Objectives. We describe methods used to mitigate the effect of race misclassification in mortality records and the data sets used to improve mortality estimates for American Indians and Alaska Natives (AI/ANs).
Methods. We linked US National Death Index (NDI) records with Indian Health Service (IHS) registration records to identify AI/AN deaths misclassified as non-AI/AN deaths. Analyses excluded decedents of Hispanic origin and focused on Contract Health Service Delivery Area (CHSDA) counties. We compared death rates for AI/AN persons and Whites across 6 US regions.
Results. IHS registration records merged to 176 137 NDI records. Misclassification of AI/AN race in mortality data ranged from 6.3% in the Southwest to 35.6% in the Southern Plains. From 1999 to 2009, the all-cause death rate in CHSDA counties for AI/AN persons varied by geographic region and was 46% greater than that for Whites. Analyses for CHSDA counties resulted in higher death rates for AI/AN persons than in all counties combined.
Conclusions. Improving race classification among AI/AN decedents strengthens AI/AN mortality data, and analyzing deaths by geographic region can aid in planning, implementation, and evaluation of efforts to reduce health disparities in this population.
Accurate health surveillance data are essential to address health disparities and to plan, implement, and evaluate disease prevention and control activities. Vital registration—the routine recording and monitoring of births and deaths within a population—provides a critical and ongoing picture of the health status of that population.1 However, the goal of producing reliable mortality estimates for American Indian and Alaska Native (AI/AN) populations has been hampered by the misclassification of race that frequently introduces bias toward underestimation of death rates using Vital Statistics data.2 Analyses that minimize misclassification of race have the potential to provide to tribes and their partners a more accurate description of the disease burden in AI/AN communities and, as a consequence, the tools to plan and implement more effective health promotion and disease prevention and control programs.
BACKGROUNDIn 2011, an estimated 6.2 million people reporting AI/AN ancestry alone or in combination with 1 or more other races lived in the United States—approximately 2% of the population.3 These people are members of—or related to—1 or more of more than 560 federally recognized or more than 200 nonfederally recognized tribes, and they represent communities with diverse languages, cultures, and histories.4 The AI/AN population is younger and poorer and has attained less education than the White population.3 Most of the AI/AN population resides west of the Mississippi River and makes up a greater proportion of the population in Alaska, Oklahoma, and other selected regions—the Southwest, the Northern Plains, and the Pacific Northwest (Figure 1). About 22% of AI/AN persons reside on tribal reservations, trust lands, or other tribally affiliated areas; approximately 59% live in urban areas.3
FIGURE 1—Percentage distribution of American Indian/Alaska Native (AI/AN) population, by county: United States, 2009.
Source. 2013 intercensal bridged single-race population estimates, US Census Bureau/Centers for Disease Control and Prevention/National Cancer Institute, (http://seer.cancer.gov/popdata).
The Indian Health Service (IHS) provides primary health care to approximately 2 million enrolled members of federally recognized tribes. The 168 IHS hospitals and clinics are located primarily on reservation lands and in a few cities with relatively large AI/AN populations. More than half of these health care facilities are managed by tribal governments under negotiated agreements with the federal government, and the rest of the facilities are operated directly by the federal government.5 An additional 34 urban health centers receive some federal funding to provide health care to the urban AI/AN population.6 Eligible AI/AN individuals can receive health care at any IHS facility, but complex rules govern and restrict delivery of contract health services for specialty medical care, such as cancer treatment, that is generally not available in IHS facilities.
Race misclassification of AI/AN persons has been reported in various public health data sets. Thoroughman et al.7 described substantial underreporting of sexually transmitted infections in Oklahoma resulting from race misclassification. Similar linkages of IHS registration records with state death records indicated a similar problem of race misclassification in mortality data.8,9 Compelling findings from the National Longitudinal Mortality Study (NLMS) have described marked race misclassification of AI/AN decedents compared with other racial groups.2 The NLMS allows investigators to compare self-identified race for a participant in the US Census Bureau’s Current Population Survey with the race classification recorded by the funeral home director at the time of death.10 It demonstrated race misclassification of nearly 30% for AI/AN decedents in comparison with other race groups.2 A previous study linking IHS patient registration records with the National Death Index (NDI) also found substantial underreporting of AI/AN race on death certificates, and the IHS uses these findings to adjust agency mortality reports.11,12 Multiple investigators have described misclassification in central cancer registries using linkages between registry records and patient registration records from the IHS.13–17 The use of such linkages has been integrated into national reporting for cancer incidence as a means for correcting misclassification of AI/AN race in central cancer registries.18
In this article, we describe methods used to mitigate the effect of race misclassification in mortality records and the data sets used to improve mortality estimates for AI/AN persons, including secular trends from 1990 to 2009, for the AI/AN population overall and by geographic region, reported in the individual articles in this supplement.
METHODS Data Sources Population estimates.This study includes population estimates and reported deaths from 1990 to 2009. Population estimates used as denominators in the rate calculations in this supplement are based on the annual time series of July 1 estimates of county populations by age, sex, race, and Hispanic origin produced by the US Census Bureau’s Population Estimates Program.19 Before 2000, the Office of Management and Budget required that federal agencies report 4 single-race categories: White, Black or African American, Asian or Pacific Islander, and AI/AN. Starting in 2000, the US Census Bureau developed annual county-level population estimates for 31 possible racial groups (5 single race and 26 multiple race) to include people who select 1, 2, 3, 4, or all 5 of the race categories. Corresponding multiple-race information was not uniformly available in state vital records (mortality data), especially in the years immediately after the transition to multiple-race reporting.20 Therefore, the US Census Bureau and the Centers for Disease Control and Prevention’s National Center for Health Statistics (NCHS) developed a method for bridging the 31 multiple-race categories used in the 2000 Census to the 4 single-race categories used in the 1990 Census by using information from the pooled 1997 to 2000 National Health Interview Surveys.20 The bridging method takes responses to the 2000 Census’ questions on race and reclassifies those responses to approximate the responses the individuals would hypothetically have given using the old single-race categories. Updated bridged single-race estimates that take into account the 2010 decennial census and population migrations during and after Hurricanes Katrina and Rita in 2005 were included as denominators in the calculations of death rates appearing in this supplement.21 Development of the bridged single-race data also makes the post-2000 race/ethnicity population estimates comparable to the pre-2000 race/ethnicity estimates and enables the reporting of a combined rate spanning 2000 as well as trend analyses.20
Death records.According to state laws and regulations, each death that occurs in that state must be registered and reported on a death certificate. For each death, a physician, coroner, or medical examiner typically provides clinical information documenting the cause of death. The funeral home director provides demographic and personal information, such as race and ethnicity. Death certificates are compiled at the state level yearly and are sent to the NCHS, where the data are edited for consistency and personal identifiers are removed. NCHS then makes this information available to the public in published reports and to the research community by providing the raw data (without identifiers) in electronic format as part of the National Vital Statistics System, which, in addition to deaths, includes births, marriages, divorces, and fetal deaths.1 The mortality data available for analysis via the National Vital Statistics System include, but are not limited to, cause of death, state of residence, age, gender, race, and ethnicity. For the years 1990 to 1998, the underlying cause of death was coded according to the International Classification of Diseases, Ninth Revision (ICD-9).22 From 1999 to 2009, the International Classification of Diseases, 10th Revision (ICD-10) was used.23 Trend analyses spanning ICD-9 and ICD-10 reporting years took into account comparability of cause-of-death recodes between the 2 revisions.24 NCHS applies a bridging algorithm nearly identical to that used by the US Census Bureau to assign a single race to decedents with multiple races reported on death certificates after 2000 such that they can be compared with deaths occurring from 1990 to 1999.25
National Death Index.The National Death Index (NDI) is a central electronic repository maintained within the NCHS of death record information on file in individual state vital statistics offices.26 NCHS works with state offices in establishing the NDI as a resource to aid epidemiologists and other investigators in the ascertainment of vital status and circumstances and cause of death. The NDI is a file of national death record information (beginning with 1979 deaths) containing personal identifiers compiled from electronic files submitted by individual state vital statistics offices. Death records are added annually, approximately 12 months from the end of the calendar year. NDI users submit as many of the following data items as possible for each study participant: first and last name, middle initial, father’s surname, Social Security number, birth date, race, sex, marital status, state of residence, and state of birth. The NDI aids investigators in determining whether individuals in their database submitted to the program have died, the state in which those deaths occurred, the date of death, the death certificate number, and cause of death.26
Indian Health Service patient registration database.In the mid-1980s, the IHS developed the Resource and Patient Management System to electronically capture clinical and public health data in IHS facilities.27 By the early 1990s, it was widely used in IHS and IHS-funded tribal facilities, many of which have now gathered personal health information for decades. Commissioned Corps officers of the US Public Health Service or non-native spouses of AI/AN persons who seek medical services at IHS facilities are excluded from the registration database by applying the IHS “Indian Status” algorithm, based on 3 variables: beneficiary code, tribal code, and blood quantum (representing the proportion of native ancestry). Individual tribes determine the degree of tribal ancestry necessary for tribal membership, which, in turn, determines eligibility to receive services from the IHS.28
Data LinkageWe submitted the IHS patient registration database to the NDI program for linkage to identify AI/AN deaths misclassified as non-AI/AN deaths. NDI conducted the linkages using a 2-step process. In the first step, the NDI program selected potential death record matches on the basis of a set of 7 matching criteria using key patient identifiers (Social Security number, first name, last name, middle initial, date of birth, sex, and date of death). The second step involved a procedure resulting in a probabilistic score or weight of a potential match pair and a suggested determination of final match status by NCHS.29 After the linkage, NDI staff sent the results to Centers for Disease Control and Prevention project staff assigned to the IHS Division of Epidemiology and Disease Prevention in Albuquerque, New Mexico, for review. Two reviewers independently examined pairs with intermediate final weights (designated as “clerical reviews”). Each reviewer assigned a status of match or nonmatch by comparing personal identifying fields in each record. Any discrepancies between these 2 reviewers were adjudicated by a third reviewer.
Analytic Data Set AI/AN Mortality Database.The AI/AN Mortality Database (AMD) includes all deaths for all races reported to NCHS from 1990 to 2009. After the linkage of IHS patient registration data with NDI and the completion of the clerical review process, we returned the state death certificate number and year of death of IHS clients to NCHS to merge back to the National Vital Statistics System mortality file. NCHS added a variable to indicate linkage to the IHS registration file (IHS link), which served as a supplemental indicator of AI/AN ancestry to the death certificate race. NCHS stripped this file of personal identifiers and death certificate numbers, and we combined the resulting file with the updated bridged single-race intercensal population estimates as denominators, thus creating the AMD.
Classification of race and ethnicity in the AI/AN Mortality Database.Current Office of Management and Budget standards include the following minimum categories for the collection of race information: AI/AN, Asian, Black or African American, Native Hawaiian or other Pacific Islander, and White.30 These race categories represent social, cultural, and political characteristics as well as ancestry and are not genetically or biologically based.30 The current Office of Management and Budget standards also allow multiple-race data collection and reporting by federal agencies.30
The reporting of decedents’ race and Hispanic ethnicity is the responsibility of funeral home directors, and NCHS has issued guidelines for the collection of this information.10 Race categories used by state vital statistics offices are specified by NCHS and correspond to the race categories used by the US Census Bureau to allow calculation of race-specific death rates. A 2003 revision to the US Standard Certificate of Death allowed the reporting of more than 1 race.31 In the AMD, we considered decedents AI/AN if they were classified as such by the NCHS in the National Vital Statistics System file or if there was a positive IHS link. Race is coded independently of Hispanic/Latino origin, and race and ethnicity are not mutually exclusive categories.30 Hispanic origin became a part of the US Standard Death Certificate in the 1989 revision. Before this revision, some states collected Hispanic origin on the death certificate; however, missing information for this item was common.2 By 1990, only Louisiana, New Hampshire, and Oklahoma were not including Hispanic origin on death certificates, and most other states had missing rates well below 1% for this field. State coverage was complete beginning in 1997.2 All of our analyses exclude data from those states for the years in which they were not reporting Hispanic ethnicity.2
During preliminary analyses, we learned that the updated bridged intercensal population estimates significantly overestimated AI/AN persons of Hispanic origin.32 Therefore, to avoid underestimating mortality and cancer incidence in AI/AN persons, reports in this supplement based on the AMD and cancer registry data linked to IHS are limited to non-Hispanic AI/AN persons. Non-Hispanic White was chosen as the most homogeneous referent group. Henceforth, the term “non-Hispanic” is omitted when discussing both groups, as has been done previously.33
Geographic CoverageMost of the analyses in this supplement were restricted to IHS Contract Health Service Delivery Area or Tribal Service Delivery Area counties (CHSDA counties) that, in general, contain federally recognized tribal reservations or off-reservation trust lands or are adjacent to them (Figure 2). For death rates restricted to CHSDA counties, data from counties within 35 states were included. CHSDA residence is used by the IHS to determine eligibility for services not directly available within the IHS. IHS also uses CHSDA designation for routine mortality reporting.12 Linkage studies involving IHS patient registration records as well as National Longitudinal Mortality Study data indicate less misclassification of race for AI/AN persons in these counties.2,34 The CHSDA counties also have higher proportions of AI/AN persons in relation to total population than do non-CHSDA counties, with 64% of the US AI/AN population residing in the 637 counties designated as CHSDA (representing 20% of the 3141 counties in the United States; Table 1). Although less geographically representative (Figure 2), analyses restricted to CHSDA counties are presented for death rates in this article for the purpose of offering improved accuracy in interpreting mortality statistics for AI/AN individuals.
FIGURE 2—States and Contract Health Service Delivery Area counties by Indian Health Service Region: United States, 2009.
Note. CHSDA = Contract Health Service Delivery Area.
TABLE 1—Population Coverage of Mortality Data for AI/AN Persons and Whites, by Geographic Area: Contract Health Service Delivery Area Counties and All Counties, United States, 2009
AI/AN White Area AI/AN Population in All Counties AI/AN Population Resident in CHSDA Counties % AI/AN Population Resident in CHSDA Counties % All Races Population That is AI/AN in CHSDA Counties White Population in All Counties White Population Resident in CHSDA Counties % All Races Population That is White in CHSDA Counties Northern Plains 439 445 284 560 64.8 3.1 37 452 086 7 857 893 21.0 Alaska 114 207 114 207 100.0 16.3 471 660 471 660 100.0 Southern Plains 468 408 357 204 76.3 7.5 16 494 077 2 939 922 17.8 Southwest 543 851 496 480 91.3 4.3 11 890 700 6 842 577 57.5 Pacific Coast 390 765 278 583 71.3 1.2 25 232 525 13 891 541 55.1 East 595 475 108 286 18.2 0.5 108 389 242 13 757 313 12.7 Total 2 552 151 1 639 320 64.2 2.3 199 930 290 45 760 906 22.9We completed analyses for all regions combined and by individual IHS region: Northern Plains, Alaska, Southern Plains, Southwest, Pacific Coast, and East (Figure 2). Identical or similar analyses using these regions have been conducted for other health-related publications focusing on AI/AN populations.35–37
Statistical MethodsAll rates, expressed per 100 000 population, were directly age adjusted, using SEER*Stat software, version 8.0.4 (National Cancer Institute, Bethesda, MD),38 to the 2000 US standard population and using, except where otherwise noted, 11 age groups (> 1 year, 1–4 years, 5–14 years, 15–24 years, 25–34 years, 35–44 years, 45–54 years, 55–64 years, 65–74 years, 75–84 years, and ≥ 85 years) in accordance with a 1998 US Department of Health and Human Services recommendation.39,40 Readers should avoid comparison of these data with published death rates adjusted using a different standard population.
Using the age-adjusted death rates, we calculated standardized rate ratios for AI/AN populations using White rates for comparison. Rate ratios calculated by the reader from rounded rates presented in the tables may not correspond to the rate ratios calculated by SEER*Stat before rounding. We calculated confidence intervals for age-adjusted rates and standardized rate ratios on the basis of methods described by Tiwari et al.41 using SEER*Stat 8.0.4 and presented them in these reports to the second decimal point.
We assessed temporal changes in annual age-adjusted incidence and death rates with joinpoint regression techniques using statistical software developed by the National Cancer Institute (Bethesda, MD).42 Joinpoint analyses spanning ICD-9 and ICD-10 were conducted only for cause-of-death recodes with comparability ratios close to 1.24
Analysis for Cancer ArticlesData sources and analytic methods for cancer incidence and stage at diagnosis in the supplement articles addressing cancer burden in AI/AN persons are explained in the cancer overview article by White et al.18
RESULTSSummary results of the linkages are presented in Table 2. Briefly, for the 176 137 decedents whose record linked to the IHS patient registration database for the period 1990–2009, misclassification of AI/AN persons as another race on the death certificate for all regions combined was 17.7% and ranged from 6.3% in the Southwest to 35.6% in the Southern Plains.34
TABLE 2—Percentage of Misclassification as a Non-AI/AN Person of NDI Decedents Linked With the IHS Patient Registration Database, by IHS Region: United States, 1990–2009
Region AI/AN Decedents, No. Non-AI/AN Decedents, No. % Misclassification Northern Plains 34 731 4033 10.4 Alaska 12 888 894 6.5 Southern Plains 22 196 12 247 35.6 Southwest 49 945 3338 6.3 Pacific Coast 19 333 7413 27.7 East 5909 3210 35.2 Total 145 002 31 135 17.7We demonstrate the impact of restricting analyses to non-Hispanic AI/AN persons in Table 3, in which the difference for all-cause death rates in non-Hispanic AI/AN decedents and all AI/AN decedents from 2000 to 2009 is shown by region and gender. The percentage of change between the 2 methods for both sexes combined ranges from a 0.6% increase in Alaska to a 25.8% increase in rates in the Pacific Coast. Similar changes between the 2 methods are also reflected in the rate ratios comparing AI/AN individuals with Whites. The time period for Table 3 begins in 2000 to match the year of onset of the overestimation of Hispanic AI/AN persons described earlier.
TABLE 3—All-Cause Death Rates for Non-Hispanic AI/AN Persons and All AI/AN Persons Compared With Non-Hispanic Whites, by IHS Region and Sex: CHSDA Counties, United States, 2000–2009
Non-Hispanic AI/ANa All AI/ANb IHS Region and Sex AI/AN Count AI/AN Rate White Count White Rate AI/AN:White RR (95% CI) AI/AN Count AI/AN Rate White Count White Rate AI/AN:White RR (95% CI) % Change in Non-Hispanic AI/AN vs All AI/AN Northern Plains Male and female 21 388 1449.9 714 793 765.6 1.89* (1.86, 1.92) 21 635 1411.4 714 793 765.6 1.84* (1.82, 1.87) 2.7 Male 11 666 1737.0 351 148 920.3 1.89* (1.84, 1.93) 11 823 1683.5 351 148 920.3 1.83* (1.79, 1.87) 3.2 Female 9722 1232.0 363 645 645.5 1.91* (1.87, 1.95) 9812 1203.3 363 645 645.5 1.86* (1.82, 1.91) 2.4 Alaska Male and female 7939 1219.0 21 785 735.3 1.66* (1.61, 1.71) 8019 1211.2 21 785 735.3 1.65* (1.60, 1.69) 0.6 Male 4400 1431.2 12 530 853.2 1.68* (1.61, 1.75) 4449 1421.0 12 530 853.2 1.67* (1.60, 1.74) 0.7 Female 3539 1040.4 9255 625.5 1.66* (1.60, 1.73) 3570 1034.4 9255 625.5 1.65* (1.59, 1.72) 0.6 Southern Plains Male and female 28 039 1317.6 326 476 927.5 1.42* (1.40, 1.44) 28 442 1288.0 326 476 927.5 1.39* (1.37, 1.41) 2.3 Male 14 717 1572.8 160 136 1098.3 1.43* (1.41, 1.46) 14 962 1534.7 160 136 1098.3 1.40* (1.37, 1.42) 2.5 Female 13 322 1121.0 166 340 791.0 1.42* (1.39, 1.44) 13 480 1096.9 166 340 791.0 1.39* (1.36, 1.41) 2.2 Southwest Male and female 30 615 1015.6 613 792 786.4 1.29* (1.28, 1.31) 32 218 944.1 613 792 786.4 1.20* (1.19, 1.22) 7.6 Male 17 332 1252.3 318 295 920.3 1.36* (1.34, 1.38) 18 337 1151.5 318 295 920.3 1.25* (1.23, 1.27) 8.8 Female 13 283 823.8 295 497 668.9 1.23* (1.21, 1.25) 13 881 772.6 295 497 668.9 1.16* (1.13, 1.18) 6.6 Pacific Coast Male and female 19 259 1097.5 1 329 768 791.8 1.39* (1.36, 1.41) 20 608 872.1 1 329 768 791.8 1.10* (1.08, 1.12) 25.8 Male 10 049 1242.4 657 797 927.5 1.34* (1.31, 1.37) 10 857 972.3 657 797 927.5 1.05* (1.02, 1.07) 27.8 Female 9210 978.8 671 971 680.9 1.44* (1.41, 1.47) 9751 786.4 671 971 680.9 1.16* (1.13, 1.18) 24.5 East Male and female 5707 826.0 1 415 749 791.3 1.04* (1.01, 1.07) 5779 719.8 1 415 749 791.3 0.91* (0.88, 0.94) 14.8 Male 2978 928.1 681 805 951.1 0.98 (0.94, 1.02) 3024 797.7 681 805 951.1 0.84* (0.80, 0.87) 16.3 Female 2729 737.3 733 944 667.9 1.10* (1.06, 1.15) 2755 649.8 733 944 667.9 0.97 (0.93, 1.01) 13.5 Total Male and female 112 947 1165.3 4 422 363 794.7 1.47* (1.46, 1.48) 116 701 1064.6 4 422 363 794.7 1.34* (1.33, 1.35) 9.5 Male 61 142 1381.3 2 181 711 942.4 1.47* (1.45, 1.48) 63 452 1250.3 2 181 711 942.4 1.33* (1.31, 1.34) 10.5 Female 51 805 990.9 2 240 652 675.8 1.47* (1.45, 1.48) 53 249 912.0 2 240 652 675.8 1.35* (1.34, 1.36) 8.7Analyses in CHSDA counties resulted in overall higher death rates for AI/AN persons than did analyses for all counties combined (Table 4). Excluding the Alaska region, which is entirely CHSDA, increases ranged from 1.8% in the Southwest region to 39.1% in the East region. The change for all regions combined was 20.9%. Updating the race field on the basis of the linkage results increased the all-cause death rates for the period 1999 to 2009 in CHSDA counties in all regions combined from 964 deaths to 1166 deaths per 100 000, a 17.3% difference.34 The largest change, 40.9%, was noted in the Southern Plains region, where the rate increased from 932 deaths to 1313 deaths per 100 000.
TABLE 4—All-Cause Death Rates for American Indians/Alaska Natives Compared With Whites, by IHS Region and Sex: United States, 1999–2009
CHSDA Counties All Counties IHS Region and Sex AI/AN Count AI/AN Rate White Count White Rate AI/AN:White RR (95% CI) AI/AN Count AI/AN Rate White Count White Rate AI/AN:White RR (95% CI) Northern Plains Male and female 23 331 1461.8 786 392 770.6 1.90* (1.87, 1.93) 31 188 1242.9 3 843 218 787.1 1.58* (1.56, 1.60) Male 12 709 1748.8 386 164 927.4 1.89* (1.84, 1.93) 16 812 1484.9 1 846 384 947.8 1.57* (1.54, 1.60) Female 10 622 1243.4 400 228 649.2 1.92* (1.88, 1.96) 14 376 1064.5 1 996 834 666.6 1.60* (1.57, 1.63) Alaska Male and female 8616 1218.6 23 621 738.2 1.65* (1.61, 1.70) 8616 1218.6 23 621 738.2 1.65* (1.61, 1.70) Male 4771 1431.6 13 600 856.8 1.67* (1.61, 1.74) 4771 1431.6 13 600 856.8 1.67* (1.61, 1.74) Female 3845 1041.2 10 021 627.3 1.66* (1.60, 1.73) 3845 1041.2 10 021 627.3 1.66* (1.60, 1.73) Southern Plains Male and female 30 421 1313.1 358 711 928.7 1.41* (1.40, 1.43) 35 130 1159.6 1 758 152 859.7 1.35* (1.33, 1.36) Male 15 946 1568.7 175 778 1102.2 1.42* (1.40, 1.45) 18 391 1359.5 858 447 1018.7 1.33* (1.31, 1.36) Female 14 475 1116.3 182 933 790.9 1.41* (1.39, 1.44) 16 739 1001.1 899 705 733.6 1.36* (1.34, 1.39) Southwest Male and female 33 325 1017.8 669 622 789.7 1.29* (1.27, 1.30) 35 366 1000.0 1 052 569 776.8 1.29* (1.27, 1.30) Male 18 836 1251.4 347 628 926.2 1.35* (1.33, 1.37) 19 916 1218.4 536 547 909.7 1.34* (1.32, 1.36) Female 14 489 828.1 321 994 670.4 1.24* (1.21, 1.26) 15 450 821.5 516 022 664.2 1.24* (1.22, 1.26) Pacific Coast Male and female 20 779 1091.5 1 459 406 796.0 1.37* (1.35, 1.39) 27 339 953.5 2 711 044 781.0 1.22* (1.21, 1.24) Male 10 875 1238.3 721 856 933.7 1.33* (1.30, 1.36) 14 379 1088.3 1 327 483 916.5 1.19* (1.16, 1.21) Female 9904 971.1 737 550 683.8 1.42* (1.39, 1.45) 12 960 842.9 1 383 561 671.4 1.26* (1.23, 1.28) East Male and female 6172 828.7 1 559 313 795.7 1.04* (1.01, 1.07) 24 738 595.7 12 136 547 824.8 0.72* (0.71, 0.73) Male 3231 939.1 750 611 957.7 0.98 (0.94, 1.02) 13 095 691.7 5 872 696 988.2 0.70* (0.69, 0.71) Female 2941 735.4 808 702 671.0 1.10* (1.06, 1.14) 11 643 518.3 6 263 851 698.3 0.74* (0.73, 0.76) Total Male and female 122 644 1165.9 4 857 065 798.8 1.46* (1.45, 1.47) 162 377 964.4 21 525 151 812.2 1.19* (1.18, 1.19) Male 66 368 1381.8 2 395 637 948.8 1.46* (1.44, 1.47) 87 364 1135.2 10 455 157 969.1 1.17* (1.16, 1.18) Female 56 276 991.5 2 461 428 678.6 1.46* (1.45, 1.47) 75 013 827.3 11 069 994 689.9 1.20* (1.19, 1.21) DISCUSSIONThe methods used in this supplement enhance AI/AN mortality surveillance by addressing race misclassification and by including analyses by geographic region. Linkages of IHS registration data to death records in the NDI and restriction of analyses to CHSDA counties are efficient ways of reducing the proportion of AI/AN decedents misclassified as non-AI/AN and reducing bias toward underestimation in mortality data among AI/AN persons. The supplement also includes data from 50 state cancer registries and the District of Columbia, including 35 of the 50 states containing CHSDA counties, and is therefore the most comprehensive analysis of cancer incidence in AI/AN populations to date.
Findings from the articles in this supplement indicate that wide regional variation is characteristic of AI/AN mortality and that region-specific data are essential to characterize AI/AN mortality patterns and disparities. In general, death rates among AI/AN populations in CHSDA counties were highest in Alaska and the Northern and Southern Plains. The wide regional variations may, in part, reflect geographic variations in environmental, social, and personal determinants of health.43 Research designed to understand regional variations in disease risk may help identify appropriate prevention and control strategies.
LimitationsSeveral limitations should be considered when interpreting the results presented in this supplement. First, although linkage with the IHS patient registration database improves the classification of race for many AI/AN decedents, the issue is not completely resolved because AI/AN people who are not members of federally recognized tribes are not eligible for IHS services and are therefore not represented in the IHS registration database. Additionally, some decedents may have been eligible for—but never used—IHS services and were therefore not included in the IHS registration database. Second, the findings from CHSDA counties highlighted in this supplement do not represent all AI/AN populations in the United States or in individual IHS regions (Table 1, Figure 2). In particular, the CHSDA portion of the East region includes only 18.2% of the total AI/AN population for that region. Furthermore, the analyses based on CHSDA designation exclude many AI/AN decedents in urban areas that are not part of a CHSDA county. AI/AN residents of urban areas differ from other AI/AN persons in poverty level, health care access, and other factors that may influence mortality trends.44 Third, these analyses revealed less variation for White than for AI/AN death rates by IHS region using data from CHSDA counties only. Perhaps alternative groupings of states or counties would reveal a different level of variation for Whites. Fourth, federally recognized tribes vary substantially in the proportion of native ancestry required for tribal membership and therefore for eligibility for IHS services. Whether and how this discrepancy in tribal membership requirements may influence some of our findings is unclear, although our findings are consistent with prior reports. Finally, although the exclusion of Hispanic AI/AN persons from the analyses reduces the overall AI/AN deaths by less than 5%, it may disproportionately exclude some tribal members in states along the US–Mexico border and possibly elsewhere who have Hispanic surnames and may be coded as Hispanic at death.
Future DirectionsLinking IHS registration records and national death records as described in this report and elsewhere in this supplement is a useful—though imperfect—tool to describe mortality patterns in AI/AN populations. To build on these efforts, we propose that the public health community and supporting agencies consider making these linkages routine until a more robust way of presenting AI/AN mortality data becomes available. Another promising approach is the development and expansion of tribal rosters, such as the Northwest tribal roster,45 to complement the IHS patient registration database and further improve race classification for AI/AN individuals through the use of data linkages. The high rate of misclassification of AI/AN race on death certificates has been documented in several prior studies9,46 and in this supplement.34 Data linkages and analysis by IHS regions restricted to CHSDA counties has allowed substantial progress in surveillance for cancer—and now mortality—in AI/AN populations, providing the most comprehensive picture of health status in this population currently available. To build on this progress, the vital statistics community and the many partners who bring mortality surveillance to fruition should continue efforts to improve race classification and routine reporting of mortality in AI/AN populations. These improved data should be readily available to the tribes and to their partners in the public health community to more effectively plan, implement, and evaluate disease control and health promotion programs that aim to improve the health and well-being of AI/AN people.
AcknowledgmentsWe thank Steve Scoppa of Information Management Services, Inc, for his valuable assistance in developing the analytic databases.
Human Participant ProtectionThe study did not involve human participants, so institutional review board approval was not necessary.
ReferencesRetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.3
