Association of guideline and policy changes with incidence of lifestyle advice and treatment for uncomplicated mild hypertension in primary care: a longitudinal cohort study in the Clinical Practice Research Datalink

Objectives Evidence to support initiation of pharmacological treatment in patients with uncomplicated (low risk) mild hypertension is inconclusive. As such, clinical guidelines are contradictory and healthcare policy has changed regularly. The aim of this study was to determine the incidence of lifestyle advice and drug therapy in this population and whether secular trends were associated with policy changes. Design Longitudinal cohort study. Setting Primary care practices contributing to the Clinical Practice Research Datalink in England. Participants Data were extracted from the linked electronic health records of patients aged 18–74 years, with stage 1 hypertension (blood pressure between 140/90 and 159/99 mm Hg), no cardiovascular disease (CVD) risk factors and no treatment, from 1998 to 2015. Patients exited if follow-up records became unavailable, they progressed to stage 2 hypertension, developed a CVD risk factor or received lifestyle advice/treatment. Primary outcome measures The association between policy changes and incidence of lifestyle advice or treatment, examined using an interrupted time-series analysis. Results A total of 108 843 patients were defined as having uncomplicated mild hypertension (mean age 51.9±12.9 years, 60.0% female). Patientsspent a median 2.6 years (IQR 0.9–5.5) in the study, after which 12.2% (95% CI 12.0% to 12.4%) were given lifestyle advice, 29.9% (95% CI 29.7% to 30.2%) were prescribed medication and 19.4% (95% CI 19.2% to 19.6%) were given both. The introduction of the quality outcomes framework (QOF) and subsequent changes to QOF indicators were followed by significant increases in the incidence of lifestyle advice. Treatment prescriptions decreased slightly over time, but were not associated with policy changes. Conclusions Despite secular trends that accord with UK guidance, many patients are still prescribed treatment for mild hypertension. Adequately powered studies are needed to determine if this is appropriate.

In order to help ensure that protocols submitted for review contain adequate information for protocol evaluation, ISAC have produced guidance on the content of protocols for research using CPRD data. This guidance is available on the CPRD website (www.cprd.com/ISAC). All protocols using CPRD data which are submitted for review by ISAC must contain information on all the areas detailed below. If a specific area required by ISAC is not applicable to your protocol, please provide the justification underneath the relevant heading.
The protocol section (next page) has pre-defined headings and the protocol must be written using these headings. Additional headings are not acceptable; however, supplementary information may be placed in one or more of the appendices providing this information is essential and an appropriate reference to it is made within the protocol. Unless very short, codes lists should be placed in an Appendix. Applications will be regarded as invalid and returned to the applicant if any of the headings below are missing or if additional sections are included.
Please note that ISAC will not consider any application where the protocol exceeds 12 pages (excluding sections A-F of the application form and annexes). Annexes should be kept to a minimum and contain only vital information that could not be provided in the main protocol section. A font-size of at least 12 should be used. Protocols not exceeding 15 pages would be acceptable if ISAC has required a resubmission where additional information is requested.
Please note, your protocol will not be reviewed by ISAC if it falls short of the above requirements. You are advised to speak to the Secretariat if you have any queries.

Voluntary registration of ISAC approved studies:
Epidemiological studies are increasingly being included in registries of research around the world, including those primarily set up for clinical trials. To increase awareness amongst researchers of ongoing research, ISAC encourages voluntary registration of epidemiological research conducted using MHRA databases. This will not replace information on ISAC approved protocols that may be published on the CPRD website. It is for the applicant to determine the most appropriate registry for their study. Please inform the ISAC secretariat that you have registered a protocol and provide the location.

A. Lay Summary (Max. 200 words)
Little is known about whether patients with moderately high blood pressure (mild hypertension) and a low risk of heart attack/stroke should be given medication to lower blood pressure. A definitive trial is unlikely to happen because it would be too expensive and require too many patients. Yet many guidelines encourage pharmacological treatment, and many patients receive it despite little evidence of benefit.
This study will examine the extent to which patients are prescribed pharmacological treatment for low risk mild hypertension, and explore the safety and effectiveness of this treatment using data from a large database of Primary Care records. Death from all causes will be compared in patients prescribed blood pressure lowering medication and those not prescribed medication, matched by characteristics such as age, sex, smoking status, prescription of other medications and history of disease. This approach will allow the effectiveness of pharmacological treatment to be examined in a large number of patients, followed-up for many years. This study will inform hypertension management guidelines, potentially leading to improved patient care.

B. Technical Summary (Max. 200 words)
Evidence to guide the pharmacological treatment of patients with uncomplicated (low cardiovascular disease risk), mild hypertension (blood pressure 140/90-159/99mmHg) is lacking. A definitive trial is unlikely due to the costs and numbers of patients required because of low event rates in this population. Yet many guidelines encourage treatment, and many patients are thought to receive it, despite little evidence of benefit.
This study will examine the extent to which patients are prescribed treatment for uncomplicated, mild hypertension in routine practice and the safety and efficacy of this treatment for using data from a large database of Primary Care records. The primary outcome of the study will be the rate of all-cause mortality in patients with uncomplicated mild hypertension prescribed antihypertensive treatment vs. those not prescribed therapy, matched using propensity scores estimating the likelihood of receiving treatment. Secondary outcomes will include the rate of cardiovascular morbidity/mortality, side effects to medication and rate of cancer in treated vs. non treated patients. These outcomes will be assessed using Cox proportional hazards modelling. The proportion of patients given lifestyle advice and/or pharmacological treatment for uncomplicated mild hypertension will also be estimated by year, and predictors of lifestyle advice/pharmacological treatment will be examined using logistic regression.

C. Objectives, Specific Aims and Rationale
Objective Examine the extent to which patients are prescribed pharmacological treatment for uncomplicated, mild hypertension in routine practice (descriptive objective) and whether such treatment is safe and effective and reducing the risk of death and cardiovascular disease (hypothesis testing).

Specific aims Descriptive objective
1. Establish the proportion of patients given lifestyle advice and/or pharmacological treatment for uncomplicated mild hypertension.

Examine the impact of the introduction of the Quality and Outcomes Framework (QOF) and NHS
health check programme on the proportion of patients given lifestyle advice and/or pharmacological treatment for uncomplicated mild hypertension. 3. Examine factors which predict the likelihood of being given lifestyle advice/pharmacological treatment for uncomplicated mild hypertension in routine practice.
Hypothesis testing 1. (Primary analysis) Examine the impact of antihypertensive prescription on all-cause mortality in patients with uncomplicated (i.e. low cardiovascular disease risk mild hypertension (i.e. sustained blood pressure between 140/90-159/99mmHg). 2. Examine the impact of antihypertensive prescription on cardiovascular disease morbidity and mortality (myocardial infarction [MI], non-MI acute coronary syndrome, stroke, heart failure) in patients with uncomplicated mild hypertension. 3. Examine the impact of antihypertensive prescription on cancer morbitdity and mortality in patients with uncomplicated mild hypertension. 4. Examine the impact of antihypertensive prescription on hospital admissions related to side effects to medication (hypotension, syncope, bradycardia, electrolyte abnormalities, injurious falls and acute kidney injury/renal failure) in patients with uncomplicated mild hypertension.

Rationale
As discussed above and below, many patients with mild hypertension receive treatment, despite little evidence of benefit. The present proposal describes two distinct analyses using data from the Clinical Practice Research Datalink (CPRD). The first will examine the extent to which guidelines for the management of uncomplicated mild hypertension are followed and how this has changed over time.
Factors which predict the likelihood of receiving antihypertensive treatment will be examined and this information will be used in a propensity score model to explore the safety and efficacy of antihypertensive treatment in these patients. These analyses will establish how many patients exist with uncomplicated mild hypertension, how it is currently managed in routine practice, and provide useful non-randomised data on treatment efficacy which will help guide future management decisions or inform whether future trial evidence is required.

D. Background
High blood pressure (hypertension) is a key risk factor for the development of cardiovascular disease 1 and a major cause of morbidity and mortality worldwide. 2 Classifications of hypertension are by definition arbitrary, but many guidelines recommend pharmacological treatment of blood pressure when it is sustained above 140/90mmHg, regardless of the underlying risk, end organ damage, diabetes or history of cardiovascular disease (table 1). [3][4][5][6][7][8] These recommendations have long been considered controversial, [9][10][11][12][13] particularly with regard to the treatment of uncomplicated (i.e. low cardiovascular disease risk) people with 'mild' hypertension (i.e. sustained blood pressure between 140/90-159/99mmHg).
There is little evidence to support initiation of pharmacological treatment in patients with uncomplicated mild hypertension and that which exists remains inconclusive. The recent SPRINT trial 14 did enrol a large number of patients with mild hypertension (6,174 patients had a baseline systolic blood pressure <145mmHg) and compared a strategy of intensive blood pressure lowering against usual care. They observed significant reductions in death and cardiovascular mortality with intensive treatment, but all enrolled patients had at least one cardiovascular risk factor (in addition to high blood pressure) and thus would not be considered 'uncomplicated'. A recent Cochrane review by Diao et al., 15 examined 8,912 patients from four clinical trials investigating the benefits of treating uncomplicated mild hypertension compared to placebo. They found no significant reduction in mortality with treatment (RR 0.85, 95% CI 0.63-1.15), nor was any reduction in coronary artery disease, stroke or total cardiovascular events observed in a subgroup of 7,080 patients. However, both the authors of that review and subsequent commentators 10 15 point to a lack of power in previous trials and meta-analyses to show significant results.
More recently, Sundström and colleagues 16 attempted to resolve this issue using individual patient data from the Blood Pressure Lowering Treatment Triallists Collaboration and aggregate data from trials identified in a systematic review. Whilst they found significant associations between blood pressure reductions with treatment and stroke or total death, the population studied still included a significant proportion of patients with diabetes, and thus would be considered high risk (not 'uncomplicated') under current clinical guidelines. 17 There are simply too few randomised data in low risk patients to make adequately powered treatment comparisons. Indeed, based on data from the Cochrane Review, a trial of treatment vs. placebo for mild hypertension would need to recruit over 100,000 patients, followed up for 5 years, to achieve statistical significance. The cost of funding such a trial mean it is unlikely to ever be conducted.
As a result, clinical guidelines make important recommendations based on 'expert opinion' alone. These guidelines are predictably contradictory, with those in the UK promoting lifestyle modification in low risk patients 17 whilst guidelines from Europe 6 and America 4 7 8 encourage prescription of drug therapy. The impact of these recommendations is not trivial. A recent study from the UK 18 suggested the costs of treating patients with uncomplicated mild hypertension could reach £229 million in the first year alone, although these estimates are based on regional data extrapolated to national population estimates. Against this background, a number of policy changes have occurred in the UK over the past 10 years which are likely to have affected general practitioner's treatment decisions. In particular, the introduction of the NHS health check in 2009 19 which will likely have increased the number of patients being diagnosed with uncomplicated mild hypertension, and QOF indicators briefly introduced in 2013 promoting treatment in this population. 20 The proposed study will establish the national prevalence of uncomplicated mild hypertension in Primary Care, how it is currently managed in routine practice, and provide useful non-randomised data on treatment efficacy which will help guide future management decisions or inform whether future trial evidence is required and feasible.

E. Study Type
Hypothesis testing.

F. Study Design
Longitudinal cohort study.

G. Sample Size
Using estimates of treatment effect size from the recent Cochrane Review by Diao et al., (risk ratio of 0.85 for all-cause mortality with treatment), and assuming a survival rate of 97.9% in the control group over 5 years, approximately 102,776 patients (51,388 in each group) and 1,600 events will be required to accurately define the relationship between antihypertensive treatment and all-cause mortality with 90% power and up to 20% withdrawal/crossover. The present analysis will utilise data from all available patients in the CPRD (with data linkage to ONS and HES) who fulfil the study eligibility criteria. We have examined the number of potentially eligible patients in a random sample of 20 general practices (413,279 patients) contributing to the CPRD database. Approximately 56,785 patients were identified with a diagnosis of hypertension, of which 8,791 (2.1%) fulfilled the study eligibility criteria (including 2,796 patients prescribed medication). The median length of follow-up in this population was 5.6 years. Extrapolating these data to the total number of patients within the CPRD database with linked data (10,022,053 patients) suggests a sample of 213,169 patients (including 67,803 patients on treatment) would be available.

H. Data Linkage Required (if applicable)
Data linkage to the ONS and Basic HES are required to define primary and secondary outcomes in the study. The ONS mortality register will be used to define the primary outcome of the study, all-cause mortality (and also censor follow-up at death). Specific linkages required will include data and ICD-10 coded cause of death (see appendix 2). Linkages to Basic Inpatient HES will be combined with data from the ONS to define secondary outcomes in the study, specifically death or admission to hospital with cardiovascular disease (defined previously), cancer or side effects to medication (hypotension, syncope, bradycardia, electrolyte abnormalities, injurious falls and acute kidney injury/renal failure) (see appendix 2). HES data will also be used to define patients eligibility for the study (e.g. previous stroke), and define the study population (e.g. ethnicity where unavailable in Primary Care records) -but only where it is recorded before the index date so as not to bias the completeness of data by those suffering an event during the study period. Data required from Basic Inpatient HES will include Primary diagnosis, secondary diagnosis, patient characteristics (e.g. sex, ethnicity), date of admission and date of discharge. All deaths and hospital admissions occurring after a patient's index date will be included. A linkage to the Index of Multiple Deprivation is required to acquire practice and patient level deciles of multiple deprivation. However, patients with missing IMD deprivation data will not be excluded, since these data will not be used to define any study outcomes.
Linkage to the MINAP dataset has not been requested due to limitations in the resources available for the study. However, this is not expected to a significant impact on the findings of the study, since: -A recent study of national database recording of non-fatal MI found that CPRD and HES alone capture 92% of all events. 21 -The MINAP dataset is unlikely to contribute any additional mortality data used for defining the primary outcome of this study which is not already captured in the ONS mortality register, CPRD or HES. -Even if the databases used do not capture all outcomes of interest, we do not see any reason to expect the likelihood of an outcome being recorded in MINAP alone to be confounded by treatment status.

I. Study Population
Individual patient data will be extracted from the medical records of all patients registered at general practices contributing to the CPRD in the UK. Patients will be entered into the study on the 'index date', and exit at the first of the following time points to occur: date at which the most recent linked data are available from the CPRD (March 2015), -Date of the most recent data upload from the practice to which a given patient is registered -Date at which a given patient transfers out of a registered CPRD practice -Date of death

Definition of the index date
The index will be defined as 12 months after the date of the third consecutive blood pressure reading between 140/90-159/99mmHg (readings must be taken within 12 months of each other) occurring after the study start date 01/01/1998. This date has been selected because it represents the date from which all relevant data linkages were first available (see below for details). Eligible patients will be those fulfilling the following criteria: -Linked general practice, Hospital Episodes Statistics (HES) and Office for National Statistics (ONS) mortality records. -Aged between 18-74 years.
-No Read code for stroke, myocardial infarction (MI), angina, coronary heart disease, peripheral vascular disease, heart failure, left ventricular hypertrophy, atrial fibrillation, diabetes or chronic kidney disease or family history of premature heart disease -Three consecutive blood pressure readings between 140/90-159/99mmHg (within 12 months of each other). -No record of any blood pressure lowering medication prescription in the 12 months prior to the third consecutive blood pressure reading between 140/90-159/99 mmHg. -Classified as a CPRD 'acceptable patient' -Registered to a CPRD practice classified as 'up-to-standard' Exclusions will be defined using read coded diagnoses listed in the appendix (see appendix 2 and 3).

J. Selection of comparison group(s) or controls Intervention group
The intervention in the present study will be sustained prescription of blood pressure lowering therapy to low risk patients with stage 1 hypertension. Patients considered in the treatment group will be those fulfilling the criteria above, who are initiated on blood pressure lowering treatment in the 12 months between the date of the third consecutive BP readings and the index date. The analysis will employ an intention-to-treat approach, so after initiation of treatment, patients will remain in the study/allocation until the last day of follow-up or until they are censored due to mortality, or cardiovascular morbidity.

Control group
Patients in the control group will be those fulfilling the eligibility criteria who were not initiated on blood pressure lowering treatment in the 12 months between the date of the third consecutive BP readings and the index date.Those prescribed therapy more than 12 months after the original index date will be included in the control group.
Patients eligible for the control and treatment groups will be matched using the index date and propensity scores, which reduce the bias introduced in observational studies because patients are not randomised to their treatment allocation. Propensity scores indicate the likelihood a patient will be prescribed treatment on the basis of their known (pre-treatment) characteristics and other known information which might influence the decision to treat. This method is advantageous because it allows patients to be matched on multiple characteristics using a single variable (the propensity score), although this means associations between specific covariates and outcomes cannot be assessed.

K. Exposures, Outcomes and Covariates Exposures
The exposure variable (blood pressure lowering treatment) in the primary analysis in this study will be defined as a prescription of an angiotensin-converting-enzyme inhibitor, angiotensin-II-receptor blocker, calcium channel blocker, thiazide or thiazide-like diuretic, beta-blocker, alpha-blocker or any other antihypertensive listed in the British National Formulary (see appendix 1). Reads codes for being offered lifestyle advice will also be extracted (see appendix 1).

Outcomes
The main aim of this study is to examine the safety and efficacy of treatment in uncomplicated, mild hypertensives. All-cause mortality has been chosen as the primary outcome for this analysis because it permits assessment of the possible benefits of treatment (reduction in death from cardiovascular disease) and the safety of treatment (in terms of risk of non-cardiovascular death) combined in a single outcome. Secondary outcomes will include: Descriptive secondary outcomes -The proportion of patients given lifestyle advice and/or pharmacological treatment for uncomplicated mild hypertension. -The likelihood of patients being given lifestyle advice and/or pharmacological treatment for uncomplicated mild hypertension before and after the introduction of QOF and the NHS health check programme. -Factors which predict the likelihood of being given lifestyle advice/ pharmacological treatment Hypothesis testing secondary outcomes -death or hospitalisation from major cardiovascular events (MI, non-MI acute coronary syndrome, stroke, heart failure or death from cardiovascular disease) using the same definition as that used in the SPRINT trial 14 -death or hospitalisation from stroke -death or hospitalisation from MI or non-MI acute coronary syndrome -death or hospitalisation from cancer -hospitalisation with suspected side effects to medication (hypotension, syncope, bradycardia, electrolyte abnormalities, injurious falls and acute kidney injury/renal failure) Outcomes will be captured from hospital admissions (including date of admission and discharge and primary diagnosis), read coded diagnoses in the CPRD and death certificates occurring after the index date using linked data from the Basic Inpatient HES and ONS mortality register (for codes see appendix 2). readings (in the preceeding 12 months), non-cardiovascular related co-morbidities (rheumatoid arthritis, high total cholesterol [>5mmol/l]) and all statin/antiplatelet medications prescribed will be extracted from the medical records of eligible patients. For subgroup analyses taking into account a patient's apparent adherence to treatment, 'daily equivalent dose' data will be extracted from CPRD and the medication possession ratio (percentage of days when medication was available [MPR]) will be estimated from the sum of tablets prescribed, divided by the number of days of follow-up.

Covariates
L. Data/ Statistical analysis The primary analysis will use a propensity score-matched longitudinal cohort study design to examine the safety and efficacy of antihypertensive treatment. Secondary analyses will include an interrupted timeseries analysis of prescriptions and lifestyle advice given to patients diagnosed with uncomplicated mild hypertension, comparing before and after the introduction of the NHS health check programme.

Descriptive objectives Aim 1 -Descriptive analyses
Descriptive statistics will be used to estimate the proportion of patients being offered lifestyle advice or therapy following a diagnosis of uncomplicated mild hypertension (aim 1). Estimates will be adjusted for age and sex and results will be stratified by year of diagnosis.

Aim 2 -Interrupted time-series analysis
The impact of the introduction of the NHS health check programme (April 2009) and changes to QOF indicators (April 2013) and on prescription rates will be examined using an interrupted time-series analysis. Rates of pharmacological treatment and or lifestyle advice will be estimated before and after these specific policy changes and the overall level and slope of the rates in the pre and post policy change periods will be compared using segmented regression.

Aim 3 -Propensity score model
Predictors of lifestyle advice and antihypertensive medication prescription following a diagnosis of uncomplicated mild hypertension (defined as three consecutive blood pressure readings between 140/90-159/99mmHg within a 12 month period) will be explored in a logistic regression model examining the association between the outcome of interest and patients and practice characteristics: age, sex, ethnicity, blood pressure, deprivation (practice level), smoking status, BMI, number of existing antihypertensive medication prescriptions, presence of statin or antiplatelet prescriptions, the presence of rheumatoid arthritis or hypercholesterolemia, calendar year of the index date and the general practice to which the patient is registered. The model examining predictors of pharmacological treatment will be used to generate a propensity score matched cohort for analysis of the primary outcome. The success of propensity score matching will be examined by comparing the characteristics patients in each group pre-and postmatching using chi-squared tests and t-tests. To assess the reduction in bias achieved by propensity score matching, the proportional reduction in the difference between each patient characteristic in the treatment and control groups will estimated and compared pre-and post-matching.

Hypothesis testing Aim 4 -Primary analysis
Using an intention-to-treat approach, the efficacy of antihypertensive treatment will be examined with Cox proportional hazards modelling, comparing all-cause mortality in those prescribed (any) antihypertensive treatment vs. those not prescribed treatment. Where hazards are not proportional, non-parametric methods such as inverse probability weights will be considered. Kaplan-Meier plots will be produced to display the cumulative incidence of all-cause mortality in the treatment group compared to the control group. Hazard ratios will be generated and adjusted by any other factors, unbalanced at baseline, not already included in the initial propensity score matching.

Aims 5-7 -Secondary analyses
Where appropriate (and sufficiently powered to detect an effect), hazard ratios will be calculated using Cox proportional hazards modelling, examining the association between treatment status and secondary endpoints in the study: major cardiovascular disease event, hospitalisation and/or death from stroke, or MI or non-MI acute coronary syndrome and pre-specified side effects to medication.

Subgroup analyses
Subgroup analyses will be conducted to examine the impact of patient adherence to medication on the effectiveness of treatment. We will stratify our analysis by quantiles of medication possession ratio and the treatment effects will be compared between groups. Other pre-specified subgroup analyses will include comparison of the treatment effect stratified by age (defined according to 10 year age bands, where possible), antihypertensive drug class (if sufficiently powered) and in males vs. females.

M. Plan for addressing confounding
The validity of propensity score matching will be examined using 'negative controls': the impact of antihypertensive treatment on an outcome not known to be affected by such treatment. Here, the impact of antihypertensive therapy on hospitalisation and/or death from cancer will be studied as the negative control (aim 6). If treatment with antihypertensives has a significant impact on this outcome, we can conclude there is something missing in the propensity score (i.e. an unmeasurable factor of propensity for treatment such as being generally unwell or an unhealthy lifestyle) causing an imbalance between the treatment and control groups, rather than a true treatment effect.

N. Plan for addressing missing data
There is potential for missing data in this study, particularly with variables such as ethnicity which are recorded with varying degrees of accuracy in routine practice. Because this analysis is focused on treatment of uncomplicated mild hypertension, accurate selection of the sample population is important and therefore any patients with insufficient data available to define their blood pressure status at baseline (index date) will be excluded. Where there is no record of blood pressure lowering, statin or antiplatelet treatment, it will be assumed the patients were not exposed to blood pressure lowering, statin or antiplatelet treatment.
Patient eligibility includes only using 'acceptable patients' in the analysis, and therefore there is no need to impute age and sex. Where there is no record of smoking history or alcohol consumption, patients will be assumed to be non-smokers and non-drinkers. Likewise, those with no record of rheumatoid arthritis or hypercholesterolemia will be assumed to have no history of these conditions. All other covariates used in the propensity score model will be imputed using multiple imputation. Ethnicity and IMD will be treated as categorical variable in the imputation model. BMI will be treated as a continuous variable.
The primary analysis examining the efficacy of treatment will be conducted by intention-to-treat, and patients moving practice after the index date (and therefore being lost to follow-up) will be censored at the point at which they are no longer active in the database.

O. Limitations of the study design, data sources and analytical methods
This study will use an observational cohort design, and as such, there is an inherent selection bias of patients in both intervention and control groups. The impact of this bias will be limited by propensity score matching but such an approach assumes that all factors influencing the propensity for medication prescription have been measured and accounted for within the score. In the present study, we will examine the validity of this assumption by studying the impact of treatment on a negative control: hospitalisation and/or death from cancer. We will conclude the assumption has been met if antihypertensive treatment is shown to have no impact on cancer outcomes, as would be expected from previous trial evidence.
Previous randomised studies examining the efficacy of treatment for uncomplicated stage 1 hypertension have been underpowered to demonstrate a true treatment effect. This study will use data from the largest linked primary care database in the world with a potential sample population of approximately 210,000 patients, sufficient to power the study for the primary and subgroup analyses. In pre-specified subgroup analyses, we will examine the impact of potential confounding factors such as patient adherence to medication.

Strategy for dealing with potential errors resulting from multiple testing
This study includes one descriptive objective to examine the extent to which patients are prescribed pharmacological treatment for uncomplicated, mild hypertension and one hypothesis testing objective to examine whether such treatment is safe and effective at reducing the risk of death and cardiovascular disease. To avoid potential errors arising from multiple testing, a primary outcome has been clearly defined as the impact of antihypertensive treatment on all-cause mortality. Secondary outcomes will examine the impact of antihypertensive treatment on major cardiovascular events (composite endpoint), death or hospitalisation from stroke, death or hospitalisation from MI or non-MI acute coronary syndrome, death or hospitalisation from cancer or hospitalisation with suspected side effects to medication (hypotension, syncope, bradycardia, electrolyte abnormalities, injurious falls and acute kidney injury/renal failure). The primary outcome will be given priority in the final analysis write-up and any related reports and presentations.
Primary and secondary outcomes will be defined using ONS, inpatient HES and read-coded primary care data. The accuracy of such outcome data has been examined previously in patients with acute myocardial infarction 21 and whilst recording of risk factor and co-morbid information was consistent across primary care, hospital admissions and disease registry records, the crude incidence of acute myocardial infarction was underestimated by up to 50% if only one data source was used, compared with using all three sources. The use of linked CPRD data in the present study is therefore crucial to ensure outcome data are ascertained accurately.

P. Patient or user group involvement (if applicable)
We have not carried out specific PPI for this particular project, although PPI work we have recently conducted for similar studies is relevant. In particular, we have conducted focus groups to examine patient views on blood pressure treatment and primary prevention of cardiovascular disease. Specific issues highlighted included that healthy older individuals, like many others, sometimes take their state of good health for granted and few considered that they might be at risk of strokes or heart attacks. The haphazard nature of blood pressure/cholesterol checks by their GP seemed confusing and many opted for an attitude of fatalism over future health. They were often informed by health experiences of friends and popular media opinion and were wary of the potential side-effects of medication (hence preferred not to take it).
These issues are pertinent in the population of interest, and evidence provided by the present analysis will support shared decision making between patients and doctors as well as informing future cardiovascular disease prevention programmes and policy.
Q. Plans for disseminating and communicating study results, including the presence or absence of any restrictions on the extent and timing of publication We expect at least two publications arising from this research: one describing the prevalence of pharmacological treatment and lifestyle advice for uncomplicated mild hypertension, and how this has changed over time. The second will explore the safety and efficacy of treatment for mild hypertension. A third publication may be possible exploring the efficacy of treatment in predefined sub-groups in detail. Findings will be presented at scientific meetings and published in journals at the earliest possible convenience. We will endeavour to publish our study findings in lay format on our University website and through lay and social media where appropriate.