BACKGROUND: The purpose of the present paper was (i) to identify trends in in-hospital mortality after transurethral resection of the prostate (TURP) in Victorian public hospitals; and (ii) to explore associations between in-hospital mortality after TURP and age, adverse events, type of admission (emergency/planned), location of the hospital (metropolitan/rural), teaching status of the hospital and length of stay. METHODS: Trends in in-hospital mortality after TURP and the associations between in-hospital mortality and the aforementioned variables were studied using International Classification of Diseases, 9th revision, Clinical Modification (ICD-9-CM) coded Victorian hospital morbidity data from public hospitals between 1987-88 and 1994-95. Crude and adjusted odds ratios (OR) and 95% confidence intervals (CI) were based on univariate and multivariate logistic regression, respectively. RESULTS: After adjustment for age, comorbidity, and other confounding variables, the trend in mortality reduction over time was highly significant (P for trend < 0.0001, 95% CI for trend: 0.84-0.95). Highly significant associations with mortality were observed for emergency admissions (OR = 1.99, P < 0.0001), presence of adverse events (OR = 2.69, P < 0.0001), length of hospital stay (P for trend < 0.0001, 95% for trend: 1.88-2.15) and age (P for trend < 0.0001; 95% CI for trend: 1.26-1.48). CONCLUSIONS: Routinely collected data from hospitals can provide tentative evidence of improved effectiveness of a surgical treatment, provided analysis takes careful account of potential sources of bias, especially those related to possible changes in case selection over time. These kinds of data should stimulate a joint effort between clinicians, quality assurance experts and epidemiologists to confirm this attribution, and to locate the causative factors.
BACKGROUND: The purpose of the present paper was (i) to identify trends in in-hospital mortality after transurethral resection of the prostate (TURP) in Victorian public hospitals; and (ii) to explore associations between in-hospital mortality after TURP and age, adverse events, type of admission (emergency/planned), location of the hospital (metropolitan/rural), teaching status of the hospital and length of stay. METHODS: Trends in in-hospital mortality after TURP and the associations between in-hospital mortality and the aforementioned variables were studied using International Classification of Diseases, 9th revision, Clinical Modification (ICD-9-CM) coded Victorian hospital morbidity data from public hospitals between 1987-88 and 1994-95. Crude and adjusted odds ratios (OR) and 95% confidence intervals (CI) were based on univariate and multivariate logistic regression, respectively. RESULTS: After adjustment for age, comorbidity, and other confounding variables, the trend in mortality reduction over time was highly significant (P for trend < 0.0001, 95% CI for trend: 0.84-0.95). Highly significant associations with mortality were observed for emergency admissions (OR = 1.99, P < 0.0001), presence of adverse events (OR = 2.69, P < 0.0001), length of hospital stay (P for trend < 0.0001, 95% for trend: 1.88-2.15) and age (P for trend < 0.0001; 95% CI for trend: 1.26-1.48). CONCLUSIONS: Routinely collected data from hospitals can provide tentative evidence of improved effectiveness of a surgical treatment, provided analysis takes careful account of potential sources of bias, especially those related to possible changes in case selection over time. These kinds of data should stimulate a joint effort between clinicians, quality assurance experts and epidemiologists to confirm this attribution, and to locate the causative factors.