Study Design and Patient Population
This single-center, retrospective cohort study aimed to assess the occurrence of new MI or stroke in patients with TBPC (SBP < 130 mmHg) versus SBPC (SBP 130-139 mmHg). We addressed only SBP rather than both SBP and DBP based on similar previous studies including the SPRINT trial [5].
The data were collected and analyzed in 2018. We accessed the health system’s data stores (for medical records) to identify all outpatient encounters regardless of specialties (not limited to Family Medicine department) with ICD-10 diagnostic codes for hypertension in 2013. This resulted in 233,622 encounters in 88,456 patients. From this group we then applied inclusion and exclusion criteria to identify our study patients.
BPs have been measured by healthcare professionals manually with using a sphygmomanometer and a stethoscope in a typical outpatient setting in the same health system. We included patients with SBP < 140 mmHg between ages 40 and 79 years. The age range was based on the recent Centers for Disease Control and Prevention report showing that adults between ages 18 and 44 years mostly had hypertension under control and used antihypertensive medications less often [13]. We excluded the population greater than or equal to age 80 due to limited life expectancy. We averaged each patient’s SBPs measured in the same year between 2013 and 2015, so that one individual should have one averaged SBP reading per year. We excluded patients if the annual SBP measure was absent or if the averaged SBP fluctuated between TBPC and SBPC during the 3 years for BP characterization (2013-2015). We excluded patients with diabetes mellitus because the disease is a significant risk factor and a confounder in MI or stroke outcome. We excluded patients with a history of MI or stroke before 2013 because these are the highest risk factors for recurrent MI or stroke and we cannot assess a true incidence of new MI and stroke if we include these patients. We also excluded patients with MI or stroke events between 2013 and 2015. The attrition diagram (Figure 1) summarizes the enrollment process.
Outcome Measures
The main outcome measures compared between the TBPC and SBPC groups were any new incidence of MI or stroke which occurred within 2 years (2016-2017). Clinical settings of the cardiovascular outcomes included outpatient, emergency department, and inpatient. For ICD-10 coding, we included MI including “late effect” codes, but we did not include angina pectoris. For stroke we counted both ischemic and hemorrhagic strokes and “late effect” codes. “Late effect” means residual sequelae after the initial acute phase of the illness has resolved [14]. We did not include transient ischemic attacks given its potential diagnostic uncertainty. We also excluded traumatic hemorrhages or vascular syndromes (e.g., vertebrobasilar artery syndrome).
Variables
All the variables were collected through the database sorting system on Epi Info 7 provided by the Centers for Disease Control and Prevention (Atlanta, GA) per statistician (LL). We obtained baseline demographic data such as age, gender, race, and body mass index (BMI). We noted smoking status, serum low-density lipoprotein (LDL) levels, glomerular filtration rate, aspirin use, antihypertensive use, and statins use. Age was categorized by decade (40-49, 50-59, 60-69, and 70-79 years old). Race/ethnicity was categorized as African American, Asian, Hispanic, white, and other/unknown. For analysis, we further dichotomized race into 2 groups of African American and non-African American based on the recent Centers for Disease Control and Prevention report showing a relatively low percentage of hypertension visits among African Americans compared to other races [13]. BMI was categorized as underweight (BMI < 18.5 kg/m2), normal (BMI 18.5-24.9 kg/m2), overweight (BMI 25.0-29.9 kg/m2), obese (BMI 30.0-39.9 kg/m2), and morbidly obese (BMI ≥ 40.0 kg/m2) based on World Health Organization criteria [15]. Smoking status was categorized as “never” or “ever.” Serum LDL was categorized as ≥ 190 mg/dl or < 190 mg/dl; the former is one of the absolute indications of statin use based on 2013 ACC/AHA guidelines [16]. Glomerular filtration rate was categorized as ≥ 30 ml/min/1.73 m2 and < 30 ml/min/1.73 m2; the latter is the cutoff value where primary care physicians are recommended to refer patients to nephrology [17]. Use of aspirin, statins, and antihypertensives was determined from medication orders and defined as yes or no. In terms of counting antihypertensives, authors sorted it into major category of antihypertensives, for example, thiazide diuretic or calcium channel blockers, and then counted each medication that falls into each category of the antihypertensives per each patient during the enrollment period. This sorting process was done by statistician (LL).
Adverse Events
Authors believed it is important to evaluate hypotension-related adverse events in the TBPC group. However, given the retrospective nature of the study with using a large cohort, we were unable to collect data on the adverse events such as lightheadedness, dizziness, syncope, or falls in the TBPC group. However, we addressed and counted SBP < 90 mmHg in the TBPC group to find out the incidence of significant hypotension.
Statistical Analysis
Sample characteristics were described using means and standard deviations for continuous variables (SBP) and frequencies (number and percentage) for categorical variables (gender, race, smoking, BMI categories, use of statin, and use of aspirin). To compare the baseline variables between TBPC and SBPC groups, we conducted chi-square tests. For multivariate analysis to determine independent predictors for MI or stroke incidence, we performed binary logistic regression entering only those variables that showed significant difference between TBPC and SBPC groups with the only exception of the variable ‘antihypertensive use versus no use’ as this variable was considered to be relevant for the outcomes because antihypertensive use can modify the cardiovascular outcomes. We additionally included ‘the number of antihypertensives’ to build a different regression model to predict outcomes with excluding ‘antihypertensive use versus no use’. All statistical analyses were performed using Epi Info 7 provided by the Centers for Disease Control and Prevention (Atlanta, GA). A P value < 0.05 was considered statistically significant.