Multimorbidity and the household context: measurement and associations with health and social care use

Abstract

BACKGROUND: Multimorbidity (multiple long-term conditions) presents significant challenges to health and social care systems worldwide, especially as populations age and more older adults live alone. The impact of multimorbidity in households on unplanned hospitalisation and transitioning to live in a care home is an under-explored area of study that might present opportunities for developing interventions.

METHODS: First, I explored the implications of making different methodological choices when measuring multimorbidity to ensure consistency in my studies. I did this by comparing the estimated prevalence of multimorbidity when varying the number and selection of long-term conditions considered in the count. Then, I examined the impact of the choice of data source (namely primary care, hospital inpatient, and primary care linked to hospital inpatient data) on multimorbidity prevalence ascertainment and the likelihood of mortality. The results of these studies informed my selection of long-term conditions and choice of data sources in the following two studies.

Second, I examined the associations between multimorbidity in households and unplanned hospitalisation and transitioning to live in a care home. To do this, I explored how the association between an individual’s multimorbidity status (categorised as 0-1, 2-3, and ≥4 long-term conditions) and the hazard of these two outcomes varied by household size (categorised as 1, 2, and ≥3 residents) in a population of adults aged ≥65. I then examined the impact of household size and the multimorbidity status of household co-residents on unplanned hospitalisations and transitioning to live in a care home in a study of adults aged ≥18. I used multistate models that incorporated household random effects and accounted for the competing risk of death, examining the outcomes of unplanned hospitalisation and transitioning to live in a care home in separate models.

I conducted all studies using large population-based datasets, including linked primary care and hospital inpatient records, from England using the Clinical Population Research Datalink and Wales using the SAIL Databank.

RESULTS: In a cross-sectional analysis of 1,168,620 individuals of all ages living in England, multimorbidity prevalence estimates varied widely depending on the number and selection of long-term conditions considered in counts. When only the two most common long-term conditions were considered, multimorbidity prevalence was 4.6% (95% CI 4.6-4.6) but was substantially higher (40.5%, 95% CI 40.4-40.6) when all the long-term conditions in the study (80) were included in the count. A ceiling where the prevalence reached over 99% of that reached when all 80 long-term conditions were included occurred when considering the 52 most prevalent long-term conditions. Using a list of long-term conditions recommended in a recent Delphi study resulted in multimorbidity prevalence estimates close to this threshold.

Using primary care linked to hospital inpatient data, compared to hospital inpatient data, multimorbidity was substantially more prevalent (32.2% versus 16.5%) in a second cross-sectional analysis, including 2,340,027 individuals of all ages living in Wales. Additionally, for the same comparison, those with multimorbidity were younger (mean age 62.5 versus 66.8), a higher proportion of women (54.2% versus 52.6%), and had a higher likelihood of 1-year mortality (in people with ≥4 versus 0-1 long-term conditions the adjusted odds ratio was 8.34 [95% CI 8.02-8.68] using linked, but 6.95 (95%CI 6.79-7.12] when the same comparison was made using hospital inpatient data).

In my cohort study of 391,686 adults living in Wales and aged ≥65, 36.8% lived alone, 54.0% lived in two-person, and 9.2% lived in three-or-more-person households. The number of long-term conditions was strongly associated with hospitalisation and care home transition (e.g., in two-person households, ≥4 long- term conditions versus 0-1 long-term conditions, adjusted hazard ratio [aHR] 2.46 [95%CI 2.42-2.50] for hospitalisation and 3.00 [95%CI 2.90-3.10] for care home).

The association between the number of long-term conditions and unplanned hospitalisation and care home transition varied by household size, with the association being less pronounced in those living alone. Household size was modestly associated with unplanned hospitalisation but more strongly associated with care home transition (e.g., in people with 0-1 long-term conditions, living alone versus in a two-person household, aHR 1.20 [95%CI 1.17-1.22] for hospitalisation and 1.45 [95%CI 1.39-1.51] for care home transition). The risk of care home transition was similar for people with 0-1 long-term conditions who lived alone and those who had 2-3 long-term conditions who lived in a two-person household.

The highest unplanned hospitalisation and care home transition rates were in those living alone (92.0 and 5.9 events/1000 person-years, respectively) in my final study, including 1,472,185 individuals aged ≥18 years and living in Wales. Event rates for both outcomes were lower in all shared households and lowest when co-residents did not have multimorbidity. In three-or-more-person households, compared to living alone, the aHR for unplanned hospitalisation was lowest when co-residents did not have multimorbidity (0.87, 95%CI 0.86-0.88) and intermediate when at least one co- resident did have multimorbidity (0.92, 95%CI 0.91-0.93). Differences were more substantial for care home transition, for example, living with co-residents without multimorbidity aHR 0.57, 95% CI 0.55-0.59, or with at least one co-resident with multimorbidity aHR 0.78, 95% CI 0.75-0.80).

CONCLUSIONS: The prevalence of multimorbidity varied substantially depending on the number of long-term conditions considered in the count and the data source used. Therefore, consistency of measurement is needed to achieve consistent results across studies.

Household size, an individual’s multimorbidity status, and their co-resident’s multimorbidity status modestly increased the risk of unplanned hospitalisation and more substantially increased the risk of transitioning to live in a care home. However, living alone was associated with the greatest hazard in all models.

Understanding the mechanisms of the associations between household composition and co-resident health status with these outcomes is important to develop targeted strategies that support healthy and independent living.