Clustering of Multiple Healthy Lifestyle Habits and Health-Related Quality of Life Among U.S. Adults With Diabetes

The BRFSS, an ongoing state-based and national surveillance system of behavioral risk factors and chronic disease conditions, is a standardized telephone survey conducted by state health departments with assistance from the CDC (24). Trained interviewers collect data on a monthly basis from an independent household probability sample in the noninstitutionalized U.S. adult population (aged ≥18 years). A detailed description of the survey design and random-sampling procedures is available elsewhere (24). According to the formula of the cooperation rate (the proportion of all respondents interviewed of all eligible units in which a respondent was selected and actually contacted), the median response rate for the 2005 BRFSS was 75.1% (25). BRFSS data have consistently been found to provide valid and reliable estimates when compared with national household surveys in the U.S. (26–28).

The four CDC HRQOL measures were derived from the original version of the Medical Outcomes Study 36-item short-form survey (SF-36) instrument (29–31) and have been validated in both healthy and disabled populations with acceptable criterion validity and reliability comparable with multiple-item SF-36 subscales (31–36). The four HRQOL questions were as follows: 1) “Would you say that in general your health is excellent, very good, good, fair, or poor? 1 = excellent, 2 = very good, 3 = good, 4 = fair, and 5 = poor.” 2) “Now thinking about your physical health, which includes physical illness and injury, for how many days during the past 30 days was your physical health not good?” 3) “Now thinking about your mental health, which includes stress, depression, and problems with emotions, for how many days during the past 30 days was your mental health not good?” 4) “During the past 30 days, for about how many days did poor physical or mental health keep you from doing your usual activities, such as self-care, work, or recreation?”

Smoking status was determined by asking participants, “Have you smoked at least 100 cigarettes in your entire life?” and “Do you now smoke cigarettes every day, some days, or not at all?” NSMK was defined as either having smoked <100 cigarettes during a person's lifetime or having smoked ≥100 cigarettes during a person's lifetime but not currently smoking. Moderate physical activity was assessed by asking participants two questions: “How many days per week do you do moderate activities for at least 10 min at a time?” and “On days when you do moderate activities for at least 10 min at a time, how much total time per day do you spend doing these activities?” Vigorous physical activity was assessed by asking participants two questions: “How many days per week do you do vigorous activities for at least 10 min at a time?” and “On days when you do vigorous activities for at least 10 min at a time, how much total time per day do you spend doing these activities?” LTPA was defined on the basis of CDC recommendations and federal physical activity guidelines (20) (i.e., moderate physical activity for ≥30 min on ≥5 days per week or vigorous physical activity for ≥20 min on ≥3 days per week). FVC5 was defined as consuming five or more servings of fruits and vegetables daily, according to the national objective in Healthy People 2010 (5).

Diabetes was determined by asking participants, “Have you ever been told by a doctor that you have diabetes?” Responses were coded as “yes,” “yes, but female told only during pregnancy,” or “no.” Gestational diabetes was coded as “no” diabetes. Age at onset of diabetes, use of insulin and oral agents, foot ulcer, and retinopathy were determined by participants' self-reports. BMI (kg/m²) was calculated with the use of self-reported weight and height. According to the World Health Organization guidelines (37), we defined obesity as BMI ≥30 kg/m². Hypertension, high cholesterol, physical disability, heart attack, angina or coronary heart disease, stroke, arthritis, and asthma were determined by asking participants whether they had ever been told by a doctor, nurse, or other health professional that they had such conditions.

We conducted analyses among men and nonpregnant women aged ≥18 years with diabetes. The prevalence estimates were calculated with use of the entire sample. The subsequent analyses for the association between clustering of multiple HLHs and HRQOL were limited to the subsample, with available data for diabetes modules, which were conducted in 40 states (appendix). Furthermore, we excluded participants with missing data for any of the demographic variables, HRQOL measures, HLHs, diabetes complications, and comorbid conditions. We calculated means for the general health rating and for unhealthy days by the number of multiple HLHs, adjusted for age. We performed direct age adjustment using the U.S. population aged ≥18 years in the year 2000 (weights: 0.129, 0.183, 0.219, 0.299, and 0.170 for age-groups 18–24, 25–34, 35–44, 45–64, and ≥65 years, respectively).

We dichotomized the responses for the self-rated general health question into poor or fair health versus good, very good, or excellent health. The 14-day cutoff point is often used as a marker for clinical depression and anxiety disorders in medical practice and research (38–40); therefore, it was used to dichotomize mentally unhealthy days as frequent mental distress (≥14 days) versus infrequent mental distress (<14 days) in our study. Physically unhealthy days and activity limitation days also were dichotomized at 14 days to be consistent with the cutoff point used for mentally unhealthy days and in line with previous studies (14,19,41). In addition, the 14-day cutoff value corresponds to the upper 10–15% of the distribution for the healthy days measures.

Logistic regression analysis was used to assess whether the clustering of multiple HLHs was independently associated with HRQOL measures. Odds ratios and 95% CIs were estimated unadjusted and adjusted for age, sex, race or ethnicity, education attainment, annual household income, as well as duration, treatment, complications, and comorbid conditions of diabetes. SUDAAN software (release 9.0; Research Triangle Institute, Research Triangle Park, NC) was used to account for the complex sampling design.

The final analytic sample (n = 16,428) consisted of 52.8% men, 67.3% non-Hispanic whites, 13.5% non-Hispanic blacks, 13.2% Hispanic subjects, and 6% were other ethnic groups. The mean age of the diabetic participants was (means ± SE) 58.9 ± 0.11 years (range 18–99). The mean duration of having diabetes was 9.6 ± 0.08 years (range 0–81).

The age-adjusted prevalence of diabetes (n = 33,320) among men and nonpregnant women aged ≥18 years (n = 356,112) was 7.8 ± 0.09%, 7.9 ± 0.14%, and 7.8 ± 0.11% in the total sample, among men, and among women, respectively. Non-Hispanic blacks (12.2 ± 0.35%; P < 0.001) and Mexican Americans (8.1 ± 0.36%; P < 0.001) had a higher prevalence of diabetes than non-Hispanic whites (7.2 ± 0.08%).

Women had a higher prevalence of having all three HLHs (P < 0.05) than men (Table 1). No significant differences were found by racial or ethnic subgroups. Prevalence rates were similar in each age-group (P > 0.05 for linear trend). People with a higher level of education (P = 0.02 for linear trend) or higher annual household income (P < 0.01) had a higher prevalence of having all three HLHs.

The age-adjusted means of the general health rating, physically unhealthy days, mentally unhealthy days, and impaired activity days were 3.3, 7.9, 5.3, and 5.2, respectively. The mean rating of general health was 3.67, 3.42, 3.16, and 3.05 among people with 0, 1, 2, and 3 HLHs, respectively. There was approximately a 2-day decrease in physically unhealthy days, mentally unhealthy days, and impaired activity days with a 1-unit increase in HLH (Fig. 1A). The age-adjusted prevalence rates of poor or fair health, ≥14 physically unhealthy days, ≥14 mentally unhealthy days, and ≥14 impaired activity days were 43.07, 27.61, 17.22, and 18.87%, respectively. There were statistically significant linear trends between the accumulation of multiple HLHs and the prevalence of HRQOL measures (all P < 0.001) (Fig. 1B).

NSMK was significantly associated with having poor or fair health (P < 0.001), having ≥14 mentally unhealthy days (P < 0.001), and having ≥14 impaired activity days (P < 0.01) but was marginally associated with having ≥14 physically unhealthy days (P = 0.07) (Table 2). LTPA was significantly associated with all four HRQOL measures (all P < 0.001). In contrast, FVC5 was not significantly associated with any of the four HRQOL measures (P values ranged from 0.21 to 0.94).

Accumulation of the three HLHs was significantly associated with a decreased likelihood of all four HRQOL measures (all P < 0.001 for linear trends) (Table 2). NSMK only, LTPA only, the combination of NSMK and LTPA, and the combination of all three HLHs were significantly associated with all four HRQOL measures (Fig. 2). After adjustment for potential confounders and comparison with those with none of the three HLHs, participants with all three HLHs were less likely to report poor or fair health (adjusted odds ratio 0.49 [95% CI 0.33–0.71]), ≥14 physically unhealthy days (0.56 [0.39–0.80]), ≥14 mentally unhealthy days (0.35 [0.23–0.55]), or ≥14 impaired activity days (0.35 [0.23–0.56]).

Using a large, nationally representative sample, we demonstrated for the first time that clustering of multiple HLHs was strongly associated with better HRQOL in diabetic adults. This association persisted after adjustment for demographic and socioeconomic characteristics, diabetes complications, diabetes severity, and diabetes-related comorbid conditions. Although NSMK or LTPA alone was associated with a significantly better HRQOL, a combination of NSMK and LTPA, or all three HLHs, appeared to further enhance the association in this diabetic population. We estimated that the prevalence of having all three HLHs (NSMK, LTPA, and FVC5) was ∼12% among U.S. adults with diabetes.

The role of NSMK, LTPA, and FVC5 in the primary prevention of cancer, cardiovascular disease, stroke, and diabetes has been widely investigated (15–18). Our results provide evidence that people with diabetes who report achieving recommended levels for these HLHs also report substantially better HRQOL. Among the three individual HLHs, NSMK and LTPA had stronger effects than FVC5 on HRQOL. Abstaining from smoking or engaging in adequate physical activity has been consistently related to reductions in all-cause mortality and morbidity and mortality from cancer, cardiovascular diseases, stroke, and type 2 diabetes (15). Our results suggest that people with diabetes who did not smoke or who participated in physical activity reported ∼2 fewer days per month in physically unhealthy days, mentally unhealthy days, and impaired activity days. We detected a significant association of FVC5 only with poor or fair health, a marginally significant association with ≥14 mentally unhealthy days in the past month, but no significant association with ≥14 physically unhealthy days or ≥14 impaired activity days. However, a combination of FVC5 with NSMK or LTPA appeared to have a stronger association with general health or physical health compared with the individual factors alone. Nevertheless, our findings do not negate the importance of sufficient fruit and vegetable consumption because of its important role in the primary prevention of cancer, cardiovascular diseases, and stroke (42–44).

Most previous studies focused on the protective effects of HLHs individually. The role of clustering of these HLHs on primary or secondary prevention rarely has been examined. In a cohort study, a greater number of multiple low-risk factors (i.e., favorable diet score, nonsmoking, moderate to vigorous exercise ≥30 min/day, BMI <25 kg/m², and alcohol ≥5 g/day) was associated with a lower relative risk of coronary heart disease events (18) or type 2 diabetes (16). The significant linear trend for the relationship between clustering of multiple HLHs and a higher level of HRQOL as shown in our study suggests that accumulation of multiple HLHs also may be related to a better quality of life in secondary prevention in the diabetic population. In fact, the proportion of people who had two or three HLHs was less than half (∼45%) among people with diabetes as indicated in our study and only 30% among people with heart disease (22). Therefore, further effort is needed to promote multiple HLHs in both healthy and diseased populations.

Several previous studies considered healthy weight (BMI between 18.5 and 25 kg/m²) as one of the HLHs in addition to NSMK, LTPA, and FVC5 (45–48). In some of these studies, >50% of the participants had at least two HLHs and >20% of the participants had at least three HLHs (45–48). In our study, we chose not to include healthy weight as an indicator of an HLH because we believe that healthy weight is an outcome of behavior and obesity is an important risk factor or comorbid condition for diabetes, which was accounted for in all of our analyses.

Our study has several limitations. First, all measures in the BRFSS are self-reported and may be subject to information bias. However, the HRQOL measures have been validated in diverse populations and have excellent retest reliability (35). Smoking status, blood pressure, height, weight, BMI, and demographic characteristics were found to be highly valid and reliable. Physical activity and fruit and vegetable consumption were found to be moderately valid and reliable (27). In addition, misclassification bias for self-reported diabetes, diabetes complications, and diabetes-related comorbid conditions may be minimal in the present study because previous studies have shown a substantial agreement between self-report and medical records for diabetes, hypertension, myocardial infarction, and stroke (49). Because misclassification bias among self-reported measures in the present study appears to be nondifferential, our findings may underestimate the true association between clustering of HLHs and HRQOL. Second, because BRFSS was a cross-sectional survey, we were unable to determine whether the observed linear trend for the relationship between clustering of HLHs and HRQOL measures actually was causal. It is possible that people with a better HRQOL are more likely to adopt HLHs, particularly LTPA. Therefore, if this association could be confirmed in future research using a prospective design or clinical trial, it may have meaningful implications in clinical practice and public health service to promote the recommended levels of the three HLHs and to substantially improve HRQOL in the diabetic population. Third, because impaired activity included self-care, work, or recreation, a slight construct overlap may be possible when examining the association of LTPA with impaired activity days. However, our results may not be influenced because all associations were adjusted for physical disability status, and similar patterns were replicated in the subsample of people free of physical disability.

In conclusion, the first goal of Healthy People 2010 is to help people of all ages increase their life expectancy and improve their quality of life. This goal may be particularly important among people with diabetes because disease severity, complications, and comorbid conditions greatly impair their quality of life. Our findings suggest that having two or more HLHs may be associated with patients' quality of life more substantially than the single factors alone. Because less than half of patients with diabetes have achieved the recommended levels of two or more HLHs, greater efforts are urgently needed to promote the adoption of multiple HLHs among people with diabetes in the U.S.

In 2005, the following 40 states had data available for the BRFSS diabetes module: Alabama; Alaska; Arizona; Arkansas; California; Colorado; Connecticut; Delaware; Washington, DC; Florida; Georgia; Idaho; Indiana; Iowa; Kentucky; Louisiana; Maine; Minnesota; Missouri; Montana; Nevada; New Hampshire; New Jersey; New Mexico; New York; North Carolina; North Dakota; Ohio; Pennsyvania; South Carolina; South Dakota; Tennessee; Texas; Utah; Vermont; Virginia; Washington; West Virginia; Wyoming; and the Virgin Islands.

We thank the BRFSS state coordinators for their assistance in data collection.

The findings and conclusions in this article are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.