Results of a Prevalence Survey
of Macomb County, 1999

The Behavioral Risk Factor and Health Needs Survey of Macomb (or Macomb BRFS) was conducted as a series of Computer Assisted Telephone Interviews (CATI) by the Survey Research Division of Michigan State University's Institute for Public Policy and Social Research. The study was designed as a cross-sectional survey of adults and households in the county.

Individuals were selected in a two stage sampling plan using random-digit dial procedures. The first stage of this plan involved randomly selecting households in which to interview an individual. The second stage involved randomly selecting a respondent from among the adults living within a selected household.

For the first stage, we purchased a stratified random-digit dial (RDD) sample of phone numbers from Survey Sampling, Inc., of Fairfield, CT. This organization enjoys an excellent reputation for producing high quality, representative samples of populations and, in fact, has been the vendor supplying BRFS samples to SRD for the Michigan Behavioral Risk Factor Survey during the past three years. Survey Sampling maintains a national telephone number database including all possible numbers whether listed or not, whether household or business, whether working or not. Based on Census information, it is estimated that 96+% of Michigan households have telephone subscriptions. To construct the sampling frame, Survey Sampling begins with the list of currently active area code + exchange combinations obtained from the telephone companies. They append to these combinations all possible blocks of 100 phone numbers represented by the first two digits of the four-digit suffix that would comprise a complete telephone number.

Blocks are linked to counties and states and zip codes based on the directory listings and information from the telephone companies. When geographic areas are defined for sampling, Survey Sampling selects telephone number blocks systematically with a random start from among those linked to the defined area. Only those blocks associated with Macomb County were included in the initial sampling frame from which actual numbers were selected. For this survey, we used a stratified sampling design so that we could improve the efficiency of the sample with respect to finding working household phone numbers.

To accomplish this, Survey Sampling compared each of these blocks against its database of updated directory listings for over 60 million households. Any block that was not represented among the directory listings or which had fewer than two listings was separated into one stratum as a sampling frame from which 1,000 telephone numbers were selected at random. Those blocks which contained two or more listed telephone numbers were separated into a second stratum as a sampling frame from which 5,625 telephone numbers were selected at random.

All of the numbers selected were then compared against Survey Sampling's database of businesses and of known disconnected numbers. Those numbers that were identified as belonging to businesses or as disconnected were flagged as such so that they did not require calling, but were transmitted to SRD with all other selected numbers as well. Of the numbers selected and not flagged as businesses or disconnected numbers, Survey Sampling expects that, on average, about 60-75% of the phone numbers generated at random will be working household numbers. The remainder of the numbers are expected to be either businesses, agencies, institutions, or not in service.

As with the BRFS, within each of the households selected, one of the adults (18 years old or older) residing in the household was randomly selected to be interviewed. A modified version of the Kish technique was used to select randomly the adult respondent. This procedure involves asking an informant within the household to list all of the adult household members and then one of the individuals listed is randomly selected based on a random number that was generated and assigned to the phone number in advance.

After the RDD sample of telephone numbers was selected by Survey Sampling, SRD had SSI cross-check all numbers against their database of listed household phone numbers. Where ever there was a match between an RDD selected number and a directory listing, SSI extracted the name and address of the phone subscriber and appended it to the telephone number included in the telephone number sample file. Matches were found for approximately 34% of the phone numbers drawn from the 2+ banks. SRD sent advance letters approximately one week prior to the release of the phone number for calling to the individual at the listed address for all those cases where a listing was obtained. The advance letter notified the household that they would be receiving a call from an SRD interviewer on behalf of the Macomb County Health Department, what the call would be about, what the research is trying to accomplish, how they were selected, and whom they could contact for verification or additional information. The letter also told them that their participation was voluntary, but that any information they provided would be kept strictly confidential. Other research indicates that such letters often reduce the number of refusals and increase cooperation with the survey effort.

The interview included a core of items taken directly from the Behavioral Risk Factor Survey (BRFS) instrument developed by the Centers for Disease Control. It also included items not included in the CDC instrument but which were added from previously administered Macomb County Behavioral Risk surveys.

The interview questions can be divided into roughly nine categories: health status items (including an overall subjective assessment of health status, days of ill health per month, obesity, hypertension, high cholesterol, diabetes, asthma, other breathing problems, and other major health problems),health care access and utilization (both medical and dental), and health behaviors divided into two categories -- those related to health risks (i.e., seatbelt non-use, smoking, alcohol consumption, drinking and driving, physical activity, and sexual behaviors associated with HIV risk) and those related to prevention or early detection (i.e., seeing health professionals regularly, getting check-ups, vaccinations, and screenings exams (i.e., mammograms, pap smears and breast exams (for women), prostate-specific antigen test (for men), blood stool occult test, digital rectal exam, and proctoscopy/sigmoidoscopy. The fourth category of items concerned spiritual health. The fifth category of items focused on health care for elderly residents (i.e., functional status of elderly household members, familiarity with elder care services, involvement in caregiving, and involvement in retirement and estate planning). The sixth category of items focused on behaviors and attitudes regarding environmental protection and recycling, while the seventh examined respondents' utilization and rating of services provided by MCHD. The final category included the demographic questions and questions regarding health promotion and worksite wellness programs provided through respondents' employers.

Interviewing began February 25, 1999 and continued through June 18, 1999. Unless contact with a household occupant was made, telephone numbers were called a minimum of 12 times across time blocks of the day and days of the week before supervisory review. Depending on the review, supervisors may have discarded the number at that point or released the number for up to three additional call attempts before discarding the number as a probable "not in service" number. If contact with the household occurred on any of the call attempts, then interviewers continued to try to contact the household until either an interview was completed or there was a refusal. In the event of an initial refusal, interviewers tried to recontact the household a number of days later to attempt to convert the household into completing the interview. Refusal conversion attempts were made in the case of initial refusals except where the initial refusal was adamant or strongly indicated little likelihood of a successful conversion.

Back To Top