Data for the present study were acquired from the United States (US) Centers for Disease Control and Prevention (CDC) website 22 as part of the 2015-2016 and 2017-2020 National Health and Nutrition Examination Survey (NHANES) cycles. Data from the two NHANES cycles utilized were gathered prior to the COVID-19 pandemic. This study was determined to be exempt from IRB review by the sponsoring university due to the nature of the secondary data analysis (IRB ID# 1505514-1).

After all inclusion and exclusion criteria were considered, the sample size consisted of 4,821 non-institutionalized US citizens who were representative of 198,781,963 US adults (Table 1). The weighted sample was 49.6% male, and the mean age of the sample was 45.92 years. The average income (poverty ratio) was 3.07 times the poverty level for the years included, and the mean Framingham 10-year cardiovascular disease risk score was 8.92%. LDL-cholesterol, fasting blood glucose, and systolic blood pressures were high in the sample (111.06 mg/dL, 109.15 mg/dL, and 120.97 mmHg, respectively), while HDL, total cholesterol, triglycerides, and diastolic blood pressures were normal (54.30 mg/dL, 187.37 mg/dL, 111.26 mg/dL, and 72.26 mmHg). The average body mass index was classified as overweight (29.64 kg/m²) and high-sensitivity C-reactive protein was slightly elevated (3.82 mg/L). Nearly half of the sample had metabolic syndrome (42.53%), one quarter were on medication for high blood pressure (22.55%), and 17.39% reported smoking.

Analyses by race/ethnicity were completed to compare the demographic variables. All races/ethnicities had statistically equivalent LDL-cholesterol, fasting blood glucose levels, and reported similar levels of physical activity. Additionally, all races/ethnicities were equivalent in frequency of diabetes. All other demographic variables indicated a significant difference by race/ethnicity.

In the main analysis of variance (ANOVA) tests, FRS and SES were compared by race/ethnicity among 4,821 subjects, who were representative of 198,781,963 Americans. The model R² was 0.0059 (Root MSE=11.15, DF=40) and FRS was found to be significantly different between races/ethnicities (DF=5, F=6.44, p=0.0002). Mexican Americans had the lowest FRS on average (Table 2) and post-hoc testing revealed that Mexican Americans had a significantly lower FRS compared to all other racial/ethnic groups except NH Asian (Table 3 and Figure 1). NH Asians had the second lowest FRS and had significantly lower risk than NH Blacks and NH Whites, but their FRS was not statistically different from other Hispanic or the other/multi-racial groups. NH Blacks were found to be statistically equivalent to NH White, other Hispanic, and other/multi-racial/ethnic groups in FRS. NH Whites were not statistically different from other Hispanic or other/multi-racial/ethnic groups. Lastly, other Hispanic and multi-racial/ethnic were not found to be statistically different. Overall, Mexican Americans had the lowest FRS whereas NH Whites had the highest.

The poverty ratio was also compared among the 6 race categories using the same sample as above. The poverty ratio was statistically different between races (DF=5, F=65.22, <0.0001). NH Asians and NH Whites had the highest poverty ratio, and the ANOVA (R²=0.112, Root MSE=1.556, DF=40) revealed that the poverty ratio was statistically equivalent between NH Asians and NH Whites. The poverty ratio was lower in NH Black, other Hispanic, and other/multi-racial/ethnic groups, and these three groups were not statistically different from one another. Mexican Americans were found to have the lowest poverty ratio and had a significantly lower SES compared to all other races/ethnicities. In additional post-hoc testing, we analyzed each component of the FRS calculation and whether it differed by race/ethnicity (Table 4).

This finding was further supported by a regression analysis including FRS as the dependent variable poverty ratio as independent variables. The p-value was insignificant (p=0.2029) and only 0.03% of the variance in FRS could be explained by the poverty ratio,

A multiple regression analysis was conducted to analyze the effects of poverty ratio and education on the FRS differences among the 6 different races and is reported in Table 4. The p-value (>0.001) suggests overall model significance. Only about 2.55 % of the variance in FRS scores between the races can be explained by race and education levels (adjusted R2= 0.02549). Although the FRS difference between the races changed from 0.000181 to 0.00215, it was still statistically significant. The independent p-values for the effect of FRS difference due to poverty ratio was 0.0103 and that due to the education level was less than 0.0001.

A Rao-Scott chi squared test indicated a significant difference between sexes by race (χ²=11.87, DF=5, p=0.0366). The logistic regression analysis used for post-hoc testing (with NH Whites as the reference group) indicated that there were a significantly greater number of females than males in the NH Black group (OR=1.24, 95%CI=1.07, 1.44, p=0.0045) as compared to the reference group, NH white. All other races/ethnicities were not statistically different than whites in terms of sex distribution.

A Rao-Scott chi squared test indicated that there was a significant difference in smoking status (χ²=71.95, DF=5, p<0.0001) by race/ethnicity. NH Black and other/multi racial/ethnic groups were more likely to smoke (OR=1.51, 95%CI=1.20, 1.92, p=0.001 and OR=2.78, 95%CI=1.84, 4.19, p<0.0001, respectively) compared to the NH white reference group. NH Asians were less likely to smoke (OR=0.53, 95%CI=0.37, 0.75, p=0.0007) compared to the reference group. Mexican and other Hispanic were not statistically different than NH whites in terms of smoking status.

A Pearson’s chi-squared test indicated a significant overall difference (χ²=37.94, DF=5, p<0.0001) in those taking medication for hypertension. Logistic regression analysis indicated that Mexican Americans and other Hispanics were less likely to be on an HTN med (OR=0.466, 95%CI=0.33, 0.66, p<0.0001 and OR=0.68, 95%CI=0.50, 0.93, p=0.0156, respectively) compared to NH White. NH Blacks were more likely to be on medications for HTN (OR=1.30, 95%CI=1.02, 1.66, p=0.0349) compared to white, and NH Asian and other/multi-racial were not statistically different than NH White in terms of HTN medication use.

A Pearson’s chi-square test was non-significant (χ²=6.86, DF=5, p=0.231) for diabetes status among the racial/ethnic groups. In logistic regression analysis, NH Black was significantly more likely to have diabetes (OR=1.304, 95%CI=1.06, 1.60, p=0.0134) compared to referent group, NH white. No other racial/ethnic groups were statistically different from the reference.

An ANOVA demonstrated that there was a significant effect of race/ethnic (p<.0001) when comparing age in the 6 race/ethnic categories. Adults 18 to 79 years of age were included in the analysis. On average, Mexican Americans were the youngest and NH Whites were oldest. Post-hoc testing demonstrated that Mexican Americans were younger, on average, compared to NH Asians (t=-3.78, p=0.0005), NH Blacks (t=-3.47, p=0.0013), NH Whites (t=-6.32, p<.0001), and other Hispanics (t=-3.43, p=0.0014). NH Whites were older, on average, than all other races/ethnicities: Mexican Americans (t=-6.32, p<.0001), NH Asians (t=-3.28, p=0.0021), NH Blacks (t=-4.67, p<.0001), other Hispanics (t=3.11, p=0.0034), and other/multi-racial (t=3.54, p=0.001).

An ANOVA demonstrated that there was a significant effect of race/ethnicity (p=0.015) when comparing total cholesterol in the 6 race categories. Overall, NH Blacks had the lowest total cholesterol, followed by Mexican Americans. NH Asians had the highest total cholesterol, followed by NH Whites and Other/Multi Racial. Post-hoc testing demonstrated that Mexican Americans had lower total cholesterol than NH Asians (t=-2.46, p=0.0185) and NH Whites (t=-2.46, p=0.0185). NH Blacks had significantly lower total cholesterol than NH Asians (t=2.86, p=0.0066), NH Whites (t=-3.35, p=0.0018), and other/multi-Racial (t=-2.28, p=0.0283) groups.

An ANOVA demonstrated that there was a significant effect of race (p<.0001) when comparing HDL-cholesterol among the 6 race/ethnicity categories. Overall, NH Blacks had the highest HDL followed by NH Asians, whereas Mexican Americans had the lowest HDL, followed by Other/Multi Racial. Post-hoc testing demonstrated that Mexican Americans had significantly lower HDL than NH Asians (t=-6.61, p<.0001), NH Blacks (t=-6.43, p<.0001), and NH Whites (t=-5.33, p<.0001). The other/multi racial/ethnic group had significantly lower HDL than NH Asians (t=4.56, p<.0001), NH Blacks (t=6.34, p<.0001), and NH Whites (t=4.6, p<.0001). NH Blacks had significantly higher HDL than NH Whites (t=2.2, p=0.0339), other Hispanics (t=5.36, p<.0001), and Other/Multi-Racial (t=6.34, p<.0001). NH Asians had significantly higher HDL than Other Hispanics (t=3.48, p=0.0012), and NH Whites had significantly higher HDL than Other Hispanics (t=3.39, p=0.0016). Mexican Americans, Other Hispanics and Other/Multi-Racial were not significantly different from one another in terms of HDL-cholesterol and NH Asians were found to be statistically similar to NH Blacks and NH Whites.

There was a significant difference in systolic blood pressure (SBP) by race (p<.0001). Overall, NH Blacks had the highest SBP (125.67, SE=0.98 mmHg) and Mexican Americans had the lowest (119.4, SE=0.80 mmHg) followed closely by NH Asians (119.79, SE=0.74 mmHg). Post hoc testing demonstrated that NH Blacks had significantly higher SBP compared to all other races: Mexican American (MA) (t=-4.67, p<.0001), NH Asian (t=-5.36, p<.0001), NH White (t=5.21, p<.0001, Other Hispanic (t=4.76, p<.0001), and Other/Multi-Racial (t=2.83, p=0.0072). No other significant relationships were found.

Racial/ethnic differences in CVD have been previously published in the literature with many studies using FRS as means to measure risk for disease. Previous studies have demonstrated equivocal findings by comparing racial/ethnic differences in CVD using the FRS 2021. Our study, using NHANES data, discovered racial differences in a large cohort of participants in FRS outcomes. MA had the lowest FRS, followed by NH Asians, while NH Blacks and NH Whites had the highest FRS, thus supporting some previous studies. 30313233 while not supporting others 34. The FRS was originally developed based on a predominantly NH White population, which raises concern about its accuracy in predicting cardiovascular disease risk for other racial/ethnic groups. Our study suggests that the FRS may overestimate or underestimate risk for certain racial/ethnic groups.

Age-Adjusted death rates from 2019 provided by the National Center for Health Statistics22 (NCHS), listed in descending order with the highest death rate in NH Blacks, followed by NH Whites, Hispanic and finally NH Asian. Though the rankings in FRS in our study and age-adjusted death rates by NCHS were slightly different, both were similar in that the NCHS death rates were higher in the racial groups with higher FRS scores (NH Blacks and NH Whites) and lower in groups with the lowest FRS scores (MA and NH Asians).

In comparing racial/ethnic differences in FRS, there was a need to ascertain SES to discover if racial/ethnic differences could be related to the poverty ratio. Our study findings report that the poverty ratio was highest in NH Whites and NH Asians, with both of those groups having the highest (NH Whites) and second lowest (NH Asians) FRS scores. The MA group had the lowest poverty ratio, but the lowest FRS score. Previous literature 35 has suggested that the poverty ratio can predict FRS, but it should be noted in our study that the highest poverty ratio group had the greatest FRS, and the lowest poverty ratio group had the lowest, and theoretically less CVD risk based on the lowest FRS. These findings suggest that the poverty ratio may not be helpful in predicting differences in FRS between racial/ethnic groups.

When comparing FRS between racial/ethnic groups, there were several counterintuitive findings from our study. First, the highest FRS was in NH Whites. It should be noted that NH Whites were also the oldest group of participants who averaged 47 years of age, suggesting that age may have played role in the FRS differences. Also, the lowest FRS group was MA which were the youngest racial/ethnic group in our study who averaged 39 years of age, supporting the possibility that age, which is unequivocal in literature 36, is a predictor of CVD and affected the outcomes of this study, considering that age is a part of the FRS calculation.

Another consideration for the findings of lowest FRS in MA may be explained by what has been termed the Hispanic Paradox or Hispanic Epidemiological Paradox. This paradox was first reported by Markides and Eschbach and later by 37 Markides and Coreil.38 The term Hispanic paradox was later introduced in 2005 in the scientific literature39. Most recently this paradox suggests that Hispanics in the United States, though having higher poverty indexes than other racial counterparts, having lower levels of CVD40. Reasons that have been offered include lower infant mortality rates41, immigrants returning to their country of origin to die, known as the salmon effect 42 and cultural and lifestyle differences40. Our study supports what is called the Hispanic paradox in that the lowest risk, when using FRS, is associated with MA, even though they scored lowest on the poverty index in our study. Finally, measurement bias43 is another factor identified in the Hispanic paradox, and using the NHANES database could have measurement bias, yet, the size of the database could help to reduce the bias associated with self-reported data.

NH Blacks had both the highest HDL (a risk factor used in FRS calculations) and lowest total cholesterol levels (also used in FRS risk calculations) between racial/ethnic groups. Normally elevated HDL levels can be associated with decreased risk of CVD and can be partially responsible for a lower FRS score when compared to NH Whites and other multiracial groups. Yet, there is emerging evidence that dysfunctional HDL 4445 may cause HDL to not participate in reverse cholesterol transport. This dysfunction is associated with higher levels of total and LDL cholesterol; however, our study does not provide support for these associations as the best cholesterol profile was discovered in NH Blacks, and the highest risk based on cholesterol values, was in MA. Yet, the corresponding FRS scores suggested the second highest FRS was in NH Blacks and the lowest FRS in MA, which are counterintuitive based on cholesterol values. Again, this may be due to differences in age between these groups as MA was the youngest racial/ethnic group. Age combined with a hypothesis of HDL dysfunction, could suggest that FRS may not be as predictive as previously reported. Though differences in cholesterol values were statistically significant, the differences were not clinically significant between most racial/ethnic groups.

It should be noted that counter to previous studies 46474849 systolic blood pressure was statistically significantly higher (though not clinically different) in NH Blacks compared to all other racial/ethnic groups. Previous literature has reported differences between racial/ethnic groups 50 with previous studies suggesting the high risk in NH Blacks, followed by Hispanics, Asian and Caucasians 51. Our study does not support these findings suggesting that systolic blood pressure did not contribute to the FRS. Additionally, it is worth noting that the clinical differences between racial/ethnic groups were not meaningful.

Finally, NH Blacks recorded the second highest smoking levels with other/multi-racial having the highest compared to other racial/ethnic groups, which normally is associated with increased FRS 45 but is not supported by our findings. This increased risk based on smoking status, combined with the lowest cholesterol levels in NH Blacks, makes for a challenging interpretation of FRS. It should be noted that when comparing sex differences between racial/ethnic groups, NH Blacks had a higher percentage of females when compared to other racial/ethnic groups.

Many of the counterintuitive findings, though not a focus of our study, may be due to shortcomings that have been identified in the literature regarding FRS 50. FRS was developed based on data collected from a predominantly Caucasian population in the United States. As a result, the accuracy of FRS in predicting the risk of CVD in other populations may be limited 28. Second, the FRS does not consider several risk factors that have since been identified as important in the development of CVD, such as family history, diabetes, chronic kidney disease and obesity 29. Third, the FRS only predicts the risk of developing CVD over a 10-year period and does not consider the lifetime risk of developing CVD 30.

Despite its limitations, the FRS has been widely used in clinical practice to assess a patient's risk of developing CVD. The FRS can help clinicians identify patients who are at high risk of developing CVD and can guide the implementation of interventions to reduce this risk 31. Interventions may include lifestyle modifications such as dietary changes and exercise, as well as pharmacological interventions such as statins 32. The FRS can also be used to monitor changes in a patient's risk of developing CVD over time and to guide the implementation of additional interventions as needed 33.

Based on these findings, the FRS could be improved to make it more inclusive for diverse populations. A comprehensive update should integrate race/ethnicity-specific variables and socioeconomic factors that more accurately reflect cardiovascular risk in non-NH White populations. The current FRS does not explicitly account for the significant differences in cardiovascular risk across racial and ethnic groups. Our findings show that Mexican Americans and NH Asians have lower FRS scores while NH Blacks and NH Whites have higher scores. However, these results do not align fully with CVD outcomes or other risk factors like cholesterol levels. Incorporating race/ethnicity into the FRS calculation would adjust for these disparities, possibly aligning the score more closely with the actual CVD risk for each group

Our study also reveals a disconnect between poverty ratio and FRS outcomes across different racial/ethnic groups. For instance, Mexican Americans had the lowest poverty ratio but also the lowest FRS, contradicting previous assumptions about the link between SES and cardiovascular risk. Updating the FRS to include SES as a weighted factor may help better capture the effects of socioeconomic stress, access to healthcare, and lifestyle factors on cardiovascular risk. This adjustment would be especially beneficial for underserved populations, where SES plays a major role in health outcomes.

Several limitations occurred in our study. Data were collected cross-sectionally at single time points through multiple NHANES cohorts limiting the ability to make causal inferences. Some data in NHANES are self-reported which is subject to under and overreporting of certain variables. The NHANES data may be subject to selection biases, including nonresponse from individuals who did not provide data, potentially affecting the representativeness of the sample. The effect of adult health on poverty ratios may also introduce bias in the analysis, as one s health status can influence healthcare costs and financial stability 46. The large sample and the sample weighting technique helps overcome information bias from self-reported data, making findings from the study of importance while yielding generalizable results.

Future studies could also include a focus on critical gene expression that may be suppressed in various racial groups. For instance, Sirtuin 1 (SIRT1) is a NAD+-dependent deacetylase that plays a crucial role in cellular regulation, influencing processes such as metabolism, inflammation, and apoptosis. Its expression is found in various tissues, including the cardiovascular system, where it modulates endothelial function and smooth muscle cell behavior.53 SIRT1 has emerged as a significant modulator of several metabolic pathways that align closely with the FRS risk factors, which include hypertension, hyperlipidemia, diabetes, and smoking.54 SIRT1 is known to influence lipid metabolism and glucose homeostasis, effectively regulating the pathways that govern insulin sensitivity and cholesterol levels. By enhancing the activity of key enzymes involved in lipid oxidation and glucose metabolism, SIRT1 plays a protective role against the metabolic dysregulation associated with these risk factors.5355 Consequently, alterations in SIRT1 expression or activity could directly impact an individual's FRS profile, highlighting its potential as a biomarker for assessing cardiovascular risk. Inclusion of SIRTI levels in clinical practice as a predictor for CVD could help enhance clinical approaches to reduce the risk of disease outcomes.53