Effect of Self-monitoring and Medication Self-titration on Systolic Blood Pressure in Hypertensive Patients at High Risk of Cardiovascular Disease: The TASMIN-SR Randomized Clinical Trial

Importance  Self-monitoring of blood pressure with self-titration of antihypertensives (self-management) results in lower blood pressure in patients with hypertension, but there are no data about patients in high-risk groups.

Objective  To determine the effect of self-monitoring with self-titration of antihypertensive medication compared with usual care on systolic blood pressure among patients with cardiovascular disease, diabetes, or chronic kidney disease.

Design, Setting, and Patients  A primary care, unblinded, randomized clinical trial involving 552 patients who were aged at least 35 years with a history of stroke, coronary heart disease, diabetes, or chronic kidney disease and with baseline blood pressure of at least 130/80 mm Hg being treated at 59 UK primary care practices was conducted between March 2011 and January 2013.

Interventions  Self-monitoring of blood pressure combined with an individualized self-titration algorithm. During the study period, the office visit blood pressure measurement target was 130/80 mm Hg and the home measurement target was 120/75 mm Hg. Control patients received usual care consisting of seeing their health care clinician for routine blood pressure measurement and adjustment of medication if necessary.

Main Outcomes and Measures  The primary outcome was the difference in systolic blood pressure between intervention and control groups at the 12-month office visit.

Results  Primary outcome data were available from 450 patients (81%). The mean baseline blood pressure was 143.1/80.5 mm Hg in the intervention group and 143.6/79.5 mm Hg in the control group. After 12 months, the mean blood pressure had decreased to 128.2/73.8 mm Hg in the intervention group and to 137.8/76.3 mm Hg in the control group, a difference of 9.2 mm Hg (95% CI, 5.7-12.7) in systolic and 3.4 mm Hg (95% CI, 1.8-5.0) in diastolic blood pressure following correction for baseline blood pressure. Multiple imputation for missing values gave similar results: the mean baseline was 143.5/80.2 mm Hg in the intervention group vs 144.2/79.9 mm Hg in the control group, and at 12 months, the mean was 128.6/73.6 mm Hg in the intervention group vs 138.2/76.4 mm Hg in the control group, with a difference of 8.8 mm Hg (95% CI, 4.9-12.7) for systolic and 3.1 mm Hg (95% CI, 0.7-5.5) for diastolic blood pressure between groups. These results were comparable in all subgroups, without excessive adverse events.

Conclusions and Relevance  Among patients with hypertension at high risk of cardiovascular disease, self-monitoring with self-titration of antihypertensive medication compared with usual care resulted in lower systolic blood pressure at 12 months.

Trial Registration  isrctn.org Identifier: ISRCTN87171227

Elevated blood pressure is the leading risk factor for global disease burden.¹Data from national and international surveys suggest that despite improvements over the last decade, significant proportions of patients have poor control of their elevated blood pressure.^2-5 Using the revised Eighth Joint National Committee (JNC 8) 2014 blood pressure guideline,⁶ which proposed less restrictive targets for adults aged 60 years or older and for those with diabetes and chronic kidney disease, the proportion of adults in the United States with treatment-eligible hypertension who met blood pressure goals was less than half for younger adults (improved from 41.2% under JNC 7 to 47.5% under JNC 8 criteria) and less than two-thirds for older adults (although improved from 40% under JNC 7 to 65.8% under JNC 8).⁷ Most management of hypertension is undertaken in primary care, where it comprises the most common long-term condition seen by family physicians, so it is appropriate that interventions are delivered in this setting.⁸ Self-monitoring is now common, with approximately a third of patients with hypertension using it in the United Kingdom and more internationally.^9,10 Trials investigating self-monitoring have shown promise in the reduction of blood pressure particularly when combined with other interventions.¹¹

The Telemonitoring and Self-Management in Hypertension 2 (TASMINH 2) trial found that self-management, comprising self-monitoring with self-titration of antihypertensives, resulted in significantly lower (5.4 mm Hg) systolic blood pressure after 1 year than did usual care.¹²However, the study included few patients with high-risk conditions such as cardiovascular disease, diabetes, or chronic kidney disease, in whom the blood pressure differences appeared to be smaller, suggesting the need for further investigation.

The potential advantage from optimal blood pressure control in patients at higher cardiovascular risk is large because the absolute benefit increases with absolute risk.¹³ Guideline recommendations for blood pressure lowering vary for different high-risk groups.^6,14-18 At the time of protocol development for this study, the British Hypertension Society and other international guidelines had suggested a blood pressure target of less than 130/80 mm Hg for patients with stroke or transient ischemic attack, diabetes, stage 3 chronic kidney disease (without proteinuria), coronary heart disease, and myocardial infarction, providing uniformity across the range of high-risk groups.^14,19

The aim of this trial was to determine whether self-management of hypertension resulted in lower blood pressure than usual care in a population of patients at high risk of cardiovascular events.

Targets and Self-Management for the Control of Blood Pressure in Stroke and at Risk Groups (TASMIN-SR) was a randomized unblinded trial with automated ascertainment of the primary end point. The trial methods have been described in detail elsewhere, but briefly, patients with a diagnostic Read code (clinical code) for at least 1 of the following: stroke or transient ischemic attack; diabetes; stage 3 chronic kidney disease (estimated glomerular filtration rate, 30-59 mL/min/m²); coronary artery bypass graft surgery; myocardial infarction or angina with poorly controlled blood pressure (last recorded practice reading >145 mm Hg systolic) who were not under the care of a specialist were identified by their family physician using electronic searches of practice clinical record systems.²⁰ (See Supplement 1 for the trial protocol.) Family physicians reviewed the invitation list and excluded patients with terminal illness, patients who were house bound, or patients they otherwise believed to be unsuitable. The remaining potentially eligible participants were invited to their local clinic for a baseline examination conducted by the research team in conjunction with the Primary Care Research Networks in central and east England.²¹

To be eligible, patients had to be aged 35 years or older, have at least 1 of the high-risk conditions (cardiovascular disease, diabetes, stage 3 chronic kidney disease, or coronary heart disease), and have a blood pressure reading during the baseline examination of at least 130/80 mm Hg. Participants were not required to have been prescribed antihypertensive medication. Patients were excluded if they could not self-monitor because of dementia or if they had a score of more than 10 on the short-orientation memory concentration test; had blood pressure greater than 180/100 mm Hg; had postural hypotension, systolic blood pressure drop of more than 20 mm Hg; took more than 3 antihypertensive medications; were participating in another blood pressure study, had participated in TASMINH 2,¹² or had a spouse who had been randomized already in the current trial; had a terminal disease; were pregnant; were receiving care for their blood pressure by a specialist rather than by a primary care physician; or had experienced an acute cardiovascular event in the previous 3 months (Figure 1).

Ethical approval was obtained from the North West—Greater Manchester East ethics committee (reference: 10/H1013/60) and site-specific research approval was obtained from the relevant primary care organizations.

After hearing the explanation of the study and giving written informed consent, eligible patients were randomized between March and December 2011 using an Internet-based system with telephone backup to either usual care or self-management and were followed up for 1 year. Minimization, a method of adaptive stratified sampling that balances the different groups of clinical trials simultaneously for several factors, was used to balance treatment allocation by family practice, sex, age, high-risk group, and baseline systolic blood pressure, factors chosen due to their potential influence on systolic blood pressure.²² One amendment was made following commencement of the study to allow reminder invitations to be sent to nonresponders. Patients randomized to usual care booked an appointment for a routine blood pressure check and medication review (including dose adjustment if required) with the participating family physician. Thereafter, blood pressure measurement, blood pressure targets, or adjustment of medication for patients receiving usual care were at the discretion of the family physician.

Patients randomized to self-management were trained to self-monitor blood pressure using a validated monitor (Microlife Watch BP Home²³) with self-titration of medication following a predetermined plan, in 2 or 3 sessions, each lasting approximately an hour. Following training, intervention patients went to their family physician to agree with the individualized 3-step plan to increase or add antihypertensive medications. This was operationalized in a paper-based algorithm including the option for additional blood tests if required. Patients took their blood pressure twice each morning for the first week of each month using simple color-coded instructions developed for the TASMINH 2 trial.¹² Four or more blood pressure readings recorded during the measurement week for 2 consecutive months that were higher than the target necessitated a change in medication pursuant to the predetermined plan. Very high or very low readings (blood pressure >180/100 mm Hg or <100 mm Hg systolic, eFigure 1 in Supplement 2) required the participant to contact his/her practice. When a medication change was needed, patients sent a paper form to their family physicians without any need for a consultation. Medication choice remained at the discretion of the family physician. If patients used all 3 steps of their management plan, they returned to their general practitioner for additional instructions.

We selected blood pressure of less than 120/75 mm Hg as a target reading for the self-titration algorithms based on the British Hypertension Society (BHS)⁹ and the Joint British Societies Guidelines¹⁵ for patients with stroke or transient ischemic attack, diabetes (in the absence of proteinuria), chronic kidney disease, or coronary heart disease.

The primary prespecified outcome was the difference between intervention and control in systolic blood pressure at 12 months, taking into account baseline blood pressure and minimization factors. Patients attended 2 follow-up research clinics at 6 and 12 months after randomization. At both baseline and follow-up visits, blood pressure was measured by a research facilitator systematically after 5 minutes of rest using a validated electronic automated sphygmomanometer (Bp-TRU blood pressure M 100 or 200).²⁴ Six blood pressure readings were taken at 1-minute intervals. The mean of the second and third readings is considered to be best practice for obtaining a clinic blood pressure reading according to many international guidelines; therefore, this was used for the primary outcome. The main analysis was also rerun using the mean of the second to sixth blood pressure readings to reduce any influence of alerting effect to cuff inflation. Outcome ascertainment was not blind to allocation but was determined independently of the clinical team by a researcher using the automatic mode of the sphygmomanometer to measure the blood pressure without the need for intervention other than to place the cuff on the patient and switch on the monitor to reduce the potential for bias.

Other baseline clinical and questionnaire data were collected at the same clinics.²⁰ Prescribed medications were recorded from the electronic patient record with quality of life, anxiety, and adverse effects measured using standard questionnaires.^25-27 To allow comparisons of the amount of antihypertensive medications taken, individual drug doses were converted into defined daily doses (a World Health Organization–defined assumed average maintenance dose per day for a drug used for its main indication in adults).²⁸

Analyses were performed using STATA version 12 (StataCorp). A sample size of 243 patients per group was estimated for 90% power assuming a standard deviation of 17 mm Hg and a difference of at least 5 mm Hg in systolic blood pressure between intervention and control groups based on data from our previous trial.¹²

Assuming a 10% dropout rate during follow-up, a sample of 270 per group was required; a dropout rate of 20% would result in more than 85% power. The primary analysis included all participants who attended 12-month follow-up and had complete data for the primary outcome, without imputation. A mixed model was used to examine differences in between-group systolic blood pressure at 12 months, adjusting for baseline blood pressure, practice (as a random effect), sex, and high-risk group. Sensitivity analyses examined the potential effect of missing data including multiple imputation and replacement of missing data by the most recent previous data or by the mean of the series. For multiple imputation, 10 multiply-imputed data sets were generated using predictive mean-matching methods under the missing at random assumptions. Planned subgroup analyses were older vs younger (65 years as the threshold), men vs women, better controlled at baseline vs worse controlled at baseline (threshold, 145 mm Hg systolic), the different risk groups, and socioeconomic status.

Of 10 764 potentially eligible patients from 59 family practices, 3353 were excluded by their family physician for being housebound, having a terminal illness, or not being thought suitable candidates. Of the remaining 7411 who were invited to participate, 1201 attended a baseline clinic and were assessed for eligibility. Of the 2003 who provided a reason for declining invitation (> 1 answer possible), 858 (43%) did not want to take part in a trial, 717 (36%) did not want to alter their own medication, and 639 (32%) did not want to measure their own blood pressure. Of the 646 patients who were excluded during the baseline examination, 350 (54%) had blood pressure readings that were not within the inclusion range and 138 (21%) withheld consent (Figure 1).

Of 555 patients randomized, 3 were randomized in error and were immediately excluded from the study, did not receive any intervention, and were not followed up or analyzed further. This left 276 patients in each group. After 12 months, 220 patients in the intervention group and 230 in the control group attended the final follow-up, providing 450 (81%) complete cases for analysis. Most who dropped out did so in the first 6 months (Figure 1). Table 1 shows that the baseline characteristics of participants were well matched between groups. Participants for whom outcome data were not available were of similar age, had similar baseline blood pressure, but were less likely to be men (eTable 1 in Supplement 2).

The primary analysis plan specified adjusted results, but because these were very similar to the unadjusted results, the latter are presented for simplicity (see eTable 2 in Supplement 2 for adjusted results). After 12 months, there was a mean systolic blood pressure difference of 9.2 mm Hg (95% CI, 5.7-12.7) between the groups (Table 2). Multiple imputation for missing values showed a marginally lower mean difference in systolic blood pressure of 8.8 mm Hg (95% CI, 4.9-12.7). Further sensitivity analyses by the last observation carried forward or the mean of the series did not materially affect the primary outcome (eTable 3 in Supplement 2). The mean of the second to sixth blood pressure readings was almost identical to the primary analysis (mean difference in systolic blood pressure at 12 months, 9.1 mm Hg; 95% CI, 5.8-12.3; eTable 4 in Supplement 2).

After 6 months, there was a mean between-group systolic blood pressure difference of 6.1 mm Hg (95% CI, 2.9-9.3). There was also a mean between-group diastolic blood pressure difference of 3.0 mm Hg (95% CI, 1.4-4.7) at 6 months and 3.4 mm Hg (95% CI, 1.8-5.1) at 12 months (Table 2). After multiple imputation the point estimates were slightly lower: the mean systolic between-group difference at 6 months was 5.5 mm Hg (95% CI, 1.6-9.5) and the mean diastolic difference was 2.7 mm Hg (95% CI, 0.4-5.1) at 6 months and 3.1 mm Hg (95% CI, 0.7-5.5) at 12 months. There were no significant differences in the primary outcome within any of the prespecified subgroups (Figure 2).

Prescription of antihypertensive drugs increased in both groups but significantly more in the intervention group: the mean defined daily doses at 12 months for the intervention group was 3.34 (95% CI, 3.1, 3.7) vs 2.61 (95% CI, 2.4-2.9) for the control group (mean difference, 0.9; 95% CI, 0.7-1.2; Table 3, adjusted results are presented in eTable 5 in Supplement 2). Comparison with the number of drug classes prescribed shows that this represents both an increase in dose and in the number of medications. The main changes seen were in the prescription of calcium channel blockers, and thiazides, which significantly increased in the intervention group compared with the control group (Table 3).

Although reported adverse symptoms were common in both groups (Table 4), there were no significant differences between control and intervention groups. Additional symptoms that could be linked to antihypertensive treatment were not significantly different between groups including dizziness, impotence, and rash. Two patients in the control group were admitted to the hospital with chest pain, 1 on 3 occasions; 3 had transient ischemic attacks; and 1 had a possible stroke. In the intervention group, 3 patients were admitted to the hospital with chest pain, 2 were admitted with arrhythmias, and 4 had transient ischemic attacks. One control patient and 1 intervention patient died during the study; neither death was judged to be study related.

There were no significant differences between groups in quality of life measured by the EQ-5D at 6 or 12 months (eTable 6 in Supplement 2).

This trial has shown for the first time, to our knowledge, that a group of high-risk individuals, with hypertension and significant cardiovascular comorbidity, are able to self-monitor and self-titrate antihypertensive treatment following a prespecified algorithm developed with their family physician and that in doing so, they achieved a clinically significant reduction in systolic and diastolic blood pressure without an increase in adverse events. These results were sustained and increasing during the 12 months of the trial. Based on systematic reviews of clinical outcome trials,¹³ the blood pressure difference observed in those self-managing would be expected to be associated with an approximate 30% reduction in stroke risk should it be sustained.^12,13

In terms of weaknesses, the follow-up of patients in the trial was not as high as hoped. Nevertheless, primary outcome data were available on more than 80% of participants, and differences in blood pressure between groups were similar whether or not missing data were accounted for in the sensitivity analyses. Given that the trial population had significant comorbidity, it is to be expected that loss to follow-up would be higher than in a hypertensive population without these comorbidities. Those lost to follow-up were more likely to be men (especially in the intervention group). Recruitment took place in 59 family practices over a wide geographical area and hence logistics were complex. Most of the dropout occurred between baseline and 6 months, particularly in the intervention group, which may have reflected patients who felt unable to continue in the trial once exposed to the intervention.

Included patients were mostly white, from a professional or skilled manual background, and were prescribed 3 or fewer antihypertensives, which might limit generalizability. Randomized groups were similar with small differences in comorbidities in favor of the intervention group. No difference in blood pressure reduction from the intervention was seen between the subgroups examined, but this may reflect inadequate statistical power. Any practice effects were taken into account in the randomization and method of analysis. Individual randomization and dropouts from the intervention group could have caused contamination between the groups, but this would have biased the results toward no effect. Similarly, high-performing practices taking part in research would have also mitigated against the observed effect.

Relatively small proportions of those potentially eligible to take part were eventually randomized. Family physicians could exclude patients from invitation to the trial who were housebound, had terminal illness, or who were thought to be unsuitable, which is likely to have included frailer patients. Nevertheless, those included were older (mean age 70 years) and had more comorbidities than our previous work (22% had 2 or more strokes, coronary heart disease, diabetes or chronic kidney disease). As with TASMINH 2, only approximately 8% of those invited to take part were randomized. Responses from more than 2000 of those who declined suggest that nonresponse reflected a combination of not wishing to take part in a trial and not wishing to self-manage. More than double the number randomized were prepared to self-manage as measured by those attending eligibility screening, and of those excluded, controlled blood pressure was the commonest reason. Taken together, for patients outside of a trial situation, we estimate that the intervention might be suitable for about 20%. The study was unblinded but ascertainment of outcome was by automated sphygmomanometer, which did not require research facilitator input other than to fit the cuff to the patient and switch on the monitor. The potential for the intervention group to become habituated to blood pressure measurement was lessened by the use of the BP-TRU monitor (which takes 6 readings at a time) for all study end points in both randomization groups and the fact that the primary outcome was almost identical whether the mean of the second and third or second to sixth blood pressure readings was used.

Patients in the intervention group were using home targets based on the then recommended clinic target of 130/80 mm Hg for all 4 groups.^17,19 In the intervening years, target recommendations have tended to roll back toward 140/90 mm Hg or higher for most conditions although UK stroke guidelines and those for diabetes in the presence of renal disease remain equivalent to those used in the trial.¹⁴ Patients in the control group received usual care without specification of target, which may have accentuated the difference between groups given that the achieved mean blood pressure in the control group was 138/76 mm Hg.

Three previous trials have considered self-monitoring with self-titration of antihypertensives.^12,29,30 Two trials, including our previous study, showed reduction of blood pressure through self-monitoring with self-titration.^12,29 The third, which used a cluster design, found that a web-based self-titration intervention increased blood pressure monitoring but did not affect blood pressure.³⁰

The current study achieved a greater blood pressure reduction than seen previously and seems to have been mediated through greater use of medication in the intervention representing both an increase in dose and in number of antihypertensive medications. Increases were particularly observed in the use of thiazide diuretics and calcium channel blockers comprising a difference of almost 1 defined daily dose between randomized groups.²⁸ Adherence to study medication—difficult to ascertain accurately and typically high in studies such as this—was not measured, but the observed difference of 9 mm Hg systolic blood pressure seen between groups is what would be expected for this degree of medication intensification.^13,31

A recently published systematic review found no other self-titration trials but showed a range of blood pressure reductions from interventions combining self-monitoring with additional support compared with usual care in other high-quality trials.¹¹ For these studies at the 12-month follow-up, there was consistent benefit with a mean net reduction in both systolic blood pressure (range, −2.1 to −8.3 mm Hg), diastolic blood pressure (range, 0.0 to −4.4 mm Hg), or both. The blood pressure differences from the current study are at the upper end of these values and broadly equivalent to that achieved in other trials of self-monitoring combined with behavioral self-management or a web-based intervention and additional pharmacist care.^32,33

This study has shown that self-monitoring with self-titration of antihypertensives is feasible and achievable in a high-risk population without special equipment and by following a modest amount of training and additional family physician input. This is a population with the most to gain in terms of reducing future cardiovascular events from optimized blood pressure control. Furthermore, despite the significantly reduced blood pressure, no additional adverse events were observed. Validated semiautomated blood pressure monitors are now widely available, costing as little as US $25 (£15, €18), meaning that with training delivered by nurses, this intervention could be implemented widely. At least 30% of patients with hypertension are already self-monitoring in the United Kingdom and more internationally,¹⁰ and there is a significant prevalence of comorbidities,³⁴ suggesting that self-management could be appropriate for many individuals, notwithstanding the issues discussed concerning generalizability.

Among hypertensive patients at high risk of cardiovascular disease, self-monitoring with self-titration of antihypertensive medication, compared with usual care, resulted in lower systolic blood pressure at 12 months. Patients at high risk of cardiovascular disease whose blood pressure is not optimally controlled could be considered for self-management.

Corresponding Author: Richard McManus, FRCGP, NIHR School for Primary Care Research, Nuffield Department of Primary Care Health Sciences, University of Oxford, Woodstock Rd, Radcliffe Observatory Quarter, Oxford, Oxfordshire OX2 6GG, United Kingdom (richard.mcmanus@phc.ox.ac.uk).

Author Contributions: Dr McManus had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: McManus, Mant, Bray, Bryan, Greenfield, Jowett, Little, Hobbs.

Acquisition, analysis, or interpretation of data: McManus, Mant, Haque, Bray, Bryan, Greenfield, Jones, Jowett, Penaloza, Schwartz, Shackleford, Shovelton, Varghese, Williams.

Drafting of the manuscript: McManus, Mant, Little, Schwartz, Shovelton, Hobbs.

Critical revision of the manuscript for important intellectual content: McManus, Mant, Haque, Bray, Bryan, Greenfield, Jones, Jowett, Penaloza, Shackleford, Varghese, Williams, Hobbs.

Statistical analysis: Haque, Bryan, Penaloza.

Obtained funding: McManus, Mant, Bryan, Greenfield, Hobbs.

Administrative, technical, or material support: Bray, Jones, Schwartz, Shackleford, Shovelton, Varghese, Hobbs.

Study supervision: McManus, Bray, Bryan, Jowett, Little, Williams.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr McManus reported that he has received equipment for research purposes from Omron and Lloyds Healthcare. Dr Williams reported that he works in academic collaboration with Healthstats, Singapore, in developing novel blood pressure monitoring approaches. Dr Hobbs reported that he has received limited research support in terms of blood pressure devices from Microlife and BP TRU and that he is supported by the NIHR SPCR, the NIHR Oxford BRC, and the NIHR CLAHRC Oxford. No other disclosures were reported.

TASMIN-SR Investigators Collaborators: PCRN Research Nurses: Somi Spannuth, Fenglin Guo, Sian Sylvester, Julie Timmins, Jacki Mundy, Kay Finney, Ann McDonald, Wendy Herring, Karla Preece, Sue Read, Emma Rayfield, Sharon Jones, Sue Maiden, Lesley Maloney, Pauline Derbyshire, Sue Smolen, Steve Hurdowar, and Mandy Aspinall.

Birmingham and Black Country: Bosworth Medical Centre, Dr Madeline Bates; Coseley Medical Practice, Dr Perumal Anandakumar; Eden Court Medical Practice, Dr Richard Edwards; Goodrest Croft Surgery, Dr Peter Giddings; Grange Hill Surgery, Dr Ramila Patel; Greenridge Surgery, Dr Nash Qamar; Jiggins Lane Surgery; Laurie Pike Medical Centre, Dr Will Murdoch; Little London Surgery, Dr Mandeep Chander; Newton Road Surgery, Dr Parag Pal; Northgate Practice, Dr Arun Singal; Norvic Family Health Centre, Dr Faheem Khan; Riverbrook Medical Centre, Dr Naresh Chauhan; Sherwood House Medical Centre, Dr Lawrence Miller; Swan Medical Centre, Dr Emamoke Ubogu; Swanswell Medical Centre, Dr Philip Schuppler; The Jacey Practice, Dr Mari Reynolds; The Lower Gornal Health Centre, Dr Naeem Malik; Wand Medical Centre, Dr Daryl Goodwin; and Yardley Wood Medical Centre, Dr Tina Banerjee.

East of England: Ambrose Avenue Group, Dr Ayotunde Ajala; Baddow Village Surgery, Dr Stephen Russell; Chafford Hundred Medical Centre, Dr Tonio Abela; Church Street Surgery, Dr Nicolette Williams; Churchfield Medical Centre, Dr Gina Johnson; Coltishall Surgery, Dr Neil Taylor; Dolphin House Surgery; Eden Surgery, Dr Rikin Patel; Great Notley Surgery, Dr Ramoo; Greensward Surgery, Dr Biju Kuriakose; John Tasker House, Dr Gervase Vernon; Leighton Road Surgery, Dr Farah Paruk; London Road Surgery; Martlesham Heath Surgery, Dr Anne Fitzgerald; Oaklands Surgery, Dr Shashi Rai; Ongar Health Centre, Dr Simon Whitehead; Orford Lodge, Dr Subhir Rohatghi; Prospect Medical Practice, Dr Nana Widmaier; Salisbury House Surgery, Dr Haseeb Wadud; St Stephensgate Medical Centre, Dr Frances Scouller; Stifford Clays, Dr Manoj Chandran; Sundon Medical Centre, Dr Haydn Williams; The Rookery Medical Centre, Dr Malini Wace; and The Surgery, Dr Nasir Jamal; Woolpit Health Centre.

West Midlands North: Ashley Surgery, Dr Denise Bladen; Penn Manor Medical Practice, Dr Manjeet Samra; and Cloisters Medical Practice, Dr Gulshan Kaul.

West Midlands South: Atherstone Surgery, Dr Trevor Gooding; Churchfields Surgery, Dr Ian Morrey; The Marches, Dr Crispin Fisher; and Trinity Court, Dr David Buckley.

Funding/Support: This article presents independent research funded by the National Institute for Health Research (NIHR) under its Programme Grants for Applied Research Programme (Grant Reference Number RP-PG 0606-1153), by the NIHR National School of Primary Care Research (NSPCR 16), and by an NIHR career development fellowship (Dr McManus). The study sponsor was the University of Birmingham.

Role of the Funder/Sponsor: The NIHR had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The views expressed in this presentation are those of the authors and not necessarily those of the National Health Service, the NIHR, or the Department of Health.

Additional Contributions: We thank Amanda Davies and Fran Palmer (University of Birmingham) for administrative work on the project and to Roger Holder, Emeritus Head of Statistics in Primary Care in (University of Birmingham), who was the original trial statistician before handing over to Dr Haque. Nicki Spillman and Beryl Caswell for their input into the Steering Group as Voluntary Lay Representatives. Data Monitoring and Ethics Committee members Dr Martyn Lewis (University of Keele, chair), Dr Philip Evans (University of Exeter), Dr Matthew Giles (University Hospitals, Oxford), none of whom received compensation. The Primary Care Research Network Central England (specifically Mrs Ros Salter, Ms Jenny Stevens, Ms Jenny Titley, Prof Jeremy Dale, Ms Sue Elwell, and Dr Mark Porcheret) and Primary Care Research Network East of England (specifically Ms Camilla Croucher, and Dr Jonathan Graffy), none of whom received compensation.

Correction: This article was corrected October 14, 2014, to include the collaborators representing West Midlands South.