A quasi-randomised, controlled, feasibility trial of GLITtER (Green Light Imaging Interpretation to Enhance Recovery)—a psychoeducational intervention for adults with low back pain attending secondary care

Introduction

Patients with low back pain (LBP) attending secondary care settings can be characterised by high pain intensity, poor function and higher frequency of poor prognosis than patients attending primary care (Karran et al., 2017b; Morsø, et al., 2013; Morsø, et al., 2014). Consequently, their contribution to the massive burden of LBP (Vos et al., 2016) is likely to be significant. General practitioners (GPs) frequently refer their patients to specialist secondary care clinics for surgical opinion (ABS, 2015; Coulter, 1998)—however surgery is recommended for only a small proportion of cases (Li & Yen, 2010; Robarts et al., 2017). For the remainder, clinicians are generally challenged to provide time-efficient, guideline-based, conservative care recommendations. Screening tools have been used to determine risk and optimal pathways for back pain patients in primary care, but systematic review data (Karran et al., 2017a) are underwhelming and they seem to offer little benefit within the context of secondary care (Karran et al., 2017b).

There is widespread current agreement in LBP clinical practice guidelines that clinicians should deliver high quality information as a key component of their management (O’Connell et al., 2016; Wong et al., 2017). In particular, patients should be reassured that their condition is not likely to be serious, that a favourable outcome is usual and that activity levels should be resumed as soon as possible (NICE, 2016; Koes et al., 2010; Qaseem et al., 2017; Stochkendahl et al., 2017). While patient reassurance can be considered a central tenet of this approach, practical guidance for how to effectively deliver reassurance in the clinical setting is lacking (Hasenbring & Pincus, 2015; Traeger et al., 2015). Furthermore, the impact of providing reassuring information on pragmatic outcomes appears to be poorly understood (Pincus & McCracken, 2013; Traeger et al., 2017).

Counter to clinicians’ efforts to reassure, the communication of spinal imaging findings have been suggested to have the potential to increase patients’ fear of re-injury and reduce their likelihood of a good outcome (Roland & Van Tulder, 1998). Adverse effects of early imaging of the lumbar spine have also been reported (Graves et al., 2012; Webster & Cifuentes, 2010). That most LBP patients attending secondary care present with spinal images (Sears et al., 2016) and that these images are routinely considered during the consultation, provides opportunity for careful consideration of how imaging results are interpreted and communicated in this setting.

Recent evidence reveals a high prevalence of common degenerative features in the imaging reports of asymptomatic adults (Brinjikji et al., 2014). This indicates that many of these features—particularly when found incidentally—should not be considered pathological and instead be regarded as normal age-related changes. Coherent with this interpretation is that degenerative findings have been found to be poorly associated with current pain (Brinjikji et al., 2014; Carragee et al., 2006; Suri et al., 2014) and prognosis (Carragee et al., 2005; De Schepper et al., 2016; Jensen et al., 2014; Steffens et al., 2014). Considering this evidence, we developed and tested an intervention framework based on using imaging findings for clinical benefit (Karran et al., 2018). This psycho-educational intervention, referred to as ‘GLITtER’ (Green Light Imaging Intervention to Enhance Recovery), involves a standardised method of communicating radiological findings in a manner designed to reassure patients and promote engagement in an active recovery.

A feasibility trial, conducted in the spinal outpatient setting at an Australian metropolitan hospital, is the crucial first step towards definitive testing of this intervention. The primary aim of this study was to determine the feasibility of recruitment and retention, assessment procedures, implementation and acceptability of the GLITtER intervention for LBP patients attending a spinal outpatient clinic. Secondary objectives included the identification of modifications needed in the design of a larger effectiveness trial and provision of data to enable calculation of an appropriately powered sample for a subsequent effectiveness trial.

Methods

Procedure

The full study procedure, including preliminary and primary screening procedures to determine patient eligibility for participation, is outlined in Fig. 1.

Confirmation of participant eligibility

At completion of the SAC assessment, all clinicians completed a 5-item checklist to confirm final eligibility for participation in this study. The intention of this checklist was to exclude patients who required subsequent investigations, interventions or surgical intervention, or patients who had pathology requiring significant caution with activity. It also confirmed that patients had functional or exercise restrictions, such that an intervention to promote activity was warranted. Participants meeting all criteria were notified of their inclusion in the study and were informed regarding subsequent follow-up procedures.

Results

Achievement of feasibility criteria

Achievement outcomes of the pre-specified feasibility are provided in Table 2 and detailed below.

Process considerations

A total of 31 participants were recruited from 15 clinic sessions. A CONSORT flow diagram detailing study recruitment is provided in Fig. 2. Of the 101 patients who met the primary eligibility criteria, 24 patients declined involvement. This resulted in a consent rate of 75%, which was considered acceptable according to our pre-specified criteria. In addition to this however, 8% did not complete signed consent forms due to unspecified reasons. An average of two participants were recruited each clinic session, which was much less than our target recruitment rate of four participants per clinic session. Further consideration and revision of recruitment feasibility is required. Acceptable participant follow-up response rates were also not achieved, with 1 and 3-month outcome data available for 36% and 42% of participants respectively.

Resource considerations

Participants completed 95% of all data items at baseline. 96% of items on the Recommended Minimum Dataset were completed and 92% of items on the TSK-11 were completed. Clinician burden was evaluated by the duration of the GLITtER consultations in comparison with standard consultations. 54 SAC consultations (11 GLITtER consultations and 43 Standard Consultations) were timed by clinicians 1 and 2. The mean duration of the GLITtER consultations was 41 min (SD = 7, minimum 30, maximum 50). The mean duration of standard consultations was 40 min (SD = 9 min, minimum 25, maximum 70). Three of the standard consultations exceeded 50 min. If these consultations were excluded, the mean duration of the Standard Consultations was reduced to 39 min (SD = 7).

Management considerations

100% of responses on the participant experience questionnaires (for participants who received a GLITtER consultation) were favourable (i.e., participants responded to the satisfaction statements with ratings of “agree” or “strongly agree”). Clinicians 1 and 2 both reported that they were “confident” that they could integrate GLITtER into standard practice on an ongoing basis (on a 4-point Likert scale ranging from “not at all confident” to “very confident”). One clinician was “very confident” and the other was “confident” that integrating GLITtER into routine consultations would enhance SAC care.

Scientific considerations

Based on a linear mixed effects model with two time points (baseline and 3 months), power = 80%, Type 1 error = 5%, an expected correlation between baseline and 3 months measurements of 0.5, a MCID of 1, and a SD of 2.1, then 53 patients are required per group. Allowing for 15% loss to follow-up, a sample size of 63 participants per group is required for an appropriately powered RCT.

Mean pain, disability, and kinaesiophobia scores at baseline and 3 month follow-up are provided in Table 4. We did not calculate change scores or conduct exploratory analysis of between-group differences (as specified a priori) due to the small sample sizes achieved in each group. No ceiling or floor effects were observed.

Table 4:

Baseline, 3 month follow-up and change scores for pain, disability and kinaesiophobia.

	Baseline scores		3 month follow-up
	GLITtER mean (SD), N	Standard mean (SD), N	GLITtER mean (SD), N	Standard mean (SD), N
Pain (NRS)	6.2 (1.8) N = 19	6.5 (2.0) N = 7	6.5 (2.0) N = 11	5.3 (2.1) N = 6
Disability (NRS)	5.9 (1.8) N = 20	4.2 (2.8) N = 7	5.6 (2.5) N = 11	4.2 (2.9) N = 6
Kinaesio phobia (TSK 11)	40.6 (7.1) N = 13	38.9 (7.2) N = 5	36.7 (7.3) N = 10	30.0 (7.8) N = 5

DOI: 10.7717/peerj.4301/table-4

Notes:

Abbreviations

SD

standard deviation

NRS

numeric rating scale

TSK

Tampa Scale for Kinaesiophobia

Discussion

This investigation has highlighted some important operational challenges impacting the feasibility of successfully conducting a future effectiveness trial of GLITtER (Green Light Imaging Interpretation to Enhance Recovery). While pre-specified targets for consent rates, resource burden and acceptability of the GLITtER intervention were met, participant recruitment was slower than anticipated and an acceptable follow-up rate was not achieved.

Recruitment of participants into the study occurred at half of the anticipated rate, due largely to fewer patients meeting the final eligibility criteria than expected. An identified contributor to this was the change to referral management processes such that patients who were unlikely to be candidates for intervention were no longer offered SAC appointments. (Instead, these patients and their General Practitioners were informed via a letter that community-based management was most appropriate.) Although the number of patients attending the SAC who were eligible for this study was low, this may not indicate that the number of individuals suitable for the GLITtER intervention is also low. Instead, it may indicate that there is need to further consider how best to access the patients for whom GLITtER may offer potential benefit. Alternatively, plans to further investigate the GLITtER intervention in the spinal outpatient setting must carefully consider patient triage procedures, allow for slower recruitment, and investigate the potential for recruitment at multiple sites. The wording on the information sheet should also be reconsidered to avoid the risk that some patients may have declined participation out of concern that they would be given reassuring information during their consultation, rather than the information that they were seeking.

We achieved a 3-month participant follow-up rate of only 42%, which failed to meet our feasibility criterion. In hindsight, our objective to achieve complete follow-up data from 95% response rate was high—a drop-out rate of more than 20% is generally considered to compromise the validity of a clinical trial unacceptably (Schulz & Grimes, 2002). Possible reasons for the high drop-out rate may have included participant improvement (such that LBP was no longer a significant concern), or patient perceptions of a lack of relevance of the intervention or the follow-up requests. The protocol for collecting follow-up data via postal questionnaires and SMS reminders was based on the procedure applied in our previous work, which achieved a 4 month follow-up rate of 89% (Karran et al., 2017b). There were two main differences in the follow-up conditions between these studies. Firstly, our prior study involved participants while they were still waiting for their clinic appointments to be scheduled which may have facilitated engagement, whereas follow-up for the current trial occurred after patients had received their clinic consultation. Secondly, for our initial study, we sent small packets of confectionery with requests for completion and return of questionnaires, whereas in the current trial we did not. Future investigations in this setting should further consider developing an enhanced protocol for achieving follow-up targets and providing participant incentives (Brueton et al., 2013). It may also be important to consider revision of the inclusion criteria to ensure that the patient’s problem is of sufficient severity to warrant engagement with the study. Unfortunately, the poor response rate attained in this study impacted the achievement of our secondary objectives. Collected data were used to cautiously inform the determination of sample size for a larger RCT, but was considered insufficient to undertake exploratory analysis of effect.

The GLITtER intervention was acceptable, of perceived benefit to both patients and clinicians, and was able to be integrated into a standard consultation without significant time burden for clinicians. The 95% data completion rate on the recommended minimal dataset for LBP research (Deyo et al., 2014) suggests that despite its length, it is not overly onerous for participants. The consent rate of 75% exceeded our target rate, however it must be acknowledged that a further 8% of patients who met the primary eligibility criteria did not either confirm or decline consent. Whilst a number of these exclusions were likely to be related to procedural or timing issues it must be also considered that some of these patients may have opted out of the study (without this action being recorded)—with impact on the overall consent rate.

This study employed rigorous scientific methods throughout. The protocol was developed in accordance with the SPIRIT checklist and pre-registered on Open Science Framework. The trial was registered on the ANZCTR and has been reported according to the CONSORT statement. Description of the GLITtER intervention followed the TIDieR recommendations for reporting of clinical interventions and the NIHs task force recommended minimum data set for LBP clinical trials was implemented. The specification of feasibility criteria prior to conducting this investigation is a further strength of this study’s design.

Several weaknesses have also been recognised. There are no standardised criteria for evaluating feasibility, or how achievement of the established criteria should be measured. We used an evidence-informed but ultimately investigator-led approach to identify and quantify our feasibility criteria, which may not have been optimal. Factors such as participant adherence to management recommendations and investigator burden could also have been considered. In addition, we consider that the ability to engage clinicians and administrative personnel is imperative to the success of a clinical trial but we did not formally assess engagement. Our sample size and our follow-up rate were smaller than we expected, which resulted in less data to evaluate our feasibility criteria than we specified as a requirement a priori. The study sample may therefore not be representative of the larger patient population, raising the risk that estimates or interpretations that arise lack precision or are potentially misleading. Last, staff training was not formalised and treatment fidelity was not assessed. While we did calculate an estimated sample size required for a RCT to definitively examine the effect of the GLITtER intervention, this result should be interpreted with caution. Encouragingly, the baseline value for the pain NRS score and the standard deviation of this measure at 3-month follow-up, were very similar to the data recorded in another LBP feasibility trial on which a sample size estimate was also based (Ellard et al., 2017). Based on the assumption that two participants from each SAC would meet final eligibility criteria for inclusion in the study and be randomised, and that clinics operate on 46 weeks of the year—it appears realistic that 92 participants could be recruited annually. Timely achievement of a recruitment target of 126 is likely to be possible, and potential also exists to consider enrolment of participants at two or more sites.

This study successfully tested for feasibility of a highly pragmatic approach to addressing the clinical challenge of providing time-efficient, low cost, guideline-informed management of patients with LBP attending secondary care. Based on analysis of our feasibility criteria, a full-scale RCT evaluating the effectiveness of the GLITtER intervention is not recommended without considering strategies that are capable of resolving the recruitment and follow-up issues that we have identified. The shift in clinical pathways raises the possibility that testing of GLITtER in primary care may be indicated. These issues are likely to affect other trials in this population and as such, we humbly suggest that the lessons we have learnt be considered recommendations for the field.

Conclusions

Failure to achieve pre-specified recruitment and follow-up rates was disappointing, however these were important outcomes of a feasibility study. It is realistic to consider that these challenges can be overcome through careful strategy, ample funding and continued partnership with health care providers.

Supplemental Information