Depression, Anxiety and Stress during COVID-19: Associations with Changes in Physical Activity, Sleep, Tobacco and Alcohol Use in Australian Adults

This was a cross-sectional observational survey, not an experiment. The researchers tested whether self-reported changes in four health behaviors (physical activity, sleep quality, alcohol intake, and cigarette smoking) since the start of the COVID-19 pandemic were associated with current levels of psychological distress (depression, anxiety, and stress). They also examined demographic factors (age, sex, income, relationship status, chronic illness) that might predict who was most affected.

There was no intervention — the study simply asked people to recall how their behaviors had changed and then measured their current mental health status. The "comparator" was people who reported no change or positive change in those behaviors.

**Sample size:** 1,491 adults

**Age:** Mean 50.5 years (standard deviation 14.9 years; range approximately 18–90+)

**Sex:** 67% female, 33% male

**Setting:** Australian adults recruited through social media (Facebook, Twitter, LinkedIn), email lists, and professional networks during April 2020 — the first month of Australia's COVID-19 lockdown

**Inclusion criteria:** Aged 18 years or older, living in Australia, able to complete an online survey in English

**Exclusion criteria:** None explicitly stated

**Key demographics:** 68% were married or in a de facto relationship; 55% had a university degree; 47% were employed full-time; 15% had a chronic illness (diabetes, heart disease, respiratory disease, or cancer)

The study used a single online survey (Qualtrics platform) with several validated instruments and custom questions:

**Depression, Anxiety, and Stress:** The Depression Anxiety Stress Scales-21 (DASS-21). This is a 21-item questionnaire with three subscales (depression, anxiety, stress), each scored 0–42. Higher scores = worse symptoms. Clinical cutoffs: Depression >9, Anxiety >7, Stress >14 indicate moderate or above severity.

**Physical activity change:** A single question: "Since the onset of the COVID-19 pandemic, has your physical activity level changed?" Response options: "increased," "decreased," or "no change." No objective measurement (no accelerometers, no activity logs).

**Sleep change:** A single question: "Since the onset of the COVID-19 pandemic, has your sleep quality changed?" Response options: "improved," "worsened," or "no change." No validated sleep scale (no PSQI, no actigraphy).

**Alcohol change:** A single question: "Since the onset of the COVID-19 pandemic, has your alcohol intake changed?" Response options: "increased," "decreased," or "no change." No quantity/frequency measure.

**Smoking change:** A single question: "Since the onset of the COVID-19 pandemic, has your cigarette smoking changed?" Response options: "increased," "decreased," or "no change." No cotinine levels or pack-year history.

**Demographics:** Age, sex, relationship status, income bracket, employment status, education level, presence of chronic illness.

**Study design:** Cross-sectional observational survey. This is a "snapshot in time" design — data on both the exposure (behavior change) and outcome (psychological distress) were collected simultaneously.

**Recruitment:** Convenience sampling via social media and professional networks. This is non-probability sampling — participants self-selected into the study.

**Duration:** The survey was open during April 2020. Each participant completed it once (approximately 15–20 minutes). There was no follow-up.

**Statistical approach:** The researchers used multiple linear regression models. They tested whether self-reported behavior change (decreased, increased, or no change) predicted DASS-21 scores, after controlling for age, sex, income, relationship status, and chronic illness. They reported unstandardized beta coefficients (B), 95% confidence intervals (CIs), and p-values. They also ran separate models for each behavior and each distress outcome (depression, anxiety, stress).

**What this design can prove:**

It can identify **associations** between behavior change and psychological distress during a specific time period

It can show which demographic groups reported the highest distress

It can generate hypotheses for future experimental or longitudinal research

**What this design cannot prove:**

**Causality:** Did worsening behaviors cause distress, or did distress cause behavior changes? The study cannot answer this because both were measured at the same time.

**Direction of effect:** The "chicken or egg" problem is unsolvable with cross-sectional data.

**Actual behavior change:** The study relied on retrospective self-report of change, not objective measurement before and during lockdown. People's memories are unreliable.

**Magnitude of change:** "Decreased physical activity" could mean going from 5 days/week to 4 days/week, or from 3 days/week to 0. The study treats all "decreases" as equivalent.

**Major methodological weaknesses:**

1. **Single-item, non-validated measures** for behavior change — no established scales, no objective verification

2. **Recall bias** — people were asked to remember how their behavior had changed over the past month, which is prone to error and mood-congruent memory (people feeling distressed may be more likely to remember negative changes)

3. **Convenience sample** — 67% female, 55% university-educated, mostly employed. Not representative of the general Australian population

4. **No pre-pandemic baseline** — the study didn't measure actual pre-COVID behavior; it asked people to estimate change retrospectively

5. **Self-selection bias** — people already interested in health or distressed by COVID may have been more likely to participate

6. **Cross-sectional design** — cannot establish temporal sequence or causality

**Prevalence of negative behavior change (self-reported):**

Physical activity decreased: 48.9% of participants

Sleep quality worsened: 40.7%

Alcohol intake increased: 26.6%

Cigarette smoking increased: 6.9%

**Demographic predictors of higher psychological distress (all p < 0.05):**

Females had higher depression, anxiety, and stress scores than males (exact beta coefficients not reported for sex alone)

People not in a relationship had higher depression (B = 1.72, 95% CI: 0.66 to 2.78, p = 0.001) and stress (B = 1.62, 95% CI: 0.56 to 2.68, p = 0.003) compared to those in a relationship

Lowest income category (<$30,000 AUD/year) had higher depression (B = 3.02, 95% CI: 1.42 to 4.62, p < 0.001), anxiety (B = 1.96, 95% CI: 0.79 to 3.13, p = 0.001), and stress (B = 2.71, 95% CI: 1.11 to 4.31, p = 0.001) compared to highest income category

Age 18–45 years had higher depression, anxiety, and stress than those aged 46+ (exact coefficients not reported for age bands)

People with a chronic illness had higher depression (B = 2.26, 95% CI: 1.00 to 3.52, p < 0.001), anxiety (B = 1.86, 95% CI: 0.93 to 2.79, p < 0.001), and stress (B = 2.28, 95% CI: 1.02 to 3.54, p < 0.001) compared to those without

**Associations between behavior change and psychological distress (primary analysis):**

*Physical activity (reference = no change):*

Decreased physical activity was associated with higher depression (B = 1.47, 95% CI: 0.59 to 2.35, p = 0.001), higher anxiety (B = 0.90, 95% CI: 0.27 to 1.53, p = 0.005), and higher stress (B = 1.42, 95% CI: 0.54 to 2.30, p = 0.002)

Increased physical activity was NOT significantly associated with any distress outcome (all p > 0.05)

*Sleep (reference = no change):*

Worsened sleep was associated with higher depression (B = 4.48, 95% CI: 3.56 to 5.40, p < 0.001), higher anxiety (B = 2.78, 95% CI: 2.12 to 3.44, p < 0.001), and higher stress (B = 4.60, 95% CI: 3.68 to 5.52, p < 0.001)

Improved sleep was associated with LOWER depression (B = -2.40, 95% CI: -4.39 to -0.41, p = 0.018) and LOWER stress (B = -2.47, 95% CI: -4.46 to -0.48, p = 0.015), but not anxiety (p = 0.076)

*Alcohol (reference = no change):*

Increased alcohol intake was associated with higher depression (B = 1.47, 95% CI: 0.47 to 2.47, p = 0.004), higher anxiety (B = 1.18, 95% CI: 0.46 to 1.90, p = 0.001), and higher stress (B = 1.85, 95% CI: 0.85 to 2.85, p < 0.001)

Decreased alcohol intake was NOT significantly associated with any distress outcome (all p > 0.05)

*Smoking (reference = no change):*

Increased smoking was associated with higher depression (B = 3.53, 95% CI: 1.18 to 5.88, p = 0.003), higher anxiety (B = 3.35, 95% CI: 1.62 to 5.08, p < 0.001), and higher stress (B = 3.85, 95% CI: 1.50 to 6.20, p = 0.001)

Decreased smoking was NOT significantly associated with any distress outcome (all p > 0.05)

**Secondary finding:** The strongest association was between worsened sleep and higher depression/stress — the beta coefficients for sleep (B ≈ 4.5) were roughly 3 times larger than those for physical activity or alcohol (B ≈ 1.5).

To translate these numbers into plain English:

**Worsened sleep** was associated with DASS-21 depression scores about **4.5 points higher** on a 0–42 scale. For context, the clinical cutoff for moderate depression is 10, and for severe depression is 21. A 4.5-point difference is roughly equivalent to moving from "mild" to "moderate" depression severity — a clinically meaningful shift.

**Decreased physical activity** was associated with depression scores about **1.5 points higher** — a smaller but still statistically detectable difference. This is roughly the difference between scoring "normal" versus "mild" on the depression scale.

**Increased alcohol intake** was associated with stress scores about **1.9 points higher** — again, a modest but consistent effect across all three distress measures.

**Increased smoking** showed the largest per-person effect (B ≈ 3.5 for depression), but only 6.9% of participants reported this change, so the population-level impact is smaller.

**Demographic effects were comparable:** Being in the lowest income bracket was associated with depression scores about **3 points higher** — roughly double the effect of decreased physical activity, and similar in magnitude to increased smoking.

**What the authors acknowledge:**

Cross-sectional design prevents causal inference

Self-report measures of behavior change are subjective and may not reflect actual change

Convenience sampling limits generalizability

The survey was conducted early in the pandemic (April 2020) — findings may not apply to later phases

Single-item measures for behavior change have unknown reliability and validity

**What a critical reader would add:**

**No pre-pandemic baseline data:** The study asked people to recall how their behavior had changed, rather than measuring behavior before and during lockdown. This is highly susceptible to recall bias — people who are currently depressed may be more likely to remember (or perceive) negative changes.

**No objective measures:** No accelerometers for physical activity, no sleep diaries or actigraphy, no alcohol biomarkers. The entire analysis rests on subjective, retrospective, single-item questions.

**Mood-congruent memory bias:** Current psychological distress likely colors people's recall of behavior change. Someone feeling depressed may be more likely to report "my sleep got worse" even if objective sleep quality was unchanged.

**No dose-response data:** "Decreased physical activity" could mean anything from "I stopped going to the gym" to "I went from running 10K daily to sitting on the couch." Without quantifying the change, we cannot know what level of behavior change matters.

**Multiple comparisons:** The authors ran 36 separate regression models (4 behaviors × 3 distress outcomes × 3 change categories). They did not adjust for multiple comparisons (e.g., Bonferroni correction), which inflates the risk of false positives.

**Low response rate denominator:** The authors don't report how many people saw the survey link versus completed it. Self-selection bias is likely — people who were already distressed or health-conscious may have been more motivated to participate.

**Single time point:** We don't know if these associations persisted, strengthened, or weakened over time. A one-month snapshot during an unprecedented global crisis may not reflect stable patterns.

**No control for pre-existing mental health:** The study didn't ask about prior diagnosis of depression, anxiety, or stress disorders. People who were already distressed before COVID may have been more likely to change their behaviors — the study can't separate pre-existing from pandemic-related distress.

For someone running their own n=1 experiment:

**What to test (specific intervention and dose):**

**Sleep quality:** Test a consistent sleep schedule (same bedtime and wake time ±30 minutes, 7–9 hours) for 2–4 weeks. Or test a specific sleep hygiene intervention: no screens 60 minutes before bed, bedroom temperature 18–20°C, no caffeine after 2 PM.

**Physical activity:** Test a minimum dose of 20–30 minutes of moderate aerobic activity (brisk walking, jogging, cycling) at least 5 days per week. Or test a specific type (e.g., outdoor walking vs. indoor workout).

**Alcohol reduction:** Test reducing alcohol intake by 50% (e.g., from 2 drinks/day to 1, or from 5 drinks/week to 2–3). Or test complete abstinence for 2–4 weeks.

**Smoking reduction:** If you smoke, test a structured reduction plan (e.g., nicotine replacement therapy, scheduled reduction by 1 cigarette every 3 days).

**Minimum meaningful duration:**

**Sleep:** 2–3 weeks minimum. Sleep changes take 7–14 days to stabilize, and mood effects may take another week to detect.

**Physical activity:** 3–4 weeks minimum. Acute mood effects occur after a single session, but sustained changes in baseline mood require consistent activity over several weeks.

**Alcohol:** 2–4 weeks minimum. Acute withdrawal effects (first 3–7