NEISS - National Electronic Injury Surveillance System
NationalElectronicInjurySurveillance Tom Schroeder 2/13System *This presentation was prepared by CPSC staff. It has not been reviewed or approved by, and may not necessarily reflect the views of, the Commission. .
Product-Related Injuries Consumer product injuries = ~ 38 million Treated in Treated in ambulatorphysicians‟ y offices care clinics Consumer product injuries treated in hospital EDs = ~ 14 million
Overview/History National sample of 96 hospitals from all U.S. hospitals with at least 6 inpatient beds and 24-hour emergency service. Each hospital reports information on emergency treatments to CPSC. Hospital coder enters data in local PC and transmits the data to CPSC over the internet. System collects ~ 400,000 product-related injury reports each year. – (additional ~ 300,000 non-CPSC injury reports each year). Multi-level system. Supports CPSC and other federal agencies.
Overview/History Has been in use for over 30 years. Last updated with new hospital sample in 1997. Data available from 1980-current through FOIA request. – 2011 data was finalized March 31, 2012. Internet – On-line estimates from 1991-2011. – Data available for download from 2002-2011.
Why Emergency Department Data? Large numbers of injuries are treated there. Data are already being captured in ED record on these cases. Hospitals have been very cooperative in supporting CPSC data collection activities. The information is timely.
NEISS Surveillance Variables Treatment date Products involved Case number (2) Age Locale Sex Type/work-related Injury diagnosis Fire dept. Body part injured involvement Disposition from Intent ED Race / ethnicity Narrative (2 lines)
Statistical Basis NEISS sample design: http://www.cpsc.gov/en/Research-- Statistics/NEISS-Injury-Data/ Follow-back investigations can be conducted within the statistical framework.
Multi-level Data Collection Capability Routine emergency department data Emergency department special study - “2nd screens” Telephone investigations On-site investigations
Analyzing NEISS On-Line Data Just Curious • Use on-line query • Read some cases • Historical national estimate
Analyzing NEISS On-Line Data Quick Analysis • Will fulfill most people‟s needs • Use minimal query to subset the data • Download the data! • Read the narratives! • Manipulate the data in a spreadsheet • form subgroups • pivot tables • always use the „weight‟ variable (sum = national estimate)
Analyzing NEISS On-Line Data Heavy Duty Statistical Analysis • Download the data! • Read the narratives! • SAS – variable formats and general programs on web-site • Use PSU, Stratum, Weight! • Proc Surveymeans • National estimates with confidence intervals • Test for statistically significant differences • Trend analysis across multiple years
Analyzing NEISS On-Line Caution Data • Always use ‘weight’ variable for national estimates! • Small sample sizes and estimates! • Consider multiple product codes • Other/Not Specfied • Unknown (locale) • Denominator Data • Census population? age adjust? • units sold? market share? • participation rates?
Analyzing NEISS On-Line Data What NEISS On-Line Does Well • Trend analysis across time • must take into consideration sample change in 1997 • Demographics of injured people (age, gender, body part, diagnosis) • Product groupings
Analyzing NEISS On-Line Data What NEISS On-Line Doesn‟t Do Well • Deaths • Non-emergency department treatment • pediatrician, school nurse, athletic trainer, etc. • Specific manufacturer • all PII removed from NEISS narratives • Detailed product information (use telephone interviews) • size and type of television • engine size of ATV • age of baby walker, crib
CPSC Stands For Safety Contact Information firstname.lastname@example.org Tom Schroeder Statistician, Division DirectorDivision of Hazard and Injury Data SystemsU.S. Consumer Product Safety Commission 4330 East West Hwy Bethesda, MD 20814