As recently as 1990, there was no reservation-wide, population-based health status information about Navajo Indians. To remedy this shortcoming, the Navajo Health and Nutrition Survey was conducted from 1991 to 1992 to assess the health and nutritional status of Navajo Reservation residents using a population-based sample. Using a three-stage design, a representative sample of reservation households was selected for inclusion. All members of selected households 12 y of age and older were invited to participate. A total of 985 people in 459 households participated in the study. Survey protocols were modeled on those of previous national surveys and included a standard blood chemistry profile, complete blood count, oral glucose tolerance test, blood pressure, anthropometric measurements, a single 24-h dietary recall and a questionnaire on health behaviors. The findings from this survey, reported in the accompanying papers, inform efforts to prevent and control chronic disease among the Navajo. Lessons learned from this survey may be of interest to those conducting similar surveys in other American Indian and Alaska Native populations.
KEY WORDS:
Navajo ·
health status ·
chronic disease ·
survey
There is little reliable population-based information available to help understand the increasing prevalence of chronic disease among the Navajo during recent decades. Several surveys of nutrition and health status of small population centers have been conducted on the Navajo Reservation in the past, but none has systematically described the entire reservation-based Navajo population, nor have any included a comprehensive examination of a wide variety of health and nutrition topics. Carpenter and Steggerda (1939)
studied food habits of five Navajo families between 1934 and 1936, and Darby et al. (1956) studied nutrition and health status in two reservation communities in 1955. More recent descriptive surveys of the prevalence of diabetes and chronic disease were conducted in a single Navajo community in 1988 (Sugarman and Percy 1989, Sugarman et al. 1990 and 1992). The National Healthand Nutrition Examination Surveys, which provide representative health status information for the United States as a whole, exclude American Indians and Alaska Natives living on reservations.
The enactment of statutory requirements for evaluating progress toward meeting the Healthy People 2000 objectives for American Indians and Alaska Natives (U.S. Department of Health and Human Services 1991, P.L. 102-573 Reauthorizing the Indian Health Improvement Act of 1976) demonstrated the dearth of useful information and fueled the interest of the Indian Health Service (IHS)4 in developing a reservation-wide, population-based assessment of the health and nutritional status of the Navajo. The IHS was particularly concerned with risk factors for chronic diseases such as diabetes, cancer, cardiovascular disease and hypertension. In 1991, the Navajo Health and Nutrition Survey was initiated by the IHS to fill some of the major gaps in knowledge about the health of this population. The information gained from the Navajo Health and Nutrition Survey was intended to provide a direct basis for making decisions about nutrition education, health care and program delivery priorities by identifying opportunities for improving the health of the Navajo as well as the reservation's public health care system. In addition, because systematic nutrition and health status information has rarely been collected for American Indian tribes (Intertribal Health Project 1996, Lee et al. 1990), this survey was also intended to assess the feasibility of and serve as a model for conducting such studies among tribes other than the Navajo.
MATERIALS AND METHODS 
Population.
The Navajo Reservation is the largest Indian reservation in the United States in both population and geographic area. In 1990, the Navajo population exceeded 200,000 (Rodgers 1993). The population has been growing at an estimated rate of 2-3%/y (Navajo Area IHS Profile 1993). The reservation encompasses ~26,109 square miles (slightly larger than the state of West Virginia) over portions of Arizona, New Mexico and Utah (Fig. 1). It is largely rural with a population density of ~7 persons per square mile.
Fig. 1.
Map of the Navajo Reservation and adjacent areas showing eight named service units (strata) and location of segments (secondary sampling units) indicated by dots. The [7] in the Tuba City service unit and the [6] in the Shiprock service unit indicate that the towns of Tuba City and Shiprock contributed multiple segments to the survey.
[View Larger Version of this Image (22K GIF file)]
The Navajo Nation government is organized into executive, legislative and judicial branches. The reservation is divided into 110 local political units called chapters, each administered by elected officials. The Navajo Nation Council (Judicial Branch) is organized into 12 committees. The Health and Social Services Committee, which commissioned the Navajo Health and Nutrition Survey, oversees issues related to health care. Health care services are provided primarily by the IHS, which serves the reservation and surrounding areas in eight administrative and geographic health care jurisdictions called service units.
The population from which the Navajo Health and Nutrition Survey sample was drawn was defined as all Navajo people at least 12 y of age who normally reside within the main Navajo Reservation or in immediately adjacent communities. The study was limited to the communities served by the eight Navajo Area IHS service units. Three Navajo chapters separate from the main reservation, enumerated separately in censuses and under the jurisdiction of the Albuquerque IHS Area Office, were not included in this survey. Planning and design of the Navajo Health and Nutrition Survey began in 1989, and the 1990 Navajo census was not published until 1993. Based on the 1980 Navajo census, we produced a direct enumeration of 130,200 Navajo residents of the survey area age 12 and older.
Sampling.
Eligible households were randomly selected using a three-stage cluster design. The design was adapted from that used for a radon survey conducted on the Navajo Reservation in 1989-1990 by the Navajo Nation Environmental Protection Agency (Navajo Nation Division of Natural Resources 1989) in cooperation with the Navajo Area IHS Office of Environmental Health and Engineering. The radon survey provided maps, aerial photographs, population distribution data and a sampling algorithm that was adapted for this survey.
Sampling was conducted in all eight service units. The two smallest service units were each paired with a nearby larger service unit because individually they would not have yielded enough respondents to provide adequate precision for service-unit-level analyses. The Crownpoint and Fort Defiance service units were paired, as were Winslow and Tuba City. These pairs and the four remaining service units were treated as the primary strata. During data analysis, appropriate sample weights for each service unit were applied to take into account the sample design (Table 1).
|
Table 1.
Sample weights for the Navajo Health and Nutrition Survey
[View Table]
|
The first stage of sampling consisted of the systematic selection of census enumeration districts within each service unit, with the probability of selection being proportional to the population of the enumeration district. Between 5 and 17 enumeration districts were chosen per service unit, depending on the population of the service unit. In The Shiprock and Tuba City service units, the towns of Shiprock and Tuba City contributed six and seven enumeration districts, respectively, to the sample; each is represented by a single dot plus a bracketed number on the map (Fig. 1). The second stage consisted of selecting one segment within each enumeration district. In the 1990 radon survey, reservation enumeration districts had been partitioned into segments with roughly equal numbers of households and one segment from each enumeration district was selected at random to be included. The Navajo Health and Nutrition Survey used the same segments previously selected in the radon survey.
The third sampling stage consisted of selecting clusters of 10 households within selected segments. Households were selected at random from the radon survey maps and aerial photos when possible; in a few cases, maps from the radon survey could not be located, necessitating creation of new segment maps. A household was defined as a group of individuals sharing income and eating together. Approximately one-half of the participants were interviewed in towns, where each household usually occupied a discrete dwelling. In rural areas, a household usually consisted of a camp composed of one or more dwellings that housed related individuals.
In each segment, all dwellings or camps on the map were numbered, and one was randomly chosen as a starting point, with that dwelling or camp and the nearest nine on the map to be included in the sample. If a selected dwelling or camp was unoccupied, the next closest was substituted. When a segment contained fewer than 10 occupied dwellings or camps, interviewing continued into a contiguous segment in the enumeration district. All residents 12 y of age and older living in selected households were targeted for interview. Based on a 1989 pilot study in the Shiprock service unit, we anticipated a nonresponse (refusal and noncontact) rate of about 20% and an average of 2.8 eligible individuals per household.
Questionnaire development.
The questionnaire used pertinent questions from a variety of other surveys, including the Second and Third National Health and Nutrition Examination Surveys (McDowell et al. 1985, National Center for Health Statistics 1994), Behavioral Risk Factor Surveillance Surveys (Remington et al. 1988), the Youth Risk Behavior Survey (Kolbe 1990), the IHS Adolescent Health Survey (Blum et al. 1992) and the IHS Health Risk Appraisal (Welty 1988 and 1989). Questions were modified to be translated easily into the Navajo language and to be culturally appropriate. The questions elicited information about education and socioeconomic status, pregnancy and childbirth, cancer screening, weight management, diabetes, hypertension, heart disease, tobacco use, alcohol and drug use, exercise and eating habits. The draft questionnaire was reviewed by the Navajo Area Institutional Review Board; modifications requested by the Board were implemented to achieve a test version of the survey instrument. A Navajo translation was developed, standardized, back-translated and then field-tested during the 1989 pilot study. Questionnaires were administered in Navajo or English according to the preference of participants.
Staff training.
Interviewers were all bilingual Navajo members of the local communities who had previous experience with venipuncture. Fourteen interviewers completed an intensive 40-h training course. The team of trainers consisted of staff from the IHS and from the National Center for Health Statistics of the Centers for Disease Control and Prevention (CDC). Following the classroom training, the interviewers were assigned to work in IHS hospital and clinic laboratories for 6 wks or until the laboratory supervisor certified them competent in collecting, processing and storing blood samples. Competency evaluations were conducted for each interviewer and periodic retraining sessions were held throughout the study period.
Survey protocols.
The scarcity of telephones and variable frequency of mail pick-up made personal visits necessary for all contacts. Interviewers therefore visited selected households to explain the study and recruit participants. If, after three attempts to recruit participants, the interviewer was unable to contact any members of a selected household, its members were not included and the next household in the segment was selected. Interviews were scheduled for a future date, usually within 1 wk of initial contact. Each household was visited again the day before the scheduled interview to remind participants to be fasting for the interview. Participation followed a fully informed consent procedure approved by the Health and Social Services Committee of the Navajo Nation Council, members of the Navajo tribal judicial system, and the Navajo Area Institutional Review Board.
Outright refusals to participate were uncommon. More frequently, individuals or families were not at home for the scheduled appointment. If families who had agreed to participate were not at home for a scheduled appointment, one subsequent attempt was made to recontact the household and reschedule the appointment. If some but not all of the eligible individuals in a household were present on the day of the interview, interviews were completed with all eligible individuals present and one additional attempt was made to recontact and reschedule with the remaining member(s) of the household. No information was available on individuals or families who refused to participate or who failed to appear for an interview after consenting to participate.
On the interview day, pairs of interviewers visited the homes and administered an oral glucose tolerance test to participants who were not pregnant or had not previously been diagnosed with diabetes mellitus. The interviewers measured blood pressure and anthropometric dimensions by using standardized protocols and calibrated equipment; interview forms were completed during the intervals between blood draws. Anthropometric measurements included standing height without shoes; weight without shoes or heavy outer clothing; mid-arm circumference; waist and hip girths; three skinfolds (triceps, suprailiac and subscapular); and wrist and elbow breadths. The interview comprised 103 questions on health risk behaviors and a 24-h diet recall. The interviewers' visits, which were conducted on weekdays, lasted 2.5-3 h. Each participant was paid $10.
Following completion of the interviews and receipt of laboratory test results, physicians reviewed all of the findings to ensure that any existing significant medical conditions were identified and that individuals were notified and referred for appropriate medical care. Individuals with atypical values were recontacted and referred for appropriate follow-up.
Blood testing.
For the oral glucose tolerance test, samples of venous blood were drawn after an overnight fast (17-24 h) and at 1- and 2-h intervals after the administration of 75 g of glucose in water. Participants excluded from the oral glucose tolerance test had only a fasting blood draw. A basic 28-item blood chemistry profile and complete blood count were performed for all participants from the fasting blood sample, including total cholesterol and tryglycerides, measured with a Technicon SMAC-III (Technicon Instrument, Terrytown, NY). Plasma glucose and high density lipoprotein levels were assayed with the Technicon RAXT (Technicon Instrument). Specimens were transported by courier daily to the reference laboratory (Corning Clinical Laboratories, El Paso, TX), which participates in the profiency program conducted by the College of American Pathologists. For a subsample of 290 participants, blood lead and selected serum nutrients were assayed; these tests were limited to participants who lived near an overnight shipping terminal and a supply of dry ice. Blood samples for the specialized tests were frozen and maintained below 
20°C until shipping. Frozen specimens were packed in dry ice and shipped overnight to the Environmental Health Laboratory, National Center for Environmental Health, CDC (Atlanta, GA) for analysis.
Data management.
Interviewers recorded all data directly onto standardized recording forms. Questionnaire data were entered by two of the interviewers following completion of all data collection with the use of EpiInfo Version 5.01B software (CDC, Atlanta, GA). The 24-h diet recalls were entered into the Food Intake Analysis System (FIAS version 2.3) software (University of Texas Health Sciences Center at Houston, Houston, TX). Data files were exported from EpiInfo and FIAS to SAS files (SAS Institute, Cary, NC) for analysis. All point estimates were calculated using sampling weights (Table 1). Standard errors were adjusted for clustering within sampling units by SUDAAN (Shah 1991).
RESULTS
Individuals living in 459 (60%) of the 760 target households participated in the survey, yielding a total of 985 participants, or 58% of the target number of 1690; this percentage serves as an approximation of the individual participation rate. The final numbers of interviews in service units were too small to support analyses at the level of the service unit. For the 985 participants, the data are more than 90% complete for the anthropometric and physiologic measurements. A small proportion of individuals declined some measurements or refused to answer some questions, but, overall, the cooperation among participants was high. Questions about physical activity were difficult for the respondents and more than half were unable to adequately quantify frequency, distance or duration of activities. Questions about individual and family income were declined by many participants. Among the health behavior and risk factor questions, alcohol and drug use was a sensitive topic, and half or more of the participants declined to answer some of those questions. Because of the high rate of nonresponse to income, alcohol and drug questions, we will not present data on these variables. In the papers in this special issue of The Journal of Nutrition, analyses are based on information from all participants who provided usable data for a particular measurement or question; thus, the sample sizes differ slightly within and among papers.
Men ages 21-64 are underrepresented relative to women in this survey, on the basis of a comparison with the 1990 Navajo Nation Census (Rogers 1993, Table 2). The sex ratio (male/female) among survey participants declines from 0.70 in census age group 21-29 y to 0.46 in census age group 50-64 y. For the age strata 21-29 y, 30-39 y, 40-49 y and 50-64 y, the 95% confidence intervals around the estimated sex ratios for the Navajo Health and Nutrition Survey participants exclude the sex ratios obtained by direct enumeration from the Navajo Nation census. Among adolescents age 14-20 y and among elders age 65 y and older, the sex ratio for the survey participants is not significantly different than that of the census. Adolescents are underrepresented relative to the distribution of age strata in the census, whereas elders are slightly overrepresented (Table 3). The sex and age distributions of the participants reflect primarily who was at home when the interview team visited, although some of the departure from the expected sex and age distributions may also be due to individual refusal to participate. However, the interviewers did not record individual absence and refusal rates within each household. Interviews were conducted on weekdays throughout the year, including the school year when adolescents might be expected to be at school. Men may have been absent on weekdays because of employment and other activities outside the home.
|
Table 2.
Comparison of the Navajo Health and Nutrition Survey sample sex ratio to 1990 Navajo Nation Census
[View Table]
|
|
Table 3.
Comparison of the Navajo Health and Nutrition Survey sample age distribution to 1990 Navajo Census1
[View Table]
|
Questions about household characteristics were answered by the self-identified head of the household; only one head-of-household responded for each survey household. Thirty-four individuals responded as the head of a one-person household; data needed to assign 29 individuals to households were missing, and these individuals did not answer the head-of-household questions themselves.
Nearly half of the heads-of-households reported that only Navajo was spoken in their household (Table 4). As the age of the head-of-household increased, the proportion who reported speaking only Navajo at home increased sharply. About two thirds of households had electricity, although we are not able to distinguish electricity supplied by power lines from generator-supplied electricity. Almost three fourths of the homes had televisions and half had VCRs as well. More heads-of-households reported owning televisions than reported electricity; we believe that some of these homes had generator-supplied electricity not reported in response to the question about electricity. About half of the heads-of-households reported having refrigerators and running water, but only 16% reported having telephones. More than 90% of all households had radios. Apart from radios and telephones, households headed by individuals age 60 and older tended to have fewer modern appliances and amenities than households headed by younger individuals, although only the differences in the presence of televisions and VCRs were statistically significant (P < 0.001). Houses were heated primarily by wood, coal or propane gas, and many homes had more than one source of heat. Fourteen percent of households participated in the Food Distribution Program for Indian Reservations (FDPIR) only, 19% received Food Stamps only, and 8% of heads-of-households reported members participating in the Special Supplemental Food Program for Women, Infants and Children (WIC) only. Although FDPIR and Food Stamp participation are mutually exclusive, WIC participants may be eligible for another support program as well. Four percent reported receiving both WIC and FDPIR, and 14% reported receiving both WIC and Food Stamps. Households headed by individuals age 60 and older participated in the FDPIR more frequently than other households, whereas households headed by individuals under age 40 more often participated in the Food Stamp program or received WIC with or without other assistance (P < 0.001). Many of these household characteristics are often useful proxies for socioeconomic status or isolation and may be associated with variation in health status. However, these characteristics were not associated with significant variation in the participants' nutritional intake, physical measurements or educational attainment in this sample (data not shown).
|
Table 4.
Characteristics of households included in the Navajo Health and Nutrition Survey, 1991-92
[View Table]
|
More than half of the households contained children under age 12 years, 41% contained teens age 12-18 y and 30% contained elders age 60 and older. Households headed by individuals age 60 and older least often contained children or teens, although one third of these households contained children and one fourth contained teens. The most common household composition was adults with children or teens (65%), followed by three-generation households (13%), adults and elders (12%), elders only (8%), and elders and children or teens (2%).
DISCUSSION
In general, those Navajo who agreed to participate in the survey accepted it well, submitting to the multiple blood draws for the oral glucose tolerance test, responding to long questionnaires and giving dietary recalls. Among the anthropometric measurements, only subscapular skinfolds were refused by many participants (8%), possibly primarily because of modesty, perhaps because it was not always possible to provide optimal privacy for these measures. Questions about individual and household income were declined by most participants, and many participants also refused questions about drug and alcohol use.
The survey was designed to yield a representative sample of Navajo living on or near the reservation. Absence from home on the day of the interview or refusal to participate produced a sample with fewer adult men age 20-59 y, fewer adolescents age 12-19 y of both sexes and more elders age 60 and older of both sexes than expected, on the basis of the 1990 Navajo census (Rogers 1993). It is possible that scheduling interviews almost exclusively on weekdays contributed to the lower participation of adult men, who might have been at work, and of adolescents, who might have been at school. Future surveys should consider the inclusion of weekend sampling. However, many Navajo are engaged in traditional farming and herding activities rather than wage employment, and their schedules are not as strongly tied to the weekday/weekend distintion as those of other groups. In addition, in the households that reported residents age 12-19, many adolescents were present and participated in the interviews, even on school days. Nevertheless, the sample is not entirely representative, and we have no data to determine whether absence or refusal were systematically related to any of the variables measured. Extrapolation from the sample concerning the characteristics of adult men and adolescents of both sexes should be interpreted with caution.
We encountered difficulties in quantifying two important variables usually associated with health status and health behavior. First, questions about physical activity proved to be ambiguous for the participants. Second, in contrast to many other populations, markers of socioeconomic status such as participation in nutritional support programs or the possession of amenities and appliances in the houshold may not be good indices of important variation in lifestyle or access to resources among Navajo living on the reservation. In future surveys, researchers might test alternative questions such as the distances from residences to grocery stores, trading posts and health care facilities; the types of surfaces on roads to residences; and transportation available to household members.
As the health care system becomes increasingly oriented toward prevention, the need for improving methods of collecting health information to support preventive care will increase. A primary purpose of this survey was to initiate a system of ongoing health status measurements that would prompt corresponding refinements in the health care delivery system. Essential to this process is an understanding of the effects of relevant environmental factors. For example, how good is access to health care, how well do providers respond to language abilities and cultural values of potential care recipients, and how available are resources for delivering and evaluating clinical programs and health information? With fewer than half of the heads of households reporting English spoken in their household, the Navajo language should be thoroughly incorporated into the delivery of services to this population. We found that telephones were inadequate for a population-based survey because only 16% of the households had telephones. We conclude that follow-up for health care services among the Navajo should also not rely on the telephone. Radio is a potentially effective way of sharing health information because more than 90% of the households had radios, although no data were collected about listening habits. Videotapes should be considered as well to communicate health information to this population because many households had televisions and VCRs.
Most information about the prevalence of disease and behavioral risk factors among American Indians and Alaska Natives is based on user data from the IHS health system. Such data are limited by the illness and service usage patterns of patients accessing health care facilities, however, and thus do not reflect the entire population. In some of the papers that follow, for example, it is apparent that IHS user statistics underestimate the magnitude of diabetes and hypertension because this survey found substantial undiagnosed diabetes and hypertension. The Navajo Health and Nutrition Survey demonstrates that population-based surveys of this type are feasible and can be useful for tribal-specific health planning. It also shows the feasibility of training local interviewers to gather the data and indicates that cultural acceptability of such surveys by study participants can be high. Other population-based surveys conducted among American Indian tribes have had similarly high response rates (Goldberg et al. 1991, Warren et al. 1990), although those studies did not include physical measurements or the collection of blood. However, in addition to periodic population-based health status surveys, it is essential to establish short-term process evaluation mechanisms that provide rapid feedback to communities and organizations implementing intervention programs.
With standardization of methodology for surveys of this type, comparisons of risk factors could be made between tribes as well as over time within tribes. This type of activity will become particularly important to support the development and evaluation of prevention and intervention programs. Repeat surveys using this design would be advisable to track progress in Navajo nutrition, health status and behavioral factors affecting health outcomes. This methodology could also be used as a template for accomplishing such an evaluation in other tribes.
ACKNOWLEDGMENTS
The investigators wish to express our thanks to the following people, without whom the study would not have been possible.
Indian Health Service Headquarters
Leo J. Nolan, M.Ed.
Acting Associate Director of Planning, Evaluation and Legislation
Navajo Area Indian Health Service
Michael W. Everett, Ph.D.
Statistician, Navajo Area Office, Window Rock, AZ
Shelley Frazier
Health Promotion Assistant, Northern Navajo Medical Center
Shiprock, NM
Navajo Nation
Larry Rodgers
Statistician, Navajo Nation Division of Community Development
Window Rock, AZ
Service Unit Field Coordinators
Geri Bahe-Hernandez, R.N.
Director of Community Health, Tuba City Service Unit
Lois J. Benson, R.D., C.D.E.
Scottsdale Memorial Hospital
Scottsdale, AZ
Kayenta Service Unit
Susan Bradway, M.S., R.D.
Public Health Nutritionist, Crownpoint Service Unit
Gay Crawford, R.D., M.P.H.
Public Health Nutritionist, Gallup Service Unit
Park W. Gloyd, M.D., M.P.H.
Director of Community Health, Chinle Service Unit
Joseph Piepmeyer, R.N., R.D., M.Ed.
Public Health Nutritionist, Winslow Service Unit
Susan Velazquez, R.D., M.P.H.
Public Health Nutritionist, Fort Defiance Service Unit
Interviewers
Vivian Bahe
Samantha Begaye
Ruby Curley
Henrietta Dale
Myra Dalton
Lucy Day
Ann Duncan
Roberta Hardy
Anna Johnson
LaRia Kanuho
Treva McKinley
Rosalita Sloan
Locita Sneddy
Ben Tsosie
Antonelle Yazzie
Elva Yazzie
Nutrition Data Entry
Patricia Osif
Kayenta Service Unit
Centers for Disease Control and Prevention
Interviewer Training
Catherine M. Loria, Ph.D.
Office of Analysis, Epidemiology, and Health Promotion
National Center for Health Statistics
Robert Kuczmarski, Dr.P.H., R.D.
National Health and Nutrition Examination Survey
National Center for Health Statistics
Editorial Support
Kelley Scanlon, Ph.D.
Laurence Grummer-Strawn, Ph.D.
Division of Nutrition and Physical Activity, National Center for
Chronic Disease Prevention and Health Promotion, Centers for
Disease Control and Prevention
Graphics Support
Rose Pecoraro
Amy Simmons
Office of the Director, National Center for Chronic Disease
Prevention and Health Promotion, Centers for Disease Control and Prevention
Technical Information Services Support
Brenda Mazzocchi, M.L.S.
Technical Information and Editorial Services Branch,
National Center for Chronic Disease Prevention and Health Promotion,
National Centers for Disease Control and Prevention
Editorial Support
Peter L. Taylor, M.B.A.
Editor, Cygnus Corporation
Rockville, MD
![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
,
,
ABSTRACT
