Aims and objectives
The aim of this pilot trial was to assess the feasibility of a definitive RCT evaluating the clinical and cost effectiveness of a health screening clinic compared to usual care, in reducing sickness absenteeism and presenteeism amongst NHS staff.
The key objectives were to:
describe recruitment rates
describe participant characteristics and assess generalisability compared to the hospital workforce
describe intervention screening assessment results
describe recommended referrals and attendance at those referrals
assess the feasibility of measuring the outcomes relating to the definitive trial, and obtain an estimate of the standard deviation (SD) of the proposed primary outcomes for a full
RCTassess levels of contamination between intervention and usual care arms to inform
RCT design (individual vs. cluster RCT)
describe and explain the fidelity to the intervention and evaluate the barriers and enablers to participation
describe the views, experiences and satisfaction of participants and those delivering the health checks
explore feasibility of the trial/service in other NHS and non-NHS settings
quantify the costs of undertaking the screening service and its consequences.
Trial design
A multicentre, two-arm, parallel group, open-label, 1 : 1 individually randomised pilot RCT of a complex intervention comparing a SHSC with usual care.
Participants
Setting
The trial was conducted in three large urban hospitals and one rural district general hospital within three NHS Hospital Trusts in the West Midlands (see Table 1), allowing the testing of practical aspects in a range of sites.
Description of study sites
Eligibility criteria
All employees in the participating hospitals who were able to provide informed consent were eligible to participate except those who had previously attended a pilot health screening clinic at QEHB or who were currently taking part in another drug trial (with the exception of COVID-related drugs/vaccines) or health and well-being trial.
Recruitment
Recruitment occurred in several phases, aiming to reflect the workforce characteristics of each hospital. Anonymous human resources (HR) data providing sociodemographic and occupational characteristics were obtained to choose wards which reflected the characteristics of the whole hospital population. Posters advertising the trial were displayed in departments/wards for a minimum of 2 weeks (or more if local arrangements required) before invitations were sent out, during which time staff could request not to receive an invitation. In order to be accessible to the full workforce, participants were informed of the trial through staff meetings, noticeboards and ward champions, and invited through multiple approaches including e-mail and personal invitation. Reminder letters were sent to non-responders after 2 weeks, with up to two further reminders if necessary. Local staff were engaged to raise trial awareness and arrange cover to allow attendance at the health screening clinic.
Consent
Potential participants joined the trial via a weblink to a custom-designed electronic trial data collection platform hosted by the Birmingham Clinical Trials Unit (BCTU). Those with limited access to the internet could complete a paper-based expression of interest through the internal mail system, then attend the clinic to complete consent and other forms online. Consent was obtained through the trial platform, via participants’ own electronic devices or using tablets provided to the health screening clinic staff, and ongoing consent was confirmed at each follow-up point. The participant information sheet (was provided with the initial invitation and was also available on the trial platform). Participants could contact the research team by telephone or request contact from the research team via the trial platform to ask any questions.
Randomisation and blinding
On completion of the baseline questionnaire, clinic staff received an alert to check eligibility criteria and consent, and participants were then randomised to intervention or usual care using an integrated randomisation module on the trial platform. Randomisation was at the level of the individual in a 1 : 1 ratio, using a minimisation algorithm (with random element) to ensure balance on the following variables:
age (< 40, ≥ 40 years)
sex (male, female, prefer not to say)
job category (categories as self-reported)
nightshift work (yes/no)
hospital.
Given the nature of the intervention, blinding of participants and nurses conducting the screening clinic was not possible.
Interventions
Intervention arm
Participants received an invitation to a SHSC, available during or outside work time (generally 7 a.m.–5 p.m., Monday–Friday), administered by trained clinic nurses [hereafter known as enhancing the Health of NHS Staff (eTHOS) nurses]. A maximum of three attempts was made to contact participants to arrange an appointment. The screening appointment consisted of two stages: (1) screening assessment for three components: mental health, MSK health and finally cardiovascular health, followed by (2) advice and/or referral, according to level of risk, to appropriate services for management according to NHS/NICE recommendations (see ). The screening assessment was delivered by nurses, with consultant physician oversight at each centre, supported by a standardised protocol and prompts from the electronic data platform. All data from participants were entered directly onto the electronic platform in a paperless system.
The eTHOS nurses explained the screening scores to participants and, where applicable, any recommended actions. Local pathways were used; these included self-referral to the staff physiotherapy service, local mental health support (e.g. healthy minds, Frontline19), lifestyle changes as directed in appropriate leaflets and websites, advice to visit their GP, recommendations to the GP that a direct referral be made for the participant and/or a request that the GP arrange a follow-up appointment with the participant. Where there were concerns regarding safety, the eTHOS nurse would immediately inform the local site principal investigator (PI) who would then contact the GP.
Participants reporting relevant diagnosed conditions currently being treated did not receive the referral element of that component. A results letter detailing findings and recommendations was sent to participants and their GPs for appropriate action.
Mental health check
Participants self-completed the GAD-739,40 (anxiety) and PHQ-938,40 (depression) screening questionnaires to assess mental health and were categorised into three levels of risk (see ). Participants without significant anxiety/depressive symptoms were offered online mindfulness advice.61 Those who were moderately affected were advised to seek support from local counselling services such as Birmingham Healthy Minds,62 and those severely affected were referred to their GP for immediate treatment.
Overview of staff health screening intervention. From National Institute for Health and Care Excellence 2016.
Musculoskeletal health check
The Keele STarT Back Screening Tool41 was self-completed to categorise risk of future disability of low back pain into three levels (low, moderate and high risk). The STarT MSK tool (also three levels) was used for non-back complaints,46 and the Orebro musculoskeletal pain screening questionnaire (OMPSQ) score63 was also calculated to enable comparison with international studies. Based on a model of stratified care, those with moderate or high scores on the STarT Back or STarT MSK tools were referred to physiotherapy teams according to NICE guidance,42–45 either directly or via their GP. Those with high scores were recommended to receive enhanced physiotherapy including CBT to address associated psychological problems. In practice, in part, due to pressures of the COVID-19 pandemic, direct referrals and CBT were largely unavailable.
Cardiovascular health check
Participants aged 40 years and over received the NHS health check, delivered face to face by the eTHOS nurses.48 This included lifestyle checks with self-completed questionnaires and clinical measures: BMI, exercise level [general practice physical activity questionnaire (GPPAQ) questionnaire],64 smoking status, alcohol intake (AUDIT C questionnaire);65 QRISK2 score66 and clinical measures [pulse, blood pressure, electrocardiogram (ECG), cholesterol, glycated haemoglobin (HbA1c), estimated glomerular filtration rate (eGFR) and creatinine tests if appropriate]. Actions included referral to GP, weight management, smoking cessation and alcohol reduction services, as well as brief advice on exercise and diet according to UK recommendations (see ). The dementia awareness component for those over 65 years was excluded as there were known to be few employees in this category. Participants aged under 40 years were assessed for lifestyle components of the NHS health check only: BMI, exercise level, smoking status and alcohol intake, with the same advice applied and referrals to appropriate services. Clinical measures and associated blood samples were not collected from participants aged under 40 years.
Occupational health
All participants with an identified health condition were also asked whether they perceived that their condition was affected by/or affected their ability to work and what adjustments at work might improve their workability. If affected, they were offered an additional optional referral to occupational health.
Usual care
Participants randomised to receive usual care did not attend the health screening clinic. Usual care consisted of standard access to medical services for management of any presenting condition (either through occupational health or their GP) and any health promotion/counselling initiatives provided for staff during the pandemic. They remained eligible for the NHS health check via their GP as usual if they were aged 40 years or older.
Outcomes
This pilot trial was originally planned to have a 52-week follow-up to fully test the processes for a definitive trial but was reduced to 26 weeks due to the pandemic-related delays.
Primary outcomes and stop/go criteria
Three coprimary outcomes were originally planned to inform criteria to progress to the definitive trial:
We defined ‘referral’ as anyone who was eligible for a referral, that is, they recorded a suitable risk value and were not already being treated for that condition.
Due to additional pandemic-related delays, following discussions with the funder, it was agreed not to extend the duration of the trial in order to complete follow-up of all participants to 6 months. This meant that only a small number of participants reached the 26-week follow-up assessment, leaving limited data available to reliably assess the third coprimary outcome on attendance. After discussion with the Trial Oversight Committee (TOC), it was agreed that acceptance of the referral (signifying intention) which was collected during the screening assessment, should be used as a proxy for attendance.
The decision to continue to a full trial was informed by progression criteria using a traffic light system as guidance (see Table 2), with flexibility built in to allow for uncertainty in cost effectiveness despite either low participation, referral or attendance rates. The three coprimary outcomes formed stop/go criteria when considered together. If any of these values fell into the ‘amber’ zone, then the trial would require modifications to proceed to full trial. If all were ‘red’, then the trial would be considered unfeasible.
Progression criteria informing decision about full trial
Secondary outcomes were collected to inform the design of the definitive RCT and any modifications required:
Secondary outcomes
Comparison of baseline characteristics of included participants and hospital population for assessing generalisability.
Description of the results of the intervention screening assessments.
Number and type of referrals to recommended services (intervention arm only).
Attendance at each individual recommended service (acceptance of referral during screening assessment, self-report at 26 weeks; intervention arm only).
Lifestyle relevant to screening intervention advice and referrals (self-report at 26 weeks compared with baseline) including Physical Activity Index measuring exercise levels (GPPAQ),
64 smoking status, weight (kg).
Acceptability of intervention to participants and health screening clinic staff (interviews within the Process Evaluation).
Feasibility of trial processes (completeness of relevant data items, interviews).
Indication of contamination (comparing pre/post data for health behaviours and health care/other service utilisation in control arm).
Outcomes related to the definitive trial
Sickness absenteeism at 26 weeks with reasons, measured by days and spells:
self-report absolute absenteeism, relative absenteeism and relative hours of work – for last 7 days and last 28 days [World Health Organization Health and Work Performance Questionnaire (WHO-HPQ)]
69self-report absenteeism using 6-month recall period
employee records of absenteeism, which will be the primary outcome of the definitive trial. Routinely collected data from the NHS Electronic Staff Record Programme, linked to employee ID.
Presenteeism at 26 weeks [self-report absolute presenteeism and relative presenteeism for last 28 days (WHO-HPQ)].
69Attendance at occupational health service at 26 weeks (self-report).
Healthcare utilisation at 26 weeks (self-report) including
GP consultations and hospital admissions.
EuroQol-5 Dimensions, five-level version (EQ-5D-5L) index value measuring health-related quality of life (HRQoL) (EuroQol EQ-5D 5-level)
70 at 26 weeks.
Resource use and costs collected by self-report questionnaire to participants on health service utilisation and information from the screening assessment regarding duration and resources used.
Data collection and management
Data were collected at several points during the trial: eligibility screening, baseline, during the screening clinic and at the 26-week follow-up (see Table 3). Study data were entered into a custom-designed REDCap (Research Electronic Data Capture) database hosted by the University of Birmingham, either directly by participants or by clinic staff.71,72
Self-reported data were collected on all participants prior to randomisation. All respondents were provided with the option to provide reasons for taking/not taking part in the study. For those who consented to take part, we collected contact details and NHS number, where available (this was optional), to assess the feasibility of future linkage to routine data such as hospital admissions, GP records, other healthcare utilisation and mortality data. In the baseline questionnaire, we collected information on demography, employment details, selected diagnosed medical conditions and current medications, absenteeism (WHO-HPQ),69 presenteeism (WHO-HPQ),69
HRQoL (EQ-5D-5L),70 smoking status, height, weight, exercise levels (GPPAQ),64 receipt of NHS health check and health service utilisation. Where the baseline remained incomplete, prompts were sent after 24 hours, 2 and 7 days, and then a final reminder giving participants the option to complete it at a later date.
At 26 weeks, participants were sent a follow-up questionnaire for online completion to report attendance at any recommended services (intervention arm only), and other outcomes as detailed above. Where follow-up questionnaires remained incomplete, one reminder was sent.
Consent was obtained from trial participants to collect the following additional linked data from HR records:
participant age, sex, ethnicity, staff group and staff grade
number of hours contracted to work
hours worked – full-time equivalent
sickness absenteeism and non-sickness absences relating to COVID-19 only, for the previous 24 months and from randomisation to 26 weeks follow-up (start/end dates, duration in days, recorded reason)
leaving date (should participant leave employment of the Trust during trial participation).
Additional anonymised data were collected at the time of site set-up from hospital HR records, at whole hospital level and for those invited, on age, sex, ethnicity, job role and number of days absent in order to assess the generalisability of the included participants.
Key data collection forms and questionnaires are available on the National Institute for Health and Care Research (NIHR) Funding and Awards website https://fundingawards.nihr.ac.uk/award/17/42/42.
Sample size
As the eTHOS trial was a pilot study, no formal sample size calculations were undertaken. The study was not designed or powered to detect a statistically significant difference in efficacy between the two trial arms. However, preliminary sample size calculations were computed for the definitive RCT. To detect a clinically important effect size for days of sickness absence of 0.1 of a SD would require 4200 participants for an individually randomised trial.
A relatively large number of participants were required for this pilot trial in order to achieve reliable estimates of the referral and attendance rates downstream along the screening pathway and to allow assessment of the intervention in different hospitals, better precision in determining the SD of the primary outcome and provision of health economic information required to persuade other hospitals to consider investing in participating in the definitive trial. We aimed to recruit 480 participants (20 per week) in 24 weeks. The target recruitment for each site differed depending on their total staff number and capacity to deliver the intervention: 170 each for QEHB and Heartlands Hospital, 93 for Birmingham Children’s Hospital and 47 for Hereford Hospital. With this sample size, the 95% confidence interval (CI) for the proportion of staff recruited could be estimated to be 4% either side of the estimate (e.g. for a 25% recruitment rate, the 95% CI would lie between 21% and 29%).
Statistical methods
The primary comparison groups were composed of those randomised to the health screening (intervention) arm and those randomised to the usual care (control) arm. All analyses used the intention-to-treat principle, that is, all participants were analysed in the treatment group to which they were randomised, irrespective of adherence or other protocol deviation. Analyses used available data only, although the number of missing data is reported in order to inform decisions regarding data collection for the definitive trial.
Primary outcomes
Recruitment rate was calculated in two ways:
Referral to a service was calculated as the percentage (with 95% CI) of participants randomised to the intervention arm that were eligible for a referral to any of the specified services (GP, physiotherapy, community psychological services). Participants were eligible for a referral if they met the score requirement for any of the mental health, MSK or cardiovascular health check assessments and were not already being treated for the referring concern.
As there were few follow-up data on self-reported attendance at referrals (because of resources restrictions forcing early trial closure), attendance at any recommended service was estimated by self-reported ‘intention’ (as agreed with the TOC). During the intervention screening, participants were given the option to accept or reject a referral to any service. The estimate of attendance at any recommended service was then determined by the proportion of participants in the intervention arm eligible for a referral that accepted that referral and presented as a percentage with 95% CI. Data from the 26-week follow-up questionnaires are provided in Appendix 2.
Secondary outcomes
Analyses were mainly descriptive, summarising binary outcomes using number of responses with percentages, and continuous outcomes with means and SDs, or medians and interquartile ranges as appropriate. No statistical modelling was undertaken, and no p-values are reported. No subgroup analyses were planned for this feasibility study.
Process evaluation
A mixed-methods process evaluation explored programme reach, fidelity of screening delivery, attendance at referrals and participants’ views of the intervention to support any modifications required for the design of the definitive trial.
Quantitative data collection and analysis
Quantitative data to support the process evaluation were obtained from:
Recruitment and intervention data, for example, response rate; proportions of those invited consenting, being randomised and attending the screening.
Baseline questionnaire to assess characteristics, for example, age, ethnicity, marital status, Index of Multiple Deprivation, educational attainment and employment role of the employees recruited to study (programme reach). We also obtained hospital-level
HR data to compare the characteristics of those invited and those consenting with the whole hospital populations.
Logs kept by the staff health screening programme recording attendance, trial database records of screening tests undertaken and duration of contacts (fidelity).
Healthcare issues identified at the employee screening, and referrals made to
GP and other services (fidelity).
Self-report of attendance, and intention to attend at recommended services from intervention screening and 26-week follow-up questionnaire (uptake).
Brief questionnaire at eligibility stage to staff not taking up the offer of the study to ascertain reasons for not participating (reach).
Findings were presented with descriptive statistics.
Qualitative data collection and analysis
Aims
The overall aim of the qualitative process evaluation was to obtain the views of relevant stakeholders on the feasibility and acceptability of the staff health screening intervention and trial and inform any protocol adaptations needed for a full trial. This was done by interviewing: NHS staff who received the trial intervention (trial participants); the eTHOS nurses who were delivering the intervention; providers of follow-on services, for example, GPs and ‘Healthy minds’; and outside agencies who could potentially benefit from the intervention, for example, other NHS organisations who might be willing to run the trial such as ambulance services and GP groups, and non-healthcare organisations who might be interested in delivering and evaluating such a service for their setting.
Objectives
Specific objectives of the interviews were to obtain:
trial participants’ views and experiences:
views and experiences of the health screening – the concept of health screening and their experiences of it
views about being offered the screening and whether or not it has enabled them to access health screening that they may not otherwise have accessed
views and experiences of the trial processes
evidence of contamination amongst the usual care arm
acceptability of being randomised to usual care
barriers and facilitators to accessing/attending the screening intervention
views and experiences of booking the appointment
views and experiences of screening location
reasons for not attending
views on confidentiality in relation to the screening and the potential for findings to be used by an employer
views on the role of occupational health services
response to referral recommendations.
research staff views and experiences:
any contextual differences in delivery of the intervention
satisfaction in relation to the training received
acceptability of the intervention in theory/practice and experiences of delivering the intervention.
providers of follow-on services:
acceptability of health checks
experience of receiving referrals, the acceptability of the process and impact of referrals on capacity
availability of appropriate referral pathways.
outside agencies’ views of potential feasibility of service:
Participants and recruitment
Due to the COVID-19 pandemic-related delays and forced truncation of the trial, recruitment for the process evaluation took place over 4 months (July–October 2021), instead of the planned 14-month period. All trial participants were eligible for interviews if they had agreed to be approached for the qualitative study. Those eligible were invited using their preferred methods of contact and the study information sheet either e-mailed or posted to them. Where recruitment targets had not been reached, one reminder was sent to those who had not yet responded.
We contacted representatives from providers of follow-on services (e.g. GPs or physiotherapists) to whom trial participants had been referred either directly or using contacts known to the trial team.
We also contacted a range of relevant outside agencies where staff health screening might be of interest in the future. This included those providing services for the hospitals involved in the trial, non-hospital-based NHS organisations, organisations independent of the NHS concerned with health care and local large businesses completely unrelated to health care as suggested by the trial team. Study information sheets were e-mailed to potential interviewees once an appropriate person within the organisation had been identified.
Participants were given the option to complete their consent form online or on paper (and post it back to the substudy team). When completing their consent, trial participants could also agree to the qualitative team contacting their GP, declining did not exclude them from participating in the 1 : 1 interview. Participants were permitted to withdraw up to 2 weeks after the interview.
Sampling
We aimed, where possible, to recruit a purposive, maximum variation sample of trial participants with a broad representation of age, gender, ethnicity, occupation, site, shift pattern and department who had had different experiences of the trial (see Table 4). The other interview groups were chosen pragmatically. Table 4 also includes details of our target numbers for recruitment, and the time frame for recruiting from the different groups as originally planned, with reasons. Regarding site, we also aimed to match the distribution of the target recruitment numbers using the ratio outlined in Table 5.
Substudy data collection targets
Target recruitment ratio from each site
However, due to COVID-19, this plan could not be adhered to and a pragmatic approach was taken instead. All eligible participants in the intervention arm were invited to participate, and invitations were sent to all controls until the target number of participants had been reached.
Data collection
Trial participants were invited to 1 : 1 focused qualitative interviews lasting approximately 30 minutes, with a choice of either zoom, telephone or face-to-face interviews if COVID-19 restrictions allowed. Other participants were offered the choice of 1 : 1 interview or a group interview with their colleagues if they preferred. Participants who did not attend screening (see Table 4) were also invited to respond to written questions via e-mail or post (see NIHR Funding and Awards website https://fundingawards.nihr.ac.uk/award/17/42/42). Semistructured topic guides (see NIHR Funding and Awards website https://fundingawards.nihr.ac.uk/award/17/42/42) guided interview conduct and were refined iteratively following initial interviews.
Data were recorded either on Zoom or an encrypted recorder and transferred via secure file transfer to an external company for transcription. Interviews were transcribed intelligent/clean verbatim and anonymised. All video files were destroyed as soon as transcription was complete.
Analysis
A thematic analysis of content was informed by the framework analytical approach.73 Analysis and discussion included the experienced qualitative team: RR, a qualitative researcher; KJ, a GP and public health consultant with experience of mixed-methods research; KR, a nurse and epidemiologist with experience of mixed-methods research; RLA, a nurse and qualitative researcher in primary care and public health; CP, a non-clinical qualitative researcher in primary care; CH, a physiotherapist postgraduate with experience of mixed-methods research; RJ, an epidemiologist with experience of mixed-methods research; PA, a public health consultant with experience of mixed-methods research; with input from the patient and public involvement (PPI) and trial oversight and collaborator’s groups to provide multiple perspectives on the data. Data collection and analysis ran concurrently so that emergent analytical themes could inform further data collection.
As soon as two trial participants’ interviews had been completed from groups B–D (see Table 4), the data collection team (RLA, CH and CP) each independently coded the transcripts using NVivo 12 (QSR International, Warrington, UK). The team then met to discuss and compare codes and agree a combined, draft coding index. The revised draft was shared with the wider qualitative team (RR, KJ, KR, RJ, PA) and further refined. The data collection team then selected two new transcripts on which to independently test the revised index; the index was further refined and the rest of the transcripts coded.
Transcripts from the other participant groups were coded using the same index with minor modifications. RLA (a nurse) collected the data for and coded the eTHOS nurses’ transcripts. CH (a physiotherapist) collected the data for and coded the physiotherapist transcripts with support from RLA. PB (non-clinical) collected the data for and coded the transcripts from the outside agencies with support from RLA. RLA collected the data for and coded the single GP transcript.
The codes were then exported to a single framework file and the data were summarised by the participant group by RLA and CP. Emergent themes were discussed with the wider qualitative team throughout.
Health economic analysis
Data collection focused on estimating the NHS resources required for screening and resulting costs and measuring the baseline quality of life of all participants using EQ-5D-5L. A descriptive analysis was then undertaken presenting information on the average costs of screening per participant and quality-of-life outcomes.
Resource use
All participating centres were contacted to collect information on their estimated time taken for each screening clinic task, the grade of the member of staff responsible and the cost of the individual blood tests. Responses were collated and a typical duration for each task was agreed upon by the study team. The amount of staff time required for each screening task was then multiplied by the relevant unit costs obtained from standard sources74 (see Table 6). Unit costs are shown in Tables 6 and 7.
Unit costs for staff roles
Costs
The cost of screening included the cost of staff time to undertake all tasks and the cost of blood tests. The total cost of blood tests per participant was weighted taking into account the proportion who received each type of blood test. Those tasks that were directly related to processing and reporting blood tests were only applied to the proportion who had at least one blood test. Costs were only applied to those who actually attended screening.
Certain tasks could be performed by healthcare professionals of different pay bands (e.g. a nurse, band 5; a nurse specialist, band 6; or a healthcare assistant, band 2 or 3). Analysis was undertaken so that the cost of screening per participant could be presented for all possible combinations, including the highest possible cost (where all tasks are carried out by the higher potential pay bands) and the lowest possible cost (for the lower pay bands).
Quality of life
The EQ-5D-5L questionnaire, containing five dimensions (mobility, self-care, usual activities, pain and discomfort, anxiety and depression), was given to participants to complete at baseline and responses were valued using the cross-walk algorithm.75 Descriptive analysis was undertaken to calculate the mean, SD and range of the overall EQ-5D-5L score, and also the frequency of responses within each dimension.
Patient and public involvement
We recruited a participant advisory group (PAG) consisting of four to six clinical, administrative and retired NHS staff who could provide us with their insights on how the eTHOS trial might be received by NHS staff. They met on a regular basis throughout the project to provide advice. The group was led by Margaret O’Hara (PPI expert coinvestigator) who assisted in recruiting members to the PAG and facilitating communication with the eTHOS team. There was also a hospital staff member of the TOC.
The general contribution of the PAG included: version testing the eTHOS database to ensure it was user-friendly, discussions on how to keep participants engaged during delays due to the COVID-19 pandemic and reviewing study documents. They were specifically asked how the eTHOS team could encourage ethnic diversity; the PAG recommended staff networks to reach out to, including ethnic minorities, faith centres and chaplaincies. It was also suggested that managers be asked to actively encourage all staff to participate. If eTHOS proceeds to full trial, the PAG recommended that funding for dedicated community engagement should be included to encourage grass root communication.
We also recruited a stakeholder advisory group representing a wider body of professionals, for example, GPs relevant to the trial. They provided advice at regular intervals; one GP highlighted that referral for certain high-risk conditions, for example, severe depression would be most effective if by direct telephone call.
Overall, the PAG and the stakeholder group were very supportive of the trial and its potential impact on the health and well-being of the NHS staff.
Data storage and confidentiality
Participants’ personal data were held securely and treated as strictly confidential, in line with the General Data Protection Regulation 2018. Electronic records were held on a secure, password protected, web-enabled customised database hosted by BCTU or on secure, password-protected University of Birmingham servers (including audio recordings). Paper records were transferred from the participating study centres to the trial office at BCTU and kept in a locked cabinet in a locked room. Any data processed outside BCTU were anonymised. Data will be stored for at least 10 years according to the University of Birmingham policy.
Monitoring
Birmingham Clinical Trials Unit, a UK Clinical Research Collaboration (UKCRC) registered trials unit, was the co-ordinating centre and was involved in the development, design and conduct of the trial since conception. A Trial Management Group oversaw research methodology, clinical trial coordination, data management, statistical analysis, compliance with Good Clinical Practice and all regulatory requirements, in conjunction with the sponsor. A TOC (combining the function of trial steering and data management committees) with an independent chair and lay representative oversaw, advised on and monitored the trial.
Impact of COVID-19 pandemic on trial delivery and summary of changes
Although ethical approval for the trial was obtained in March 2020, it was not able to commence due to the COVID-19 pandemic. Two hospital sites granted permission to commence the trial in December 2020, but after a short period, in February 2021, recruitment was paused again due to a spike in COVID-19 cases. The trial recommenced in May 2021 in all four sites. A 6-month extension to the trial was awarded from the funder until October 2021. The original trial was planned to include a 52-week follow-up; however, due to the delays, a 26-week follow-up was then agreed instead, but only a few people reached this time point before trial closure. Changes therefore were:
shortened total recruitment period and reduced number of participants recruited
limited 26-week follow-up data (both self-report and hospital personnel data)
no 52-week follow-up data (including no information to inform likely effect size)
modification to definition of attendance at referrals
modification to data collection documents to include reference to COVID-19 illness and adjusted working arrangements
reduced number of qualitative interviews, with adjusted time frame and sampling
health economic analysis reduced in scope
modifications to trial delivery (inability to promote the trial during face-to-face staff meetings, part of the screening intervention carried out online in some cases, all qualitative interviews carried out online)
reduced scope for consulting
PPI and stakeholder groups.
A list of protocol amendments is found in Appendix 1.
