Effect of core versus enhanced implementation strategies on adherence to a clinical pathway for managing anxiety and depression in cancer patients in routine care: a cluster randomised controlled trial

Clinical pathways (CPs) are standardised, evidence-based multidisciplinary management plans, which identify an appropriate sequence of clinical interventions, timeframes, milestones and expected outcomes for one or more patient groups [1]. By their operational nature, CPs provide a level of detail over and above that provided in clinical guidelines. CPs are increasingly used in healthcare to achieve optimal, evidence-based and cost-effective outcomes, [2, 3] increase hospital efficiency [4, 5], decrease operational costs [6, 7], reduce lengths of stay [8], and decrease mortality rates [9].

However, while CPs have been shown to improve patient outcomes, this is not always the case [10], possibly due to poor uptake or incomplete adherence to CPs resulting from environmental, system or practitioner barriers [11, 12]. Deviations from CPs may reflect beneficial tailoring to the needs of individual patients but may also result in reduced quality of patient care, [5, 13] increased staff burden due to additional steps required to address deficiencies [3], and increased staff resistance to future health service change.

Due to these negative outcomes, increasing research effort has explored strategies to promote uptake and adherent use of health service interventions [14], culminating in sophisticated, evidence-based implementation frameworks, such as the Promoting Action Research in Health Services framework (PARiHS) [15] and the Consolidated Framework for Implementing Research (CFIR) [16]. These frameworks propose that, along with evidence, context and intervention characteristics, facilitation is key to implementation success. Facilitation encompasses resources and processes offered both internally and by the research team, to support staff in implementing interventions such as CPs. While there is increasing understanding of the nature of facilitation strategies, there is very little clarity regarding the optimal dose of facilitation required. Without such data it is difficult for health systems to plan implementation efforts when introducing new interventions into health systems, including CPs, into routine care.

Our group developed a CP for screening, assessment and management of anxiety and depression in adult cancer patients (the ADAPT CP) to guide best practice in Australia [17]. The ADAPT CP is based on evidence review, refined through stakeholder interviews [18] and a Delphi consensus process [19]. The ADAPT CP follows a stepped care model incorporating iterative screening at recommended intervals, with triage to one of five steps, each with a recommended management plan (from universal care and self-management for those with minimal or mild levels of anxiety and/or depression, to specialist care for those with severe anxiety and/or depression), with review and change in step where necessary. Evidence-based recommendations on staff responsibilities, content and timing of interventions, are provided for each step and tailored to available resources [17]. Drawing on the PARiHS and CFIR frameworks to support CP implementation, we developed an online portal (the ADAPT Portal) [20] to operationalise as many processes as possible, increase efficiency and reduce staff time and burden. We also developed staff education modules [21], patient information, and an online cognitive-behavioural intervention for mild to moderate anxiety and depression [22] to address concerns raised by health professionals in an earlier barrier analysis [18].

Our team has reported extensively on implementation preparation, processes and outcomes of the ADAPT trial [23], including organisational readiness for change [24], identification of barriers and facilitators [25], selection and definitions of evaluation outcomes [26], staff perspectives on acceptability and appropriateness [27] and tailoring of implementation processes according to site needs [28]. Forthcoming publications will address implementation costs and fidelity, sustainability of the ADAPT CP and effect of the ADAPT CP on health service use.

In this cluster trial, our primary objective was to determine whether “dose” of implementation strategy (Core versus Enhanced) effects staff adherence to the ADAPT CP. Our secondary objective was to examine the effect of step allocation (overall, and by implementation support arm) on total and component adherence scores.

Our primary hypothesis was that the proportion of eligible patients screened, for whom sites achieve acceptable adherence to the ADAPT CP (defined as yes: having completed ≥ 70% of the recommended ADAPT CP components versus no.: < 70%) would be greater in the Enhanced implementation strategy arm than in the Core implementation strategy arm.

Secondary hypotheses included that:

Block randomisation resulted in 4 sites allocated to the Core arm, and 8 to the Enhanced (see Table 3 for site characteristics). Characteristics were evenly distributed across arms except for funding type, with all Enhanced arms being publically funded, while half the Core sites had a private or private/mixed funding model.

There were 1280 patients registered on the ADAPT Portal, 745 in the Core and 535 in the Enhanced services. Of registered patients, 696 (54%) went on to complete at least one screening, 63% in Core and 42% in Enhanced services. The most common reasons across arms for patients not screening were non-response to the screening invitation (41% of non-screeners), or patient decline (22% of non-screeners); specific reasons for non-response and decline were not recorded. As patients could complete more than one screen, a total of 1323 screening events (883 in Core and 440 in Enhanced services) were recorded (see Fig. 1).

There were slightly more females (62.6%) in the Core arm compared to the Enhanced arm (52.4%) (Table 4). Mean age of patients (63.4 years) was similar across arms. The most common cancer diagnoses were breast and genitourinary cancer. At the time of registration into the ADAPT CP, most patients had a diagnosis of stage 2/3 (regional spread) or stage 4 (distant spread) cancer. A higher (non-significant) proportion of patients completed two or more screenings in the Core (53.9%) than Enhanced (45.8%) arm.

Seventy-four percent of patients were categorised as having minimal or mild anxiety/depression after their first screen, with the remainder evenly distributed between steps 3–4 (there were no step 5 s). Following triage conversations, most step allocations were maintained or downgraded to a lesser step by staff in both Core and Enhanced arms, resulting in 63 at step 3 (9.1%) and 26 at step 4 (3.7%).

Our finding that implementation strategy as a main effect did not impact adherence is in contrast to the findings of Almatar et al. [34], who found that additional monthly feedback did improve adherence to a clinical pathway for community-acquired pneumonia over education alone. Notably, champions at Core arm sites did also receive (uncurated) monthly feedback, which may have reduced our ability to detect differences between the arms.

However, our cut-off for adherence, while based on similar studies, is somewhat arbitrary. When we used a continuous outcome, providing maximum power, and included the interaction term between implementation strategy and severity step, we found a marginally significant difference between arms for adherence to the ADAPT CP for patients with higher or more severe levels of anxiety and/or depression. Controlling for other potential confounders, staff in the Enhanced arm adhered to about 7.5 percentage points more components of the CP than those in the Core arm, for patients with step 3 anxiety/depression. While statistically non-significant due to considerable variability, there was a 10 percentage-point difference in adherence between arms for patients with step 4 anxiety/depression. Thus, the Enhanced implementation effort did appear to assist services to respond to patients with more severe anxiety/depression, who arguably need help more specialist care than those with minimal to mild anxiety/depression. However, this result must be interpreted with caution, given the marginally significant result.

These results provide some of the first empirical data on the extent of implementation effort required to ensure success in implementing health service change in the form of a novel CP. Our study suggests that even when great care has been taken to design and launch a new CP, adhering to principles described in implementation science frameworks, ongoing assistance during implementation facilitates greater staff adherence. While frameworks such as the Consolidated Framework for Implementing Research (CFIR) [16] do emphasise ongoing processes, particularly tracking progress and refining an intervention across time to simplify and enable execution, this longitudinal aspect of implementation has received relatively little attention in the implementation science literature to date [15, 16]. Cancer services in our Enhanced arm received ongoing progress reports; scheduled monthly meetings to discuss reports, identify and overcome barriers, identify training and support needs of local champions and deliver these, and highlight issues for attention and promotion to the wider team. In addition, ongoing awareness campaigns were run and regular newsletters with progress updates and tailored strategies to address identified service-specific barriers were sent to all staff. Further research is required to identify any additional strategies that could promote uptake of health service initiatives over time.

If longitudinal implementation effort is required to achieve health service change, adequate resourcing will be needed not only for initial effort, but for sustainment. Our own health economic data (paper under review) will inform appropriate planning. Unfortunately to date, it appears that health systems often fail to dedicate sufficient resourcing to integrate and sustain implementation efforts, particularly those focused on improving mental health [35]. Both funding and sufficient staffing are required during implementation of novel CPs. In Australia, as elsewhere, there are significant psycho-oncology workforce shortages, with few hospital positions allocated to this role.

All patients included in this data set had completed screening at least once. Subsequent adherence was highest across both arms of the study for triage (85–100% across anxiety/depression steps). Staff followed up concerning patient scores on screening measures, by having a discussion with them to clarify their concerns and needs. Adherence to re-screening, whilst lower than for triage, was also relatively high (54–78%). Screening for distress using patient-reported outcome measures has been promoted for many years [36, 37], so it is unsurprising that staff found these aspects of the ADAPT CP the easiest to adhere to, while triage is a clinical strategy with which most staff are familiar.

However, adherence to interim ADAPT CP steps (tracking whether patients took up referrals, were happy with care received and improved or required re-referral, and obtaining a treatment discharge summary) was poor. These components, laid out in the ADAPT CP, are particularly important for patients with more severe morbidity who require referral. Yet we observed a highly significant difference for the main effect of anxiety/depression step on adherence, with lower adherence overall for those with greater anxiety/depression.

While we did not directly assess reasons for this lower adherence to interim CP components in the current trial, we can speculate regarding causes. Oncology staff are likely less familiar with these components, than with initial screening. There may also be a lack of referral pathways, given psychosocial workforce limitations, and staff may therefore be referring out to community services with whom they have less established communication channels. Furthermore, these steps can require more time and be frustrating if the specialist staff to whom patients have been referred respond slowly or not at all to requests for information and discharge summaries. Integrating psychological care related to cancer diagnosis that occurs outside of the cancer centre could be complex and require careful planning and system changes. Staff may need targeted support in integrating these CP components within existing workflow processes.

It is also possible that these steps occurred but were not well documented in either the ADAPT Portal or patient records. However, without documentation, steps to address non-uptake of referral or need for new referral may not consistently occur, impacting patient outcomes [5, 13]. An advocacy effort to promote these aspects of quality care is now needed, given the success of earlier advocacy efforts to entrench screening into routine care.

Our study had significant strengths. We employed a cluster randomised controlled trial design, that allowed us to clearly compare two implementation strategies. We recruited diverse oncology services across urban and regional areas, and public and private health care systems. We collected detailed data on staff adherence to each of the CP components and controlled for potential site and patient confounders. Reporting complied with CONSORT reporting standards [38] (see Additional file 1).

The study also had limitations. While our study compared a Core implementation strategy with an Enhanced one that provided implementation support across one year, it is not clear what intensity of implementation support is required to achieve improved adherence. Perhaps less ongoing support may have been sufficient, or more would have achieved an even better outcome. Future studies could compare more or less intense implementation efforts, compare length of supported needed and tease out which strategies were most helpful.

We were not able to obtain accurate data on the proportion of all eligible patients who were invited by staff to participate in the ADAPT CP. The implementation strategy delivered may have effected the level of staff commitment, time and energy devoted to encouraging patients to screen for anxiety and depression, and this should be explored in future trials.

An important consideration is how much statistical power we had to detect an effect if one was present. This information is best captured in the bounds of the confidence intervals for the effect sizes. Although there was no effect of implementation arm on adherence as a binary outcome (Table 6), the reasonable range of possible effect sizes that the implementation could have had is an odd’s ratio between 0.85 and 1.59. Thus it is possible that we failed to detect an effect (within the confidence interval bounds) that was present. However, as there are very few implementation studies in Psycho-Oncology on which to base effect size estimates, it is difficult to interpret this potential effect size with confidence. This demonstrates the need for future work in this area to provide more precise estimates of effect sizes.

The ADAPT Portal was not able to integrate with the electronic medical records (EMRs) used in participating hospitals, due to diversity of systems used across services. This likely reduced staff compliance in recording their actions within the Portal in line with instructions. Thus, adherence to the ADAPT CP may have been higher than we could detect. However, we did conduct exhaustive medical record reviews to supplement the portal data and ensure our adherence data was as accurate as possible. Nevertheless, future studies would benefit from integrating data collection methods within existing hospital EMR systems.

In summary, this study suggested that an Enhanced implementation strategy that supported staff over 1 year may support staff to be more adherent to a clinical pathway for anxiety/depression, at least for patients with more severe anxiety/depression. Wensing and colleagues [39] noted that with the increasing pace of scientific discovery, and the number of new programs and technologies being introduced into the health care system each year, successful and efficient integration of these interventions into routine care is increasingly required. Our data suggest that commitment of sufficient staff and resources to support staff during implementation efforts will increase our success in incorporating evidence into patient care and improving outcomes.