Dose-Response Meta-Analysis
Dose-Response Meta-Analysis (Non-Linear Dose-Response Synthesis) · Also known as: Dose-Response Relationship, Non-Linear Meta-Analysis, Dose-Effect Synthesis
Dose-response meta-analysis is a specialized evidence synthesis method that models the relationship between exposure dose (or intensity, duration, quantity) and health outcome across multiple studies, assessing whether effects follow a linear trend, nonlinear curve, or threshold pattern. Pioneered by Greenland and Longnecker (1992) and refined by Orsini et al. (2012), dose-response meta-analysis answers critical questions like 'Does cardiovascular disease risk increase consistently with salt intake, or is there a threshold above which risk plateaus?' and 'Does the benefit of physical activity increase linearly with exercise duration, or do diminishing returns occur at high doses?' This method is essential for risk assessment, policy-setting on safe exposure limits, and optimizing treatment dosing.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use dose-response meta-analysis when (1) your research question concerns how an outcome varies with dose or exposure amount (e.g., 'How does cancer risk increase with cumulative smoking exposure?'), (2) multiple studies provide dose-stratified data or multiple dose arms, (3) understanding the dose-response shape is important for policy or clinical decisions (safe exposure limits, optimal drug dosing, public health recommendations), or (4) you want to compare dose-response shapes across studies and identify optimal or threshold doses. Common applications include nutritional epidemiology (salt, alcohol, dietary fiber and health outcomes), occupational exposure (toxin exposure and disease risk), drug dosing (increasing therapeutic benefit with higher doses up to a point, then diminishing returns or toxicity), and lifestyle interventions (how exercise duration relates to health benefits).
Strengths & limitations
- Identifies optimal or threshold doses: dose-response curves can reveal sweet spots (maximum benefit at moderate dose) or thresholds (no benefit below a dose level, benefit plateaus above it).
- Utilizes full dose information across studies, not just dichotomized comparisons, increasing statistical power and precision.
- Detects non-linear relationships that simple linear meta-analysis would miss, providing more accurate risk or benefit assessment.
- Informs policy and clinical decisions: regulatory agencies use dose-response meta-analyses to set safe exposure limits (WHO, EPA); clinicians use them to inform optimal drug dosing.
- Integrates multiple studies with varying dose ranges, synthesizing a broader exposure spectrum than any single study provides.
- Requires dose-stratified data from included studies, which is not always available. Some studies report only a single comparison (high vs low dose) without intermediate categories.
- Assumes a consistent dose metric across studies, which may not hold if studies measure exposure differently (e.g., pack-years of smoking vs cigarettes per day; absolute dose vs dose per kilogram body weight).
- Statistical power depends on number of dose categories across all studies. If most studies provide only two dose categories, power to detect non-linear shapes is limited.
- Heterogeneity in dose-response shape across studies may indicate effect modification or true variability, complicating interpretation. If studies show wildly different curves, combining them into a single synthesis may be misleading.
Frequently asked
What is the difference between linear and non-linear dose-response relationships?
Linear dose-response means effect increases (or decreases) in a straight line with dose: each additional unit of dose produces the same change in effect. Non-linear means the relationship curves: initial dose increases may produce large effect changes; higher doses may produce smaller additional changes (plateauing), or extremely high doses may reverse direction (J or U-shaped, where high doses are harmful). Test which model fits better using likelihood ratio test.
Can I include studies with only two dose categories in dose-response meta-analysis?
Yes, but recognize they contribute less to shape estimation. Studies with multiple dose categories (e.g., 5 dose levels) directly inform whether the dose-response curve is linear or curved. Studies with two categories (e.g., high vs low dose) provide only one point on the curve. Include all studies but weight them accordingly or perform sensitivity analyses excluding studies with sparse dose categories.
What if studies use different dose metrics (e.g., grams per day vs milligrams per kilogram)?
Standardize dose metrics before meta-analysis. Convert all to a common unit (e.g., all to grams per day). If conversion is not possible (e.g., some studies report pack-years of smoking without providing conversion to cigarettes per day), perform sensitivity analyses stratified by dose metric, or exclude studies that cannot be standardized. Document conversions explicitly.
How do I interpret a J-shaped or U-shaped dose-response curve?
J or U-shapes indicate effect is lowest at intermediate doses and increases (or decreases, for U-shapes) at both low and high doses. This pattern is common for alcohol (cardiovascular protective at moderate doses, harmful at very high doses). Interpretation requires caution: (1) Rule out reverse causality (did sick people reduce intake?); (2) Examine whether curve is driven by a few studies or consistent across studies; (3) Assess biological plausibility. Avoid recommending that people consume alcohol for health benefit; the relationship is complex and population-dependent.
Sources
- Greenland, S., & Longnecker, M. P. (1992). Methods for trend estimation of environmental health risks, with application to exposure to contaminated groundwater. Statistics in Medicine, 11(14‐15), 1837–1847. link ↗
- Orsini, N., Li, R., Wolk, A., Khudyakov, P., & Spiegelman, D. (2012). Meta-analysis for linear and nonlinear dose-response relations: examples, an evaluation of approximations, and software. American Journal of Epidemiology, 175(1), 66–73. DOI: 10.1093/aje/kwr265 ↗
- Berlin, J. A., Longnecker, M. P., & Greenland, S. (1993). Meta-analysis of epidemiologic dose-response studies. American Journal of Epidemiology, 140(1), 75–82. link ↗
How to cite this page
ScholarGate. (2026, June 4). Dose-Response Meta-Analysis (Non-Linear Dose-Response Synthesis). ScholarGate. https://scholargate.app/en/evidence-synthesis/dose-response-meta-analysis
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Meta-RegressionMeta Analysis↔ compare