Studying Injuries Using the National Hospital Discharge Survey

Studying Injuries Using the National Hospital Discharge Survey Marni Hall, Ph. D. Hospital Care Statistics Branch, Division of Health Care Statistics

Outline of this presentation • Present general information about the design of the National Hospital Discharge Survey (NHDS) • Discuss decisions that have to be made when designing injury research using NHDS • Highlight issues particularly related to trend analyses in

Upcoming Chartbook Trends in Injury Hospitalization, United States, 1979 -2001 by Melissa Heinen, Marni Hall, Manon Boudreault, and Lois Fingerhut

NHDS a national probability sample of shortstay non-federal hospitals - conducted every year since 1965 provides data on discharges or hospitalizations – not individuals 2002 data - now available 2003 data - available in the winter

NHDS Design Three stage design Geographic Units Hospitals Discharges

Data Collection Automated – 40% Manual – 60%

Patient Data • • • Sex Race Age Expected source of payment Discharge status – including deaths

Facility Characteristics • Geographic region • Bed size • Ownership

Medical Data Coded according to the International Classification of Diseases, 9 th Revision, Clinical Modification (ICD-9 -CM) • Diagnoses • Surgical and nonsurgical procedures

Additional Variables • • • Days of care Month of admission/discharge Weight DRG – diagnosis-related group Available since 2001 NHDS – Source of Admission – Type of Admission

2002 NHDS 445 hospitals and 327, 000 discharges were sampled Weighted number of discharges was 33. 7 million

Weights • Must use weighted data to obtain national estimates • Each record has a weight • Must calculate the sum the weights of the records – one line of programming

Estimates have standard errors A standard error is the sampling variability that occurs by chance because only a sample rather than the entire universe is surveyed

For more design information Plan and Operation of the National Hospital Discharge Survey: 1988 Redesign Vital and Health Statistics, 1(39). 2000 http: //www. cdc. gov/nchs/data/series/sr_01/ sr 01_039. pdf

Designing Injury Research Project • Determine what injury definition will be used – all injuries or selected injuries – what ICD-9 -CM codes • Decide how you will count injury patients • Select the data you will report • Evaluate whether and how external cause codes will be used

Injury definitions • Definition developed by injury experts, e. g. the State and Territorial Injury Prevention Directors Association (STIPDA) reported in Consensus recommendations for using hospital discharge data for injury surveillance, 2003 • Use existing categorization of codes – e. g. the Barell Matrix which defines injuries by type and body region in Injury Prevention, 8, 2002 • Injury and Poisoning Chapter of the ICD-9 CM – codes 800 to 999 – includes “true injuries” and “medical injuries”

Diagnoses selected for study should • Have a specific ICD-9 -CM code(s) • Be relatively common in hospitalized patients or you will have to combine data over multiple years in order to get reliable estimates

Reliability • To be reliable, estimates must be based on at least 30 records, and have a relative standard error of less than 30 percent - these usually produce weighted estimates of less than 5, 000 • Estimates based on 30 -59 records may be unreliable and should be used with caution – these usually produce weighted estimates from 5, 000 -9, 000

Counting injury patients If you want the number of patients hospitalized because of an injury – count the first-listed diagnoses If you want the number of hospitalized patients who have one or more injuries – count any-listed diagnoses

Counting injuries If you want the number of injuries and not the number of patients – count all-listed diagnoses

Hospital discharges with fractures, 2002 1, 609, 000 1, 387, 000 995, 000 Principal or first listed Any listed All listed

Injury chartbook includes the following data • • Injury discharges by age and sex Type and body region of injuries Average number of diagnoses Days of care/average length of stay Expected source of payment Discharge disposition Percent with, and types of, external cause codes

Choices involving external cause codes • Evaluate the percent of injury patients with external cause codes – how complete is it? • If you decide you will use external cause codes, will you use just the first code or all codes? • Consider using a previously developed categorization, e. g. - the External Cause of Injury Matrix

Challenges in studying trend data • • Coding changes over time Size and availability of data files Presenting the data Interpreting the data

Different versions of the International Classification of Diseases • 8 th revision used 1970 -78 • 9 th revision used 1979 -2004 – with addenda since 1986

Size and availability of NHDS data • Single year files can be downloaded from the NHDS website and unzipped using free software. These include DRG’s. • Multiple year files are on CD’s, rather than our website, due to their large size. They can be obtained by calling our office (301 -458 -4321). They do not include DRG’s.

Presenting the data Age adjustment - Eliminates the differences in observed rates that result from age differences in population composition over time

Presenting the data Use of the log scale - allows the presentation of estimates with a very wide range on the same graph and facilitates comparison of the percent change of estimates over time

Presenting the data Measures of change AAPC – Average annual percent change from 1979 -2001 APC – Average percent change for 19792001

Interpreting the results Health service system changes which contributed to the decrease in hospitalization overall and for injuries • • Reform of Medicare hospital payment Increased utilization review of hospital care Growth in managed care Expansion and coverage of ambulatory surgery

Interpreting the results Injury prevention activities which contributed to the decrease in hospitalization for injuries • Encouragement of the use of safety belts and helmets • Safer automobiles (air bags) • Safer roads • Improvements in home and workplace safety • Poison control centers

For more information: • Check our website www. cdc. gov/nchs/about/major/hdasd/nhds. htm • Phone: 301 -458 -4321 • Fax: 301 -458 -4032 • Email: NHDS@cdc. gov