Quantitative data collection & sample.

Quantitative data were collected at all six time points (T1 –T6). Between April and June, participants received four surveys (T1 –T4) with an interval of seven days between each survey. In this period, regular, non-COVID care was largely downscaled in the participating hospitals. After the fourth survey, all participants received a digital voucher (€10) for their participation. In August, participants received the fifth survey (T5). At this time, non-COVID hospital care had mostly upscaled again. In October, participants received the sixth and final survey (T6). Around this time, the number of hospitalised COVID-19 patients was progressively increasing, accompanied by partial downscaling of non-COVID care.

The study population (N = 46) consisted of eight consultants (17.4%), 15 junior doctors (32.6%), and 23 nurses (50.0%). Participants were on average 35.4 years old (SD = 9.8), and 78.3% were female (n = 36). One participant withdrew their participation after T1. Thirty-seven participants completed all six surveys (T1 –T6), four participants completed five surveys, and four participants completed four surveys. Forty participants completed the last survey (T6). In total, we received 259 surveys, resulting in 6.16% (17/276) missing data. An additional number of 14 (5.41%) surveys did not include the basic need satisfaction and frustration scale, because this scale was only assessed if the participant had worked during the previous week (i.e. some participants were off-duty during one of the measurements). The completion rate was 88.8% (245 out of 276 surveys).

Quantitative measures.

Demographic variables. We collected the following demographic characteristics of participants: age, sex, professional occupation, department where they normally worked in non-COVID times, and average hours worked per week regularly.

Need satisfaction & frustration. Work-related autonomy, competence, and relatedness satisfaction and frustration were measured six times with the diary version of the BPNS/F scale [26, 28, 29]. The response scale ranged from 1 (not true at all) to 5 (completely true). Example items included: “Last week, I felt a sense of choice and freedom in my job” (autonomy satisfaction; aggregated α = .63, with a range between α = .34 to α = .85), “Last week, I felt confident that I could do things well” (competence satisfaction; aggregated α = .83, with a range between α = .28 to α = .93), “Last week, I felt connected with people who care for me, and for whom I care” (relatedness satisfaction; aggregated α = .84, with a range between α = .77 to α = .90), “Last week, most of the things I did felt like ‘I had to’” (autonomy frustration; aggregated α = .73, with a range between α = .60 to α = .83), “Last week, I felt disappointed with many of my performances” (competence frustration; aggregated α = .73, with a range between α = .54 to α = .83), and “Last week, I felt excluded from my team” (relatedness frustration; aggregated α = .78, with a range between α = .72 to α = .85). Higher scores indicated more satisfaction or frustration of work-related needs, respectively.

Psychological distress. We assessed participants’ perceived psychological distress with the 12-item version of the General Health Questionnaire (GHQ) [30]. We asked participants to rate their perceived psychological distress over the last seven days compared to pre-COVID-19 (before March 2020). Example items from the GHQ-12 are “Have you been feeling reasonably happy over the past seven days, all things considered” (1 = happier than before; 2 = as happy as before; 3 = less happy than before; 4 = much less happy than before) and “Have you felt that you were playing a useful part in things over the past seven days” (1 = more useful than before; 2 = just as useful as before; 3 = less useful than before; 4 = much less useful than before). Participants rated all items on a four-option multiple-choice scale. We used the average GHQ-12 Likert score for statistical analysis and the bimodal scoring method (0-0-1-1, range 0–12), to screen for psychological distress among participants, with a threshold of 1 / 2. Higher scores indicated more psychological distress than before the COVID-19 pandemic. A bimodal score > 1 indicated that participants experienced somewhat more psychological distress than before COVID-19. A bimodal score > 2 indicated that the participant experienced more severe psychological distress than before the COVID-19 pandemic, which may lead to mental health disorders, and might be considered for follow-up mental health testing [31]. The reliability estimates ranged between α = .65 and α = .82, with an aggregated average of α = .76.

Quantitative data analysis.

Descriptive data analysis was performed with SPSS version 26. We used MLwiN version 3.05 [32] to employ multilevel analyses.

Qualitative data collection & sample.

The qualitative data (audio diaries) were collected during the first four time points (T1 –T4). The participants were instructed to record 1 to 3 audio diaries per week over a period of four weeks (see S2 File). Fifteen participants (ten nurses, two consultants, and three junior doctors) shared 60 audio diaries, 13 of them were female (86.7%).

Qualitative measures.

Audio diaries. To explore participants’ recent work experiences during the COVID-19 pandemic, we asked participants to record experiences that had a significant impact on them, either positive or negative (see S2 File for all prompts). All shared entries were in Dutch, with a duration of one to eight minutes, and were transcribed verbatim before analysis.

Thematic analysis.

We used a constructivist perspective to analyse the qualitative data, assuming that each participant shared their perspective on reality [33]. We used the six phases of thematic analysis to structure our analysis [34] (see S3 File for additional information). Data analysis started with familiarisation with the data by reading and re-reading all of the transcripts. Second, the transcripts were open-coded. Third, we developed a deductive thematic framework with autonomy, competence, and relatedness satisfaction and frustration, respectively. Fourth, we used the thematic framework for deductive analysis and discussed and reviewed the themes. Fifth, we (re)defined and interpreted the themes, leading to the results section in this paper. Finally, we wrote the draft paper and triangulated the qualitative and quantitative data. We used Atlas.ti software version 8 (ATLAS.ti Scientific Software Development GmbH, Berlin, Germany) to support data organisation and management.

Multilevel analyses.

Our data were hierarchically structured, with six repeated measures (T1 –T6, Level 1) nested in participants (Level 2). We created seven null models (see S1 Table), one for each variable, to examine the distribution of variance at the within- (Level 1) and between-person level (Level 2). In all null models, the within-person variance was larger than the between-person variance. Given these significant differences, multilevel analysis was appropriate to test our hypotheses.

Hypothesis testing.

Our first hypothesis predicted a positive relation between need frustration and psychological distress (Hypothesis 1a), and a negative relation between need satisfaction and psychological distress (Hypothesis 1b). First, we tested the variance in psychological distress that was explained by need frustration (Table 2, Model 1). Multilevel analysis showed that need frustration explained 9.0% variance at the between-person level and 9.0% variance at the within-person level. This model provided a significantly better fit than the null model (χ²_{change null model} (3) = 45.81; p < .001). We observed a significant increase in psychological distress for autonomy frustration (p = .005), competence frustration (p < .001), and relatedness frustration (p = .008).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Summary of the model estimates for psychological distress.

https://doi.org/10.1371/journal.pone.0255510.t002

Second, we tested the variance in psychological distress that was explained by need satisfaction (Table 2, Model 2). Multilevel analysis showed that need satisfaction explained 1.1% variance at the between-person level, and 7.9% variance at the within-person level. This model provided a significantly better fit than the null model (χ²_{change null model} (3) = 30.69, p < .001). We observed a significant decrease in psychological distress for autonomy satisfaction (p = .007), and relatedness satisfaction (p < .001), but not for competence satisfaction (p = .089).

Third, we combined Model 1 and Model 2 to assess if frustrated and satisfied needs explained additional variance in psychological distress (Table 2, Model 3). This model explained 5.6% variance of psychological distress at the between-person level, and 12.4% at the within-person level. Model 3 provided a significantly better fit than the need satisfaction only model (χ²_{change 2} (3) = 22.85; p < .001), and the need frustration only model (χ²_{change 1} (3) = 7.73; p = .052). We observed a significant decrease in psychological distress for relatedness satisfaction (p = .025), but not for competence satisfaction (p = .323), or autonomy satisfaction (p = .530). We observed a significant increase in psychological distress for autonomy frustration (p = .029), and competence frustration (p < .001), but not for relatedness frustration (p = .363). Hence, we conclude that Hypothesis 1 was only partly supported: unique variance of psychological distress was explained by autonomy and competence frustration (Hypothesis 1a), and by relatedness satisfaction (Hypothesis 1b).

In our second hypothesis, we predicted that the level of perceived psychological distress among frontline workers was higher than before the COVID-19 pandemic (Hypothesis 2a), but that these levels would decrease over time when COVID-19 infection rates declined (Hypothesis 2b). As indicated, we asked the participants to rate their perceived psychological distress over the last seven days compared to pre-COVID-19 (i.e. before March 2020). To test the second hypothesis, we first calculated perceived psychological distress among frontline workers and used the cut-off scores of Goldberg et al. [31]. Table 3 shows that relative to pre-COVID-19, a substantial percentage (27.9% - 50.0%) of the participants experienced more psychological distress across different time points, which supports Hypothesis 2a. More specifically, 7.0% - 20.0% experienced somewhat more psychological distress (score > 1–2), and 13.3–30.0% experienced more severe psychological distress (score > 2) during COVID-19. Mann-Whitney tests indicated that, across all six time points, the group of participants that experienced more psychological distress (GHQ > 1) had a significantly higher median score (p < .001) than the group of participants that did not experience more psychological stress (GHQ ≤ 1) than before COVID-19 (see S2 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Perceived psychological distress across time points.

https://doi.org/10.1371/journal.pone.0255510.t003

Note that Table 3 also shows that psychological distress fluctuated over time. We tested these fluctuations (Hypothesis 2b), with multilevel analysis (see S3 Table). The results indicated a trend that perceived psychological distress was somewhat lower on T2 (p = .081) and T4 (p = .050) compared to T1. This model did not explain more variance than the null model with psychological distress only (χ²_{change null model} (5) = 7.52; p = .185). Hence, we conclude that Hypothesis 2 was partially supported. A substantial number of participants perceived more psychological distress during COVID-19 than before, but we found no statistically significant changes in psychological distress over time.

Need frustrating experiences.

In many need frustrating experiences, the unpredictable course of COVID-19 or organisational logistics played a role. In particular, the need for autonomy was frustrated, illustrated by participants who felt unable to support patients or organise their work effectively. Limited experience with COVID-19 (competence frustration) and frequently changing teams (relatedness frustration) further frustrated their needs. The use of personal protective equipment (PPE) complicated collaboration between colleagues or communication with patients and family members. Organisational logistics (e.g. work schedules, rules and regulations, communication) led participants to experience little control and influence over their work and work-life balance (autonomy frustration).

COVID-19 care. The unpredictable course of COVID-19 had an enormous impact on many participants. Participants perceived limited possibilities to effectively support patients and often felt defeated when patients deteriorated quickly despite the care they provided. Multiple participants experienced that ‘fate’ predicted the course of the disease and they could not (entirely) rely on their clinical skills and experience due to the absence of a concise treatment.

“So, in that sense, it is a constantly unpredictable situation and that, yes, sometimes that causes me to be extra worried for patients while I would normally not be for such patients. […] Anyway, I now find it scary, and now I notice that I-, um, for example, last night I was awake, and I started thinking whether I really did everything right. Whether I should have done extra checks.” (P3)

These worries increased when participants worked in frequently changing teams or when teams were stressed or frustrated. This hampered the connection and trust among colleagues. Some participants reflected that they needed some time to feel comfortable in a new role and new team. Others experienced more difficulty relying on the skills of colleagues they did not know, which impeded their feelings of trust.

Another example from a nurse illustrated how they felt helpless to comfort a patient who just returned from the ICU to the COVID-ward. This patient woke up panicked, and started to hyperventilate. Despite all their efforts, the doctors and nurses were unable to comfort the patient and take their shortness of breath and anxiety away. The nurse reflected afterwards: “I cried when I got home, just because you só badly want someone to–to be comfortable and to help them–and not being able to help someone, that you only can just stand by them, that makes me feel helpless. (P43).

This feeling of inability to comfort patients was strengthened by the use of PPE, which impeded contact and communication with patients, but also made daily care more tiring than normal: “Because they also only see, uh, a blue suit and a mask and a hairnet and they do not see the person behind that, so it is harder to reassure [patients] than normally.” (P2). On top of that, the obligatory use of PPE and strict rules to prevent the spreading of the disease impeded participants to lend a helping hand or support colleagues. This was especially hard during busy shifts, where participants saw their colleagues struggle to manage their work while they could only watch.

Organisational logistics. Organisational logistics frustrated participants’ needs in different ways. We found two categories related to communication that impeded need support, namely experiences associated with visiting arrangements and experiences associated with changing rules and regulations for the staff.

First, the strict visiting arrangements frustrated participants’ autonomy and competence. Many participants shared experiences where they needed to bring bad news. Normally, family members would visit the hospital for this and get the opportunity to say goodbye to their loved ones. During COVID-19, many participants had to bring bad news to family members under time pressure–because the patient deteriorated quickly and needed to be transferred to the ICU–and by telephone–because of the strict visiting arrangements: “It is very difficult to give an emotional response [by telephone], uh, such as putting an arm around them […] or looking them in the eyes to check how they handle this news.” (P6) This challenged their perceived professional effectiveness of communication and provision of emotional support. In addition, visiting arrangements were changed frequently, but this was not always effectively communicated. This led to miscommunication and frustration within teams. For example, a pilot period to allow some visitors on the COVID-wards was not clearly communicated with the nurses, which resulted in fewer visitors than were possible and confusion within the team.

Second, limited involvement of the staff in organisational logistics especially frustrated participants’ autonomy. Many participants felt insufficiently involved in decision-making or incompletely informed by the organisation. Regulations from the hospital organisation led to (some degree of) uncertainty, unrest, or frustration for many participants. For example, it was unclear if or when participants would return to their own specialty to provide care for non-COVID patients. Some participants felt uncomfortable, useless or bored during quiet COVID-shifts with very few hospitalised patients. Multiple narratives included participants not being able to influence scheduling and rosters being available only last minute, in combination with frequently working on different wards with different teams. Some participants described a feeling of being ‘toyed with’:

“For me personally, I-, it also exhausts me that you just don’t know where you stand. Especially if you have a family with children at home, a husband who in fact also has to work, and everyone has to adapt to my schedule ad hoc.” (P6)

When miscommunications occurred about such regulations, nurses and junior doctors experienced no choice but to accept it because they stood lower in the overall hierarchy. In one case, participants heard that they needed to take care of more patients in the near future. This resulted in doubts about the future quality of care and impaired work satisfaction of nurses:

“Uh, we have heard that we need to care for uh many more patients uh in total. Instead of uh 30 patients, a total of 60 approximately. […] Uh, and there is unrest about that. We are concerned that the quality will decrease, that we will have less work satisfaction, uh, that- that care will be compromised and that this is insufficiently acknowledged.” (P29)

Need supportive experiences.

In many need supportive experiences, a feeling of connectedness with colleagues, patients, or family members played a central role. It seemed that the need for relatedness was vital to cope with emotionally demanding situations, such as deteriorating patients or the changes and uncertainties that were caused by COVID-19. Relatedness satisfaction seemed to promote competence and autonomy in the workplace, possibly due to an open and positive atmosphere within teams. Examples of autonomy support were more implicit and often related to a sense of competence. For example, knowledge and practical experience of dealing with COVID-19 (competence satisfaction) stimulated participants to organise care for patients in a way they felt was best (autonomy satisfaction).

Interaction with colleagues. Most participants felt connected with their colleagues. The uncertainties and challenges of working on COVID-wards promoted a sense of camaraderie, e.g. “You have to do it as a team, and that’s the strength of uh working on wards like these" (P33). This supported participants in coping with the emotionally demanding care that they had to provide and helped them share successes with their colleagues.

"Yes, I sometimes find it difficult, to then- you are also in a kind of social isolation, and well, I am happy to see my colleagues. Luckily, I get a lot of support from my colleagues, and we were also able to talk about it well. Be able to evaluate afterward how intense it is.” (P1)

Although some participants, at first, felt incompetent to care for COVID-19 patients, their confidence was built through training, using protocols, and open consultation with colleagues. There seemed to be an increased level of equality among colleagues because everyone at times struggled with the unpredictable course or treatment of COVID-19. An open atmosphere and a strong team spirit made it easier to ask colleagues for advice, build trust in each other, and address areas of improvement in daily care. Moreover, participants felt free to share their expertise with colleagues from a different professional background.

“And I think that because of that, um, yes, people also feel more valued, because you’re not only doing regular COVID care, but you’re also be- being called upon your expertise from your own profession” (P3)

Interaction with patients & family members. After a couple of weeks of working on a COVID-ward, participants felt increasingly confident in taking care of COVID-patients. Gaining more experience helped them to better explain to patients and family members what might happen during the disease, despite its unpredictability. Many participants shared that they felt closely connected to their patients, for example, through the in-depth conversations they had with patients. This strengthened their ability to emotionally or mentally support patients. Participants felt the freedom to take the time to talk to patients, because they cared for a limited number of patients each day:

“Uh, it was uh, nice to support a patient on a social level and to have a conversation about that. And I really hope I’ve helped him. Uh, and what makes my profession so valuable–it’s uh–you’ve not become a physician to just prescribe pills, but you’re in a position to talk to people and uh try to find answers to their existential questions in life together.” (P6)

The strong bond that participants felt with their patients seemed to stimulate autonomous behaviour. Participants used visiting policies, although they were strict and not always clearly communicated, to arrange short moments of contact between patients and their family members, or organised that patients were transferred to (rehabilitation) clinics near their families. They also used technical innovations, e.g. video calling, to establish valuable contact between patients and their families. As a participant (P6) stated, “Uh, it was kind-of an eye-opener for me, at first I was hesitant because you think ‘will this work for effective communication?’. But actually, it was uh a nice method to communicate.” In addition, another participant (P29) explained: “Uh, it’s special to see that uh, loved ones can share some time through video-calling. But it’s also difficult to see them so far apart, in completely different worlds.”