Association of Total Medication Burden with Intensive and Standard Blood Pressure Control and Clinical Outcomes

Abstract

Total medication burden (antihypertensive and non-antihypertensive medications) may be associated with poor systolic blood pressure (SBP) control. We investigated the association of baseline medication burden and clinical outcomes, and whether the effect of the SBP intervention varied according to baseline medication burden in the Systolic Blood Pressure Intervention Trial. Participants were randomized to intensive or standard SBP goal (below 120 or 140 mmHg, respectively); n=3,769 participants with high baseline medication burden (≥5 medications) and n=5,592 with low burden (<5 medications). Primary outcome: differences in SBP. Secondary outcomes: 8-item Morisky Medication Adherence Scale, and modified Treatment Satisfaction Questionnaire for Medications measured at baseline and 12 months; and incident cardiovascular disease (CVD) events and SAEs throughout the trial. Participants in the intensive group with high versus low medication burden were less likely to achieve their SBP goal at 12 months (Relative Risk [RR] 0.91, 95%CI 0.85-0.97), but not in the standard group (RR 0.98, 95%CI 0.93, 1.03; P_interaction<0.001). High medication burden was associated with increased CVD events (Hazard ratio [HR] 1.39, 95%CI 1.14-1.70) and SAEs (HR 1.34, 95%CI 1.24-1.45), but the effect of intensive versus standard treatment did not vary between medication burden groups (P_interaction>0.5). Medication burden had minimal association with adherence or satisfaction. High baseline medication burden was associated with worse intensive SBP control and higher rates of CVD events and SAEs. The relative benefits and risks of intensive SBP goals were similar regardless of medication burden.

Graphical Abstract

Summary

Given the prevalence of hypertension among US adults, strategies are needed to reduce inappropriate medication burden among individuals treated for hypertension. Intensive SBP control remains a beneficial strategy to reduce CVD events regardless of medication burden status.

INTRODUCTION

Despite widespread availability of safe, effective, and low-cost antihypertensive medications, blood pressure (BP) control rates remain unacceptably low.¹ High total daily medication burden, including antihypertensive and non-antihypertensive medications, may hinder control rates. Because achieving BP control often requires two or more antihypertensive medications, and hypertension often coincides with other comorbidities, patients with hypertension are likely to have high medication burden, often defined as taking five or more prescription medications.² Although high medication burden may be clinically indicated and appropriate, it is associated with adverse events, medication non-adherence, reduced functional status, and other negative clinical outcomes.^3–6

In the Systolic Blood Pressure Intervention Trial (SPRINT), targeting an intensive versus standard systolic BP (SBP) goal significantly reduced risk of cardiovascular disease (CVD) events and all-cause mortality without negatively impacting patient-reported adherence, treatment satisfaction, and serious adverse events (SAEs).^7,8 While the mean number of antihypertensive medications used per participant was 2.8 in the intensive group and 1.8 in the standard group, the association between total medication burden at baseline, including non-antihypertensive medications, and the likelihood of achieving and maintaining the assigned SBP goal within each treatment group is unknown.

We sought to determine the association between total prescription medication burden at baseline (including antihypertensive and non-antihypertensive medications) and SBP control, CVD events, SAEs, medication adherence, and treatment satisfaction. A secondary objective was to determine if total medication burden at baseline modified the effect of intensive vs standard treatment goals on CVD events and SAEs.

METHODS

Study Design and Participants

In accordance with NIH policy, data will be shared through the National Heart, Lung, and Blood Institute data repository in spring 2019 (http://www.biolincc.nhlbi.nih.gov/home). The rationale, design, and primary results of SPRINT have been previously published.^7,9 Briefly, SPRINT was a multicenter, randomized trial comparing two SBP treatment goals (an intensive goal of <120 mmHg versus a standard goal of <140 mmHg) in participants with hypertension at high CVD risk without diabetes or a history of stroke. Participants were age ≥50 years and had one or more high-CVD risk conditions (history of clinical or subclinical CVD other than stroke, estimated glomerular filtration rate [eGFR] of 20–59 mL/min/1.73 m² using the four-variable Modification of Diet in Renal Disease [MDRD] equation,¹⁰ 10-year risk for CVD ≥15% calculated using the Framingham Risk Score (FRS) for general clinical practice,¹¹ and/or age ≥75 years). Those with diabetes, previous stroke, heart failure, proteinuria ≥1 g/day, or an eGFR <20 mL/min/1.73 m² were excluded.

Trained SPRINT personnel ascertained participants’ baseline sociodemographic data, comorbidities,– and medications during the screening or randomization visit. We restricted the present analysis to participants with complete outcome and covariate data for each analysis (e.g., a participant with SBP data but missing adherence data at the 12-month visit would be included in the SBP analysis, but not the adherence analysis).

Each site’s institutional review board approved the main study protocol, and written informed consent was obtained from each participant. Several agencies funded the trial without involvement in the analysis or interpretation of the present post-hoc study (see Acknowledgements). The SPRINT Steering Committee and Publications subcommittee reviewed and approved this manuscript on behalf of the SPRINT Research Group.

Assessment of Baseline Medication Burden

Trained study personnel recorded data on participants’ complete medication profiles (e.g., name, total daily dose) at baseline using information obtained from the participant, family member, or guardian (e.g., pill-bottle review), or if available, current medication profiles in the electronic health record (see Protocol and Supplemental Methods for description of data cleaning.)¹² Medications were classified via route of administration, Food and Drug Administration-approved antihypertensive status, and over-the-counter status. Single-pill combinations were classified into individual components (e.g., “atorvastatin/amlodipine” was classified as two distinct components, “atorvastatin” and “amlodipine”). Each active ingredient in combination products were counted as individual medications because we could not reasonably determine if a combination product was used unless its brand name was recorded.

High medication burden was defined as five or more different prescription medications recorded at the baseline visit. There is no universally-accepted definition of “high” medication burden; however, five or more concomitantly-prescribed medications is frequently cited in the literature^2,13,14 (see Supplemental Methods for detailed discussion.) The primary analysis included antihypertensive and non-antihypertensive prescriptions of all formulations and routes of administration, but to avoid exposure misclassification, we excluded over-the-counter medications and herbal supplements as these agents are inconsistently documented in medication reviews.¹⁵

Outcomes

We evaluated the following outcomes: 1) SBP, 2) CVD events (a composite of myocardial infarction, acute coronary syndrome not resulting in myocardial infarction, stroke, acute decompensated heart failure, or death from cardiovascular causes^7,12), 3) all SAEs and SAEs of interest (i.e., hypotension, syncope, injurious falls, electrolyte abnormalities, or bradycardia⁷), 4) self-reported medication adherence scores measured by the 8-item Morisky Medication Adherence Scale (MMAS-8),^16–18 and 5) satisfaction with hypertension care and treatment using the modified Treatment Satisfaction Questionnaire for Medication (TSQM).¹⁹ SBP levels, medication adherence, and treatment satisfaction were evaluated at 12 and 48 months post-randomization. SBP levels were assessed in three ways: mean, mean change from baseline, and the proportion below the randomized treatment goal. Definitions, measurements, and adjudication of these outcomes have been previously described (see Supplemental Methods.)^7–9,20

Covariates

Covariates were selected based on the potential to serve as confounders of the association between baseline total medication burden and the primary and secondary outcomes in the analysis, guided by clinical knowledge and prior evidence.^5,21–23 Baseline covariates included age, sex, race or ethnic group, SBP, diastolic BP (DBP), eGFR, urine albumin-to-creatinine ratio, blood glucose, total cholesterol, high-density lipoprotein (HDL) cholesterol, statin use, aspirin use, smoking status, body mass index (BMI), presence of metabolic syndrome, FRS, number of comorbidities, atrial fibrillation/flutter, myocardial infarction, heart failure, peripheral vascular disease, MMAS-8, treatment satisfaction scores, and depression. Definitions and measurement of these covariates have been previously described.^7,9,20

Statistical Analysis

For the primary objective of determining the association between baseline medication burden and outcomes, analyses were performed within randomized treatment groups (i.e., intensive and standard). Modified Poisson regression with robust error variance was used to calculate risk ratios (RRs) for SBP control and to calculate adjusted percentages for medication adherence and treatment satisfaction associated with high or low baseline medication burden. Mean SBP and change in SBP from baseline to 12 months and 48 months was analyzed using an ordinary least squares regression model. Cox proportional hazards regression was used to calculate hazard ratios (HRs) for SAEs and CVD events associated with high or low baseline medication burden. The proportional hazards assumption was tested for violations by modeling the product of the medication burden group and the log of follow-up time as an interaction term. To determine the interaction between randomization treatment group and medication burden group, we calculated each outcome stratified by the randomized treatment group and included the product term (randomized treatment group×medication burden group) in each regression model in the full sample.

To determine if baseline medication burden modified the effect of intensive versus standard treatment, Cox proportional hazards regression was used to calculate HRs for CVD events and SAEs associated with intensive vs standard treatment within high or low baseline medication burden. The potential presence of an interaction between randomization treatment group and medication burden group was evaluated as described in the primary objective.

Several sensitivity analyses were planned for the present analysis a priori. To investigate whether the findings were sensitive to our medication burden definition, we repeated the analyses with various definitions: (1) including documented over-the-counter medications, (2) categorizing the number of baseline medications into ordinal categories and by quintiles, (3) excluding antihypertensive medications, and (4) restricting to only antihypertensives (by thresholds or tertiles). Because medication number is correlated with number of comorbidities, we repeated analyses stratified by the number of comorbidities at baseline, although number of comorbidities was included as a covariate in the primary analysis. Finally, two SAE sensitivity analyses were performed: first, we restricted the medication burden definition to non-antihypertensive medications only; and second, we restricted the outcome to only SAEs that the investigator deemed to be definitely or probably related to the treatment.

Adjusted percentages were calculated using STATA v.13.3 (StataCorp LLC, College Station, TX), and all other analyses were performed using SAS v.9.4 (SAS Institute Inc., Cary, NC).

RESULTS

Participant characteristics

Among 9,361 study-eligible participants, 5,592 (59.7%) had low medication burden (median 3, IQR 2,4) and 3,769 (40.3%) had high medication burden (median 6, IQR 5,8) at baseline (Figure 1; Table S1 and Figures S1–S3 in the online-only data supplement). Those with high medication burden were more likely to be older, have lower SBP and DBP, demonstrate worse kidney function, use statins, and have more comorbidities at baseline compared to those with low medication burden (Table 1). These patterns were similar in the intensive and standard treatment groups (all P_interaction>0.05).

Figure 1: CONSORT flow diagram.

CVD=cardiovascular disease; SAE=serious adverse event; SBP=systolic blood pressure

^aMissing outcome or covariate data, see methods

Table 1:

Baseline Characteristics by Treatment Group and Number of Baseline Medications

Characteristic	Intensive treatment			Standard treatment			P interaction *
	No. of baseline medications		P-value	No. of baseline medications		P-value
	< 5	≥ 5		< 5	≥ 5
Participants, n	2793	1885		2800	1883		0.93
Age, years	66.8±9.3	69.5±9.2	<.001	66.9±9.5	69.3±9.3	<.001	0.31
Female sex	998 (35.7)	686 (36.4)	0.64	977 (34.9)	671 (35.6)	0.60	0.97
Race or ethnic group			<.001			<.001	0.61
Non-Hispanic black	860 (30.8)	519 (27.5)		893 (31.9)	530 (28.1)
Hispanic	355 (12.7)	148 (7.9)		330 (11.8)	151 (8.0)
Non-Hispanic white	1513 (54.2)	1185 (62.9)		1519 (54.3)	1182 (62.8)
Other	65 (2.3)	33 (1.8)		58 (2.1)	20 (1.1)
Increased CVD risk criteria	2352 (84.2)	1708 (90.6)	<.001	2360 (84.3)	1706 (90.6)	<.001	0.96
Age ≥ 75 yr	656 (23.5)	661 (35.1)	<.001	699 (25.0)	620 (32.9)	<.001	0.06
Chronic kidney disease	606 (21.7)	723 (38.4)	<.001	601 (21.5)	715 (38.0)	<.001	0.98
Cardiovascular disease	394 (14.1)	546 (29.0)	<.001	395 (14.1)	542 (28.8)	<.001	0.93
Clinical	291 (10.4)	448 (23.8)	<.001	275 (9.8)	447 (23.7)	<.001	0.58
Subclinical	103 (3.7)	98 (5.2)	0.01	120 (4.3)	95 (5.0)	0.22	0.35
FRS ≥ 15%	2133 (76.9)	1426 (75.7)	0.37	2107 (75.8)	1438 (76.7)	0.49	0.26
Systolic blood pressure, mmHg	140.9±16.0	137.9±15.2	<.001	140.7±15.2	138.1±15.5	<.001	0.47
Diastolic blood pressure, mmHg	79.9±11.7	75.7±11.7	<.001	79.8±11.9	75.4±11.7	<.001	0.71
Serum creatinine, mg/dL	0.98 [0.84,1.15]	1.04 [0.88,1.28]	<.001	0.99 [0.84,1.16]	1.07 [0.89,1.29]	<.001	0.22
eGFR, mL/min/1.73m2	74.3 [62.3,86.3]	66.4 [52.9,81.2]	<.001	73.9 [62.1,86.6]	66.3 [52.5,80.4]	<.001	0.87
Urine albumin-to-creatinine ratio, g/mg	9.0 [5.5,18.2]	11.1 [6.1,28.3]	<.001	8.8 [5.4,18.7]	10.7 [5.9,27.3]	<.001	0.42
Blood glucose, mg/dL	97.0 [90.0,105.0]	98.0 [91.0,106.0]	0.003	97.0 [91.0,104.0]	98.0 [91.0,105.0]	0.05	0.74
Total cholesterol, mg/dL	193.0 [169.0,221.0]	176.0 [151.0,204.0]	<.001	194.0 [169.0,220.0]	177.0 [153.0,205.0]	<.001	0.98
HDL cholesterol, mg/dL	51.0 [43.0,61.0]	50.0 [42.0,60.0]	0.05	50.0 [43.0,61.0]	50.0 [42.0,60.0]	0.01	0.44
Smoking status			<.001			<.001	0.37
Never	1310 (46.9)	743 (39.4)		1320 (47.1)	749 (39.8)
Former	1058 (37.9)	917 (48.6)		1088 (38.9)	910 (48.3)
Current	414 (14.8)	225 (11.9)		380 (13.6)	221 (11.7)
Missing	11 (0.4)	0 (0.0)		12 (0.4)	3 (0.2)
Body mass index, kg/m2	29.6±5.6	30.4±6.0	<.001	29.5±5.6	30.3±5.9	<.001	0.84
Metabolic syndrome	1199 (42.9)	885 (46.9)	0.01	1214 (43.4)	916 (48.6)	<.001	0.55
Statin use	816 (29.2)	1167 (61.9)	<.001	873 (31.2)	1207 (64.1)	<.001	0.99
Aspirin use	1336 (47.8)	1150 (61.0)	<.001	1312 (46.9)	1116 (59.3)	<.001	0.70
Mean FRS, %	24.8±12.6	24.8±12.6	0.96	24.7±12.3	24.9±12.8	0.54	0.69
No. of comorbidities, median [IQR]	3.0 [2.0,5.0]	6.0 [4.0,8.0]	<.001	3.0 [2.0,5.0]	6.0 [4.0,8.0]	<.001	0.41
No. of medications, median [IQR]	3.0 [2.0,4.0]	6.0 [5.0,8.0]	<.001	3.0 [2.0,4.0]	6.0 [5.0,8.0]	<.001	1.00
Atrial Fibrillation/Flutter	150 (5.4)	240 (12.8)	<.001	155 (5.6)	209 (11.1)	<.001	0.23
Myocardial Infarction	120 (4.3)	218 (11.6)	<.001	127 (4.5)	201 (10.7)	<.001	0.38
Heart Failure	44 (1.6)	122 (6.5)	<.001	45 (1.6)	115 (6.1)	<.001	0.75
Peripheral Vascular Disease	107 (3.8)	143 (7.6)	<.001	101 (3.6)	152 (8.1)	<.001	0.49

CVD=cardiovascular disease; eGFR=estimated glomerular filtration rate; FRS=Framingham Risk Score; HDL=high-density lipoprotein; IQR=interquartile range; SD=standard deviation

All values are no.(%) or means±SD unless noted otherwise

^*P-value for treatment randomization×medication burden status

Association of baseline medication burden and SBP, CVD events, SAEs, medication adherence, and treatment satisfaction

In the intensive treatment group at 12 months post-randomization, participants with high versus low baseline medication burden had a clinically small but statistically higher mean SBP level (122.5±14.3 mmHg versus 120.6±13.1 mmHg, P<0.001), experienced a smaller change in SBP from baseline (−15.3±18.2 mmHg versus −20.1±18.6 mmHg, P<0.001), and were less likely to achieve the assigned SBP goal (51.8% versus 58.3%, RR 0.91, 95% CI 0.85-0.97; Table 2). However, in the standard treatment group at 12 months, no statistically significant differences were observed between the high and low medication burden groups on achieved SBP level (136.2±14.8 mmHg versus 136.2±12.8 mmHg, P=0.63), change in SBP from baseline (−2.3±19.2 mmHg versus −4.5±17.7 mmHg, P=0.63), or likelihood of achieving the assigned SBP goal (63.6% versus 64.8%, RR 0.98, 95% CI 0.93-1.03). The P_interaction between randomized treatment group and medication burden group were <0.001 for each of the three SBP metrics. Of 1,469 participants with available SBP data at 48 months post-randomization, the association between medication burden and SBP metrics was similar to that which was observed at 12 months for both randomization groups, although not statistically significant (Table S2). SBP differences between medication burden groups observed in the intensive treatment group were persistent throughout the trial (Figure 2, Panel A). Further, in the intensive treatment group, those with high medication burden were able to achieve lower SBP than those with high medication burden in the standard group throughout the trial. Differences in the likelihood of achieving the assigned SBP goal between the treatment groups were observed at 4-5 total baseline medications when medication burden was categorized by quintile (Figure 2, Panel B; Figure S4).

Table 2:

Blood Pressure and Clinical Outcomes by Treatment Group and Number of Baseline Medications

Outcomes	Intensive treatment			Standard treatment			P interaction *
	No. of baseline medications		P-value or RR/HR (95% CI)	No. of baseline medications		P-value or RR/HR (95% CI)
	< 5	≥ 5		< 5	≥ 5
Participants with SBP value at 12 months	(N=2,368)	(N=1,611)		(N=2,353)	(N=1,575)
SBP, mmHg	120.6±13.1	122.5±14.3	<0.001	136.2±12.8	136.2±14.8	0.63	<0.001
SBP change, mmHg	−20.1±18.6	−15.3±18.2	<0.001	−4.5±17.7	−2.3±19.2	0.63	<0.001
Below randomization goal	1381 (58.3)	835 (51.8)	0.91 (0.85-0.97)	1524 (64.8)	1001 (63.6)	0.98 (0.93-1.03)	<0.001
All Participants	(N=2,630)	(N=1,781)		(N=2,609)	(N=1,770)
CVD Events†	98 (3.8)	136 (7.7)	1.32 (0.98-1.78)	138 (5.4)	170 (9.7)	1.47 (1.13-1.92)	0.53
Any SAE‡	831 (31.9)	872 (49.3)	1.32 (1.18-1.47)	786 (30.5)	853 (48.4)	1.35 (1.21-1.51)	0.59

All values are no. (%) or means ± SD unless noted otherwise

CI=confidence interval, CVD=cardiovascular disease, HR=hazard ratio, RR=risk ratio, SAE=serious adverse event, SBP=systolic blood pressure, SD=standard deviation

^*P-value for treatment randomization×medication burden status

^†Composite of myocardial infarction, acute coronary syndrome, stroke, heart failure, or death from cardiovascular causes

^‡Defined per main SPRINT paper^7,12

Figure 2: SBP and risk ratios for achieving SBP control by medication burden and treatment group.

Panel A: Mean SBP

Panel B: Adjusted risk ratios and 95% confidence intervals for achieving SBP goal at 12 months by quintile of medication burden

Participants with high versus low baseline medication burden were more likely to experience a composite CVD event outcome (HR 1.39, 95% CI 1.14-1.70; Figure 3, Panel A; Table 2/Table S3 for stratified analysis) and any SAE (HR 1.34, 95% CI 1.24-1.45; Table 2/Table S4 for stratified anlysis). The incidence of CVD events was progressively higher at higher levels of baseline medication burden (Figure 3, Panel B; Figure S4). These associations were similar regardless of treatment group (all P_interaction>0.5). When stratified by treatment group, the association between high medication burden and increased risk of experiencing a CVD event (intensive: HR 1.32 [95% CI 0.98-1.78], standard: HR 1.47 [95% CI 1.13-1.92], P_interaction=0.53) and SAE (intensive: HR 1.32 [95%CI 1.18-1.47], standard: HR 1.35 [95%CI 1.21-1.51]; P_interaction=0.59) remained. These findings were not driven by increased associations with individual components of each composite measure.

Figure 3: Cumulative hazard plot and hazard ratios for CVD events outcome by medication burden and treatment group.

Panel A: Cumulative hazards for CVD event

Panel B: Adjusted risk ratios and 95% confidence intervals for experiencing CVD events by quintile of medication burden

At 12 months post-randomization, more participants in the standard treatment group with high medication burden reported “high” medication adherence (score of 8 on MMAS-8) compared to those with low medication burden (P<0.001); this association was not observed in the intensive treatment group (P_interaction<0.001; Table S5). In both treatment groups, no significant differences were observed between high and low medication burden groups on MMAS-8 scores at 48 months or on TSQM responses at 12 and 48 months post-randomization (all P_interaction>0.05, Tables S6–S8).

Effect modification of intensive SBP treatment by baseline medication burden

The beneficial effects of intensive versus standard SBP treatment on CVD events (i.e., reduced with intensive treatment group) and SAEs (i.e., no difference between treatment groups) were similar among those with and without high baseline medication burden (Table S9; all P_interaction>0.5).

Sensitivity analyses

A detailed description of the results of all sensitivity analyses is provided in the Supplementary Results. Overall, the results of all sensitivity analyses were qualitatively similar to the main analysis (Figures S5–S7 and Tables S10–S16).

DISCUSSION

In this post-hoc analysis of SPRINT, high medication burden at baseline was significantly associated with a lower likelihood of achieving the intensive, but not the standard, SBP target at 12-months of follow-up. Within the intensive treatment group, participants with high versus low prescription medication burden had an approximately 10% lower likelihood of achieving an SBP of <120 mmHg at 12 months; high burden participants also had a higher mean SBP and a smaller change in SBP from baseline. In both treatment groups, high medication burden was associated with increased likelihood of experiencing a CVD event or an SAE. However, SAE rates were similar between treatment groups, and baseline medication burden did not impact the beneficial reduction in CVD events and mortality observed in the intensive treatment group. There were minimal differences in patient-reported adherence and treatment satisfaction between participants with high and low medication burden in both groups.

In the context of an aging population with increasing multi-morbidity, high medication burden prevalence is expected to rise. Up to 50% of patients aged 65 or older take five or more medications.^24–28 Data from the National Health and Nutrition Examination Survey (NHANES) demonstrate an increasing prevalence of high medication burden over time in the entire population, from 8.2% in 1999-2000 to 15% in 2011-2012.²⁶ Consistent with the current analysis, high medication burden has been associated with increased risks of experiencing an adverse drug event, drug-drug interactions, impaired balance and falls, frailty and disability, cognitive decline, hip fracture, delirium, hospital admission, and mortality.^5,6,29–38 As comorbidity burden increases, clinicians are challenged with choosing medications that are most appropriate. While our analysis included adjustment for number of comorbidities, specific comorbidities, and age, we could not distinguish between inappropriate and appropriate medications in this study, and this remains an area that requires more research.

Within the intensive treatment group, we observed a statistically significant difference in 12-month SBP of 1.9 mmHg between the between the high and low medication burden groups. Although small, differences of such magnitude have the potential to have vast population health impacts. For example, data extrapolated from the INTERSALT study suggest that by lowering population-wide average SBP by 2.2 mmHg, risk for coronary death and stroke death would decrease by 4% and 6%, respectively.³⁹ Moreover, pooled projections from clinical trials estimate that for every 5 mmHg reduction in SBP, risk for experiencing the composite cardiovascular endpoint decreases by 13%.⁴⁰

Because there is no universal definition of high medication burden, we conducted several sensitivity analyses. Qualitatively, results were similar when including over-the-counter medications in the definition of medication burden and stratifying by the number of comorbidities at baseline, but the association with SBP control was not statistically significant when excluding antihypertensive medications from the definition. Furthermore, when categorizing medication burden into ordinal categories, ≥4 and ≥6 total medications was the threshold at which SBP control begins to diminish in the intensive and standard treatment groups, respectively (Figure 2, Panel B and Figure S4).

Similar to previous analyses on patient-reported outcomes in SPRINT, medication burden was not associated with clinically significant differences in medication adherence or treatment satisfaction at 12 or 48 months.⁸ This could be due to the measurement of these outcomes at 12 and 48 months, which largely ignore the effects of therapy intensification that occurred in the early phases of the trial (i.e., visits at 3, 6, and 9 months). Additionally, several patient- and provider-specific factors influence medication selection, adherence and treatment satisfaction, which complicate the achievement of clinical outcomes. These factors provide a complex clinical picture for managing hypertension and include cultural preferences, health literacy level, cost or financial burdens, medication allergies/interolerances or previous experiences with antihypertensives, the patient-provider fiducial relationship, potential for SAEs, and physician medication preferences, among others.^41–43 These factors could not be accounted for in our analyses; further research is needed on how to optimize these factors to improve hypertension treatment and outcomes.

There are several study limitations worth noting. Although our results were consistent across sensitivity analyses and we included an extensive list of covariates in our adjusted models, there is an inherent risk of confounding as participants with higher baseline medication burden are also more likely to be the highest risk participants with more comorbidities. We did not adjust our results for multiple comparisons, increasing the type I error risk. Low participant numbers at long-term follow-up (i.e., 48 and 60 months post-randomization) limited our ability to determine long-term associations with medication burden because the trial was stopped early (median follow-up of 3.2 years) due to overwhelming benefit of intensive treatment. For this reason, we did not re-classify participants’ medication burden status beyond the baseline visit, although this may be of interest for future analyses. There is potential for misclassification of baseline medication burden status, particularly due to inconsistent collection or patient reporting of over-the-counter products, herbal supplements, and non-oral medications. Medication data collection within clinical research is time-consuming and arduous, and standardization solutions are needed.^44,45 We were not able to detect and properly account for single-pill combination medications, which may have confounded the adherence analyses and led to the observed null association; thus, our analysis should be interpreted as investigating associations with medication burden rather than those of daily pill burden. Furthermore, we could not distinguish medications intended for short-term use (e.g., antibiotics) or account for the use of medications which may increase BP (e.g., nonsteroidal anti-inflammatory drugs), which may have affected our results. Finally, differences in medication burden across race/ethnicity, sex, and comorbidity strata were not performed here due to small sample sizes but should be explored in future analyses.

PERSPECTIVES

In SPRINT, high medication burden was common and associated with a lower probability of achieving intensive SBP control. Medication burden was not associated with patient-reported medication adherence or treatment satisfaction. Importantly, the beneficial effects of intensive SBP control for reducing CVD events observed in the primary analysis persist irrespective of medication burden without a differential association with SAEs. Nonetheless, high medication burden was associated with increased risks of CVD events and SAEs irrespective of treatment group. Whether this reflects medication polytoxicities, patient behavior, drug-drug interactions or an inability to statistically control for medication use as a marker of sicker patients with multiple comorbidities, or a combination of these, is unresolved and deserving of further study.

Novelty and Significance:

What Is New?

• High medication burden (five or more prescription medications) was common and associated with increased risk of CVD events and SAEs in both treatment groups of a randomized clinical trial.

• High medication burden was significantly associated with a higher SBP at 12 months and reduced likelihood of achieving an intensive SBP goal (<120 mmHg).

• The beneficial effects of intensive SBP control (i.e., reduced CVD events) persisted irrespective of medication burden status.

What Is Relevant?

• Adults with hypertension often take non-antihypertensive and antihypertensive medications to control comorbidities, thereby increasing risks associated with high medication burden (i.e., reduced medication adherence and increased serious adverse events).