Disease Change and the Role of Medicine: The Navajo Experience

Appendix I: Data and Methods

APPENDIX I DATA AND METHODS In a number of places throughout this study,multiple regression is used with rates of measures of hospital utilization or mortality as the dependent variables. This appendix describes the sources of data and the typesof analyses thathavebeen used. Several sources of data have been used: (1) hospital discharge records of Navajo patients residing on the reservation who have been seen in Indian Service and contract hospitals in the Navajo, Phoenix, and Albuquerque areas of the Indian Health Service from fiscal years 1972 through 1978; (2)population estimatesof Navajos residing in the eighteen land management districts of the reservation in 1975; (3) economic data gathered in 1974 by the Survey Research Center at Brigham Young University;and (4) death certificates from the Navajo area from 1972 to 1978.1 shall describe each briefly in turn. The hospital records consist of discharge sheets which have been computerized and include age, sex, tribe, communityof residence, primary, secondary, and tertiary diagnoses, types of surgical procedures done, dates of admission and discharge, and some additional information. Because the IHS isthe major provider of health care to Navajos on the reservation, and because most of the care not provided directly is paid for by contract funds, reporting of hospitalizations is virtually complete. The one important exception appears to beworkmen'scompensation cases whichare often cared for in non-IHS facilities and paid for by a third party that does not report to the Indian HealthService. In the present instance, the loss of cases isthought to be minimal. 190 I Appendixes Unfortunately, these records do not include such information as marital status, parity, occupation, and income. Population estimates for land management districts are based upon tribal enrollment data and school censuses for 1975 and include only total population and no information on age structure. Methods of estimation are described in detail elsewhere (Davis and Kunitz 1978). It must be pointed out that Navajo population has been notoriously difficult to estimate and enumerate. There isgood reason to believe that the U.S.Census has consistently undercounted Navajos, some estimates suggest by as much as 20 or 25 percent. By contrast, the Bureau of Indian Affairs population register is known to have overestimated the population (U.S. Bureau of the Census 1977). The figures I have used are based upon estimates made bythe Navajo Tribe's Office of Information and Statistics in consultation with the BIA and the Census Bureau (Faich 1976). When these are projected from 1975 to 1980 (by adding births and subtracting deaths), the result is a reservation population upwards of 148,000. Preliminary analyses of the 1980 U.S. Census gives a figure of about 128,000, approximately 14 percent less. In order to calculate hospitalization and mortality rates, the 1975 estimate of land management district populations has been used. In calculating age-specific rates of certain causes of mortality in chapter 3, a 1978 estimation of age and sex structure of the population based upon the 1975 figures has been used. In each case it must be kept in mind that the denominator (population size) may be as much as 15 percent overestimated.Thus, the rates I have estimated may be as much as 15 percent lower than they are in reality. In point of fact, the difference is likely to be less than 15 percent as the census enumeration is still probably somewhat low despite major efforts at improving itsaccuracy. The 1974 economic survey used land managementdistricts as sampling frames. Considering the difficulties described above in enumerating the population, developing an adequate sampling frame would be a formidable task indeed. Undoubtedly there are inaccuracies in the survey. Nonetheless, it is the only one availablein recent yearsthat gathered comparable data from all areas of the reservation and as such is of considerable value. Except for average distance to the nearest hospitalproviding surgery or simply general services, the independent variables are Appendixes I 191 drawn from this source. They are displayed in table 2.9 (chapter 2). The 1974 economic survey (Wistisen et al. 1975) collected data on a verylarge number of variables,onlya fewof which have been used in the present study. Those chosen reflect socioeconomic characteristics of the populations in different land management districts (median family and per capita income, proportion of income from wages and welfare,proportions of men and women working full-time, average educational level); housing conditions that might influence health status(proportion of...