National and Subnational Patterns of Cause of Death in Iran 1990-2015: Applied Methods.

BACKGROUND Causes of death statistics provide crucial health intelligence in national and international communities. An efficient death registration system provides reliable information for health policy system. In many developing countries, death registration systems face a degree of misclassification and incompleteness. There are many impediments to putting an estimate of cause-specific death rates. Addressing those challenges could prevent misleading results. METHODS Our data was collected by Ministry of Health and Medical Education, Tehran and Isfahan cemeteries from 1995 to 2010. After converting ICD codes of Iran's death registration into GBD codes, 170 underlying causes of deaths were recognized in the available data. A wide range of methods were applied for preparing the data. We used several statistical models to estimate mortality rates in age-sex-province groups for all causes of deaths. The considerable number of combinations for age, sex, cause of death, year, and province variables made further complicated model selection and evaluation of the results. RESULTS Totally, 58.91% of deaths were related to males. The majority of cases of death were classified as NCDs (77.83%) and injuries (14.80%). We extrapolated 71.76% and 14.71% of causes of death by mixed effect model, spline model with parameter 0.9 and 0.6, respectively. CONCLUSION A comprehensive and unique registration system is able to solve many DRS issues. It is necessary to assess the quality and validity of cause of death data. Scientific methods like analyzing mortality level and cause-of-death data are used to provide an overview for better decisions.