|
The WISQARS estimates are based on trained coding of visits at a sample of ER across the country; this raw data is then used to produce national estimates
The estimates, of course, are not particularly convincing when sample sizes are small, and querying the WISQARS database will produce red asterisks beside estimates supported by relatively little data
If, for example, you query WISQARS for unintentional nonfatal gunshot injuries, with estimates separated by year and age group, you will find most estimates flagged with red asterisks; if you query only by year, the bins are larger, and the estimates more reliable, so red asterisks do not appear. So we can really draw absolutely no conclusions whatsoever about trends in unintentional nonfatal gunshot injuries to (say) toddlers (0 - 4) from the data
Suppressing age groups, and examining only the data by year (2001 - 2008), will eliminate the red asterisks. But the estimates fluctuate year by year, with rather large 95% confidence intervals: in fact, the estimate for any one given year 2001 - 2008 falls within the confidence interval of every other year 2001 - 2008
So the WISQARS estimates for unintentional nonfatal gunshot injuries don't support conclusions about trends in recent years
|