Table 1 presents prevalence estimates and 95% confidence intervals (CIs) for HRQOL indicators by demographic, health condition, and health risk variables. Overall, 13.7% had fair or poor general health. Results for the 8 indicators based on the criterion of 14 or more days of poor health in the past month (hereafter referred to as recent frequent [RF] poor health) were as follows: 4.8% had RF activity limitations due to a physical or mental health problem; 9.7% had RF poor physical health; 7.2% had RF pain related activity limitations; 27.9% reported RF lack of energy; 9.5% had RF poor mental health; 8.0% were RF sad, blue or depressed; 12.6% were RF worried, tense or anxious; and 23.9% reported RF inadequate sleep or rest. A major depressive episode in the past 12 months was experienced by 7.6%.
Each of the independent variables was significantly associated with 4 or more indicators of poor HRQOL. The following are general observations about prevalence patterns identified across HRQOL indicators for each of the independent variables.
Age
The prevalence of three of the indicators for poor physical health (fair/poor general health, RF physically unhealthy days, and RF pain related activity limitation) increased with increasing age, while the prevalence of indicators for RF poor mental health decreased with increasing age. All differences were statistically significant. There were no significant age related differences for RF physical or mental health related activity limitations and RF lack of energy.
Sex
Women reported a higher prevalence than men for all of the HRQOL indicators. All differences were statistically significant except for "RF pain related activity limitations" and "RF lack of rest/sleep".
Race/Hispanic ethnicity
It was necessary to group all non-white, non-Hispanic respondents into a single "other" category due to small sample sizes. There were significant differences by race/Hispanic ethnicity for 5 of the 10 indicators. "Hispanics" had the highest percent with "poor/fair" general health, but did not have the highest percent for other indicators of RF poor HRQOL. The "Other" group had the highest percentages for "RF lack of energy", "RF lack of rest/sleep", for "RF sad/blue depressed" and for MDE and these were significant differences.
Income
The prevalence of all poor HRQOL indicators increased with decreasing levels of annual household income. All differences were statistically significant with the exception of "RF lack of rest/sleep".
Employment
Statistically significant differences occurred for all poor HRQOL indicators in relation to employment status. Respondents who were unable to work had higher rates than other employment categories for all indicators. Respondents who were unemployed also had higher rates than others for most of the poor HRQOL indicators although they were very similar to retired persons for poor general health, RF physically unhealthy days, and RF pain related activity limitations.
Tobacco use
Current smokers had significantly higher rates than non-smokers for all poor HRQOL indicators except for general health status.
Physical inactivity
Adults who were physically inactive had significantly higher rates for all poor HRQOL indicators than adults who reported engaging in leisure time physical activity.
Chronic alcohol use
Chronic drinkers were significantly less likely to report poor general health or poor physical health than those who were not chronic drinkers, but significantly more likely to report "RF worried tense or anxious" or "RF lack of rest or sleep".
Diabetes and asthma
People who had been told by a physician that they had diabetes or asthma had significantly higher percentages for all indicators of poor HRQOL than persons without these conditions, with the exception that diabetics and non-diabetes were equally likely to report RF lack of rest or sleep.
Obesity and disability
Obese persons and disabled persons had higher percentages for all poor HRQOL indicators than non-obese or non-disabled persons, and all differences were statistically significant.
Table 2 presents the crude (unadjusted) odds ratios (CORs), and 95% confidence intervals (CIs) for demographic, health condition, and health risk variables regressed on HRQOL indicators. The CORs make it possible to evaluate the strength of the relationship between independent variable categories and dependent variables without adjusting for confounding effects. The highest CORs occurred among those unable to work for all but two of the HRQOL indicators. Those with disabilities had the next highest CORs for all but three indicators of poor mental health. The CORs for those indicators were comparable or only slightly less than those for the unemployed.
Table 3 presents the adjusted odds ratios (AORs), and 95% confidence intervals (CIs), for demographic, health condition, and health risk variables regressed on HRQOL indicators using multivariable logistic regression (MLR) after multiple imputation. The same set of co-variables was used for adjustment in each MLR model.
In comparing the AORs with the CORs, the number of variables significantly related to the HRQOL variables decreased after adjustment, and almost all of the significant AORs were lower than the CORs, some substantially, indicating the presence of confounding effects (compare Table 2 and 3) (attached as Additional File 4 and 5).
The AORs in most instances were greater than 1, indicating a higher probability of poor HRQOL when compared with the reference group. However, in a few instances the AORs were less than 1 indicating decreased risk compared with the reference group.
Significantly increased AORs occurred for five or more of the poor HRQOL indicators for seven of the independent variables: have disability, unable to work, no leisure physical activity, have asthma, household income <$25 K, current smoker and unemployed. Age 65+ had decreased AORs associated with the indicators of poor mental health.
Being physically disabled had significant increased AORs for all ten of the HRQOL indicators and had the highest AORs associated with the variables related to physical health, e.g. general health status (4.5), RF activity limitation (10.6), RF poor physical health (5.3), and RF pain related activity limitation (7.8). It was also the strongest predictor for four of the six indicators of poor mental health – RF lack of energy (3.4), RF worried/tense/anxious (2.6), RF lack of rest/sleep (2.8), and major depressive episode (3.3) and is significantly elevated for RF poor mental health (2.2) and for RF sad/blue/depressed (2.5).
After the MRL, the AORs were much lower than the CORs for "unable to work" indicating that other independent variables (e.g. disability) are important confounders for being unable to work. Nonetheless, being unable to work was the second strongest predictor of both poor physical and mental HRQOL, with AORs between 1.9 and 5.0. It had the highest AORs of any variable for RF poor mental health (2.7) and for RF sad/blue/depressed (3.7). Being unable to work was not a significant predictor for RF lack of rest/sleep (1.1).
"No leisure time physical activity" was a significant predictor (with AORs between 1.5 and 1.8) for eight of the ten poor HRQOL indicators, the exceptions being RF lack of rest/sleep and MDE.
Current asthma was a significant predictor for six of 10 indicators including poor general health status (1.9), RF activity limitation (1.8), poor physical health (2.4), RF lack of energy (1.5), RF worried/tense/anxious (1.6), and MDE (2.2).
Being unemployed significantly increases the odds for poor general health (2.4), RF activity limitations (2.2) and for four of the mental health indicators – poor mental health (1.8) RF sad/blue/depressed (1.9), RF worried/tense/anxious (2.3) and MDE (1.8).
Persons with household incomes less than $50,000 had significantly greater odds (AORs 1.7 – 2.9) than higher income persons for poor/fair general health, RF sad/blue/depressed, and for RF worried/tense/anxious. Having household income less than $25,000 was a significant predictor as well for RF lack of energy (1.7), and RF poor mental health (1.8).
Hispanic ethnicity was a strong predictor for poor/fair general health (4.1) but this finding seemed anomalous when compared with the significantly decreased risk for Hispanics for RF lack of energy (0.7) and with the non-significant results for all other indicators.
Diabetics experience double the odds (2.1) of poor/fair general health compared with non-diabetics, and have increased odds of having RF pain related activity limitation (1.7) and for RF sad/blue/depressed (1.7).
Current smoking was a predictor for five mental health indicators: RF poor mental health (1.5), RF sad/blue/depressed (1.8), RF worried/tense/anxious (1.7), RF lack of rest/sleep (1.4) and MDE (1.9) but not for RF lack of energy nor for any of the poor physical health indicators.
Females had significantly increased odds compared with males for four mental health indicators: RF poor mental health (1.4), RF worried/ tense/ anxious (1.4), RF lack of rest/sleep (1.3) and MDE (1.7) but not for RF sad /blue /depressed.
The AORs for poor general health increased with each of the two older age groups – from 2.1 for those ages 45 – 64 to 2.5 for those ages 65 and older. However, when other variables are controlled, the oldest age group was not at increased risk for other indicators of poor HRQOL. In fact, this age group had decreased odds (0.5) for RF activity limitations when compared with younger adults as well as significantly decreased odds compared with younger adults for each of the indicators for poor mental health. The AORs for RF lack of rest/sleep was less than one for both groups of older adults (0.6 and 0.3), in relation to the reference group aged 18 – 44. Those 45–64 had significantly greater odds (1.5) than the reference group for RF pain related activity limitations.
Obesity did not have significant elevated AORs for any of the HRQOL indicators other than for RF poor/fair general health (1.4) and RF lack of energy (1.3). Chronic drinking was a significant predictor only for RF worried/tense/anxious days (2.0).