The Role of Health-Related Lifestyle Attitudes in Understanding Health Behaviors and Outcomes

Health systems throughout the world increasingly recognize health-related quality of life (HRQoL) as a key patient-centered health outcome. HRQoL is particularly important in conditions which impair daily living (e.g. Parkinson’s Disease) and in those that are largely subjective in nature (e.g. pain). HRQoL, when assessed using a preference-based measure, is an essential component in the calculation of quality adjusted life years (QALYs). QALYs gained form the denominator in an incremental cost-effectiveness ratio (ICER), which is the most common result derived in a cost-utility analysis and is used to inform patient and system-level decisions and incorporated into pricing and reimbursement decisions. HRQoL can be viewed as an individual’s perception of his/her own health. An individual’s perception is often influenced by attitudes and beliefs, which in turn also influence health behaviors and overall health. This dissertation examines the role of health and life-style related attitudes (HLA) in understanding health behaviors and outcomes through three separate studies. The studies evaluated the validity of measures of HRQoL and HLA in a community population and examined the relationship between HLA and health behaviors, self-assessment of health and life-satisfaction. The first chapter introduces key study concepts including HLA, self-assessment and life-satisfaction along with a conceptual framework that facilitates the understanding of the relationship between HLA, health behaviors and outcomes (including self-assessment of health and life satisfaction). The introductory chapter highlights the gaps in the evidence, research objectives and hypotheses that were examined in subsequent chapters, and explains the significance of the research. All the studies are based on secondary analyses from a large scale community health study conducted in Kirklees, United Kingdom, in 2012. The second chapter examines the properties of the recently developed EQ-5D-5L as a measure of population health in a large community-based survey. The assessment of the construct validity of the EQ-5D-5L in a community sample helps to facilitate in the use of EQ-5D-5L in general population health studies. A secondary aim of the study was to compare EQ-5D-5L summary scores based on an English value set (5L-ENG), crosswalk scores based on 3L values (XW), and scores based on equal weighting (5L-EWS). Properties of the EQ-5D-5L that were examined included discriminative ability, strength of correlations with related measures of health, and statistical efficiency, as assessed with relative efficiency (RE) ratios between summary scores from known group comparisons, using the XW as the referent. Correlations between EQ-5D dimensions and other measures of health ranged from r =0.56-0.73. Compared to the XW, the RE of the 5L-ENG and 5L-EWS were similar or higher (Age:RE5L-ENG=0.96,RE5L-EWS=1.36; BMI:RE5L-ENG=1.00,RE5L-EWS=1.0; Number of prescription medications:RE5L-ENG=1.00,RE5L-EWS=1.18; Number of Chronic conditions:RE5L-ENG=1.00,RE5L-EWS=1.61; General Health:RE5L-ENG=1.13,RE5L-EWS=1.20). The results provided evidence to support the validity of EQ-5D-5L in large community based surveys. If statistical efficiency of a summary score is a desirable property, the EWS should be considered for non-QALY purposes. The third chapter describes the second study which sought to determine the extent to which health and lifestyle related attitudes (HLA) explained health behaviors and outcomes. If certain HLAs are associated with specific health behaviors, it raises the possibility of using HLA based segmentation to improve the uptake of health interventions designed, for instance, to reduce drug abuse. Individuals were characterized based on HLA using two different approaches: (1) as five discrete segments; and (2) as 3 single-item scales (“importance of health”, “involvement in health” and “control of health”). The association between each HLA approach and several potentially modifiable behaviors/conditions was assessed using logistic regression. When respondents were categorized according to HLA segments, 30.8% were health conscious realists, 15.0% balanced compensators, 17.7% life-for-todays, 10.0% hedonistic immortals and 26.4% unconfident fatalists. Correlational analysis between HLA segments and trait scales produced low scores and moderate negative correlations between specific HLA traits and segments, which indicated the approaches were complementary. The model fit between HLA and health behaviors/conditions was weak and except for mental health problems, was similar across the approaches. Both approaches demonstrated convergent and divergent validity, e.g. unconfident fatalists and those with low levels of the trait “involvement in health” were significantly more likely to report poor diet [OR=2.27 (95%CI: 2.02, 2.55), OR=1.59 (95%CI: 1.45, 1.74), respectively. The HLA segmentation approach performed better as a predictor of certain behaviors (e.g. tobacco use) than others (e.g. exercise). Both the segment and the trait methods had comparable performance in explaining health behaviors and outcomes. Thus, both provided viable approaches to identify sub-groups of the population who are more likely to engage in certain negative health behaviors that could be targeted for tailored healthcare interventions. The fourth chapter describes the third study, which examined the relationship between HLA and self-assessment of health and life-satisfaction. Evaluation of the relationship between HLA and health behaviors and self-assessment of HRQoL and life satisfaction may help to identify groups which systematically differ in their valuation of health, despite of being in similar health state. In this study, individuals were characterized based on HLA segments [Health Conscious Realists(HCR), Balanced Compensators(BC), Live for Todays(LFT), Hedonistic Immortals(HI), and Unconfident Fatalists(UF)] using an algorithm for the 19-item Healthy Foundations Lifestyle Segmentation questionnaire. Associations between HLA and outcomes of interest [(i) self-assessment of health (EQ-VAS), (ii) differences between personal (i.e. EQ-VAS) and societal valuation of health (Dev), and (iii) life-satisfaction] were examined by multivariate regression analyses, adjusting for current health and relevant individual characteristics. The societal valuation was based on the preference-based algorithm for the EQ-5D-5L. Dev was examined to evaluate if HLA segments differ systematically between personal valuation of health and societal valuation of the health state. In the analytical sample (n=9,130), compared to HCR respondents (reference): (i) mean VAS scores for UF, LFT and BC respondents were 9.02 (p<0.001) and 1.96 points lower (p=0.002) and 1.15 points higher (p<0.01) respectively; (ii) Dev score in UF respondents was 0.027 lower (p<0.001), and (iii) BC respondents were less likely (OR:0.8,95%CI:0.6, 0.9), while UF were more likely (OR:3.5, 95%CI: 2.9, 4.2) to report low lifesatisfaction. Thus, HLA were associated with systematic differences in self-assessment of health and life-satisfaction, most notably UF, who rated their health much more negatively and had much lower life satisfaction, even after adjusting for characteristics such as age and gender. Results suggest that unconfident fatalists, who compromised 30% of survey respondents, are an important subgroup in the community who may need to be separately targeted for health care interventions, particularly behavioral. Importantly from a health services research perspective, the assessment of self reported HRQoL in this subgroup using measures of health and well-being may need to be separately considered in analysis and interpretation. Further research on this topic is warranted given the increasing use of patient-reported outcomes in health systems, particularly in conditions in which UF are disproportionately represented. The fifth chapter provides an overall conclusion to the dissertation. The characterization of individuals based on HLA's implications for health promotion and self assessment of health is discussed, and future research directions are proposed.