Why English? | Reading MOOSE Meta-Analyses of Observational Studies Through Search, Bias, Heterogeneity and Conclusions

People searching for a MOOSE checklist, meta-analysis of observational studies or how to interpret heterogeneity often encounter a forest plot before they understand the evidence assembled beneath it. English matters because a reader must keep study design, exposure, outcome, confounding, search coverage, model choice and uncertainty attached to every pooled estimate.

The official EQUATOR Network MOOSE record, checked on 5 November 2026, identifies MOOSE as reporting guidance for meta-analyses of observational studies in epidemiology. The original JAMA consensus statement organises its proposed checklist across background, search strategy, methods, results, discussion and conclusions.

MOOSE improves reporting; it does not remove confounding from the primary studies or make a pooled association causal. The existing STROBE owner covers individual observational reports. This article owns the additional decisions involved when observational studies are selected, coded and combined.

---

Define the Epidemiological Problem

A MOOSE report begins with the problem the meta-analysis is intended to clarify.

Meta-analysis decision

Find the exposure, outcome, population and reason existing observational results require synthesis. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

A review asks whether long-term night-shift exposure is associated with a defined cardiovascular outcome in working adults.

Bias and interpretation check

A topical association can be framed too broadly. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Rewrite the problem as a testable relationship. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

State the Hypothesis

A prespecified hypothesis sets a boundary around analysis and interpretation.

Meta-analysis decision

Look for the expected direction, eligible comparison and any planned effect modifiers. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

The authors predict stronger association with longer exposure duration and label later occupation-specific analyses exploratory.

Bias and interpretation check

A hypothesis does not prevent selective analysis unless deviations are reported. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Compare it with the final models. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Define Every Outcome

Outcome labels must refer to comparable clinical or epidemiological events.

Meta-analysis decision

Find case definitions, ascertainment methods and time windows. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

The review separates incident disease from mortality rather than pooling both under “cardiovascular events”.

Bias and interpretation check

Similar names can hide different constructs. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Build an outcome-definition table before trusting a combined estimate. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Define the Exposure or Intervention

Observational meta-analysis depends on how exposure was measured and categorised.

Meta-analysis decision

Look for units, thresholds, duration, timing and reference group. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

Weekly hours are converted carefully, while studies using only job title are analysed separately.

Bias and interpretation check

Conversion can create false comparability. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Keep original definitions alongside harmonised categories. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Specify Eligible Study Designs

Cohort, case-control and cross-sectional studies answer different temporal questions.

Meta-analysis decision

Find the eligible designs and whether they are combined or stratified. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

Prospective cohorts are pooled separately from retrospective case-control studies.

Bias and interpretation check

A statistical model cannot erase design differences. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Link each design to its likely biases. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Define the Study Population

Population criteria determine the domain of the conclusion.

Meta-analysis decision

Look for age, occupation, geography, risk status and exclusion conditions. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

The review excludes populations with a pre-existing outcome when estimating incident risk.

Bias and interpretation check

A broad population label can hide major baseline differences. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Compare the pooled population with the reader’s intended use. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Report Searcher Qualifications

MOOSE asks who designed and performed the search.

Meta-analysis decision

Find involvement of information specialists or investigators and their relevant expertise. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

A librarian develops database syntax while epidemiologists define exposure and outcome concepts.

Bias and interpretation check

Credentials do not guarantee comprehensive retrieval. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Inspect the actual strategy as well as the team. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Give Search Dates and Terms

A reproducible search needs a coverage period and query language.

Meta-analysis decision

Look for dates, keywords, controlled vocabulary, fields and updates. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

The supplement provides database-specific syntax and a final search date rather than a short topic phrase.

Bias and interpretation check

An old but well-reported search can still be out of date. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Date the evidence boundary. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Search Databases and Registries

Different sources index different disciplines and publication types.

Meta-analysis decision

Find every database, registry and platform used. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

The review combines biomedical and occupational-health databases rather than assuming one source is sufficient.

Bias and interpretation check

Many database names can mask overlapping coverage. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Assess complementarity. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Pursue Unpublished and Grey Evidence

Unpublished studies may affect estimates and publication-bias assessment.

Meta-analysis decision

Look for conference material, theses, reports, registries and author contact. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

The team obtains an unpublished adjusted estimate from a cohort report and explains its verification.

Bias and interpretation check

Grey evidence may have limited detail. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Report retrieval and appraisal with the same care as journal studies. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Address Language Restrictions

Language restrictions can shape which settings and results enter the review.

Meta-analysis decision

Find eligible languages, translation methods and reasons for exclusions. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

A multilingual team screens non-English abstracts before deciding whether full translation is required.

Bias and interpretation check

English-only selection may create bias and reduce global relevance. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Quantify excluded records and discuss direction of risk. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

List Excluded Studies With Reasons

A transparent exclusion list helps readers inspect boundary decisions.

Meta-analysis decision

Look for full-text exclusions and specific justifications. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

A study is excluded because its exposure precedes the outcome ambiguously, not because its result is inconvenient.

Bias and interpretation check

A reason can still be inconsistently applied. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Compare similar borderline cases. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Contact Study Authors

Author contact can clarify overlapping cohorts, missing adjustments and unpublished results.

Meta-analysis decision

Find what was requested, how often and the response. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

Investigators confirm that two papers analyse the same cohort years, preventing double-counting.

Bias and interpretation check

Non-response leaves uncertainty and may be selective. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Record both successful and unsuccessful contact. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Judge Relevance to the Hypothesis

Included studies should be appropriate for the question, not merely mention the exposure and outcome.

Meta-analysis decision

Look for temporal order, population fit and comparable constructs. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

A cross-sectional study is retained for prevalence context but not pooled with incident-risk cohorts.

Bias and interpretation check

Broad inclusion can increase data while weakening meaning. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Explain which studies answer which part of the question. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Explain Data Selection and Coding

Reviewers choose which estimates, categories and covariates to extract.

Meta-analysis decision

Find a decision hierarchy and clinical or epidemiological rationale. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

The most adjusted model is not chosen automatically; the team prefers models that avoid adjusting for a mediator.

Bias and interpretation check

Coding convenience can alter the pooled effect. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. State the causal reasoning behind model selection. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Document Independent Classification

Multiple reviewers, blinding and agreement checks can make coding disagreements visible.

Meta-analysis decision

Look for duplicate extraction, calibration and conflict resolution. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

Reviewers independently classify study design and confounder adjustment before reconciling differences.

Bias and interpretation check

Agreement does not prove the classification scheme is correct. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Preserve reasons for difficult decisions. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Assess Confounding

Confounding is central in observational evidence.

Meta-analysis decision

Find the major confounders, how primary studies handled them and whether adjusted estimates are comparable. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

Studies adjusting for age and smoking are separated from minimally adjusted estimates.

Bias and interpretation check

Different adjustment sets may estimate different quantities. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Avoid treating “adjusted” as one uniform category. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Assess Study Quality and Risk

The review should examine design features that could bias individual estimates.

Meta-analysis decision

Look for domain judgements, blinding where relevant and planned use of those judgements. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

Sensitivity analysis excludes studies with self-reported outcomes and compares the pooled result.

Bias and interpretation check

A total quality score can conceal a decisive bias. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Keep domain-level reasons visible. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Investigate Heterogeneity

Variation among estimates can reveal different populations, definitions or biases.

Meta-analysis decision

Find statistical measures, visual inspection and substantive explanations. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

High heterogeneity leads the team to examine exposure measurement and follow-up rather than merely switching models.

Bias and interpretation check

A random-effects model does not explain inconsistency. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Treat heterogeneity as a question. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Describe the Statistical Model

Readers need a replicable account of effect measures and pooling.

Meta-analysis decision

Find transformations, fixed or random effects, weighting, correlation handling and software. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

Odds ratios are not casually treated as risk ratios when the outcome is common.

Bias and interpretation check

Sophisticated software cannot repair incompatible inputs. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Reconstruct the estimand and assumptions. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Show Study Estimates and the Pooled Result

A graphic should display individual estimates, uncertainty and the combined result.

Meta-analysis decision

Check labels, direction, scale and whether study weights are interpretable. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

The forest plot uses a logarithmic scale and clearly marks which side indicates higher risk.

Bias and interpretation check

A diamond can attract attention away from inconsistent studies. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Read the spread before the summary. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Provide a Study Characteristics Table

A table lets readers compare design, population, exposure, outcome and adjustment.

Meta-analysis decision

Look for the variables needed to explain heterogeneity. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

The table reveals that the largest studies use administrative codes while smaller studies use adjudicated outcomes.

Bias and interpretation check

A table can be comprehensive but still omit decision-relevant detail. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Use it to generate alternative explanations. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Run Sensitivity Analyses

Sensitivity analysis tests whether conclusions depend on reasonable choices.

Meta-analysis decision

Find exclusions, model alternatives and prespecified subgroup analyses. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

The association weakens when studies with unclear temporal order are removed.

Bias and interpretation check

Running many analyses can create selective emphasis. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Report the full planned set and interpret patterns cautiously. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Report Statistical Uncertainty

Every pooled estimate requires an interval and a statement of what it represents.

Meta-analysis decision

Look for confidence or credible intervals and prediction intervals where appropriate. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

A narrow mean-effect interval coexists with a wide prediction interval, signalling uncertain performance in a new setting.

Bias and interpretation check

Precision is not validity. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Keep bias and heterogeneity beside the interval. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Examine Publication Bias

Missing studies can distort the visible literature.

Meta-analysis decision

Find funnel plots or models with their assumptions and limitations. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

The review notes that asymmetry may reflect small-study differences as well as publication bias.

Bias and interpretation check

A statistical test has low power with few studies. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Use several pieces of evidence rather than one p-value. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Justify Important Exclusions

The discussion should revisit exclusions that could change conclusions.

Meta-analysis decision

Look for language, design, exposure and publication-status restrictions. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

The authors explain how excluding cross-sectional studies narrows the claim to temporal associations.

Bias and interpretation check

A reasonable restriction still limits generalisation. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. State the trade-off. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Consider Alternative Explanations

Observational associations may reflect confounding, reverse causation, selection or measurement error.

Meta-analysis decision

Find plausible alternatives and evidence that strengthens or weakens them. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

A dose-response pattern is discussed alongside the possibility that healthier workers select out of high exposure.

Bias and interpretation check

Plausibility is not proof. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Rank explanations by evidence rather than preference. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Generalise Within the Evidence Domain

Conclusions should match the populations, exposures and designs synthesised.

Meta-analysis decision

Look for explicit boundaries rather than a universal statement. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

A review of adult workers in high-income settings does not claim the same association for adolescents or informal work.

Bias and interpretation check

A large combined sample does not widen the domain automatically. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Name the represented settings. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Set a Future Research Agenda

A good synthesis identifies which uncertainty matters next.

Meta-analysis decision

Find needs for better measurement, confounder control, prospective design or underrepresented populations. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

The review calls for harmonised exposure measures rather than simply requesting more studies.

Bias and interpretation check

A generic call adds little. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Tie each recommendation to a documented evidence gap. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Disclose Funding and Influence

Funding and sponsor roles can shape review choices.

Meta-analysis decision

Find support, material assistance and involvement in protocol, analysis and publication. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

Industry data access is disclosed while coding and interpretation remain under investigator control.

Bias and interpretation check

Disclosure enables scrutiny but does not settle validity. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Connect the relationship to decisions it could affect. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Check the Protocol and Registration Record

A protocol can show which questions and methods existed before the results were known.

Meta-analysis decision

Look for registration, dated amendments and explanations of deviations. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

The published review adds a subgroup after protocol registration and labels it exploratory with a reason.

Bias and interpretation check

Registration does not guarantee that the plan was complete or followed. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Compare timestamps, planned outcomes and final analyses. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Detect Overlapping Study Populations

Several papers may analyse the same cohort, creating a risk of counting participants more than once.

Meta-analysis decision

Find cohort names, recruitment periods, sites and author clarification. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

Two publications use the same national registry years, so the review selects one compatible estimate for the main pool.

Bias and interpretation check

Different paper titles can disguise shared data. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Build a cohort-overlap table. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Respect the Unit of Analysis

Clustered, repeated or multilevel observations require methods that reflect dependence.

Meta-analysis decision

Look for the level at which participants, schools, clinics or regions contribute estimates. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

A study reporting several outcomes from one cohort does not become several independent studies.

Bias and interpretation check

Ignoring dependence can make intervals too narrow. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Trace each estimate back to its independent evidence unit. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Did You Know? A Prediction Interval Answers a New-Setting Question

A confidence interval around a mean pooled effect and a prediction interval for a future setting answer different questions.

Meta-analysis decision

Find whether between-study variation is incorporated and whether enough studies support the calculation. In a MOOSE report, every pooled estimate depends on choices about population, exposure, outcomes, eligible designs, search coverage, coding, confounding and statistical model. Precise English keeps those choices attached to the number.

The mean association is positive, yet the prediction interval crosses no association, warning that a new setting may differ.

Bias and interpretation check

Prediction intervals inherit model assumptions and unstable heterogeneity estimates. A pooled estimate can be numerically precise while still reflecting biased primary studies or inconsistent definitions. Use them as a transfer prompt, not a guarantee. State the decision, its rationale, the evidence affected and the alternative explanation it leaves open.

---

Frequently Asked Questions

What is MOOSE?

MOOSE is reporting guidance for meta-analyses of observational studies in epidemiology.

Is MOOSE a statistical model?

No. It is a reporting framework covering the question, search, methods, results and conclusions.

Why are observational studies difficult to pool?

They may differ in populations, definitions, designs, confounder adjustment and bias.

Does a random-effects model solve heterogeneity?

No. It allows variation in a model but does not explain why studies differ.

What is confounding?

It is distortion of an exposure–outcome association by other factors related to both.

Why can adjusted estimates differ?

Primary studies may adjust for different variables and may estimate different causal quantities.

What does a forest plot show?

Individual study estimates and uncertainty, often with a pooled estimate; it does not by itself establish validity.

What is publication bias?

It is distortion when the visible literature differs from all studies conducted, often because results influence publication.

Can a precise pooled estimate still be wrong?

Yes. Precision does not remove systematic bias or incompatible definitions.

How should a student summarise a MOOSE report?

State the question, designs, definitions, search coverage, confounding approach, heterogeneity, model, pooled estimate and key alternative explanations.

---

Useful Next Reading

• Read the official EQUATOR Network MOOSE record.
• Read the original MOOSE consensus statement and checklist.
• Read individual observational reports with STROBE.
• Check outcome definitions and forest plots in RevMan Web.

Closing thought: MOOSE becomes practical when the pooled estimate is the end of a visible chain of definitions, searches, coding decisions, bias checks and explanations for variation—not the beginning of the story.