Gender differences in prevalence of myocardial infarction in rural West Texans

Abstract

Background

Heart disease is the leading cause of death in the United States. Incidence rates of myocardial infarction (MI) in rural West Texas signify a lack of effective, risk-specific prevention programs. The purpose of this study was to identify gender-specific risk factors for MI in rural West Texans.

Subjects and methods

Hospital patient data for those with and without a history of MI were obtained from the Project FRONTIER database for rural West Texas counties. We used statistical software, such as SPSS, R, and WinBUGS to detect and understand the nature of MI risk factors. Statistical methods including t-tests, Chi-squared, logistic regression, and a Bayesian approach were utilized to analyze data.

Results

MI significant risk factors obtained for females were systolic blood pressure (p = 0.002), diastolic blood pressure (p = 0.004), pulse (p = 0.015), and smoking (p = 0.002). For males, these were glucose (p = 0.022), age (p = 0.050), body fat (p = 0.034), and smoking (p = 0.017). The mean risk parameter followed a normal distribution, while the precision parameter depicted skew for both sexes.

Conclusions

Gender-specific differences in MI risk factors exist, and incorporating such variables can guide relevant policymaking to reduce MI incidence in rural West Texans.

Background

Although incidence rates of cardiovascular disease (CVD) have steadily decreased since 1920, it remains the leading cause of death in the United States (US) today (AHA 2017). The American Heart Association focuses attention on five key types of CVD, namely atrial fibrillation, congestive heart failure, coronary heart disease (CHD), high blood pressure, and stroke (AHA 2017). As of 2015, 41.5% of the US population had at least one of these conditions, with the projection for 2035 being 45% or 131.2 million people (AHA 2017). CHD is the most common type of CVD and develops as atherosclerotic plaque builds up within the coronary arteries. Myocardial infarctions (MIs), commonly known as heart attacks, occur when such a blockage in one or more of these arteries cuts off the blood supply to the heart, thereby damaging the heart muscle. It is mentioned that MI versus no MI is not the same as CVD versus no CVD. Also, after a diagnosis of MI has been made, treatment would probably be started, so a patient with an MI might be more likely to have had their risk factors modified towards better values than a patient without an MI. This could be a particular problem in places with poor access to primary care since people without symptoms may be less likely to receive treatment for risk factors in these rural areas than those with symptoms. It was estimated that 785,000 Americans will have a new coronary attack each year, 470,000 Americans will have a recurrent attack, and 195,000 Americans will have a silent first MI (Rogers et al. 2011). Although only about 16% of MIs are fatal, they, like other CVDs, still have a major economic effect (Mozaffarian et al. 2014). In 2016, CVD cost the US $555 billion, and by 2035 this is projected to increase to $1.1 trillion (AHA 2017). A better understanding of modifiable risk factors could aid in the reduction and costs of CVD.

Texas has the 19th highest CVD death rate compared to the rest of the country, and in 2013, CVD killed 40,203 Texans (AHA 2015). However, MI rates in Texas (3.9%) are similar to the rest of the nation (4.4%). Between 2012 and 2014, hospitalizations due to acute MIs were surprisingly high in a few rural West Texas counties (CDC 2012). Urban–rural differences exist for various forms of health and social conditions in rural West Texas, such as CVD, geriatric-related diseases, lack of primary care physicians, and lack of health insurance (Rohrer et al. 2003; Taylor et al. 2002). Previous studies have found that mortality rates due to CHD in rural counties are declining at a slower rate compared to urban areas (Kulshreshtma et al. 2014). Furthermore, rural populations in the South of Texas had higher age-adjusted CHD mortality rates compared with urban areas in the South of Texas (Kulshreshtma et al. 2014).

Risk factors for CVD, and thus for an MI, range from lifestyle choices to genetic factors, such as family history of heart disease or genetic predispositions. Demographic variables including age and sex have been shown to have a positive impact on one’s risk of developing CVD (Mozaffarian et al. 2014). Women have been shown to have poorer prognosis for MI when compared to men, for example, Zhang et al. 2012; Sadowski et al. 2011. In particular, women have a higher mortality rate for MI in-hospital when compared to men of all ages, which may be related to regionalization of care and corresponding policies (Zhang et al. 2012; Kilbourne et al. 2011).

Biological risk factors for MI include high body mass index (BMI), cholesterol levels, triglyceride levels, low-density lipoprotein (LDL) levels, blood pressure, and low high-density lipoprotein (HDL) levels (Mozaffarian et al. 2014). These risk factors are also associated with other chronic diseases, like diabetes mellitus, history of stroke, and obesity (Mozaffarian et al. 2014). Although higher fruit, vegetable, and legume intake are associated with lower risk of total mortality, they are not significant for cardiovascular-specific mortality (Miller et al. 2017). Smoking has also been proven to increase one’s risk of having an MI (Gidding et al. 1995; Houston et al. 2006; Pletcher et al. 2006). Furthermore, work-related psychosocial factors and job-related stress have also been found to be predictors of CVD (Becher et al. 2018; Pais et al. 2018). Given that Americans living in rural areas have a higher risk of dying from MI than Americans living in urban areas, an assessment of how risk factors affect the occurrence of MIs within rural counties could aid in identifying if such influences change based on location and cultural differences (CDC 2017).

Project FRONTIER (Facing Rural Obstacles to Healthcare Now Through Intervention, Education, and Research) is a longitudinal study initiated in 2006 and involving the three rural West Texas counties of Bailey, Parmer, and Cochran. This project aims to observe the long-term impact of a variety of chronic diseases on the physical, mental, and cognitive health of adults 40 years and older. Participants in this study often lacked access to affordable healthcare and human services because of their rural and isolated geographic location. Information collected could be used to develop programs for effective disease management and those focused on helping preserve cognitive function throughout lifespan. This will in turn improve the quality of overall health of individuals living in rural West Texas (FRONTIER 2006). County-specific data analysis was not appropriate for this study due to the small sample sizes for each individual county.

A number of studies have been carried out on the prevalence of MIs in rural areas, with several mentioning that geography can be one of the most significant risk factors for MI and CHD (Baldwin et al. 2004; Sheikh & Bullock et al. 2001; Taylor et al. 2002). In fact, inequalities are often magnified in rural areas, and disparities such as low socioeconomic status, food deserts, and lack of primary care physicians have also been shown to increase the risk of MI in persons over the age of 50 (Taylor et al. 2002; Pedigo et al. 2011). While studies looking at risk of MI in different geographic areas are more common, there are very few which address risk factors in the development of CVD specifically in rural West Texas. The purpose of the study was therefore to determine whether there are significant gender differences in MI risk factors in rural West Texas counties, and to identify the nature of the distributional pattern of risk parameter differences by gender.

Methods

Data source and study population

Project FRONTIER conducted data collection in a hospital setting for the three rural West Texas counties of Bailey, Cochran, and Parmer (Fig. 1). County descriptions including ethnicity percentage, health insurance percentage, poverty percentage, number of primary care physicians, and patient to primary care physician ratio are given in Table 1. All three counties had higher poverty rates than Texas overall (19%) as well as a higher patient to primary care physician ratio compared to the national benchmark of 603:1.

Fig. 1.

Age-adjusted mortality rates per 100,000 for ischemic heart disease by county, Texas, 2006–2015. Project FRONTIER collected data from three rural West Texas counties: Parmer County (n = 582), Bailey County (n = 201), and Cochran County (n = 426)

Table 1.

County description of ethnicity, health insurance, poverty, and physician care profiles

County	Caucasian (%)	Age > 64 w/o health in- surance (%)	In poverty (%)	No. of primary care physicians	Patient to primary care ratio
Bailey	93	33	19.9	10	707:1
Cochran	90.9	33.5	23.6	2	2008:1
Parmer	94.2	31.7	14.9	–	1326:1

– indicates that data is not available

No. = number

w/o = without

Due to the scarcity of physicians and the number of persons experiencing poor health, the study of chronic illnesses such as heart disease in these counties could prove very useful in identifying some of the most commonly seen risk factors for MIs in this geographic region (TDHHS 2013; Philips 2016a, b, c; USCB 2010).

Sample size and power calculation

The sample size was calculated using G*Power software (version 3.1.1). It was determined that 128 participants, 64 in each independent group, was sufficient to compare mean differences of continuous measurements with alpha (α) = 0.05, median effect size = 0.50, and power = 80% when running the independent sample t-test. It was calculated that a total of 88 subjects would be sufficient to detect a statistically significant relationship for two discrete variables with alpha (α) = 0.05, median effect size = 0.30, and power = 80% when running a Chi-squared test for a contingency table. Furthermore, it was determined that 435 subjects were required for logistic regression based on 13.3% males with MI, 5.1% females with MI, alpha (α) = 0.05, and a two-sided testing procedure.

Data was extracted from the Project FRONTIER database (n = 1527) in a hospital setting where annual patient check-ups were conducted within three counties, Bailey (n = 201), Cochran (n = 426), and Parmer (n = 582). Patients with missing data on MI history were removed by manual data cleaning, thus reducing the sample size (n = 1209). Among 382 men, 51 (13.4%) had a history of MI and 331 (86.6%) did not. Among 827 women, 42 (5.1%) had a history of MI and 785 (94.9%) did not. For both genders, survival times of MI patients were included in the study. The counties were combined due to their geographic and racial/ethnic similarities.

Statistical analysis

A summary of statistics was obtained for both men and women with and without a history of MI. IBM SPSS software (version 25.0) was used to perform descriptive and inferential statistics. Statistical methods such as independent sample t-tests for continuous variables and a Chi-squared test for discrete variables were used. Since the dependent variable was categorical (MI — “yes,” MI — “no”), a logistic regression method was performed to calculate odds ratios and 95% confidence intervals, and to determine an association between risk factors and the occurrence of an MI. To reduce the effect of multicollinearity, we conducted correlation analysis among independent variables and removed those which had a variance inflation factor (VIF) greater than 3 (William 2015).

Bayesian analysis

The Bayesian Markov chain Monte Carlo (MCMC) (Gelman et al. 2004) is a useful method to obtain the posterior distributions of risk parameters by interfacing R on WinBUGS software.

Assuming the observed continuous risk variable (y) is independent and normally distributed with mean (mu), then unknown variance (tau) and its sampling distribution follows y~normal (mu, tau). Since the distribution of y is interrelated with two parameters, we assumed the following were the prior distributions for mu and tau: mu~normal (theta’, tau.theta’) and tau~gamma (alpha, beta); and sigma = √(1/tau). Since we had a discrete risk variable for smoking, namely a “yes” or “no” response, the sampling distribution followed y~d-bi-nomial (theta, n) and assumed a prior distribution, theta~beta (a, b). After numerous iterations using the WinBUGS14 and R software, we were able to obtain estimates for each of the parameters mu, tau, sigma, and theta. While carrying out scientific studies, it is not uncommon that investigators have some previous knowledge from pilot studies about certain unknowns in the field of investigation. Diffuse or assumed prior can be used for tau and theta if information is unknown. Distributional patterns of MI risk variables will shed light on the hierarchy of significance of risk factors of having an MI, which in turn can guide policymaking to target particular lifestyle choices and biological risk factors in order to drive the reduction of MI incidence. Armed with the knowledge of which factors contribute the most to MI prevalence, public health practitioners can design suitable prevention programs which focus on these particular risks and most effectively cut down rates of MI in these rural vulnerable counties.

Results

Table 2 is a comparison of MI and no MI between men and women with respect to the mean risk factors cholesterol, HDL, glucose, LDL, triglycerides, abdominal circumference, percent body fat, BMI, height, weight, percent oxygen saturation, systolic blood pressure (SBP), diastolic blood pressure (DBP), and pulse.

Table 2.

Myocardial infarction (MI) risk profile of study participants by gender

	MI			No MI
	Men (Mean ± SD)	Women (Mean ± SD)	P value	Men (Mean ± SD)	Women (Mean ± SD)	P value
Cholesterol (mg/dl)	171.3 ± 38.6	189.0 ± 44.6	0.046*	188.0 ± 44.0	195.9 ± 40.1	0.004**
HDL (mg/dl)	42.1 ± 10.5	49.5 ± 18.6	0.020*	43.9 ± 13.3	52.5 ± 14.3	0.0001***
Glucose (mg/dl)	130.0 ± 61.2	129.9 ± 71.9	0.993	112.1 ± 36.1	111.1 ± 47.7	0.745
LDL (mg/dl)	97.2 ± 32.7	104.0 ± 40.6	0.378	112.8 ± 37.4	114.4 ± 35.1	0.505
Triglycerides (mg/dl)	174.7 ± 76.5	184.5 ± 119.9	0.637	172.5 ± 134.8	153.8 ± 97.5	0.010**
Abdominal circumference (inches)	40.3 ± 4.9	39.9 ± 5.2	0.678	39.6 ± 5.1	38.1 ± 6.4	0.0001***
Body fat (%)	28.0 ± 8.7	41.1 ± 3.9	0.0001***	28.8 ± 6.5	38.9 ± 7.5	0.0001***
BMI (km/m2)	29.7 ± 5.2	31.9 ± 7.0	0.081	29.5 ± 5.0	30.8 ± 6.6	0.001**
Height (inches)	68.8 ± 3.2	62.2 ± 3.2	0.0001***	67.8 ± 3.6	62.8 ± 3.2	0.0001***
Weight (lbs)	199.2 ± 35.0	175.1 ± 38.2	0.002**	192.7 ± 35.4	173.1 ± 39.4	0.0001***
Age (years)	65.3 ± 11.3	63.5 ± 11.3	0.233	59.8 ± 11.7	58.0 ± 12.8	0.030*
Blood oxygen (%)	94.8 ± 3.3	95.3 ± 1.9	0.418	95.1 ± 2.6	95.5 ± 2.7	0.044*
Pulse (beats/sec)	69.7 ± 11.3	69.2 ± 13.8	0.848	71.8 ± 12.9	74.3 ± 11.6	0.003**
DBP (mmHg)	75.1 ± 10.9	71.7 ± 12.2	0.173	77.9 ± 12.6	73.6 ± 11.2	0.0001***
SBP (mmHg)	133.2 ± 19.2	133.9 ± 21.4	0.871	130.8 ± 19.4	125.3 ± 16.7	0.0001***

^*p ≤ 0.05

^**≤ 0.01

^***≤ 0.0001

p value for comparing means (independent sample t-test)

SD = standard deviation

DBP = diastolic blood pressure

SBP = systolic blood pressure

HDL = high density lipoprotein

In case of no MI, women had higher cholesterol (p = 0.004), HDL (p = 0.0001), body fat (p = 0.0001), BMI (p = 0.001), blood oxygen (p = 0.044), and pulse (p = 0.003) compared with men. On the other hand, men had higher triglycerides (p = 0.010), abdominal circumference (p = 0.0001), height (p = 0.0001), weight (p = 0.0001), age (p = 0.030), DBP (p = 0.0001), and SBP (p = 0.0001) compared with women.

For patients with an MI, men had greater height (p < 0.0001) and weight (p = 0.002), while women had higher cholesterol (p = 0.046), HDL (p = 0.020), and body fat (p < 0.0001).

Fig. 2 is a visual representation of gender differences for those with MI. The means of the risk variables cholesterol (p < 0.05), HDL (p < 0.05), weight (p < 0.01), body fat (p < 0.001), and height (p < 0.001) were significantly different between men and women. High variations were observed among the variables of cholesterol, HDL, body fat percentage, height, and weight, while lower variations were seen for glucose, abdominal circumference, blood oxygen, pulse, and BMI (Fig. 2).

Fig. 2.

Risk factor differences between genders for those who have had a myocardial infarction within three rural West Texas counties. p * ≤ 0.05; ** ≤ 0.01; *** ≤ 0.001, p value for comparing means (independent sample t-test). BMI = body mass index; LDL = low density lipoprotein; HDL = high density lipoprotein; SBP = systolic blood pressure; DBP = diastolic blood pressure. Data were expressed in the standard units of each variable. Mean significant risk factors for cholesterol (p = 0.046) and HDL (p = 0.020) at the 5% level, for weight (p = 0.002) at the 1% level, and for body fat (p = 0.0001) and height (p = 0.0001) at the 0.1% level

Table 3 and Fig. 3 show the results of men and women with and without an MI in terms of the risk factor variables. For men with and without MI, there were statistically significant mean differences for the variables cholesterol (p = 0.012), glucose (p = 0.004), LDL (p = 0.006), and age (p = 0.002), with high variation among them (Table 3). In general, men with MI had higher glucose levels and were older while those without MI had higher cholesterol and LDL (Fig. 3a). Furthermore, blood oxygen, height, BMI, body fat, and abdominal circumference did not show any significant variance (Fig. 3a).

Table 3.

Myocardial infarction (MI) risk profile of study participants within men and women

	Men			Women
	MI (Mean ± SD)	No MI (Mean ± SD)	P value	MI (Mean ± SD)	No MI (Mean ± SD	P value
Cholesterol (mg/dl)	171.3 ± 38.6	188.0 ± 44.0	0.012*	189.0 ± 44.6	195.9 ± 40.1	0.281
HDL (mg/dl)	42.1 ± 10.5	43.9 ± 13.3	0.366	49.5 ± 18.6	52.5 ± 14.3	0.184
Glucose (mg/dl)	130.0 ± 61.2	112.1 ± 36.1	0.004**	129.9 ± 71.9	111.1 ± 47.7	0.016*
LDL (mg/dl)	97.2 ± 32.7	112.8 ± 37.4	0.006**	104.0 ± 40.6	114.4 ± 35.1	0.065
Triglycerides (mg/dl)	174.7 ± 76.5	172.5 ± 134.8	0.910	184.5 ± 119.9	153.8 ± 97.5	0.050*
Abdominal circumference (inches)	40.3 ± 4.9	39.6 ± 5.1	0.403	39.9 ± 5.2	38.1 ± 6.4	0.082
Body fat (%)	28.0 ± 8.7	28.8 ± 6.5	0.450	41.1 ± 3.9	38.9 ± 7.5	0.077
BMI (kg/m2)	29.7 ± 5.2	29.5 ± 5.0	0.773	31.9 ± 7.0	30.8 ± 6.6	0.277
Height (inches)	68.8 ± 3.2	67.8 ± 3.6	0.077	62.2 ± 3.2	62.8 ± 3.2	0.251
Weight (lbs)	199.2 ± 35.0	192.7 ± 35.4	0.222	175.1 ± 38.2	173.1 ± 39.4	0.747
Age (years)	65.3 ± 11.3	59.8 ± 11.7	0.002**	62.5 ± 11.3	58.0 ± 12.8	0.027*
Blood oxygen (%)	94.8 ± 3.3	95.1 ± 2.6	0.457	95.3 ± 1.9	95.5 ± 2.7	0.677
Pulse (beats/sec)	69.7 ± 11.3	71.8 ± 12.9	0.290	69.2 ± 13.8	74.3 ± 11.6	0.008**
DBP (mmHg)	75.1 ± 10.9	77.9 ± 12.6	0.142	71.7 ± 12.2	73.6 ± 11.2	0.298
SBP (mmHg)	133.2 ± 19.2	130.8 ± 19.4	0.410	133.9 ± 21.4	125.3 ± 16.7	0.001**

^*p ≤ 0.05

^**≤ 0.01

p value for comparing means (independent sample t-test)

SD = standard deviation

DBP = diastolic blood pressure

SBP = systolic blood pressure

HDL = high density lipoprotein

Fig. 3.

Risk factor differences or variations between those who have and have not had a myocardial infarction (MI) for men (a) and women (b) within three rural West Texas counties. p * ≤ 0.05; ** ≤ 0.01; *** ≤ 0.00, p value for comparing means (independent sample t-test). BMI = body mass index; LDL = low density lipoprotein; HDL = high density lipoprotein; SBP = systolic blood pressure; DBP = diastolic blood pressure. Data were expressed in the standard units of each variable. Mean significant risk factors for men (a) with MI: cholesterol (p = 0.012), glucose (p = 0.004), LDL (p = 0.006), and age (p = 0.002); and for women (b) with MI: glucose (p = 0.016), triglycerides (p = 0.050), age (p = 0.027), pulse (p = 0.008), and SBP (p = 0.001)

For women with and without MI, there were statistically significant mean differences in glucose (p = 0.016), triglycerides (p = 0.050), age (p = 0.027), SBP (p = 0.001), and pulse (p = 0.008) (Table 3). Women with an MI had higher glucose, triglycerides, age, and SBP, and lower pulse compared to women without a history of MI (Fig. 3b). Low variation existed for the risk factor variables of blood oxygen, BMI, abdominal circumference, and height (Fig. 3b).

Table 4 contains the results of the logistic regression for both men and women. For men, we found that glucose was the only risk variable constituting a significant risk factor. A modified analysis with VIF values ≥ 3 (age, HDL, glucose, triglycerides, abdominal circumference, body fat, blood oxygen, DBP, SBP, pulse, and smoking status) was conducted in order to address any potential multicollinearity problems among independent variables. Based on the logistic regression results, glucose (odds ratio (OR)= 1.009, p = 0.022), age (OR = 1.056, p = 0.050), body fat (OR = 0.952, p = 0.034), and smoking status (OR = 3.415, p = 0.017) all proved to be significant risk factors for men (Table 4).

Table 4.

Gender-specific logistic regression results for risk factors of myocardial infarction (MI)

	Beta (β) coefficient	SE	Wald statistic	Odds ratio	P value	95% CI
Men
HDL (mg/dl)	−0.012	0.016	0.616	0.988	0.433	(0.958, 1.019)
Glucose (mg/dl)	0.008	0.004	5.222	1.009	0.022*	(1.001, 1.016)
Triglycerides (mg/dl)	0.000	0.001	0.000	1.000	0.997	(0.997, 1.003)
Abdominal circumference (inches)	0.029	0.038	0.606	1.030	0.437	(0.956, 1.109)
Body fat (%)	−0.049	0.023	4.476	0.952	0.034*	(0.909, 0.996)
Age (years)	0.033	0.018	3.784	1.036	0.050*	(1.000, 1.074)
Blood oxygen (%)	−0.043	0.058	0.562	0.958	0.454	(0.855, 1.072)
SBP (mmHg)	−0.021	0.015	1.867	0.979	0.172	(0.950, 1.009)
DBP (mmHg)	−0.015	0.024	0.418	0.985	0.518	(0.940, 1.032)
Smoking status	1.228	0.514	5.713	3.415	0.017*	(1.247, 9.351)
Women
Glucose (mg/dl)	0.002	0.004	0.141	1.002	0.708	(0.993, 1.010)
Body fat (%)	0.023	0.033	0.481	1.023	0.488	(0.959, 1.092)
Abdominal circumference (inches)	0.029	0.036	0.638	1.029	0.424	(0.959, 1.105)
Age (years)	0.008	0.016	0.241	1.008	0.623	(0.976, 1.041)
Pulse (beats/min)	−0.043	0.018	5.867	0.958	0.015*	(0.926, 0.992)
SBP (mmHg)	0.046	0.015	9.317	1.047	0.002**	(1.016, 1.078)
DBP (mmHg)	−0.065	0.023	8.289	0.937	0.004**	(0.896, 0.979)
Smoking status	1.156	0.380	9.233	3.176	0.002**	(1.507, 6.692)

^*p ≤ 0.05

^**≤ 0.011

SE = standard error

CI = confidence interval

SD = standard deviation

DBP = diastolic blood pressure

SBP = systolic blood pressure

HDL = high density lipoprotein

For the women’s data, we found three variables to be significant risk factors, namely SBP, DBP, and pulse. To reduce the effect of multicollinearity, we then improved the model by ignoring independent variables which had a VIF value > 3. We again ran a logistic regression model based on reduced risk variables for glucose, body fat, abdominal circumference, age, SBP, DBP, pulse, and smoking status. This analysis revealed SBP (OR = 1.045, p = 0.002), DBP (OR =0.937, p = 0.004), pulse (OR = 0.958, p = 0.015), and smoking status (OR = 3.176, p = 0.002) as significant risk factors for women (Table 4).

We also performed a Chi-squared analysis to find whether there was a relationship between gender (with MI and no MI) and hypertension, hyperlipidemia, and diabetes. We only found a significant correlation between diabetes and women (p = 0.036).

Figure 4 depicts the frequency distribution of patients for each identified significant risk factor in both men and women. In men, glucose, body fat, age, and smoking status were the significant risk factors identified by the logistic regression analysis. The majority of the men had normal glucose levels (44%), were overweight (88%), and were smokers (71%). Their ages ranged from < 50 to 75+. In women, SBP, DBP, pulse, and smoking status were the significant risk factors. The majority of women had Stage 1 hypertension for SBP (39%), Stage 2 hypertension for DBP (73%), had a pulse rate between 70 and 79 (31%), and were non-smokers (61%).

Fig. 4.

Gender-specific frequencies for risk factors of myocardial infarction (MI) for men and women. MI for men (a), n = 371, including glucose level (1), body fat (2), age (3), and smoking status (4), and for women (b), n = 825, including systolic blood pressure (SBP) (1), diastolic blood pressure (DBP) (2), pulse rate (3), and smoking status (4). Glucose levels: low < 120 mg/dl, normal 120–140 mg/dl, impaired, 140–160 mg/dl, and diabetic >200 mg/dl. Body fat (men): athletic 6–13%, fitness 14–17%, average 18–24%, obese > 25%. SBP ranges: normal < 120 mmHg, elevated 120–129 mmHg, HTN stage 1 130–139 mmHg, HTN stage 2 140–179 mmHg, HTN crisis > 180 mmHg. DBP ranges: normal < 80 mmHg, HTN stage 1 80–89 mmHg, HTN stage 2 > 90 mmHg

Table 5 shows the summary results generated after 90,000 iterative samples for the posterior parameters based on significant risk factors.

Table 5.

Summary statistics for mean and precision parameters of myocardial infarction (MI) risk variables for men and women

	Mean	SD	95% CI
Men
Glucose (mg/dl)
mu	166	5.703	(154.4, 176.8)
sigma	39.56	4.222	(32.3, 48.81)
tau	6.605 E −4	1.378 E −4	(4.198 E −4, 9.582 E −4)
Body fat (%)
mu	29.74	0.8299	(31.36, 628.1)
sigma	5.061	0.607	(4.56, 6.934)
tau	0.03298	0.006996	(0.0208, 0.04809)
Age (years)
mu	65.1	1.606	(61.92, 68.23)
sigma	11.44	1.17	(9.416, 13.99)
tau	0.007881	0.001581	(0.005113, 0.01128)
Smoking status
theta	0.002212	1.23 E −4	(0.001979, 0.002209)
Women
SBP (mmHg)
mu	132.4	3.401	(125.6, 138.9)
sigma	21.9	2.508	(0.00906, 27.45)
tau	0.002166	4.841 E −4	(0.001327, 0.003215)
DBP (mmHg)
mu	71.46	1.933	(67.63, 75.23)
sigma	12.46	1.42	(10.05, 15.6)
tau	0.006683	0.001487	(0.004109, 0.009899)
Pulse (beats/min)
mu	68.87	2.206	(64.51, 73.18)
sigma	14.05	1.621	(11.3, 17.64)
tau	0.005262	0.001185	(0.003215, 0.007828)
Smoking status
theta	5.48 E −4 δ	2.84 E −5	(4.942 E −4, 6.051 E −4)

^δ5.48 E ⁻⁴ = 0.000548 (since 1/10,000 = 0.0001 = E ⁻⁴; 1.2000 = 0.0005 = 5 E ⁻⁴)

SD = standard deviation

CI = confidence interval

DBP = diastolic blood pressure

SBP = systolic blood pressure

Model for y~dnorm(mu, tau) and diffuse prior for mu and tau: mu~dnorm (0.00, 0.001), tau~dgamma(0.001,0.001), and sigma = sqrt (1/tau). The initial values were assumed to be 0.50 for mu and 0.0011 for tau

All simulations performed to 90,000 iterations

Figure 5 displays the posterior probability density for the parameters for each significant risk factor. Point estimators and credible intervals for the parameters can be determined from the posterior probability distribution. Figure 5 also exhibits whether the distribution of location parameter mu is normal or symmetric, and whether the distributions of precision parameters sigma and tau are asymmetric or skewed to the right. The parameter distribution for glucose, body fat, age, and smoking risk factor for men, and SBP, DBP, pulse, and smoking status for women can be approximated to normal for a large sample.

Fig. 5.

Gender-specific pattern of parameters for significant myocardial infarction risk factors. Normal distribution of mean parameter mu and skewed distribution for precision parameters tau, and sigma for significant risk factors for men (a), n = 371, including glucose level (1), body fat (2), age (3), and smoking status (4); and for women (b), n = 825, including systolic blood pressure (SBP) (1), diastolic blood pressure (DBP) (2), pulse rate (3), and smoking status (4)

Discussion

As a result of this Project FRONTIER study, gender differences became apparent in the prevalence of MI and its associated risk factors within rural West Texas. For patients with a history of MI, women had higher levels of cholesterol, HDL, and body fat compared to men. Previous literature has recognized gender differences due to pharmacological treatment and hormone response to an acute MI, which may play an indirect role in such blood levels (Pinkham et al. 2012; Jakobsen et al. 2012). For those without a history of MI, cholesterol level, HDL level, and blood oxygen were higher in women, whereas triglyceride level, SBP, and DBP were higher in men. Although there were only a few risk factor differences among those men with and without MI, examining the mean levels for the two categories gave rise to a larger issue. We found that mean SBP levels were at hypertensive levels according to the 2017 AHA guidelines, indicating that a number of men might be hypertensive, but not diagnosed as such (Whelton et al. 2018). Furthermore, body fat and BMI levels indicated that the majority of the patients were bordering on obesity, which is a major risk factor of CVD (Rosengren et al. 2005).

When comparing men who have and have not experienced a cardiac event, it seems as though rural West Texan men were becoming healthier following an MI due to significant decreases in total cholesterol, HDL, LDL, and triglycerides. Men with MI had glucose levels that were in diabetic ranges, while men without MI were within a normal range. Interestingly, a history of diabetes was not found to be a significant independent risk factor of an MI. This anomaly should be examined further in future studies. Other influences that significantly increased the odds of experiencing a cardiac event were height, age, and smoking status, as reinforced through a logistic regression analysis.

When examining cases in women with MI compared to those without MI, glucose levels, triglyceride levels, and SBP were higher, while pulse was lower. This is interesting, as you would expect those women with a history of MI with a lower pulse to be more physically active. This would indicate healthy lifestyle choices and lower cholesterol, glucose, and LDL levels, but this was not the case. Previous studies have found non-diabetic hyperglycemia to be highly associated with risk of MI, which may be due to stress levels (Gerstein et al. 1999; Shah et al. 2012). Furthermore, race/ethnic backgrounds, gender, and age all play a role in determining how elevated triglyceride levels puts one at a greater risk of MI and even death (Nordestgaard et al. 2007; Essilfe et al. 2016). In line with current literature, hypertension and diabetes incidence is higher in women overall compared to men (Hemingway et al. 2008; Egiziano et al. 2013). Such studies found relatively similar MI incidence rates between genders, which they indicate may be due to high MI mortality rates among women (Hemingway et al. 2008; Egiziano et al. 2013). This study did not include patients who had experienced fatal MI, and thus might not be a complete representation of the number of women who had had a heart attack in rural West Texas. Other significant risk factors for women in developing an MI were SBP, DBP, pulse rate, and smoking status. These findings are significant, since cardiovascular complications due to hypertension are greater in women than men (Anastos et al. 1991). With treatment, female all-cause mortality can decrease by 30%, thus illustrating a need for effective public health intervention within rural West Texas (Robitaille 1996).

Previous studies have found that diabetes was an independent predictor for MI, but none have identified sex-based differences (Tian et al. 2013). In the present study, diabetes was found to be an independent risk factor for women only. Due to multicollinearity and high VIF values, it was removed from the logistic regression model. Previous studies also found that diabetes was associated with greater MI mortality rates, especially for men (Wannamethee et al. 1995; Huxley et al. 2006). Although numerous articles have stressed this association, these findings were not present within our rural West Texas patients. One possible explanation is that the prevalence of unrecognized MIs for patients with diabetes is 2–7% higher than for those without diabetes (Scirica 2013). Diabetes might play a larger role, but due to the unrecognized and undiagnosed nature of the event, it goes unnoticed.

Predictive inferences for future trends were calculated using posterior distributions for the parameters of glucose, height, age, and smoking status for men, and SBP, DBP, pulse, and smoking status for women. A data-based normal distribution was generated; the sample was simulated 90,000 times and predictive curves were obtained. From these curves, it was possible to characterize the future trends of the prognosis of disease development for each variable as they pertained to each gender following an MI. The distribution of the parameters (mu and theta) for significant variables was normal, while the others (tau and sigma) were skewed.

The relationship between smoking status and risk of MI has been established for decades (Wilhelmsson et al. 1975). More recently, gender differences have also been identified, indicating that women are more likely to be affected by the toxins that are expelled from cigarettes and develop an MI at a younger age, thus decreasing their years of potential life by half as much compared to men (Prescott et al. 1998; Grundtvig et al. 2009). The present study found that patients of both sexes who have smoked in the past had a significantly higher risk of developing an MI.

Although it is commonly known that hypertension and hyperlipidemia are risk factors for CVD, we did not find similar results in this study for either men or women from a Chi-squared analysis. This may be due to a lack of proper knowledge of diagnosis prior to one’s MI, and may thus be a limitation of the study. Men had higher rates of smoking than women, and both men and women smokers showed a statistical association with MI. It was also verified that diabetes had an independent association with MI for women.

There is no other study found that has examined MI prevalence and its associated gender-specific risk factors within rural West Texan populations, so our findings are novel to that degree. More research needs to be conducted on under-served communities and rural West Texas in general in order to develop a better understanding of such MI differences in both genders. These findings can be generalized to rural Texas or other rural communities with similar demographics. Furthermore, occupational, environmental, racial/ethnic, and sociodemographic variables must be taken into consideration in order to grasp a more complete illustration of the health status of rural West Texans.

Study limitations

There are several limitations in this study. First, data were based on prevalent cases identified in a hospital setting. Berksonian bias may therefore apply, one consequence of which is that the information may not be generalizable to the population from which the cases were drawn (Berkson 1946; Snoep et al. 2014). Second, the data did not include information from persons who died from MI. By focusing on survivors, the data may be biased toward persons with less severe forms of the disease. Third, there was no longitudinal data in this report about the effects of treatment for those with MI and how this may have resulted in lower risk factor levels. Fourth, there was no review as to the appropriateness of treatment and follow-up. Fifth, it is not known if the present data reflect standardized collection measures. Specifically, there was no information as to whether values such as blood glucose and cholesterol were obtained in a fasting state. There were also no stated protocols for blood pressure measurement, for example details regarding how health care provider(s) performed measurements, time of most recent medication dosage, number of readings, resting time before and between measurements, how many measurements were taken, type of device and quality/maintenance of the device (Giorgini et al. 2014), cuff size, position of the patient, and other factors. Similarly, there was no documentation available as to protocols and/or quality assurance for abdominal measurement. Finally, pulse but not heart rate was recorded, meaning that cases in which heart rate was different from pulse rate would be missed. Moreover, since the data were clinic based and clinic access may be limited, it cannot be certain that the data reflect values for the county populations as a whole. Another set of limitations involves patient recall bias on the sequence of events. Years of diagnosis were often missing, and specific diagnoses were sometimes incorrect on the part of the patient. Other relevant comorbidities such as physical activity level, dietary choice, genetic predisposition, and certain heart disorders were not available in the present data and are still being collected by the Project FRONTIER.

Conclusions

This study considered Project FRONTIER adult patient data obtained from three similar geographic counties, Bailey, Cochran, and Parmer, in order to identify and understand the risk factors for MI, specifically in rural West Texas. Glucose, age, body fat, and smoking were identified as significant risk factors for men, and SBP, DBP, pulse rate, and smoking were found to be significant risk factors for women. Population-based, analytic epidemiologic study would be needed to confirm or deny hypotheses based on these observations.

Due to the findings concerning smoking as a significant risk factor for both genders, we recommend population-based epidemiologic research to estimate the potential benefit of targeted health care and public health efforts within rural West Texas communities, to specifically help improve the health of men and women through a reduction in smoking. If such targeting is supported by future research, methods to decrease smoking rates might include stricter smoking bans within restaurant and bar settings and/or city-wide, and the promotion of nicotine gum for current addicts. We also found that economically dis-advantaged patients and those lacking health insurance often did not receive adequate or appropriate care for their diagnosis, which can adversely affect their prognosis. Recommendations may include improving treatment options for these individuals and increasing policy efforts to ensure more comprehensive screening coverage. Further research is needed to identify genetic, environmental, and other related demographic and socioeconomic risk factors that contribute to MI.

The 2013–2017 Texas Plan to Reduce Cardiovascular Disease and Stroke, initiated by the Texas Department of State Health Services (TDSHS), aims to improve risk assessment and screening in vulnerable Texan areas as well as improve treatment programs and outreach. Not only does the policy aim to focus on individuals who are at a higher risk of stroke or heart attack, it also aims to reduce inequalities and poor lifestyle choices which can lead to a greater risk. Such targets include adult obesity, smoking, alcoholism and substance use disorders, the use of tobacco, lack of physical activity, and unhealthy food choices. Policies such as these are priceless in terms of improving health outcomes in vulnerable populations as well as across the state (TDSHS 2017).