Streamlined mail-based methods for large randomised trials: lessons learnt from the ASCEND study
Reliable assessment of the effects of an intervention usually requires large randomised trials but such studies are becoming increasingly complex and costly to run. ‘Streamlined’ trials are needed in which every aspect of the trial design and conduct is simplified, retaining only those elements needed to answer the research question and ensure the safety of the individual participants. In this review we discuss how the trial ‘A Study of Cardiovascular Events iN Diabetes’ (ASCEND) was streamlined. The study included a two-by-two factorial design: it assessed the effects of low-dose aspirin and, separately, supplementation with n-3 fatty acids on serious vascular events in 15,480 people with diabetes but no overt cardiovascular disease. Other key streamlined design features, such as mail-based recruitment and follow-up, mainly by post, with no in-person visits and use of a run-in period, are also described. We go on to discuss the success of the study and other studies that have employed a similar mail-based approach, and the type of clinical trials that are suitable for mail-based design. Finally, we consider the limitations of the study, and how these could be circumvented in future studies. ASCEND randomised large numbers of eligible participants, achieved good adherence rates and almost complete follow-up at a fraction of the cost of traditional clinic-based trials. Such studies are necessary if researchers are to address the important clinical questions most relevant to improving health.
KeywordsAspirin Cardiovascular Fatty acid Methodology Randomised Review Streamlined
A Study of Cardiovascular Events iN Diabetes
Electronic health record
Reliable assessment of treatment effects requires randomised trials with unbiased assessment of large numbers of outcomes, complete follow-up of all randomised participants and adequate adherence to the allocated treatment. However, randomised studies often include complex procedures that can hinder achievement of these aims. Large trials assessing serious clinical outcomes (Phase 3 or 4 studies) have become so costly that many important research questions go unanswered, while some potentially useful therapeutic agents are not brought to clinical testing [1, 2]. Some investigators have suggested that observational analyses of large, complex, routinely collected datasets can be used to assess the effects of treatments without randomisation. However, such methods can lead to erroneous estimates of treatment effects since differences exist between those treated and not treated that are not captured within the dataset, and the bias resulting from these confounding factors can be as large, if not larger, than the treatment effect itself .
To obtain randomised trial evidence at feasible costs, ‘streamlined’ methods are needed. In such trials, every aspect of the study design and procedures is simplified as much as possible, retaining only the elements needed to answer the central research question and keep the individual participants safe . Strategies to improve efficiency in trials include conduct in a small number of countries, simple eligibility criteria, limited data collection and the adoption of risk-based monitoring [5, 6]. However, despite these measures, the cost of running large streamlined clinic-based outcome studies  remains high and such studies can often only be carried out in partnership with the pharmaceutical industry. ‘Registry trials’ in which participants are identified, randomised and followed-up within an existing registry can be highly cost-effective [8, 9] and have been used in a number of settings, including percutaneous coronary procedures [10, 11, 12] and acute myocardial infarction . However, such designs are not suitable or possible for many research questions, and other approaches to streamlining trials are needed.
Key streamlined design features of the ASCEND trial
Two-by-two factorial design enabled two questions to be answered
Recruitment and assessment
Clinic-based with multiple sites
No clinics. Mail-based recruitment and follow-up, mainly by post
Conducted within the UK
Under-used in many trials
Two month placebo run-in managed by postal mail improved later adherence
Complex criteria requiring blood or other tests and review of medical notes
Simple criteria that can be determined from participant self-report
Complex assessments, including collection of biological samples in all participants
Limited data collection
Optional blood and urine sampling
Double adjudication by Clinical Outcomes Committee
Single adjudication using pragmatic procedures
Seldom undertaken and rarely possible in international trials
Low-cost post-trial follow-up through linkage with routine healthcare records
Streamlined design features of ASCEND
Two-by-two factorial design
By using a two-by-two factorial design, ASCEND was able to assess the effects of two independent treatments on serious vascular events within the same population [14, 15]. Provided the effects of each treatment are moderate, such factorial designs have little impact on the statistical power of each analysis  and bring obvious efficiency savings by addressing two questions in one trial. It is frequently suggested that they are vulnerable to an interaction between the two treatments under study; however, if real, such interactions would be important to detect, and such a design is the most efficient way to do so. Despite the use of factorial designs in clinical trials for decades, this valuable method of increasing efficiency in randomised studies remains under-used.
Mail-based trial design with no in-person visits
Large randomised trials conducted by mail
Study (year completed)
Aspirin, β carotene
Aspirin, vitamin E
Vitamin C, vitamin E, β carotene
Aspirin, n-3 FA
Female health professionals
Women with, or at high risk of CVD
Mean or median follow-up (years)
Characteristics of clinical trials suitable for mail-based design
Mail-based design suitable
Mail-based design unsuitable
National healthcare system with access to centralised medical records available
Local healthcare systems comprising multiple providers with no centralised medical records
Eligibility can be identified from routine healthcare records and/or patient report (e.g. prior stroke or myocardial infarction)
Eligibility requires detailed review of medical notes (e.g. heart failure with reduced ejection fraction) or recent blood results (e.g. kidney function)
Suitable for administration in remote settings (e.g. oral medication)
Requires administration in a healthcare setting (e.g. devices) or in-person participant education (e.g. injectable therapies)
No requirement for regular blood or other tests to monitor safety
Requires regular blood or other tests to monitor safety
Can be identified from routine healthcare records or self-report (e.g. cardiovascular events)
Require in-person assessment (e.g. measurement of physical function)
ASCEND was conducted in a single country, which reduced cost and complexity. International collaboration is often necessary when recruiting individuals with a rare disease, but many trials in common diseases involve multiple countries. This is often driven by requirements of regulatory authorities to obtain trial data within particular countries, marketing strategies for new drugs and a perception that results from international trials are more generalisable than those from trials conducted in a single country. However, the proportional effects of treatment are not likely to vary by country unless there are substantial differences in disease pathophysiology or healthcare provision that modify the effects of the treatment. Obtaining the necessary regulatory and ethics committee approvals for international studies adds substantial complexity and delay , and operational procedures must be arranged for each country.
Furthermore, by recruiting participants entirely within the UK, low-cost long-term follow-up can be conducted via linkage to electronic health records (EHRs). This is particularly relevant for the aspirin vs placebo comparison, given the recently raised hypothesis that aspirin may reduce long-term cancer risk . In ASCEND, pre-specified cancer analyses will be undertaken in 2022 and 2027 based on cancers identified from the UK Cancer Registry and from coded hospitalisation data.
Complex eligibility criteria, frequently requiring manual review of the participant’s medical record and trial-specific blood or other tests, such as imaging, can have unhelpful consequences. The procedures required to implement such protocols can be time consuming and expensive and participants be may inappropriately excluded, limiting generalisability. To be eligible for ASCEND, potential participants had to fulfil just six criteria. They had to be at least 40 years of age at the time the screening questionnaire was sent, have a clinical diagnosis of diabetes mellitus, no clear indication for aspirin (e.g. diagnosed occlusive arterial disease) and no clear contraindication to aspirin (e.g. recent gastrointestinal haemorrhage or peptic ulcer; use of anti-coagulant; or history of aspirin allergy). Participants had to be free of severe comorbid disease that might limit their ability to remain in a long-term study, and both the participant and their doctor had to be uncertain about whether antiplatelet or n-3 fatty acid therapy conferred worthwhile benefit.
Maintenance of adherence to the study intervention in randomised trials is critical to ensure adequate statistical power to detect plausible moderate effects. Although ethical standards require trial participants to be free to stop any aspect of participation at any time, every effort should be made to randomise only those participants who are likely to remain adherent for the duration of the study. A run-in period prior to randomisation during which treatment (active or placebo, depending on the circumstance) is taken provides a useful method of identifying participants prior to randomisation who are likely to be non-adherent so that they can be excluded. A run-in period may be particularly valuable in mail-based studies since research staff are unable to make a detailed face-to-face assessment of likely long-term adherence. In ASCEND, more than 40% of eligible individuals entering the placebo run-in period were not randomised, predominantly because they were unwilling to continue in the trial or failed to respond to questionnaires. If these individuals had been randomised immediately after assessment of eligibility, the resulting reduction in treatment adherence would have led to catastrophic loss of statistical power .
Limited baseline assessments
In many trials, complex baseline assessments can include long questionnaires, clinical measurements, collection of biological samples and other tests. The rationale for these assessments includes confirmation of eligibility, characterisation of the cohort and to allow subgroup analyses. While some trials may require biochemical or physical measures to determine eligibility, or as part of the outcome assessments, a random sample of the population is often sufficient. In ASCEND, baseline data were limited to self-reported information about diabetes history, cardiovascular risk factors and non-study medication use. The non-mandatory blood and urine samples collected at the participants’ local general practices, which were mailed to the central laboratory, enabled further characterisation of around two-thirds of the randomised cohort, which allowed assessment of the effects of the treatment in various subgroups defined by laboratory results (e.g. HbA1c). If this approach is used in a large randomised trial, the proportion of individuals without a valid sample at baseline should be equally distributed between the treatment groups, so that subgroup analyses can simply include a separate subgroup for those individuals without a valid result.
In a typical clinic-based trial, adjudication of cardiovascular outcomes often does not alter the main findings but does consume substantial resource . In ASCEND, the ability of participants to accurately report cardiovascular and bleeding events, the key primary, secondary and safety outcomes, was unknown and so such outcomes were adjudicated. Provided adjudication is undertaken blind to treatment allocation, misclassification of small numbers of events does not introduce bias and has little effect on statistical power . Therefore, a streamlined approach was taken, with limited collection of supporting documents from the participants’ general practitioners (GPs) and adjudication of each event by a single trained adjudicator blind to treatment allocation. Unrefuted participant-reported events (i.e. those for which the adjudication procedure was unable to confirm or refute the event because of lack of information) were included in the analysis. The validity of this approach for future studies depends on the outcomes considered. For some outcomes, such as hypoglycaemia, confirmation from GP records may not be useful and for outcomes that are not well reported by participants, inclusion of unrefuted cases might not be appropriate. Analyses assessing the impact of adjudication on the main ASCEND trial results will be detailed in a later publication.
Lessons learnt from the practical implementation of the ASCEND study design
Lessons learnt from the practical implementation of the ASCEND study design
Identification of potentially eligible individuals
Searches of healthcare data (e.g. retinopathy registries) and GP lists
Obtaining permissions and managing formats for multiple datasets caused delay
Search of National datasets (e.g. Hospital Episode Statistics in England and Wales)
Linkage with additional datasets to refine eligibility prior to invitation
Invitation to take part
Large numbers of centrally mailed invitation letters sent by post
Response rate low
Requirement to identify large numbers of potentially eligible individuals
Assessment of eligibility
Participant self-report on mailed screening questionnaire
Substantial resource needed to process and check completed questionnaires
Online or app-based questionnaires
Mailed consent form with telephone support from study nurses and doctors
Changing regulatory framework may require a ‘prior interview’
Online or app-based consent materials with interactive questions supported by Skype or telephone interview
Optional blood collection (at local general practice, mailed to a central laboratory)
Postal delays meant delayed processing of samples
Response rate ~70%
Linkage to routine healthcare lab results
Collection of follow-up data
Mailed or online 6-monthly follow-up questionnaires
Systematic chasing of non-responders
Substantial resource needed to process and check questionnaires and clarify inconsistencies/missing answers
Online or app-based participant questionnaires
Ascertainment of outcomes from routine healthcare data
Coordinating centre staff contacted those wishing to stop treatment to encourage continuing
Required substantial resource at the coordinating centre
Direct discussion with participants likely to still be required
Participant identification and invitation
The ASCEND recruitment procedures are described in detail elsewhere . Potentially eligible patients with diabetes were identified from electronic searches of centrally held diabetes registers (e.g. for retinopathy screening) in collaboration with local investigators, for which permission was obtained under UK national privacy legislation (Section 251 of the Health and Social Care Act). This allowed details of identified patients to be transferred securely to the coordinating centre in order to generate invitations on behalf of the collaborating investigator. Participants were also identified from GP-held registries, facilitated by local research networks. Practices searched their electronic records for eligible individuals with diabetes and mailed them an invitation pack (including an invitation letter, screening questionnaire including the consent form, information leaflet and freepost envelope) that had been provided preassembled by the coordinating centre. Obtaining a sufficient number of potentially eligible patients for invitation required collation of 58 separate diabetic register datasets and liaison with 785 general practices.
Assessment of eligibility, consent and pre-randomisation run-in period
Around 400,000 people were mailed an invitation pack by post and around one third returned a valid screening form. Screening forms were logged, checked and scanned using optical character recognition software. Eligibility for the study was determined from the participant’s responses on the screening questionnaire. The questionnaire contained binary questions [e.g. ‘Has a doctor ever told you that you had a stroke or ministroke (sometimes called TIA)?’] along with some questions requesting a written description from the participant. These written responses were reviewed by a clinician, with further details sought by telephone as needed. The screening and other ASCEND questionnaires are available online .
Consent required the person to confirm and sign that they had understood the implications of entering the trial having read the information leaflet and having had questions relating to trial entry, if any, addressed by telephone. Future trials will need to establish cost-effective methods of establishing valid consent without an in-person visit and regulations made sufficiently flexible to allow this. Such methods could include an online or app-based video describing the study, with a participant-completed quiz to test understanding and support available by telephone, e-mail or skype to answer questions .
Consenting and eligible individuals entered a 2 month run-in period and were sent a run-in pack of medication containing placebo aspirin and placebo n-3 fatty acids along with a medication information sheet and a copy of their signed consent form. An automated letter informed their GP of the patient’s potential enrolment and asked them to contact the coordinating centre if they did not wish their patient to be randomised. Approximately 2 months after entering the run-in period, participants were sent a randomisation questionnaire to confirm eligibility and willingness to continue.
The overall ‘conversion rate’ (i.e. the proportion of those invited who were randomised) was low. Around one-third of those invited returned a valid screening form but, of these, only about one fifth entered the run-in period (6% of those invited) since the majority of respondents declined to take part. Approximately 40% of those entering the run-in period were not randomised [22, 23]. Overall, 4% of those invited (15,480 participants) were randomised. Future studies using similar invitation methods need to ensure that large numbers of potentially eligible people are available to invite.
Few trials that use traditional clinic-based recruitment collect data on how many potential participants were approached for enrolment. The conversion rate from invitation to randomisation in ASCEND was similar to the other mail-based studies included in Table 2 and to a recently completed clinic-based cardiovascular trial which recruited 8000 participants in the UK by mailed invitation . In ASCEND, available data for those who declined to take part was limited but those living in more deprived areas and older individuals were less likely to return a screening questionnaire indicating willingness to take part . Future studies recruiting from within a single national dataset could undertake more detailed comparisons between those who did and did not take part. However, while differences in social deprivation, age, comorbidity or other factors between trial participants and the target patient population will affect the likely absolute harms or benefits, the proportional effects of treatment are likely to be generalisable.
Managing the mail-based recruitment process required considerable resources at the coordinating centre, including a 24 h freephone service for participants or their doctors. Bespoke computer programs were developed to enable and control all aspects of the trial. Dedicated trial managers, nurses and clinicians performed manual checks of the data and contacted the participants for clarification where necessary.
Recruitment to ASCEND took 6 years but could be faster in future. Identification of potentially eligible individuals from a single national dataset would bring substantial efficiency savings. Linkage with additional datasets, such as prescribing data, could refine eligibility prior to invitation and use of online or app-based forms to collect information about eligibility would eliminate many of the manual checks that were required to process the ASCEND screening forms.
Approximately 2–4 weeks after entering the run-in period, participants were sent a blood and urine sample kit and asked to take it to their GP or other usual phlebotomy service. The kit included an information leaflet, a blood collection form (which included a section for recording of blood pressure, pulse, height and weight by the practice nurse) and a consent section (which included a request for consent for long-term storage of samples for future research). Samples were returned to the coordinating centre by mail. Previous stability studies have shown that a wide range of analytes (including HbA1c, lipids and cystatin C) and genetic polymorphisms can be reliably measured in mailed samples at ambient temperature . Sixty-three per cent of randomised participants returned a usable blood and/or urine sample within 4 days of sample collection and provided consent for analyses. Those who returned a sample were generally representative of the whole study population .
In future, linkage to routine laboratory data could increase efficiency and save money. In an ongoing registry trial in haemodialysis patients, local hospital blood results are obtained via the UK Renal Registry , while in a diabetes trial in Scotland (Clinical Trials.gov Identifier NCT03439345), linkage to the Scottish Care Information Diabetes Collaboration allows the study to collect routine blood and urine results.
Collection of follow-up information
Every 6 months, participants were sent a short questionnaire asking about clinical events and admissions to hospital, adherence to the study treatments and use of anticoagulants or non-study antiplatelet agents. Each returned questionnaire was checked and scanned, with automated checks for missing or inconsistent responses. Any written descriptions of events or reasons for stopping the study treatment were reviewed by a study clinician and coded using bespoke software with clarification sought from the participant as necessary. Once coding was complete, details of any serious adverse events reported by participants were populated into the relevant tables within the database. Unless the participant or their doctor had requested that they stop their study treatment, further packs were mailed every 6 months. Serious adverse event reports could also be created by the coordinating centre clinicians if contacted directly by the participant or their doctor.
Throughout the study, the response to the 6-monthly participant follow-up questionnaire was generally over 90%. However, achieving these response rates required considerable work. Participants were sent up to two reminder questionnaires, following which, if no response was received, staff at the coordinating centre made several attempts to contact the participant by phone before mailing a standard letter. If necessary, a call was made to an ‘alternative contact’ whose details were requested on the randomisation questionnaire. If participants still failed to respond, follow-up information was requested from their GP. In addition, the work involved in checking and processing paper follow-up forms was substantial. During the trial an online follow-up questionnaire was developed, which included in-built validation to reduce errors and inconsistencies. Participants who provided a valid e-mail address were given the option of receiving an e-mail link to a secure online version of the follow-up questionnaire. By the end of the study about a quarter of those who were still completing follow-up questionnaires were doing so online.
Future studies should aim to use participant-completed electronic data capture with either online or app-based questionnaires with or without additional use of linked EHR data for outcome ascertainment. In the ongoing Aspirin Dosing: A Patient-centric Trial Assessing Benefits and Long-term Effectiveness (ADAPTABLE) study, a streamlined web-based trial comparing two doses of aspirin in people with prior cardiovascular disease, vascular and bleeding events are assessed from participant reports supplemented by routine hospital claims data, with further investigation reserved for those events where a discrepancy exists . However, future trials in countries with a single health provider or with good record linkage systems could potentially rely entirely on linkage to the EHRs, as is planned for an ongoing trial of allopurinol vs placebo in 5000 people with ischaemic heart disease in the UK . Systems developed by regulatory agencies to link large numbers of health datasets for post-marketing surveillance, such as the EU-ADR web platform  and the FDA Sentinel Initiative , could be used to obtain outcomes for trials and embedded randomised trials are already part of the FDA Sentinel Initiative .
However, several hurdles remain before EHRs can replace clinical-based or questionnaire-based data collection for trials. First, careful validation must be undertaken for each outcome. Most cardiovascular outcomes assessed in ASCEND are accurately captured in the EHRs , but some outcomes (e.g. hypoglycaemia or transient ischaemic attack) may be difficult to assess from routinely collected data. For trials of investigational medicinal products, it may be necessary to receive data in a timely fashion to meet regulatory reporting requirements. Furthermore, open access to trial data may be limited by the data sharing agreements with the providers of linked healthcare data. In ASCEND, routinely collected data about cancer, cause of death and admissions to National Health Service (NHS) hospitals in the UK are available for almost all randomised participants and further publications will assess whether comparable results can be obtained using only EHR data.
Maintaining adherence with study treatment
In ASCEND, any participant who expressed a wish to stop the study treatment or reported low adherence on their questionnaire was contacted by telephone to encourage continuation where appropriate. Participants who stopped the aspirin or placebo treatment were encouraged to continue with the n-3 fatty acids or placebo capsule and vice versa. Engagement was maintained with participants using regular newsletters and a study website. Using these procedures, weighted average adherence to study therapy while at risk of an event was 70% for the aspirin or placebo tablet and 76% for the n-3 fatty acids or placebo capsule over the mean follow-up duration of 7.4 years [14, 15], which is comparable to many clinic-based trials. Although future studies may make use of app-based or online communications and reminders to encourage adherence, it is likely that personal discussion with individuals to address their concerns will still be required to ensure good rates of adherence.
Verification of participant identity
A key challenge for mail-based studies is the verification of the participant’s identity and the authenticity of participant-completed responses. In ASCEND, potentially eligible participants were identified within routine healthcare records and therefore identifiers, such as date of birth, were known prior to invitation. For each questionnaire, participants were asked to record their date of birth as a verification check and visual inspection of the signature was undertaken by the coordinating centre staff. Future studies, particularly those in which data are collected online or via mobile device applications, must develop systems to verify the identity of respondents.
The key streamlined design features of ASCEND allowed a robust result to emerge at relatively low cost (estimated at less than £10 million ). However, future studies could recruit large numbers of participants and collect high-quality data more efficiently, for example, by using EHR data for both identification of participants and outcome ascertainment, and by collecting participant-reported data by online or app-based questionnaires. ASCEND has demonstrated that streamlined methods can obtain high-quality randomised trial evidence, sufficient to meet regulatory requirements and directly influence national and international guidelines . Hopefully similar approaches will be adopted by other researchers, since trials need to become less complex and costly to run if important clinical questions most relevant to improving health are to be answered.
We thank the participants, the collaborating doctors and general practitioners, Steering Committee and Data Monitoring Committee and all the administrative and clinical staff who have helped run the study.
All authors were responsible for drafting the article and revising it critically for important intellectual content. All authors approved the version to be published.
The ASCEND study is funded by grants to the University of Oxford from the British Heart Foundation (SP/03/002, SP/08/010/25939, SP/14/3/31114, PG/05/013/18296). The Clinical Trial Service Unit at the University of Oxford receives support from the UK Medical Research Council (which funds the MRC Population Health Research Unit in a strategic partnership with the University of Oxford), the British Heart Foundation, and Cancer Research UK. We are grateful to Bayer Pharma AG who provided aspirin and placebo tablets and to Solvay, Abbott and Mylan who provided the n-3 FAs and placebo and for all companies for providing funding for drug packaging and to Bayer Healthcare LLC who contributed funding.
All authors work in the Clinical Trial Service Unit & Epidemiological Studies Unit of the Nuffield Department of Population Health at the University of Oxford. The Clinical Trial Service Unit & Epidemiological Studies Unit has a staff policy of not taking any personal payments directly or indirectly from industry (with reimbursement sought only for the costs of travel and accommodation to attend scientific meetings). It has received research grants from Abbott, AstraZeneca, Bayer, Boehringer Ingelheim, GlaxoSmithKline, The Medicines Company, Merck, Mylan, Novartis, Pfizer, Roche, Schering, and Solvay, which are governed by University of Oxford contracts that protect their independence.
- 10.Rao SV, Hess CN, Barham B et al (2014) A registry-based randomized trial comparing radial and femoral approaches in women undergoing percutaneous coronary intervention: the SAFE-PCI for women (study of access site for enhancement of PCI for women) trial. JACC Cardiovasc Interv 7(8):857–867. https://doi.org/10.1016/j.jcin.2014.04.007 CrossRefPubMedGoogle Scholar
- 14.ASCEND Collaborative Group (2018) Effects of aspirin for primary prevention in persons with diabetes mellitus. N Engl J Med 379(16):1529–1539Google Scholar
- 15.ASCEND Collaborative Group (2018) Effects of n-3 fatty acid supplements in diabetes mellitus. N Engl J Med 379(16):1540–1550Google Scholar
- 23.Bowman L, Mafham M, Stevens W et al (2018) ASCEND: A Study of Cardiovascular Events iN Diabetes: characteristics of a randomized trial of aspirin and of omega-3 fatty acid supplementation in 15,480 people with diabetes. Am Heart J 198:135–144. https://doi.org/10.1016/j.ahj.2017.12.006 CrossRefPubMedPubMedCentralGoogle Scholar
- 26.Rothwell PM, Price JF, Fowkes FG et al (2012) Short-term effects of daily aspirin on cancer incidence, mortality, and non-vascular death: analysis of the time course of risks and benefits in 51 randomised controlled trials. Lancet 379(9826):1602–1612. https://doi.org/10.1016/S0140-6736(11)61720-0 CrossRefPubMedGoogle Scholar
- 27.Ndounga Diakou LA, Trinquart L, Hróbjartsson A et al (2016) Comparison of central adjudication of outcomes and onsite outcome assessment on treatment effect estimates. Cochrane Database Syst Rev (3):MR000043. https://doi.org/10.1002/14651858.MR000043.pub2
- 33.Mackenzie IS, Ford I, Walker A et al (2016) ALL-HEART study group. Multicentre, prospective, randomised, open-label, blinded end point trial of the efficacy of allopurinol therapy in improving cardiovascular outcomes in patients with ischaemic heart disease: protocol of the ALL-HEART study. BMJ Open 6(9):e013774. https://doi.org/10.1136/bmjopen-2016-013774 CrossRefPubMedPubMedCentralGoogle Scholar
- 37.Cosentino F, Grant PJ, Aboyans V et al (2019) ESC Guidelines on diabetes, pre-diabetes, and cardiovascular diseases developed in collaboration with the EASD. Eur Heart J pii:ehz486Google Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.