SAT0598 PREDICTORS OF HEALTH-RELATED QUALITY OF LIFE IN PATIENTS WITH MUSCULOSKELETAL DISEASES: A LONGITUDINAL ANALYSIS FROM AN ELECTRONIC HEALTH RECORD DATABASE

Background:Musculoskeletal diseases (MSKs) are the major cause of disability, affecting the patient’s daily activities as well as their Quality of Life (QoL).Objectives:To analyze in a cohort of patients with MSKs the influence of demographic and clinical predictors in repeated measures of Health-Related QoL (HRQoL) obtained from a departmental electronic health record (EHR)Methods:Patients attending the Hospital Clínico San Carlos rheumatology outpatient clinic (HCSC-MSK cohort) from April 1st, 2007 to November 30th, 2017 who were seen at least twice were included in this study. Visits separated more than one year from previous or next visit were excluded. Our primary outcome was HRQoL collected in each patient’s visit using the Rosser Classification Index (it ranges between 0 and 1, the higher value, the higher HRQoL. It is calculated as the combination of two subscales: disability and distress). Demographic and clinical variables, such as diagnoses, treatments, and comorbidities were included as predictors. Variable selection was carried out in 4 steps. First, we developed a bivariate Generalized Estimating Equations model, selecting those variables with a p-value < 0.15. Second, those with a pair-wise absolute correlation < 0.95 were carried to the next step. Third, multivariate GEE analyses were performed including clinical related variables, and those independently associated with our primary outcome were selected with a p-value < 0.15. At least one predictor of each multivariate analyses was included even if none of them had p-value < 0.15. In this case the one, with the lowest p-value was chosen. Fourth, a grid search approach was followed to determine the penalization value and a penalized GEE model was implemented including all selected variables and using an independence correlation structure. Those with an estimate greater than 10^-3 and a z-robust > 1.96 were considered to be independently associated with the HRQoL.Results:A total of 18,187 patients with 95,960 outpatient visits were included in this study. From the initial 410 predictors, 289, 241, 153, and 34 were selected in the first, second, third, fourth and fifth steps. The median age was 56.4 years. The major part of the patients were women (69.4%), and the medianRosservalue was 0.986. More common diagnoses, treatments and comorbidities were back pain, tendinitis, pain in joint; analgesia, gastric protector and NSAIDs; hypertension, dyslipidemia and diabetes respectively. The predictors with a greater negative impact in HRQoL were the use of 3rd level analgesics and azathioprine, a presence of kidney failure, fibromyalgia, and ischemic heart disease. Conversely, use of symptomatic slow action drugs for osteoarthritis, statins, lowering uric acid drugs, a diagnose of mixed connective tissue disease, and better HRQoL in the past six months were independently associated with a positive impact in the HRQoL.Conclusion:We have identified several diagnoses, treatments and comorbidites independently associated with HRQoL in a cohort of patients followed up in a rheumatology outpatient clinic. This represent a first step in the implementation of value-based care for MSK patients, as we can now review the procedures associated with a worse HRQoL in an attempt to improve them.References:Predictor selection. Dx: Diagnoses. DxE: Diagnoses Episode. DxTFT: Diagnoses Total Follow-up Time.DxTET: Diagnoses Total Episode Follow-up Time T: Treatment. DxE: Treatment Episode. DxTFT:Treatment Total Follow-up Time. DxTET: Treatment Total Episode Follow-up Time. XM: Comorbiditiesand Concomitant Treatments. DM: Demographics. HS: Health Status.Disclosure of Interests:None declared