AHRQ unveils inpatient pediatric database

Covers 2,500 hospitals and nearly 2 million stays

The U.S. Agency for Healthcare Research and Quality (AHRQ), in Rockville, MD, has introduced the nation’s first database on the hospital inpatient care of America’s children. The Kids’ Inpatient Database (KID) was developed to make national and regional estimates of children’s treatment, including surgery and other procedures, and for estimating treatment outcomes and hospital charges.

The database, drawn from about 1.9 million children’s hospital inpatient stays at more than 2,500 hospitals across the United States in 1997, includes pediatric patients from birth through age 18. The large sample allows the analysis of hospital care and charges for common conditions in children, such as respiratory diseases and injuries, as well as rare conditions, such as congenital abnormalities. It also enhances researchers’ ability to study infrequently used procedures, such as bone marrow biopsy and organ transplantation.

"The decision to assemble this database was made about 18 months ago," says Anne Elixhauser, PhD, senior research scientist with AHRQ and the project officer. It was assembled as part of the Health Care (Cost) and Utilization Project (HCUP), a partnership between the federal government, state data organizations, and hospital organizations. Each state was responsible for assembling an all-payer discharge database from hospitals, including Medicare, Medicaid, privately insured, and uninsured patients.

"In the year 1997, we had 22 participating states; the KID program was completely voluntary," Elixhauser explains. "We then transformed all the data into a uniform format, since every state has a different way of doing things. This way, they can be used for cross-state analysis." The current information represents about 60% of all pediatric hospital discharges in the United States. Elixhauser says she hopes that for the 2000 data year, she will have 80% of all discharges.

More specific information needed

For many years now, a database has existed called the Nationwide Inpatient Sample, or NIS. "It’s a huge sample of 1,000 hospitals and 7 million discharges every year," notes Elixhauser. However, she says, AHRQ realized this database had its limitations. "We couldn’t look at specific subpopulations in the detail we wanted, and children were one prime example," she explains. "If we wanted to look at something very rare, like certain childhood cancers, we could not do that using NIS. We realized, however, that we were able to go back to the state data and draw samples just of children. This was our first pilot project to see how it could work."

The database information is available in two formats: HCUPnet and KID. HCUPnet is available free of charge on the AHRQ web site: www.ahrq.gov. (Select HCUPnet, followed by "Start HCUP net," and then click on "Children’s Hospital Stays Only.") The KID database, which includes all of the information assembled, comes in ASCII format, and is readily translatable into many applications. The purchase price is $220 for a year’s worth of data, but its use also requires statistical analysis software like SAS, SPSS, or STATA.

Although it is more limited than the KID database, there are a number of ways in which quality managers can make use of the free HCUP net, says Elixhauser. "For example, you can obtain simple descriptive statistics, like what the average charge is for hospitals to treat congenital cardiac anomalies, or how many kids had specific types of cancer, or what percentage of kids are uninsured."

"You can search for statistics that revolve around diagnoses or procedures. You can even break it down by children’s characteristics, like age groups or gender, or compare different types of hospitals. For example, you could look at the length of stay for certain conditions or charges or mortality rates." In addition, she says, you could research admissions for certain conditions, or identify cases with complication of procedures via ICD-9 codes. "It is primarily valuable in getting overall statistics," she adds. "You could benchmark outcomes, such as mortality rates for specific conditions, or charges for specific procedures." The advantage of the KID database, she explains, is that it allows the ability to conduct more in-depth research and perform risk-adjustment activities.

Later this fall, the database will be expanded to include information on HCUP quality indicators. "We’ve got a contract with Stanford to develop indicators based on discharge data," Elixhauser reports. "They have given us the first couple of modules — mortality, volume of procedures, and utilization of procedures (appropriateness), including ambulatory care sensitive conditions. These quality indicators will provide additional benchmarks for quality managers."

Need more information?

For more information, contact: Anne Elixhauser, PhD, Senior Research Scientist, Agency for Healthcare Research and Quality, Rockville, MD. Telephone: (301) 594-6815. E-mail: hcup@ahrq.gov.