Comparative Effectiveness of Antihypertensive Therapeutic Classes and Treatment Strategies in the Initiation of Therapy in Primary Care Patients: A Distributed Ambulatory Research in Therapeutics Network (DARTNet) Study

Data Source

This study was conducted using data from primary care clinics participating in the Distributed Ambulatory Research in Therapeutics Network (DARTNet) collaborative, a federated network of electronic health record (EHR) data that has as one of its objectives the facilitation of observational comparative effectiveness research.^12⇓–14 DARTNet, in collaboration with QED Clinical, Inc. (doing business as CINA; http://www.cina-us.com/), has developed data extraction, transformation, and loading (ETL) processes that allow aggregation of data from disparate EHRs into a limited database. All data were imported nightly from the organization EHR to a relational clinical data repository (CDR) located behind the firewall of each organization. The CINA software used for ETL was already in place and being used by each organization to produce reports of clinical decision support and population management at the point of care.

Data validation was largely the responsibility of CINA, as the ETL vendor in place at each organization before the initiation of this project. Because CINA provides software tools that utilize data from the CDR in the course of clinical care and decision making, CINA has several processes in place to ensure the reliability and validity of the data that is contained within the CDR. Data reliability testing by CINA includes the following: (1) patient-level sampling comparing the data imported into the CDR with the source data as it is represented in the EHR; (2) daily use of the data in the CDR in clinical practice through the point-of-care clinical decision support tool and population management tools provided by CINA; and (3) data reliability testing with data extraction for research analysis.

The DARNet Cardiovascular Risk Reduction Learning Community was designed to provide patient-specific clinical decision support at the point of care and an audit with feedback on national guidelines (Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure [JNC7]) for the control of blood pressure (unpublished data). The Cardiovascular Risk Reduction Learning Community limited data set obtained from the DARTNet collaborative also was used for the current study, which includes 33 primary care clinics from 10 health organizations and approximately 154 clinicians providing care to more than 250,000 patients.

Data use agreements for access to the limited data set were obtained from each organization, and a waiver of informed consent and Health Insurance Portability and Accountability Act authorization were approved by the American Academy of Family Physicians' institutional review board and the Colorado Multiple Institutional Review Board of the University of Colorado Denver.

Data Collection and Cleaning

The data used in the present analyses included patient demographics, height, weight, blood pressure, comorbidities (International Classification of Diseases, 9th revision, codes from the problem lists and reasons for visit), medications, laboratory data, and dates of encounters between August 2001 and August 2011. Data on patient race/ethnicity and frequency of medication dosing were sparsely populated and were not included in our analyses.

For continuous variables, physiologically implausible values were identified by clinicians examining the distributions of the variables. These consensus-derived, physiologically implausible values were systolic blood pressure <50 or >260 mm Hg, diastolic blood pressure <0 or >200 mm Hg, height <45 or >90 inches, weight <50 or >500 lb, and serum creatinine <0.2 or >20 mg/dL and were excluded in the current study. The proportion of values deleted varied from 0.005% for systolic blood pressure to 0.5% for serum creatinine. In addition, height and weight were missing for 4.2% and 0.3% of patients, respectively, and were replaced with sex-specific mean values for height (women, 63.9 inches; men, 69.8 inches) and weight (women, 178.9 lb; men, 212.9 lb).

Definitions

A diagnosis of hyperlipidemia was defined as an active International Classification of Diseases, 9th revision, code from 272.xx during at least one visit, hypertension as a code from 401 to 405.xx or 437.2 during at least one visit, diabetes mellitus as codes from 250.xx during at least 2 visits or ≥1 antidiabetic medication, and chronic kidney disease (CKD) as a code from 403 to 404.xx, 581 to 582.x, 585 to 586.x, V45.11, V45.12, or V56.x during at least one visit or a calculated glomerular filtration rate¹⁵ of >60 mL/min/1.73 m². Patients were assumed to be white for purposes of calculating glomerular filtration rate. Finally, therapeutic goals were defined per the JNC7 (<130/80 mm Hg for patients with CKD or diabetes mellitus and <140/90 for all others).⁵

Antihypertensive agents were categorized into the following therapeutic classes: angiotensin-converting enzyme inhibitor (ACEI), angiotensin II receptor blockers (ARB), cardioselective β-blockers, calcium channel blockers (CCBs), and thiazides and thiazide-like diuretics (thiazide) using the Medi-Span Master Drug Data Base version 2.5 (Medi-Span/Wolters Kluwer Health, Indianapolis, IN). FDCs and FECs were defined as being composed of 2 of the 5 monotherapy therapeutic classes. All other antihypertensive therapeutic classes were excluded because of numbers insufficient for adequate analyses.

Patient Inclusion

Figure 1 defines the inclusion criteria for this study. The index date was defined as the date of the first prescription of an antihypertensive agent, before which the patient had been followed with no antihypertensive prescriptions for ≥12 months. The index blood pressure was the value closest to and falling within the 2-month interval before the initiation of antihypertensive therapy. If the patient's treatment was classified as FEC, the second drug must have been started within 3 days of the first. The blood pressure value used to assess reduction and goal attainment from the index pressure was the one with the latest date/time stamp within the time period of 1 week to 1 year following the index time point during which no change had been made in the antihypertensive drug regime.

• Download figure
• Open in new tab

Figure 1.

Timeline defining key events in the study for patients initially using antihypertensive agents. BP, blood pressure; Rx, prescription.

Figure 2 shows the STROBE diagram for inclusion of patients in our analyses. There were 79,176 patients ≥18 years old who were receiving at least one antihypertensive agent. Exclusions were as follows: (1) 51,454 patients classified as prevalent users of antihypertensive agents because there was no period ≥12 months in duration when they received no antihypertensive agents; (2) 4,391 patients who had no blood pressure measured in the 2 months before initiation of antihypertensive therapy; (3) 10,711 patients whose blood pressure was at goal in the 2 months before initiating antihypertensive therapy; (4) 2,451 patients taking ≥3 antihypertensive agents; and (5) 1,493 patients with no blood pressure measurements available following the initiation of antihypertensive therapy. The final analytic subset consists of 8,676 patients (11.0%) of the 79,176 patients ≥18 years old who had received at least one prescription for an antihypertensive agent and had elevated blood pressures (as defined by the JNC7) before or on the index date. The outcomes evaluated in this study were changes in follow-up systolic and diastolic blood pressures from index blood pressures and JNC7 therapeutic goal attainment rates at follow-up.

• Download figure
• Open in new tab

Figure 2.

STROBE diagram of included patients with initial use of antihypertensive agents (anti-HTN). Apt, appointment; BP, blood pressure; CKD, chronic kidney disease.

Statistical Analyses

To characterize the study population, we calculated descriptive statistics using means and standard deviations for continuous variables and compared the different antihypertensive therapeutic classes using analyses of variance. For categorical variables, frequencies and percentages were calculated and logistic regression was used to compare patient characteristics between the groups.

We used analysis of covariance to model mean reductions in blood pressures and logistic regression to compare the proportion of patients achieving blood pressure control in each group. To account for patient differences, we constructed multivariable prediction models using the covariates listed in Table 1, which were chosen using clinical judgment and prior research. All covariates were included in each risk-adjusted model, except in the case of mean reductions in blood pressures, for which baseline diastolic blood pressure was not included in the model of mean reductions in systolic blood pressure and vice versa. Hierarchical linear models including clinic and organization also were evaluated and gave similar results (data not shown).

Risk-adjusted outcomes for each antihypertensive therapeutic class were calculated from predicted values obtained from models fitted with covariates only. Means and standard deviations of the predicted values were calculated to obtain risk-adjusted average reductions in blood pressures. For risk-adjusted control rates, expected rates of control for each antihypertensive therapeutic class were compared with the observed rates of control as an observed-to-expected ratio, which was converted to standardized rates by multiplying each observed-to-expected ratio by the observed rate of control for all patients across all antihypertensive therapeutic classes.

For all outcomes, P values were adjusted for multiple tests using the Bonferroni method. All statistical tests were considered to be significant at a 2-sided P < .05. All analyses were performed using SAS software version 9.3 (SAS Inc., Cary, NC).