Invasive Species Compendium

Detailed coverage of invasive species threatening livelihoods and the environment worldwide

Abstract

Impacts of misclassification on Lyme disease surveillance.

Abstract

In Maryland, Lyme disease (LD) is the most widely reported tickborne disease. All laboratories and healthcare providers are required to report LD cases to the local health department. Given the large volume of LD reports, the nuances of diagnosing and reporting LD, and the effort required for investigations by local health department staff, surveillance for LD is burdensome and subject to underreporting. To determine the degree to which misclassification occurs in Maryland, we reviewed medical records for a sample of LD reports from 2009. We characterized what proportion of suspected and "not a case" reports could be reclassified as confirmed or probable once additional information was obtained from medical record review, explored the reasons for misclassification, and determined multipliers for a more accurate number of LD cases. We reviewed medical records for reports originally classified as suspected (n=44) and "not a case" (n=92). Of these 136 records, 31 (23%) suspected cases and "not a case" reports were reclassified. We calculated multipliers and applied them to the case counts from 2009, and estimate an additional 269 confirmed and probable cases, a 13.3% increase. Reasons for misclassification fell into three general categories: lack of clinical or diagnostic information from the provider; surveillance process errors; and incomplete information provided on laboratory reports. These multipliers can be used to calculate a better approximation of the true number of LD cases in Maryland, but these multipliers only account for underreporting due to misclassification, and do not account for cases that are not reported at all (e.g., LD diagnoses based on erythema migrans alone that are not reported) or for cases that are not investigated. Knowing that misclassification of cases occurs during the existing LD surveillance process underscores the complexities of LD surveillance, which further reinforces the need to find alternative approaches to LD surveillance.