Skip Navigation LinksSkip Navigation Links
Centers for Disease Control and Prevention
Safer Healthier People
Blue White
Blue White
bottom curve
CDC Home Search Health Topics A-Z spacer spacer
Blue curve MMWR spacer

Classification of American Indian Race on Birth and Infant Death Certificates -- California and Montana

The accuracy of infant mortality rates and other indices of the health of populations depends on the consistency of information collected from separate sources (e.g., birth and death certificates). Inconsistent recording of basic information such as race and ethnicity has resulted in underestimation of mortality among minority populations, particularly minority populations other than blacks (1). This report summarizes studies in California and Montana that describe and measure the magnitude of differences in the recording of race for American Indians/Alaskan Natives (AI/ANs) on birth and infant death certificates. California

Reported infant mortality rates in California often have been based on information obtained separately from birth and death certificates. To assess the accuracy of the reported rates for the state's AI/ANs, the California Area Office of the Indian Health Service (IHS) and the California Department of Health Services used the state's Birth Cohort File in which birth and death certificate information of individual infants are linked. Infant mortality rates for 1984-1988 were calculated from the Birth Cohort File using two methods for defining infant race: the pre-1989 algorithm from CDC's National Center for Health Statistics * and the IHS algorithm **. With both methods, information on parents' race was obtained from birth certificates. Numerators for infant mortality rates were deaths linked to birth certificates that met the CDC and IHS definitions for AI/AN births. Denominators were composed of all AI/AN births meeting the CDC and IHS definitions using birth certificates. All births and deaths were restricted to infants whose mothers were California residents at the time of the birth. Record linkage for this population was 100% for this study.

The aggregate number of AI/AN births for 1984-1988 was 28,668 when applying the IHS definition to unlinked birth certificate files (i.e., when death certificate data were not linked to birth certificates). In the linked Birth Cohort Files (which allow an additional year of data collection to ensure the accuracy of the data) the number of AI/AN births was 27,588 when tabulated using the CDC definition, and 29,030 using the IHS definition.

In every year, the number of AI/AN infant deaths in California was greater when calculated from the linked file than from the unlinked death certificate file, and greatest when the IHS definition was applied. The unlinked Death Statistics Master File for 1984-1988 included data from a total of 111 death certificates for AI/AN infants. In comparison, the linked file included 298 deaths when the CDC definition was used (a difference of 168%) and 315 when the IHS definition was used (a difference of 184%).

For 1984-1988, the infant mortality rate based on unlinked data was 3.9 deaths per 1000 live births. The rate was 2.8 times greater when calculated using the linked files (10.8 and 10.9 per 1000 using the CDC and IHS definitions, respectively). Both of these rates are comparable to that for AI/AN infants in all areas served by the IHS (10.9 per 1000 for AI/AN infants). Montana

In Montana, resident birth and infant death records linked by the state vital registrar since 1980 were used to estimate errors in the reporting of race (2,3). All linked birth and infant death records for 1980-1989 were analyzed; less than 0.1% of these records in the state were not linked. Infant race at birth was tabulated based on 1) the mother's race -- the procedure used as of 1989 in statistics tabulated by CDC -- and 2) the algorithm used before 1989. Infant race at death was tabulated by three methods:

  1. race as reported on the infant's death certificate -- the standard method, used in the absence of linked records; 2) the mother's race as recorded on the birth certificate; and 3) the infant's race derived from the CDC algorithm used before 1989.

The number of AI/AN births based on the CDC definition (14,893) was higher than that based on mother's race (12,749). For 1980-1989, there were 1285 infant deaths of all races in Montana. For 42 (3.3%) of these, the race recorded on the death certificate differed from that on the birth certificate. Of 1036 infants who were classified as white at birth, seven (0.7%) were classified otherwise at death (one as black and six as AI/AN). In comparison, of 232 infants classified as AI/AN at birth, 29 (12.5%) were classified as white at death. The number of AI/AN infant deaths for the decade varied according to the definition of the decedents' race: 210 were reported on the death certificate, 232 were ascertained using the pre-1989 CDC algorithm, and 202 were ascertained by assignment of the mother's race.

Determination of the race at death based on the death record produced two different estimates of AI/AN infant mortality (Table 1): 16.5 per 1000 when mother's race was used to define race at birth and 14.1 per 1000 when the pre-1989 CDC algorithm was used. Use of the pre-1989 CDC algorithm to define race at death produced two higher estimates: 18.2 when the mother's race was used to define race at birth and 15.6 using the pre-1989 CDC algorithm. Finally, the rates were lowest when race of the infant at death was defined as the mother's race: 15.8 when mother's race was used to define race at birth and 13.6 when the infant's race at birth was defined by the pre-1989 CDC algorithm. When the mother's race determined from the birth record was used as the denominator, estimated AI/AN infant mortality rates were always higher than those obtained when the pre-1989 CDC algorithm was used, independent of the method of assigning race to the infant (Table 1). The infant mortality rate for the same period calculated by standard procedures with unlinked vital records was 13.9 per 1000 live births.

Reported by: CC Watson, MPH, California Area Office, Indian Health Svc, Sacramento. T Bennett, DrPH, Dept of Maternal and Child Health, Univ of North Carolina at Chapel Hill. FW Reed, PhD, WH McBroom, PhD, Center for Population Research, Univ of Montana, Missoula. SD Helgerson, MD, Health Care Financing Administration (Region X), Seattle. Div of Surveillance and Epidemiology, Epidemiology Program Office, CDC.

Editorial Note

Editorial Note: From 1973 through 1987, the reported U.S. AI/AN infant mortality rate decreased from 22 to 11 infant deaths per 1000 live births (4). In addition, during this period cause-specific mortality rates for AI/ANs decreased 77% for gastrointestinal diseases, 76% for tuberculosis, 59% for pneumonia and influenza, and 54% for unintentional injuries (4). Although these trends in mortality rates suggest substantial improvements in public health status for AI/ANs, the findings in this report indicate that race-specific mortality rates can substantially underestimate disease burden when race coding is inconsistent in vital records.

Race has not been consistently defined or ascertained in public health and related data sets (5). For example, when infant mortality rates are calculated, the denominator generally consists of the number of births in the same year as that in which the deaths occurred. For many other mortality and morbidity rates, denominators are commonly derived from U.S. Census data that rely on respondents' self-identification. In comparison, race coding for numerator data may reflect a wider range of methods, including designation of race by next-of-kin, a coroner, or other person who certifies the death. Similarly, race coding for reportable diseases may represent the independent designation of a health-care provider.

The effect of the inconsistent race coding associated with these practices may be to obscure race-specific disease and injury burdens. The studies described in this report and in others (1,6) illustrate how the impact of inconsistent race coding between data sets can be estimated by linking the data sets. Previous studies linking IHS records to state death files, tumor and injury registries, and a registry for end-stage renal disease have documented underascertainment of the disease burden for AI/AN populations (7-11).

Vital records and registries are essential data sources for monitoring progress toward national health objectives for the year 2000 (12). The Indian health amendments of 1992 *** require attention to 61 AI/AN health objectives. Improved national and local surveillance of AI/AN mortality rates and other health indicators is necessary for achieving these objectives. Improvement of surveillance efforts will require assessment of fundamental concepts such as "race," "ancestry," and "ethnicity" and the meaning of these terms in different populations. These improvements also will require the use of consistent categories and measurement procedures between data-collection sources and agencies.

Different definitions of race identification may serve different purposes. For example, the IHS definition corresponds to definitions of tribal membership associated with health-care benefits. The revised definition of an infant's race at birth as that of its mother (now used at the national level for purposes of published statistics) is consistent for all races. Access to national linked birth and infant death computer files, currently available for infants born from 1983 through 1987, will facilitate the computation of vital statistics with consistent definitions of race at birth and death (13).


  1. Hahn RA, Mulinare J, Teutsch SM. Inconsistencies in coding of race and ethnicity between birth and death in US infants: a new look at infant mortality, 1983 through 1985. JAMA 1992;267:259-63.

  2. Reed FW, McBroom WH, Sperry SH, Helgerson SD. Implications of inaccurate infant mortality data. The IHS Primary Care Provider 1992;17:29-32.

  3. McBroom WH, Reed FW, Sperry SH, Helgerson SD. Ethnic misclassification, substance use, and their sequelae. Presented at the fifth Annual Indian Health Service Research Conference, Tucson, Arizona, May 5-7, 1992.

  4. Indian Health Service. Trends in Indian health, 1992. Rockville, Maryland: US Department of Health and Human Services, Public Health Service, 1992:63.

  5. Hahn RA. The state of federal health statistics on racial and ethnic groups. JAMA 1992;267:268-71.

  6. Hahn RA, Mendlein JM, Helgerson SD. Differential classification of American Indian race on birth and death certificates, U.S. reservation states, 1983-1985. The IHS Primary Care Provider 1993;18:8-11.

  7. Kimball EH, Frost F. Estimate of racial misclassification error on Washington State death certificates and Puget Sound Cancer Registry. Presented at fourth Annual Indian Health Service Research Conference, Tucson, Arizona, April 22-24, 1991.

  8. Frost F, Taylor V, Fries E. Race misclassification of Native Americans in a Surveillance, Epidemiology, and End Results registry. J Natl Cancer Inst 1992;84:957-62.

  9. Bleed DM, Risser DR, Sperry S, Hellhake D, Helgerson SD. Cancer incidence and survival among American Indians registered for Indian Health Service care in Montana, 1982-1987. J Natl Cancer Inst 1992;84:1500-5.

  10. Sugarman JR, Soderberg R, Gordon JE, Rivara FP. Racial misclassification of American Indians: its effect on injury rates in Oregon, 1989 through 1990. Am J Public Health (in press).

  11. Sugarman JR, Lawson L. The effect of racial misclassification on estimates of end-stage renal disease among American Indians and Alaska Natives, Pacific Northwest, 1988-1990. Am J Kidney Dis 1993;24:383-6.

  12. Public Health Service. Healthy people 2000: national health promotion and disease prevention objectives -- full report, with commentary. Washington, DC: US Department of Health and Human Services, Public Health Service, 1991; DHHS publication no. (PHS)91-50212.

  13. Prager, K, Flinchum GA, Johnson DP. The NCHS pilot project to link birth and infant death records: stage 1. Public Health Rep 1987;102:216-23.

    • Based on the pre-1989 CDC algorithm in use during the study period, infants were determined to be AI/AN if on the birth certificate 1) the father was coded AI/AN and the mother was not coded Hawaiian or 2) if the mother was coded AI/AN and the father was coded AI/AN, white, or race unknown. ** According to the IHS procedure, infants were coded as AI/AN if the mother and/or father were identified as AI/AN on the birth certificate. *** Public Law 102-573.

Disclaimer   All MMWR HTML versions of articles are electronic conversions from ASCII text into HTML. This conversion may have resulted in character translation or format errors in the HTML version. Users should not rely on this HTML document, but are referred to the electronic PDF version and/or the original MMWR paper copy for the official text, figures, and tables. An original paper copy of this issue can be obtained from the Superintendent of Documents, U.S. Government Printing Office (GPO), Washington, DC 20402-9371; telephone: (202) 512-1800. Contact GPO for current prices.

**Questions or messages regarding errors in formatting should be addressed to

Page converted: 09/19/98


Safer, Healthier People

Morbidity and Mortality Weekly Report
Centers for Disease Control and Prevention
1600 Clifton Rd, MailStop E-90, Atlanta, GA 30333, U.S.A


Department of Health
and Human Services

This page last reviewed 5/2/01