Cause-specific mortality in knee, hip and hand osteoarthritis

Abstract

Purpose: There has been much focus on cardiovascular mortality in persons with knee or hip osteoarthritis (OA), but there is paucity of information about other causes of death, i.e., causes that are over- or under-represented in persons with OA. We used a large population-based cohort to study cause-specific mortality in knee, hip and hand OA as compared to the general population. Methods: Using the Swedish Population Register, which contains information about vital events (births, deaths) and changes in residential address, we identified all persons aged 45 to 84 years resident in Skåne (southern Sweden, total population 1.15 million) in the year 2003. We used data from Skåne Health Care Register (SHR), a regional legislative, mandatory register based on physicians’ International Classification of Diseases (ICD) 10 diagnostic codes to identify persons who received the diagnosis of knee OA (ICD-10: M17), hip OA (M16) or hand OA (M18, M15.1, M15.2), at least once between 1998 and 2003. Using the Cause of Death Register we followed all persons from Jan 1st 2004 until relocation outside of the region, death, or Dec 31st 2014, whichever comes first. For each person we identified the underlying cause of death (registered with ICD-10 codes) and classified it into: cardiovascular, neoplasms, diabetes, infectious diseases, dementia, or other causes. We calculated hazard ratios (HR) of cause-specific mortality using a multistate Cox proportional hazard model adjusted for age and sex, disposable income, marital status, highest level of education reached, being born outside Sweden, residential area (all provided by Statistics Sweden) and diagnosed obesity (as registered in SHR). We evaluated the proportional hazards assumptions using plots of Schoenfeld residuals. If violations of this assumption were detected, we estimated the HRs after partitioning of the time scale. Results: We identified 15 901 persons with prevalent OA in knee, 9347 in hip and 4004 in hand among 469 177 persons in the Skåne population. Persons with OA were older, more often female and had more comorbidities than persons without. (Table 1) Over the 11 years of follow-up the crude mortality rates per 1000 person-years were 32 in knee OA, 40 in hip OA, 20 in hand OA and 21 in the general population. For most causes of death and joints with OA, we found no evidence of increased mortality, with HRs close to 1. (Table 2) For knee and hip OA and death from cardiovascular disease, the hazards were non proportional. HRs for death from cardiovascular disease were below 1 during the first 4 years of follow-up. However, during 5 to 8 years of follow-up the HRs (95% confidence interval) were 0.97 (0.89, 1.03) for knee OA and 1.11 (1.02, 1.20) for hip OA, and during 9 to 11 years of follow-up 1.19 (1.10, 1.28) and 1.13 (1.03, 1.24) for persons with knee and hip OA, respectively. We also found some evidence of increased mortality from diabetes in persons with knee OA (HR 1.21 [1.01, 1.44]). However, this association was partially explained by confounding from type I diabetes as registered in SHR (HR additionally adjusted for type I diabetes 1.08 [0.90, 1.29]). Conclusions: The risk of death from cardiovascular diseases increases with duration of knee and hip OA, but not hand OA. Elevated mortality from diabetes, partly explained by type I diabetes, reinforces the need for better understanding of the association between diabetes and OA.