Appendix A: Methods and Sample Results
The Michigan Behavioral Risk Factor Surveys are conducted through a cooperative agreement with the Centers for Disease Control and Prevention and have followed the overall CDC survey protocol51 for the Behavioral Risk Factor Surveillance System (BRFSS) since 1987. In 1996, two BRFSs were conducted in Michigan by the Institute for Public Policy and Social Research at Michigan State University for the Michigan Department of Community Health.
Both 1996 BRFSs were conducted using the same sampling design and the same survey protocol. Each month in 1996, a random sample of approximately 210 Michigan adults 18 years and older was selected for each of the two surveys and interviewed by telephone from January through December.
A disproportionate stratified random sample was used for the 1996 BRFSs, with the stratification based on phone bank density. The higher density stratum included all possible phone numbers in 100-banks with at least two directory-listed numbers; the lower density stratum included those in 100-banks with one or fewer directory-listed numbers. The call design consisted of up to 15 calls to reach a household and up to a total of 20 calls to select and interview the eligible respondent. Calling was distributed across a combination of daytime hours during the week, weekday evenings, and weekends. Once a household was contacted, the intrahousehold selection procedure involved the enumeration of all household members 18 years and older and the random selection of one adult member as the eligible respondent. The monthly target sample size was 210 completed interviews, resulting in an annual sample size of approximately 2500 per survey.
The survey instruments were designed so that a core set of questions dealing with some of the main BRFS risk indicators overlapped. This overlapping instrument design allowed for more precise estimates for these main indicators, as well as allowing for a broad range of questions to be included. Combined data from both 1996 BRFSs were used in the preparation of this report. The core of the first 1996 Michigan BRFS (BRFS-1) instrument was developed by the CDC and included questions on health status, health care access, diabetes, cigarette smoking, breast and cervical cancer screening, body weight and height, HIV/AIDS, leisure-time physical activity, fruit and vegetable consumption, and weight control.g The BRFS-1 instrument also included state-added questions on high fat food consumption, injury prevention (smoke detectors, child car safety seats, bike helmets), activity limitations, and oral health. The second 1996 Michigan BRFS (BRFS-2) questionnaire included an abbreviated core of questions selected from the CDC core. In addition, the BRFS-2 instrument also included questions about blood pressure, blood cholesterol, non-leisure-time physical activity, cardiovascular disease risk prevention counseling, aspirin use, estrogen use, contraceptive methods, and alcohol consumption.
The prevalence estimates for overweight presented in this report were based on body mass index (BMI) as calculated from the self-reported weight and height measurements. Body mass index is defined as weight (in kilograms) divided by height (in meters) squared [weight in kg/(height in meters)2]. The BMI reference standards used to categorize weight status were based on the Second National Health and Nutrition Examination Survey 1976-1980 (NHANES II) sex-specific BMI distributions for persons 20-29 years of age.47 The definition of "overweight" was a BMI greater than or equal to the 85th percentile (i.e., 27.8 kg/m2 or greater for men and 27.3 kg/m2 or greater for women). Errors in self-reported weight and height appear to be related to gender, age, and overweight status,48,49 which tend to result in an underestimate of the prevalence of overweight using self-reported measurement data.
g Questions on blood pressure, cholesterol, colorectal cancer screening, alcohol consumption, injury prevention, and immunizations were not included in the 1996 core CDC questionnaire; they are rotated in and out of the CDC core questionnaire on a biennial basis and were included again in the 1997 CDC BRFSS instrument.
The 1996 BRFS data were initially weighted to adjust for the probabilities of selection. A poststratification weighting factor that conditionally adjusted for the 1995 Michigan intercensal population estimates by age, sex, and race was then applied after the initial weighting. Both the initial weighting factor and the poststratification weighting factor were trimmed at approximately the 3rd and 97th percentile of their distributions to minimize weighting loss.
Calculations of the prevalence figures and confidence interval limits were performed using SUDAAN, a statistical computing program that was designed for analyzing data from multistage sample surveys.50
Unless otherwise specified, respondents who answered that they did not know or refused to answer were not included in the calculation of percentages of the population considered "at risk." Weighted proportions of respondents who reported risk behaviors are presented in this report by categories of age, sex, race, education, and income. The age, sex, race, and educational attainment refer to the respondent. Income, however, reflected the reported annual income of the household from all sources, regardless of the respondent's income and the number of individuals in the household. Sample sizes used to calculate the estimates included in this report varied and are reflected in the confidence intervals presented with each estimate.
A total of 13,005 telephone numbers were used for the 1996 Michigan BRFS-1. The final call dispositions for the sample numbers fell into the following categories: 2,476 completed interviews, 1,415 refusals, 5,890 nonworking numbers, 598 ring-no-answers, 1,784 businesses, 251 households reached but no members eligible, 429 eligible respondents selected but not interviewed, 34 informants/eligible respondents with language barriers, 25 interviews terminated at some point, 21 busy numbers, and 82 informants/eligible respondents unable to participate. Seventy respondents started but did not complete the interview (e.g. interview discontinued because of impairment or language barrier; interviewer discontinued and then upon callback the phone number was nonworking, respondent refused, or was unable to be reached again). These "incomplete" respondents were included in the final call distribution categories above, and the data from these records were included in the analyses to the extent possible. The CASRO (Council of American Survey Research Organizations) response rate,51 which includes a portion of the dispositions with unknown eligibility in the denominator of the rate, was 54.3%. Fifty-three percent (52.5%) of all household contacts resulted in a completed interview; 30.0% of all household contacts refused.
For the 1996 BRFS-2, a total of 12,968 telephone numbers were used. The final call dispositions for the sample numbers fell into the following categories: 2,463 completed interviews, 1,314 refusals, 5,765 nonworking numbers, 531 ring-no-answers, 1,909 businesses, 195 households reached but no members eligible, 608 eligible respondents selected but not interviewed, 85 informants/eligible respondents with language barriers, 13 interviews terminated at some point, 10 busy numbers, and 75 informants/eligible respondents unable to participate. Seventy-seven respondents started but did not complete the interview; they were included in the analyses to the extent possible. The CASRO response rate for BRFS-2 was 53.7%. Nearly 52 percent (51.8%) of all household contacts resulted in a completed interview; 27.6% of all household contacts refused.
The distributions of the samples from the 1996 Michigan BRFS-1 and BRFS-2 combined are presented in Tables A.1 (by age, gender and race) and A.2 (by education, household income, employment status among those aged 20-64 years, and marital status). The unweighted sample tended to underrepresent younger adults, males, and African Americans. After weighting, the sample underrepresented those with less formal education and those living in households with lower annual incomes, as compared with distributions from the 1990 census.
Interpretation of Results and Limitations of the Survey
All results from the 1996 Michigan BRFS presented in this report have been weighted as described above and can be interpreted as estimates of risk and healthful behavior prevalence among the general adult population of Michigan. As with all survey estimates, these estimates are subject to sampling error (as well as other types of error) and therefore might differ from other survey estimates even if the surveys were conducted during the same time period, using the same target population and survey protocol. The confidence intervals presented are an attempt to quantify this sampling error and to aid the reader in using these results as one set of estimates for the prevalence of health risk behaviors among the Michigan adult population.
One limitation of telephone surveys is the lack of coverage of that portion of the population who live in households without telephones. The magnitude of bias from this undercoverage would be dependent upon both the magnitude of the undercoverage and the magnitude of the difference between the characteristics of those covered and those not covered by the sampling frame. The literature indicates that differences do exist between those who live in households with and without telephones and that "the most powerful correlate of telephone subscription is family income."52 Therefore, the potential for undercoverage bias would probably be greatest among those risk prevalence estimates that are associated with household income status; this undercoverage bias would tend to result in overestimation of those risks that tend to increase with higher income categories and underestimation of those risks that are inversely related to income. According to figures from the 1990 census, approximately four percent of Michigan households do not have telephones. This nontelephone rate, however, is not consistent across all subgroups within Michigan. Approximately nine percent of African-American households in Michigan do not have telephones.53
Another potential source of bias results from the self-reported nature of these data. It would be expected that respondents might tend to underreport health risk behaviors, and especially those risk behaviors that are illegal or socially unacceptable. Compared with estimates obtained from observational studies, median BRFSS estimates for seat belt use were approximately 20 percentage points higher, when defined by "always or nearly always" using seat belts,54 while median BRFSS estimates were only 2-5 percentage points higher when seat belt use was based on "always" using.55 Despite the problems of underreporting of alcohol consumption, state-level comparisons of alcohol sales data have been shown to correlate with the BRFSS self-reported estimates of heavy drinking, binge drinking, and drinking and driving.56 Self-reporting biases have been shown for height and weight, the magnitude and direction of which appear to vary with gender, age, and relative weight status.49 In a study from the New York BRFS, prevalence of selected self-reported, CVD-related risk factors tended to underestimate prevalence as estimated from measured data.57
Overall, however, the disadvantages of the telephone survey methodology are thought to be outweighed by the advantages, which include the quality control over data collection made possible by a computer-assisted-telephone-interviewing system, cost efficiency, and the speed at which data can be collected.58,59
1. CDC. Addition of prevalence of cigarette smoking as a nationally
notifiable condition. MMWR
1996;45:537.
2. US Department of Health and Human Services. The health
consequences of smoking: Nicotine
addiction: A report of the Surgeon General. US Dept of Health and Human Services
publication, 1988 [(PHS) 88-8406]. Rockville MD, Office on Smoking and Health.
3. McGinnis JM, Foege WH. Actual causes of death in the United States. J Am Med Assoc 1993;270:2207- 2212.
4. U.S. Department of Health and Human Services. Reducing the Health Consequences of Smoking - 25 Years of Progress, A report of the Surgeon General. Washington: US Government Printing Office, 1989. (DHHS Publ. No [CDC] 89-8411.)
5. Dobson AJ, et al. How soon after quitting smoking does risk of heart attack decline? Journal of Clinical Epidemiology 1991;44:1247-1253.
6. Troiano RP, Frongillo EA, Sobal J, Levitsky DA. The relationship
between body weight and mortality: a quantitative analysis of combined information from
existing studies. Int Obes Relat Metab Disord
1996;20:63-75.
7. Seidell JC, et al. Overweight, underweight, and mortality. A prospective study of 48,287 men and women. Arch Intern Med 1996;156:958-963.
8. Manson JE, Willett WC, Stampfer MJ, Colditz GA, Hunter DJ, Hankinson SE, Hennekens CH, Speizer FE. Body weight and mortality among women. N Eng J Med 1995;333:677-685.
9. Lee IM, Manson JE, Hennekens CH, Paffenbarger RS. Body weight and mortality: A 27-year follow-up of middle-aged men. J Am Med Assoc 1993;270:2823-2828.
10. National Institutes of Health Consensus Development Panel on the
Health Implications of Obesity. Health implications of obesity: NIH consensus development
conference statement. Ann Intern Med
1985;103:1073-1077.
11. Pi-Sunyer FX. Medical hazards of obesity. Ann Intern Med 1993;119:655-660.
12. US Department of Health and Human Services. The Surgeon General's Report on Nutrition and Health. Washington, DC: US Government Printing Office, 1988. (DHHS pub no (PHS) 88-50210)
13. Kuczmarski RJ, Flegal KM, Campbell SM, Johnson CL. Increasing prevalence of overweight among US adults. J Am Med Assoc;1994:205-211.
14. Div. Health Examination Statistics, NCHS. Update: prevalence of overweight among children, adolescents, and adults _ United States, 1988-1994. MMWR 1997;46:199-202.
15. Galuska DA, Serdula M, Pamuk E, Siegel PZ, Byers T. Trends in overweight among US adults from 1987 to 1993: A multistate telephone survey. Am J Pub Health 1996;86:1729-1735.
16. Troiano RP, et al. Overweight prevalence and trends for children and adolescents. The National Health and Nutrition Examination Surveys, 1963 to 1991. Arch Pediatr Adolesc Med 1995;149:1085-1091.
17. Giddings SS, et al. Effects of secular trends in obesity on coronary risk factors in children: the Bogalusa Heart Study. J Pediatr 1995;127:868-874.
18. NIH Technology Assessment Conference Panel. Methods for voluntary
weight loss and control. Ann
Intern Med 1993;119:764-770.
19. Blair SN. Evidence for success of exercise in weight loss and control. Ann Int Med 1993;119:702-706.
20. US Department of Health and Human Services. Physical activity and health: A report of the surgeon general. Atlanta, GA: Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, The President's Council on Physical Fitness and Sports, 1996.
21. Blair SN, Kampert JB, Kohl HW, Barlow CE, Macera CA, Paffenbarger RS, Gibbons LW. Influences of cardiorespiratory fitness and other precursors on cardiovascular disease and all-cause mortality in men and women. J Am Med Assoc 1996;276:205-210.
22. Pate RR, Pratt M, Blair SN, Haskell WL, et al. Physical activity
and public health. J Am Med Assoc
1995;273:402-407.
23. Genest J Jr, Cohn JS. Clustering of cardiovascular risk factors: targeting high-risk individuals. Amer J of Cardiol 1995;76:8A-20A.
24. Davis MA, Neuhaus JM, Moritz DJ, Lein D, Barclay JD, Murphy SP. Health behaviors and survival among middle-aged and older men and women in the NHANES 1 epidemiologic follow-up study. Prev Med 1994;23:369-376.
25. Wynder EL. From the discovery of risk factors for coronary artery disease to the application of preventive measures. Am J Med Sci 1995;Dec.310 Suppl 1:S119-122.
26. Hunter SM, Bao W, Berenson GS. Understanding the development of
behavior risk factors for
cardiovascular disease in youth: The Bogalusa heart study. Am J Med Sci 1995;Dec
310 Suppl 1:S114- 118.
27. Sibilski C, Wucherer W. Behavioral risk factors of cardiovascular disease:A study of the citizens of Franklin, Wisconsin. Wis Med J 1995;Dec 94(12):687-689.
28. Centers for Disease Control and Prevention. Update: Alcohol-related traffic fatalities-United States, 1982- 1993.MMWR 1994;43(47):861-867.
29. Michigan Department of Community Health, Office of the State Registrar and Division of Health Statistics. Michigan Health Statistics, 1995. Lansing MI: MDCH, 1996.
30. National Diabetes Data Group. Diabetes in America. Bethesda, MD: National Institutes of Health, 1985. (U.S. DHHS Publ. [NIH] 85-1468.)
31. American Diabetes Association. Clinical Practice Recommendations
1998. Diabetes Care 1998;21
(Suppl 1): S1-S98.
32. Kaplan GA. Perceived health and mortality: A nine-year follow-up of
the human population laboratory
cohort. American Journal of Epidemiology 1963:117:292-304.
33. Linn MW, et al. Self-assessed health, impairment, and disability in Anglo, Black and Cuban elderly. Medical Care 1980;18:282-288.
34. Centers for Disease Control and Prevention. Breast cancer incidence and mortality - United States, 1992. MMWR 1996;45(39):833-837.
35. American Cancer Society. Cancer facts and figures 1996. Atlanta: American Cancer Society, 1996; publication no. 5008.96.
36. American Cancer Society. Cancer facts and figures 1998. Atlanta: American Cancer Society, 1998; publication no. 5008.98.
37. American Cancer Society. Cancer facts and figures - 1997. Atlanta: American Cancer Society, 1997; publication no. 5008.97.
38. Schofield MJ, et al. Notification and follow-up of Pap test
results: Current practice and women's
preferences. Preventive medicine 1994;23:276-283.
39. National Cancer Institute. Breast and cervical cancer screening among underserved women. Archives of Family Medicine 1995;4:617-624.
40. Institute of Medicine. Access to Health Care in America, 1993.
41. Valdiserri RO, et al. Community planning: A national strategy to
improve HIV prevention programs.
Journal of Community Health 1995;20(2):87-100.
42. Holtgrave DR, et al. An overview of the effectiveness and efficiency of HIV prevention programs. Public Health Reports 1995;110(2):134-146.
43. Webb CA, Yuhas JA. Ability of WIC clientele to estimate food quantities. J Am Diet Assoc 1988;88:601- 602.
44. Suitor CJ, Gardner J, Willett WC. A comparison of food frequency and diet recall methods in studies of nutrient intake of low-income pregnant women. J Am Diet Assoc 1989;89:1786-1794.
45. Block G, Dresser CM, Hartman AM, Carroll MD. Nutrient sources in
the American diet: Quantitative
data from the NHANES II survey. Am J Epidem 1985;122:27-40.
46. Office of Surveillance and Analysis, Center for Chronic Disease Prevention and Health Promotion. BRFSS Operations Manual 1989. Atlanta, GA: Centers for Disease Control, 1989.
47. Najjar MF, Rowland M. Anthropometric reference data and prevalence
of overweight, United States,
1976-80. Vital and Health Statistics, series 11, no. 238. Washington, DC: Public
Health Service, 1987. (DHHS Pub. No. (PHS)87-1688.)
48. Rowland ML. Self-reported weight and height. Am J Clin Nutr 1990;52:1125-1133.
49. Rowland ML. Reporting bias in height and weight data. Statistical Bulletin Metropolitan Life Insurance Company 1989;70(2):2-11.
50. Shah BV, Barnwell BG, Bieler GS. SUDAAN User's Manual, Version 6.4, Second Edition. Research Triangle Park, NC: Research Triangle Institute, 1996.
51. Office of Surveillance and Analysis, Center for Chronic Disease Prevention and Health Promotion. BRFSS Operations Manual 1989. Atlanta, GA: Centers for Disease Control, 1989.
52. Groves RM. Survey errors and survey costs. New York: John Wiley & Sons, 1989, p. 117.
53. U.S. Bureau of the Census (1990) (MI-STF#A/STF4A).
54. Robertson LS. The validity of self-reported behavioral risk factors: seatbelt and alcohol use. J Trauma 1992;32:58-59.
55. Nelson DE. Validity of self-reported data on injury prevention behavior: lessons from observational and self-reported surveys of safety belt use in the US. Injury Prev 1996;2:67-69.
56. Smith PF, Remington PL, Williamson DF, Anda RF. A comparison of alcohol sales data with survey data on self-reported alcohol use in 21 states. AJPH 1990;80:309-312.
57. Bowlin SJ, Morrill BD, Nafziger AN, Jenkins PL, Lewis C, Pearson TA. Validity of cardiovascular disease risk factors assessed by telephone survey: The Behavioral Risk Factor Survey. J Clin Epidemiol 1993;46:511-517.
58. Frey JH. Survey Research by Telephone. Newbury Park, CA: Sage, 1990.
59. Lavrakas PJ. Telephone Survey Methods. Newbury Park, CA: Sage, 1993.