The prevalence and association of stress with sleep quality among medical students

This was an observational, cross-sectional study—not an experiment. The researchers did not test any intervention. Instead, they measured two things simultaneously in a single group of medical students:

• Sleep quality (the outcome of interest), measured using the Pittsburgh Sleep Quality Index (PSQI)
• Stress level (the predictor), measured using the Kessler Psychological Distress Scale (K10)

They also collected demographic and academic information: age, sex, year of study, and cumulative grade point average (GPA). The goal was to see whether stress and poor sleep were associated, and whether other factors (like GPA) predicted sleep quality.

There was no comparator group, no random assignment, and no blinding—because this was a survey, not a controlled trial.

• Sample size: 700 medical students (from a total pool of ~1,200 eligible students)
• Population: Male and female medical students at King Saud bin Abdulaziz University for Health Sciences in Riyadh, Saudi Arabia
• Age range: Not explicitly reported, but typical medical student age range is 18–25
• Sex distribution: 50.4% male, 49.6% female (roughly equal)
• Academic years: All five years of the medical program were represented
• Exclusion criteria: Students who did not complete the questionnaire were excluded (no other exclusions reported)

The sample was drawn using stratified random sampling by academic year and sex, which means they tried to get a representative slice of the entire student body. However, because this is a single university in one country, the results may not generalize to medical students elsewhere or to non-medical populations.

Two validated questionnaires were used:

Pittsburgh Sleep Quality Index (PSQI):

• A 19-item self-report questionnaire that assesses sleep quality over the past month
• Yields a global score from 0 to 21 (lower = better sleep)
• A score >5 indicates "poor sleep quality"
• Components include: sleep duration, sleep latency (how long to fall asleep), sleep efficiency (time asleep vs. time in bed), sleep disturbances, use of sleep medication, and daytime dysfunction
• The PSQI has been validated against polysomnography (lab-based sleep measurement) and has high test-retest reliability

Kessler Psychological Distress Scale (K10):

• A 10-item questionnaire measuring non-specific psychological distress (anxiety and depression symptoms) over the past month
• Scores range from 10 to 50 (higher = more distress)
• A score ≥20 is considered "high stress" (the cutoff used in this study)
• The K10 is widely used in population surveys and has good sensitivity for detecting common mental disorders

Additional variables collected:

• Age, sex, academic year
• Cumulative GPA (on a 5-point scale, where 4.25 is roughly a B+ average)
• Self-reported height and weight (to calculate BMI, though BMI results were not a focus of the paper)

Important measurement limitation: Both sleep quality and stress were measured via self-report at a single time point. This means the study captures perceived stress and perceived sleep quality, not objective measures. People who are stressed may also perceive their sleep as worse than it objectively is (negative reporting bias), which could inflate the association.

Study design

This was a cross-sectional observational study. The researchers surveyed a sample of medical students at one point in time (during the 2015–2016 academic year). They did not follow students over time, nor did they assign anyone to a treatment or control group.

Sampling and recruitment

• Stratified random sampling by academic year and sex
• Questionnaires were distributed during class time to maximize response rate
• 700 out of ~1,200 eligible students completed the survey (response rate ~58%)
• Data were collected anonymously

Statistical approach

• Descriptive statistics (means, percentages) to report prevalence
• Chi-square tests to examine associations between categorical variables (e.g., stress yes/no vs. poor sleep yes/no)
• Logistic regression to calculate odds ratios (OR) for predictors of poor sleep quality
• Variables entered into the regression model: stress level, sex, academic year, GPA, and BMI
• Significance set at p < 0.05

What this design can and cannot prove

What it can prove:

• The prevalence of poor sleep and stress in this population at this time
• That stress and poor sleep are statistically associated (they co-occur more often than chance)

What it cannot prove:

• Causality: Because both variables were measured at the same time, we cannot tell whether stress causes poor sleep, poor sleep causes stress, or both are caused by a third factor (e.g., exam pressure, caffeine use, or a noisy dormitory). This is the classic "chicken or egg" problem of cross-sectional data.
• Direction of effect: Even if there is a causal relationship, we don't know which direction it runs. Stress could disrupt sleep; poor sleep could impair coping and increase stress; or both could be true in a vicious cycle.
• Generalizability: Results from Saudi medical students may not apply to other populations (different cultures, different academic systems, different ages).
• Objective sleep quality: Self-reported sleep quality does not always match objective measures like actigraphy or polysomnography. People with depression or anxiety may overestimate sleep problems.

Major methodological weaknesses

1. Self-report bias: Both stress and sleep were measured by questionnaire. Someone who is stressed may also report worse sleep simply because they are more aware of negative experiences.
2. Single time point: Cannot assess change over time or establish temporal order.
3. Moderate response rate (58%): Non-responders may have different sleep or stress patterns than responders, introducing selection bias.
4. No objective sleep measure: No actigraphy, no sleep diaries, no polysomnography.
5. Single institution: Results may reflect local culture, curriculum, or dormitory conditions rather than universal truths.

Primary outcome: Prevalence of poor sleep quality

• 76% of medical students had PSQI scores >5, indicating poor sleep quality
• This is substantially higher than the general population (typically 20–30% in adult samples)

Primary outcome: Prevalence of high stress

• 53% of students scored ≥20 on the K10, indicating high psychological distress
• This is also elevated compared to general population norms (typically 10–20%)

Association between stress and sleep quality

• Among students with high stress, 87% had poor sleep quality
• Among students without high stress, 64% had poor sleep quality
• Chi-square test: p < 0.001 (statistically significant association)
• Logistic regression: Students who were NOT suffering from stress were less likely to have poor sleep quality (OR = 0.28, p < 0.001)
  • Translation: The odds of poor sleep were 72% lower in students without high stress compared to those with high stress
  • Alternatively: Students with high stress had about 3.6 times the odds of poor sleep compared to those without high stress (1/0.28 ≈ 3.57)

GPA and sleep quality

• Students with a cumulative GPA less than 4.25 (on a 5-point scale) had 3.83 times the odds of poor sleep quality compared to students with GPA ≥ 4.25 (OR = 3.83, p = 0.01)
• This was a secondary finding—the study was not designed to test GPA effects specifically

Other variables

• Sex: No significant association with sleep quality (p > 0.05)
• Academic year: No significant association (p > 0.05)
• BMI: No significant association (p > 0.05)

Summary of logistic regression model

The final model included stress, GPA, sex, academic year, and BMI. Only stress and GPA were significant predictors of poor sleep quality.

The key finding is that stress and poor sleep are strongly linked, but the cross-sectional design prevents us from knowing which causes which.

In plain English:

• If you are a medical student with high stress, you have roughly 3.6 times the odds of having poor sleep compared to a student without high stress
• If your GPA is below 4.25 (roughly a B+ average), you have nearly 4 times the odds of poor sleep compared to a student with a higher GPA

However, these are odds ratios, not risk ratios. Because poor sleep is very common (76% overall), the odds ratio overestimates the relative risk. A more intuitive way to state it: among stressed students, 87% had poor sleep vs. 64% among non-stressed students—an absolute difference of 23 percentage points.

The GPA finding is interesting but should be interpreted cautiously. It could mean that poor sleep hurts academic performance, that struggling academically causes stress which disrupts sleep, or that a third factor (e.g., poor time management, high anxiety) affects both GPA and sleep.

Acknowledged by authors

• Cross-sectional design cannot establish causality
• Single institution limits generalizability
• Self-report measures may introduce bias
• Possible confounding by unmeasured variables (e.g., caffeine intake, exercise, social support, living conditions)

Additional critical notes

• No objective sleep measurement: The PSQI asks about sleep over the past month, which relies on memory and may be inaccurate. People with depression or anxiety tend to overestimate sleep problems.
• No objective stress measurement: No cortisol, heart rate variability, or other physiological stress markers.
• No control for exam period: Data collection timing relative to exams is not reported. Stress and sleep vary dramatically across the academic calendar.
• Moderate response rate (58%): If students with worse sleep or higher stress were less likely to complete the survey (or more likely to skip class that day), the true prevalence could be even higher—or lower if the opposite occurred.
• Cultural specificity: Saudi medical students live in a particular cultural and educational context. Dormitory arrangements, social expectations, and curriculum structure may differ substantially from other countries.
• No adjustment for multiple comparisons: Several statistical tests were run, which inflates the risk of false positives.
• GPA cutoff (4.25) was chosen post-hoc: The authors do not justify why 4.25 was used as the threshold. This could be data-driven rather than hypothesis-driven.

For someone running their own n=1 experiment to explore the relationship between stress and sleep:

What to test

• Specific intervention: Try a structured stress-reduction technique (e.g., 10-minute mindfulness meditation before bed, or a 20-minute evening walk) versus your usual routine
• Dose: Start with daily practice for at least 2 weeks
• Comparator: Your own baseline (no intervention) measured for 1–2 weeks before starting

Minimum meaningful duration

• Baseline phase: 7–14 days of tracking without any changes to your routine
• Intervention phase: At least 14–21 days of consistent practice
• Total experiment: 3–5 weeks minimum
• Why: Sleep patterns take time to shift, and stress levels fluctuate day-to-day. A single week may capture random variation rather than true change.

What to measure (specific metrics)

• Sleep quality: Use the PSQI (self-administered online) at the beginning and end of each phase. A change of 2–3 points is considered clinically meaningful.
• Sleep diary (daily): Record bedtime, wake time, estimated time to fall asleep, number of night awakenings, and subjective sleep quality (1–10 scale)
• Stress level: Use the K10 or a simpler 1–10 daily stress rating. Also consider measuring cortisol awakening response (saliva sample within 30 minutes of waking) if you have access to testing kits
• Objective sleep tracker: If available, use a wrist actigraphy device (e.g., Fitbit, Oura Ring, or research-grade device) to measure sleep duration, sleep efficiency, and sleep latency
• Confounders to track daily: Caffeine intake (mg and timing), alcohol, exercise (type and duration), screen time before bed, room temperature, noise level, academic workload (hours studying), and social stressors

Key confounds to control for

1. Caffeine: Record all sources (coffee, tea, soda, energy drinks, chocolate) and timing. Caffeine has a half-life of 4–6 hours and can disrupt sleep even if consumed 6+ hours before bed.
2. Alcohol: Even small amounts can fragment sleep and reduce REM sleep.
3. Exercise timing: Vigorous exercise within 2 hours of bed can raise core body temperature and delay sleep onset.
4. Screen exposure: Blue light from phones/laptops suppresses melatonin. Track screen time in the hour before bed.
5. Academic schedule: Exams, deadlines, and presentations will spike stress and disrupt sleep regardless of your intervention. Note these in your diary.
6. Weekend vs. weekday: Sleep patterns often differ. Analyze weekdays and weekends separately.
7. Menstrual cycle (if applicable): Sleep quality and stress sensitivity vary across the cycle. Track cycle phase.

What a positive result would look like

• PSQI score drops by ≥2 points (e.g., from 8 to 6 or below the clinical cutoff of 5)
• Sleep latency decreases by ≥10 minutes (you fall asleep faster)
• Sleep efficiency increases to ≥85% (time asleep divided by time in bed)
• Daily stress ratings drop by ≥1.5 points on a 1–10 scale
• Consistency: The improvement should be visible in your daily sleep diary, not just in the before/after questionnaires. Look for a clear shift in the trend line after starting the intervention.
• Replication: If possible, repeat the experiment (intervention, then return to baseline, then intervention again) to confirm the effect is real and not due to random fluctuation or the passage of time.

Important caveat from this study

This paper shows that stress and poor sleep are strongly correlated, but it cannot tell you whether reducing stress will improve your sleep, or whether improving your sleep will reduce your stress. In your n=1 experiment, you can test both directions:

• Test A: Reduce stress (e.g., meditation, exercise, therapy) and measure sleep
• Test B: Improve sleep hygiene (e.g., fixed bedtime, no screens, cool room) and measure stress

If both interventions improve both outcomes, you have evidence for a bidirectional relationship—and you can choose whichever intervention is easier to sustain.