Methodology for Validating Completeness of Communicable Disease Reporting in Maricopa County, Arizona

BACKGROUND:   The Maricopa County Department of Public Health (MCDPH) communicable disease surveillance system is based on mandatory reporting of more than 80 diseases and conditions by healthcare providers and laboratories.  Hospitals/healthcare facilities rely on Infection Preventionists to report communicable diseases to public health.  Complete disease reporting is crucial to accurately measure and compare disease incidence rates, control spread of the disease, identify outbreaks, and assess disease burden. This data was evaluated to determine completeness of disease reporting of hospitalized or emergency department cases for selected morbidities using a novel, collaborative methodology.

METHODS:   We conducted a comprehensive study of reporting completeness using the Arizona Hospital Discharge Data (HDD) with an analysis of all communicable diseases required to be reported within 24 hours, within 1 day and select diseases required to be reported within 5 days. All patients among 48 reporting health care facilities in MC during 2009 who were assigned an International Classification of Diseases, 9th Revision, Clinical Modification diagnosis code (ICD-9 Code) for a state-required reportable communicable disease were matched to public health communicable disease reporting database (CDR database) using a SAS program. Sensitivity was calculated by dividing the number of matched cases in CDR by the total number of cases pulled from HDD. Unmatched cases were manually compared against the CDR database to determine if there was a matched case that was reported but missed by the program. Cases identified by this process were added to the list of matched cases. Additionally, medical record review of unmatched records was performed to identify cases that should have been reported.

RESULTS:   Estimated initial sensitivity of reporting among healthcare facilities was 43%. After implementation of manual matching and record review the sensitivity of reporting increased to 84%. Thirty-six percent of healthcare facilities showed 100% sensitivity; fifty-eight percent showed (50-99)% sensitivity; six percent showed (0-49)%  sensitivity. Disease-specific reporting sensitivity ranged from 0% to 100% for both 24 hour and within 1 day reportable diseases, as well as for the select 5 day reportable diseases with medians of 82%, and 71% respectively. Diseases of significant public health importance, which were not reported were addressed with the reporting healthcare providers.

CONCLUSIONS:   Validation of passive surveillance systems are essential to ensure completeness and validity of the data. Manual matching and medical record review enhanced the automated validation using a SAS program and improved accuracy of evaluation.