Calculating Injury and Poisoning Estimates
1997-present
Questions in the Injury and Poisoning section of the NHIS have a recall period of the “last 3 months.” To estimate the annual number of injury or poisoning episodes using episodes with three months or less elapsing between the episode and the date the injury/poisoning questions were asked, each weighted count should be multiplied by 4. Annual estimates of the number of people injured or poisoned cannot be produced from these data due to the limited 3-month reference period. Estimating the number of persons injured or poisoned using the annualizing method described above (i.e., multiplying the estimate by the number of time periods in a year) would assume that the same individuals experienced injuries or poisonings at the same rate over the year.
Research by Warner et al. (3) recommends that, beginning with 2004 NHIS data, analysts use a 5-week reference period to calculate annual estimates of injuries and poisonings even though the questions still use a recall period of three months. To calculate an annual estimate of the number of injuries and poisonings using this reference period, the weighted number of episodes with five weeks or less between the episode and the date the injury/poisoning questions were asked should be multiplied by 10.4. Using the 5-week reference period has been made relatively simple due to the imputation of non-complete dates (in days) of injuries or poisonings beginning with the 2004 data. If analysts wish to use the 5-week reference period prior to 2004, they must impute non-complete dates themselves.
To increase the number of observations or respondents for the same number of variables, and thus increase the precision of estimates, analysts may want to combine data from successive years of the NHIS when the questions remain essentially the same over the years being combined. As mentioned in previous sections, the injury and poisoning questions have changed over the years between 1997 and the present. As a result, analysts should carefully examine the injury and poisoning questions for the years they want to combine. In general though, the questions remained the same for years 1997-1999, 2000-2003, and 2004 to the present. For more detailed information about merging multiple years of data, see the Survey Description Documents for the years of interest. Links to these documents can be found in the Guide for Data Users.
1963-1996
Background
From 1963 to 1996, injuries occurring in the two weeks prior to the interview were used to estimate the annualized number of injuries occurring to the U.S. population, the kinds of injuries, and where they happened. In cases where more than one injury occurred in one episode, one injury would be identified as the “first”; single injuries were “first” injuries by default. When injuries were classified as “first” injuries, the data were used in NCHS reports and by analysts to estimate the annualized number of injury episodes occurring in the U.S. The same data were also used to calculate quarterly numbers and rates of injuries using only the records for a calendar quarter instead of all records.
The disability days occurring in the previous two weeks that were associated with injuries (regardless of the date of the injury) were used to estimate bed days, work and school loss days, etc. due to injuries. Similarly, hospitalizations in the past six months linked to injuries were used to estimate injury-related hospitalizations until 1982 when the NHIS stopped asking the conditions related to hospitalizations. In the case of disability days and hospitalizations, those impairments caused by injuries were also included in the estimates of days.
Calculating Estimates
To calculate the annualized number of injuries, all condition records for those injuries that occurred in the two weeks prior to the interview were used; this included both the “acute” injuries and those few chronic impairments caused by injuries that occurred in this time period. The weights used to annualize these data were the final basic annual (person) weight on the condition record and the “6.5” weight, which is a weighting factor that corrects for the number of two week periods in a quarter. The denominator used for calculating rates was the person record using the final basic annual (person) weight. Population estimates must be calculated using the person record to account for everyone in the population.
To calculate the quarterly number of injuries, all condition records in a given quarter for those injuries that occurred in the two weeks prior to the interview were used; this included both the “acute” injuries and those few chronic impairments caused by injuries that occurred in this time period. The weights used to annualize these data were the final basic quarterly (person) weight on the condition record and the “6.5” weight, a weighting factor that corrected for the number of two week periods in a quarter.
To calculate disability days, all condition records identified as injuries (including the impairments caused by injuries) and not necessarily occurring in the past two weeks but which had associated disability days that did occur in the previous two weeks were used. The number of days on these records multiplied by the final basic annual weight and the “6.5” weight were used to calculate the estimates. The denominator used for calculating rates was the person record using the final basic annual (person) weight. Population estimates must be calculated using the person record to account for everyone in the population. Quarterly estimates of days were calculated as described above.