in creating the forest plots, which were calculated in STATA v17 using the following formula: standard error = (log upper CI-log lower CI)/3.92. 47The statistical heterogeneity across studies was estimated using I 2 statistics.The I 2 values of <50%, between 50% and 74%, and ≥75% were interpreted as low, moderate, and high levels of statistical heterogeneity, respectively. 48

| Assessment of publication bias
The publication bias was assessed using a funnel plot, provided at least 10 studies were included in the meta-analysis. 49| RESULTS

| Inclusion of studies
Eight thousand seven hundred eighty-one records were identified from the literature search, and all were in the English language.After the removal of duplicates, 7378 records were left for the title and abstract screening.Following the title and abstract screening, 93 records were left for the full-text screening.3][54][55][56][57][58][59][60][61][62] Figure 1 shows the process of study selection and inclusion.

| Characteristics of included studies
The characteristics of the included studies are presented in Table 1.
Five studies were conducted on a nationally representative sample, 19,50,57,61,65 five in the northern states, 18,54,55,58,60 three in the eastern states, 1,52,53 two in the southern states, 56,62 and one in both northern and southern states. 51All included studies were crosssectional and conducted in the community except for four (one each was conducted in residential care, 56 primary care, 62 both community and primary care, 18 and community, residential care, primary care, secondary care, and tertiary care). 58The studies were published in 2004 and after.The sample size of older adults in the included studies varied from 148 to 42,756.The mean age of older adults in the included studies ranged from 66 to 75 years.Broadly, sociodemographic, lifestyle, and health conditions-related factors were explored in the studies.All included studies used self-reported data on exposures except for one (which used medical records). 62The same definition of multimorbidity (i.e., the coexistence of two or more chronic health conditions in an individual) was reported in 12 studies, whereas it was unclear in three studies, 53,55,60 and in one study, no definition was provided. 50Nine studies used self-reported data on multimorbidity, 1,19,51,52,[56][57][58][59]61 one used medical records, 62 four used both self-reported data and medical records, 18,[53][54][55] one screened selected chronic health conditions, 60 and one had not reported the details. 50

| Methodological quality of included studies
The methodological quality of the included studies is presented in Table 2.The critical appraisal scores varied from 38% to 88%.[59]61,62 The inclusion criteria were clearly defined in all the studies except for one. 52All included studies described study participants and settings in detail except for one (where the exact study location was unclear). 551][62] The definition of multimorbidity was unclear in three studies, 53,55,60 and one had not defined it. 50Two studies did not identify the confounders, 53,54 and four had not stated strategies to deal with them. 18,52,54,60Only two studies assessed multimorbidity using a reliable and valid method. 53,60Appropriate statistical analysis, such as multivariable logistic regression, was used in all included studies except for four. 18,53,54,60I G U R E 1 PRISMA flow diagram for systematic reviews which included searches of databases, registers, and other sources.*See Supplementary Materials.
T A B L E 1 Characteristics of included studies.| 9 of 20

| Meta-analysis
Fifteen studies were included in the meta-analysis.

Economic dependency
The pooled odds of multimorbidity were higher in economically dependent people compared to economically independent people (OR: 1.54; 95% CI: 1.21-1.97).No statistical heterogeneity was found across studies (I 2 0%) (see Figure 10).F I G U R E 5 Forest plot of the association between religion and multimorbidity.Hindu religion was the reference group, and other categories (i.e., Muslims, Sikhs, Christians, and others) were combined to form the other group.*Unadjusted ORs and 95% CI.

Work status
The pooled odds of multimorbidity were lower in working people compared to those not working (OR: 0.51; 95% CI: 0.36-0.72).High statistical heterogeneity was found across studies (I 2 95%) (see Figure 11).
F I G U R E 6 Forest plot of the association between education and multimorbidity.No education or illiteracy was combined as the reference group, and primary school, secondary school, and higher education were combined to form the other group.*Unadjusted ORs and 95% CI.
F I G U R E 7 Forest plot of the association between marital status and multimorbidity.Married was the reference group, and other categories (i.e., single, divorced, separated, and widowed) were combined to form the other group.*Unadjusted ORs and 95% CI.
F I G U R E 8 Forest plot of the association between family type and multimorbidity.The nuclear family was the reference group, and the joint family was the other group.*Unadjusted ORs and 95% CI.
F I G U R E 9 Forest plot of the association between living arrangements and multimorbidity.Living alone was the reference group, and other categories (i.e., living with a spouse, living with children, living with both spouse and children, and living with others) were combined to form the other group.*Unadjusted ORs and 95% CI.
F I G U R E 10 Forest plot of the association between economic dependency and multimorbidity.Economically independent was the reference group, and partially dependent and totally dependent were combined to form the other group (i.e., economically dependent).*Unadjusted ORs and 95% CI.
F I G U R E 11 Forest plot of the association between work status and multimorbidity.Not working was the reference group, and working was the other group.*Unadjusted ORs and 95% CI.
F I G U R E 12 Forest plot of the association between socioeconomic status and multimorbidity.Lower categories of socioeconomic status (i.e., low, lower, and upper lower) were combined as the reference group, and higher categories of socioeconomic status (i.e., middle and upper) were combined to form the other group.*Unadjusted ORs and 95% CI.

| Publication bias
Publication bias was detected in the funnel plot for age but not for sex as a risk factor (see Figure 20 and Figure 21).

| DISCUSSION
We conducted a systematic review and meta-analysis on risk factors of multimorbidity among older adults in India.Broadly, sociodemographic, lifestyle, and health conditions-related factors were explored in the included studies.The pooled odds of multimorbidity were higher in people aged ≥70 years, females, single, divorced, separated, and widowed, economically dependent, not working, and smokers.
In our review, higher age was found to be associated with multimorbidity.This finding is consistent with systematic reviews of studies conducted globally. 5,14,16,37,63Aging is a universal process that is accompanied by a decline in anatomical, immunological, and cognitive functions as a result of changes at the cellular level. 64,65As individuals age, the number of chronic conditions, their severity, and associated adverse consequences like disability, become more profound and complex. 12,13In our review, female sex was found to be associated with multimorbidity, which is consistent with available global evidence. 16,63,66A possible Forest plot of the association between wealth index and multimorbidity.Lower categories of wealth index (i.e., poor, poorer, and poorest) were combined as the reference group, and higher categories of wealth index (i.e., middle and upper) were combined to form the other group.*Unadjusted ORs and 95% CI.  explanation could be inadequate access and utilization of healthy lifestyle practices and healthcare facilities for females in India 67,68 due to factors like socio-cultural issues or personal choices. 67males usually have a higher life expectancy than males 16,67 as they are more likely to suffer from nonfatal diseases. 16In this review, single, divorced, separated, and widowed people had higher odds of multimorbidity.The finding is in line with a crosssectional study conducted in Nepal, a neighboring country. 69nerally, married people tend to have better physical and mental health due to the emotional and financial support they receive from their partner. 69,70However, factors like relationship quality and length could be also important.In our review, economic F I G U R E 20 Funnel plot to assess publication bias for age as a risk factor.
dependency was found to be associated with multimorbidity.
Economic dependency can deprive individuals of a healthy lifestyle and receive high-quality healthcare, which could explain the above-mentioned association. 71Similarly, in this review, people who were not working had higher odds of multimorbidity.This is consistent with a systematic review of studies conducted in the WHO Eastern Mediterranean countries. 72Those who do not work can struggle with finances as well as physical and mental health. 72,73In our review, smoking was found to be associated with multimorbidity.This is consistent with another systematic review of studies conducted globally. 37Smoking is an unhealthy lifestyle that predisposes individuals to the development of several health conditions. 74The cumulative toxic effects of smoking have a detrimental impact on health, particularly on the respiratory and cardiovascular systems. 75In other words, these synthesized sociodemographic and lifestyle factors should be taken into consideration when developing health interventions for addressing multimorbidity among older adults in India.
To the best of our knowledge, this is the first systematic review to synthesize the existing evidence on risk factors of multimorbidity among older adults in India.A robust systematic review process was followed, and several databases were searched for published and unpublished studies without any date and language restrictions and using comprehensive search strategies.
Although a standardized critical appraisal tool was used in this review, the assessment of methodological quality is subjective to a large extent.However, the inter-rater reliability was 96%.In the absence of adjusted ORs, unadjusted ORs were used in the metaanalysis.The included studies were geographically well-distributed across different states of India.Some of the studies were conducted on a nationally representative sample with a large sample size thus, giving a largely complete picture.In terms of generalizability, the findings could be valid in similar populations, settings, and contexts.To update this systematic review in the future, more primary studies should be conducted on other potential risk factors.For example, several factors (e.g., family history of diabetes, family history of hypertension, and level of physical activity) could not be included in any meta-analysis due to being reported in single studies. 1,18,52,55,57,61In addition, none of the included studies explored genetic and environmental factors.
The included studies were all cross-sectional, and thus, there is a need to conduct longitudinal studies to explore causality.Rather than relying completely on self-reported data in primary studies, exposures and outcomes should be measured objectively.For example, screening people to identify chronic health conditions and reviewing medical records for medical and surgical history.
In conclusion, this systematic review and meta-analysis provided F I G U R E 21 Funnel plot to assess publication bias for sex as a risk factor.

F I G U R E 3
Forest plot of the association between sex and multimorbidity.Studies reporting either sex or gender were combined as 'sex'.Male was the reference group, and female was the other group.*Unadjusted ORs and 95% CI.F I G U R E 4 Forest plot of the association between social caste and multimorbidity.The general category of social caste was the reference group, and other categories (i.e., scheduled castes, scheduled tribes, and other backward classes) were combined to form the other group.*Unadjusted ORs and 95% CI.

F
I G U R E Forest plot of the association between place of residence and multimorbidity.The rural area was the reference group, and the urban area was the other group.*Unadjusted ORs and 95% CI.F I G U R E Forest plot of the association between geographical region and multimorbidity.Northern India was the reference group, and other regions in India were combined as the other group.*Unadjusted ORs and 95% CI.

F
I G U R E Forest plot of the association between smoking and multimorbidity.No smoking was the reference category, and smoking was the other group.*Unadjusted ORs and 95% CI.F I G U R E Forest plot of the association between tobacco consumption and multimorbidity.No tobacco consumption was the reference group, and tobacco consumption was the other group.*Unadjusted ORs and 95% CI.

F I G U R E 18
Forest plot of the association between alcohol consumption and multimorbidity.No alcohol consumption was the reference group, and alcohol consumption was the other group.*Unadjusted ORs and 95% CI.F I G U R E 19Forest plot of the association between BMI and multimorbidity.Lower two categories of BMI (i.e., underweight and normal) were combined as the reference group, and higher two categories of BMI (i.e., overweight and obesity) were combined to form the other group.*Unadjusted ORs and 95% CI.

a
comprehensive picture of the problem by synthesizing the existing evidence on risk factors of multimorbidity among older adults in India.These synthesized sociodemographic and lifestyle factors should be taken into consideration when developing health interventions for addressing multimorbidity among older adults in India, and more specifically, in people aged ≥70 years, females, single, divorced, separated, and widowed, economically dependent and not working, and smokers.AUTHOR CONTRIBUTIONS Nikita Goel: Conceptualization; data curation; formal analysis; investigation; methodology; project administration; resources; software; validation; visualization; writing-original draft; writing-review and editing.Isha Biswas: Data curation; formal analysis; methodology; software; validation; writing-review and editing.Kaushik Chattopadhyay: Conceptualization; investigation; methodology; resources; software; supervision; validation; visualization; writingoriginal draft; writing-review and editing.ACKNOWLEDGMENTS Ms. Ella Wharton, the librarian at the University of Nottingham, for contributing to the search strategies.The authors have no funding to report.
Methodological quality of included studies.
*≥60 years only.°Significant (S)/nonsignificant (NS) as calculated by reviewers.^Significant (S)/nonsignificant (NS) as reported by study authors.T A B L E 2