Mental Health Response to the COVID-19 Outbreak in China

This is not an intervention study but a cross-sectional survey and commentary. The authors measured the prevalence of mental health problems (anxiety, depression, insomnia) among outpatients seeking care at psychiatry, neurology, and sleep medicine departments during the COVID-19 outbreak. They also assessed how many existing psychiatric patients experienced deterioration, could not access care, reduced or stopped medication, and whether they used newly available online mental health services.

**Sample size:** 2,065 outpatients who completed the survey (out of 3,441 approached, a 60% response rate)

**Population:** Patients seeking care in the Departments of Psychiatry, Neurology, and Sleep Medicine at West China Hospital of Sichuan University, Chengdu, China

**Breakdown:** 589 new patients (first visit) and 1,476 existing patients (follow-up visits)

**Among existing patients:** 1,434 had preexisting psychiatric disorders (depression, bipolar disorder, schizophrenia, anxiety, insomnia, psychosis)

**Timing:** Survey conducted February 25 to March 9, 2020 — during the peak of China's lockdown and quarantine measures

**Setting:** Urban tertiary-care hospital in a major Chinese city; not a general population sample

**Anxiety:** Generalized Anxiety Disorder 7-item scale (GAD-7). Cutoff for "clinically significant anxiety" was total score ≥5 (range 0–21, higher = worse). Note: The standard clinical cutoff is usually ≥10 for moderate anxiety; the authors used a lower threshold of ≥5, which captures mild and above.

**Depression:** Patient Health Questionnaire 9-item scale (PHQ-9). Cutoff for "clinically significant depression" was total score ≥5 (range 0–27, higher = worse). Again, standard cutoff for moderate depression is ≥10; ≥5 captures mild symptoms.

**Insomnia:** Insomnia Severity Index (ISI). Cutoff for "clinically significant insomnia" was total score ≥8 (range 0–28, higher = worse). Standard cutoff for clinical insomnia is ≥15; ≥8 captures subthreshold insomnia.

**Deterioration in preexisting conditions:** Single self-report question: whether their mental health condition had worsened related to the pandemic (yes/no)

**Access to care:** Self-reported questions about whether they could receive timely diagnosis/treatment, whether they reduced or stopped medication, and whether they sought online help

**Data collection:** Self-report questionnaire administered via WeChat-based survey program (Questionnaire Star) — an online platform, not in-person interviews

**Study design:** This is a cross-sectional survey with a commentary component. The authors conducted a single-wave, self-report questionnaire among a convenience sample of outpatients at one hospital during a specific 13-day period. There is no control group, no randomisation, no blinding, and no follow-up.

**Why this design matters:** Cross-sectional surveys are useful for estimating the prevalence of a condition at a single point in time. They can identify potential problems (e.g., "many patients report worsening symptoms") but cannot prove causation. For example, the authors cannot determine whether the pandemic *caused* the mental health deterioration or whether these patients would have worsened anyway due to the natural course of their illness. The survey also cannot track changes over time — it's a snapshot, not a movie.

**What this design can prove:**

The proportion of outpatients meeting symptom thresholds for anxiety, depression, and insomnia during the early pandemic period

The proportion of existing patients who *perceived* their condition had worsened

The proportion who reported barriers to care (transport restrictions, medication access)

**What this design cannot prove:**

That the pandemic *caused* these mental health problems (no pre-pandemic baseline for comparison)

That these rates are higher than normal (no control group from before the pandemic)

That the findings generalise to other hospitals, rural areas, or countries with different lockdown policies

Any causal relationship between specific pandemic stressors and specific outcomes

**Major methodological weaknesses:**

1. **No pre-pandemic baseline:** Without knowing the baseline prevalence of anxiety, depression, and insomnia in this same outpatient population before COVID-19, we cannot attribute the observed rates to the pandemic. For context, the lifetime prevalence of any mental disorder in Chinese adults is 16.6% (from the paper's citation), but that's a different metric.

2. **Low symptom thresholds:** Using GAD-7 ≥5 and PHQ-9 ≥5 (instead of the standard ≥10) inflates prevalence rates. Many people with mild, transient distress would be counted as "cases." The insomnia cutoff of ≥8 on the ISI is also below the clinical threshold of ≥15.

3. **Self-report only:** No clinician-administered diagnostic interviews. Self-report scales overestimate prevalence compared to structured clinical interviews.

4. **Single hospital, single city:** West China Hospital is a major tertiary referral centre in Chengdu. Patients there may have more severe illness or better access to care than the general population. Results may not apply to rural areas or smaller hospitals.

5. **60% response rate:** 1,376 out of 3,441 patients did not complete the survey. Non-responders may have different mental health status (e.g., more severe illness, less tech-savvy, or less distressed).

6. **No demographic breakdown:** The paper does not report age, sex, socioeconomic status, or specific psychiatric diagnoses for the full sample, making it impossible to assess subgroup differences.

7. **No statistical testing:** The authors report percentages but no confidence intervals, p-values, or effect sizes for comparisons (e.g., between new and existing patients, or between diagnostic groups).

**Primary outcomes (prevalence of mental health problems in all outpatients, N=2,065):**

Anxiety (GAD-7 ≥5): 25.5% of patients

Depression (PHQ-9 ≥5): 16.9% of patients

Insomnia (ISI ≥8): 26.2% of patients

**Secondary outcomes (among existing patients with preexisting psychiatric disorders, N=1,434):**

20.9% (N=300) reported deterioration of their mental health condition related to the pandemic

22.0% (N=315) could not receive routine psychiatric care because of suspended hospital visits

18.1% (N=259) self-reduced medication dosages

17.2% (N=247) stopped taking medication entirely because they could not access prescriptions

**Among new patients (N=589):**

24.5% (N=144) could not receive timely diagnosis and treatment, including those with anxiety (N=46), depression (N=37), insomnia (N=79), and psychosis (N=21)

**Use of online services:**

Only 7.4% of patients with mental disorders (N=136 out of 1,434) had sought online help for medical care, despite a free online outpatient service being available from January 26, 2020

**Note on effect sizes and statistics:** The paper reports only raw percentages. No confidence intervals, p-values, or effect sizes are provided. The authors do not compare these rates to any control group or pre-pandemic baseline. The findings are purely descriptive.

Because this is a cross-sectional survey without a comparison group, we cannot calculate an effect size in the traditional sense (e.g., Cohen's d, risk ratio). However, we can contextualise the raw numbers:

**Anxiety (25.5%):** Roughly 1 in 4 outpatients scored above the mild threshold for anxiety. For comparison, pre-pandemic estimates of anxiety disorders in Chinese general hospital outpatients ranged from 8–15% depending on the setting and cutoff. The 25.5% figure is likely inflated by the low cutoff (GAD-7 ≥5 instead of ≥10).

**Depression (16.9%):** About 1 in 6 outpatients scored above the mild threshold. Pre-pandemic estimates for depressive disorders in Chinese hospital outpatients were around 10–12%.

**Insomnia (26.2%):** About 1 in 4 outpatients reported insomnia symptoms above the subthreshold cutoff. Pre-pandemic insomnia prevalence in Chinese adults was approximately 15–20%.

**Medication disruption:** Nearly 1 in 5 existing patients reduced or stopped their psychiatric medication. This is a clinically significant finding because abrupt discontinuation of antidepressants, antipsychotics, or mood stabilisers can trigger relapse, withdrawal symptoms, or worsening of the underlying condition.

**Online service uptake:** Only 7.4% used available telemedicine, meaning over 92% of patients with mental disorders did not access digital care despite barriers to in-person visits. This suggests a massive gap between service availability and actual utilisation.

**Acknowledged by authors:**

The authors note that "data are still being gathered" and that "long-term outcomes... need further evaluation." They do not claim their findings are definitive.

They mention that the survey was conducted at a single hospital and that results may not be generalisable.

**Additional critical limitations:**

1. **No pre-pandemic comparison:** The most fundamental limitation. Without knowing baseline rates in this same population, we cannot attribute the observed prevalence to the pandemic. The paper's title implies a "mental health response to COVID-19," but the design cannot support that causal claim.

2. **Low symptom thresholds inflate prevalence:** Using GAD-7 ≥5 (instead of ≥10) and PHQ-9 ≥5 (instead of ≥10) likely doubles or triples the reported prevalence compared to standard clinical cutoffs. Many of these "cases" may represent normal stress reactions, not clinical disorders.

3. **Self-report bias:** Patients may over-report symptoms due to the salience of the pandemic, or under-report due to stigma. The survey was anonymous but administered through a hospital-affiliated WeChat account, which may have influenced responses.

4. **Single time point:** The survey captures only a 13-day window. Mental health fluctuates; a single snapshot cannot distinguish transient distress from persistent disorder.

5. **No diagnostic verification:** The GAD-7, PHQ-9, and ISI are screening tools, not diagnostic instruments. They have high sensitivity but low specificity — meaning they catch many false positives.

6. **Selection bias:** Patients who were well enough to complete an online survey may be less severely ill than those who did not respond. Conversely, those with severe anxiety or depression may have been more motivated to participate. The net direction of bias is unknown.

7. **No data on non-responders:** 40% of approached patients did not complete the survey. If non-responders had worse mental health (e.g., too depressed to fill out forms), the reported prevalence is an underestimate. If they had better mental health (e.g., not worried about COVID), it's an overestimate.

8. **No adjustment for confounders:** The paper does not control for age, sex, socioeconomic status, severity of preexisting illness, or other factors that could influence mental health outcomes.

9. **Industry funding:** The authors report no financial relationships with commercial interests, which is a strength. However, the study was conducted at a single institution with no external funding mentioned, which limits resources for rigorous methodology.

10. **Generalisability:** West China Hospital is a leading tertiary hospital in a major city. Patients there likely have higher education, income, and digital literacy than the average Chinese citizen. Rural populations, migrant workers, and elderly individuals (who may not use WeChat) are not represented.

For someone running their own n=1 experiment on mental health during a crisis or lockdown:

### What to test

**Intervention:** Structured daily routine (fixed wake/sleep times, scheduled meals, dedicated work/leisure blocks) vs. unstructured days. Alternatively: daily mindfulness practice (10–20 minutes) vs. no practice. Or: daily social connection (video call with a friend) vs. no scheduled connection.

**Dose:** For routine: maintain for at least 14 consecutive days. For mindfulness: 10–20 minutes daily. For social connection: one 15–30 minute video call per day.

**Comparator:** Your own baseline (measure for 7 days before starting the intervention) or alternating weeks (e.g., week 1: routine, week 2: no routine, repeat).

### Minimum meaningful duration

**At least 14 days per condition.** Mental health changes during crises can be rapid, but a minimum of two weeks allows you to see trends beyond daily fluctuations. For medication-related experiments (e.g., checking if you can maintain adherence during a lockdown), track for the entire duration of the disruption.

### What to measure (specific metrics)

**Daily mood:** Use a 0–10 scale (0 = worst ever, 10 = best ever) for anxiety, depression, and irritability. Record once in the evening.

**Sleep:** Use the Insomnia Severity Index (ISI) once per week (free online). Also track: time to fall asleep, total sleep time, number of night awakenings (daily sleep diary).

**Medication adherence:** If you take psychiatric medication, log daily: did you take the prescribed dose? (yes/no). Track any skipped or reduced doses.

**Access to care:** Note any barriers: pharmacy closed, unable to get prescription, fear of leaving home, telehealth appointment made or missed.

**Social connection:** Log number of meaningful conversations per day (≥10 minutes) and subjective feeling of loneliness (0–10 scale).

**Physical activity:** Steps per day (phone pedometer) or minutes of intentional exercise.

### Key confounds to control for

**News exposure:** Track minutes per day consuming COVID-19 news. High exposure is strongly associated with anxiety. Try to keep this constant across conditions.

**Alcohol/caffeine use:** Both affect sleep and mood. Log daily consumption and keep stable.

**Work status:** Were you working from home, furloughed, or essential worker? Changes in work status are a major confound.

**Social isolation:** Number of people in your household, whether you can leave the house, and frequency of in-person contact. These are structural factors you may not be able to control, but you should measure them.

**Physical health:** Any COVID-19 symptoms, other illness, or chronic pain. Log daily.

**Time of year:** Daylight hours affect mood. If your experiment spans March (spring equinox) to June (summer solstice), seasonal effects may confound results.

### What a positive result would look like

**For a routine intervention:** Your average daily anxiety score drops by ≥1.5 points (on 0–10 scale) during routine weeks compared to non-routine weeks. Your sleep onset latency decreases by ≥10 minutes. Your ISI score drops by ≥3 points.

**For mindfulness:** Your average daily depression score drops by ≥1 point. You report fewer "bad days" (score ≤3 on mood scale) — e.g., from 4 out of 14 days to 1 out of 14 days.

**For medication adherence:** You maintain 100% adherence during the intervention period vs. missing ≥2 doses during the control period.

**Important caveat:** Because this is an n=1 experiment, you cannot generalise to anyone else. A "positive result" means you personally feel better, sleep better, or maintain treatment. Use statistical process control (plot your daily scores on a run chart) to see if changes exceed normal day-to-day variation. A change of ≥2 standard deviations from your baseline mean is a reasonable threshold for a meaningful effect.

**Bottom line for self-experimenters:** The key lesson from this paper is not about specific interventions but about the *systemic vulnerability* of mental health care during disruptions. If you rely on regular medication, therapy, or clinic visits, have a backup plan: a 30-day emergency supply of medication, a telehealth provider identified in advance, and a written crisis plan. Test your backup plan *before* a crisis hits — run a 1-week simulation