The target survey sample was 3,500 beneficiaries, separated into the following 4 strata: women aged 21 to 64 years of age enrolled in the direct care system (n = 1,250); women aged 21 to 64 years enrolled in the purchased (commercial) care network (n = 1,250); men aged 51 to 64 years enrolled in the direct care system (n = 500); and men aged 51 to 64 years enrolled in the purchased care network (n = 500). The random sample was drawn from an overall population of about 35,000 members. Sampling was performed without replacement until the target number of surveys was achieved. Survey completion was defined as the respondent having reached the end of the survey questionnaire but not necessarily having answered every question.
Data Elements
The preventive health survey collected information on beneficiaries’ knowledge of and satisfaction with their PCM, the primary location where they sought health care in the previous 12 months, preference for scheduling cancer screening tests, and general knowledge about the frequency and type of screening for breast, cervical, and colorectal cancers. Responses were scored based on guidelines effective as of 2009. In addition, the survey collected information on the beneficiary’s overall health status, current age, highest level of education achieved, current employment status, place of residence (on or off a military installation), race, and whether the beneficiary carried other health insurance aside from TRICARE.
Survey Mode and Fielding
A sampling population of eligible beneficiaries was created from a database of all TRICARE Prime beneficiaries. An automated system was used to randomly draw potential participants from the sample. Survey interviewers were given the beneficiary’s name and telephone number but no other identifiable information. Phone numbers from the sample were dialed up to 6 times before the number was classified as a “no answer.” Interviewers read to each beneficiary a statement describing the survey and participation risk and benefits and explained that participation was voluntary and the participant could end the survey at any time without penalty or prejudice. The survey commenced only after verbal consent was obtained.
Sample Weighting and Statistical Analysis
Each survey record was weighted to control for potential bias associated with unequal rates of noncoverage and nonresponse in the sampled population. A design weight was calculated as the ratio of the frame size and the sample size in each stratum. For each stratum, an adjusted response rate (RR) was calculated as the number of completed surveys divided by the number of eligible respondents. Since all respondents were eligible, the RR was not adjusted. The ratio of the design weight to the adjusted RR was calculated and assigned to each survey.
Frequency distributions and descriptive statistics were calculated for all close-ended survey items. Open-ended survey items were summarized and assessed qualitatively. When appropriate, open-ended responses were categorized and included in descriptive analyses. No formal statistical testing was performed.
Results
A total of 6,563 beneficiaries were contacted, and 3,688 agreed to participate (56%), resulting in 3,500 TRICARE beneficiaries completing the survey (95% completion rate), of whom 71% (2,500) were female. The overall cooperation rates were similar across the 4 strata. Interviews ceased once 3,500 surveys were completed. The largest distribution of respondents was aged between 55 and 64 years (37%) (Table 1). Respondents aged 21 to 24 years comprised the smallest percentage of the sample (7%). Nearly a third of respondents were dependents of ADSMs (30%), another 30% were retirees, and most respondents self-identified as white (Table 1).