State Syndromic Surveillance Systems

For syndromic data, if alerts are generated based on a statistical criterion, such as 2.5 standard deviations above the weekly mean, the following adjustments may be useful:

Rural areas are likely to generate a high number of alerts since counts (and means) will be low and just a very small number of excess cases will set off an alert. Therefore, setting a minimum number of registrations may be a necessary modification.

1. Use an adequate lookback period, such as 84 days, to ensure normal fluctuations do not generate an alert.
2. If possible, consider incorporating trends typically seen by day of the week. If weekend counts are normally higher for example, the system should compare a weekend day to past weekend days. Some clinics or hospital departments may close on weekends, and some EDs have reduced weekend/holiday staffing, while for other data sources, weekend counts typically spike; typical data patterns should be considered by system users even if algorithms adjust for them.
3. Lags in reporting from a given facility may be of concern. Specifying several days with non-zero registration counts before an alert can be generated can help avoid alerts generated after a “data dump”. Customary reporting lags may allow such adjustment, while the investigator should be aware of temporary or unusual lags.
4. Visits can be viewed by hospital location or patient location. Hospital location will track the occurrence of the acute event and may be more helpful for targeting responses with naloxone. Patient location can be an indicator of long-term burden in residents and may help with planning for medication for opiod use disorder (MOUD) or case management.

Note that when syndromic surveillance visits are labeled as an overdose or opioid overdose they should not be interpreted as absolute counts of confirmed events. The visit is labeled as such based on information provided in the chief complaint and diagnosis fields. Variations in data quality can have an impact on trends and counts. If less detail is provided in chief complaint or diagnosis text fields by certain hospitals, for example, fewer visits may be labeled as overdoses. In some cases, where not enough information is available to detect an opioid overdose specifically, a more general query, or all drug overdose, may be a better syndrome definition to monitor.