Type 2 diabetes mellitus is a progressive disorder, with many patients eventually requiring insulin therapy. Alongside management of risk factors such as blood-pressure, smoking and lipids, tight glycaemic control significantly reduces the risks of diabetes-related micro- and macro-vascular complications [1–3]. As a result of these findings, health professionals have, until recently, been encouraged to adopt aggressive approaches to controlling blood glucose levels amongst their type 2 diabetes patients, including early initiation onto insulin.

The 4-T study was a large, 3-year multi-centre, open label, UK-based randomized trial. It was developed because of a lack of clinical knowledge at the time about the most appropriate ways to initiate insulin therapies in patients with type 2 diabetes, and how to help them achieve optimal blood glucose control on intensive insulin treatments. To do this, the trial evaluated and compared the impact of three different licensed analogue insulin regimens in patients amongst whom oral medication was no longer sufficient to maintain good glycaemic control [4, 5]. The three insulin regimens investigated were basal (insulin determir, normally taken once a day), biphasic (insulin aspart 30, taken twice daily) and prandial (insulin aspart, taken three times a day), and were all already widely used in routine clinical practice.

The trial employed a treat-to-target approach with a glycaemic target of HbA_1c ≤6.5%, in line with clinical recommendations at the time of trial development [6, 7]. This treat-to-target approach was used to help and enable patients to achieve tight blood glucose control and allow the different insulin regimens to be compared in terms of ease of use, acceptability, clinical outcomes (e.g. weight effects, risk of hypoglycemia and impact on microalbuminuria), and patient adherence. A second insulin formulation was added in the second and third years if HbA_1c levels of ≤6.5% were not achieved. To promote attainment of the trial's glycaemic target, an on online Trial Management System (TMS) was used in all centres. Staff were required to enter patients' clinical data (e.g. weight, self-reported blood glucose readings over a three day period, and reported episodes of hypoglycemia) into the TMS. The TMS then employed a standardized algorithm [8] to determine the insulin doses that needed to be prescribed in order for patients to attain or maintain the glycaemic target, although, in the event of hypoglycaemia or other untoward factors, trial staff were allowed occasional discretion in the prescription of insulin doses. During the trial, patients received a comprehensive monitoring and support package comprising 17 clinic visits and at least 11 telephone contacts over three years. Patients were also encouraged to alter insulin doses themselves in between contacts and in response to self monitoring of blood glucose (SMBG) readings to help attain the trial target. The trial commenced November 2004 and final closeout took place in July 2009.

Despite the intensive input provided during 4-T, there was limited attainment of the glycaemic target across all three treatment groups. At one year, 17% of the biphasic, 24% of the prandial and 8% of the basal group attained an HbA_1c ≤6.5%. At three years, the corresponding proportions were 32%, 45% and 43% [4, 5]. Similar outcomes have been observed in other trials involving intensive insulin therapies [9–12] and in routine clinical situations, recent data indicating that control may be sub-optimal in up to 75% of insulin-treated patients [13, 14]. This has prompted calls for the behavioral factors to be considered and explored [5]. However, research has predominantly focused on attitudes towards, and barriers to, initiating insulin therapy [15–18]. There is a very limited understanding of how patients and health professionals perceive and respond to tight glycaemic targets and the issues and challenges encountered attaining or sustaining tight glycaemic control after insulin treatment has been initiated. Similarly, while there is an extensive trial literature, this has predominantly focused on patients' motivations for trial participation [19–22] or patient and/or staff understandings of randomization and/or equipoise [23–25], and the implications of these for facilitating recruitment and informed consent [19, 26]. Little is known about patients' and health professionals' experiences of trial delivery and why treatments may or may not be adhered to in the context of a clinical trial.

In this paper we report findings of an interview study involving patients and health professionals who had participated in 4-T. This study was conducted to understand why there was limited attainment of the trial target for glycaemic control across all three treatment groups and over time. Key objectives were to inform interpretation of trial outcomes and provide recommendations for future trial delivery.

Qualitative methods are recommended when little is known about the area of investigation. In-depth interviews were used in this study as these encourage participants to display their own understandings and meanings, and permit themes and hypotheses which might not have been anticipated to be identified and explored [27].

Findings from the main 4-T trial showed that endpoint glycated haemoglobin were similar for patients randomised to biphasic (median: 7.1%), prandial (median: 6.8%) and basal (median: 6.9%) insulin-based therapies (P = 0.28) [5]. Consistent with these findings, analysis of the qualitative interviews did not reveal widespread or systematic differences in reported adherence amongst patients in the three trial arms. Staff accounts, similarly, did not reveal obvious differences in the treatment, advice and care given to patients in different arms of the trial (hence data reported below cut across the three trial arms). However, whereas patients tended to present relatively unproblematic accounts of participating in, and taking their insulin during, the trial, staff highlighted various difficulties and dilemmas which arose during trial delivery. In particular, staff described experiencing a conflict between their roles and responsibilities as researchers and clinicians. This conflict became most apparent when they were required to treat to the trial's tight glycaemic target.

Below, we begin with patients' experiences of taking part in 4-T before considering staff perspectives and views. We will show that staff accounts need to be considered alongside those of patients to fully understand why there was limited attainment of the trial's glycaemic target.

This study sought to understand why there was limited attainment of the primary endpoint (HbA_1c ≤6.5%) in the Treating to Target in Type 2 Diabetes Trial (4-T). Patient interviews did not reveal extensive and intentional acts of treatment non-adherence which would have provided a quick and easy answer to this question. To the contrary, patients generally made efforts to adhere to their insulin regimens during the trial, in some cases by skipping or titrating doses to avoid hypoglycaemia. Some also believed they had done well during the trial despite not reaching the glycaemic target. While previous research has attempted to explain poor adherence to diabetes and other treatments by focusing upon individual patient characteristics, experiences and life circumstances [31], this study highlights the importance of considering the perspectives and experiences of staff. By locating patients' accounts alongside those of 4-T staff, this study has highlighted how staff mediated and buffered patients' experiences of using insulin treatments during the trial, their perceptions of success, and their awareness of the trial target. It has further been shown that the potential impact of staff behaviours on trial outcomes arose in part from the conflict they experienced between their roles and responsibilities as researchers and practitioners and from the requirement to follow a protocol which sought to standardise practice. While they sometimes addressed this conflict by revising their clinical practice in light of trial experiences (e.g. saying they would prescribe larger insulin doses after using the TMS), more frequently, and especially when staff came to the trial with diabetes clinical experience, they adapted the trial protocol to align it with their clinical practices and experiences. This led to instances of staff not following the TMS when recommended doses were perceived as presenting risks to patient trust and safety, or offering patients enhanced care outwith the protocol. While such interventions may have helped to curb or mitigate intentional treatment non-adherence by patients and/or to foster trial retention, they also appear to be a major factor responsible for the limited attainment of the 4-T glycaemic target.

Staff ambivalence about using decision support software and following standardised algorithms has been observed in other healthcare settings, particularly when recommendations contradict professional judgement arising from clinical experience [32, 33]. Hence staff attempts to tailor and individualise patient care during 4-T may not simply have been a consequence of working in a trial setting, and may reflect a more general difficulty health professionals encounter when required to follow protocols and clinical guidelines which seek to regulate and standardize their practice [34]. Like this study, research has also found that triallists may struggle to exchange their role of providers of individualised care with that of researchers required to follow standardised trial procedures [25, 35], resulting, for instance, in the provision of a 'quasi-clinical service' in the research setting [35, 36]. However, while many trials involve new treatments, the insulin therapies compared in 4-T were already widely used in clinical practice. Hence 4-T staff may have had more clinical experience to draw upon than is typical in a trial and may have experienced a greater role conflict as a consequence. Indeed, the data suggest that a large part of staff ambivalence about their participation in 4-T arose from the trial's requirement to treat to a lower target than that routinely used in their clinical practice. Hence the timing of the trial and its location within UK may have impacted on its outcomes. In the UK, management of type 2 diabetes is strongly influenced by the requirements of the UK quality and outcomes framework (QOF). Until recently, the financially incentivised target set by QOF was 7.5%. While the target was reduced to 7% in April 2009 [37], in line with EASD recommendations, it is still higher than that used in 4-T and recommended in other countries [6, 7]. Furthermore, it must be noted that, for reasons of patient safety, 4-T staff were allowed a limited degree of leeway in prescribing the dosage as recommended by the TMS. A degree of discretion is often built into protocols of pragmatic trials and complex interventions. However, granting 4-T staff the scope to adjust the TMS' recommendations may have been interpreted by some, particularly the more clinically experienced, as permission to individualise treatments more widely than was originally intended by those who designed the trial.

While the specific context of 4-T delivery needs to be taken into account, this study does raise important issues for the conduct of future trials, especially those, such as 4-T, where staff may be required to follow procedures to assess individual patient safety before implementing mandated therapy changes. Fidelity to protocols is essential for successful trial delivery [38] and this study, alongside others [25, 39], highlights the importance of exploring, and addressing, staff perspectives and views. One way to avoid or resolve the role conflict highlighted by 4-T staff and observed elsewhere [35, 36] may be to appoint front-line staff who do not have an extensive clinical background or training in the specific area of trial investigation. However, this approach may not be feasible as changes in the research environment have resulted in an increased reliance on nurses working on trials who have primarily clinical rather than research experience. While trial staff might benefit from greater regulation and oversight to foster protocol adherence, such as the recording and feedback of the frequency of their overrides, it has already been shown that these regulatory procedures are unlikely to curb individual decision-making and ensure standardization of practice [33]. Hence multifaceted interventions may be needed to support staff, reduce ambiguity in the courses of action they perceive as open to them [40], and help them 'suspend belief' and provide treatment in accordance with a protocol which may be outside current guidelines of standard clinical care. This may include the provision of more training and education in principles of trial design, aims, and methods, to promote a better understanding of why trial protocols need to be followed and of concepts such as equipoise [25]. Since increasing numbers of clinical staff are becoming involved in clinical trials, this type of education could be made an integral part of all specialist nurse training. Another approach which has already been used to support delivery of trials involving complex interventions is to set aside resources to help promote protocol fidelity. These include extensive training aligned to a manual, use of role play to ensure familiarity with options available in the protocol, recording of interventions, and review of these recordings with a lead facilitator or supervisor to discuss best practice in following the protocol [41, 42]. However, this type of approach may be complex and costly to deliver. An alternative is to focus on leadership of research in the clinical setting. One way to do this would be to appoint a cadre of senior research staff who are not directly involved in trial delivery and are available to provide regular support to research nurses. In providing support and supervision, it would be possible to identify issues arising during trial delivery and help resolve them. The building of qualitative work into the early stages of a trial may also be an effective way of identifying staff dilemmas and/or better ways of delivering a trial to promote adherence and retention [43].

A key strength of this study, as already indicated, is that it drew upon the perspectives of staff as well as patients, enabling a clear and comprehensive understanding to be gained of why there was limited attainment of the trial's primary endpoint. Indeed, we would recommend that multiple perspectives (staff and patient) are considered in all future work seeking to understand and/or promote treatment adherence, whether this be in a trial or clinical setting. The use of a multi-site recruitment strategy, which is unusual in qualitative research, increases the potential generalisability of the findings. By using an opt-in procedure (a medical ethics committee requirement), and only interviewing trial completers, the sample may have been biased towards patients with more positive, or extreme, experiences of the trial and injecting insulin. However, this potential limitation was overcome by purposively selecting a patient sample broadly representative of trial participants in terms of glycaemic control and (non-)attainment of the trial's glycaemic target.

There is a growing trend towards trials which involve the use of licensed treatments, as well as those in which staff are allowed some discretion in exercising or following a protocol (e.g. trials involving complex interventions). As 4-T is a good example of this trend, this study has wide relevance for the rigor with which future trials can be delivered and provides important insights into some of the issues faced by trial staff. The findings highlight that the delivery of trials such as 4-T can be affected by what are currently poorly recognised and debated issues, such as role conflicts experienced by trial staff. These conflicts may be exacerbated when trial protocols are implemented by staff who undertake clinical work alongside, and/or in the same area as, the trial. Our findings suggest that, to promote fidelity to trial protocols, front line staff may benefit from more education, training and support, especially when they are required to provide treatment which is outside current guidelines of standard clinical care. Similarly, involving clinical staff in the development of trial protocols and procedures may help limit the possibility of tensions arising during trial delivery.