Study population
The Gothenburg H70 Birth Cohort Studies (the H70 studies) are ongoing population-based longitudinal studies of health and ageing. Full details of these studies have been reported elsewhere [12-16]. In brief, initiated in 1971, the H70 studies are a series of cohort studies of older men and women living in Gothenburg, Sweden. Seventy-year-old men and women listed in national population registers in Gothenburg were systematically selected based on their birth dates and they underwent extensive medical, social, psychiatric, and physical examinations.
For the purpose of the present study, baseline data on the 1930 birth cohort collected during 2000 was used (n = 524, response rate 70%). Forty-seven percent of non-participants was surveyed for a shorter health interview. Participants and non-participants were similar regarding self-rated health, history of myocardial infarction, diabetes or smoking status, but married men were significantly overrepresented among participating men [17].
All participants gave informed consent to participate in the study. The study complied with the Declaration of Helsinki principles, and was approved by the Regional Ethics Committee for Medical Research at the University of Gothenburg.
Assessment of loneliness
Self-perceived feeling of loneliness was assessed by a single question as ‘do you feel lonely?’ There were four alternative responses where 1 indicated never feeling lonely, 2 seldom, 3 sometimes, and 4 very often (see Table 1 for sample distribution). The four categories were then merged into a dichotomous variable as 0 = not lonely (responses 1-2), and 1 = lonely (responses 3-4).
Table 1. Number and percentages of men and women in each of the four response categories of self-perceived feelings of loneliness.
|
All
N
|
Never
(response 1)
n (%)
|
Seldom
(response 2)
n (%)
|
Sometimes
(response 3)
n (%)
|
Often
(response 4)
n (%)
|
Men
|
240
|
147 (61.3)
|
52 (21.7)
|
39 (16.3)
|
2 (0.8)
|
Women
|
272
|
104 (38.2)
|
84 (30.9)
|
70 (25.7)
|
14 (5.1)
|
All
|
512
|
251 (49.0)
|
136 (26.6)
|
109 (21.3)
|
16 (3.1)
|
Mortality
Based on unique personal identification numbers and using the Swedish national registers (the national population register and the national cause of death register), cohorts were followed for 12 years from the date of their baseline examination or until death. Cardiovascular deaths were those with International Classification of Diseases, 10th Revision (ICD-10) codes I.00-I.99.
Other covariates
Adverse socioeconomic status, health and health related behavioral factors that have previously been shown to be associated with loneliness [18-20] were included as possible confounding factors. Current perceived economic situation was assessed using a seven-point scale ranging from excellent to very bad (coded from 1 to 7). The seven-point scale was then merged into three categories: good (scale points 1-3, excellent, very good, good), Average (scale point 4), and poor (scale points 5-7, not very good, bad, very bad). Living alone was categorized as individuals who are single, or divorced, or widowed and live alone versus individuals who live with a partner (married/cohabiting/having partner but lives separate or live-apart-together or occasionally live together, Swedish term is ‘särbo’) or with someone else. Smoking status was categorized as current smoker (regular or occasional), previous smoker, and never smoker. Leisure time physical activity was defined as moderate/regular versus inactive. Alcohol consumption was measured with questions regarding weekly consumption of beer, wine, and spirits in centiliters (cl) during the past month. Based on these volumes, average weekly grams of alcohol consumption were calculated using conversion factors based on average alcohol concentration by volume (spirits 1 cl = 3 g, wine 1 cl = 1 g, beer >3.5% 1 cl= 1/3 g). Body mass index (BMI) was calculated from measured weight and height (weight in kg/height in m2). Previous history of having (yes/no) cancer, diabetes, coronary heart disease and stroke was based on self-report as well as from medical examinations conducted by a study physician. Systolic and diastolic blood pressure (SBP, DBP) were measured in the sitting position after a minimum of 5 min of rest. Blood samples were drawn from an antecubital vein and serum triglyceride measurement was determined according to standard laboratory procedures. Impaired mobility was defined based on a six-item scale of activities of daily living (ADL). The ADL scale measured self-reported difficulties in performing daily life activities including transferring, dressing, bathing, using toilet, eating, and continence. Each item was coded as 0 = no need of help from another person, and 1 = need help. A composite index was created by summing up all the six items ranging from 0 to 6 (need no help to need help in all six activities). The index was then dichotomized as 0 (no impaired mobility) and 1 (impaired mobility, scale 1-6). Based on symptoms elucidated during a psychiatric examination, major depression was diagnosed according to the DSM-5 criteria (American Psychiatric Association) [21], and minor depression according to the DSM-IV research criteria [22]. Definition of these variables has been described previously [23].
Statistical analysis
Using descriptive statistics, differences in the distribution of baseline characteristics in men and women according to their loneliness status were examined using the Pearson x²-test for categorical variables and Student’s t-test for continuous variables. Descriptive statistics are presented as percentages or mean values with standard deviations (SD). All P-values are two-sided and values of < 0.05 were considered statistically significant. The survival function for the 12-year period according to loneliness status was assessed using the Kaplan Meier method, and the log-rank test was used to evaluate group differences. Cox proportional hazard regression models were used to study the association between loneliness status at baseline and cardiovascular and all-cause- mortality during 12-year follow-up. Both unadjusted and multivariable adjusted regressions were carried out separately for cardiovascular- and all-cause mortality. Factors that were shown to be associated with loneliness were included in multivariable models. Estimates derived from Cox regressions are presented in graphical format showing hazard ratios (HR) and 95% CI. Statistical analyses were performed using SPSS, Windows version 25.0 (SPSS Inc., Chicago, IL, USA) and graphics were produced using R version 3.4.3 (The R Foundation for Statistical Computing).