Stroke Survivors Experience Elevated Levels of Loneliness: A Multi-Year Analysis of the National Survey for Wales

Abstract

Objective: Despite clinical observation that stroke survivors frequently experience loneliness, there is no large-scale empirical evidence to support this observation. Therefore, the primary objective of this research was to provide the first large-scale and comprehensive estimate of loneliness in the stroke survivor population.

Method: To address this issue, we completed two preregistered analyses of a nationally representative annual survey (N > 21,000). A two-phase approach was adopted combining both exploratory (Study 1) and confirmatory (Study 2) phases. The benefit of such an approach is that replication is built into the design, which considerably strengthens the inferences that can be made.

Results: Across two separate cohorts, the results consistently showed that human stroke survivors report higher levels of loneliness compared with healthy individuals, and this relationship could not be accounted for by demographic factors (e.g., age, sex) or objective measures of social isolation (e.g., marital status, number of household members).

Conclusions: These findings demonstrate that elevated levels of loneliness poststroke are robust in that they replicate in large nationally representative samples and cannot be reduced to objective measures of social isolation. The work has clinical and societal relevance by suggesting that loneliness poststroke is unlikely to be adequately "treated" if only the quantity and not the quality of social experiences are considered.

Keywords: Acquired brain injury; Loneliness; National Survey for Wales; Rehabilitation; Stroke.

Fig. 1

Histograms (A) and violin plots (B) of loneliness scores in Study 1. (A) Histogram representing the proportion of reported loneliness score for each health condition. (B) Violin plot representing the mean and 95% CI for each health control group in addition to the distribution of scores. Healthy = no longstanding illness; ill = any longstanding illness.

Fig. 2

Forest plots depicting mean loneliness scores and 95% CI for each health condition. (A) Results for Study 1 (2016/17), (B) Results for Study 2 (2017/18). Color separates broadly defined brain (orange) and non-brain conditions (green). Dashed black lines demonstrate average loneliness score for healthy (left) and any longstanding illness group (right).

Fig. 3

Parameter estimates (odds ratios) across studies 1 (16_17) and 2 (17_18). Odds ratios were derived using logistic regression with self-reported loneliness as the dependent variable. The loneliness scale was scored from 0 to 6, which was recoded as zero if less than three and one otherwise. The reference category for gender was male. Age was included as a continuous variable. Mar_stat = marital status, which included three levels: single (reference category), married, and ended. There were five levels of deprivation, with the most deprived group included as the reference category (Q = quintiles). People refers to the number of people within the household and was included as a continuous variable. Internet yes refers to those individuals who reported having an internet connection (reference category = no connection). Pop_density refers to the population density of the household’s location with the sparsest density (<2,000 people) as the reference category. Health refers to health status/condition with the reference category being healthy individuals who report no longstanding illness. Point estimates (dots) represent the mean estimate odds ratios and whiskers represent 95% confidence intervals. For more information see Tables 4 and 5.

Fig. 4

Histograms (A) and violin plots (B) of loneliness scores in Study 2. (A) Histogram representing the proportion of reported loneliness score for each health condition. (B) Violin plot representing the mean and 95% CI for each health control group in addition to the distribution of scores. Healthy = no longstanding illness, Ill = any longstanding illness.

Fig. 5

Meta-analytical results. (A) Mean difference in loneliness scores between those with a stroke and those with no longstanding illnesses. (B) Mean difference in loneliness scores between those with a stroke and those with arthritis. (C) Mean difference in loneliness scores between those with a stroke and those with any longstanding illness. Orange indicates the random effects model estimates for mean difference scores across both studies (Studies 1 and 2). Healthy = no longstanding illness, Ill = any longstanding illness.