Incorporating health workers’ perspectives into a WHO guideline on personal protective equipment developed during an Ebola virus disease outbreak

Approach

The 2014–2016 EVD outbreak in West Africa was initially declared a Public Health Emergency of International Concern in early August 2014, coinciding with the decision to develop a WHO rapid advice guideline on the selection and use of PPE for EVD care in outbreaks. We electronically surveyed international frontline physicians and nurses who participated in foreign medical teams deployed to the affected countries in early stages of the EVD outbreak. The pragmatic approach was necessary given that this survey was developed and delivered at the height of outbreak and that WHO had very limited time available in which to produce guidance.

Survey

The online, 23-item survey was developed specifically for this study (Supplementary File 1). The first section consisted of multiple-choice questions examining participant demographic characteristics, role, and experience with PPE in West Africa. The next section addressed health worker exposure to the following specific components of PPE: eye protection (goggles/face shields), nose and mouth protection (medical mask/particulate respirator), gloves (single/double gloves), body covering (gowns/coveralls), foot wear (boots/closed shoes), and head covering (hair cover/hoods). In subsequent sections, we used a four or five-point Likert-scale to examine participants’ perceptions about the impact of each PPE item on the following domains: safety, communication, ability to provide patient care, personal wellbeing (heat and dehydration), and comfort. In addition, for each of the items, participants could provide free-text comments on open-ended questions to describe any difficulties or to provide suggestions on how PPE could be improved. The final section explored specific training needs and confidence in PPE. The last question asked participants to compare two sets of PPE available in West Africa shown side-by-side in a picture: one was composed of lighter items and the other had more robust components.

Five experts reviewed the study protocol and questionnaire during the development phase. Subsequently, three clinicians with experience in the EVD outbreak in West Africa similar to that of the sampling frame field-tested the survey for consistency, readability, completeness, and question sequencing. The final version of the online survey incorporated all relevant feedback and comments. We obtained expedited approval of the study protocol and survey from the WHO Ethics Review Committee (RPC690).

We contacted potential participants via email. The first email explained the objectives, expected time commitment, and provided a link to the informed consent form and online survey on Survey Monkey^®. Participation was voluntary and implied informed consent. A follow-up email in 5 days reminded potential participants of the deadline (10 days after launching). Participants could withdraw from the study at any time without providing any justification.

Participants

The study population consisted of international frontline physicians and nurses with direct field experience caring for EVD patients in West Africa. Our sampling frame targeted international physicians and nurses deployed by WHO and Médecins Sans Frontières (MSF) to West Africa between March and September 2014. We used maximum variation purposeful sampling, a non-probability sampling strategy, to capture a wide range of health worker perspectives and experiences in two organizations and four different countries affected by the EVD outbreak. Health workers were reached through a contact individual in each organization (MSF and WHO) who directly emailed potential participants. Physicians and nurses from the affected countries and from other international organizations were not included for pragmatic reasons given the extreme time constraints and infeasibility of obtaining additional organizational approvals in the available timeline. An initial communication error lead to the contact of other groups of health workers that did not have frontline clinical experience. The perspectives of these workers were considered for WHO quality improvement efforts, but were excluded from this analysis as these groups were not part of the approved sampling frame for this study.

Data analyses

Participants could indicate their experience with more than one item for each PPE component (e.g., both goggles and face shields for eye protection). For the purpose of statistical analysis, we considered each participant’s experience with a PPE item unique and independent. We analysed closed-ended questions with STATA 10 (StataCorp. 2007. College Station, TX) using counts, proportions, and the Chi-square test when comparisons were appropriate.

Two independent researchers analysed the answers to the open-ended questions using an iterative and reflexive process. This encompassed close reading and re-reading of the answers using constant comparison within and across different participants to identify key topics. The researchers then grouped the interpretations and understanding of the participant’s ideas and selected quotes to represent these findings, discussing discrepancies to achieve agreement.

Immediately after data collection with the Survey Monkey^® instrument, all information was downloaded to an anonymized spreadsheet, removed from the online database. All analyses were performed on de-identified data.

Informing rapid advice guideline recommendations

The rapid advice guideline was developed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach^9,10. With this approach, clinical and public health recommendations are based on a systematic review and critical appraisal of the evidence on benefits and harms of an intervention, and an assessment of the balance between the two. Other considerations are also taken into account when an expert panel formulates recommendations, including feasibility, acceptability and resource implications of the intervention options, and the effects on equity across subpopulations. The relative value of the potential outcomes of the intervention options and the values and preferences of persons affected by the intervention are also important considerations. The findings of the survey were presented at the guideline development meeting and incorporated into evidence-to-decision tables (Supplementary File 2) to inform the formulation of recommendations for PPE components in the context of an EVD outbreak. Evidence-to-decision tables followed the GRADE-DECIDE¹¹ approach and were populated by the WHO guideline development team in preparation for the expert panel meeting. These tables were key instruments used to present multiple sources of information to the guideline expert panel, helping to structure the discussion and to document the final judgements and decisions that underpin each recommendation.

PPE use and perceived risks and effects

For each of the different components of PPE, one item was used by the majority of survey participants (Table 2). For example, 42 (95%) of participants had experience using goggles, while only seven (16%) had used a face shield (some participants had experience with both types of eye protection). Generally, health workers felt at low or extremely low risk regardless of the type of PPE used. PPE, particularly goggles, particulate respirators, and medical masks or hoods, impaired communication (Table 2). A reduction in the ability to provide care was predominantly related to eye protection equipment - both face shields and goggles. Heat and dehydration were a significant or major issue for 31 participants using goggles (76%) compared to two (29%) using a face shield (p=0.02), and for 27 (64%) using a hood compared to none using a hair cover (p=0.02). Heat and dehydration also were a significant or major issue for the majority of individuals using a gown (n=11, 73%) or coverall (n=26, 87%); however, there was no significant difference between the two groups (p=0.41). Goggles were considered more uncomfortable (n=29, 71%) than face shields (n=2, 29%, p=0.08) (Table 2).

Experiences with PPE and suggestions for improvement

Participants indicated that fogging of goggles or face shields was a major issue, affecting visibility and potentially creating a hazard for health workers as well as patients. There was some indication that fogging was a bigger issue with goggles and a few participants indicated that they would have preferred a face shield. Two participants indicated that the goggles caused pain after using them for extended periods. A number of participants noted that goggles did not cover sufficient skin of the face and there were requests for larger goggles, which would have the added advantage of greater visibility. Other issues were the poor quality of face shield and goggles, poor fit of goggles, and the logistical challenges of waiting to clean and dry re-usable goggles. One respondent summarized it as follows: “The goggles (are) not so comfortable and (they) felt like the "unsafe" part of the PPE. They move easily, hurt on the head, and affect vision in a negative way due to sweat, etc.”.

Medical mask and the particulate respirator were reported to cause difficulty breathing when wet (due to sweat or condensation). One participant doubted the mask’s effectiveness when wet. Two participants were of the opinion that respirators were excessive since EVD is not airborne.

The main problem regarding gloves was the risk of having them slip down, allowing fluids to contact the skin as illustrated by the following respondent: “Some people found using tape over gloves (the second pair) useful as sometimes they did roll down during arduous patient care activity and in the end I also did this”. Other participants also attempted to solve this problem by taping gloves to the coverall, however this occasionally resulted in the tearing of gloves or the coverall. It was also mentioned that gloves were not long enough and that they tore easily.

Many participants indicated the need for lighter suits with better ventilation. As one respondent commented: “During the dry season and if it was a sunny day it became quickly unbearable to stay too long (in the ETU). Ebola patients need lots of care and support, full PPE hinders this process. We need lighter and cooler PPE to be able to provide better care and stay longer inside (the ETU). Full PPE causes heat exhaustion and dehydration”. Difficulties included finding the right size coverall – in several instances the available coveralls were too small, leaving the health worker to opt for a coverall of lesser quality or have difficulties removing the coverall. A number of health workers indicated that they had difficulty taking off the coverall. Specific issues included having to remove the face shield first, leaving the eyes and face unprotected while undressing from the coverall, and problems taking off the coverall over large rubber boots. One respondent mentioned that coveralls with attached shoe covers could increase the risk of tripping. One respondent commented that boots were too big causing difficulty walking on irregular ground. As for reusable items (goggles and boots), it was mentioned that the time required to fully decontaminate and dry them sometimes brought challenges and put pressure on the team.

Training on PPE use

A third of survey participants had received formal training over 2 to 3 days (n=15, 34%) and four (9%) reported training duration of more than 3 days. On the other hand, 20% (n=9) had received no formal or on-the-job training and another 20% (n=9) reported training for 2 hours or less. The remaining 15% of study participants (n=7) had training of one day or less. A number of participants commented that they would have liked to have had training, more formal training, or longer training. Others indicated that they would have liked to receive training before their departure, or before arriving at the treatment centre. The training topics that the survey participants would have liked included: removal of PPE and how to manage eye glasses. One health worker recommended weekly refresher training, especially in the light of frequent equipment changes, which may impact the order items are put on and taken off. Another health worker commented: “I believe that only experienced people can teach about Ebola. Teaching on the use of PPE is not about dressing and undressing. It is about using a set of behaviours with it and the understanding of all the underlying water and sanitation principles and applying them”. Regarding hand hygiene, alcohol-based hand-rub was not always available and there was conflicting information in different settings about which product to use.

Confidence using PPE and preferences

The majority of participants (n=26, 59%) were very confident that they were using PPE correctly, 17 (39%) were reasonably confident and 1 (2%) was not very confident. Generally, participants were least confident about goggles (fogging, moving/displacing), medical masks and particulate respirators (difficulty breathing, becoming uncomfortable), and gloves (rolling down, tearing). Removing PPE was also an area that people felt less confident about (e.g., taking arms and feet out of a coverall, lack of face protection during undressing if the face shield was worn outside the hood). As one health worker illustrated: “Taking off the (Tyvek suit) coverall was difficult due to my height; it required me to wiggle out of it more than the average person”. A respondent also mentioned feeling less confident working in the screening area where much lighter PPE was worn, while possibly also being exposed to infectious patients.

When asked to indicate their preference regarding two sets of PPE depicted in a picture, 8 (18%) participants preferred the PPE that was composed of lighter items, 33 (77%) participants preferred the more robust components, 2 (5%) did not have a preference and one participant did not respond to the question.

Implications of the survey findings

Overall, our findings showed that health workers perceive a balance between transmission protection and the ability to effectively care for complex patients while using PPE. The survey highlighted a slight preference of health workers for face shields compared to goggles because of less fogging, easier communication and better fit. There was no strong preference for one item of PPE over the other for all other PPE components. Given the variation in preferences for different components of PPE and the absence of data on comparative effectiveness, it may be important to provide a choice for health workers. This was, in fact, a guiding principle during the development of the PPE guidelines. Several issues raised by survey participants should be relatively straightforward to address, making a major contribution to health worker safety and comfort, such as providing a sufficient range of sizes, choice of equipment, and adequate training on how to put on and take off PPE in the conditions that will be faced in the field.

Challenges

We experienced a number of challenges planning and executing this study. We had to develop a survey questionnaire de novo with limited time for field testing. Although this likely had a minimal impact on the results, we noted two questions that participants appeared to have difficulty comprehending (questions 11 and 23; see Supplementary File 1); if we had had more time for field testing we could have revised the questionnaire before formal data collection began. While our aim was to include only health workers who had provided direct patient care, such as nurses and physicians, given a communication error early in the study, we invited to participate and consequently received responses from workers without direct clinical experience who had been deployed to the EVD outbreak. Because these workers were not part of our pre-defined sampling frame, we excluded their responses from the analysis. Similarly, our survey failed to take into account the fact that PPE consists of different components such as eye protection, nose and mouth protection, gloves and body coverings that work together to protect the health worker from the risk of infection. In the first part of our questionnaire we asked how the survey participant experienced individual components of PPE (e.g., goggles or face mask). However, it is difficult to review these components as isolated items, separate from the rest of the PPE. As one survey participant noted: “It is the combination of the respirator and the face shield which is difficult. One or the other would be manageable but, both together meant major impairment”. Another survey participant commented: “The coverall would probably be better tolerated if we could breathe easier and see without problems”. In addition, although we compared gowns and coveralls, we did not specify or ask about the materials the body coverings were made of, its level of fluid resistance, or whether the head cover was attached or not. Such issues can have a significant impact on health workers’ experiences. It also became clear that solutions to an issue with one component of PPE could compromise the safety of another element of PPE. For example, participants mentioned that they would improvise and tape gloves to the coverall in order to prevent them from slipping down, but then the coverall would tear when removing the tape. Finally, the combination of different components of PPE may change the order in which PPE items are put on and taken off, thus end-users may perform donning and doffing procedures that are different than the training they received. This is particularly relevant if there are frequent changes in the availability of specific types of PPE, as was the case early in the outbreak response.

Study limitations

Most of the limitations of this study were caused by pragmatic decisions the research team had to make in order to complete the study in the available time. This was in and of itself an invaluable learning experience for undertaking similar projects in the future. Specifically, we had to include only international health workers deployed by WHO and MSF in our study; therefore, we did not collect information on the values and preferences of local health workers and health workers deployed by other organizations. There were two important reasons as to why we selected our sampling frame. First, we carried out the survey at the height of the EVD epidemic when local doctors and nurses were fully engaged in the response efforts and we refrained from removing them from their primary work. Internationally recruited health workers on the other hand, were usually deployed for shorter periods and could thus participate when they returned home. Second, we had little time in which to execute the survey before the guideline meeting and we anticipated that it would be a lengthier and more complex process to identify and recruit local health workers. Thus, the findings of this survey may not be applicable to local health workers. In addition, generalizability of our findings to other international health workers involved in the Ebola response may be limited due to the small size of our purposive sample.

Study strengths

In the context of the most challenging of research settings, our study proceeded very efficiently and effectively in several regards. Peer reviewers for both the study protocol and draft survey made very helpful comments within 1 to 2 days. The WHO Ethics Review Committee approved the survey in less than two weeks. By reaching out to several key managers and opinion leaders from the two organizations, we were quickly able to identify frontline clinicians that were part of the sampling frame. The online format of the survey allowed us to quickly reach a larger number of health workers in different countries who had recent personal experience with different types of PPE in the EVD outbreak. The combination of different types of questions in our survey also worked well. Closed and Likert-scale questions made analysis of trade-offs and comparisons of health workers’ preferences possible while open-ended questions allowed the survey participants to share additional thoughts and perspective in more depth.