Background Multimorbidity is an important patient-centered concept that needs to be considered when deciding on diagnostic or therapeutic strategies for typical rheumatoid arthritis (RA) patients. In chronic diseases like RA, health related quality of life (HRQol) is the main outcome. An index reflecting multimorbidity (MMI) that is based on HRQoL would be novel, as existing indices are commonly comorbidity indices based on more unidimensional outcomes, such as mortality, costs or function. A MMI would be helpful to better address the disease-related aspect of patients' overall well-being.
Objectives To develop a multimorbidity index (MMI) based on HRQoL.
Methods The index was developed in an RA cohort, the Brigham and Women's Hospital Rheumatoid Arthritis Sequential Study (BRASS). 40 specific morbidities recommended as core by a systematic literature review were identified using ICD-9 codes; the primary outcome was EQ-5D. Two types of MMI were calculated, one by simply enumerating concomitant morbidities (MMIc), and another one by weighing mobid conditions based on their association with HRQoL (using EQ5D) in a multiple linear regression analysis (weights derived from β-coefficients of the regression model) (MMIw). Performance of both MMI was compared to the Charlson comorbidity index (CCI).
Results In total 544 out of 876 patients were multimorbid, defined as ≥2 morbid conditions. MMIc ranged from 1-16 (median 2; 25th/75th percentile 1/3); MMIw ranged from 0 – 38 (mean±SD 3.8±6.1). Both indices were similarly associated with EQ-5D, and significantly more strongly than the established CCI (Correlation coefficient r (95CI)%: MMIc -0.21 (-0.15 to -0.28); MMIw -0.32 (-0.26 to -0.38), CCI -0.10 (-0.03 to -0.15); p<0.01 each). The relationship of MMI and EQ-5D followed a linear trend, shown in the figure. R2 obtained by linear regression models using EQ-5D as dependent variable and the two Indices as independent variable, adjusted for age and gender was highest for MMI (R2: MMIc 0.05, MMIw 0.11, CCI 0.01). When accounting for RA disease activity (using Clinical disease activity index CDAI) R2 increased (R2 MMIc 0.18; MMIw 0.22; CCI 0.17), still showing highest values of MMI compared to CCI. The predictive validity of the MMI was robust as demonstrated by the agreement of predicted values of EQ5D after one year with observed values of EQ5D after one year (Spearman r: MMIc 0.34, MMIw 0.37; both p<0.01).
Conclusions In our cohort, the novel MMI showed a better relationship with HRQol than the well known CCI. A simple enuneration of morbid conditions is similarly effective as a HRQoL weighted index, and could therefore be useful to control for the effect of multimorbidity on patient's overall well being.
Disclosure of Interest None declared
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.