How should the certainty of evidence for the prevalence and risk factors of shift‑work disorder in nurses be interpreted?

Interpreting Certainty of Evidence for Shift-Work Disorder in Nurses

The certainty of evidence for shift-work disorder (SWD) prevalence and risk factors in nurses should be interpreted as moderate to low, with substantial heterogeneity across studies driven primarily by inconsistent diagnostic criteria and assessment methods, though recent meta-analyses provide increasingly robust prevalence estimates around 45.5%.

Understanding the Evidence Quality

The body of evidence reveals significant methodological challenges that directly impact certainty assessments:

Diagnostic Heterogeneity as the Primary Limitation

The choice of diagnostic criteria substantially affects reported prevalence rates, creating the largest source of uncertainty 1, 2. Studies using:

• ICSD-2 criteria: 32.9% prevalence
• ICSD-3 criteria: 40.2% prevalence
• Symptom-based scales/questionnaires: 54.2% prevalence

This 21-percentage-point variation from diagnostic method alone represents a critical limitation when interpreting the evidence 3. The most recent systematic review found prevalence ranging from 4.8% to 44.3% depending on work schedule, with overall rates of 32.4-37.6% varying by assessment method 4.

Evolving Quality of Exposure Assessment

The quality and consistency of shift work exposure measurement has improved substantially over time 1. Earlier studies suffered from:

• Crude exposure definitions (simply "employed in shift work")
• Inconsistent categorization of evening versus night work
• Potential misclassification of controls who may have included night workers
• Limited assessment of shift intensity, duration, and rotation patterns

More recent case-control studies demonstrate superior exposure assessment methods, with the largest and highest-quality studies now given greater prominence in evaluations 1.

Interpreting the Most Recent Meta-Analysis

The 2026 systematic review and meta-analysis provides the strongest current evidence 3:

• Pooled prevalence: 45.5% (95% CI: 39.2-51.8%)
• Sample: 24 observational studies, 15,479 nurses
• Geographic variation: Africa (51.0%), Asia (46.5%) highest
• Schedule-specific: Three-shift model showed 61.7% prevalence

This represents moderate-certainty evidence because while the sample size is substantial and methodology rigorous, the I² heterogeneity remains high across studies, reflecting genuine differences in populations, work environments, and measurement approaches.

Risk Factor Evidence: Certainty Levels

High-Certainty Risk Factors

The following associations demonstrate consistent, dose-response relationships across multiple studies 3, 5:

• Number of night shifts per year (OR = 1.010-1.020 per shift)
• Quick return shifts (<11 hours between shifts: OR = 2.593)
• Depression symptoms (OR = 1.178)
• Fatigue (OR = 1.145)
• Anxiety (OR = 1.073)

These factors show reproducible effects with tight confidence intervals and biological plausibility.

Moderate-Certainty Risk Factors

• Age (OR = 1.025): Consistent direction but smaller effect size
• Stress levels (OR = 1.019): Measured variably across studies
• Chronotype: Associated but measurement inconsistent 4, 5

Low-Certainty Associations

Melatonin levels and circadian gene expression show inconsistent relationships with cancer risk despite strong experimental evidence for circadian disruption mechanisms 1. Only a few studies have examined circadian gene variations, with no consistent evidence emerging 1.

Critical Methodological Concerns

Residual Confounding

Early studies, particularly those examining long-term health outcomes like cancer, suffered from:

• Inadequate control for lifestyle factors
• Limited data on potential confounders
• Small numbers of exposed cases
• Lack of exposure-response relationships 1

Bias Cannot Be Excluded

Given the heterogeneity of findings across study designs, bias cannot be excluded with reasonable confidence 1. Cohort studies show less consistent results than case-control studies, suggesting potential selection or recall bias in retrospective designs.

Practical Interpretation Framework

When evaluating evidence certainty for clinical or organizational decision-making:

1. For prevalence estimates: Use the most recent meta-analytic pooled estimate (45.5%) but recognize 95% CI spans 39.2-51.8%, indicating meaningful uncertainty 3

2. For risk factor identification: Prioritize factors with:

   • Dose-response relationships (night shift frequency, quick returns)
   • Consistent effects across study designs
   • Biological plausibility (mental health, fatigue)

3. For intervention planning: Focus on modifiable organizational factors with strongest evidence:

   • Reducing rapid shift rotations
   • Limiting consecutive night shifts
   • Ensuring adequate recovery time between shifts
   • Addressing mental health support 3

4. Acknowledge measurement limitations: Standardized assessment tools remain lacking, with validation studies only recently emerging 6, 2

Geographic and Population Considerations

Certainty varies by geographic region and healthcare system 3. African and Asian studies show higher prevalence, potentially reflecting:

• Different staffing ratios
• Longer shift durations
• Less regulated work schedules
• Cultural differences in symptom reporting

This geographic heterogeneity reduces certainty when generalizing findings across settings.

The Assessment Tool Problem

The lack of validated, standardized assessment methodology represents the single greatest threat to evidence certainty 2. Until consensus emerges on diagnostic criteria and measurement approaches, comparative research remains challenging and certainty assessments must account for this fundamental limitation.

The SWDI shows promise with excellent internal consistency (α = .94) and good convergent/divergent validity 6, but requires broader validation across populations and settings before becoming a reference standard.