The identification of needs and development of best practice guidance for the psychological support of frontline healthcare workers during and after COVID-19: A protocol for the FLoWS project

Population

Across all work packages, a FLW will be defined as a person working in healthcare whose role during the pandemic was likely to require repeated contact with patients with COVID-19, whether confirmed or suspected (Lai et al., 2020; Nguyen et al., 2020). This is likely to include doctors, nurses, and allied health professionals working in emergency departments, ICUs, and COVID-19 designated facilities and inpatient wards, but does not exclude other departments, primary care facilities, or those who work in the community. It may also include essential non-clinical staff working in these areas, such as technical, administrative, cleaning and catering staff.

Participants for both studies will be recruited through hospital communications to staff (e.g. emails, noticeboards), professional organisations and training colleges, and by asking FLWs to share details with colleagues. Participants are likely to participate in either the online questionnaire or the interview, but may participate in both if they so wish. A truly representative sample of FLWs is unlikely to be reached given the broad range of roles that are eligible to participate, but responses will be monitored and outreach strategies adjusted to avoid significant over- or under-representation of any one subgroup.

Inclusion Criteria

Work packages 1 and 2 will utilise the same inclusion criteria, outlined below.

Work Package 1: Prevalence and determinants of psychological distress in Irish and Italian frontline workers

The aim of this work package is to determine the proportion of FLWs with elevated symptoms of psychological distress to indicate the magnitude of psychological suffering in Irish and Italian FLWs. This work package also aims to evaluate the extent to which moral injury, demographic and occupational factors predict psychological distress in Irish and Italian FLWs, thereby providing an indication of groups at-risk of psychological distress. To do this, a cross-sectional design will be employed, whereby psychometric measures and questionnaires are administered to Irish and Italian FLWs. Each participant will complete an online survey comprised of psychometric scales to assess symptoms of depression (PHQ-9; Kroenke et al., 2001), anxiety (GAD-7; Spitzer et al., 2006), insomnia (ISI-7; Bastien et al., 2001), post-traumatic stress (PCL-5; Blevins et al., 2015), post-traumatic growth (Post-traumatic Growth Inventory; Tedeschi & Calhoun, 1996), adjustment disorder (ADNM-8; Kazlauskas et al., 2018), and moral injury (Moral Injury Events Scale; Nash et al., 2013). The survey will also include a questionnaire to record age, gender, marital status, country of residence, type of residence, occupation title, professional experience, and type of hospital. Additional areas of interest as suggested by clinicians are perceptions of personal readiness for their usual or redeployed roles, perceived control over the influx of patients, availability and utility of training, institutional support, and impacts on personal life and living arrangements. An additional series of questions will be provided to assess these perceptions. Finally, to ensure that any recommendations made within the guidelines reflect support services that FLWs would realistically avail of, participants will be asked to preferentially rank a series of support services based on the recommendations of clinicians (e.g. psychological support, more PPE) which they either did avail of, or would have availed of, throughout the pandemic. Participants will be asked to rank these support services from being the most to least preferential based on their perceived usefulness for alleviating psychological distress throughout the pandemic. Efforts will be made in the design of the questionnaire to reduce participant burden and ensure time needed to complete is kept to a minimum.

To ensure the sample size is large enough to facilitate multivariate analyses of the range of predictors being assessed (demographics, occupational factors, moral injury), the following formula was used: N=100+50i, where N and i respectively denote the sample size and number of predictors being included in the models (Bujang et al., 2018). Based on approximately 20 coefficients being generated within the models, a minimum of 1,100 participants, recruited across Ireland and Italy, is necessary to yield representative parameter estimates and avoid model over-fitting. This sample size is also sufficient to generate prevalence estimates of the psychological distress symptoms being assessed using the traditional prevalence formula: N = Z²P(1 − P) / d², where the sample size N is generated using the conventional values of 95% and 5% for the level of confidence Z and precision d, respectively, and the prevalence value P based on existing estimates (i.e., 22.8% for depression, 23.3% for anxiety, 34.32% for insomnia, 71.5% for distress; Lai et al., 2020; Pappa et al., 2020). A convenience sample of FLWs will be recruited through health service networks in Ireland and Italy.

While an earlier version of this protocol included plans for a second questionnaire carried out 6 months later, the ever-changing course of the pandemic and the additional burden this would create for FLWs led to this plan being adapted. A second questionnaire will now not be carried out in the course of this project.

Research questions. 1. What is the proportion of Irish and Italian FLWs with mild, moderate, and severe symptoms of depression, anxiety, insomnia, and distress at time of assessment and after 6 months?

2. What are the predictors of depression, anxiety, insomnia, and distress in Irish and Italian FLWs at time of assessment and after 6 months?

Data analysis plan. Descriptive statistics for the total sample will comprise of means with standard deviations for normally distributed continuous variables, medians with interquartile ranges for non-normal continuous variables, and frequencies with proportions for categorical variables. The prevalence of psychological distress symptoms will be determined using the proportion of respondents with either mild symptoms, moderate or severe symptoms in accordance with the standardised cut-off scores for each measure.

To provide a preliminary indication of the determinants of the psychological distress, a series of independent sample t-tests, Mann-Whitney U-tests, and Chi-square tests will be used, where appropriate, to compare the distributions of psychological distress symptoms between each demographic and occupational group. To determine the independent influence of these factors in predicting psychological distress, multiple logistic regressions will be conducted using each psychological distress symptom as a dichotomized outcome variable based on the standardised cut-off score for each measure. For each symptom, one logistic regression will examine correlates of at least mild symptoms, and a second logistic regression will examine the correlates of at least moderate to severe symptoms. Each logistic regression will control for the geographical region where the participants are employed (i.e., Ireland or Italy), thereby indicating whether one cohort is at higher odds of experiencing psychological distress, but also allowing for an analysis of the demographic/occupational predictors of psychological distress independent of geographical location. Each logistic regression will also include the participants’ occupation, whether they are employed in a rural or urban setting, gender, years of professional experience, and marital status. Odds ratios (OR) and 95% confidence intervals (95% CI) will be used to determine whether any of the geographical, occupational, or demographic characteristics of the sample are independently predictive of higher odds of experiencing psychological distress. Exploratory moderation analyses will also be considered to determine whether the influence of predictor variables on the outcome variable varies depending on the status of other variables (e.g., whether the influence of occupational status on psychological distress differs between the Irish and Italian participants), and whether these interactions significantly contribute to the fit of the model using the maximum likelihood method.

Additional descriptive analyses will also examine the proportion of participants reporting differing levels of readiness for their professional roles during the pandemic, whether they received training, their perceptions of the suitability of the training, whether they received institutional support, and the extent to which the pandemic adversely affected their family life. Finally, the preferentially ranked support services will be examined by determining which services yielded the largest number of participants ranking them as being the most preferential.

Work Package 2: Qualitative needs assessment of Irish and Italian frontline workers

The aim of this work package is to qualitatively evaluate FLW’s first-hand accounts of their primary sources of stress and resilience during the pandemic, their perceptions of barriers and facilitators to psychological care, and insights into whether and how psychological services can alleviate stress in FLWs. To do this, a qualitative interpretative design using thematic framework analysis will be employed, using a purposive sampling and snowballing strategy to recruit participants. Interviewees will be sent an invitation with a link to the information form via e-mail. They will then be contacted by a member of the research team who will explain the purpose of the interview and schedule an appointment. In line with social distancing guidelines, electronic consent will be sought from all participants to prevent face-to-face contact. Interviews will take approximately one hour. All interviews will be audio-recorded and transcribed verbatim. To ensure the quality of the research findings, the sample size for the semi-structured interviews will range between 20 to 30 participants, recruited across Ireland and Italy, in line with existing guides on sampling requirements for qualitative interviewing (Francis et al., 2010; Green & Thorogood, 2018; Ritchie et al., 2003). The exact number recruited will be determined by the point at which the data reaches saturation (Guest et al., 2006).

Semi-structured interviews will be conducted with individual participants via phone or a secure web-based platform. Semi-structured interviews were chosen as they are flexible and allow participants to elaborate on issues that they want to speak more about, yet still provide a good structure for comparability (Jamshed, 2014). Semi-structured interviews also allow the researcher and participant to engage in a dialogue whereby initial questions are modified in the light of the participant’s responses and the investigator is able to probe interesting and important areas that arise. An interviewing topic guide of open-ended questions will be used to flexibly guide the interviews and to minimize variability, and enhance comparability, across the interviews. This interview guide will be developed based on a topic list that was developed from the research questions.

Research questions

Data analysis plan. Interview transcripts will be analysed using thematic framework analysis. The framework method, as described by Gale et al. (2013), was chosen for this study as it can compare data across cases as well as within individual cases and allows for the inclusion of a priori as well as emerging themes. It also provides a highly systematic method for categorizing and organizing qualitative data as the procedure for applying framework analysis consists of seven distinct stages: transcription, familiarization, coding, developing an analytical framework, charting and interpreting (see Hanlon et al., 2020 for details on the method). Having a systematic method such as this will be useful to ensure consistency as there will be multiple researchers working on the project. The framework method also produces highly structured outputs of summarised data, which is beneficial for projects such as this where the research team will need to manage a large data set while also obtaining a holistic, descriptive overview of the entire data set.

The researchers are cognisant of the necessity for quality assurance and rigour throughout the data collection, analysis and write up stages of this work package. Each researcher involved in interviewing participants and analysing the data has the necessary education, skills, and experience that is required for conducting high quality qualitative research. Rigour will be established using the criteria of Reliability, Validity, Dependability, and Confirmability. In line with guidelines outlined by Morse et al. (2002) and Whittemore et al. (2001), this will be achieved with care in the application of research practices to ensure replicability and consistency, providing verbatim transcriptions of the interviews, demonstrating saturation, clearly articulating data analysis decisions, reflexive journaling by the researchers following each interview, providing an audit trail, and demonstrating evidence (i.e. using participant quotes and a synthesis of researchers perspectives) that supports interpretations.

Work Package 3: Development of best practice guidelines

The aim of this package is to utilise the findings from the earlier work packages to draft best practice guidelines for alleviating psychological distress in FLWs. Researchers with a diverse range of expertise, including psychiatrists, clinical psychologists, health psychologists, physicians, and nurses will all provide input for interpreting the findings and determining how they can be realistically and effectively applied to alleviate psychological distress in FLWs. Qualitative data will be used to identify modifiable sources of stress, methods to overcome perceived barriers to psychological care, and common sources of resilience which can be used to foster effective coping mechanisms in less resilient workers. Quantitative data will be used to identify the most prevalent psychological distress symptoms in FLWs, and the most effective evidence-based treatments for alleviating those symptoms can be recommended accordingly. Through identifying determinants of psychological distress, recommendations can be made concerning which at-risk groups warrant additional attention for psychological care. The extent to which the drafted guidelines are generalized across different cultures (i.e., Irish and Italian) and occupational groups will be dependent on the congruency between the quantitative and qualitative findings of the groups assessed. For example, should certain demographics be more indicative of psychological distress in the Irish cohort, or FLWs in the Italian cohort report different psychological needs, the guidelines will be developed to reflect these differences.

The drafted best practice guidelines will be validated using the Delphi method, whereby a purposively selected panel of national and international stakeholders will review the guidelines and amend any of the content based on their insights. Eligible members of this panel will include providers of psychological support (e.g., clinical psychologists, psychotherapists, psychiatrists) and FLWs who are being targeted for provision of psychological support (e.g., doctors, nurses), to ensure the feasibility and acceptability of implementing suggestions is considered. To ensure the guidelines are applicable to FLWs with varying levels of seniority and differing experiences throughout the pandemic, efforts will be made to recruit both junior and senior FLWs, and FLWs who either remained in their usual position or were re-deployed during to the pandemic (with FLWs as defined as in WP1 and 2 eligible for inclusion). Panel members will be recruited through searching relevant national and international organisations, literature, and recommendations from the project team and Delphi participants. Although there is no standardized sample size for the Delphi approach, a minimum of 10 panel members per area of expertise is recommended to ensure reliability in the group’s judgements (Okoli & Pawlowksi, 2004). Accordingly, a minimum sample of 10 providers of psychological support and 10 FLWs will be targeted for the panel, although this may be subject to change based on the availability of participants, and the ability to recruit within a short timeframe.

In accordance with recommendations on how to reach optimal decision making using the Delphi method, all panel members will provide insights anonymously and work autonomously, moderated by a member of the research team with experience in managing discussions acting as a panel facilitator (Jorm, 2015). The round one Delphi survey will ask panel members to rate their level of agreement with each of the recommendations on a 5-point Likert scale (‘strongly disagree’ to ‘strongly agree’). Open text space for participants to give reasons for their evaluations, and suggestions for changes and additional content will be provided. For each recommendation outlined within the guidelines, a summary statistic will be used to quantify the proportion of panel members that either agree or disagree with the usefulness or practicality of each corresponding strategy, where 80% agreement is required for consensus (Jorm, 2015). Should a large proportion of panel members disagree with the usefulness or practicality of a proposed strategy, their recommendations will be used to amend such strategies until a consensus is reached, or that strategy is omitted from the final set of guidelines.