Longitudinal beta regression models for analyzing health-related quality of life scores over time

BMC Medical Research Methodology

Rolf Holle²  Angela Döring¹  Matthias Hunger² 

[1] Helmholtz Zentrum München, German Research Center for Environmental Health (GmbH), Institute of Epidemiology II, Ingolstädter Landstr. 1, Neuherberg, 85764, Germany;Helmholtz Zentrum München, German Research Center for Environmental Health (GmbH), Institute of Health Economics and Health Care Management, Ingolstädter Landstr 1, Neuherberg, 85764, Germany

关键词: Marginal model;    Mixed model;    Longitudinal study;    Beta regression;    Health-related quality of life;   

Others  :  1126702
DOI  :  10.1186/1471-2288-12-144

 received in 2012-04-04, accepted in 2012-09-12,  发布年份 2012

【 摘 要 】

Background

Health-related quality of life (HRQL) has become an increasingly important outcome parameter in clinical trials and epidemiological research. HRQL scores are typically bounded at both ends of the scale and often highly skewed. Several regression techniques have been proposed to model such data in cross-sectional studies, however, methods applicable in longitudinal research are less well researched. This study examined the use of beta regression models for analyzing longitudinal HRQL data using two empirical examples with distributional features typically encountered in practice.

Methods

We used SF-6D utility data from a German older age cohort study and stroke-specific HRQL data from a randomized controlled trial. We described the conceptual differences between mixed and marginal beta regression models and compared both models to the commonly used linear mixed model in terms of overall fit and predictive accuracy.

Results

At any measurement time, the beta distribution fitted the SF-6D utility data and stroke-specific HRQL data better than the normal distribution. The mixed beta model showed better likelihood-based fit statistics than the linear mixed model and respected the boundedness of the outcome variable. However, it tended to underestimate the true mean at the upper part of the distribution. Adjusted group means from marginal beta model and linear mixed model were nearly identical but differences could be observed with respect to standard errors.

Conclusions

Understanding the conceptual differences between mixed and marginal beta regression models is important for their proper use in the analysis of longitudinal HRQL data. Beta regression fits the typical distribution of HRQL data better than linear mixed models, however, if focus is on estimating group mean scores rather than making individual predictions, the two methods might not differ substantially.

【 授权许可】

   
2012 Hunger et al.; licensee BioMed Central Ltd.

【 图 表 】

Figure 1.

Figure 2.

Figure 3.

【 参考文献 】

• [1]Calvert M, Blazeby J, Revicki D, Moher D, Brundage M: Reporting quality of life in clinical trials: a CONSORT extension. Lancet 2011, 378(9804):1684-1685.
• [2]Konig HH, Bernert S, Angermeyer MC, Matschinger H, Martinez M, Vilagut G, Haro JM, de Girolamo G, de Graaf R, Kovess V, Alonso J: Comparison of population health status in six european countries: results of a representative survey using the EQ-5D questionnaire. Med Care 2009, 47(2):255-261.
• [3]Basu A, Manca A: Regression estimators for generic health-related quality of life and quality-adjusted life years. Med Decis Making 2012, 32(1):56-69.
• [4]Pullenayegum EM, Tarride JE, Xie F, Goeree R, Gerstein HC, O'Reilly D: Analysis of health utility data when some subjects attain the upper bound of 1: are Tobit and CLAD models appropriate? Value Health 2010, 13(4):487-494.
• [5]Austin PC: A comparison of methods for analyzing health-related quality-of-life measures. Value Health 2002, 5(4):329-337.
• [6]Li L, Fu AZ: Some methodological issues with the analysis of preference-based EQ-5D index score. Health Serv Outcomes Res Method 2009, 9:162-176.
• [7]Ferrari SLP, Cribari-Neto F: Beta Regression or Modeling Rates and Proportions. J Appl Statist 2004, 31(7):799-815.
• [8]Smithson M, Verkuilen J: A better lemon squeezer? Maximum-likelihood regression with beta-distributed dependent variables. Psychol Methods 2006, 11(1):54-71.
• [9]Bruche M, González-Aguado C: Recovery rates, default probabilities and the credit cycle. J Banking Finance 2010, 34(4):754-764.
• [10]Hunger M, Baumert J, Holle R: Analysis of SF-6D index data: is beta regression appropriate? Value Health 2011, 14(5):759-767.
• [11]Koch CG, Khandwala F, Blackstone EH: Health-related quality of life after cardiac surgery. Semin Cardiothorac Vasc Anesth 2008, 12(3):203-217.
• [12]McCulloch CE: Repeated measures ANOVA, R.I.P.? Chance 2005, 18(3):29-33.
• [13]Schaafsma FG, Kurrle SE, Quine S, Lockwood K, Cameron ID: Wearing hip protectors does not reduce health-related quality of life in older people. Age Ageing 2012, 41(1):121-125.
• [14]Barrera M, Atenafu E, Hancock K: Longitudinal health-related quality of life outcomes and related factors after pediatric SCT. Bone Marrow Transplant 2009, 44(4):249-256.
• [15]Chen H, Cohen P: Using individual growth model to analyze the change in quality of life from adolescence to adulthood. Health Qual Life Outcomes 2006, 4:10.
• [16]Verkuilen J, Smithson M: Mixed and mixture regression models for continuous bounded responses using the beta distribution. J Educ Behav Stat 2012, 37(1):82-113.
• [17]Zimprich D: Modeling change in skewed variables using mixed beta regression models. Res Hum Dev 2010, 7(1):9-26.
• [18]Fitzmaurice GM, Laird NM, Ware JH: Applied Longitudinal Analysis. 2nd edition. Wiley, New York; 2011.
• [19]Hu FB, Goldberg J, Hedeker D, Flay BR, Pentz MA: Comparison of population-averaged and subject-specific approaches for analyzing repeated binary outcomes. Am J Epidemiol 1998, 147(7):694-703.
• [20]Holle R, Happich M, Lowel H, Wichmann HE: KORA–a research platform for population based health research. Gesundheitswesen 2005, 67(Suppl 1):S19-25.
• [21]Brazier JE, Roberts J: The estimation of a preference-based measure of health from the SF-12. Med Care 2004, 42(9):851-859.
• [22]Duncan PW, Wallace D, Lai SM, Johnson D, Embretson S, Laster LJ: The stroke impact scale version 2.0. evaluation of reliability, validity, and sensitivity to change. Stroke 1999, 30(10):2131-2140.
• [23]Hunger M, Sabariego C, Stollenwerk B, Cieza A, Leidl R: Validity, reliability and responsiveness of the EQ-5D in German stroke patients undergoing rehabilitation. Qual Life Res 2011, 21(7):1205-1216.
• [24]Rollnik JD, Janosch U: Current trends in the length of stay in neurological early rehabilitation. Dtsch Arztebl Int 2010, 107(16):286-292.
• [25]Brown LD: Fundamentals of statistical exponential families: with applications in statistical decision theory. Volume 9. Hayward, California: Institute of Mathematical Statistics - Lecture Notes; 1986.
• [26]Cribari-Neto F, Zeileis A: Beta Regression in R. J Stat Softw 2010, 34(2):1-24.
• [27]Aitchison J: The statistical analysis of compositional data. J Roy Stat Soc B Met 1982, 44(2):139-177.
• [28]Carriere I, Bouyer J: Choosing marginal or random-effects models for longitudinal binary responses: application to self-reported disability among older persons. BMC Med Res Methodol 2002, 2:15.
• [29]Gardiner JC, Luo Z, Roman LA: Fixed effects, random effects and GEE: what are the differences? Stat Med 2009, 28(2):221-239.
• [30]Hahn EA, Cella D, Dobrez DG, Weiss BD, Du H, Lai JS, Victorson D, Garcia SF: The impact of literacy on health-related quality of life measurement and outcomes in cancer outpatients. Qual Life Res 2007, 16(3):495-507.
• [31]Levy AR, Christensen TL, Johnson JA: Utility values for symptomatic non-severe hypoglycaemia elicited from persons with and without diabetes in Canada and the United Kingdom. Health Qual Life Outcomes 2008, 6:73.
• [32]Papke LE, Wooldridge JM: Econometric methods for fractional response variables with an application to 401(k) plan participation rates. J Appl Econometrics 1996, 11:619-632.
• [33]Liang KY, Zeger S: Longitudinal Data Analysis Using Generalized Linear Models. Biometrika 1986, 73(1):13-22.
• [34]Zeger SL, Liang KY: Longitudinal data analysis for discrete and continuous outcomes. Biometrics 1986, 42(1):121-130.
• [35]Molenberghs G, Verbeke G: Models for Discrete Longitudinal Data. New York: Springer; 2005.
• [36]Ibrahim JG, Molenberghs G: Missing data methods in longitudinal studies: a review. Test (Madr) 2009, 18(1):1-43.
• [37]Little RJA, Rubin DB: Statistical analysis with missing data. 2nd edition. New York: Wiley; 2002.
• [38]Horton NJ, Lipsitz SR: Review of software to fit generalized estimating equation (gee) regression models. Am Stat 1999, 53:160-169.
• [39]Lipsitz SR, Molenberghs G, Fitzmaurice GM, Ibrahim J: GEE with Gaussian estimation of the correlations when data are incomplete. Biometrics 2000, 56(2):528-536.
• [40]Robins JM, Rotnitzky A, Zhao LP: Analysis of semiparametric regression models for repeated outcomes in the presence of missing data. J Am Stat Assoc 1995, 90(429):106-121.
• [41]Dean CB, Nielsen JD: Generalized linear mixed models: a review and some extensions. Lifetime Data Anal 2007, 13(4):497-512.
• [42]Cox DR, Snell EJ: Analysis of Binary Data. 2nd edition. Boca Raton: Chapman & Hall; 1989.
• [43]Espinheira PL, Ferrari SLP, Cribari-Neto F: On beta regression residuals. J Appl Statist 2008, 35:407-419.
• [44]Ferrucci L, Bandinelli S, Guralnik JM, Lamponi M, Bertini C, Falchini M, Baroni A: Recovery of functional status after stroke. a postrehabilitation follow-up study. Stroke 1993, 24(2):200-205.
• [45]Frank B, Schlote A, Hasenbein U, Wallesch CW: Prognosis and prognostic factors in ADL-dependent stroke patients during their first in-patient rehabilitation–a prospective multicentre study. Disabil Rehabil 2006, 28(21):1311-1318.
• [46]Johnson JA, Pickard AS, Conner-Spady B: Longitudinal changes in health-related quality of life of people with diabetes compared to those without chronic conditions. In Proceedings of the 17th EuroQol Plenary Meeting. Edited by Badia X. Pamplona, Spain; 2010:193-206.