Should Years of Schooling Be Used to Guide Treatment of Coronary Risk Factors?

INTRODUCTION

The relationship between lower socioeconomic status measured by education, income, or occupation and mortality is well established.^1–,⁴ Socioeconomic disparities in morbidity and mortality from coronary heart disease (CHD) represent a major cause of socioeconomic disparities in overall mortality in the United States.^5–,⁸

Evidence suggests that the relationship between lower socioeconomic status and CHD is, in part, independent of traditional CHD risk factors, including hypertension, elevated cholesterol levels, diabetes mellitus, and smoking.^9–,¹² Despite these findings, risk-based treatment guidelines, such as the Adult Treatment Panel III (ATP III) from the National Cholesterol Education Program,¹⁰ do not account for the effects of socioeconomic status on overall risk of death from CHD. Yet, if lower socioeconomic status independently predicts CHD risk, then it could be used to identify persons in need of earlier or more intensive management of modifiable risk factors, such as elevated low-density lipoprotein (LDL) levels.

In this study, we compare the independent risk associated with low educational level with traditional CHD risk factors. Specifically, we use nationally representative data from the United States to compare the rates of CHD death within 10 years based upon these factors.

METHODS

RESULTS

Ascertainment of risk factors and 10-year status was available for 6,637 persons (96%). The prevalence of the risk factors by education is shown in Table 1⇓. In a multivariate analysis, significant predictors of CHD mortality in order of relative risk were older age, systolic blood pressure, diabetes, cholesterol level, having less than 12 years of education, being male, and smoking (Table 2⇓). The addition of race to the models had little effect on the relative risks for any of the risk factors (results not shown). There were no statistically significant interactions among the key variables and no problems with collinearity.

The predicted marginal risk of CHD associated with these risk factors is also displayed in Table 2⇑. Compared with persons having more than 12 years of education, persons with less than 12 years of education had a 2.4 % (95% CI, 1.2–3.6) greater absolute risk of 10-year CHD mortality. This effect is comparable to being male compared with being female (2.3%; 95 % CI, 1.2–3.4), and age 65 years and older compared with 35 to 44 years (2.8%; 95% CI, 1.6–3.9). The inclusion of persons with CHD in the analysis did not appreciably alter the results (results not shown).

Analyses stratified by educational level are shown in Table 3⇓. At each level of cholesterol, the adjusted marginal predicted risk of CHD mortality is greater for persons with fewer than 12 years of education than for persons with more than 12 years. Based on these adjusted predictive margins, and assuming 100% effectiveness of cholesterol-reducing medication, the numbers needed to treat to reduce the mortality risk for those with a cholesterol level of greater than 280 mg/dL to that of persons with a cholesterol level of less than 200 mg/dL is 19 for those with fewer than 12 years of education, and 111 for those with more than 12 years.

DISCUSSION

The results show that low educational level predicts 10-year CHD mortality independently of traditional risk factors. Notably, the risk associated with low education is comparable in magnitude to many of the traditional risk factors, including cholesterol level, smoking status, sex, and age, that are used in current treatment guidelines. These findings suggest the possibility that an educational level of less than 12 years may be used to identify persons at higher absolute risk of death from CHD who would not otherwise be identified for treatment.

The incorporation of low education into treatment guidelines could potentially reduce socioeconomic disparities in CHD mortality. Although education cannot be modified medically, it can be used, as are other nonmodifiable sociodemographic risk factors such as age and sex, to select persons for more intensive management of modifiable risk factors. Subgroup analyses of clinical trials suggest that persons of lower socioeconomic status derive as much, if not more, benefit from CHD interventions than persons of higher socioeconomic status.²¹

The clinical application of risk stratification based on lower socioeconomic status represents a novel strategy for addressing socioeconomic disparities in conditions other than CHD; it applies to any condition where treatment decisions are based on risk and where socioeconomic status is a risk factor. Such a strategy suggests more resources be directed toward primary or secondary prevention for diseases that occur at higher rates or at earlier ages among persons of lower socioeconomic status.

Use of an educational cutoff represents a simple, clinically applicable decision rule. Questions about education are less sensitive than questions about family income and are more reliably reported. Education is also subject to less change with time than income, which is strongly linked to employment status. Nonetheless, we observed similar results in our study using household income (results not shown).

The potential limitations of this study merit discussion. First, baseline assessments on subjects in NHANES I were obtained more than 25 years ago. Potentially, the relation between socioeconomic status and CHD could have changed during this period. Available evidence suggests, however, that socioeconomic disparities in CHD mortality have increased, not decreased, with time.^22–,²⁴ Second, low-density lipoprotein or high-density lipoprotein levels were not assessed. Given the modest and inconsistent relationship between years of education and these lipoprotein levels,^25,26 appreciable confounding by these factors is unlikely. Third, previous studies have shown that socioeconomic status is also associated with nontraditional CHD risk factors not included in this analysis, such as fibrinogen,²⁷ C-reactive protein,²⁸ apolipoproteins,²⁷ hostility,^29,30 psychological distress,³¹ social support,³² passive smoking,³³ and job control.³⁴ These unmeasured factors could account for the effect of education on mortality. Even so, however, none of these factors is included in ATP III guidelines, so their relevance to the question of including education in these guidelines is moot.

Fourth, it is possible that our findings are attributable to greater modification in traditional risk factors among more educated persons after baseline assessment of these factors. For example, rates of smoking cessation are significantly higher among those with higher education³⁵; higher rates of subsequent smoking cessation by more educated persons may contribute to their lower CHD death rate. Given the limitations of the data, we cannot exclude this possibility; yet, if the goal of risk-based decision models is to identify persons in need of more intensive treatment, the pathway or mechanism that produces this heightened risk is not pertinent. The relevant question is whether the addition of a particular risk factor to the model improves prediction of the outcome.

Last, showing that education independently predicts risk at all cholesterol levels in an observational study does not definitively establish using a lower treatment threshold for less-educated persons to improve outcomes.³⁶ Our view is that if the model observed here is validated in other cohorts, then the potential benefits of including low education as a risk factor in reducing both total CHD mortality and disparities in mortality likely outweigh the risks.

In summary, low socioeconomic status as measured by fewer than 12 years of education is associated with a CHD mortality risk comparable to traditional CHD risk factors. Use of a low educational level to identify persons at higher risk, who are not otherwise identifiable under current guidelines, may facilitate progress toward the elimination of socioeconomic disparities in health.