Explanation Of NEISS Estimates Obtained Through The CPSC Web-site

Background

CPSC’s National Electronic Injury Surveillance System (NEISS) is a probability sample of hospitals in the U.S. and its territories that have at least six beds and an emergency department. The NEISS is a stratified sample based on emergency department size and geographic location. The emergency department size is categorized by the annual number of emergency department visits reported by each hospital. Currently, there is also a stratum of children’s hospitals. Patient information is collected nightly from each NEISS hospital for every patient treated in the emergency department for an injury associated with consumer products. National estimates are made of the total number of product-related injuries treated in U.S. hospital emergency departments based on the NEISS data collected from these hospitals. Estimates from NEISS are available through CPSC’s web-site for user specified products and/or other classifications of interest. When a web-site user requests estimates for specified products and/or groups of interest, the following will appear:

- Historical Estimate
- Adjusted Estimate
- Sample Count
- Variance (CV)

Data from each hospital are weighted based on the sample design. The historical estimate is simply the sum of the weights from the NEISS cases of interest. **These estimates may differ from those published by CPSC for a similar query, since CPSC analyses may take more factors and information into consideration.**

On January 1, 1997 a new NEISS sample was implemented. NEISS data were collected from hospitals in both the new and old NEISS samples during an 'overlap period' from January 1, 1997 - September 30, 1997. By comparing national estimates from the new sample to national estimates from the old sample during the overlap period, to some extent we can determine and adjust for the effects of the sample change on the estimates. Adjustment factors are calculated for years prior to 1997 and used when comparing estimates in different years from the two NEISS samples. The adjusted estimate is simply the adjustment factor for the year multiplied by the historical estimate. The adjusted estimates should always be used to compare results from 1997 forward with earlier time periods.

The sample count is the number of cases collected in the NEISS for the products and/or groups of interest that are used to calculate the estimate. **As for historical estimates, the number of cases may differ from those published by CPSC for a similar query, since CPSC analyses may take more factors and information into consideration.**

Because the NEISS is a sample of hospitals and not a census, the estimates from NEISS are indeed ‘estimates’ with an associated variability. Variances of estimates are calculated based on the NEISS sample design and given as a coefficient of variation (CV). The CV of an estimate is the square root of the variance divided by the estimate.

The estimate and CV are used to calculate confidence intervals around the estimate. For example an estimate of 100,000 with a CV of .10 yields a 95% confidence interval of:

The variance (CV) listed on the web-site is the variance of the historical estimate. The variance of the adjusted estimated is currently not available on the web-site. The user should use the variance given as the variance for both the historical and adjusted estimates.

CPSC considers a national estimate unstable and potentially unreliable when:

- the estimate is less than 1,200 or
- the number of records used is less than 20, or
- the CV exceeds 33%

Results from all on-line queries that do not meet these three stability requirements are not available and are listed as 'N/A'.