BACKGROUND: Cases of notifiable conditions reported to CDC are classified by MMWR week. CDC recommends that jurisdictions assign MMWR week according to the variable “event date.” However, jurisdictions may vary in how they assign “event date” (e.g., by diagnosis date or date of report to state/local health department). If MMWR week is derived from a date distant from date of infection, MMWR week may be a suboptimal measure of incidence. We investigated how MMWR week might affect reporting using reported cases of chlamydia, a sexually transmitted infection.
METHODS: We reviewed chlamydia cases assigned to MMWR weeks in 2013–2014 in thirteen jurisdictions that had 1) non-missing values for specimen collection date and date of first report to CDC for ≥90% of cases and 2) 100% concordance between assigned MMWR week and “event date” variable. We used “event date type” variable to classify the type of date used to assign MMWR week. We used the specimen collection date (a separate variable that is not included in “event date type”) as closest indication of incidence date. We compared observed MMWR weeks based on “event date” to hypothetical MMWR weeks based on date of specimen collection to assess differential classification.
RESULTS: Of 703,265 total chlamydia cases, 73.7% were assigned to an MMWR week using the date of first report to the state or local health department. Another 23.9% cases were assigned using the lab result date, 1.9% cases were assigned using diagnosis date, 0.1% cases were assigned using symptom onset date, and 0.3% cases had unknown or invalid event dates. When we instead used specimen collection date to assign MMWR week, 0.7% cases were classified in a different year. We observed differences in classification of ≥1 week in 59.5% cases and ≥4 weeks in 9.0% cases when comparing the observed event date to the specimen collection date.
CONCLUSIONS: A majority of chlamydia cases reported by these jurisdictions were based on date reported to the state or local health department, which may be a poor proxy for incidence. Although a small proportion of cases were classified in a different year when using specimen collection date as the gold standard, over half were classified in different weeks, which may affect analyses involving more granular time periods such as aberration detection or quarterly reporting. To standardize reporting, national guidance should be provided and date types representing a closer approximation to disease incidence should be prioritized.