How to use Newcastle-Ottawa Scale (NOS) – Adapted for Experimental Studies
The Newcastle-Ottawa Scale (NOS) is primarily designed for non-randomized studies (e.g., cohort and case-control studies), but it can be adapted for experimental studies (e.g., quasi-experimental or non-randomized trials). Below is a step-by-step guide on how to use it effectively.
1. Understand the NOS Structure (Adapted for Experimental Studies)
The NOS assesses studies based on three domains, each with sub-questions rated by stars (★). The maximum score is 9★ (highest quality).
| Domain | Key Criteria (Adapted for Experimental Studies) |
|---|---|
| Selection (4★ max) | – Representativeness of exposed cohort – Selection of non-exposed group – Ascertainment of exposure – Demonstration that outcome was not present at start |
| Comparability (2★ max) | – Controls for key confounders (e.g., age, sex, baseline differences) – Additional adjustments (if applicable) |
| Outcome (3★ max) | – Independent/objective outcome assessment – Adequate follow-up duration – Completeness of follow-up |
2. Step-by-Step Appraisal Process
A. Selection (Max 4★)
- Representativeness of the Exposed Group (1★)
- ★ If the sample represents the target population (e.g., similar demographics, disease severity).
- Example: “Patients were recruited from multiple clinics, reflecting real-world diversity.”
- Selection of the Non-Exposed Group (1★)
- ★ If the control group is drawn from the same source as the experimental group.
- Example: “Controls were matched from the same hospital database.”
- Ascertainment of Exposure (1★)
- ★ If exposure (intervention) is reliably recorded (e.g., medical records, standardized protocols).
- Example: “Treatment allocation was documented in trial logs.”
- Demonstration That Outcome Was Not Present at Baseline (1★)
- ★ If the study confirms no outcome differences between groups at baseline.
- Example: “Baseline scores were similar (p > 0.05).”
B. Comparability (Max 2★)
- Controls for Confounders (1–2★)
- ★ If adjusted for key variables (e.g., age, sex, comorbidities).
- ★★ If adjusted for multiple major confounders.
- Example: “Analysis controlled for age, BMI, and baseline severity.”
C. Outcome (Max 3★)
- Independent/Blinded Outcome Assessment (1★)
- ★ If outcomes were assessed blindly or objectively.
- Example: “Outcomes were evaluated by an independent blinded assessor.”
- Adequate Follow-Up Duration (1★)
- ★ If follow-up was long enough for outcomes to occur.
- Example: “6-month follow-up was sufficient for detecting treatment effects.”
- Completeness of Follow-Up (1★)
- ★ If ≥80% follow-up rate or losses explained.
- Example: “90% completed follow-up; dropouts were random.”
3. Scoring & Interpretation
- Total Score = Selection (4★) + Comparability (2★) + Outcome (3★)
- Interpretation:
- ≥7★ = High quality
- 5–6★ = Moderate quality
- ≤4★ = Low quality
Example Assessment
| Domain | Criteria | Stars (★) |
|---|---|---|
| Selection | Representativeness, matched controls | ★★★★ (4/4) |
| Comparability | Adjusted for age, sex, severity | ★★ (2/2) |
| Outcome | Blinded assessment, 85% follow-up | ★★★ (3/3) |
| Total | 9★ (High Quality) |
4. When to Use NOS for Experimental Studies
- Quasi-experimental studies (non-randomized trials)
- Before-after studies (if control group is present)
- Comparative effectiveness research (real-world evidence)
Limitations:
- Not ideal for RCTs (use Cochrane RoB 2 instead).
- Subjectivity in star allocation (use multiple reviewers for consistency).
5. Comparison with Other Tools
| Tool | Best For | Max Score |
|---|---|---|
| NOS (Adapted) | Non-randomized/quasi-experimental | 9★ |
| Cochrane RoB 2 | RCTs | Low/Some/High risk |
| SIGN RCT Checklist | RCTs (guideline development) | ++/+/- |
Final Tips
✔ Use alongside other tools (e.g., ROBINS-I for non-randomized studies).
✔ Pre-define scoring rules to reduce subjectivity.
✔ Report NOS scores in systematic reviews for transparency.
Would you like a real-world example of an NOS appraisal for a quasi-experimental study?
