DESCRIPTION OF THE NATIONAL HOSPITAL DISCHARGE SURVEY Introduction This document and its appendixes allow one to use the l990 data file which contains the data collected by the National Hospital Discharge Survey conducted by the National Center for Health Statistics. The National Hospital Discharge Survey provides an annual sample of hospital discharge records, collecting medical and demographic information for calculating statistics on hospital utilization. The survey consists of data from sampled inpatients discharged from a national sample of non- Federal short-stay hospitals. For a description of the survey design and data collection procedures, see below. For a more detailed description of the survey design, data collection procedures, and the estimation process, see Reference l. Publications based on the data for each survey year can be obtained from the Government Printing Office. History To provide more complete and precise information on the utilization of the Nation's hospitals and on the nature and treatment of illness among the hospitalized population, the NCHS in l962 began exploring possibilities for surveying morbidity in hospitals. A national advisory group was established. The NCHS conducted planning discussions with other elements of the Public Health Service. Hospitalization material from the Survey Research Center of the University of Michigan, the American Hospital Association, and the Professional Activities Study was examined and evaluated. In l963, a study by the School of Public Health of the University of Pittsburgh under contract to the NCHS demonstrated the feasibility of an NHDS type of program. An additional pilot study using enumerators from the Bureau of the Census was conducted in late l964 and confirmed the University of Pittsburgh's findings. Finally, with advice and support from the American Hospital Association, the American Medical Association, individual experts, other professional groups, and elements of the U.S. Public Health Service, the NCHS initiated the National Hospital Discharge Survey in l964. SURVEY METHODOLOGY Source of the Data The National Hospital Discharge Survey (NHDS) covers discharges from noninstitutional hospitals, exclusive of Federal, military, and Veterans Administrative hospitals, located in the 50 States and the District of Columbia. Only short-stay hospitals (hospitals with an average length of stay for all patients of less than 30 days) or those whose specialty is general (medical or surgical) or children's general are included in the survey. These hospitals must Page 1 also have six or more beds staffed for patient use. These criteria, used from 1988 through the current survey year, differ slightly from those used prior to 1988. Beginning with 1988, the NHDS sampling frame consisted of hospitals that were listed in the April 1987 SMG Hospital Market Tape (2), met the above criteria, and began accepting patients by August 1987. For 1990 the sample consisted of 542 hospitals. Of the 542 hospitals, 23 were found to be out of scope (ineligible) because they went out of business or otherwise failed to meet the criteria for the NHDS universe. Of the 519 inscope (eligible) hospitals, 474 hospitals responded to the survey. Sample design and data collection The NCHS has conducted the NHDS continuously since 1965. The original sample was selected in 1964 from a frame of short-stay hospitals listed in the National Master Facility Inventory. That sample was updated periodically with samples of hospitals that opened later. Sample hospitals were selected with probabilities ranging from certainty for the largest hospitals to 1 in 40 for the smallest hospitals. Within each sample hospital, a systematic random sample of discharges was selected. A report on the design and development of the original NHDS was published (1). In 1988, the NHDS was redesigned to provide geographic sampling comparability with other surveys conducted by the NCHS; to update the sample of hospitals selected into the survey; and to maximize the use of data collected through automated systems. The redesigned NHDS sample included with certainty all hospitals with 1,000 or more beds or 40,000 or more discharges annually. The remaining sample of hospitals was based on a stratified, three- stage design. The first stage consisted of selection of 112 primary sampling units (PSU's) that comprised a probability subsample of PSU's used in the 1985-94 National Health Interview Survey. The second stage consisted of selection of non-certainty hospitals from the sample PSU's. At the third stage a sample of discharges was selected by a systematic random sampling technique. These changes in the survey may affect trend data, that is, some of the differences between NHDS statistics based on the 1965-87 sample and statistics based on the sample drawn for the new design may be due to sampling error rather than changes in hospital utilization. For 1990, the NHDS sampling frame consists of eligible hospitals which began accepting inpatients before September 1, 1987, and were listed in the April 1987 SMG Hospital Market Data Tape (SMG Marketing Group, Inc. 1989). Two data collection procedures were used for the survey. The first was a manual system of sample selection and data abstraction, used for approximately 63 percent of the responding hospitals. The second was an automated method, used for approximately 37 percent of Page 2 the respondent hospitals, that involved the purchase of computerized data tapes from abstracting service organizations, state data systems, or hospitals. In the manual system, the sample selection and the transcription of information from the hospital records to abstract forms were performed at the hospitals. Of the hospitals using this system in 1990, about two thirds had the work performed by their own medical records staff. In the remaining hospitals using the manual system, personnel of the U.S. Bureau of the Census did the work on behalf of NCHS. The completed forms, along with sample selection control sheets, were forwarded to NCHS for coding, editing, and weighting. For the automated system, NCHS purchased tapes containing machine-readable medical record data from which records were systematically sampled by NCHS. The Medical data contains items relating to the personal characteristics of the patient, including birth date or age, sex, race, and marital status, but not name and address; administrative information, including admission and discharge dates, discharge status, and medical record number; and medical information, including diagnoses and surgical and nonsurgical operations or procedures. Since 1977, patient zip code, expected source of payment, and dates of surgery have also been collected. (The medical record number, date of birth, and patient zip code are confidential information and are not available to the public.) Medical Coding and Edit The medical information that was recorded manually on the sample patient abstracts was coded centrally by NCHS staff. A maximum of seven diagnostic codes was assigned for each sample abstract; in addition, if the medical information included surgical or nonsurgical procedures, a maximum of four codes for these procedures was assigned. The coding system currently used for coding the diagnoses (see Appendix 1) and procedures (see Appendix 2) on the medical abstract forms, as well as the data that appear on the commercial abstracting services data tapes, is the International Classification of Diseases, 9th Revision, Clinical Modification, or ICD-9-CM (3). All of the diagnostic codes and most of the procedure codes in the ICD-9-CM are utilized with the exception of selected procedures (see Appendix 3 for a list of procedure codes not used in 1990). NHDS usually presents diagnoses and procedures in the order they are listed on the abstract form or obtained from abstract services; however, there are exceptions. For women discharged after a delivery, a code of V27 from the supplemental classification is entered as the first-listed code. In another exception, a decision was made to reorder some acute myocardial infarction diagnoses. If an acute myocardial infarction is listed with other circulatory diagnoses and is other than the first entry, it is reordered to Page 3 first position. If a symptom appears as a first- listed code and a diagnosis appears as a secondary code, the diagnosis replaces the symptom which is moved back. Following conversion of the data on the medical abstract to computer tape and combining it with the automated data tapes, a final medical edit was accomplished by computer inspection and by a manual review of rejected records. Priority was given to medical information in the editing decision. Users of the National Hospital Discharge Survey (NHDS) diagnostic and/or procedure data, which is coded to the ICD-9-CM, must take into account annual ICD-9-CM addenda. The addenda lists new codes, new fourth or fifth digits to existing codes, as well as other modifications. A list of the changes for October 1986, October 1987, October 1988, October 1989, and October 1990 are listed in Appendix 4. The Uniform Hospital Discharge Data Set (UHDDS) Starting with 1979 data, the NHDS has followed guidelines of the Uniform Hospital Discharge Data Set (UHDDS) within the confines of its contractual agreement with participating hospitals. The UHDDS is a minimum data set of items uniformly defined (4). These items were selected on the basis of their usefulness to a broad range of organizations and agencies requiring hospital information, uniformity of definition, and general availability from medical records and abstract services. Population Estimates Population estimates provided by the U.S. Bureau of the Census are built into the SETS system to allow rate calculations for age, race, and sex. The estimates are of the U.S. civilian resident population on July l of the data year. These population estimates are consistent with those published in "Current Population Reports," Series P-25; however, these populations are not official population estimates of the Bureau of the Census. Measurement Errors As in any survey, results are subject to nonsampling or measurement errors, which include errors due to hospital nonresponse, missing abstracts, information incompletely or inaccurately recorded on abstract forms, and processing errors. Less than 2 percent of the discharge records failed to include age or sex of patient or date of admission or discharge. If the hospital record did not state age or sex of patient, it was imputed by assigning the patient an age or sex consistent with the age or sex of other sampled patients with the same diagnostic code. If the dates of admission or discharge were not given, and if they could not be obtained from the monthly sample listing sheet transmitted by the sample hospital, a length of stay was imputed by assigning the patient a stay characteristic of Page 4 the stays of other patients of the same age. Other edit and imputation procedures may have been applied to data in the NHDS collected in automated form. Sampling errors and Rounding of Numbers The standard error is primarily a measure of sampling variability that occurs by chance because only a sample rather than the entire universe is surveyed. The relative standard error of the estimate is obtained by dividing the standard error by the estimate itself. The resulting value is multiplied by 100, so the relative standard error is expressed as a percent of the estimate. Estimates of sampling variability were calculated with SESUDAAN software, which computes standard errors by using a first-order Taylor series approximation of the deviation of estimates from their expected values. A description of the software and the approach it uses was published by Shah (5). Relative Standard Errors for Aggregate Estimates Approximate relative standard errors for aggregate estimates are presented. To derive error estimates that would be applicable to a wide variety of statistics, numerous variances were calculated and a best fit formula was produced. The curves were based on an empirically determined relationship between the size of an estimate X and its relative variance. The relative standard error is then derived by taking the square root of the relative variance. The relative standard error of an estimate X [RSE(X)], expressed as a percent, may be calculated from the formula: _________ RSE(X) = 100 \/a + b/X with a and b provided in Appendix 5. Relative Standard Error for Estimates of Percents Approximate relative standard errors (in percent) for estimates of percents may be calculated from Appendix 5. Relative standard errors for a percent, 100 p (0 (133,000 + 7,969) 125,031 <-> 140,969 Presentation of Estimates Publication of estimates for the NHDS is based on the relative standard error of the estimate and the number of sample records on which the estimate is based (referred to as the sample size). Estimates are not presented in NCHS reports unless a reasonable assumption regarding the probability distribution of the sampling error is possible. Based on consideration of the complex sample design of the NHDS, the following guidelines are used for presenting the NHDS estimates: If the sample size is less than 30, the value of the estimate is not reported. If the sample size is 30-59, the value of the estimate is reported but should not be assumed reliable. If the sample size is 60 or more and the relative standard error is less than 30 percent, the estimate is reported. If the sample size is 60 or more but the relative standard error is over 30 percent, the estimate is reported but should not be assumed reliable. How to Use the Data File The NHDS records are weighted to allow inflation to national or regional estimates. To produce an estimate of the number of discharges, the Weight variable for Discharges, WGT, must be summed for the desired records. To produce an estimate for number of days of care, the Weight variable for Days of Care, DOCWGT, must be Page 6 summed for the desired records (DOCWGT is the variable WGT multiplied by the Length of Stay, LOS, variable). Appendix 6 contains unweighted frequencies and Appendix 7 contains weighted frequencies for selected descriptive variables on the data file. These may be used as a cross-check when processing the data on the user's system. Monthly and Seasonal Estimates Under the New Design An important difference between the old and new designs is the method used for adjusting weights for sampled cases from responding hospitals for individual months. In the old design, weights for responding hospitals were adjusted monthly to account for nonresponse. In the new design, this approach is maintained to account for hospitals which do not respond for the year. However, for partially responding hospitals (submitting between 6 through 11 months of data), weights are adjusted for the responding months to produce annual estimates. A result of this adjustment is that monthly and seasonal estimates from the NHDS are not strictly valid, but the effect is believed to be small. In the 1990 NHDS, 86 percent of the 474 responding hospitals provided data for all twelve months, and 94 percent provided at least 9 months of data. Diagnosis-Related Groups (DRGs).--Many users of the NHDS data files have expressed an interest in converting the data to DRGs. This has been done using DRG Grouper Programs obtained from the Health Care Financing Administration. The DRGs and the DRG Grouper Programs were developed outside of the National Center for Health Statistics; any questions about DRGs, other than specific questions about how they relate to NHDS data, should be addressed elsewhere. Questions Questions concerning data on the file should be directed to Maria Owings, Ph.D., Hospital Care Statistics Branch, Division of Health Care Statistics, National Center for Health Statistics, Presidential Building, Room 956, 6525 Belcrest Road, Hyattsville, Maryland 20782, (301)-436-7125. REFERENCES 1. National Center for Health Statistics: Development of the design of the NCHS Hospital Discharge Survey, by W. R. Simmons. Vital and Health Statistics. PHS Pub. No. l000, Series 2-No. 39. Public Health Service. Washington. U.S. Government Printing Office, Sept. l970. 2. SMG Hospital Marketing Group, Inc. 1989. Hospital Market Database. Healthcare Information Specialists, 1342 North LaSalle Drive, Chicago, Illinois. Page 7 3. National Center for Health Statistics: International Classification of Diseases, 9th Revision, Clinical Modification. DHHS Pub. No. (PHS) 80-l260. Public Health Service. Washington. U.S. Government Printing Office, Sept. l980. 4. Office of the Secretary, Department of Health and Human Services: Health Information Policy Council: 1984 Revision of the Uniform Hospital Discharge Data Set. Federal Register, Volume 50, No. 147. July 31, 1985. 5. Shah, B.V. 1981. SESUDAAN: Standard Errors Program for Computing of Standardized Rates from Sample Survey Data. Research Triangle Institute. Research Triangle Park, N.C. Page 8 c:\hdssets\testst3.txt - Page 9