The experiences of health-care providers during the COVID-19 crisis in China: a qualitative study

This was not an intervention study. The researchers tested no drug, device, or behavioural intervention. Instead, they explored the lived experiences of health-care providers (HCPs) who were redeployed to COVID-19 wards during the early outbreak in Hubei province, China. The "intervention" was the naturalistic experience of working in a crisis setting. The "outcome" was the thematic content of HCPs' narratives about their challenges, coping strategies, and perceptions of duty. There were no comparators (no control group, no pre-post measurement). The study used semi-structured telephone interviews to collect qualitative data, then analysed transcripts using a phenomenological method to identify recurring themes.

**Sample size:** 13 health-care providers (9 nurses, 4 physicians)

**Population:** Frontline HCPs working in COVID-19-designated hospitals in Hubei province, China

**Setting:** Five hospitals in Hubei province (the epicentre of the early outbreak). All participants were recruited using purposive and snowball sampling. Inclusion criteria: physicians or nurses who had been working on COVID-19 wards for at least one week. No specific age range, sex distribution, or years of experience were reported in the abstract (the full text may contain these details). All participants were Chinese, working in a single province during a specific 2-week period (Feb 10–15, 2020). This is a highly specific, non-representative sample: all were volunteers who agreed to be interviewed by telephone during a crisis, which likely selects for more articulate, less overwhelmed, or more motivated individuals.

No quantitative instruments were used. The study used:

**Semi-structured, in-depth telephone interviews** (conducted Feb 10–15, 2020). The interview guide was not described in the abstract but would have included open-ended questions about experiences, challenges, and coping.

**Verbatim transcription** of audio-recorded interviews.

**Haase's adaptation of Colaizzi's phenomenological method** for data analysis. This is a structured qualitative analysis technique involving: reading transcripts to get a sense of the whole, extracting significant statements, formulating meanings, clustering themes, developing an exhaustive description, and returning to participants for validation (though the abstract does not confirm member checking was done).

**No quantitative scales, no physiological measures, no behavioural tracking.** The only "measurement" is the researchers' interpretation of spoken narratives.

**Study design:** Qualitative phenomenological study using semi-structured interviews. This is an exploratory, descriptive design, not an experimental or quasi-experimental one.

**Sampling:** Purposive and snowball sampling. Purposive means researchers deliberately selected participants who had relevant experience (working on COVID-19 wards). Snowball means initial participants referred colleagues. This is appropriate for qualitative research aiming to access a hard-to-reach population during a crisis, but it introduces selection bias: participants are likely those with stronger opinions, more time, or better relationships with the researchers.

**Data collection:** Telephone interviews (not in-person) due to infection control restrictions. This is a pragmatic adaptation but may reduce rapport and depth compared to face-to-face interviews. The interview period was only 6 days (Feb 10–15, 2020), which captures a very narrow snapshot of the rapidly evolving crisis.

**Data analysis:** Haase's adaptation of Colaizzi's phenomenological method. This is a well-established qualitative analysis technique. However, the abstract does not report whether two or more researchers independently coded the data (inter-coder reliability), whether saturation was reached (the point where new interviews yield no new themes), or whether participants reviewed and confirmed the themes (member checking). With only 13 participants, saturation is questionable—typical phenomenological studies require 10–20 participants, but the narrow sampling frame and short interview window raise concerns.

**What this design can and cannot prove:**

**Can prove:** That certain themes (duty, exhaustion, resilience) were present in the narratives of these 13 HCPs at this specific time and place. It can generate hypotheses about what frontline workers experience during a pandemic.

**Cannot prove:** That these themes are representative of all HCPs in Hubei, China, or elsewhere. Cannot quantify the prevalence or severity of any experience (e.g., "most nurses felt exhausted" is not supported—only that some did). Cannot establish causal relationships (e.g., that protective gear caused exhaustion—only that participants reported both). Cannot compare groups (e.g., nurses vs. physicians, or experienced vs. novice staff). Cannot measure change over time (single interview per participant). Cannot provide effect sizes, confidence intervals, or p-values—this is purely descriptive.

**Major methodological weaknesses:**

Very small sample (n=13) from a single province during a 6-day window.

No random sampling—selection bias is high.

No blinding (interviewers knew the purpose and likely had expectations).

No quantitative measures to triangulate self-reports.

No follow-up to see if experiences changed as the crisis evolved.

The abstract does not report whether the researchers had prior relationships with participants (e.g., colleagues), which could bias responses.

No mention of ethical approval or informed consent in the abstract (though likely obtained).

Three theme categories emerged from the analysis. Each contains sub-themes. The abstract provides no numerical data (no percentages, no counts of how many participants endorsed each theme). Findings are purely descriptive.

**Theme 1: "Being fully responsible for patients' wellbeing—'this is my duty'"**

HCPs volunteered to work on COVID-19 wards (though the abstract does not specify whether all 13 volunteered or some were assigned).

They tried their best to provide care despite the circumstances.

Nurses had a crucial role in providing intensive care and assisting with activities of daily living (e.g., feeding, bathing, toileting patients who were too ill to do these themselves).

No numbers: we do not know how many nurses vs. physicians expressed this theme, or whether it was universal.

**Theme 2: "Challenges of working on COVID-19 wards"**

Working in a totally new context (many had no infectious disease expertise).

Exhaustion due to heavy workloads and protective gear (e.g., wearing N95 masks and gowns for long shifts, which can cause dehydration, heat stress, and difficulty breathing).

Fear of becoming infected and infecting others (family, colleagues, patients).

Feeling powerless to handle patients' conditions (e.g., rapid deterioration, lack of effective treatments early in the pandemic).

Managing relationships in this stressful situation (e.g., conflicts with colleagues, strained family relationships due to quarantine).

No numbers: we do not know which challenge was most common or severe.

**Theme 3: "Resilience amid challenges"**

Sources of social support: from colleagues, family, friends, and the broader community (e.g., public appreciation, donations).

Self-management strategies: e.g., exercise, meditation, maintaining routines, peer debriefing.

"Transcendence": participants reported gaining meaning, personal growth, or a sense of purpose from the experience (similar to post-traumatic growth).

No numbers: we do not know what proportion of participants reported resilience vs. ongoing distress.

**No primary vs. secondary outcomes**—this is a qualitative study with no pre-specified hypotheses or outcome hierarchy.

This study does not report effect sizes. There are no quantitative outcomes to translate. The "effect" is the presence of themes in narratives. In plain English: when asked about their experiences, these 13 HCPs consistently talked about duty, exhaustion, fear, and resilience. But we cannot say how intense these experiences were, how common they were among all HCPs, or how they compared to other stressful work environments. A reader might infer that exhaustion and fear were universal, but the study design cannot support that claim.

**Acknowledged by authors (from abstract):**

The study was conducted in the early stages of the outbreak (Feb 2020) and may not reflect later experiences.

The sample was limited to Hubei province, China.

The authors call for comprehensive support and training, implying the findings are generalisable, but they do not explicitly list limitations in the abstract.

**Critical reader observations:**

1. **Sample size and representativeness:** n=13 is very small. With purposive and snowball sampling, the sample is likely biased toward HCPs who were more willing to talk, more reflective, or less overwhelmed. Those who were too exhausted, traumatised, or busy to participate are excluded.

2. **No quantitative data:** No prevalence rates, no severity scores, no comparison groups. We cannot know if these themes are universal or idiosyncratic.

3. **Single time point:** Interviews were conducted over 6 days. The crisis evolved rapidly—experiences in early February may differ from late February or March.

4. **Telephone interviews:** May reduce depth and rapport. Non-verbal cues are lost. Participants may self-censor more on the phone.

5. **Researcher bias:** The interviewers knew the study purpose and likely had expectations. Without blinding or structured protocols, leading questions could shape responses.

6. **No member checking reported:** Participants did not confirm whether the themes accurately captured their experiences.

7. **No conflict of interest disclosure:** The abstract lists funding sources but does not state whether authors had prior relationships with participants.

8. **Language and cultural context:** Interviews were conducted in Chinese, then translated for publication. Translation may lose nuance. Cultural norms (e.g., collectivism, respect for authority) may influence what participants are willing to say.

9. **No longitudinal follow-up:** We do not know if these HCPs developed PTSD, burnout, or long-term resilience.

10. **No comparison to non-COVID HCPs:** Without a control group of HCPs working in non-crisis settings, we cannot attribute these experiences specifically to COVID-19.

For someone running their own n=1 experiment (e.g., testing a resilience intervention, a shift schedule change, or a coping strategy during a stressful period):

**What to test:**

If you are a health-care worker (or anyone in a high-stress, high-exposure role), test a structured resilience protocol: e.g., daily 10-minute mindfulness meditation, a peer debriefing session after each shift, or a 15-minute exercise break during work.

Alternatively, test a protective gear schedule: e.g., 4-hour shifts with mandatory 30-minute breaks vs. 6-hour shifts with no breaks, measuring subjective exhaustion and cognitive performance.

**Minimum meaningful duration:**

For a resilience intervention: at least 2 weeks (14 days) to allow habituation and to capture variation in workload. The study suggests exhaustion builds over days to weeks, so a shorter trial may miss effects.

For a shift schedule test: at least 1 week per condition (e.g., 5 shifts of 4 hours vs. 5 shifts of 6 hours), with a 2-day washout between conditions.

**What to measure (specific metrics):**

**Primary outcome:** Subjective exhaustion (use a validated single-item scale: "On a scale of 0–10, how exhausted do you feel right now?" rated before and after each shift).

**Secondary outcomes:**

- Fear of infection (single item: "How worried are you about getting infected? 0–10")

- Sense of duty/meaning (single item: "How meaningful does your work feel today? 0–10")

- Cognitive performance (e.g., 2-minute digit span test or reaction time test on a phone app, done after each shift)

- Sleep quality (use the Pittsburgh Sleep Quality Index [PSQI] once per week, or a daily sleep diary: bedtime, wake time, number of awakenings, subjective sleep quality 0–10)

- Physical symptoms (headache, dehydration, skin irritation from gear—rate 0–10 daily)

**Objective measure:** Wear a heart rate variability (HRV) monitor (e.g., Polar H10 chest strap) for 24 hours on the last day of each condition. Lower HRV indicates higher stress and exhaustion.

**Key confounds to control for:**

**Shift timing:** Night shifts vs. day shifts produce different exhaustion levels. Compare only same-shift types.

**Patient acuity:** A shift with many critical patients is more stressful. Track number of critical patients per shift and include as a covariate.

**Prior sleep:** Poor sleep the night before will inflate exhaustion. Measure and control for prior night's sleep duration.

**Caffeine and food intake:** Track coffee/tea consumption and meal timing. Caffeine can mask exhaustion.

**Social support:** Note whether you debriefed with colleagues or family after each shift. This is a major confound identified in the study.

**Protective gear type:** If you switch between N95 masks and surgical masks, or between full gowns and aprons, note this. The study highlights gear as a source of exhaustion.

**Baseline mental health:** Measure your baseline anxiety and depression (e.g., GAD-7 and PHQ-9) before starting the experiment. Those with higher baseline distress may respond differently.

**What a positive result would look like:**

For a resilience intervention: a consistent reduction of ≥2 points on the 0–10 exhaustion scale after shifts, and a ≥1 point increase in sense of meaning, sustained over the 2-week period. HRV would increase (higher HRV = lower stress) by ≥5 ms in the RMSSD metric.

For a shift schedule test: the shorter-shift condition would show ≥2 points lower exhaustion, ≥1 point lower fear of infection, and better cognitive performance (e.g., 10% fewer errors on the digit span test).

A positive result is not just statistical significance (which you cannot calculate with n=1) but a clear, consistent pattern across repeated measurements. For example, if exhaustion is 6/10 after every 6-hour shift and 4/10 after every 4-hour shift, that is a meaningful difference.

**Watch for placebo effects:** If you expect the intervention to work, you may report lower exhaustion due to expectation. Use a blinded design if possible (e.g., have a friend randomise you to condition without telling you which is which, or use a sham intervention like a 10-minute quiet sitting vs. 10-minute meditation).

**Document everything:** Keep a daily log of all confounds. If you get a positive result but also changed your diet, sleep, or workload during the experiment, the result is uninterpretable.

**Bottom line for self-experimenters:** This study provides rich qualitative hypotheses but zero quantitative data. If you want to test whether a resilience intervention works for you during a high-stress period, you must design your own rigorous n=1 experiment with daily measures, control for confounds, and run it for at least 2 weeks per condition. Do not assume that because these 13 HCPs reported resilience, any intervention will work for you. The study's real value is in identifying what to measure (exhaustion, fear, meaning, social support) and what confounds to watch for (gear, shift length, patient acuity, prior sleep).