Is a specialist breathlessness service more effective and cost-effective for patients with advanced cancer and their carers than standard care? Findings of a mixed-method randomised controlled trial

We recruited patients with advanced cancer referred to BIS into a Phase III mixed-method single-blind pragmatic fast-track (waiting list) RCT of BIS versus standard care (November 2008 to January 2012). Ethical approval was given by by Cambridgeshire 2 NHS REC (Ref:08/H0308/157); RCT registration at ClinicalTrials.gov NCT00678405 and Current Controlled Trials ISRCTN04119516.

Participants were randomised to one of two groups using randomly permuted blocks of random size two, four and six, generated by the study statistician and concealed within sealed opaque envelopes until allocation notification by the intervention deliverer. The fast-track (intervention) group received BIS immediately; the waiting-list (control) group received BIS after two-weeks. All participants received standard, including palliative, care. Data collection-design facilitated researcher-blinding to group allocation for the collection of primary and key secondary outcomes at the key measurement point, that is, planned unblinding occurred during the two-week follow up interview (t3) only after collection of this outcome data and prior to qualitative data collection about the intervention.

Consecutive cancer patients referred to BIS (from primary or secondary care) were invited to participate by letter. Patients were eligible if they met BIS referral criteria (that is, diagnosed appropriately-treated cause of breathlessness, troubled by breathlessness in spite of optimisation of underlying illness, and might benefit from a self-management programme) and excluded if they had received BIS previously. Recruited patients were asked to identify who gave them the most help and support at home and these informal carers were also invited to participate. All participating patients and informal carers gave informed consent. Patients who were unwilling to participate in the trial continued to have access to BIS.

A sample size of 60 randomised patients (26 analysed per arm, allowing for dropout) provided 80% power to detect a 2-point difference in mean distress at two-weeks between groups (SD = 2.5, alpha = 5%), with increased precision anticipated from adjustment for baseline.

Patient distress due to breathlessness [4],[5] (the primary outcome on which the trial was powered) was measured using a numerical rating scale (NRS). Other key patient-reported variables included disease-specific health related quality of life (Chronic Respiratory Questionnaire: CRQ [25]), and anxiety and depression (Hospital Anxiety and Depression Scale: HADS [26]). Key carer-reported outcome measures included an NRS for carer distress due to patient breathlessness and HADS. A generic health status measure (EQ-5D [27]) and measure of service use (Client Services Receipt Inventory (CSRI) [28]) were administered for health economic analyses. Brief qualitative topic-guided interviews were also conducted with all patients and carers to explore expectations and experiences of BIS. All outcomes were participant-reported; the full outcomes list is reported elsewhere [24].

Participating patients and carers completed a baseline interview (t1: week 1) before randomisation. These mixed-method interviews included the quantitative patient- and carer-reported measures and qualitative interviews described above (carers were interviewed separately where possible). A two-week follow up interview (t3: week 3) was designed to represent completion of BIS for the intervention arm, or end of the waiting-list period prior to BIS for controls. A final interview (t5: week 5) was conducted four-weeks from baseline; this represented two-weeks after BIS for the intervention arm and completion of BIS for controls. All interviews were conducted in home-settings.

Complete case intention-to-treat analyses were conducted using a linear regression model; each outcome was adjusted for its baseline. A 5% level of statistical significance was used. Costs were calculated combining service use data (CSRI) for eight-weeks and two-weeks prior to baseline and t3, respectively, with UK 2011/2012 unit costs [29]. Informal care (unpaid hours/week from family/friends performing specific tasks) was valued at average UK wages (£11.21/hour) [30]. Costs of BIS visits were estimated at £91 (based on specialist nurse contacts which averaged the rehabilitation specialists’ wages) and phone contacts at one-quarter of this. Costs were compared using regression models, controlling for baseline, and using bootstrap methods with 1,000 resamples to address the likely skewed distribution. Sensitivity analyses were performed around these key costs by increasing/decreasing costs by 25% and 50%. Costs, with and without informal care, were combined with the primary outcome and EQ-5D-derived quality-adjusted life years (QALYs), with uncertainty explored using cost-effectiveness planes [31]. Cost-effectiveness planes are a graphical way of comparing one intervention to another or to usual care. The vertical axis of the plane shows the extra costs incurred by an intervention while the horizontal axis shows the extra outcomes achieved; both of these could be positive or negative resulting in four quadrants on the plane. Bootstrap methods can then be used to produce a large number of cost and outcome combinations which can be plotted on the plane to show the probability of the intervention resulting in higher costs and better outcomes, higher costs and worse outcomes, lower costs and better outcomes or lower costs and worse outcomes.

Qualitative interview data were transcribed and anonymised. Two analytic approaches were taken to the comprehensive qualitative dataset. First, transcripts of all post-intervention interviews (fast-track group t3s and control t5s; n = 53, no qualitative interview for one patient) were categorised into one of three intervention impact levels by three analysts working independently (Level 1: Significant impact - clearly stated BIS made a difference; Level 2: Some impact - no major change recognised, but valued specific aspects of BIS; Level 3: No impact - BIS made no difference at all). Categorisation commenced with a small number of interviews. Analysts then compared categorisations, discussing and resolving differences, clarifying level-definitions and data interpretation, before repeating for all remaining interviews.

Second, as qualitative analysis of this sized dataset (n = 53) would be unmanageable, 20 patient (and associated carer) intervention arm transcripts were purposefully sampled against a novel stratified four-cell matrix of primary outcome changes by t3, to achieve a maximum diversity sample [32]. The four cells represented: (Cell 1) patients who improved most on primary outcome (who, predictably, had high baseline scores; Biggest Improvers); (Cell 2) patients with high baseline scores (to match Cell 1) but who improved least (Limited Improvers); (Cell 3) patients who worsened (who transpired to have low-middling baseline scores; Worseners); and (Cell 4) patients with closest match to Cell 3 baseline scores but who improved most (Moderate Improvers). Anonymised interview transcripts for this purposive sample were imported into NVivo software [33], to facilitate framework analysis [34]. This principally descriptive analysis explored the nature of BIS impacts, which aspects were valued, and possible mechanisms of impact.

Comparison of change in patient distress due to breathlessness (primary outcome; NRS range 0 to 10) by the key outcome measurement point (t3) showed that patients randomised to the intervention arm achieved a significantly greater, 1.68-point, reduction compared with 0.23-point reduction for controls: adjusted difference of -1.29 (95% CI: -2.57 to -0.005), P = 0.049 (Table 2). There was little change in carer distress.

Mean CRQ mastery scores improved only negligibly by t3 on the intervention arm and remained stable for controls (not statistically significant; Table 2). No significant differences were found between trial arms to t3 on other CRQ domains (dyspnoea, fatigue or emotional function). Mean anxiety scores (HADS) remained fairly stable to t3 (both arms). Mean depression scores decreased slightly by t3 in the intervention arm, increasing slightly for controls (not statistically significant). There was little change in other patient or carer outcomes.

Categorisation of qualitative interviews inferred that for 68% (n = 36) of patients, or patient-carer dyads, BIS had had a significant impact (Level 1). A further 28% (n = 15) indicated BIS had had some impact (that is, no major change, but they valued specific aspects of BIS; Level 2) and 4% (n = 2) reported no impact (Level 3).

Table 3 shows the Impact Categorisation Levels for the purposively-sampled qualitative interviews, sampled for maximum diversity of change on primary outcome to t3. Given the skewed distribution of the levels it is unsurprising that most were Level 1 and 2, even among primary outcome `Worseners’ (Cell 3).

Qualitative analysis of the purposively-sampled interviews identified both the nature of BIS impacts and which aspects of the BIS model were valued, as well as the possible mechanisms of impact. Impacts described included patients and carers noting reductions in fear, anxiety, worry and feelings of panic, and greater confidence about breathlessness. Patients reported managing their breathlessness better, and some carers living separately from patients reported that the latter contacted them less.

Patients and carers valued specific aspects of the BIS model. The multi-disciplinary staff expertise was repeatedly noted: their knowledge about breathlessness, and strategies suggested to manage it, made clear they understood difficulties of life with breathlessness. Participants described particularly helpful characteristics of BIS staff, such as being relaxed and easy to talk to, listening and reassuring. They valued the time BIS gave them to talk about breathlessness with an expert. Being seen at home was especially helpful, as was the positive `can do’ approach of BIS and unexpected attention given to carers.

Patients and carers identified interventions delivered by BIS that were therapeutic including: providing and teaching use of a handheld fan; encouragement of exercise (including pedometer provision/goal-setting); coaching in breathing techniques, positioning, pacing and relaxation (including provision of BIS’ mindfulness-based body-scan CD and teaching visualisation techniques); occupational therapy aids; information and education (verbal, printed sheets and hand-drawn diagrams) - learning that `being breathless won’t kill me’ was cited as particularly liberating. Other valued actions included medication changes, referral-on to other services (for example, hospice day care) and advising on daily strategies to ease breathlessness (often described as `lots of little things’).

Explanatory analysis suggests it was not only the provision of these interventions that was important: how they were delivered was key to their impact. The data suggest they were delivered through the provision of knowledge, with specialist expertise, which increased patients’ and carers’ confidence. For example, some reported receiving handheld fans from other clinicians but identified the way BIS delivered this intervention as different: BIS explained how and when to use the fan and how it might work, so legitimising what at first appeared an unlikely effective intervention.

Thus, the mechanism of impact appeared to relate to improved knowledge, enhancing patients’ and carers’ understanding and their confidence in living with the symptom. BIS acknowledged, and thereby validated, breathlessness; patients and carers reported no longer feeling alone. Box 2 provides illustrative quotes for gaining knowledge and confidence, and some participant-identified interventions.

Reviewing transcripts for the categorisation exercise identified that more than half of patients had further contacts planned with BIS beyond the key measurement of the primary outcome: 52% (28/54) described planned contacts beyond t3. Thus, any improvement in the primary outcome at t3 was early impact and there may have been further benefit beyond this.

Table 4 shows contacts with services (face-to-face or telephone) and associated costs. At baseline most patients received hospital, general practitioner and informal care. Costs for some individual services were quite different between arms, but (in common with most other economic evaluations) zero costs were common and standard deviations high. Total costs at baseline were higher in the intervention arm. Between baseline and t3 more than half received hospital care (both arms) and fewer in the intervention group used inpatient or informal care (Table 4). Costs during follow-up were notably lower, mainly due to the shorter cost-period compared to baseline. Adjusting for baseline costs, the intervention group had health/social care costs on average £211 less than controls (95% CI, - £918 to £310), decreasing to £182 and £154 when intervention costs were increased by 25% and 50%, respectively (intervention remained dominant). Total costs (including informal care) were £354 less for the intervention group (95% CI, - £1020 to £246).

Lower health/social care costs and better primary outcome results for the intervention arm indicated dominance over the control condition. Cost-effectiveness planes showed a 66.4% likelihood that the intervention would have lower health/social care costs and better outcomes in terms of reduced distress due to breathlessness than the control condition and a 30.9% likelihood of higher health/social care costs and better outcomes (likelihood of lower costs and worse outcomes 1.4%; likelihood of higher costs and worse outcomes 1.3%). The figures for total costs (including informal care) were 80.9% and 16.4%, respectively (likelihood of lower costs and worse outcomes 2.2%; likelihood of higher costs and worse outcomes 0.5%). The maximum QALY-gain over a two-week period was 0.038. The intervention resulted in an incremental QALY-gain of 0.0002 (95% CI, -0.001 to 0.002). While technically indicating dominance for the intervention, the difference was small and variation substantial. This is reflected in the cost-effectiveness plane showing BIS having a 40.5% likelihood of lower health/social care costs and a greater QALY-gain than the control condition, a 21.4% likelihood of higher costs and a greater QALY-gain, a 27.3% likelihood of lower costs and a smaller QALY-gain and a 10.8% likelihood of higher costs and a smaller QALY-gain. With informal care included the figures were 50.9%, 11.0%, 32.2% and 5.9%, respectively. The cost-effectiveness planes are available as supplementary material [see Additional file 1: Figures S1 to S4].