Effect of External Variables on the Performance of the Geriatric Comorbidity Score Derived from Prescription Claims in the Community-Dwelling Elderly
Comorbidity scores based on medical or prescription claims data are frequently used to control for confounding in pharmacoepidemiological studies. Performance of such scores in predicting the risk of death in community-dwelling elderly adults may be compromised by the absence of mental health and socioeconomic characteristics not captured in claims data.
The aim of the study was to assess the impact of adding mental health status and sociodemographic characteristics to the Geriatric Comorbidity Score (GCS), a score derived from prescription claims data in the Quebec community-dwelling elderly population.
We used the cohort study from the longitudinal Quebec Seniors’ Health Survey (n = 1,494) conducted between 2005 and 2006. For each participant, we obtained mental health and socioeconomic characteristics through validated questionnaires, which we linked with the medical and prescription claims databases of the Quebec Health Insurance Agency [Régie de l’assurance maladie du Québec (RAMQ)]. The main study outcome was death within 1 year, ascertained using the Quebec death registry. The GCS was calculated from prescription claims data, with the c statistic as a measure of performance. Using backward stepwise selection, external variables (marital status, region, family income, social support, daily hassles, perceived physical and mental health status, presence of mental health disorders) were added to the logistic regression model and the marginal effect assessed by comparing the c statistic with and without each covariate.
Over 1 year, 77 deaths (5.15 %) were reported. The c statistic for the GCS was calculated as 0.67 (95 % confidence interval 0.64, 0.70). Addition of sex and age to the score yielded a 2.4 % increase. The variable with the greatest impact on the c statistic was marital status (6.1 % increase). Though important contributors, social support and perceived mental health status did not significantly improve performance of the score.
While sex, age and marital status significantly improved performance of a predictive score in the community-dwelling elderly population, the absence of data on mental and physical health status did not appear to compromise the validity of claims-based scores. Combining comorbidity scores with other methods to control for confounding thus remains a useful tool in pharmacoepidemiological research.
The authors are grateful to Danielle Buch, medical writer, for revision and editing of the manuscript. The authors also wish to thank the Régie de l’assurance maladie du Québec (RAMQ). The authors wish to thank the Canadian Institute of Health Research (CIHR) for the doctoral fellowship received by S.G.B., and the Reseau Quebecois de recherche sur le vieillissement, the Reseau Quebecois de recherche sur l’Utilisation du Medicament and the Fonds de recherché en Sante du Quebec for their funding contributions to the study. Y.M. has received several grants from, and has acted as a consultant for, organizations in the pharma industry and has served on two data and safety monitoring boards.
Conflict of interest
The authors have no conflict of interest to declare that are directly relevant to the content of this study.
- 12.Grenier S, Préville M, Boyer R, et al. The impact of DSM-IV symptom and clinical significance criteria on the prevalence estimates of subthreshold and threshold anxiety in the older adult population. Am J Geriat Psychiat. 2010;19(4):316–26.Google Scholar
- 14.Préville M, Boyer R, Grenier S, Dube M, Voyer P, Punti R, et al. The epidemiology of psychiatric disorders in Quebec’s older adult population. Can J Psychiatry Revue canadienne de psychiatrie. 2008;53(12):822–32.Google Scholar
- 16.Boyle PJ, Feng Z, Raab GM. Does widowhood increase mortality risk?: testing for selection effects by comparing causes of spousal death. Epidemiology. 2011;22(1):1–5.Google Scholar
- 17.Beland SG, Ducruet T, Tannenbaum C, et al. Development and validation of an Geriatric Comorbidity Score based of drug use. Pharmacoepidem DrS, vol. 20 (Issue Supplement S1); 2011. p. S1–382.Google Scholar
- 18.Quebec Prescription Drug Insurance Plan. Prescription claims database [data used: January 2011]. Québec: Régie de l’Assurance Maladie du Quebec; 2012.Google Scholar
- 20.Romano PS, Roos LL, Jollis JG. Adapting a clinical comorbidity index for use with ICD-9-CM administrative data: differing perspectives. J Clin Epidemiol. 1993;46(10):1075–9 (discussion 81–90).Google Scholar
- 23.Death records. Québec: Institut de la statistique du Québec; 2010.Google Scholar
- 24.International classification of disease. 9th ed. Geneva: World health Organization; 1977.Google Scholar
- 26.Vézina JGL. L’Échelle des Embêtements: une étude de validation française du “Hassles Scale” pour les personnes âgées. Communication presented at the 49th Annual Meeting of the Canadian Psychological Association. 1988; Montreal.Google Scholar
- 28.Hosmer D, Lemeshow S. Applied logistic regression. 2nd ed. Wiley Series in probability statistics. Hoboken: Wiley; 2000.Google Scholar
- 29.American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 4th ed. Washington, D.C: American Psychiatric Press; 1994.Google Scholar
- 30.Manzoli L, Villari P, Pirone GM, Boccia A. Marital status and mortality in the elderly: a systematic review and meta-analysis. Soc Sci Med. 2007;64(1):77–94.Google Scholar