Role of Healthcare Professionals in the Success of Blood Pressure Control Interventions in Patients with Hypertension: a Meta-Analysis

Abstract

Background:

Globally, only 13.8% of hypertensive patients have their blood pressure (BP) controlled. Trials testing interventions to overcome barriers to BP control have produced mixed results. Type of healthcare professional delivering the intervention may play an important role in intervention success. The goal of this meta-analysis is to determine which healthcare professionals are most effective at delivering BP reduction interventions.

Methods:

We searched Medline and Embase (to December 2023) for randomized controlled trials of interventions targeting barriers to hypertension control reporting who led intervention delivery. One hundred articles worldwide with 116 comparisons and 90,474 hypertensive participants were included. Trials were grouped by healthcare professional, and the effects of the intervention on systolic and diastolic BP were combined using random effects models and generalized estimating equations.

Results:

Pharmacist-, community health worker (CHW)-, and health educator-led interventions resulted in the greatest systolic BP reductions of −7.3 mmHg (95% CI: −9.1, −5.6), −7.1 mmHg (−10.8, −3.4), and −5.2 mmHg (−7.8, −2.6), respectively. Interventions led by multiple healthcare professionals, nurses, and physicians also resulted in significant systolic BP reductions of −4.2 mmHg (−6.1, −2.4), −3.0 mmHg (−4.2, −1.9), and −2.4 mmHg (−3.4, −1.5), respectively. Similarly, the greatest diastolic BP reductions were −3.9 mmHg (−5.2, −2.5) for pharmacist-led and −3.7 mmHg (−6.6, −0.8) for CHW-led interventions. In pairwise comparisons, pharmacist were significantly more effective than multiple healthcare professionals, nurses, and physicians at inte.

Conclusions:

Pharmacists and CHWs are most effective at leading BP intervention implementation and should be prioritized in future hypertension control efforts.

INTRODUCTION

Hypertension is the leading modifiable risk factor for cardiovascular disease and all-cause mortality and affects 31.1% of adults, worldwide.^1–3 Despite proven effective pharmaceutical and lifestyle interventions, in 2010 only 13.8% of people with hypertension had their blood pressure (BP) controlled, globally.³ Barriers to effective BP control exist at the patient, provider, healthcare system, and community levels.^4–6 Randomized controlled trials testing strategies to overcome these barriers have produced heterogeneous results.⁷ Better understanding characteristics of successful implementation strategies is critical for future interventions and, ultimately, improved BP control.

One common barrier to effective BP control is lack of available time for primary care physicians to adequately conduct medication titration and coaching for lifestyle modification due to demanding clinic schedules and competing demands of co-morbidity management during brief clinic visits.^5,8 Implementation strategies, such as team-based care and task shifting, address this barrier by assigning responsibility for health coaching and/or medication titration to other professionals, including pharmacists, nurses, community health workers (CHWs), and health educators.^7,9,10 Prior trials and meta-analyses have demonstrated that interventions led by non-physicians can be effective for BP control, but no direct comparison between intervention leaders has been conducted.^11–14 The objective of the current analysis is to determine the comparative effectiveness of interventions delivered by different healthcare professionals for BP reduction among patients with hypertension in order to optimize the effectiveness of future interventions for BP control.

METHODS

Study Selection

Investigators searched MEDLINE and Embase from inception to December 20, 2023 using search terms for BP or hypertension and an extensive list of terms to identify studies testing implementation strategies to overcome barriers to hypertension control, including task shifting and task sharing, patient health coaching, and healthcare provider guideline adherence training. Searches were restricted to clinical trials in adult humans, and no articles were excluded due to publication language. Details of the comprehensive literature search strategy have been published previously,⁷ and search strategies are available in Tables S1 and S2. Additional articles were identified from manual searching of references from review articles, meta-analyses, and original publications.

An article was eligible for inclusion if it: 1) was a randomized-controlled trial; 2) was conducted in adults with hypertension defined as use of antihypertensive medications or guideline-defined BP criteria (minimum ≥ 130/80 mm Hg); 3) the net change in systolic or diastolic BP was a main trial outcome; 4) BP was ascertained using standardized BP measurement techniques; 5) the intervention targets barriers to hypertension control in real world settings (healthcare delivery, worksite, communities, and pharmacies) and included direct patient interaction; 6) trial duration was at least 6 months; 7) the control group received usual care or minimal education; 8) variance of BP changes (or data to calculate it) was reported; 9) clustering was accounted for in the analysis if the trial was cluster-randomized; and, 10) qualifications of the individuals delivering the intervention were reported. Two investigators independently reviewed all abstracts for inclusion and further evaluated full texts to select a final set of included articles. All disagreements regarding eligibility were settled by consensus.

Data Extraction

Investigators independently extracted data in duplicate from included trials using a standardized data collection form. Authors were contacted for relevant data if not reported in published articles. Baseline participant characteristics, trial components, intervention details, BP measurement methods, and BP study outcomes were extracted. Results from duplicate extraction were compared and discrepancies were addressed through consensus. If trials reported BP changes at more than one time point, data collected closest to the end of the intervention period was used.

The Cochrane risk-of-bias tool for randomized trials (RoB 2) and the extension for cluster-randomized trials were used to assess the risk of bias in included trials.¹⁵ Data on trial methods and results of included papers were reviewed to assess risk of bias in randomization, intervention assignment, missing outcome data, outcome measurement, and reported results.

Trials were divided into categories based on the type of healthcare professional who led intervention delivery: pharmacists, nurses, CHWs, physicians, health educators, and multiple healthcare professionals. Health educators include dieticians and trained study staff who conducted health coaching. The multiple healthcare professional team category includes interventions that were delivered by a group or that had more than one member delivering different intervention components with no clear intervention leader.

Trials were eligible for inclusion if they had control groups receiving usual care or minimal education. Usual care is defined as hypertension management by usual care providers with no trial intervention, and minimal education is defined as brief educational sessions or materials provided to patients and/or providers.

Five types of implementation strategies were used in the eligible trials: health coaching, team-based care with non-physician titration, team-based care with physician titration, physician training for improved care delivery, and multilevel strategies without team-based care.^7,16,17 Health coaching interventions are characterized by multiple sessions of health education and motivation for lifestyle modification and/or medication adherence.^7,18 Team-based care is defined as collaborative care provision by at least two team members working closely with the patient to accomplish hypertension treatment goals.^7,19 Team-based care interventions were further divided into those with medication titration performed by a physician and those with titration done by another team member. Physician training consists of strategies to improve patient-provider interaction and/or quality of care through education, audit and feedback, and/or treatment decision support.^7,20 Lastly, multilevel strategies without team-based care are strategies in which barriers to hypertension control are targeted at two or more of the patient, provider, and system levels but without a team-based care component.

Data Synthesis and Statistical Analysis

Mean net change in systolic and diastolic BP and associated variance were calculated from available data for each comparison reported in included trials. The mean net change is defined as the difference in differences between intervention and control groups of follow-up minus baseline BP. The difference between follow-up and baseline BP and related variance was also extracted and calculated within each treatment arm for use in adjusted models to compare treatment effects among healthcare professional categories.

For healthcare professional categories where all treatment arms have independent study populations, inverse variance-weighted random effects models with the Sidik-Jonkman residual heterogeneity estimator and Knapp-Hartung small sample adjustment were used to calculate pooled mean differences in systolic and diastolic BP within each healthcare professional category.^21–23 In some categories, two or more intervention groups were compared to the same control group, and robust variance estimation was used in the analysis to address non-independence of estimates.²⁴ The Q test was used to evaluate heterogeneity. To conduct a thorough assessment of publication bias, both the Begg’s rank correlation test and the Egger’s weighted linear regression test were used.²⁵ Subgroup analyses of pharmacist-led interventions were conducted to compare community- and retail-based interventions to clinic-based interventions.

Generalized estimating equations were used to quantify BP reduction in each healthcare professional category after adjustment for important baseline covariates and to conduct pairwise direct comparisons between healthcare professional categories.⁷ Analyses were adjusted for trial-level baseline percent male, mean age, mean systolic BP, and trial duration, and if the control group was usual care or minimal education. Estimates were additionally adjusted for type of intervention in a sensitivity analysis to determine the effect of intervention leader independent of the type of intervention. Changes in mean BP and associated standard errors in each randomized comparison arm were used in these analyses to compare intervention leader groups to a common control group. These models included estimates from each trial arm of all included trials (N=207 observations for systolic BP). An exchangeable correlation matrix was used for estimates within a trial, and trials were treated as clusters to maintain randomized comparisons. Models used weights exported from a random effects meta-analysis including all BP changes and standard errors from all treatment arms to account for within and between trial variance. Additional information on analysis methods have been published previously.⁷

Analyses were conducted and forest plots created using the metafor (version 4.4–0), robumeta (version 2.1), and forestplot (version 3.1.3) packages in R 4.0.2 (R Project for Statistical Computing) and PROC GENMOD in SAS 9.4 (SAS Institute). This meta-analysis is reported according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) 2020 statement (PRISMA 2020 Checklist available in Table S3).²⁶ Search and analysis methods used here have been previously published,⁷ but the protocol for this meta-analysis was not registered. The protocol, data, and other materials are available from the authors upon request. All data were previously published, aggregated, and de-identified, so informed consent and IRB approval were not required.

RESULTS

A total of 8,547 citations were identified from our searches after duplicates were excluded, and 1,270 articles were included for full-text review (Figure 1). One hundred articles reporting 116 comparisons and including 90,474 participants met our eligibility criteria and were included in analyses. Details of individual included studies are available in Table S4. Sample size per intervention arm ranged from 12 to 17,407 participants (median: 132), and the median study arm-specific mean age was 60 years (range: 33–71.2). Median study arm mean baseline systolic and diastolic BP are 147 mm Hg (range: 124–174) and 87 mm Hg (range: 70–97), respectively. Median trial duration was 9 months (range: 6 months - 2 years), 36% (N=36) of trials were cluster-randomized, and 63% (N=63) had usual care control groups. Forty-three percent (N=50) of included comparisons tested health coaching strategies, 13% (N=15) tested team-based care with non-physician titration, 21% (N=24) tested team-based care with physician titration, 12% (N=14) tested physician training and support, and 11% (N=13) tested multilevel strategies without team-based care. The number of comparisons per intervention leader category ranged from 10 for CHWs to 30 for pharmacists and from 2,798 study participants in educator-led interventions to 38,943 in CHW-led interventions (Table 1).

Figure 1. Flowchart of Study Selection.

BP – blood pressure

Table 1.

Characteristics of Included Trials by Healthcare Professional Delivering the Intervention

Healthcare professionals	Number of comparisons	Number of participants	Baseline SBP Range, mmHg	Baseline DBP Range, mmHg	Mean Age Range, years	Range of Men, %	Range of Trial Duration	Study Design, %	Strategy Type, %	Control Categories, %
Pharmacists	30	7,476	130–174	76–97	39–68	22–100	6–12 months	Parallel: 100%, Cluster-Randomized: 28%	Health Coaching: 20%; TBC with Non-Physician Titration: 30%; TBC with Physician Titration: 37%; Multilevel without TBC: 13%	Usual Care: 67%, Minimal Education: 33%
Nurses	24	9,453	123–162	70–91	52–71	0–99	6–24 months	Parallel: 100%, Cluster-Randomized: 25%	Health Coaching: 54%; TBC with Non-Physician Titration: 12%; TBC with Physician Titration: 29%; Multilevel without TBC: 4%	Usual Care: 83%, Minimal Education: 17%
Community Health Workers	10	38,943	128–154	76–94	50–66	23–47	6–24 months	Parallel: 100%, Cluster-Randomized: 70%	Health Coaching: 60%; TBC with Non-Physician Titration: 20%; TBC with Physician Titration: 10%; Multilevel without TBC: 10%	Usual Care: 40%, Minimal Education: 60%
Health Educators	11	2,798	136–159	75–97	33–66	12–99	6–18 months	Parallel: 91%, Factorial: 9%, Cluster-Randomized: 9%	Health Coaching: 82%; TBC with Physician Titration: 9%; Multilevel without TBC: 9%	Usual Care: 36%, Minimal Education: 64%
Multiple Healthcare Professionals	17	7,693	132–167	73–96	50–71	19–96	6–24 months	Parallel: 100%, Cluster-Randomized: 44%	Health Coaching: 53%; TBC with Non-Physician Titration: 6%; TBC with Physician Titration: 18%; Multilevel without TBC: 24%	Usual Care: 71%, Minimal Education: 29%
Physicians	24	24,111	127–162	74–95	54–69	20–97	6–24 months	Parallel: 100%, Cluster-Randomized: 75%	Health Coaching: 29%; TBC with Physician Titration: 4%; Physician Training to Improve Care Delivery: 58%; Multilevel without TBC: 8%	Usual Care: 71%, Minimal Education: 29%

SBP - Systolic Blood Pressure; DBP - Diastolic Blood Pressure; TBC - Team-based Care

The risk of bias assessment for individually and cluster randomized trials are presented in Figure S1 and Figure S2, respectively. Overall, included trials were of high quality. Due to the nature of the interventions and delivery in real-world settings, many studies were not able to mask those delivering the intervention or the study participants. Additionally, some cluster trials did not enroll all study participants before randomization. Further, many older trials did not have pre-specified statistical analysis plans.

In unadjusted analyses, pharmacist- and CHW-led interventions resulted in the greatest net change in systolic BP of −7.6 mm Hg (95% Confidence Interval, CI: −9.5, −5.6) and −8.0 mm Hg (95% CI: −12.4, −3.5), respectively. (Figure 2). Health educators had the next largest change in systolic BP of −4.6 mm Hg (95% CI: −7.1, −2.1) followed by multiple healthcare professional teams that had a change in systolic BP of −4.1 mm Hg (95% CI: −6.2, −2.0). Nurse- and physician-led interventions had the smallest net changes in systolic BP of −3.0 mm Hg (95% CI: −4.7, −1.3) and −2.3 mm Hg (95% CI: −3.5, −1.2), respectively.

Figure 2. Mean Net Systolic Blood Pressure Reduction and 95% Confidence Intervals by Healthcare Professional.

(A) Pharmacist, (B) Community Health Worker, (C) Health Educator, (D) Multiple Healthcare Professionals, (E) Nurse, and (F) Physician

Similarly, pharmacist and CHW-led interventions had the greatest net changes in diastolic BP compared to other intervention leaders [−3.8 mm Hg (95% CI: −5.1, −2.5) and −3.1 mm Hg (95% CI: −5.1, −1.0), respectively] (Figure 3). Health Educator-, nurse-, physician- and multiple healthcare professional-led interventions resulted in significant, modest changes in net diastolic BP of −1.9 mm Hg (95% CI: −3.5, −0.4), −1.6 mm Hg (−3.0, −0.3), −1.2 mm Hg (95% CI: −1.9, −0.5), and −1.4 mm Hg (95% CI: −2.4, −0.3), respectively.

Figure 3. Mean Net Diastolic Blood Pressure Reduction and 95% Confidence Intervals by Healthcare Professional.

(A) Pharmacist, (B) Community Health Worker, (C) Health Educator, (D) Multiple Healthcare Professionals, (E) Nurse, and (F) Physician

Pharmacist-led interventions in clinic and retail settings (n=6 comparisons) resulted in net systolic and diastolic BP reductions of −9.8 mm Hg (95% CI: −15.9, −3.7) and −4.9 mm Hg (95% CI: −9.0, −0.8), while those in clinic and hospital settings (n=24 comparisons) resulted in reductions of −7.0 mm Hg (95% CI: −9.0, −5.1) and −3.5 mm Hg (95% CI: −4.8, −2.2) in systolic and diastolic BP. Differences between subgroups were not statistically significant (p = 0.3 for systolic and p = 0.5 for diastolic).

All meta-analyses had some evidence of heterogeneity with a Q test p-value range of <0.0001 to 0.01. No tests for publication bias were statistically significant [Begg P range: 0.08–0.86, Egger P range: 0.06–0.90].

After multivariable adjustment for trial-level baseline covariates and all other healthcare professional categories, pharmacist- and CHW-led interventions remain the most effective for systolic BP reduction [−7.3 mm Hg (95% CI: −9.1, −5.6) and −7.1 mm Hg (95% CI: −10.8, −3.4, respectively] (Figure 4). Health educator- and multiple healthcare professional-led interventions are also associated with substantial reductions in systolic BP of −5.2 mm Hg (95% CI: −7.8, −2.6) and −4.2 mm Hg (95% CI: −6.1, −2.4), respectively. Nurse- [−3.0 mm Hg (95% CI: −4.2, −1.9)] and physician-led [−2.4 mm Hg (95% CI: −3.4, −1.5)] interventions are associated with more modest but significant reductions in BP. Similarly, pharmacist- and CHW-led interventions resulted in the greatest reductions in diastolic BP of −3.9 mm Hg (95% CI: −5.2, −2.5) and −3.7 (95% CI: −6.6, −0.8), respectively. Health educator-, multiple healthcare professional-, nurse-, and physician-led interventions also resulted in significant reductions in diastolic BP. After further adjustment for the type of intervention delivered by healthcare professionals, BP reduction associated with pharmacist-led interventions was slightly attenuated. In contrast, BP reductions associated with other healthcare professional-led interventions were slightly greater (Figure S3).

Figure 4. Adjusted Mean Net Systolic (A) and Diastolic (B) Blood Pressure Reduction Associated with Healthcare Professional Delivering the Intervention.

BP – blood pressure

Adjusted for sex, age, baseline BP, trial duration, type of control group, and other healthcare professional categories. Boxes are weighted by sample size.

Multivariable-adjusted pairwise comparisons between intervention leader categories demonstrate that pharmacist-led interventions result in significantly greater systolic BP reductions of −3.1, −4.3, and −4.9 mm Hg compared to multiple healthcare professional-, nurse-, and physician-led interventions, respectively (Table S5). CHWs and health educators also have significantly greater reductions in systolic BP compared to physicians. Pharmacist-led interventions also resulted in significantly greater reductions in diastolic BP compared to interventions led by multiple healthcare professionals, nurses, and physicians of −2.5, −2.1, and −2.6 mm Hg, respectively.

DISCUSSION

Significant BP reductions can be achieved with interventions targeting barriers to BP control regardless of the type of healthcare professional leading the intervention. However, interventions lead by pharmacists and CHWs resulted in the largest reductions in BP. Further, nurse- and physician-led interventions resulted in comparatively modest BP reductions compared to other healthcare professionals. This is the first meta-analysis to directly compare the effectiveness of interventions delivered by different types of professionals for BP reduction. These findings have important public health implications. There is a critical need for effective interventions to improve BP control in real-world clinic and community settings. Optimizing intervention delivery, including who is responsible for leading intervention delivery, could result in more effective intervention implementation.

Pharmacist-led interventions resulted in the largest reduction in systolic and diastolic BP in both adjusted and unadjusted analyses. Pairwise comparisons also showed that pharmacist-led interventions result in statistically greater reductions in BP compared to those led by nurses, physicians, and multiple leaders. Similar to our results, a previous meta-analysis from 2014 found a net systolic BP reduction of −7.6 mm Hg and a net diastolic BP reduction of −3.9 mm Hg.¹⁴ In addition, a meta-analysis of community pharmacy medication review programs found a net reduction of −6.8 mm Hg in systolic BP and −2.1 mm Hg in diastolic BP.²⁷ Our prior work has shown that team-based care approaches in which someone other than the physician can titrate medications result in larger reductions in BP than any other approaches.⁷ Given the ability of pharmacists to titrate medications, this approach is common in pharmacist-led interventions, which may partially explain the large net BP reduction in these trials.^28–35 After adjusting for type of intervention, the effect of pharmacist-led interventions on BP was attenuated but still had a net reduction in BP of over 6 mm Hg, suggesting pharmacists are effective at delivering interventions even without medication titration. In many interventions, pharmacists operate as part of team-based care to supplement physicians for health coaching and medication support.³⁶ Because medication adherence and side effects are barriers to taking medications as prescribed at the patient level, pharmacist medication expertise makes them uniquely suited to counsel patients in these areas.³⁷ A prior meta-analysis found that task sharing with pharmacists resulted in greater BP reduction (−8.12 mm Hg) compared to task sharing with other health care professionals, including dietitians, CHWs, and nurses.¹¹ Although the difference between groups was not statistically significant, our results suggest that community/retail settings for pharmacist-led interventions have greater BP reduction than those delivered in clinic/hospital settings. Further work is needed to explore the impact of setting on the success of pharmacist-led interventions.

CHW-led interventions resulted in the second highest systolic BP reduction in adjusted analyses and the highest reduction after further adjustment for type of intervention. They also resulted in statistically greater reductions in systolic BP compared to physician-led interventions in pairwise comparisons. CHWs have extensive cultural understanding combined with some job-related training who are primarily tasked with delivering culturally-appropriate health-related interventions in a given community.³⁸ They play an important role in intervention delivery by overcoming cultural barriers and serving as a bridge between patients and the healthcare system.³⁹ Further, they are a cost-efficient alternative to interventions delivered by pharmacists, physicians, and other highly specialized occupations and could lead to better intervention sustainability, because they are embedded in the communities they serve.^39,40

Nurse- and physician-led interventions resulted in the smallest net reductions in BP. These smaller reductions compared to other categories of intervention delivery leaders is likely due to system-level barriers of lack of reimbursement for health counseling and insufficient time for health coaching in clinic visits that are not overcome by adding additional intervention-related responsibilities to physicians and nurses.⁴¹ Clinical inertia is also a common physician-level barrier to BP control stemming from uncertainty of patients’ true BP and medication adherence, overestimation of care, lack of training/understanding of clinical guidelines, and competing demands.⁴² The comparatively small improvements in BP for interventions led by physicians compared to other healthcare professionals suggests that intervention components are needed to address this barrier. Team-based care approaches, where other healthcare professionals absorb some of the responsibilities of hypertension management, can be effective at providing support for antihypertensive medications and health coaching outside of the normal clinic work flow.¹²

Our study has several strengths. It is a large meta-analysis of 100 articles including 90,474 hypertension patients with 116 comparisons of randomized trials of interventions to overcome barriers to BP control in patients with hypertension. All studies report usual care or minimal education control groups so that head-to-head comparisons can be made between intervention delivery leaders. Our study also has several limitations. Only 23% of identified studies are from low- and middle-income countries, so it is not clear if the findings observed here are globally relevant. Despite the large number of studies identified, there are modest sample sizes in some intervention delivery categories, such as only 10 studies with CHW-led interventions and 11 with educator-led interventions. Further, significant heterogeneity was observed for all comparison, although that is expected given the diverse nature of the intervention components and contexts of included studies. In addition, because our analyses rely on published reports, many subgroup analyses, such as comparing the intervention effect by hypertension severity, are not possible.

In conclusion, pharmacists and CHWs are the most effective healthcare professionals for implementation of interventions to improve BP control. Interventions led by health educators, multiple healthcare professionals, nurses, and physicians also resulted in significant but smaller net BP reduction. Interventions delivered by pharmacists and CHWs should be prioritized in clinical practice and in community settings to improve hypertension control.

What is Known.

• Multicomponent interventions targeting barriers to hypertension control can be effective for blood pressure reduction.

• It is unclear if the type of healthcare provider who delivers these interventions impacts implementation success.

What the Study Adds.

• Interventions delivered by pharmacists and community health workers resulted in the greatest reductions in systolic and diastolic blood pressure.

• Interventions led by health educators, multiple healthcare providers, nurses, and physicians also resulted in significant but more modest reductions in blood pressure.

Non-standard Abbreviations and Acronyms

BP

blood pressure

CHW

community health worker

CI

confidence interval