Improving Surveillance of Diabetes Deaths through EHR Linkage *

BACKGROUND:  

In 2006, the Utah Department of Health (UDOH) developed an Electronic Death Entry Network (EDEN) for submitting death certificates. Although state mortality databases are rich data sources for chronic disease surveillance, diabetes deaths are historically underreported.  UDOH and Intermountain Healthcare (IH), Utah’s largest healthcare delivery system, recognized the potential for improving reporting of diabetes deaths by linking EDEN to patient electronic health records (EHRs). An interface was developed that allowed physicians to complete their portions of death certificates within their patients’ EHR. An exploratory study was conducted to compare reporting between deaths that were certified using the EDEN/EHR interface and those using EDEN only. Key outcomes were: (1) number of ICD-10 codes on death certificates with diabetes as any listed cause; (2) timeliness of certifying the death certificate, and (3) sensitivity of reporting diabetes as listed in EHRs.

METHODS:  UDOH and IH examined death certificates with any listing of ICD-10 codes E10-E14 among Utah residents, aged 13 and over, that occurred between September 2013 and August 2014 (N=1,623). Information and timeliness for certificates submitted through the enhanced EDEN/EHR interface (N=417) were compared with certificates submitted through the EDEN only system (N=1,206). Wilcoxon and chi-square tests were used to identify statistically significant differences.

RESULTS:  

(1) The number of ICD-10 codes on death certificates was significantly higher for death certificates submitted using the EDEN/EHR interface than for the EDEN only system (median 5 vs.4; p<.0001); (2) The percentage of deaths certified within 72 hours was significantly higher in the EDEN/EHR system than in the EDEN only system (48.1% vs.79.1%;  p<.0001); (3) Certificates submitted through the EDEN/EHR system had a lower percentage listing diabetes as underlying cause of death (2.4% vs. 3.7%) but a higher percentage of diabetes listed as any cause (10.3% vs. 8.7%). Neither of these differences were statistically significant.

CONCLUSIONS:   Partnerships with private healthcare industries can produce innovative strategies that benefit public health. Submission of death certificates through an EHR interface was associated with an increased number of ICD-10 codes listed on death certificates and improved timeliness of certification. Findings suggest that the interface enhances reporting of diabetes as a contributing cause of death.  Additional research is needed to explain the lower percentage of diabetes as underlying cause in the EDEN/EHR system. This research will be pursued during Phase II of this project. Phase II results will be shared in the CSTE poster presentation, if selected.