Estimating Contributions of Social and Behavioral Factors to Cardiovascular Disease, Cancer, COPD, and Unintentional-Injury Mortality Disparities by Psychological Distress in the United States: A Blinder-Oaxaca Decomposition Analysis of the 1997-2014 NHIS-NDI Record Linkage Study

Data

The data for this study are derived from the National Health Interview Survey (NHIS) prospectively linked to the National Death Index (NDI).³⁷ As a nationally representative, annual cross-sectional household interview survey, NHIS provides demographic, socioeconomic, and health characteristics of the civilian, non-institutionalized population in the US. The National Center for Health Statistics (NCHS) developed public-use versions of NHIS prospectively linked with death certificate records from the NDI. For this study, we used the latest available 1997-2014 public-use linked mortality file containing 18 years of mortality follow-up data from the date of survey participation through December 31, 2014.³⁸

Sample

The study sample consisted of adults aged 18 and older from the years 1997 to 2014 NHIS sample adult files. Since the main decomposition analysis allows only dichotomous group variable, we restricted the study sample to adults with no PD (K6=0) and adults with SPD (K6=13+) as defined below. The final pooled eligible sample size, after excluding missing values, was 263,825; the listwise deletion accounted for 1.61% of the total sample size of 268,145. For missing values, accounting for more than 1% of the total observations, that is for poverty status (18.74%), body mass index (BMI) (3.54%), and alcohol consumption (1.66%), we created missing categories so as to not lose a substantial number of observations for the analysis.

Outcome Measurement

We analyzed crude death rates for cancer (International Classification of Diseases-10^th Revision (ICD-10): C00-C97), CVD (heart disease and stroke combined, 100-109, 111, 113, 120-151, 160-169), COPD (J40-J47), and unintentional injury (V01-X59, Y85-Y86), the top five leading causes of death in the US in 2017.³⁹ All mortality outcomes were dichotomized, with I being dead and zero equaling alive.

Psychological Distress Classification

Psychological distress (PD) was measured by the Kessler 6 (K6) nonspecific distress scale³¹ of six symptoms. Respondents were asked: “during the past 30 days, how often do you feel 1) so sad that nothing could cheer you up; 2) nervous; 3) restless or fidgety; 4) hopeless; 5) that everything was an effort; 6) worthless.” We created a 5-level categorical variable with scores of 0, 1-2, 3-5, 6-12, and 13-24 from the K6 scale.²¹ Since Blinder-Oaxaca allows only a dichotomous group comparison, we only used the sample with a score of 0 as a group with no PD and those with a score of 13 or higher as a group with SPD.

Risk Factors

Based on the previous literature, we selected the following risk factors for model estimation: age, sex, race/ethnicity, nativity/immigrant status, education, marital status, poverty status, housing tenure, region of residence, BMI, smoking status, and alcohol consumption.^{18,21,22,40-43} Age was used as a continuous variable ranging from 18 to ≥85 years. Race/ethnicity was defined by six categories as non-Hispanic White, non-Hispanic Black, Hispanic, non-Hispanic American Indian/Alaska Native, non-Hispanic Asian/Pacific Islander, and non-Hispanic other races. Nativity/immigrant status was categorized as US- born, who was born in one of the 50 states or the District of Columbia, or foreign-born. Educational attainment was defined by four categories as less than high school diploma, high school diploma, some college, and college degree or more. Marital status was categorized as currently married, widowed, divorced/separated, and never married. Poverty status was defined by five categories based on the ratio of family income to the US Census Bureau's federal poverty threshold (<100%; ≥100% & <200%; ≥200% & <400%; ≥400%; missing). Housing tenure was dichotomized, with 1 being renters and zero equaling homeowners. Region of residence was defined by four categories, Northeast, Midwest, South, and West. BMI was defined by four categories, <25, 25-29, 30-39, ≥40, and missing. Smoking status was defined by three categories as never, former, and current smokers. Alcohol consumption was defined by four categories: lifetime abstainer, former drinker, current drinker, and unknown.

Analytic Approach

The Blinder-Oaxaca decomposition analysis was conducted to quantify the contributions of individual characteristics in explaining the observed mortality differences between adults with no PD and those with SPD.^44,45

In the equation above, ӯ_SPD and ӯ_noPD represent means of the rates of deaths for adults with SPD and those with no PD, and x̄_SPDand x̄_{no PD} are vectors of group-specific means of the individual-level characteristics studied. β̂* is the vector of coefficients from the pooled linear probability models that reflects the average outcome difference associated with each variable. β̂_SPD is the adjusted difference in outcomes between adults with no PD and those with SPD that remains unexplained by covariates. The “explained” portion of the decomposition reflects the change in the mean of mortality rates for adults with SPD, compared to those with no PD, due to the differences in the individual-level characteristics. The percentage of the outcome difference explained by each covariate can be calculated by dividing the explained difference by the total difference of Blinder-Oaxaca decomposition analysis: model 1 included only sociodemographic characteristics; model 2 included both socio-demographic and health behavioral characteristics as contributors, to see confounding effects between sociodemographic and health behavioral characteristics.

Collinearity between model covariates was assessed using variance inflation factors, and all values across models were within an acceptable range (≤1.61). Complex survey design procedures, svy in Stata, were used, to account for clustering, multiple stages of selection, and disproportionate sampling. To correct the bias from the ineligible adults for linkage to the NDI due to insufficient identifying data, we used eligibility-adjusted weights developed by NCHS, instead of the standard sample weight.⁴⁶ The sample weights were adjusted by dividing by the number of pooling years. All the analyses were conducted by Stata 16 and the Blinder-Oaxaca model was fitted using the Oaxaca procedure.⁴⁷

Mortality and Individual Characteristics by No Psychological Distress and Serious Psychological Distress

As shown in Table 1, adults with SPD had higher CVD, cancer, COPD, and unintentional injury mortality, compared to those with no PD. Regarding sociodemographic characteristics and health behaviors, adults with SPD were more likely to be female (no PD vs. SPD: 47% vs. 62%, P<0.001), non-Hispanic Blacks (12% vs. 13%), Hispanics (14% vs. 15%), or American Indians/Alaska Natives (0.6% vs. 1.3%), US-born (83% vs. 87%), with less than high school education (13% vs. 29%), never married (22% vs. 25%), below poverty level (8% vs. 28%), renters (28% vs. 47%), less with higher BMI ≥40 (3% vs. 8%), current smokers (17% vs. 42%), and former drinkers (13% vs. 24%).

Blinder-Oaxaca Decomposition

Higher CVD mortality for adults with SPD was explained by the measured sociodemographic factors by 45.75% in Model 1 and by both sociodemographic and behavioral factors by 50.09% in Model 2 (Table 2). In Model 1, the lower educational level for adults with SPD accounted for 29.49% of the total 1.23 percentage point mortality disparity. Lower-income, higher proportions of renters and higher proportions of non-currently- married individuals among adults with SPD also contributed to CVD mortality disparities by 16.39%, 7.63%, and 5.98%, respectively. Higher proportions of females (-10.58%) and younger individuals (-4.17%) favored adults with SPD and decreased CVD mortality disparities between adults with no PD and those with SPD. In Model 2, the contributions of education, poverty status, marital status, and housing tenure slightly decreased, indicating that health behaviors such as smoking and alcohol consumption partly accounted for the impact of sociodemographic factors. Higher proportions of current smokers and former drinkers contributed to CVD mortality disparities by 9.76% and 6.30%, respectively, while higher proportion of BMI among adults with SPD decreased CVD mortality by 3.08% (Table 2, Figure 1). Cancer mortality difference (1.00 percentage point, 95% CI = -1.33, -0.68) between adults with no psychological disparities and those with SPD was explained by 37.46% in Model 1 and 71.25% of the difference was explained by Model 2, in total (Table 3). In Model 2, around half of the cancer mortality disparity (0.48 percentage points, or 47.37%) could be attributed to the higher rate of current smoking among adults with SPD. Lower educational level, lower-income, and higher proportions of former or current drinkers and US- born also contributed to higher cancer mortality for adults with SPD by 30.98%, 10.56%, 6.77%, and 2.24%, respectively (Table 3, Figure 1). However, younger age, higher proportion of females and racial/ethnic minorities, and higher BMI mitigated the cancer mortality disparities by 4.69%, 11.87%, 1.70%, and 5.16%, respectively.

Close

Figure 1:

Explained proportions of CVD, cancer, COPD, unintentional-injury mortality disparities between adults without psychological distress (PD) and with serious psychological distress (SPD), 1997-2014

Source: The difference in outcome explained was estimated using the Blinder-Oaxaca linear decomposition analysis. The estimates come from the Model 2 results in Tables 2 to 5.

Export to PPT

Higher COPD mortality for adults with SPD was explained by the measured sociodemographic factors by 24.35% in Model 1, and explained by both sociodemographic and behavioral factors by 36.77% in Model 2 (Table 4). The highest contribution to COPD mortality disparity was made by the higher rate of current smoking (17.48%) among adults with SPD. The higher COPD mortality for adults with SPD was also attributable to the lower level of education and income, and higher proportions of current and former drinkers, renters, and foreign- born among adults with SPD, by 9.80%, 7.05%, 2.93%, 2.73%, and 1.08%, respectively (Table 4, Figure 1). Younger age, higher proportion of females and racial/ethnic minorities, and higher BMI mitigated the COPD mortality disparities by 1.20%, 2.47%, 0.67%, and 1.94%, respectively.

Unintentional injury mortality difference (0.54 percentage points, 95% CI = -0.70, -0.38) between adults with no PD and those with SPD was explained by measured variables in Model 1 by 15.4% and 20.19% of the difference was explained by Model 2 (Table 5). In Model 2, higher proportion of current smokers, lower education level, lower-income, and higher proportion of non-currently married individuals contributed to higher unintentional injury mortality for adults with SPD, by 10.78%, 9.44%, 6.82%, and 3.75% respectively (Table 5, Figure 1). Higher proportions of females and higher BMI mitigated the disparities in unintentional injury mortality by 7.65% and 2.9%.

To check the robustness of the results by follow-up years due to potential time-varying risk factors, a sensitivity analysis was conducted for cancer mortality to assess variations in the contributions of individual characteristics by using 2, 5, and 10 years of mortality follow-up. The longer the follow-up years, the larger the component explained by measured variables (Table 6). This might be because the longer follow-up allows for a greater manifestation of the lagged effect of smoking, BMI, drinking, and socioeconomic disadvantage on mortality.

Limitations

This study has some limitations. First, our study used the NHIS sample eligible for linkage to the NDI, which is, on average, 94.8% of the original NHIS survey participants.⁴⁶ Excluding records ineligible for linkage may lead to biased mortality estimates. To address potential bias, we used the eligibility- adjusted sampling weight to account for the NHIS- NDI mismatches.⁴⁶ Second, while our Blinder-Oaxaca decomposition included a variety of risk factors of cause-specific mortality, there could be other potential factors of mortality, confounding existing risk factors. Third, in the NHIS-NDI database, all the risk factors were available at baseline as of the survey date, while SES, behavioral-risk factors, and PD are expected to vary over the long follow-up. It is worthwhile to examine the extent of changes in SES or behavioral factors from childhood, working- age, to older ages contributing to the disparities in mortality by SPD status, using the life course approach⁵⁴ with longitudinal data.

Compliance with Ethical Standards