Can we use mortality as proxy of prevalence? －an attempt to find out different disease iceberg shape

• Main Authors: Kuo Ting-Yueh, 郭庭籥
• Other Authors: Meng-Chih Lee, MD, Ph.D.
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/99048582649464604754

碩士 === 中山醫學大學 === 醫學研究所 === 92 === Background: To obtain original figures for prevalence data is very time and money consuming. Many scholars have attempted of using mortality data, which is less expensive and routine available, as an proxy of prevalence rate. Nevertheless, it was not without problem to use morality data as the proxy of prevalence rate. Objective: To determine what different kinds of diseases in which mortality data was more suitable as the proxy of prevalence rate. For those not a good proxy of prevalence rate we attempted to determine the relationships among number of cases with specific diagnosis according to 1) main discharge diagnosis (MD) in last hospitalization before death; 2) all discharge diagnoses (AD) in last hospitalization before death; 3) underlying cause of death (UCD) on death certificate; 4) multiple cause of death (MCD) on death certificate. Methods: All Death certificates issued in Chung Shan Medical University Hospital from January 1, 1998 through December 31, 2000 were extracted for this study. Using identification number we linked each death to the electronic hospitalization records. We first calculated the proportions in which the denominator was the number of MD with one specific diagnosis and the nominator was the number of MCD or UCD with one specific diagnosis in which it was appeared in the death certificate. We then calculated the ratios between number of MCD and AD with specific diagnosis with number of UCD with the specific diagnosis. The proportions and ratios were analysis by age and sex. Results: Diagnoses with high proportion between number of MCD and UCD with number of MD included: Leukemia, Stomach cancer, Lung cancer, Breast cancer, Cervical cancer, Aortic aneurysm, Myocardial disease. Diagnoses with low proportion were Fracture neck of femur, Peptic ulcer, Intracranial injury, Anemia, Parkinson’s disease, Hypertension. Diagnoses with intermediate proportions were Nephritis, nephrosis and nephrotic syndrome, Pneumonia, Asthma, Tuberculosis, Cirrhosis of liver, Stroke, Diabetes, Ischaemic heart disease. The results were similar if we stratified the analyses by sex and age. Diagnoses with high ratios between number of MCD and AD with number of UCD included: Diabetes, Peptic ulcer, Pneumonia, Hypertension, Nephritis, nephrosis and nephrotic syndrome. Diagnoses with ratios near one were Leukemia, Stomach cancer, Lung cancer, Aortic aneurysm, Myocardial disease. The ratios in the intermediate included Asthma, Tuberculosis, Cirrhosis of liver, Stroke, Ischaemic heart disease. The results were the same when stratified by sex and age. Conclusion: According to the above findings we concluded that 1) diagnoses like Leukemia, Stomach cancer, Lung cancer, Breast cancer, Cervical cancer, Aortic aneurysm, Myocardial infarction were highly appropriate to use number of UCD in the death certificate as the proxy of prevalence. 2) Diagnoses such as Diabetes, Pneumonia, Nephritis, nephrosis and nephrotic syndrome, Asthma, Tuberculosis, Cirrhosis of liver, Stroke, Ischaemic heart disease were moderate appropriate in using number of UCD in death certificate as the proxy of prevalence. 3) Diagnoses like Peptic ulcer and Hypertension were very inappropriate to use number of UCD in death certificate as the proxy of prevalence.