Bayesian Hierarchical Models for Meta-Analysis of Quality-of-Life Outcomes: An Application in Multimorbidity
Health-related quality of life (HRQoL) is a key outcome in cost-utility analyses, which are commonly used to inform healthcare decisions. Different instruments exist to evaluate HRQoL, however while some jurisdictions have a preferred system, no gold standard exists. Standard meta-analysis struggles with the variety of outcome measures, which may result in the exclusion of potentially relevant evidence.
Using a case study in multimorbidity, the objective of this analysis is to illustrate how a Bayesian hierarchical model can be used to combine data across different instruments. The outcome of interest is the slope relating HRQoL to the number of coexisting conditions.
We propose a three-level Bayesian hierarchical model to systematically include a large number of studies evaluating HRQoL using multiple instruments. Random effects assumptions yield instrument-level estimates benefitting from borrowing strength across the evidence base. This is particularly useful where little evidence is available for the outcome of choice for further evaluation.
Our analysis estimated a reduction in quality of life of 3.8–4.1% per additional condition depending on HRQoL instrument. Uncertainty was reduced by approximately 80% for the instrument with the least evidence.
Bayesian hierarchical models may provide a useful modelling approach to systematically synthesize data from HRQoL studies.
SSc extracted the data, conducted the statistical analysis, interpreted the results, and wrote the first draft of the manuscript. TM conducted the literature review, extracted and checked the data, and provided detailed feedback on the manuscript. RA provided important input on the application of HRQoL in CUA, and provided detailed feedback on the manuscript. MvA extracted data and provided feedback on the manuscript. SSt and MZ reviewed and provided important feedback both during the analysis phase and for the manuscript.
Compliance with Ethical Standards
No funding was received to conduct this study.
Conflict of interest
Susanne Schmitz, Tatjana Makovski, Roisin Adams, Marjan van den Akker, Saverio Stranges, and Maurice P. Zeegers have no conflicts of interest to report in respect of this study.
- 2.Drummond MF, et al. Methods for the economic evaluation of health care programmes. Oxford: Oxford University Press; 2015.Google Scholar
- 3.Ware JE Jr, Sherbourne CD. The MOS 36-item short-form health survey (SF-36): I. Conceptual framework and item selection. Medical Care. 1992;30(6):473–83.Google Scholar
- 5.Oppe M, Devlin NJ, Szende A. EQ-5D value sets: inventory, comparative review and user guide. New York: Springer; 2007.Google Scholar
- 13.Earnshaw J, Lewis G. NICE guide to the methods of technology appraisal. New York: Springer; 2008.Google Scholar
- 14.National Centre for Pharmacoeconomics Ireland. Applicant template for submission of full pharmacoeconomic assessments to the National Centre for Pharmacoeconomics [cited 11 Sep 2019]. http://www.ncpe.ie/submission-process/submission-templates/format-of-full-submissions/.
- 18.van den Akker M, Buntinx F, Knottnerus JA. Comorbidity or multimorbidity: what’s in a name? A review of literature. Eur J Gen Pract. 1996;2(2):65–70.Google Scholar
- 25.Borenstein M, et al. Introduction to meta-analysis. New York: Wiley; 2011.Google Scholar
- 26.Deeks JJ, Higgins JP, Altman DG. Analysing data and undertaking meta-analyses. In: Higgins JP, Green S, editors. Cochrane handbook for systematic reviews of interventions: Cochrane book series. New York: Wiley; 2008. p. 243–96.Google Scholar
- 27.DerSimonian R, Laird N. Meta-analysis in clinical trials. Controll Clin Trials. 1986;7(3):177–88.Google Scholar
- 31.Dias S, et al. NICE DSU technical support document 2: a generalised linear modelling framework for pairwise and network meta-analysis of randomised controlled trials. London: National Institute for Health and Clinical Excellence; 2011.Google Scholar
- 39.Core Team R. R: a language and environment for statistical computing. Vienna: R Core Team; 2013.Google Scholar
- 41.Lunn DJ, et al. WinBUGS—a Bayesian modelling framework: concepts, structure, and extensibility. Stat Comput. 2000;10(4):325–37.Google Scholar
- 42.Sturtz S, Ligges U, Gelman AE. R2WinBUGS: a package for running WinBUGS from R. J Stat Softw. 2005;12(3).Google Scholar
- 44.Viechtbauer W. Conducting meta-analyses in R with the metafor package. J Stat Softw. 2010;36(3):1–48.Google Scholar
- 56.Kim KI, Lee JH, Kim CH. Impaired health-related quality of life in elderly women is associated with multimorbidity: results from the korean national health and nutrition examination survey. Gender Med. 2012;9(5):309–18.Google Scholar
- 57.Lang K, et al. The impact of multimorbidity on quality of life among midlife women: findings from a US nationally representative survey. J Womens Health (Larchmt). 2015;24(5):374–83.Google Scholar
- 73.Prazeres F, Santiago L. Relationship between health-related quality of life, perceived family support and unmet health needs in adult patients with multimorbidity attending primary care in Portugal: a multicentre cross-sectional study. Health Qual Life Outcomes. 2016;14(1):156.PubMedPubMedCentralGoogle Scholar
- 76.McCullagh L, et al. Examining the feasibility and utility of estimating partial expected value of perfect information (via a nonparametric approach) as part of the reimbursement decision-making process in Ireland: application to drugs for cancer. PharmacoEconomics. 2017;35(11):1177–85.PubMedGoogle Scholar