Cross-National Variation in Quality of Life of Care-Dependent Elders in Europe: A Two-Step Approach Combining Multilevel Regression and Fuzzy-Set QCA

Abstract

Cross-national variation in subjective quality of life (sQoL) among older long-term care recipients (LTCRs) in Europe is substantial but rarely subject to explicit theoretical or empirical consideration. We suggest combining multilevel regression analysis with fuzzy-set qualitative comparative analysis (QCA) in a two-step approach given the small number of countries for which cross-national survey data are available and the considerable number of potentially relevant and likely interdependent country-level characteristics. Based on data from the Survey of Health, Ageing and Retirement in Europe (SHARE 2012), sQoL in 2,470 older LTCRs from 14 countries was analyzed. Results of multilevel regression analyses (1) confirmed known individual-level determinants of sQoL and provided a weighted country-level outcome to be explained by fuzzy-set QCA, where (2) we found both necessary and sufficient (combinations of) country-level conditions for high- and low-quality-of-life countries in Europe. In combination, these two distinct methodological tools allowed for a more comprehensive analysis of country-level characteristics on sQoL in older LTCRs in Europe.

Keywords:

• cross-national
• long-term care recipients
• multilevel regression
• QCA
• quality of life

Notes

Simply aggregating individual observations and conducting an ecological analysis of countries was not considered due to the resulting information loss, problematic ecological inferences, and the fact that most of the variance in individual-level outcomes such as sQoL is usually found within countries, that is, between individuals, rather than between countries.

Notation of Boolean algebra: Logical “and” = (*), logical “or” = (+), negation = (∼), “if then” = (→).

Respectively includes simplifying assumptions regarding easy counterfactual cases (Ragin 2008).

We do not discuss recent methodological contributions regarding the analysis of multilevel data within QCA (e.g., Thiem 2014) as we use “single-level” QCA in this research as subsequent step of analysis after multilevel regression modeling, which provides the random variance component on which the subsequent QCA outcome is based.

This study uses data from SHARE wave 4 release 1.1.1, as of March 28, 2013 (DOI: 10.6103/SHARE.w4.111). The SHARE data collection was primarily funded by the European Commission through the Fifth Framework Programme (project QLK6-CT-2001-00360 in the thematic program Quality of Life), through the Sixth Framework Programme (projects SHARE-I3, RII-CT-2006-062193, COMPARE, CIT5- CT-2005-028857, and SHARELIFE, CIT4-CT-2006-028812), and through the Seventh Framework Programme (SHARE-PREP, No. 211909, SHARE-LEAP, No. 227822 and SHARE M4, No. 261982). Additional funding from the U.S. National Institute on Aging (U01 AG09740-13S2, P01 AG005842, P01 AG08291, P30 AG12815, R21 AG025169, Y1-AG-4553-01, IAG BSR06-11 and OGHA 04-064), from the German Ministry of Education and Research, and from various national sources is gratefully acknowledged (see www.share-project.org for a full list of funding institutions).

Instrumental support was provided either co- or extra-residentially, or in both forms. In contrast to co-residential care and assistance, support from outside the household referred to the household level. Therefore, in the case of a cohabiting partner, the receipt of extra-residential care was assigned to those respondents who reported functional limitations. When none of the partners reported such limitations, the receipt of care was assigned to the partner with the lower self-reported health status. In rare cases of no reported functional limitations and the same level of self-reported health in partners, the family respondent was selected.

We used a flat prior and 25,000 iterations, which, with a burn-in period of 5,000 and thinning by factor 10, resulted in an effective Bayesian sample size of 2,000. Optical inspection, low autocorrelation, and low R-hat (<1.01) indicated proper convergence. Country differences are represented by posterior mean values and standard deviations. Due to asymmetric distributions of the random variance component, posterior median and credible intervals were used for the calculation of ICC.

Retrieved July 23, 2015 (http://data.worldbank.org/about/country-and-lending-groups#High_income).

The slightly more complex solution formula including HTS appeared when excluding the Netherlands or Spain for HQOL. For ∼HQOL, excluding one of the Eastern or Southern European countries (except Slovenia) resulted in the slightly more complex solution including ∼HSS.

Additional information

Notes on contributors

Erwin Stolz

Erwin Stolz is an assistant professor at the Institute of Social Medicine and Epidemiology at the Medical University of Graz and a lecturer in the Department of Sociology at the University of Salzburg. His research interests are sociology of aging, long-term care, social policy, social inequality, and cross-national research.