Evaluation Of The Arkansas Department Of Health Pertussis Surveillance System

BACKGROUND:   Pertussis is a vaccine-preventable bacterial respiratory illness caused by the agent Bordetella pertussis. The Arkansas Department of Health (ADH) receives pertussis reports from several sources, including laboratories, physicians, and schools. ADH investigates suspect cases, identifies contacts, facilitates prophylaxis, and determines necessity of exclusion from school or work. Additionally, ADH notifies the Centers for Disease Control and Prevention (CDC) of confirmed and probable cases using the NEDSS (National Electronic Disease Surveillance System) Base System (NBS). There is considerable year-to-year variation in pertussis cases in Arkansas, but approximately 200 cases per year are identified.

METHODS:   Using the 2001 CDC Updated Guidelines for Evaluating Public Health Surveillance Systems, the pertussis surveillance system was evaluated for simplicity, acceptability, flexibility, data quality, and timeliness. Sensitivity was calculated by matching available reports in NBS from 2009-2011 to hospital discharge data from 2009-2011.  

RESULTS:   Pertussis surveillance is highly complex, as the case definition has multiple components, including several laboratory tests that must be paired with clinical symptoms to determine case status. Furthermore, case ascertainment requires passage through multiple levels of reporting. System flexibility and acceptability are excellent, as ADH receives multiple types of case reports and several reporting sources regularly participate in surveillance. Although many records are missing information, key variables, such as case status, symptoms, and laboratory results are available for the majority of investigations. Although a subset of cases were systematically misclassified in NBS from 2009-2011, there was excellent overall agreement (kappa coefficient=0.82) between case status assigned in NBS and case status calculated using an algorithm based on symptoms and laboratory results. The median amount of time from initial notification of cases to completion of case investigations decreased from 69 to 28 days from 2009 to 2011. The surveillance system detected only 52% of hospitalized persons diagnosed with pertussis.

CONCLUSIONS:   There is significant underreporting of pertussis in Arkansas. The widespread use of inappropriate laboratory tests not included in the case definition likely contributes to the false negative designation of many cases. Additionally, many cases may be misclassified due to the operational complexity and stringency of the case definition. Overall, data quality is fair. Sensitivity in detection of hospitalized cases is low, likely due to the passive design of the surveillance system and underreporting by hospitals. Since NBS was introduced in 2009, ADH has been successful in implementing mechanisms to improve the timeliness of investigations. Evaluation findings will be used to guide improvements in pertussis surveillance in Arkansas.