Validation of a Health System Measure To Capture Intensive Medication Treatment of Hypertension

Abstract

Importance:

Blood pressure (BP) targets are the sole measure of high-quality hypertension care in health systems. However, BP alone does not reflect intensity of pharmacologic treatment, which should be carefully managed in older patients.

Objectives:

To develop and validate an electronic medical record (EMR) data-only algorithm (using pharmacy and BP data) to capture Intensive Hypertension Care (IHC), defined as ≥3 BP medications AND BP <120 mmHg. In addition, we sought to identify conditions that contribute to greater IHC, either through greater algorithm false-positive IHC, or by contributing clinically to delivering more IHC.

Setting:

Veterans Health Administration (VHA)

Participants:

319 older (age ≥65) randomly-selected Veterans with intensively-treated hypertension, July 2009-June 2011 who had 3625 primary care visits.

Exposure:

We calibrated and measured the algorithm for IHC (Algorithm-IHC).

Main Outcomes and Measures:

For each visit, we determined the reference standard, Clinical-IHC, by detailed chart review of free-text clinical notes. We calculated the correlation in BP medication count between the EHR versus the reference standard, and the sensitivity and specificity of Algorithm-IHC. In addition, we measured presence vs absence of Contributing Conditions (CC) acting in combination with hypertension management to result in IHC (IHC+CC): (1) an acute condition that lowered BP (e.g., dehydration), (2) another condition requiring a BP target lower than the standard 140 mmHg (e.g., diabetes), or (3) the Veteran needing a BP-lowering medication for a non-hypertension condition (e.g., beta blocker for atrial fibrillation) resulting in low BP.

Results:

For 319 patients with 3625 visits (mean age 75.6 years; 99.1% male), one-quarter (25.13%) of the visits had Clinical-IHC by the reference standard. The algorithm for determining medication count was highly correlated with the reference (r=.84). Sensitivity of clinical IHC was 92.2% [95% CI 90.3%–93.9%] and specificity was 97.2% [96.5%–97.8%], suggesting that Clinical-IHC can be identified from routinely-collected data. Only 2.07% of the 3635 visits were Algorithm-IHC false-positives, 1.52% involved IHC+CC, and 3.45% involved either false-positive or IHC+CC. Among select comorbid conditions, congestive heart failure (5.24%) was most associated with a pre-specified >5% combined false-positive or IHC+CC rate.

Conclusions and Relevance:

Health system data can be used reliably to gauge IHC.

Background

Lower blood pressure (BP), and the intensive treatment with BP medications required to achieve it, prevents strokes and death.^1–5 However, if achieving low BP requires multiple medicines, this could increase side effects and added treatment burden. Our current system of performance measures is designed to increase BP control, but it doesn’t discourage intensive treatment that far exceeds the target, e.g., by ≥20mmHg without cardiovascular risk. In addition, BP declines naturally before death^6,7 and may be less beneficial among frailer adults.⁸ Therefore, further research is needed to support more clinically-nuanced care for older adults who might not need such intensive hypertension care (IHC). Before such research can be conducted in health systems, however, the first step is to determine whether IHC using multiple BP medications can be identified using health system data.

We have previously described approaches to assess both BP level and medication treatment simultaneously from extractable electronic health record (EHR) data fields to improve large-scale health system efforts to achieve appropriate BP treatment.^9–12 There are, however, reasons why the EHR measure could be misleading about the actual care provided. A patient may fill a medication, but then later discontinue it due to side effects, therefore appearing to be on more medications than the pharmacy data would suggest.

We conducted this study to develop and validate the accuracy of an electronic measure of Intensive Hypertension Care (IHC) for older patients in the Veterans Health Administration (VHA). In addition, we sought to understand which co-morbid conditions, in presence of hypertension treatment, could potentially contribute synergistically to the delivery of IHC, e.g., by increasing measurement error or because their management also requires BP medication treatment. The results would allow patients with those conditions to be analyzed separately or excluded from future quality improvement initiatives or health outcomes research.

Methods

Data

We used 2009–13 national VA Clinical Data Warehouse (CDW) data, including all outpatient clinical encounters, vital signs, diagnoses, pharmacy records, and a file of non-VHA medications. We also linked 2009–2013 Medicare Part D drug claims to CDW data. Informed consent was waived (IRB#2015-286).

Sample

We identified a cohort of older patients (age ≥65 years) with hypertension receiving VHA primary care during a two-year study period (July 1, 2011-June 30, 2013). Eligibility was determined during July 1, 2009-June 30,2011 by ≥1 hypertension (ICD 401.x) diagnosis and established primary care (≥2 primary care visits with ≥1 visits in 2009–10).

Unit of analysis was each eligible visit with physicians, nurses, advanced practice providers, pharmacists, and social workers in primary care (family medicine, general/internal medicine, geriatrics, mental health primary care) or specialty care that manage hypertension (nephrology, endocrinology, cardiology, neurology outpatient non-emergency care) (eTable 1). We defined eligible visits as days when BP measurements occurred on the same day as eligible services. We excluded visits occurring on a day with nursing home or hospital care. SAS code is provided for constructing the measures (eTable 2).

In order to enhance our sample with a greater opportunity to study variations in IHC than in a general VA sample, we used a preliminary algorithm (i.e., prior to calibration) of IHC to identify patients with a continuous period of intensive hypertension care:¹² SBP <120 mmHg on ≥2 consecutive primary care visits, with both visits preceded by VHA BP medication fills of ≥3 different classes (any dose) within 100 days. Out of 1,260,150 older VHA patients with hypertension, the 66,412 Veterans (4.8%) meeting these criteria had an average of 11.5 visits during the 2 years. Only 12% had uninterrupted IHC throughout the observation time. The overall mixture of visits was balanced between 79,704 episodes of contiguous IHC and 55,539 isolated IHC visits. From this sampling frame, we randomly-selected 420 patients for further study.

Overview of study design (Figure 1)

Figure 1. Algorithm Development and Validation Strategy.

Flow of data for 1.26 million older Veterans, from data source to sampling for chart review, to analytic sample.

* We considered a visit as a calendar day on which patient had an eligible encounter AND a recorded BP measurement

BP = Blood pressure

This is an observational validation study. We first developed an algorithm to measure IHC using EHR extractable data fields only (the “test”). Then we reviewed the text-based clinical notes for documented BP medications, BP, and the clinical care provided (the “reference standard”). Last, we measured the accuracy of the IHC EHR data algorithm against the reference.

Algorithm development using pharmacy and BP EHR data only (the Test)

To calibrate the medication algorithm, we structured the EHR data as a longitudinal series of BP medication fill events (from VA data only) before and after each outpatient visit (Figure 2). BP-lowering medications were organized into classes: (1) angiotensin agents as a single class containing both angiotensin converting enzyme inhibitors [ACEI] and angiotensin receptor blockers [ARB], (2) calcium channel blockers, (3) thiazide and (4) potassium-sparing diuretics, (5) beta blockers, (6) centrally-acting alpha-agonists, (7) other vasodilators, and (8) direct renin blockers (eTable 3). Fills of different medications within a class would be considered a switch between the medications rather than an additional medication.

Figure 2. Final BP medication identification algorithm.

A patient is considered to be taking a continuous medication on the date of valid visit (primary or specialties that manage BP) if there is a fill prior to the visit within 186 days (the pre-fill) AND a fill after the visit within 186 days (the re-fill) within the same class of medications. Pre-filled medications without a matching re-fill are considered to be discontinued medications. Among discontinued medications, there are two types: new and old. If the visit occurs within 80% of the day supply of a new pre-filled medication (no fills within that class in the year prior to the pre-fill) then the medication is most likely to be active on the day of the visit. The visit must occur within 90% of the day supply of old pre-filled medication to be considered to be active.

At each visit, we considered whether or not a patient was on a BP medication class using the pharmacy fill data prior to (the “pre-fill”) and after each visit (the “re-fill”). First, we considered continuously-refilled BP medications. A pre-fill and a re-fill in the same class (Figure 2) regardless of the dose was considered a match. For example, if patient had a lisinopril pre-fill, then a subsequent losartan fill (representing a switch within class) would be considered as the matched re-fill. We calibrated the algorithm on a subsample of 30 patients who had 455 visits and 1749 medication fill events. We first varied the criteria for days between the pre-fill and the eligible visit (look-back period) between 120 and 365 days to assess changes in false-positives (positive algorithm test but the chart review determined non-use) and false-negatives (negative algorithm test but the chart review documented use). Second, we similarly varied the re-fill criteria. We allowed for time greater than the typical 90-day supply in order to bridge over a refill at a non-VA pharmacy or verbal instructions to decrease the dose (e.g., pill-splitting which would increase the effective supply).

Next, we developed criteria for discontinued medications. If we found a medication with a pre-fill but no re-fill, then we considered the medication to be potentially discontinued, i.e., not active on the visit day. For each medication, we calculated time elapsed since the pre-fill date divided by number of supplied days, resulting in a percent of days’ supply elapsed by the visit day (range from 0% to more than 100%). For example, if the pre-fill was for a 90-day supply and the visit was 30 days since the fill, then 33% of the supply days were elapsed. We then calibrated the most accurate threshold in percent of days’ supply elapsed. We calibrated these thresholds separately depending on whether the potentially discontinued medication was chronic (defined as pre-fill with a prior fill within the past year) or new. Compared to continued medications, fewer medications were discontinued, so we had to use the entire analytic sample (n=3625 visits) to calibrate the discontinuation algorithm.

After algorithm calibration, we considered additional pharmacy data from two additional sources. First, we considered the non-VA medication file maintained voluntarily by VA clinicians. Despite reducing false negative rate by 7.5%, this data source worsened false positives by 45%. Therefore, we dropped this data source from further analysis. Second, 67 (21.0%) of the patients also utilized Medicare Part D. The proportion of visits with ≥3 medications increased from 63.81% in the VA-only pharmacy data to 67.82% after using Medicare D data, so we retained Part D data for subsequent analysis.

Medical record review (Reference-standard)

Trained chart abstractors reviewed free-text clinical notes to obtain key elements for comparison with EHR data. First, abstractors added documented BPs on eligible visits not already captured in the EHR data, e.g., if manually rechecked during the visit or from a home BP log on the same day as the visit. Next, to determine the daily BP medications, reviewers read all eligible visit notes as well as interval notes between eligible visits (i.e., telephone notes, emergency visits, hospital discharge summaries), including scanned non-VA notes.

Medical Record Review of Visits for Contributing Conditions

We reviewed the clinical notes for three types of Contributing Conditions (CC), circumstances where a condition contributed to having IHC at the visit (i.e., IHC+CC), i.e., acting in combination synergistically with hypertension treatment. The three IHC+CC situations were: (1) an acute condition spuriously lowering BP (e.g., dehydration), (2) another chronic condition requiring a BP target lower than <140 mmHg (e.g., diabetes), and (3) need to take a BP-lowering medication for a non-hypertension chronic condition (e.g., a beta blocker for heart rate control in atrial fibrillation) that was documented as causing low BP. More than one CC could be coded for any IHC+CC visit.

Ten percent of the records were abstracted by two coders for interrater reliability in presence vs absence of each class of BP medications and any CCs versus no CCs (pooled kappa .92).

Analysis

We used the chart review as the reference standard and the medication algorithm as the test to count the number of medications at each visit. If more than one SBP was measured on a day, then we calculated the SBP as the mean, in order to reflect underlying true BP. If the clinical notes documented additional SBPs (i.e., rechecked by the clinician, or noted on a home BP log), then it was possible for the reference to differ from the EHR-only SBP.

We calculated (1) agreement (kappa) between the EHR and reference standard for the presence versus absence of each BP medication class and (2) the correlation in total number of meds at each visit. Next, we classified each visit as representing IHC or not (≥3 BP classes AND mean SBP <120 mmHg) by the reference standard and by the final developed algorithm (Figure 2). We calculated sensitivity, specificity, positive predictive value, and negative predictive value of the Algorithm-IHC. All calculations adjusted standard errors by clustering of visits within patient. Then, we added the false positive rate with the IHC+CC rate, to estimate the percent of visits that the Algorithm-IHC potentially overestimated hypertension care intensity. We calculated this summary measure in the overall sample and by chronic condition subsamples (diabetes, coronary heart disease, cardiovascular disease, arrhythmia, heart failure (HF) chronic kidney disease, psychiatric conditions and benign prostatic hypertrophy (BPH)). We set a 5% rate as the clinically-significant proportion of false positives (either due to a data error or because a comorbidity was the reason for IHC). We hypothesized that IHC would be more likely to be overestimated (due to false positive and IHC+CC) for conditions also treated with BP-lowering medications (e.g. HF). The chronic conditions were determined by any coded visit diagnoses during the observation window. Then, all visits for that patient were considered associated with those conditions.

Results

On the randomly-sampled 420 patients, we used 31 charts to train chart abstractors. We discarded these charts. Of the remaining 389 patients, we excluded 58 patients who died during the observation period, 7 who spent ≥10% of the study time in inpatient/nursing home care, 3 patients enrolled in a hypertension treatment trial, and 2 for missing Medicare pharmacy files (Figure 1). The analytic sample of 3625 visits for the remaining 319 patients (mean 11.4 visits per patient) was 99.1% male, with a mean age of 75.6 years. The prevalence of other chronic conditions was comparable to the general older VA population:¹³ 30.1% for diabetes, 19.5% for congestive heart failure, and 28.7% for coronary artery disease (Table 1). The reference standard mean count of BP medications was 2.58 (SD 0.96). The mean SBP was 121.8 mmHg (SD 16.4). Of the 3625 visits, 911 visits (25.1%) met both SBP (<120 mmHg) and medication criteria (≥3 medication classes) for IHC by the reference-standard chart review (Table 1).

Table 1.

Sample Characteristics (N=3,625 visits)

Variable		Mean (SD, range) or N (%)
Age (years)		75.6 (7.2, 65–96)
Gender (Male)		3592 (99.1%)
Low BP < 120 mmHg (y/n) (mean of all BPs documented in the clinical notes)		1721 (47.5%)
On 3+ medications (by chart review reference standard)		1895 (52.2%)
Low BP and 3+ BP meds (by chart review reference standard)		911 (25.1%)
Chronic condition	Diabetes mellitus	1090 (30.1%)
	Atherosclerotic heart, cerebrovascular, or peripheral vascular disease	1042 (28.7%)
	Arrhythmia	738 (20.4%)
	Heart failure (any type) or any valve disease	706 (19.5%)
	Kidney disease	372 (10.3%)
	Psychiatric disease	341 (9.4%)
	Benign prostatic hypertrophy	261 (7.2%)
Medication Class (any during study time period)	Beta blocker	2913 (80.4%)
	Angiotensin converting enzyme inhibitor or Angiotensin receptor blocker	2856 (78.8%)
	Calcium channel blocker	1524 (42.0%)
	Thiazide diuretic	1195 (33.0%)
	Potassium sparing diuretic	584 (16.1%)
	Vasodilator	140 (3.9%)
	Centrally-acting alpha blocker	117 (3.2%)
	Direct renin blocker	24 (0.7%)

Algorithm development to determine active medication use on each visit day

For calibration of continuous refills, on the subset of 30 patients over 455 visits, we found a steady decrease in false negatives when broadening the lookback window from 14.2% at 120 days to 10.8% at 365 days. The false positive rate was low and varied little when the lookback window changed, starting at 1.19% at 120 days, increasing to 1.27% at 136 days, but increasing to 1.34% between 180 and 200 days. We set our pre-fill and refill at 186 days (31 days x 6 months).

On the full sample, the 3625 visits were associated with 10,594 medications with valid pre-fills within 186 days. After 9017 (85.1%) of those pre-filled medications were matched with a valid refill, we calibrated the remaining prefilled medications without a valid refill (1,577 medications, 14.9%) as potentially discontinued medications (eFigure 1). The false positive and negative rates were optimized if the visit date occurred before 80% of the days’ supply elapsed for discontinuation of new medications and before 90% for old medications. The discontinuation criteria restored 511 (4.8%) true positive medications in comparison to an algorithm that required medications to be refilled in order to be counted. These analyses resulted in the final algorithm (Figure 2).

Validating algorithm to identify IHC

On the analytic sample of 319 persons, the count of BP medications in the algorithm (mean 2.59 [SD .98], range 0–6) versus the reference (mean 2.58 [SD .96], range 0–6) was highly correlated (r=.84 [95%CI .79–.89]). The correlation was also high for the algorithm applied to VA-only (i.e., without Medicare Part D) pharmacy data (r=.84). The most common medications were ACEI/ARBs (79.2%) and beta-blockers (80.0%). The agreement (kappa) for each class ranged from .83 (beta-blocker and ACEI/ARBs) to .96 for potassium-sparing diuretics.

We found positive Algorithm-IHC (the test) at 915 visits (25.3% of the total visits) versus Clinical-IHC (the reference standard) at 911 visits (25.1% of the visits). The sensitivity was 92.2% [95% CI 89.3%–95.1%], specificity 97.2% [96.1%–98.3%], positive predictive value 91.8% [88.5%–95.1%], and negative predictive value97.4% [96.4%–98.4%], suggesting that IHC could be validly identified from routinely collected data. Of the total 3625 visits, 75 visits (2.1%) were false positives due to error in EHR data.

Conditions Contributing to IHC

A positive Algorithm-IHC visit could also have a Contributing Condition (IHC+CC) (Table 2). Of the 3625 visits, the chart review identified 174 visits (4.8% of the 3625 visits) with a potential CC, but only 55 (1.5%) visits were also Algorithm-IHC positive (Table 2): 16 (0.44%) had an acute CC, 16 (0.44%) had an intentionally-low BP target due to a CC, and 24 (0.66%) had a treatment tradeoff due to treatment of the CC.

Table 2:

Acute and chronic conditions contributing to intensive hypertension care

Conditions contributing to IHC (CC)	CC identified in medical record review, n (% of total N=3625 visits)	Algorithm-positive IHC visits with contributing condition (IHC+CC), n (% of total N=3625 visits)
Acute cause of low BP	31 (.9%)	16 (.4%)
Intensified BP goal to lower than 140 mmHg to treat a comorbid condition	97 (2.7%)	16 (.4%)
Low BP caused by a medication necessary to treat another condition	52 (1.4%)	24 (.7%)
Any of the three CCs	174 (4.8%)	55 (1.5%)

There were 125 visits (Table 3) that were potentially overestimated due to false positive or IHC+CC (3.5% of the total visits). Of the comorbid conditions, only 37 patients with congestive heart failure had a total potential combined rate of either false positive or IHC+CC that exceeded 5% (5.2%). The next highest conditions with overestimated IHC were prostate disease (nearly 5%) and psychiatric disease (4.1%).

Table 3.

Top 7 chronic condition categories associated with false positive Algorithm-IHC and Contributing Conditions to IHC

Diagnoses	Total visits (row denominator)	Algorithm IHC Visits	Potentially overestimated IHC visits (%)
			False positives	Visit with with Contributing Condition to IHC	Total
All patients	3,625	915 (25.2%)	75 (2.1%)	55 (1.5%)	125 (3.5%)
DM	1,090	271 (24.9%)	14 (1.3%)	19 (1.7%)	33 (3.0%)
CAD/CVD/PAD	1,042	234 (24.7%)	12 (1.2%)	21 (2.0%)	31 (3.0%)
Arrhythmia	738	182 (24%)	8 (1.1%)	21 (2.9%)	27 (3.7%)
CHF/Valve	706	219 (31%)	14 (2.0%)	26 (2.7%)	37 (5.2%)
Chronic kidney disease	372	81 (21.8%)	4 (1.1%)	0 (0.0%)	4 (1.1%)
Psychiatric condition	341	97 (28.5%)	6 (1.8%)	8 (2.4%)	14 (4.1%)
BPH	261	57 (21.8%)	4 (1.5%)	9 (3.5%)	13 (5.0%)

Condition categories were assigned to patients based on any outpatient visits with International Classification of Diseases-9 during the study period. All visits for the patient were assigned to that condition. Visits could be in more than one row.

EHR (Electronic Health Record) data error: False positives occurring when using discrete BP and pharmacy fill data from the VA electronic health record, compared to the reference standard chart review

DM: Diabetes

CAD Coronary Artery Disease

CVD: Cerebrovascular Disease

PAD: Peripheral Arterial Disease

Arrhythmia: includes atrial fibrillation and flutter

CHF: Congestive Heart Failure includes low and preserved ejection fraction and non-specific type

Valve: any valvular disease

Psychiatric disease: includes depression and bipolar disease, post-traumatic stress disorder

BPH: benign prostatic hypertrophy

Other conditions tested but not displayed due to number of visits < 100 and IHC visits were also <10: eye problem, fall or gait impairment, failure to thrive, neurologic condition, orthostatic hypertension, venous/lower extremity edema.

Discussion

In this validation study using large national health care system data, we developed a method to capture Intensive Hypertension Care (IHC), defined as ≥3 BP medications with SBP <120 mmHg. Our calibrated algorithm for IHC had sensitivity of 92.2% and specificity of 97.2%, with an overall low total rate of false positives (2.1% of all visits).

We were also able to test whether certain comorbid conditions contributed to false positive IHC or situations where a non-hypertension condition contributed or mixed effects with hypertension treatment to result in clinical IHC. Patients with CHF were the most likely to have visits of either type of overestimation, thus future applications of IHC may need to stratify by patient condition or remove patients with CHF. We may have underestimated IHC+CC rate due to small sample size. The next highest condition, BPH may be specific to the VHA, due to terazosin’s wide use in the VHA as “double-duty” as a hypertension medication. The third ranked condition, psychiatric disease, may have been due to clonidine dual-use for post-traumatic stress disorder.

This study builds upon methods previously developed to use health-system data to measure the adequacy of clinical action, not only by measuring the percent of patients with BP below a certain threshold, but also capturing the intensity of BP medications prescribed.¹² By integrating BP values and medications, we gain a more comprehensive view of the care provided to patients, by a provider, or by a health system than by using BP alone. Using only BP as the target, e.g., BP < 140/90 mmHg (the current performance measure in the VA and by health insurances) is a far simpler computation. By capturing medication intensity, it is possible to capture more clinical nuance: from intense treatment (low BP and multiple medications), to mild undertreatment (moderate hypertension on treatment), to severe undertreatment (no medications despite high BP). This method paves the way beyond dichotomous BP targets for future research based on intensity of care.

BP combined with medication data yields greater understanding of systems-based management than BP alone. We have found previously that intensively-treated BP for patients with diabetes is rarely de-intensified, with little effect by advanced age or mortality risk.{Sussman} Research by others have captured IHT accurately by computerized administration record in more acute settings: intensifying hypertension regimens upon hospital discharge worsens clinical outcomes of older adults,¹⁴ and intensely treated patients in post-acute nursing homes with history of falls are less likely to fall again after deintensification.¹⁵ This study now broadens our understanding of administrative data accuracy for capturing IHC in the outpatient setting. This method will also be important in coming years after the SPRINT trial has redefined what we consider to be beneficial, yet high-intensity treatment. How we defined IHC (BP <120 mmHg on ≥3 BP medications) is consistent with the SPRINT intervention, in which the goal was to attain a target BP of <120 mmHg by increasing medications prescribed.² We note here that the period of time covered by these analyses preceded publication of the SPRINT results, during a time when SBP goal within VA was <140 mm Hg. These patients may have been treated intensively above and beyond the VA goals at the time, but in the post-SPRINT era, this degree of intensity may more commonly attained for many older patients.

The strength of this study was its national patient representation and its use of multiple sources of data. The algorithm was calibrated using only VA pharmacy data. Adding Medicare Part D data added to the complete picture, however was not critical to the accuracy of the algorithm. Therefore, our results would be applicable to health systems or insurance plan populations for whom most of the pharmacy data is obtainable. Second, our method calibrated the algorithm to capture IHC at any patient visit, therefore allowing flexibility for future outcomes studies to study the combination of visits needed to construct measures of short or long-term exposure to IHC. For example, one could study patients with year-over-year change in IHC (by comparing last visit IHC of each year), or select patients with consecutive visits of IHC to identify those with consistent long-term IHC.

There are several limitations that must be acknowledged. First, we had to ensure that our sample was intensively treated and had variability between visits in order to have an opportunity to test for the absence or presence of IHC. Therefore, we studied the Veterans with generally intensive BP care, possibly resulting in a false-positive rate that is higher than in the rest of the VA. Second, for a portion of our algorithm (medications that were discontinued), we needed to use the entire dataset for calibration, then validated back using the same sample, therefore potentially overestimating the algorithm accuracy. Third, we were unable to distinguish heart failure patients with reduced (HFrEF), which we expect would be more likely than preserved ejection fraction to contribute to finding IHC (IHC+CC), because treatment of HFrEF with BP-lowering medications BP below usual hypertension goals. Fourth, our algorithm could not capture “free samples” of BP medications dispensed by non-VA clinics and BP medications purchased at lower cost than VA co-payments (e.g., $5 generic programs). Last, this older Veteran population was predominantly male, therefore we could not test for gender differences in algorithm accuracy. An older female cohort might be treated less aggressively for comorbid conditions,¹⁶ potentially limiting our IHC+CC analysis among women.

This study’s focus on medication counts is the first step in validating a high-intensity BP regimen, but there is still room for further research regarding dosing intensity. Current guidelines for comorbid conditions now recommend medications, such as low-dose ACEIs for renal protection in diabetes, thus dosing intensity will be more important in the context of guideline-consistent care.

In conclusion, we have developed and validated an algorithm to identify patients with IHC in health systems. This method can be used to compare IHC between providers or systems or to improve appropriateness of patient care.

Key Points.

Question:

Can health system data be used to accurately detect patients receiving intensive hypertension care with multiple blood pressure medications?

Findings:

In this observational cohort study, we developed an algorithm based on clinic-measured blood pressure and pharmacy fills that can detect intensive hypertension care delivered at any visit. Using chart review of clinical notes as the reference standard for patients ≥age 65 years receiving ≥3 medications with systolic blood pressure of 120 mmHg lower, the algorithm was 92% sensitive and 97% specific.

Meaning:

This research provides a resource for health care systems to measure high-intensity care for quality comparison or as a research tool.