You're staring at a pyramid. Worth adding: maybe it's the classic evidence-based medicine triangle. Maybe it's a GRADE framework diagram. Now, either way, there it is: systematic reviews and meta-analyses sitting pretty at the top. Randomized controlled trials right below them. Cohort studies, case-control studies, case series, expert opinion — all stacked in descending order of trustworthiness.
Then you spot "literature review" somewhere near the bottom. Or maybe it's not on the pyramid at all.
So what level of evidence is a literature review? The short answer: it depends entirely on what kind of literature review you're talking about. And most people asking this question don't realize there's more than one kind Simple, but easy to overlook. Worth knowing..
What Is a Literature Review
At its core, a literature review is exactly what it sounds like — a review of the literature. You gather published work on a topic, read it, synthesize it, and write about what you found. That said, that's the process. But the product can look wildly different depending on who's doing it, why they're doing it, and how rigorous they're being.
Here's where the confusion starts. In real terms, in nursing school, you might write a "literature review" for a term paper. Which means in a PhD dissertation, you write a "literature review" chapter. Day to day, a journal might publish a "literature review" as a standalone article. And Cochrane publishes "systematic reviews" — which are also, technically, literature reviews Which is the point..
They're not the same thing. Not even close.
Narrative reviews — the traditional kind
This is what most people mean when they say "literature review" without any modifier. Plus, a narrative review (sometimes called a traditional or non-systematic review) is essentially an expert's guided tour of a topic. The author picks papers they know, maybe searches a database or two, reads them, and writes a coherent narrative about the state of the field.
There's no required search strategy. That's why no PRISMA flowchart. No inclusion/exclusion criteria. No risk of bias assessment. The author decides what matters and what doesn't Most people skip this — try not to..
That doesn't make it bad. Because of that, a great narrative review by a genuine expert can be incredibly valuable — they spot patterns, connect dots, identify gaps that a rigid systematic approach might miss. But it's inherently subjective. Two experts writing narrative reviews on the same topic will produce different articles. That's a feature, not a bug, but it also means you can't treat it as definitive evidence And that's really what it comes down to. That alone is useful..
Systematic reviews — a different beast entirely
A systematic review is a literature review, but it follows a strict, pre-specified protocol. You define a research question (usually PICO format). That's why you search comprehensively — multiple databases, grey literature, trial registries, reference lists. That said, you screen results against explicit inclusion/exclusion criteria. You assess risk of bias for every included study. You extract data systematically. And if the studies are similar enough, you meta-analyze them.
The key word is "systematic.Here's the thing — " The process is designed to be reproducible. Another team following the same protocol should get the same results.
This is why systematic reviews sit at the top of evidence pyramids. They're not opinion. They're a research methodology in their own right Worth keeping that in mind..
Scoping reviews, rapid reviews, umbrella reviews — the middle ground
There's a whole taxonomy now. Scoping reviews map the literature broadly — "what's out there on this topic?" — without necessarily assessing quality or synthesizing findings quantitatively. And rapid reviews use systematic methods but with shortcuts (single reviewer, limited databases, shorter timeframe) to answer urgent questions. Umbrella reviews (or overviews of reviews) synthesize multiple systematic reviews on the same topic Not complicated — just consistent..
Each has its place. Each has a different evidence level.
Why It Matters / Why People Care
You're probably reading this because you need to cite something. Maybe you're grading evidence for a clinical decision. Maybe you're writing a guideline. Maybe you're peer reviewing a paper and the authors cited a narrative review as "Level I evidence Still holds up..
That's a problem.
Evidence hierarchies exist for a reason. Plus, when we make clinical decisions, write guidelines, or set policy, we want the least biased estimate of effect. In practice, both are useful. Narrative reviews give us someone's informed perspective. That said, systematic reviews of RCTs give us that — or at least, they get us closest. But they're not interchangeable.
Here's what goes wrong when people conflate them:
Guideline committees sometimes cite narrative reviews as if they were systematic reviews. The result? Recommendations based on cherry-picked evidence.
Students and early-career researchers waste months writing narrative reviews thinking they're doing systematic reviews — or vice versa, they attempt a systematic review without understanding the methodological rigor required.
Clinicians read a "review article" in a high-impact journal and assume it's a comprehensive, unbiased summary. Often it's a narrative review written by a key opinion leader with industry ties. That doesn't make it wrong. But it makes it not a systematic review.
Journal editors sometimes publish narrative reviews without labeling them clearly. "Review" in the article type field tells you nothing about methodology Worth keeping that in mind..
The evidence level determines how much weight a review should carry in decision-making. That's why this distinction isn't academic pedantry — it's practical That alone is useful..
How Evidence Levels Actually Work
Most evidence hierarchies follow a similar logic, though the exact labels vary. Here's a common framework (adapted from Oxford Centre for Evidence-Based Medicine, GRADE, and others):
Level I — Systematic reviews of RCTs (or individual RCTs with narrow confidence intervals)
Level II — Systematic reviews of cohort studies / individual cohort studies / low-quality RCTs
Level III — Case-control studies / systematic reviews of case-control studies
Level IV — Case series / case reports
Level V — Expert opinion / narrative reviews / consensus statements
Notice where "narrative review" lands. Same as expert opinion. That's why level V. Because methodologically, that's what it is — an expert's synthesis, filtered through their experience and biases.
But a systematic review of RCTs? Also, level I. On top of that, a systematic review of cohort studies? Level II. The methodology determines the level, not the word "review" in the title.
What about GRADE?
GRADE (Grading of Recommendations Assessment, Development and Evaluation) doesn't use simple levels. It rates certainty of evidence across four domains: risk of bias, inconsistency, indirectness, imprecision, and publication bias. A systematic review starts as "high certainty" if it includes RCTs — but gets downgraded for flaws. A narrative review doesn't enter GRADE at all because it's not a body of evidence; it's a summary of evidence (or opinion) Still holds up..
This matters. If you're doing a GRADE evidence profile, you don't include narrative reviews as evidence. You include the studies the narrative review discusses — if you can find and assess them yourself.
The JBI levels
Joanna Briggs Institute has its own hierarchy. Level 2: Systematic reviews of quasi-experimental studies. Level 3: Systematic reviews of observational studies. And so on. Level 4: Individual experimental studies. Level 1: Systematic reviews of experimental studies. Narrative reviews don't appear as a level — they're considered "background information" or "expert opinion Not complicated — just consistent. Simple as that..
Common Mistakes / What Most People Get Wrong
Mistake 1: Assuming "review article" = systematic review
Journals publish plenty of narrative reviews and label them "Review.No methods section? " The article type field is useless for this. Worth adding: it's narrative. On top of that, you have to read the methods section. " Some even publish systematic reviews and label them "Review.Methods section describes a protocol, search strategy, screening process, risk of bias assessment?
Mistake 2: Conflating Study Design with Quality
A frequent error is treating study design as synonymous with quality. While RCTs (Level I) are generally more rigorous than case series (Level IV), not all RCTs are created equal. An RCT with poor randomization, inadequate blinding, or high dropout rates may provide less reliable evidence than a well-conducted cohort study. Plus, gRADE addresses this by allowing systematic reviews of RCTs to be downgraded based on methodological flaws. Similarly, JBI emphasizes that individual studies must be critically appraised for validity before being included in evidence syntheses. Quality assessment tools like the Cochrane Risk of Bias tool or ROBINS-I for non-randomized studies are essential for distinguishing strong findings from flawed ones, regardless of their position in the hierarchy.
Easier said than done, but still worth knowing.
Mistake 3: Dismissing Expert Opinion Too Quickly
While expert opinion (Level V) is the lowest tier, it plays a critical role in areas with sparse evidence or when translating research into practice. Here's a good example: clinical guidelines often rely on expert consensus to bridge gaps where high-quality studies are lacking. Still, this should not be mistaken for evidence. Day to day, narrative reviews and opinion pieces can offer valuable context or hypotheses but must be weighed against empirical data. In GRADE, expert input is used to inform recommendations but is never classified as evidence itself. The key is recognizing when expert judgment complements versus replaces rigorous research But it adds up..
Mistake 4: Ignoring Context and Applicability
Evidence hierarchies prioritize internal validity (how well a study was conducted) but often overlook external validity (how applicable the results are to real-world scenarios). A Level I systematic review of RCTs might focus on a narrow population or controlled setting, making its findings less relevant to diverse clinical or policy contexts. To give you an idea, an RCT on a new drug tested in a homogeneous group of young adults may not translate to elderly patients with comorbidities. Critical appraisal should always include assessing whether the evidence aligns with the specific question, population, and setting at hand. Tools like the PICO framework (Population, Intervention, Comparison, Outcome) help ensure this alignment.
Mistake 5: Treating Hierarchies as Static
Evidence quality is dynamic, not fixed. A systematic review initially rated as high certainty may later be downgraded if new studies reveal inconsistencies or biases. To give you an idea, a meta-analysis of RCTs might initially suggest strong evidence for a treatment, but subsequent trials could highlight adverse effects or lack of efficacy in subgroups. But similarly, emerging methodologies like real-world evidence or machine learning-based analyses challenge traditional hierarchies by offering new ways to synthesize data. Staying current with evolving evidence and reassessing prior conclusions is crucial, especially in rapidly advancing fields like medicine or technology.
Conclusion
Understanding evidence hierarchies requires more than memorizing labels—it demands critical thinking about methodology, context, and applicability. While frameworks like GRADE and JBI provide structured ways to evaluate certainty and rigor, they are tools to guide, not replace, thoughtful analysis. Because of that, narrative reviews and expert opinions, though lower in the hierarchy, still have roles in hypothesis generation and practical decision-making. In real terms, conversely, high-level evidence must be scrutinized for quality and relevance before being applied uncritically. By avoiding common pitfalls and embracing a nuanced approach, researchers, clinicians, and policymakers can better handle the complexities of evidence-based practice and ensure decisions are grounded in both methodological soundness and real-world utility That alone is useful..
