Using linked clinical and hospital morbidity data to assess risk and outcomes of primary lower limb total joint replacement in elderly men.

Abstract

Background: Osteoarthritis is the most common musculoskeletal disorder affecting elderly Australians and is a leading cause of lower limb total joint replacement (TJR). The incidence of TJR has risen substantially over the past two decades, reflecting the ageing population, and increases in the prevalence of risk factors such as obesity. Primary TJR is considered to be relatively safe with low rates of adverse outcomes, however, there is increasing evidence that elderly, and male patients who undergo the procedure may be at higher risk for postoperative complications and mortality. The retrospective cohort studies presented in this thesis used data, drawn from Health In Men Study (HIMS), that were linked with Western Australia (WA) linked data system to assess risk and outcomes of primary TJR in a large population-based cohort of men. The studies closely examined three issues - obesity, co-morbidities, and smoking - about which there is continuing debate in regard to their association with the risk of undergoing the procedure, and their roles as determinants of outcome of TJR. These risk factors are particularly important because they are amenable to modification. Objectives: The main objectives of this thesis were: 1. To validate WA hospital morbidity data (HMD) and to assess the performance of HMD-based co-morbidity adjustment methods in predicting mortality among men undergoing elective primary TJR. 2. To assess risk of undergoing elective primary TJR in elderly men. 3. To assess risk of adverse outcomes following elective primary TJR including: • in-hospital complications, • prolonged length of stay in hospital (LOS), • all-cause readmission, and • short- and long-term mortality. 4. To assess the role of obesity in predicting postoperative complications following TJR. Methods: The electronic records of 12,203 men from HIMS were linked with WA HMD, Cancer Registry, Mental Health Services System and mortality records. Linkage with hospital morbidity data was done to identify TJR, in-hospital complications, LOS, and readmission in the target population. Significant morbidity was retrieved from HMD in the period 1970-2007. Multivariable analyses including logistic, Cox proportional hazards, and competing risk regressions were undertaken to assess study outcomes. Main findings: • WA HMD are more likely to identify major co-morbidities and major operations with relatively high sensitivities and positive predictive values than co-morbidities of a less serious nature. • Co-morbidity as recorded in HMD, irrespective of method used to measure it, independently increased risk of adverse outcomes. Model discrimination of 5-year mortality following TJR improved by 13% when HMD-based Deyo-Charlson index (Deyo-CI) was added to the baseline model that only accounted for age (Harrell's C: 0.69 for baseline model vs. 0.78 for model including age and Deyo-CI). • A dose-response relationship between both weight and smoking, and risk of TJR was observed. Being overweight independently increased the risk, while smoking lowered it. Engaging in vigorous exercise and having a high socioeconomic status were associated with higher risk of TJR. • Of the 819 men who had had elective TJR, 331 (40.4%) developed an in-hospital complication of which 155 were major. Weight, co-morbidity, and minor complications independently predicted major complications. Any in-hospital complications significantly increased risk of short- and long-term mortality. Similarly, weight was independently associated with 5-year mortality following TJR. • Length of stay in hospital was significantly longer in the overweight or obese and those who had had a total knee replacement [TKR] (compared with total hip replacement [THR]) and these two groups were more likely to be readmitted. All-cause readmission was also significantly high among the socioeconomically disadvantaged patients. • All-cause 90-day and 1-year readmission following TJR independently increased risk of postoperative 5-year mortality. • Augmenting HMD with actual weight and height significantly improved the model fit when predicting major in-hospital complications following TJR. Conclusions:  HMD-based co-morbidity adjustment methods (Deyo-Charlson, Enhanced-Charlson or Elixhauser) significantly improve HMD-based predictive models and are appropriate in epidemiological research.  Compared to men with normal weight, the obese are at higher risk of undergoing elective TJR and are more likely to develop major complications, stay longer in hospital and be readmitted following the procedure.  Adding minimal information to routinely collected HMD improves the latter's predictive ability. This study suggests making actual weight and height mandatory variables in any HMD system.