Stepwise mass screening for atrial fibrillation using N-terminal pro b-type natriuretic peptide: the STROKESTOP II study design

Aim

Atrial fibrillation (AF) is the most prevalent clinical arrhythmia and a major risk factor for ischaemic stroke. Treatment with oral anticoagulants (OACs) reduces the risk of stroke by two thirds in AF patients with risk factors. Due to its often paroxysmal and asymptomatic presentation, AF is sometimes challenging to diagnose. So far, AF screening studies have applied opportunistic or systematic screening, most often using a single 12-lead electrocardiogram (ECG) recording or ambulatory ECG. We hypothesise that the biomarker N-terminal pro b-type natriuretic peptide (NT-proBNP) is a valuable adjunct in population based AF screening.

Methods

We are conducting a randomized population-based study on AF screening using ambulatory ECG recording where the decision to use prolonged intermittent ECG recording is directed by NT-proBNP levels, the STROKESTOP II trial. The entire population of inhabitants 75 or 76 years of age (n = 28 712) in the capital region of Sweden will be randomized 1:1 to intervention or control group. In the intervention group NT-proBNP will be analysed in all without previously known AF. Those with NT-proBNP ≤ 125 pg/L will make a single one lead ECG recording, participants with NTproBNP ≥ 125 np/L will be instructed to record ECG for 30 s at least twice daily for 2 weeks with a handheld ambulatory ECG recorder. Participants with newly diagnosed or undertreated AF will be referred to a cardiologist and offered OAC treatment. Primary endpoint is incidence of stroke or systemic embolus, during a 5 year follow-up period in the control group vs the group invited to screening.

Introduction

Atrial fibrillation (AF) is the most prevalent clinical arrhythmia and a major risk factor for ischaemic stroke. A considerable share of patients has no symptoms from AF, often referred to as silent AF.¹ The size of this share is not very well known and estimates vary dependent on the extent of electrocardiogram (ECG) monitoring.^1–3 Stroke risk in connection with AF is not dependent on the presence of symptoms.

Although some contemporary large AF studies have reported a possible increase in stroke risk in patients with permanent AF in comparison to paroxysmal AF, there is still insufficient data at hand to support different treatment thresholds for different AF patterns. Brief and most often silent episodes of paroxysmal AF are associated with increased stroke risk, mainly reported from pacemaker studies.^2,4,5 Other risk factors, of which the most important are included in the risk stratification scheme CHA₂DS₂-VASc [Congestive heart failure, Hypertension, Age > 75 years (doubled), Diabetes Mellitus, Stroke (doubled), Vascular disease, Age >65 years, Sex category (female gender)], decide stroke risk in AF.⁶

Atrial fibrillation is an important and prevalent health problem that can be asymptomatic and as there is highly effective treatment for stroke reduction, AF seems to fulfil most criteria for population screening.⁷ Indeed the latest European Society of Cardiology guidelines recommend screening for AF using pulse palpation in individuals ≥65 years of age.⁸ Screening for AF has been undertaken with several different methods and in different populations svenn. Very few population based screening studies have so far utilized ambulatory ECG technology which seem necessary when paroxysms of AF are to be detected.⁹

We have previously reported on the yield and feasibility of using intermittent handheld ECG recordings in population-based AF screening for stroke reduction purposes.^3,10,11 In these studies, screening with handheld ECG increased the AF prevalence in a 75-year old population with at least 30%. Oral anticoagulant (OAC) treatment acceptability among diagnosed participants was high. However, the ECG screening procedure used in our previous studies required 2 weeks of intermittent ECG recording from all participants and all those recordings were interpreted manually.

The biomarker N-terminal pro b-type natriuretic peptide (NT-proBNP) is a validated tool for heart failure diagnostics. Increased levels of NT-proBNP are associated with increased myocardial wall stress and the test is used to diagnose or exclude heart failure. However, NT-proBNP levels are also elevated in patients with AF and NT-proBNP elevation can predict development of AF.^12,13 In patients with known AF, NT-proBNP levels seem to be in proportion to stroke risk.¹⁴

In the STROKESTOP I trial, 886 of the 7173 participants were studied with regard to NT-proBNP levels in connection to their handheld ECG screening. Levels of NT-proBNP were significantly higher among participants diagnosed AF. Using receiver operating characteristic regression (ROC-curves) to determine a cut-off level for NT-proBNP for AF screening, a value of 125 ng/L was considered as an optimal level.¹⁵

We are now launching a randomized study in the first quarter 2016 using NT-proBNP followed by handheld ECG in a stepwise, population based randomized AF screening study aiming at stroke prevention through treatment with OAC in patients diagnosed with AF and through initiating treatment in patients already diagnosed with AF but undertreated with regard to OAC.

Hypothesis

Population-based screening for AF in 75-year-old individuals using NT-proBNP and handheld ECG recordings in a stepwise screening procedure will reduce stroke incidence in the screened population cost-efficiently.

Design

Randomized, controlled non-blinded interventional cohort study.

Ethics

The study has been approved by the regional Ethics Committee of Stockholm (2015/2097-31/1). The study will be conducted in compliance with the Declaration of Helsinki. All participants will receive written information and thereafter sign informed consent before inclusion.

Methods

Study population

All individuals born in 1940 or 1941 and living in the Stockholm Region (n = 28 712). Individuals are identified by their 10-digit personal identification number assigned to all citizens in Sweden.

Randomization

Sample size calculation

The power calculation is based on the following assumptions; the stroke rate in the control group is 5.5% in 5 years (data based on Swedish National registries). In the screening group the power calculation is based on an uptake between 50 and 70%, and the stroke frequency in the group not attending screening despite invitation is 5.5%, and the group screened where no AF is found varies from 2 to 4%, and the group screened where AF is found is 10% (30% risk of stroke during a 5-year period, reduced by 70% with oral anticoagulant therapy). If participation is 50% and 5-year stroke risk among screened participants without AF is 3%, there is a 98% power to detect a significant difference between control and intervention groups with estimated annual stroke risk 5.5% vs. 4.4% in these groups with intention to treat analysis.

Screening procedure

Individuals in the intervention group will receive an invitation to participate by mail. Participants not responding will receive two reminders by mail. Approximately 150 individuals will be invited weekly to one of six screening centres run by Karolinska Trial Alliance.

At the screening centre, participants are informed about the study orally and in writing. All participants have to sign informed consent before entering the study. After completing written informed consent, participants are asked to report their medical history with regard to AF, stroke, transient ischaemic attack, systemic emboli, heart failure, hypertension, diabetes, vascular disease, and palpitation symptoms. Data on height and weight will be collected. Participants also have to report if they are on any antithrombotic treatment. A pulse palpation for 30 s will be made and recorded as regular or irregular. Blood pressure using an automated sphygmomanometer (Boso medicus uno, Bosch + Sohn GmbH, Jungingen, Germany) with an alarm for irregular pulse will be measured.

All participants with a blood pressure exceeding 140 mm Hg systolic and/or 90 mm diastolic will be given a referral note to their general practitioner’s office.

Participants with a prior diagnosis of AF will be asked if they are on treatment with anticoagulants. If they are not a referral to a cardiologist will be made to ensure treatment according to guidelines.

Individuals without known AF will have blood samples drawn and analysed with regard to NT-proBNP using point-of-care analysis. All NT-proBNP data will be dichotomized into two groups, below 125 pg/L, ‘low-risk’ and above 125 pg/L, ‘high-risk’. Blood samples will also be stored in biobank for future analysis of biomarkers within coagulation, inflammation, and myocardial stress.

All individuals with NT-proBNP ≥ 125 pg/L and without known AF will be instructed to make intermittent ECG recordings with a Zenicor II handheld device (Zenicor Medical Systems, Stockholm, Sweden) at least twice daily for 2 weeks. Participants who make this extended recording are instructed at their index visit to return the Zenicor device after 2 weeks and that they will be contacted by study staff if there were any arrhythmias in need of further investigation or treatment.

Individuals with a NT-proBNP ≤125 pg/L will do one initial 1-lead ECG with the Zenicor II handheld device, and if no AF is detected on this single ECG recording they will not undergo further ECG screening.

Individuals where new AF is discovered on index ECG or during intermittent ECG recordings will be offered a structured follow-up by a cardiologist, initiating OAC treatment if appropriate.

All participants with NTproBNP ≥ 900 pg/L and no previous diagnose of congestive heart failure will be offered an echocardiographic and clinical examination in order to diagnose subclinical structural heart disease.

Electrocardiogram monitoring

All participants will make at least one ECG recording using the Zenicor handheld ECG on their index visit. Participants without known AF and with NT-proBNP ≥ 125 pg/L will be instructed to make at least two daily ECG recordings during 2 weeks with the handheld ECG recorder. The Zenicor ECG device records lead I and automatically transmits the encrypted recording to a database to which access is password protected. The Zenicor handheld ECG recorder has been validated and used in several previous AF screening studies.^3,10,16–18

Definition of atrial fibrillation

Atrial fibrillation will be defined as at least one episode of completely irregular rhythm with no organized or regular atrial activity and a duration of 30 s. Participants with episodes of irregular supraventricular rhythm suggestive of AF but with a duration less than 30 s but at least four beats will be subject to follow-up within a substudy.

Inclusion period

First quarter 2016–Third quarter 2017.

Duration of follow-up

The entire screening population, including participants and non-participants in the intervention group and the control group will be subject to long-term follow up 5 years after screening. Start of follow-up for participants will be the date of the screening visit (index date), for non-participants, a random date for inclusion will be assigned to each individual in such a way that mean follow-up and distribution of days from the inclusion to end date of the observation period will be the same as in the participating cohort.

An interim analysis will be made 3 years after screening. Data on endpoints and prescriptions of OAC will be collected from the National Patient Register and from the National Prescription Drugs register.

Primary outcome

Primary outcome is ischaemic stroke or systemic embolism in the group invited to screening compared to the control group.

Secondary outcome is composite of ischaemic stroke, systemic embolism, and mortality in the group invited to screening compared to the control group.

Secondary objectives

1. To study socioeconomic status among participants and non-participants in the control group, reported as education level, immigration status, marital status, and disposable income.

2. To report echocardiographic data in 100 participants with no previous diagnosis of heart failure and NT-proBNP >125 ng/L and NT-proBNP < 900 ng/L expressed as(EF; E/A ratio; E/e’ ratio; left ventricular mass with eccentricity index; left atrial volume and dimensions; systolic and diastolic strain; Kovacs M; Tei index; duration of A wave velocity in pulmonary vein flow in relation to the A wave velocity in transmittal flow; non-invasive arterial elastance and LV stiffness; LV chamber level (Ees) as well at the myocardium (circumferential midwall fractional shortening (mFS) and the pre-load recruitable stroke work by use of the 2-shell method of Shimizu.

3. An algorithm for digital automated detection of ECG recordings with normal sinus rhythm in the Zenicor ECG system has been developed by Cardiolund Research AB. A real-life validation of this algorithm will be done within STROKESTOP II. Validation of an automated detection of ECG recordings will be made by calculation of sensitivity and specificity.

4. The relationship between participant-reported symptoms and newly discovered AF will be studied as diagnostic yield among asymptomatic and symptomatic participants.

5. The value of pulse palpation as compared with ECG detection of AF will be studied as sensitivity and specificity as compared with ECG data.

6. The prognostic implication of short supraventricular bursts consisting of at least four beats will be studied as the share of patients developing clinical evident AF and the share diagnosed with silent AF during 7 days of continuous ECG event recording.

Cost effectiveness

Our previous study STROKESTOP I has been assessed with regard to cost effectiveness.¹⁹ This analysis concluded that eight fewer strokes, 11 more life-years and 12 more quality-adjusted life years (QALYs) would be gained by screening 1000 individuals. A similar cost effectiveness analysis will be made after the closure of the screening phase of STROKESTOP II.

Detection of events and comorbidity

Information on previous illness, comorbidities, and events during follow-up for the entire study cohort will be collected from the Swedish National Patient Register using International Classification of Disease, 10th revision (ICD-10) going back to 1997. Data will be collected according to Table 1. Data from this register will also be used in calculating individual risk scores for stroke and bleeding according to the risk stratification systems of CHADS₂, CHA₂DS₂-VASc, and HASBLED. The Swedish National Patient Register has been previously validated with very good coherence between clinical course and discharge diagnoses.²⁰

Socioeconomic data will be collected from Statistics Sweden and will comprise of the following variables: marital status, immigrant status, educational level, and disposable income.

Antithrombotic treatment

During follow-up, data from the National Drug Register will be collected. This register contains data from all pharmacies in Sweden. Individual data on collection date, dosage, and quantity on all dispensed prescriptions are available since 2005. The dispensing of OAC for both intervention and control group will thus be available. Data from the National Drug Register will be collected according to Table 2.

Exposure to anticoagulants

Days of exposure for oral anticoagulants is counted from index date.

Exposure to dabigatran, rivaroxaban, and apixaban will be estimated from the expected number of days the dispensed quantity would last if taken as intended. The strength of the first dispenzation of the drug will define dosage (e.g. dabigatran 110 mg means 110 mg bid etc).

Warfarin exposure will be approximated in the following way: each prescription is assumed to cover the consumption for 3 months (Swedish regulations state that each prescription should be for the estimated consumption during 3 months). If a new purchase of warfarin is made within 90 days of the foregoing, treatment is considered as continuous. If the purchase occurs after 90 days, but before 180 days a grace period will be applied (e.g. the patient may have a very low dose) and treatment will be considered as continuous. If there is no purchase within 180 days, warfarin exposure will be considered as terminated 90 days after the previous purchase. If there is another purchase after day 180, a new exposure period will start.

Statistical analysis

Primary endpoint (rate of stroke and systemic emboli in intervention and control groups in 5 years) will be analysed on intention-to-treat basis. However, an analysis on treatment basis (including only participants) will also be made. For primary endpoints Cox-regression will be performed, adjusted for potential confounders such as gender and presence of cardiovascular risk factors included in the CHA2DS2-VASc score, as well as use of oral anticoagulants. Kaplan–Meier survival analysis will be used in long-term analysis of primary endpoint. Crude AF detection rates and rates of anticoagulation treatment initiation will be reported.

Clinical relevance

Atrial fibrillation is a gross contributor to stroke risk. Despite the advent of non-vitamin K anticoagulants, there is still a widespread underuse of anticoagulants among patients with AF, and the share of patients with silent AF add to the share of undertreated patients since most of them will remain undiagnosed. Screening for AF in risk groups has the potential of reducing stroke incidence, and we believe that the proposed screening procedure will make AF screening simpler, more targeted and more cost effective.

The screening procedure validated in this study is expected to be simpler since one third of the participants with the lowest risk of incident AF do not have to make ambulatory ECG recordings, which also diminishes the burden of ECG interpretation. The cost of a point-of-care venous blood test will be added to screening expenditures. However, we calculate that there will be a cost reduction of 80 Euro per participants not making the 2-week ECG recording, which is a cost reduction of 35%. Participants with NT-proBNP < 125 ng/L (low-risk group) who has paroxysmal AF will have a lower stroke risk than participants in the high-risk group which will make the consequences of a missed AF diagnosis smaller.

Furthermore, the automated ECG interpretation algorithm is expected to make a significant reduction of workload.

Today, there are numerous devices available over the counter for self-assessment of the heart rhythm, of which some record ECG and some record pulse. However, most of these devices do not come with an interpretation service or availability of other health care services such as prescription of treatment. Therefore, we believe that AF screening will have the best yield when available to the entire target population and undertaken in a structured context addressing all steps in the screening chain, not only ECG detection.

Funding

The STROKESTOP II trial is supported by Roche Diagnostics, Carl Bennet AB, The Swedish Heart and Lung Foundation, The Stockholm County Council and the Swedish Order of St John.

Conflict of interest: J.E. has received consultancy fees from Sanofi and Pfizer, lecture fees from Astra Zeneca, Boehringer-Ingelheim, Medtronic and Bristol-Myers Squibb, travel expenses from Boehringer-Ingelheim and Sanofi. E.S. has received lecture fees from MSD, Pfizer, Boehringer-Ingelheim and Sanofi, and a research grant from Boehringer-Ingelheim. L.F. received research grants and/or lecture fees from Boehringer-Ingelheim, Sanofi-Aventis, Bristol-Myers-Squibb, Pfizer, Bayer, and St Jude Medical. F.A. has received lecture fees from Bristol-Myers-Squibb, Boehringer-Ingelheim and Bayer. V.F. reports grants and study collaboration from Medtronic, and study collaboration from St Jude Medical. M.R. reports consultancy fees/research grants/lecture honoraria from Bayer, Pfizer, Boehringer-Ingelheim, Bristol Myers Squibb, Medtronic, St JudeMedical and Zenicor Inc. i. K.K.G. and T.F. report no conflicts of interest.

References