Reporting Risk Ratios

Reporting Risk, Rates, and Ratios in Academic Manuscripts: Principles and Best Practices

In academic-based manuscripts reputable and accurate reporting of statistical measures such as risks, rates and ratios is important because it aids biomedical and public health research provides much information on the probability of outcomes and strength of associations for researchers, clinicians, and most importantly, policymakers to make evidence-based decisions.

Principles and Best Practices

Recent Post

Introduction

In academic-based manuscripts reputable and accurate reporting of statistical measures such as risks, rates and ratios is important because it aids biomedical and public health research provides much information on the probability of outcomes and strength of associations for researchers, clinicians, and most importantly, policymakers to make evidence-based decisions. Nevertheless, variation within and ambiguity in reporting can lead to misinterpretation of results, poor decision making and lack of scientific robustness. This paper discusses some important principles for reporting these measures correctly and highlights what is current best practice while drawing inspiration from peer-reviewed literatures.

Clearly Define the Type of Measure

When reporting statistics, the first thing you should do is explicitly say how one measure differs from another. In general, we distinguish between three broad categories of measures:

  • Risk: This includes absolute risk (AR), relative risk (RR), and risk differences.
  • Rate: This may include incidence rates, prevalence rates, mortality rates.
  • Ratio: This includes odds ratios (ORs), hazard ratios (HRs), and rate ratios.

Each type of measure has a different meaning and use. For example, relative risk is most often used in cohort studies to describe the likelihood of an event between a group that was exposed and a group that was not (e.g., risks of developing disease), whereas odds ratios are used in case-control studies (Park & Han, 2022; Kerr et al., 2023). Therefore, if you use the wrong measure or even fail to define your measure clearly, you may seriously miscommunicate information.

  • Transparent reporting of the numerator (the count) and denominator (the at-risk population) is vital for reproducibility and use of the information. According to Chao et al. (2023), one of the most frequent sources of confusion in statistical reporting is poorly designed or inconsistent denominators, especially when rates or risks are being quantified. For example, an incidence rate should be defined as the number of new cases (numerator) divided by total person-time at risk (the denominator).
  • In case-control studies, while it may not always be possible to calculate absolute risks, knowing how to construct odds ratios from case-control studies can be constructed and distinguished; however, providing careful description of the populations being considered is still required (Kerr et al., 2023). Studies should define what each of the groups is intended to represent, to not over- or under-estimate associations.
    • Time is a necessary component in any measure of rate or risk. If the timeframe is not reported, the reader cannot assess whether the described risk relates to an acute or chronic incident. The risk of cardiovascular mortality over 10 years, could be very dissimilar to that of a 6-month risk.
    • Murad et al. (2025) demonstrate how the ambiguous recognition of follow-up periods will inevitably lead to inappropriate comparisons and unintentionally wrong clinical implications. Therefore, any measures of risk, rate, and ratio need to denote the period over which the data was collected or the outcome was observed.
  • In supporting comparisons across studies and populations, rates should be allocated to standard population units (e.g., per 100, per 1,000, or per 100,000 population). This is especially important in public health when comparing incidence and prevalence over time or across regions.
  • Lininger et al. (2024) suggest that it causes significant challenge to communicate rates without a population multiplier and obscures the size of a problem, whilst also being able to sensationalize or trivialize something. Standardizing applies in turn makes it clearer and enables greater comparison across research.

    • Include a Measure of Precision

  • Precision tells us how confident we can be about the point estimate of a statistical measure. Confidence intervals (CIs), especially 95 % CIs, are the gold standard for reporting the uncertainty surrounding a risk, rate, or ratio.
  • Murad et al. (2025) have recently provided evidence that CIs for odds ratios (ORs) and risk ratios (RRs) could also predict the power of a study (i.e., whether the study had sufficient power).
  • A narrow CI is a more reliable indicator of the point estimates; and a wide CI, is lacking precision, potentially due to low sample size or high variability. CIs can also provide evidence in a meta-analysis whether we are satisfied with the optimal information size (OIS), which can make a meaningful difference in how to evaluate the quality of evidence.
  • Richardson et al. (2025) also mention the practical use of CIs, in both statistical terms, as well as clinical and policy, to help elucidate uncertainty and inform decisions. While p-values provide little to no information about the size of an effect or its direction, CIs truly add rigor to the interpretation of results, as CIs provide more nuance and information.

    • Relative vs. Absolute Measures

  • Another common problem is the tendency to report the relative risks (e.g., relative risk, odds ratio) without reporting the absolute risk or risk difference. When anyone relies solely on relative measures, the clinical significance of the findings may be overstated, especially when the baseline risk is quite low.
  • Newland (2024) points out that risk communication is incomplete without both absolute and relative risks. For instance, a relative risk reduction of 50% sounds substantial, but what if the absolute risk was lowered from 2% to 1%; the impact in the “real-world” would obviously be small.
  • Davies et al. (2022), in their study on violence risk assessment tools, essentially made the same point by saying that risk ratios only add to the understanding of `prediction tools` when used in conjunction with the baseline probabilities.

    • Choosing the Right Measure: Risk Ratio vs. Odds Ratio

  • There remains ongoing debate about the appropriate use of odds ratios (ORs) versus risk ratios (RRs). Although ORs are typically used in logistic regression and in case-control studies, they are prone to overestimating the association when the outcome is common. A couple of methodological studies (Colnet et al., 2023; Ning et al., 2022) proposed approaches for estimating RRs from studies that have typically reported ORs, including case doubling and modified Poisson regression methods.
  • Thus, at a minimum, authors should clearly state their methodological choices to produce ORs or RRs and provide both if feasible, or at least provide a justification if only one is provided.

    • Summary of Best Practices

    Based upon the current literature and reporting guidelines, we have identified best practices:
    Reporting Table
    What to Report Why It's Important
    Statistical measure type Clarifies analytic approach and comparability
    Denominator and numerator Promotes transparency and reproducibility
    Time frame Places findings in temporal context
    Population unit Standardizes interpretation across studies
    Confidence intervals Shows statistical precision and reliability

    Conclusion

    Communicating risk, rate, and ratios in a consistent and accurate way is not just a technical problem- it’s related to scientific and accountability integrity in manuscript statistics. The longer the research informs clinical care and public health decisions, the increasingly aware and respectful we will be of clarity, transparency, and rigor. Researchers will strengthen their research impact, validity, and future interpretability by using best practices, and referring to contemporary peer-reviewed guidelines.

    References

    1. Chao, Y. S., Wu, C. J., Po, J. Y., Huang, S. Y., Wu, H. C., Hsu, H. T., Cheng, Y. P., Lai, Y. C., & Chen, W. C. (2023). The upper limits of risk ratios and recommendations for reporting risk ratios, odds ratios, and rate ratios. Cureus, 15(4), e37799. https://doi.org/10.7759/cureus.37799

    2. Murad, M. H., Tomlinson, G. A., Brignardello-Petersen, R., Wang, Z., & Lin, L. (2025). Confidence intervals of the relative risk and odds ratio can predict when the optimal information size in a meta-analysis is not met. Journal of Clinical Epidemiology, 179, 111653. https://doi.org/10.1016/j.jclinepi.2024.111653

    3. Kerr, S., Greenland, S., Jeffrey, K., Millington, T., Bedston, S., Ritchie, L., Simpson, C. R., Fagbamigbe, A. F., Kurdi, A., Robertson, C., Sheikh, A., & Rudan, I. (2023). Understanding and reporting odds ratios as rate-ratio estimates in case-control studies. Journal of Global Health, 13, 04101. https://doi.org/10.7189/jogh.13.04101

    4. Richardson, R., Kanellopoulou, A., & Dwan, K. (2025). Risk ratios, odds ratios and the risk difference. BMJ Evidence-Based Medicine, 30(1), 66–67.

    5. Park, S. H., & Han, K. (2022). How to clearly and accurately report odds ratio and hazard ratio in diagnostic research studies? Korean Journal of Radiology, 23(8), 777. https://doi.org/10.3348/kjr.2022.0237

    6. Newland, M. C. (2024). The proper calculation of risk ratios: How and why. Perspectives on Behavior Science, 47(4), 803–814. https://doi.org/10.1007/s40614-024-00379-z

    7. Davies, S. T., Helmus, L. M., & Quinsey, V. L. (2022). Improving risk communication: Developing risk ratios for the VRAG-R. Journal of Interpersonal Violence, 37(1–2), 835–862. https://doi.org/10.1177/0886260520910342

    8. Colnet, B., Josse, J., Varoquaux, G., & Scornet, E. (2023). Risk ratio, odds ratio, risk difference… Which causal measure is easier to generalize? arXiv preprint, arXiv:2303.16008. https://doi.org/10.48550/arXiv.2303.16008

    9. Ning, Y., Lam, A., & Reilly, M. (2022). Estimating risk ratio from any standard epidemiological design by doubling the cases. BMC Medical Research Methodology, 22(1), 157. https://doi.org/10.1186/s12874-022-01676-7

    10. Lininger, M. R., Root, H. J., Camplain, R., & Barger, S. D. (2024). Describing the appropriate use and interpretation of odds and risk ratios. Research in Sports Medicine, 32(3), 504–510. https://doi.org/10.1080/15438627.2023.2178529