HAP-FAST: a feasibility study incorporating qualitative, mechanistic and costing sub-studies alongside a randomised pilot trial comparing chest x-ray to low-dose CT scan and empirical antibiotics to antibiotics guided by the BIOFIRE(R) FILM ARRAY(R) pneumonia plus panel in adults with suspected non-ventilator-associated hospital-acquired pneumonia

Participants and sample size

Since the aim is to assess feasibility, a sample size justification is given rather than a calculation. We aim to recruit 220 adult participants based on prospective audits of HAP in the UK Northwest revealing between 600 and 1000 cases per year at each of our recruiting sites and assuming 30% of cases are eligible of whom 40% are recruited to the trial. Recruitment targets will likely be affected by the seasonality of HAP, with a greater burden in winter and seasonal variation in pathogens, and thus, we aim to recruit across the majority of a calendar year.

HAP is potentially severe as evidenced by the in-patient mortality of 24%. NICE-recommend treatment is commenced within 4 hours. Clinicians therefore face a narrow timeframe during which patients must be clinically assessed and diagnostic tests must be ordered, completed, reported, interpreted and acted on. Patients with nv-HAP frequently have impaired mental capacity due to underlying cognitive impairment or acute delirium. Therefore, due to the emergency nature of HAP, in common with research in other emergency settings such as trauma and intensive care, HAP-FAST uses a Research Without Prior Consent (RWPC) model.24–26 The use of RWPC for nv-HAP trials has been studied previously and deemed acceptable by patients and the public.26

At the point of suspecting nv-HAP, treating clinicians at the recruiting sites can randomise, carry out the interventions and obtain the initial sample set. Randomisation leads to an automatic email alerting the site research team who then obtain written informed consent from the patient or for those lacking capacity from a personal or professional proxy before discharge. Every effort will be made to obtain written informed consent after discharge if a patient is discharged before consent is obtained. Patients who decline to provide consent or no longer wish to continue in the study will be withdrawn. Data collected up to the point of withdrawal will be included in the analysis, and permission will be sought to collect data from routine assessments to complete some outcome data.

Pilot study eligibility criteria

Eligibility criteria for stage 1 randomisation to CXR versus CT and stage 2 randomisation to FAPP or no FAPP can be seen in table 1. Patients who are ineligible for randomisation to stage 2 will still be able to participate in the trial.

Interventions and treatments

Participants are initially randomised between a standard-care CXR and low-dose, non-contrast, thoracic CT scan. If the clinician decides to give antibiotics to treat nv-HAP and the participant can produce a sputum sample before the administration of the second dose of antibiotics, they are further randomised between sputum testing by FAPP alongside local, standard of care microbiological processing or standard processing alone—no FAPP. A study-specific antibiotic guideline has been produced and approved by all recruiting sites for use with the results of the FAPP. It is anticipated that patients randomised to standard microbiological testing will receive an empirical antibiotic prescription supported by usual microbiological tests. Additional advice regarding the antibiotic treatment is available from microbiology specialists in line with local policies. Participants who cannot provide sputum and who are not randomised at stage 2 will be managed as per usual care. These interventions are summarised in table 2 and figure 1.

Figure 1

Pilot sequential multiple assignment randomised trial design.

Outcome measures

A key objective of HAP-FAST is to gather data to inform the choice of outcome measure for a fully powered RCT. We searched the COMET database for core outcome sets in HAP trials.27 Some groups advocate all-cause mortality assessed on a non-inferiority basis.28 However, others argue that discerning the mortality attributable to HAP, as opposed to underlying comorbidity, is difficult without unfeasibly large trials.29 One group proposed a hierarchical, composite, primary outcome of survival at day 28 and ‘clinical cure’ between days 7–10 but unfortunately did not provide a pragmatic definition of clinical cure.30 A group convened by the FDA suggested using mortality plus resolution of symptoms.31 HAP-FAST will therefore evaluate a range of outcomes including mortality, antibiotic usage and clinical cure incorporating a pneumonia-specific patient-reported outcome measure called the CAP-SYM score.

Pilot study randomisation

The pilot study has been designed as a sequential, multiple assignment, randomised trial (SMART) with a 1:1 allocation ratio, with the purpose to address study objectives 1–5.32 The randomisation list has been created by an independent statistician, and participant allocations are generated by completion of the web-based randomisation platform. The SMART study design is presented schematically in figure 1.

Pilot study blinding

The study is open-labelled, and treating clinicians, researchers and participants will know which intervention is being administered via the web-based randomisation process.

Pilot study outcome measures and participant timeline

Baseline and outcome data are collected at distinct time points according to the schedule in table 3 and in the supplementary table 4. Participants will be assessed by the study team daily until day 10 to track symptomatic recovery, changes in the quality of life (QOL) and determine time to clinical cure. Participants will have symptoms and QOL assessed on day 28 as an in- or out-patient. Follow-up will be conducted as a phone call 90 days (±14 days) following entry into the study to assess symptoms and QOL and to remind them to return a survey booklet on health and social care use up to day 90.

Pilot study data analysis

All analyses will be carried out on an intention-to-treat basis, retaining all participants in their initially randomised groups, irrespective of any protocol deviations. The focus of analysis will be to assess the feasibility and recruitment for each participating site and overall pilot study as well as assessments of efficacy for each outcome for treatment arm comparisons of CXR versus CT (figure 1, group 1–4 vs group 5–8) and FAPP versus no FAPP (figure 1– group 1+5 vs groups 2 and 6). No inference will be drawn—all results will be treated as hypothesis generation.

Continuous data will be presented using median (IQR) and mean (SD) as appropriate, with boxplots summarising measurements at each time point by the treatment group. Categorical data will be presented as frequencies and percentages. Time-to-event data will be presented with Kaplan–Meier curves and summarised by median (95% CI) if possible.

As much information as possible will be collected about the reasons for missing outcome data; this will be used to inform any imputation approaches employed in the analysis. Such methods will be fully described in the full statistical analysis plan, which will be written before conducting any comparative analysis of the treatment arms, including methods employed for missing data.

Clinicians

This qualitative sub-study will address objectives 5, 7, 8 and 10 to evaluate human factors involved in the delivery of the study, clinician willingness to recruit participants and adherence to antibiotic guidelines as per the study protocol (table 3).26 33 A range of clinical, allied health professional and research staff will be invited to participate in focus groups of approximately eight participants. Focus groups will be topic guided, yet conversational and exploratory and conducted in a comfortable private environment.

Patients and carers

This qualitative sub-study will address objectives 9 and 11 to evaluate patient willingness to participate in the study and their experience from recruitment to study follow-up (table 3).34 Approximately 15 participants (five from each of the three recruiting trusts) will be purposively recruited for in-depth semi-structured interviews based on age, gender and underlying comorbidity class (medical admission, surgical admission, acute admission). Relatives and carers of some study participants will also be interviewed.