Prevalence and correlates of loneliness in the later life—insights from Longitudinal Ageing Study in India (LASI) Wave-1

ABSTRACT

Context:

Loneliness is a psychosocial problem that can affect anyone at any age in any circumstances, but it has become a serious issue of concern among the elderly.

Aim:

The present study aims to examine the overall and state and union-territories-wise prevalence of loneliness among India’s elderly population and exploring the correlates of loneliness in old age.

Method:

The study is based on secondary data from Longitudinal Ageing Study in India (LASI) Wave-1 (2017-2018) data collected from a total of 30,394 older adults aged 60 and above excluding missing responses. Frequent loneliness was measured using a single question asking how often you feel alone.

Statistical Analysis Used:

Descriptive statistics, the Chi-square test, and bivariate logistic regression were used to meet the objectives. The statistical significance was set at <0.05.

Results:

13.4% of elderly reported frequent loneliness. The prevalence of loneliness among elderly was found to be maximum in Andaman and Nicobar Island and minimum in Tripura. Age, current marital status, region, religion, MPCE quintile, number of chronic conditions, number of Activities of Daily Living (ADL), and Instrumental Activities of Daily Living (IADL) difficulties, involvement in physical activity, involvement in yoga/meditation, number of social networks, participation in social activities, and involvement in leisure activities like reading books, watching TV, or using computer were found to be significantly associated with loneliness.

Conclusion:

Not being in a marital union, having more than two chronic conditions, more than two ADL and IADL difficulties, being physically inactive, non-involvement in Yoga/Meditation, and small social network are associated with higher loneliness.

INTRODUCTION

Loneliness and social isolation have become the most significant and serious issues among the elderly. The two are often talked about interchangeably, but there are important distinctions between the two. Social isolation is an objective state—measured on the basis of the quantity of social relationships and social contacts, while loneliness is a subjective experience. Loneliness is a negative and unpleasant feeling related to the deficient social relations, a perceived gap between the quality and quantity of relationships that we have and those we want. Loneliness not necessarily means being alone; instead, it is the perception of being alone and isolated that affects them.

The population of the elderly is growing fast, both in terms of proportion and absolute numbers. As the population of elderly has grown, the number of individuals experiencing loneliness often or all of the time has also increased. Changing living situations (living alone or living with relations and non-relations) is the main demographic cause of loneliness. Studies show that loneliness is found to be more among those who live alone.[1] Some other factors like widowhood, increasing dependency ratio as well as economic dependency, type of family, social network, place of residence, and population migrations are some of the significant correlates of loneliness. Women are found to be more at risk for loneliness and isolation than men.[1,2]

Over recent years, there has been growing public attention to loneliness among elderly and this has been accompanied by a shift in our understanding of its impact and in particular its implications on mental and physical health.

It is generally accepted that consistent, overwhelming, and pervasive loneliness develops stress and ultimately culminates into serious physical morbidities. Loneliness can cause depression in old age. It is known to be more dangerous than smoking. The effect of loneliness and isolation can be as harmful to health as smoking 15 cigarettes a day and is more damaging than obesity.[3] High degree of loneliness precipitates suicidal ideation and parasuicide. Lonely people are at higher risk of the inception of disability. Loneliness puts individuals at higher risk of cognitive decline. A study found that lonely people have 64 per cent more chance of developing clinical dementia.[4] It adversely affects the immune and cardiovascular systems. Loneliness affects the physical and mental health. It is generally an accepted notion that loneliness causes decline of well-being of an individual and has an adverse effect on physical health, possibly through immunological impairment or neuro-endocrine changes. Loneliness is thus among the hidden causes of hospitalization and of placement in nursing homes. It is also responsible for sleep problems, disturbed appetite, and so on. Loneliness increases risk for developing biological dysfunctions, psychological distress, behavioral problems, and personality and adaptation disorders, such as overconsumption of alcohol, loss of self-esteem, extreme forms of anxiety, powerlessness, and stress. Loneliness enhances the likelihood of mortality and functional decline.[5]

The devastating impacts which loneliness puts on our mental and physical health make it an issue we can ill-afford to ignore. But loneliness is also a deeply personal experience—unique to every individual, a problem with different causes and different consequences for each and every one of us. This makes addressing loneliness a challenge. Most of us have experienced or will experience loneliness at some point in our lives, but for many, it is temporary. Sadly though, for a large number of older people, loneliness defines and devastates their lives. Everyone understands that there is need for action, but it is less clear how we can most effectively respond to such a personal problem; there is no “one-size-fits-all’ approach that will prevent everyone from feeling lonely.[5]

Mother Teresa once said: “the greatest disease … is not TB or leprosy; it is being unwanted, unloved, and uncared for. We can cure physical diseases with medicine, but the only cure for loneliness, despair, and hopelessness is love.”[6,7]

Loneliness within a few decades had evolved from a personal misfortune into a social epidemic.[3] It has now become an important public health concern worldwide. Loneliness, therefore, needs to be handled as a disease; not just as a situation or a symptom of a disease or merely a social concept and should find its place in the classification of psychiatric disorders. It is the major problem associated with old age and, therefore, needs proper and holistic strategy for its diagnosis, care, and management.

Hence, we carried out the study with the objective of finding the prevalence of loneliness in the Indian elderly and assessing the correlates of loneliness in old age.

SUBJECTS AND METHODS

The study is based on data from the Longitudinal Ageing Study in India (LASI) Wave-1, which is a full-scale national survey of aging population of India, consisting of a representative sample of 72,250 individuals aged 45 years and above and their spouses across all States and Union Territories of India except Sikkim, conducted between 2017 and 2018. It was a cross-sectional study, which adopted the three-stage sampling design in rural areas and the four-stage design in urban areas.

The first stage of the sampling process involved the selection of Primary Sampling Units (PSUs) or subdistricts (Tehsils/Talukas) in each state. In the second stage, Secondary Sampling Units (SSUs) or villages from rural areas and wards from urban areas were selected from the chosen PSUs. For rural areas, the third stage involved the selection of households from the selected villages, while for urban areas, one more stage was added. In the third stage of urban sampling, one Census Enumeration Block (CEB) was randomly selected from each chosen urban ward, and households from this CEB were selected in the fourth stage. The adoption of a four-stage sample design in urban areas was due to the large size of urban wards, making it difficult to list all households.

Trained field investigators interviewed the sampled households and individuals through the interview schedule.

Informed written consent was obtained from household and individual interview respondents in their preferred language, with the LASI instrument professionally translated into 16 regional languages to improve response rates and data quality. For illiterate respondents, the consent form was read aloud, and they were asked to provide a signature or inked fingerprint. A signed copy of the consent form was provided to the respondents.

Proxy interviews were conducted for participants who were unable to give the full interview, usually due to cognitive limitations.

Ethics approvals were obtained from various collaborating organizations, including the Indian Council of Medical Research (ICMR), Delhi; IRB, International Institute for Population Sciences (IIPS), Mumbai; IRB, Harvard T.H. Chan School of Public Health (HSPH), Boston; IRB, University of Southern California (USC), Los Angeles; IRB, ICMR-National AIDS Research Institute (NARI), Pune; and IRB, Regional Geriatric Centres (RGCs), MoHFW.[8]

The LASI Wave-1 dataset was used after filtering for respondents with age 60 years and above. Some new variables were computed by transforming and recoding existing variables. Respondents whose responses were missing were excluded. The effective sample size was therefore 30,394 older adults aged 60 years and above.

Outcome variable

The LASI questionnaire does not contain all the three questions of the well recognized three-item UCLA loneliness scale. Hence, the single-item question: “How often you feel alone?” from the 10-item Centre for Epidemiological Studies Depression Scale (CES-D-10) was used as an outcome variable, which is the self-reported loneliness. The response of “rarely” and “sometimes” were coded as 0 which indicated “not lonely” and response of “often” or “most/all of the time” was coded as 1 which indicated “frequent loneliness.”

Independent variables

The independent variables included in the study that are studied for association with loneliness were chosen after studying the existing literature on loneliness and on the basis of the availability of suitable questions in the LASI questionnaire to assess that variable. A list of the variables and their corresponding categories can be found in supplementary material I.

Statistical analysis

The data analysis was performed using SPSS version 20.0. Descriptive statistics (frequency distribution, and prevalence) were calculated. The Chi-square test was used to test associations between frequently reported loneliness and respondent’s socio-economic, demographic, and health-related characteristics for identifying the predictors of loneliness. Binary logistic regression was performed with a 95% confidence interval to determine significant associations between frequently reported loneliness and each of predictors obtained from the Chi-square test after adjusting the effect of all other predictor variables.

RESULTS

Table 1 represents the socio-economic, demographic, and health-related characteristics of elderly study subjects. Among 30,394 study subjects included in the study, 14558 (47.9%) were males and 15836 (52.1%) were females. More than half of the study subjects belonged to the age group of 60-69 years (61%), had no or only primary level of education (84.8%), were married (63.7%), were from Southern India (24%), were Hindu (73.3%), belonged to rural region (66%), and belonged to poorer monthly per-capita expenditure quintile (20.6%). A greater percentage of elderly reported that they were currently not working (70%). Nearly 12% of elderly reported no chronic condition, whereas 17% of elderly were suffering from single chronic disease. Moreover, nearly half of the elderly (52.2%) had more than two chronic conditions. Only 7.1% of elderly had more than 2 ADL difficulties; however, majority of them reported no ADL difficulty. The percentage is quite higher for IADL difficulties wherein 24.5% of elderly reported having more than 2 IADL difficulties. More than half of the respondents (60%) were engaged in either of the moderate or vigorous physical activity or both. Very few elderly (only 13%) reported to be involved in activities like yoga, meditation, asana, pranayama, etc., Most of the elderly had only one person with whom they share most of their personal matters. Majority of elderly (94.2%) did not participate in any of the social activities. However, nearly two-third respondents reported that they are involved in reading books, watching TV, or using computer (65%).

Table 1.

Socio-economic, demographic, and health-related characteristics of the elderly

Socio-demographic variables		Count	Percent
Age	60-69	18534	61.0
	70-79	8763	28.8
	≥80	3097	10.2
Sex of Respondent	Male	14558	47.9
	Female	15836	52.1
Education	No education/Primary	25763	84.8
	Secondary	3222	10.6
	Higher	1409	4.6
Current Marital Status	Not in marital union	11023	36.3
	Currently in marital union	19371	63.7
Region	North	5659	18.6
	Central	4079	13.4
	East	5600	18.4
	North-East	3618	11.9
	West	4139	13.6
	South	7299	24.0
Religion	Hindu	22283	73.3
	Muslim	3586	11.8
	Other	4525	14.9
Place of residence	Rural	20064	66.0
	Urban	10330	34.0
MPCE quintile	Poorest	6229	20.5
	Poorer	6252	20.6
	Middle	6213	20.4
	Richer	5979	19.7
	Richest	5721	18.8
Currently working	No	21279	70.0
	Yes	9115	30.0
Number of chronic diseases	No chronic condition	3520	11.6
	One no chronic condition	5277	17.4
	Two chronic conditions	5745	18.9
	More than two chronic conditions	15852	52.2
ADL category	0	24180	79.6
	1	2651	8.7
	2	1412	4.6
	>2	2151	7.1
IADL category	0	17189	56.6
	1	3362	11.1
	2	2391	7.9
	>2	7452	24.5
Physical activity	No	12327	40.6
	Yes	18067	59.4
Involvement in activities like yoga, meditation, asana, pranayama, etc.	No	26534	87.3
	Yes	3860	12.7
Number of social network (with whom they share most of their personal matters)	0	2853	9.4
	1	19529	64.3
	2	5919	19.5
	>2	2093	6.9
Number of social activities participated	0	28629	94.2
	1	562	1.8
	2	586	1.9
	>2	617	2.0
Involvement in leisure activities like reading books, watching TV, or using computer	No	10572	34.8
	Yes	19822	65.2

Abbreviations: ADL—Activities of Daily Living, IADL—Instrumental Activities of Daily Living, MPCE—Monthly per-capita expenditure

Prevalence of loneliness

Out of 30,394 elderly included in the study, 4076 elderly (i.e. 13.4%) reported frequent feeling of loneliness.

Figure 1 shows State and Union Territory-wise prevalence of loneliness among elderly in India. Among states, Jammu and Kashmir had highest prevalence of loneliness among elderly, while Tripura had lowest prevalence of loneliness among elderly in India.

Figure 1.

State and Union Territory-Wise Prevalence of Loneliness among Elderly in India

Among Union Territories, Andaman and Nicobar Islands had highest prevalence of loneliness among elderly, while Daman and Diu had a lowest prevalence of loneliness among elderly in India.

Table 2 depicts the distribution of elderly population characteristics with the prevalence of loneliness. The prevalence of loneliness was found to be increasing with increasing age. Elderly women reported more loneliness than men. Elderly having lower education reported more loneliness. Similarly, elderly who were not in marital union, belongs to Hindu religion, from a rural area, not employed, had more chronic conditions, facing more ADL and IADL difficulties, were physically inactive, were not doing yoga/meditation, and had few social networks reported more loneliness. A U-shaped relationship was found between loneliness and MPCE quintile. Poorest and richest elderly showed higher loneliness, while elderly having middle MPCE quintile showed relatively lower loneliness. Elderly who were involved in activities like reading books, watching TV, and using computer reported less loneliness than those who were not. Number of social activities participated also showed U-shaped trend. The elderly who did not participated in any social activities or those who participate in more than two social activities showed higher loneliness than those who participated in only one or two social activities.

Table 2.

Distribution of elderly population according to prevalence of loneliness and their Chi-square estimates

		Prevalence of loneliness among elderly (%)	Chi-square test
			Degrees of freedom	Chi-square	P
Age	60-69	12.2	2	71.81	0.001*
	70-79	14.5
	≥80	17.4
Sex of Respondent	Male	11.6	1	81.73	0.001*
	Female	15.1
Education	No education/Primary	14	2	50.15	0.001*
	Secondary	10.5
	Higher	9.5
Current Marital Status	Not in marital union	19.3	1	523.91	0.001*
	Currently in marital union	10
Region	North	13.7	5	333.20	0.001*
	Central	17.4
	East	11.9
	North-East	7.8
	West	9
	South	17.4
Religion	Hindu	14.1	2	38.68	0.001*
	Muslim	12.2
	Other	10.9
Place of residence	Rural	13.9	1	10.76	0.001*
	Urban	12.5
MPCE quintile	Poorest	14.4	4	15.93	0.003*
	Poorer	13.5
	Middle	12.3
	Richer	12.9
	Richest	14
Currently working	No	14.4	1	58.03	0.001*
	Yes	11.1
Number of chronic diseases	No chronic condition	10.7	3	52.54	0.001*
	One no chronic condition	12.5
	Two chronic conditions	12.4
	More than two chronic conditions	14.7
ADL category	0	12	3	340.20	<0.001*
	1	14.7
	2	17.3
	>2	25.6
IADL category	0	10.9	3	328.56	0.001*
	1	12.4
	2	14.6
	>2	19.4
Physical activity	No	16	1	121.77	0.001*
	Yes	11.6
Involvement in activities like yoga, meditation, asana, pranayama, etc.	No	13.9	1	45.65	0.001*
	Yes	9.9
Number of social networks	0	23	3	270.81	0.001*
	1	12.9
	2	11.7
	>2	9.7
Number of social activities participated	0	13.5	3	8.18	0.043*
	1	11.4
	2	10.1
	>2	12.8
Reading books, watching TV, or using computer	No	16.7	1	149.73	0.001*
	Yes	11.7

Abbreviations: ADL—Activities of Daily Living, IADL—Instrumental Activities of Daily Living, MPCE—Monthly per-capita expenditure. “*” refers significant at 0.05 level of significance

The Chi-square test result showed significant association between loneliness and all of the independent variables (i.e. age, sex, education, marital status, region, religion, place of residence, MPCE quintile, current working status, number of chronic conditions, number of ADL and IADL difficulties, involvement in physical activity, involvement in yoga/meditation, number of social network, participation in social activities, and involvement in activities like reading books, watching TV, or using computer) [refer to Table 2]. Hence, all were taken as predictors for logistic regression analysis.

The logistic regression analysis found that out of all predictors identified by the Chi-square test only age, current marital status, region, religion, MPCE quintile, number of chronic conditions, number of ADL and IADL difficulties, involvement in physical activity, involvement in yoga/meditation, number of social networks, participation in social activities, and involvement in leisure activities like reading books, watching TV, or using computer were significant predictors of loneliness. The goodness-of-fit statistic, i.e. Nagelkerke R-square, for the model was 0.083. Supplementary material II contains the ROC curve for different predictors.

The odds of loneliness were 1.161 times more in the age group of 60-69 as compared to age group of 80 and above, and 1.86 times more in elderly who are currently not in marital union as compared to those who are currently in a marital union. However, as compared to elderly living in southern India, the elderly from northern, eastern, north-eastern, and western India had lower odds of loneliness. Elderly belonging to Muslim religion had lower odds of loneliness than other religion (excluding Hindus). The elderly respondents belonging to poorest, poorer, middle, and richer MPCE quintile had lower odds of loneliness than elderly belonging to richest MPCE quintile. The elderly having no, one, or two chronic conditions had lower odd of loneliness than elderly with more than two chronic conditions. The elderly adults who have no, one, or two ADL difficulties had lower odds of loneliness than elderly having more than two ADL difficulties. Similarly, elderly having no or one IADL difficulty had lower odds of loneliness than elderly having more than two IADL difficulties. The elderly respondents who were not physically active had 1.173 times more odds of loneliness than those who were physically active and elderly who were not doing Yoga/meditation/Pranayama had 1.148 times more odds of loneliness than those who reported doing Yoga/meditation/Pranayama. The elderly who had no social networks had twice more odds of loneliness than elderly having more than two social networks and elderly who have only one social network had 1.343 times twice more odds of loneliness than elderly having more than two social networks. Similarly, elderly who did not participated in any or only involved in any one of the social activities had lower odds of loneliness than those who participated in more than two social activities. Elderly who reported not doing any of the activities like reading books, watching TV, or using computer had 1.331 times more odd of loneliness than those involved in these activities. [Refer to Table 3]

Table 3.

Bivariate Logistic Regression Estimate for Loneliness by Socio-Economic and Demographic Variables among Elderly in India

		Adjusted Odds Ratio	P	95% C.I. for Odds Ratio
				Lower	Upper
Age	60-69	1.161	0.012*	1.033	1.305
	70-79	1.107	0.088	0.985	1.245
	≥80	Ref	Ref	Ref	Ref
Sex of Respondent	Male	1.000	0.995	0.923	1.083
	Female	Ref	Ref	Ref	Ref
Education	No education/Primary	1.164	0.131	0.956	1.416
	Secondary	1.079	0.490	0.870	1.339
	Higher	Ref	Ref	Ref	Ref
Current Marital Status	Not in marital union	1.859	0.000*	1.720	2.009
	Currently in marital union	Ref	Ref	Ref	Ref
Region	North	0.811	0.000*	0.732	0.899
	Central	0.996	0.937	0.892	1.112
	East	0.657	0.000*	0.589	0.732
	North-East	0.447	0.000*	0.385	0.520
	West	0.475	0.000*	0.418	0.539
	South	Ref	Ref	Ref	Ref
Religion	Hindu	1.066	0.279	0.950	1.196
	Muslim	0.819	0.009*	0.705	0.951
	Other	Ref	Ref	Ref	Ref
Place of residence	Rural	1.043	0.307	0.962	1.132
	Urban	Ref	Ref	Ref	Ref
MPCE quintile	Poorest	0.877	0.021*	0.784	0.980
	Poorer	0.877	0.019*	0.785	0.979
	Middle	0.808	0.000*	0.723	0.902
	Richer	0.872	0.015*	0.781	0.974
	Richest
Currently working	No	0.978	0.623	0.896	1.068
	Yes	Ref	Ref	Ref	Ref
Number of chronic diseases	No chronic condition	0.755	0.000*	0.668	0.855
	One no chronic condition	0.883	0.013*	0.801	0.974
	Two chronic conditions	0.868	0.003*	0.791	0.953
	More than two chronic conditions	Ref	Ref	Ref	Ref
ADL category	0	0.554	0.000*	0.489	0.627
	1	0.638	0.000*	0.548	0.744
	2	0.712	0.000*	0.598	0.848
	>2
IADL category	0	0.782	0.000*	0.711	0.859
	1	0.814	0.001*	0.717	0.923
	2	0.904	0.140	0.791	1.034
	>2	Ref	Ref	Ref	Ref
Physical activity	No	1.173	0.000*	1.090	1.263
	Yes	Ref	Ref	Ref	Ref
Involvement in activities like yoga, meditation, asana, pranayama etc.	No	1.148	0.023*	1.020	1.293
	Yes	Ref	Ref	Ref	Ref
Number of social networks	0	2.195	0.000*	1.844	2.613
	1	1.343	0.000*	1.152	1.567
	2	1.158	0.087	0.979	1.370
	>2	Ref	Ref	Ref	Ref
Number of social activities participated	0	0.635	0.000*	0.495	0.816
	1	0.679	0.035*	0.473	0.974
	2	0.713	0.071	0.494	1.029
	>2	Ref	Ref	Ref	Ref
Reading books, watching TV, or using computer	No	1.331	0.000*	1.231	1.438
	Yes	Ref	Ref	Ref	Ref

Last was reference category. Abbreviations: ADL—Activities of Daily Living, IADL—Instrumental Activities of Daily Living, MPCE—Monthly per-capita expenditure, CI—Confidence interval, Ref—Reference category. “*”refers significant at 0.05 level of significance

DISCUSSION

The study was carried out with aim of finding out the prevalence and correlates of loneliness in later life. The prevalence of loneliness among elderly in India was found to be 13.4%, which is relatively lower than that reported in the study conducted in Bengaluru City, Karnataka, India,[1] and from another study which was based on WHO SAGE data.[9] Similar to the study,[1] our study also found that higher age, being illiterate or having lower education, either spouse separated or dead and the presence of difficulties in daily living competencies are risk factors for loneliness.

The results revealed that age, marital status, region, religion, MPCE quintile, number of chronic conditions, number of ADL and IADL difficulties, involvement in physical activity, involvement in yoga/meditation, number of social networks, participation in social activities, and involvement in leisure activities like reading books, watching TV, or using computer were significant predictors of loneliness.

The findings showed that loneliness was more prevalent among older adults in the age group of 60-69 years and those who were currently not in a marital union. However, older adults living in southern India had higher odds of loneliness than those living in other regions. Elderly belonging to Muslim religion had lower odds of loneliness than other religion (excluding Hindus). Moreover, those belonging to the richest MPCE quintile had higher odds of loneliness than those belonging to the poorest, poorer, middle, and richer MPCE quintile.

The study also found that older adults with more chronic conditions, ADL and IADL difficulties, and those who were not physically active and not involved in yoga/meditation had higher odds of loneliness. Similarly, those with fewer social networks and less participation in social activities had higher odds of loneliness. Furthermore, older adults who did not engage in leisure activities such as reading books, watching TV, or using a computer were more likely to experience loneliness.

In line with previous research conducted in a rural clinic attached to a tertiary care center in North India[10] and another study,[11] our research found that older adults who live with non-communicable diseases/multi-morbidity face a greater risk of experiencing loneliness. Additionally, our study revealed a significant association between physical inactivity and loneliness, which aligns with the findings of a study[12] that reported higher loneliness rates in Indian older adults due to their sedentary lifestyle. However, it is important to note that non-communicable diseases can also result from physical inactivity, indicating that there may be some confounding factors that remain unexplored. Furthermore, our findings contrast with the results of a study conducted in Yazd, Iran,[13] which reported an association between social participation and loneliness. In contrast, our study found that those who do not participate in any activities or engage in only one activity have lower odds of experiencing loneliness.

However, the study suffered several limitations. Firstly, to measure the outcome variable, i.e. loneliness, no well-recognized standard scale could be used as the LASI questionnaire does not have such questions. Hence, single-item question was used to measure loneliness.

Secondly, loneliness is known to be affected by not only quantity of social networks but also by the quality of social networks. However, our study only takes into account the quantity and not quality due to the absence of relevant quality of social network-specific questions.

Thirdly, personality characteristics such as shyness, introversion, neuroticism, high expectations, and demand can also influence subjective feeling of loneliness, which is also not taken into account in our study due to the use of secondary data that does not ask such questions to measure these traits. Lastly, as our study was cross-sectional, hence, we are not able to make any kind of causal relationship between covariate and outcome variables.

Although our study has some limitations, it offers valuable insights into the factors associated with loneliness among the elderly. We conducted an extensive review of the existing literature on the topic and included a diverse range of variables to examine their association with loneliness among the elderly population throughout India. These findings can be utilized to develop targeted interventions that address this issue.

CONCLUSION

The present study’s findings indicate that loneliness is becoming a prevalent issue among the Indian elderly. Thus, there is a great need of identifying the determinants of loneliness among the elderly in India. This is necessary for the proper management of loneliness so as to ensure healthy aging. Future research is also needed with different study designs to have an in-depth insight into the problem and identify potential risk factors from the correlates of loneliness. In addition, possible steps should be undertaken for the management of loneliness.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.

SUPPLEMENTARY MATERIAL I

The independent variables in the study: The independent variables studied for association with loneliness were chosen on the basis of availability of suitable questions in LASI questionnaire. Following are the variables and its coding procedure:

1. Age recoded as “60-69”, “70-79” and “80 and above”,

2. Gender categorized as “Male” and “Female”,

3. Education recoded as “No education/Primary”, “Secondary” and “Higher”,

4. Marital status recoded as currently in “marital union” and “not in marital union”,

5. Religion recoded as “Hindu”, “Muslim” and “Others”

6. Place of residence categorized as “Rural” and “Urban”

7. Region was recoded as “North”, “Central”, “East”, “North-East”, “West” and “South”. North included states of Chandigarh, Delhi, Haryana, Himachal Pradesh, Jammu & Kashmir, Punjab, Rajasthan, and Uttarakhand. Central included states of Chhattisgarh, Madhya Pradesh, and Uttar Pradesh. East included states of Bihar, Jharkhand, Odisha, and West Bengal. Northeast included states of Arunachal Pradesh, Assam, Manipur, Meghalaya, Mizoram, Nagaland, and Tripura. West included states of Dadra & Nagar Haveli, Daman & Diu, Goa, Gujarat, and Maharashtra. South included states of Andaman & Nicobar Islands, Andhra Pradesh, Karnataka, Kerala, Lakshadweep, Puducherry, Tamil Nadu, and Telangana.[1]

8. Monthly Per-capita expenditure (MPCE) quintile categorized as “Poorer”, “Poorest”, “Middle”, “Richer”, and “Richest”. [1]

9. Currently working or not recoded as 0 (No and missing) and 1 (Yes).

10. Number of chronic conditions categorized as 0 (no chronic condition), 1 (single chronic condition), 2 (two chronic conditions) and 2+ (more than two chronic conditions. The variable was obtained from the question “Have you ever diagnosed with the particular disease?” The diseases were – Hypertension, diabetes, Cancer, Lung disease, Heart disease, Stroke, Bone/Joint disease, Neurological or psychiatric problem, High cholesterol, Thyroid, gastrointestinal, skin diseases, Urogenital disease, Visual impairment and eye-related problems, Hearing problems and Oral health. Responses were self-reported and measured in Yes (coded as 1) and No (recoded as 0).

11. Activities of Daily Living (ADL) measuring everyday competence (ability to perform normal daily activities). To measure ADL limitation, respondents were asked if they were having any of the following limitations that last for more than three months like difficulty with dressing, walking across the room, bathing, eating, getting in or out of bed, or using the toilet (including getting up and down). It was then recoded as 0, 1, 2 and more than 2 according to number of difficulties they reported. [1]

12. Instrumental Activities of Daily Living (IADL) measuring competences that are not necessarily related to the fundamental functioning of a person, but allows an individual to live independently in a community. They are necessary for independent functioning of the elderly in the community. They were assessed by asking respondents if they were having any difficulties that last more than three months, such as preparing a hot meal, shopping for groceries, making a telephone call, taking medications, doing work around the house or garden, managing money (such as paying bills and keeping track of expenses), and getting around or finding an address in unfamiliar places. It was then recoded as 0, 1, 2 and more than 2 according to number of difficulties they reported. [1]

13. Physical Activity: Those who are either engaged in moderate physical activity or vigorous physical activity or an equivalent combination of moderate- and vigorous-intensity activity were taken as physically active while those who are not engaged in any type of moderate or vigorous physical activity for a given time throughout the week were taken as physically inactive. [1]

    For moderate physical activity, respondents were asked about their engagement in moderately energetic activities such as cleaning house, washing clothes, fetching water, drawing water from a well, gardening, walking at a moderate pace, bicycling at a regular pace, and floor or stretching exercises. For vigorous activity, respondents were asked about their involvement in running or jogging, swimming, going to a health centre/gym, cycling, digging with a spade or shovel, heavy lifting, chopping, farm work, fast bicycling, and cycling with loads. The responses “Every day” and “More than once a week” were recoded as 1 and rest were recoded as 0.

14. Yoga Practice, Meditation, Asana, Pranayama: It was assessed by asking participants ‘have you engaged in activities like yoga, meditation, asana, pranayama, and similar activities’, frequency of activities; the responses “Every day” and “More than once a week” were recoded as 1 and rest were recoded as 0.

15. Number of social network assessed by asking the respondents - whom they share most of their personal matters with from a list that included their spouse, children or grandchildren, son-in-law or daughter-in-law, siblings, friends, and others.

16. Number of social activities participated was assessed by asking respondents if they “are a member of any social organization or group where they can interact with other individuals, groups, or institutions”, with “Yes” coded as 1 and “No” coded as 0 and how frequently they “eat out of the house (Restaurant/Hotel)”, “Go to park/beach for relaxing/entertainment”, “Play cards or indoor games”, “Play out door games/sports/exercise/jog/yoga”, “Visit relatives /friends”, “Attend cultural performances /shows/Cinema”, “Attend religious functions /events such as bhajan/satsang/prayer”, “Attend political/community/organization group meetings” with responses “Rarely” and “Never” coded as 0 and rest coded as 1. New variable was formed by taking total of these responses and it was then recoded as 0, 1, 2 and more than 2 according to number of social activities they participated.

17. Participation in diverse leisure activities like reading books, watching TV or using computer [2]: It was assessed by asking respondents how frequently they “Read books/newspapers/magazines”, “Watch television/listen radio”, “Use a computer for e-mail/net surfing”. “Rarely” and “Never” coded as 0 and rest coded as 1. New variable was formed by taking total of these responses and then recoding it as 1(if sum is at least 1) otherwise 0.

SUPPLEMENTARY MATERIAL II

Supplementary Material II.

ROC Curve and associated summary statistics

Area Under the Curve (AUC)
Test result variable (s)	Area
Age	0.534*
Sex of Respondent	0.538*
Education	0.479
Current Marital Status	0.407
Region	0.506*
Religion	0.477
MPCE quintile	0.495
Currently working	0.471
Number of Chronic disease	0.531*
ADL category	0.554*
IADL category	0.575*
Physical activity	0.454
Involvement in activities like yoga, meditation, asana, pranayama etc.	0.481
Number of social network	0.447
Number of social activities participated	0.495
Reading books, watching TV or using computer	0.451

*denotes better classifier than random classification. Any classifier with an AUC greater than 0.5 (AUC of 0.5 indicating a random classification) is considered to perform better than random. AUC value below 0.5 indicates that the classifier is performing worse than random

Note: The steeper the ROC curve, the better the performance of the classifier, as it indicates a higher true positive rate (TPR) for a lower false positive rate (FPR).

The position of the ROC curve relative to the reference line indicates the overall performance of the classifier. If the curve is closer to the upper left corner, the classifier performs better, and if it is closer to the lower right corner, the classifier performs worse.