Assessment of Hypertension Complications and Health Service Use 5 Years After Implementation of a Multicomponent Intervention

Key Points Question Is a protocol-driven, team-based, multicomponent hypertension management program implemented in the public primary care setting associated with fewer complications and mortality in patients with hypertension? Findings In this cohort study that included 212 707 adults with uncomplicated hypertension, participation in the Risk Assessment and Management Program for Hypertension, offered in addition to usual primary care, was associated with clinically significant reductions in incidences of hypertension-related complications, all-cause mortality, and hospital-based health service use after 5 years. Meaning This study’s results suggest that a protocol-driven, team-based hypertension management program incorporated into the public primary care setting is a viable strategy to reduce disease burden on health care systems due to hypertension.


Introduction
Hypertension is the leading global risk factor for mortality and morbidity resulting from coronary heart disease (CHD), stroke, and end-stage kidney disease (ESKD). 1 One study 2 found that only 1 in 3 patients receiving treatment for hypertension worldwide achieved blood pressure (BP) control.
Team-based care, standardized management protocols, clinician training, and patient empowerment have shown effectiveness for BP reduction, [3][4][5] whereas multilevel, multicomponent interventions have been most effective for lowering systolic BP (SBP). 6However, the effect of implementing these strategies on the population level is unclear, and evidence for longer-term effects (>24 months) on cardiovascular outcomes, mortality, or health service use remains sparse.
Hong Kong has a dual health care system, with a universal-type public system providing more than 90% of hospital-based services, 70% of chronic disease care, and approximately 30% of acute episodic care. 7The public system is heavily strained due to an aging population, increasing prevalence of chronic diseases, and limited manpower. 8Recent studies have found that, on average, each public outpatient clinic consultation lasted for 3 to 5 minutes, 9 while each patient presented with 1.3 health problems including both new complaints and chronic diseases. 10Consequently, limited time can be spent on chronic disease care, contributing to therapeutic inertia among clinicians, poor patient-clinician communication, and patients' nonadherence to treatment due to paucity of education and structured support.
According to an internal report, more than 45% of patients with hypertension receiving public primary care in Hong Kong had uncontrolled BP.To improve the quality of hypertension care, the territory-wide Risk Assessment and Management Program for Hypertension (RAMP-HT) was launched in 2011 by the Hospital Authority, the organization that manages public health services in Hong Kong. 11The RAMP-HT is a multilevel, multicomponent, team-based intervention, targeting total cardiovascular disease (CVD) risk control of primary care patients with uncomplicated hypertension, offered in addition to usual physician-led care.Three specific services were offered: (1)   protocol-driven, nurse-led risk assessment and risk-stratified management linked to electronic action reminder system; (2) nurse intervention for patient empowerment; and (3) specialist consultations to manage resistant hypertension.
Preliminary results found that RAMP-HT participants with suboptimal BP at baseline were more likely to achieve target BP (odds ratio [OR], 1.18) and low-density lipoprotein cholesterol (LDL-C) levels (OR, 1.13), along with modest decreases in SBP, LDL-C, and predicted 10-year CVD risk compared with usual care patients after 12 months. 12It remains uncertain whether short-term reduction of these surrogate CVD markers translates to morbidity and mortality reduction over time.
This study aimed to compare hypertension-related complications and health service use at 5 years among patients with hypertension managed with RAMP-HT vs usual care.

Study Design
This territory-wide prospective matched cohort study compared the risks of CVD, ESKD, all-cause mortality, and frequencies of public health service use after 5 years between RAMP-HT participants and patients with hypertension receiving usual public primary care in Hong Kong.This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Ethics
The research protocol was approved by the Clinical and Research Ethics Committees of all 7 Hospital Authority geographical clusters in Hong Kong.Anonymous data were extracted through the Hospital Authority-Clinical Management System (HA-CMS); thus, informed consent from participants was not required in accordance with 45 CFR §46.

RAMP-HT
The setting and workflow of RAMP-HT has been described in detail previously 11,12 and in eMethods 1 in Supplement 1.In brief, RAMP-HT introduced 3 specific services to augment usual care: nurse-led risk assessment, nurse intervention, and specialist consultation.The risk assessment was repeated every 12 to 30 months, whereas nurse intervention and specialist consultation were offered when necessary.
All participants underwent a standardized assessment on CVD risk, hypertensive complications, and self-care and were stratified into having low, medium, or high CVD risk according to the Joint British Societies' (JBS2) risk calculator. 13The care-manager nurse then prepared a care plan and coordinated follow-up interventions for participants according to protocol (eFigure 1 in Supplement 1).The care plan was recorded on an electronic record platform with an action reminder system accessible at any public primary care clinic (ie, general outpatient clinic [GOPC]) to support team members' clinical decision-making, including GOPC physicians providing usual care.
Participants with adherence issues or specific risk factors were referred for nurse interventions, while patients with resistant hypertension were referred for additional specialist consultations.All participants continued to receive usual care every 8 to 16 weeks at GOPCs.The RAMP-HT was intended for all adult patients with uncomplicated hypertension managed at GOPCs.However, due to the large number of patients, limited resources, and limited manpower, recruitment could only be performed in stages.Consequently, there was an opportunity window to compare the outcomes of RAMP-HT and usual care.

Usual Care
Patients who had not been enrolled in RAMP-HT continued to receive physician-led usual care at a GOPC every 8 to 16 weeks.During a typical consultation, the physician could review a patient's BP and control of other risk factors, titrate medication, advise on lifestyle, arrange assessment, or refer to allied health professionals as appropriate, according to the Hong Kong reference framework for hypertension in primary care. 14Accesses to medications, laboratory tests, and allied health services were identical for both RAMP-HT and usual care patients.

Study Participants
Patient inclusion criteria were (1) aged 18 years or older; (2) diagnosed with hypertension, defined by the International Classification of Primary Care, 2nd Edition (ICPC-2) code of K86; (3) not diagnosed with diabetes, CVD, or ESKD (defined in next section) on or before baseline; and (4) receiving hypertension care from any GOPCs.RAMP-HT participants were patients who enrolled in RAMP-HT between October 1, 2011, and September 30, 2013; usual care patients were those who had visited any GOPC at least once for hypertension care within the same period, but had not been enrolled in the RAMP-HT by September 30, 2017.Baseline dates for RAMP-HT and usual care groups were defined as the first date of attending a RAMP-HT risk assessment session and GOPC consultation during the aforementioned period, respectively.Each patient was observed from their baseline date

Outcome Measures
The

Statistical Analysis
Missing baseline covariates were handled by multiple imputation.Missing values were imputed 5 times by chained equation method using all known baseline covariates and event outcomes. 16The same analysis was performed for each data set and the 5 sets of results were combined using Rubin rules. 17 reduce potential selection bias, all RAMP-HT participants and usual care patients were matched by propensity score-fine stratification weighting, which is an extension of propensity score matching that combines propensity score stratification with weighting technique. 18The propensity score for each patient was generated by fitting a logistic regression with the patient's corresponding RAMP-HT or usual care group as a dependent variable and all baseline covariates as independent variables.The overall analytic sample was then stratified into 1000 quintiles of the propensity score, and a weight for each individual was generated based on their corresponding stratum and treatment assignment based on inverse probability of treatment weights. 19To apply this matching approach, the MMWS package in Stata was used. 20ter weighting, summary statistics were described as mean (SD) or frequency (proportion).
Balance of baseline covariates between the 2 groups was further assessed using standardized mean difference (SMD) to show the magnitude of difference in characteristics between groups.SMD of less than 0.2 implies sufficient balance between groups. 21To evaluate the changes in clinical The number needed to treat for each outcome was calculated based on corresponding hazard ratios (HR). 22Frequencies of health service use were compared using negative binomial regression adjusted for baseline characteristics and corresponding incidence rate ratios (IRRs) were calculated.
Eight sensitivity analyses were conducted to evaluate robustness of main results.First, to minimize reverse causality, matched patients with less than 1-year follow-up were excluded.

Results
A total of 212 707 primary care patients with uncomplicated hypertension (108 045 patients in the RAMP-HT group; 104 662 patients in the usual care group) were matched with fine stratification weightings after multiple imputations and included in the analyses.The patient inclusion flow was illustrated in eFigure 2 in Supplement 1. Data completion rates for all baseline covariates ranged from 69.9% to 100% (eTable 2 in Supplement 1).Baseline characteristics of the RAMP-HT and usual care groups before and after multiple imputations before matching (eTable 3 in Supplement 1), and propensity scores distributions between RAMP-HT group and usual care group (eFigure 3 in Supplement 1) were similar.4).Results from the 8 sensitivity analyses were similar to the main analyses (eTables 6 and 7 in Supplement 1).
Further analyses showed greater risk reduction in the incidences of adverse events and less hospital-based service use (eFigure 5 in Supplement 1) in RAMP-HT participants regardless of subgroups.Participants aged 80 years or older and those with fasting glucose of at least 6.1 mmol/L exhibited a smaller reduction in risk of adverse events than other groups.

Discussion
Implementation of the RAMP-HT added onto to public usual care was associated with significant reductions in the absolute risks of incident CVD (−8.0%),ESKD (−1.6%), and all-cause mortality (−10.0%)among primary care patients with hypertension compared with usual care alone.A corresponding decrease in hospital-based service use was observed.
Reductions in CVD and mortality associated with RAMP-HT might partly be attributed to the synergistic effects of BP and LDL-C control.RAMP-HT participants had a higher likelihood of achieving BP and LDL-C control compared with usual care patients after 5 years, where a greater proportion of RAMP-HT participants were prescribed statins, amid increased statins use in both groups since introduction of the drug in GOPC formulary shortly before launch of the program.
RAMP-HT adopted a treat-to-target approach for total CVD risk management, thus explaining for the relatively small absolute differences in BP and LDL-C between groups.However, absolute risk reductions (1.6% to 14.5%) for adverse outcomes were substantial.These findings were comparable to the SPLINT trial, where a specialist nurse-led hypertension/dyslipidaemia clinic resulted in significantly lower mortality (OR, 0.55; 95% CI, 0.32-0.92)after 1 year despite similar BP reduction between groups. 24While the magnitude of risk reduction in this study was considerably greater than many large-scale drug trials, 25-27 it was consistent with findings from the HOPE-3 trial 28 and ASCOT-LLA trial 29 where use of statins in addition to BP-lowering medications reduced risk of CVD events by 40% to 50% in individuals with hypertension.A meta-analysis of 61 prospective studies further demonstrated an approximately additive effect of BP and total cholesterol reduction on lowering ischemic heart disease mortality, 30 reinforcing the importance of total CVD risk reduction among patients with hypertension to prevent adverse outcomes.
The add-on nurse-led risk assessment and intervention sessions contributed further through patient empowerment on self-care, including self-BP monitoring, smoking cessation, dietary modifications, physical activity, medication adherence and help-seeking behavior. 11,31The potential contribution of these interventions might be proxied from the greater likelihood for RAMP-HT participants to be nonsmokers, and have all 5 CVD risks being at or below target values after 5 years.
The strategic risk-stratified management also allowed for better care coordination and resource allocation.High-risk RAMP-HT participants with active health problems were channeled to attend early GOPC follow-up for review and management, hereby explaining for their higher GOPC attendance but reduced use of other hospital-based health care services.
From a post hoc analysis of the landmark Steno-2 trial in patients with diabetes, multifactorial intervention demonstrated modest improvements in disease parameters but significant relative risk reduction in CVD (59%) and all-cause mortality (46%) after a mean follow-up of 13 years. 32Similarly, the observed reductions in CVD and mortality in this study might represent a function of change in practices on multiple levels beyond disease parameter control.In addition to overcoming clinicians' treatment inertia through clinician training and use of action reminder prompts, and empowering patients' self-care capacity through extra contact time with nurses and the allied health care team, task-shifting also allowed more time for physicians to promptly recognize and manage other complex or urgent issues during the time-constrained consultation, which might contribute to the lower incidence of non-CVD mortality in RAMP-HT participants.

parameters between baseline and 5
years, values at 5 years were determined by the last available value between 4.5 years and 5.5 years after baseline.Any drop-out or loss to-follow-up before the included period were excluded from further analyses.The proportions of patients having each of 5 CVD risk factors at or below target values in RAMP-HT group were compared with those of the usual care group using logistic regression, adjusted for baseline value.For each clinical outcome measures, 5-year cumulative incidences, incidence rates, and corresponding absolute risk reduction in each group were reported.Kaplan-Meier survival curves for event incidences were compared by log-rank JAMA Network Open | Health Policy test.Multivariable Cox proportional hazards regressions were performed to estimate the association of RAMP-HT with the risk of each event outcome adjusted for baseline characteristics.Proportional hazards assumption was checked by examining plots of the scaled Schoenfeld residuals against time for the covariates.Multicollinearity was assessed by variance inflation factor.All models satisfied the proportional hazards assumption and no multicollinearity existed (variance inflation factor = 1.94).

Downloaded From: https://jamanetwork.com/ on 09/23/2023
primary composite outcome was any CVD events, including CHD, heart failure or stroke; ESKD; Second, one-to-one propensity score matching with multiple imputation without fine stratification weighting was used.Third, all eligible RAMP-HT participants and usual care patients were included in the analysis after multiple imputation without propensity score matching or weightings.Fourth, fine stratification weighting without multiple imputation was performed.Fifth, propensity score matching without multiple imputation was conducted.Sixth, a complete case analysis including only patients with complete data sets was performed.Lastly, additional analyses using data sets with 25 and 50 multiple imputations with propensity score fine stratification weightings were performed.
23l outcomes were further compared in subgroups, stratified by gender, age, smoking status, SBP, fasting glucose, LDL-C, eGFR, body mass index (BMI, calculated as weight in kilograms divided by height in meters squared), estimated Framingham 10-year CVD risk, the Charlson Comorbidity Index, and number and classes of antihypertension drugs and lipid-lowering agents prescribed.Fine stratification weightings were carried out for each subgroup analysis.Interactions between the RAMP-HT effect and each group were assessed, significance of interactions were based on Hommeladjusted P values.23Alltests were 2-tailed and P < .05 was considered statistically significant.Statistical analyses were performed with Stata version 13.0 (StataCorp) from January 2019 to March 2023.

Table 1
statins was observed in the RAMP-HT group (RAMP-HT: from 7.8% to 39.4% vs usual care: from 7.3% to 32.9%) (eTable 5 in Supplement 1).The 5-year cumulative incidence of the primary composite outcome was 11.8% (n = 12 784) in RAMP-HT group and 26.3% (n = 27 514) in the usual care group (Table3), corresponding to an absolute risk reduction of 14.5% for RAMP-HT participants.RAMP-HT participants had an absolute risk reduction of 8.0% in CVD, 1.6% in ESKD, and 10.0% in all-cause mortality.Between-group differences in cumulative hazards for adverse outcomes were found as early as 1 year after baseline (Figure), and continued at least up to the fifth year.After adjusting for all baseline covariates, 64 012 RAMP-HT participants (67.3%) had LDL-C levels at or below target (vs 46 048 [61.8%] in the usual care group (Table 2).After adjustments for baseline values, RAMP-HT participants had 36% higher odds (OR, 1.36; 95% CI, 1.33-1.39) of having all 5 CVD risk factors being at or below target values compared to usual care patients.A greater increase in the proportion of patients prescribed JAMA Network Open | Health Policy Hypertension Complications and Health Service Use 5 Years After Starting Multicomponent Intervention JAMA Network Open.2023;6(5):e2315064.doi:10.1001/jamanetworkopen.2023.15064(Reprinted) May 24, 2023 5/14 Downloaded From: https://jamanetwork.com/ on 09/23/2023

Table 2 .
Proportion of Patients in the RAMP-HT or Usual Care Achieving Clinical Targets at Baseline and 5 Years SI unit conversion factors:To convert glucose to mg/dL, divide by 0.0555; to convert LDL-C to mg/dL, divide by 0.0259.a OR were adjusted for the corresponding target of clinical outcome at baseline.

Table 3 .
Outcome Events at 5 Years in RAMP-HT Participants and Usual Care Patients Hazard ratios were adjusted by gender, age, smoking status, systolic blood pressure, diastolic blood pressure, fasting glucose, low-density lipoprotein cholesterol, TC/HDL-C ratio, triglyceride, body mass index, estimated glomerular filtration rate, Charlson Comorbidity Index and the usages of angiotensin converting enzyme inhibitor/angiotensin receptor blocker, β-blocker, calcium channel blocker, diuretic, other antihypertensive drugs, statin and fibrate at baseline.Strengths of this study included the inclusion of a large representative cohort of primary care patients with hypertension over the long-term.Our observations reflected how such a model of care benefitted all groups of patients with different levels of literacy, motivation, and baseline health.Several sophisticated analytical techniques including multiple imputations, fine stratification weightings and adjustments of inclusion criteria were used to minimize imbalances in baseline Figure.Cumulative Hazards for Cardiovascular Disease, End-Stage Kidney Disease, All-Cause Mortality, and All Composite Events of Primary Outcome Between RAMP-HT and Usual Care Groups Abbreviations: ARR, absolute risk reduction; CHD, coronary heart disease; CVD, cardiovascular disease; ESKD, end-stage kidney disease; HR, hazard ratio; NNT, number needed to treat; RAMP-HT, Risk Assessment and Management Program for Hypertension.a

Table 4 .
Public Health Service Use of RAMP-HT Participants and Usual Care Patients at 5 Years Gupta A, Mackay J, Whitehouse A, et al.Long-term mortality after blood pressure-lowering and lipid-lowering treatment in patients with hypertension in the Anglo-Scandinavian Cardiac Outcomes Trial (ASCOT) Legacy study:16-year follow-up results of a randomised factorial trial.Lancet.2018;392(10153):1127-1137.doi:10.1016/S0140-6736(18)31776-826.Group UPDS; UK Prospective Diabetes Study Group.Tight blood pressure control and risk of macrovascular and microvascular complications in type 2 diabetes: UKPDS 38.BMJ.1998;317(7160):703-713. doi:10.1136/bmj.Sever PS, Dahlöf B, Poulter NR, et al.Prevention of coronary and stroke events with atorvastatin in hypertensive patients who have average or lower-than-average cholesterol concentrations, in the Anglo-Scandinavian Cardiac Outcomes Trial--Lipid Lowering Arm (ASCOT-LLA): a multicentre randomised controlled trial.Fu SN, Dao MC, Luk W, et al.A cluster-randomized study on the Risk Assessment and Management Program for home blood pressure monitoring in an older population with inadequate health literacy.J Clin Hypertens (Greenwich).2020;22(9):1565-1576.doi:10.1111/jch.1398732.Gaede P, Lund-Andersen H, Parving HH, Pedersen O. Effect of a multifactorial intervention on mortality in type 2 diabetes.N Engl J Med. 2008;358(6):580-591. doi:10.1056/NEJMoa070624533.Yoshida K, Hernández-Díaz S, Solomon DH, et al.Matching weights to simultaneously compare three treatment groups: comparison to three-way matching.Epidemiology.2017;28(3):387-395.doi:10.1097/EDE.000000000000062734.Desai RJ, Franklin JM.Alternative approaches for confounding adjustment in observational studies using weighting based on the propensity score: a primer for practitioners.BMJ.2019;367:l5657.doi:10.1136/bmj.l565735.Wong L, Lee MK, Mak HT, et al.Accuracy and completeness of ICPC coding for chronic disease in general outpatient clinics.HK Pract.2010;32:129-135.Setting of RAMP-HT and Usual Care eMethods 2. Baseline Covariates eFigure 1. Risk Assessment & Management Program -Hypertension (RAMP-HT) Workflow eFigure 2. Patient Inclusion Flow Chart eFigure 3. Propensity Scores Distributions of RAMP-HT Group and Usual Care GroupeFigure 4. Means of Systolic Blood Pressure (SBP), Diastolic Blood Pressure (DBP), Fasting Glucose (FG), Low Density Lipoprotein -Cholesterol (LDL-C) and Body Mass Index (BMI) of RAMP-HT Group and Usual Care Group eFigure 5. Adjusted Hazard Ratios (HRs) of RAMP-HT Participants Over Usual Care Patients Associated With the Incidences of Cardiovascular Diseases (CVD), End-Stage Renal Disease (ESRD), All-Cause Mortality and All Composite Event in Selected Subgroups by Multivariable Cox Proportional Hazards Regressions eFigure 6. Adjusted Incident Rate Ratios (IRRs) of RAMP-HT Participants Over Usual Care Patients Associated With the Number of Hospitalization, Accident and Emergency Department (AED) Attendance, Special Outpatient Clinic (SOPC) and General Outpatient Clinic (GOPC) Attendance in Selected Subgroups by Negative Binomial Regressions eTable 1. Definition of the Event Outcome Measures eTable 2. Data Completion Rates of Matched RAMP-HT Participants and Usual Care Patients at Baseline and 5-Year Follow-up eTable 3. Baseline Characteristics Between RAMP-HT Participants and Usual Care Patients Before and After Multiple Imputation Without Fine Stratification Weightings eTable 4. Frequency of RAMP-HT Service Attendances Among RAMP-HT Participants eTable 5. Patients' Characteristics Between RAMP-HT Participants and Usual Care Patients at 5-Year Follow-up After Fine Stratification Weightings JAMA Network Open | Health Policy Hypertension Complications and Health Service Use 5 Years After Starting Multicomponent Intervention Sensitivity Analyses on Comparisons of All Outcome Events, Diabetes Mellitus, Cardiovascular Disease, End Stage Renal Disease and All-Cause Mortality Between RAMP-HT Participants and Usual Care Patients eTable 7. Sensitivity Analyses on Comparisons of Service Use Between RAMP-HT Participants and Usual Care Patients Abbreviations: AED, accident and emergency department; GOPC, general outpatient clinics; IRR, incidence rate ratio; RAMP-HT, Risk Assessment and Management Program for Hypertension; SOPC, specialist outpatient clinics.aAllincidencerate ratios were adjusted by gender, age, smoking status, systolic blood pressure, diastolic blood pressure, fasting glucose, LDL-C, TC/HDL-C ratio, triglyceride, BMI, eGFR, Charlson Comorbidity Index and the usages of ACEI/ARB, β-blocker, CCB, diuretic, other antihypertensive drugs, statin and fibrate, and the corresponding number of service uses at baseline.25.JAMA Network Open.2023;6(5):e2315064.doi:10.1001/jamanetworkopen.2023.15064(Reprinted) May 24, 2023 13/14 Downloaded From: https://jamanetwork.com/ on 09/23/2023 eTable 6.