advertisement
Abstract #24651 Published in IGR 11-4
How and why automated medical databases should be used to create background occurrences for drug outcomes and safety studies: Glaucoma in thin (UK)
Maguire A; Blak BTValue in Health 2009; 12: A489
OBJECTIVES: When estimating expected occurrences of events in patient populations, background rates from published literature are often applied thereby creating comparative statistics such as the Standardized Incidence Ratio. However, in order to adequately estimate the occurrence of events the probabilities of the events have to be available according to the main risk factors for the event. Also the patient population needs to have the same granularity in order to marry the risk factor specific rates to the patient characteristics. Through the example of safety outcomes for glaucoma therapy, we describe an approach to create appropriate background occurrences for the outcomes of heart failure (HF), asthma/COPD exacerbation and death. METHODS: A cross section of glaucoma patients were identified on January 1, 2007 from The Health Improvement Network nullTHINnull, a UK primary care database widely used for medical research. As well as age and sex, patients were categorized according to risk factors for HF (Hx HF, hypertension) and asthma/COPD (Hx asthma/COPD, smoking and diabetes). General population rates of these for the same categories events were calculated from the three million people registered with THIN on January 1, 2006. These rates were assigned to each glaucoma patient and the expected number of events over one year for 10,000 patients was calculated. UK national statistics provided the death rates. RESULTS: The median age of the 31,535 glaucoma patients was 74 years and 46% were male. The prevalence of HF was 4%, asthma 13%, COPD 6% and smoking 12%. For 10,000 patients we can expect 64 HF events, 139 asthma/COPD exacerbations and 524 deaths each year. Due to different patient characteristics, the expected occurrences were lower in timolol treated patients as opposed to latanoprost users. CONCLUSIONS: The use of databases provides more appropriate and granular background rates than from published rates. The approach can be used in risk-benefit models.
A. Maguire. United BioSource Corporation, London, LondonUnited Kingdom.
Classification:
14 Costing studies; pharmacoeconomics