Article Text
Abstract
Background Providing informal care has been linked with poor health but has not previously been studied across a whole population. We aimed to study the association between informal care provision and self-reported poor health.
Method We used data from the UK 2001 Census. The relationship between informal caregiving and poor health was modelled using logistic regression, adjusting for age, sex, marital status, ethnicity, economic activity and educational attainment.
Results We included 44 465 833 individuals free from permanent sickness or disability. 5 451 902 (12.3%) participants reported providing informal care to another person. There was an association between provision of informal caregiving and self-reported poor health; OR 1.100, 95% CI 1.096 to 1.103. This association remained after adjustment for age, sex, ethnic group, marital status, economic activity and educational attainment. The association also increased with the amount of care provided (hours per week).
Conclusions Around one in eight of the UK population reports that he or she is an informal caregiver. This activity is associated with poor health, particularly in those providing over 20 h care per week.
- Census informal caregivers logistic regression poor health
Statistics from Altmetric.com
Introduction
About one in eight of the population of the UK are caregivers.1 Providing care to another person has been implicated as a risk factor for poor health.2 However, population-based studies3–5 and systematic reviews6 assessing the impact of exposure to informal caregiving on health status either focus on a subset of the caregiving population3 ,4 or have not taken into account factors known to influence health status.5
We examined the relationship between self-reported health status and informal caregiving in the whole UK population.
Methods
We wanted to determine whether the prevalence of reported poor health: (a) was higher in those exposed to providing informal care than those who are not providing care, (b) increases with exposure to caregiving and (c) could be explained by other confounding variables.
We used data from the UK Census 2001 from respondents aged 16 and over. For the first time, the 2001 Census included a question on the number of hours care per week provided to 'family members, friends, neighbours or others because of long-term physical or mental ill health or disability, or problems related to old age'.7 The primary outcome was self-reported health status, rated as ‘good’, ‘fairly good’ or ‘not good’. We excluded people classified as 'permanently sick and disabled' (n=2 464 676, 5.3%).
The number of variables available for inclusion was limited by the Office for National Statistics to avoid issues related to confidentiality and disclosure.8 We adjusted for age, sex, marital status, ethnicity, economic activity and education based on our knowledge of factors likely to influence health status.9–11
We used univariate analyses to examine the association between reported health (and other covariates) and caregiving exposure. As the proportional odds assumption was not fulfilled when the three self-reported health categories were used, we dichotomised the reported health status responses to good/fairly good and not good (poor). We examined the association between poor health and three levels of caregiving exposure (no care was the reference category). ORs for ‘poor’ health were computed for each caregiving exposure level using binary logistic regression. Additional analyses adjusted for age, sex, marital status, ethnicity, economic activity and education. Identical categorisation of economic activity and education for individuals aged 75 and over made estimation of their individual regression coefficients unreliable. Therefore, we performed two analyses: (1) all ages not adjusting for economic activity and educational qualifications (Model 1) and (2) age groups 16–74 adjusting for all covariates (Model 2). Statistical analyses were performed using Minitab 15 Statistical Software (Minitab Inc., State College, PA) and SAS V.9.1.
Results
The UK population on census day 2001 was 58 789 194 of whom 46 930 509 (80%) were aged 16 or over and 44 465 833 (76%) were not classified as permanently sick or disabled. Within this population, 5 451 902 (12.3%) reported providing care to a family member, friend, neighbour or others and 3 725 048 (8.4%) reported their health to be poor. Among non-caregivers, 3 232 301 (8.3%) reported poor health compared with 492 747 (9.0%) of those who provide care (OR 1.100, 95% CI 1.096 to 1.103).
In univariate analyses, the odds of poor health were increased (p<0.001) as hours of caregiving rose (p<0.001): with age (p<0.001), in female subjects (p<0.001), in white ethnic groups (p<0.001), in groups classified as economically inactive (p<0.001) and as the level of educational attainment decreased (p<0.001).
The full binary logistic regression models are presented in table 1. Model 1, correcting for basic demographic characteristics, shows that the association between caregiving and poor health is attenuated but remains highly significant. Model 2 introduces economic activity and educational attainment; this again reduces the strength of the association. We found no important interaction among caregiving exposure and age, sex, marital status, ethnic group, economic activity and educational attainment. The model R-squared values, which indicate the proportion of variance explained, are 15.6% (Model 1) and 13.3% (Model 2). A sensitivity analysis, including participants classified in the economic activity category as permanently sick and disabled, showed very similar results.
Discussion
The main finding of our study is that people who report providing informal care have an increased risk of reporting poor health compared with those who do not provide informal care. This was not explained by potential confounders. We found an apparent dose–response relationship, with those providing the highest number of hours care reporting the worst health, which is consistent with results from other studies that have linked informal caregiving with adverse health outcomes.3 ,4 ,5 Although the health risk associated with providing informal care appears relatively modest, a large number of people are exposed to this risk. Therefore, these findings have important societal and public health implications.
There are a number of possible mechanisms for how informal caregiving could cause poor health. Potential ‘occupational hazards’ that caregivers may be exposed to include physical or psychosocial stressors4–6 (eg, expectations or values that exceed their knowledge, skills or capacity) that might be risk factors for physical and mental health problems.11 ,12 In addition, informal caregivers who are stressed have been reported to have impaired immunity,13 healing14 and increased risk of hypertension, heart disease and death.
One limitation of our study is that temporal associations cannot be inferred.15 In addition, we have compared currently exposed individuals with currently unexposed individuals which does not account for duration or past history of caregiving. Further, there may be a relation between health status and the occurrence of caregiving behaviour . While we were able to adjust the data to account for a number of key socio-economic and demographic factors, the Census dataset does not include all the potentially important risks for poor health (eg, smoking, blood pressure, diet, exercise). Although we adjusted for socio-economic status, with which these risks are associated, confounding may nevertheless provide a partial explanation of our study findings. However, our study has several strengths. The UK 2001 Census is unique in that it gathered data in a whole population, representing all stages of health, disease, illness and exposures of interest. General health and caregiving were recorded using validated and explicit measures. A simple, one item self-reported health question similar to that used in this UK 2001 Census has been associated with mortality.13 The large size of our study allows precise estimates of effect.
One of the key challenges facing countries with a burgeoning ageing population and a concomitant rise in prevalence of disability and dependence is meeting the needs of this population without placing excessive demands on informal carers. This dilemma has direct implications for health and social service systems. Currently, the national General Medical Services GP contract awards points for establishing of a system to identify carers and liaise with outside agencies; however, creation of such a system is voluntary. Formal surveillance systems offer the opportunity to monitor trends in the rate of occurrence of ill health. These can be used to anticipate needs, inform plans for health or social care and guide preventive and therapeutic strategies for carers. Moreover, information from surveillance could guide future research priorities. However, informal care is a complex and dynamic chronic exposure; use of a clear, unambiguous definition of informal care and an appropriate system for measuring (and reviewing) the unpaid care exposure is fundamental to the validity of any surveillance system.
Conclusions
We found an independent association between provision of informal caregiving and self-reported poor health. This association increased with exposure and remained after adjustment for age, sex, ethnic group, marital status, economic activity and education. These results suggest that there is a substantial burden of ill health associated with the common task of caregiving. This carries important public health implications, with a need to develop systems for identifying people at risk and effective interventions to prevent ill health.
What is already known on this subject
Informal caregiving has been implicated as a risk factor for poor health but previous studies have had a limited focus and often did not account for other factors known to influence health status.
What this study adds
We found that providing informal care for 20+ hours per week was associated with a substantial increase in self-reported poor health. This association remained after adjusting for age, sex, marital status, education, ethnic group and economic activity.
Footnotes
PSI Licence: C200900034.
Funding This work was supported by Chief Scientist Office Scotland.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.