Natural causes: Do physicians code for uncertainty In cause of death?



Journal Title

Journal ISSN

Volume Title



Cause of death (COD) determination by physicians on the death certificate can be difficult when the conditions leading to death are unclear or uncertain. The Center for Disease Control (CDC) recommends the classification for these deaths should be “undetermined natural causes” (ICD-10 code R99). Physicians may incorrectly code COD in patients who die of undetermined natural causes, instead listing heart disease (ICD-10 I00-I09, I11, I13, I20-I51)), or another COD, on the death certificate. This may have important implications in the calculation of mortality statistics, particularly when using the Mortality Medical Data System (MMDS), a system to automate the entry, classification, and retrieval of cause-of-death information reported on death certificates. \r\n \r\nForty-two primary care physicians participated in a randomized, controlled trial assessing completion of the death certificate for a patient dying of undetermined natural causes. The physicians were randomized into two groups: the first group reported the COD using an open-ended format, and the second group selected COD from a list of potential choices (including “undetermined natural causes”) for the death certificate COD fields. The study has two aims. The first aim is to determine the percentage of physicians that correctly code the death certificate, and to see if responses are sensitive to having “non-specific” COD as a possible response. The second aim is to assess the reliability of coding COD by comparing physician reported COD to the MMDS transformations of those reports.\r\n\r\nOverall, 62.5% of respondents chose an underlying COD that was categorized as “symptoms, signs and abnormal clinical and laboratory findings, not elsewhere classified,” (ICD-10 codes R00-R99), and the remaining 37.5% reported a specific COD, such as heart disease (I00-I09, I11, I13, I20-I51). There was no statistical difference [÷2 (1, N = 40) = 1.932, p = 0.165] in the frequency of nonspecific R codes used by the respondents for the underlying COD completing either the open-ended format (52% nonspecific, 48% specific) compared to the close-ended format (74% nonspecific, 26% specific). \r\n\r\nProcessing with the Mortality Medical Data System (MMDS) resulted in an underlying COD category of 27.5% non-specific R codes and 72.5% specific causes (60% were the heart disease category), nearly opposite the numbers for the original physician-reported underlying COD. The MMDS-generated COD did not show a statistically significant difference [÷2 (1, N = 40) = 2.489, p = 0.115] for frequency of R code generation between the open-ended format (38% nonspecific, 62% specific) and the close-ended format (16% nonspecific, 84% specific) .\r\n\r\nThis study suggests that while the majority of physicians may correctly code for an underlying COD using codes R00-R99 for a person dying of undetermined natural causes, software used for national statistical reporting may be biased to report the underlying COD category as heart disease (I00-I09, I11, I13, I20-I51).\r\n



MMDS, death certificate inaccuracies