Confidence Intervals and Odd Ratios

By Tanvi Patel, SRT

Fields requiring research and presentation of measurable data use confidence intervals and odds ratios. These figures allow for data analysis and interpretation in order to make tangible conclusions. As respiratory therapists, it is important to understand the concept of confidence intervals and odds ratios, as the scope of practice and research of respiratory-related topics increases.

Defining Confidence Intervals (CI) and Odds Ratio (OR)

Confidence intervals are numerical ranges that assure the researchers and target audience of the precision of the data (Clarke, 2012). Essentially, it is a range of values within which it is highly probable that the true value lies. For example, if a parliamentary election was being conducted and the prediction was that a particular party would get 62% of the votes, a 95% confidence interval would suggest a range of 59% to 64% (Clarke, 2012). This means that there is a 95% chance that the party in question will receive 62% of the votes. Most confidence intervals are 95%, but some studies use 99% confidence intervals instead (Szumilas, 2010).

An odds ratio (OR) is a figure that gives the probability of an outcome depending on a particular variable of interest (Szumilas, 2010). For example, if you are assessing the likelihood of particular groups contracting a disease, you would look at the odds ratio. If the odds ratio is greater than 1, the risk of contracting this disease is greater in that particular group  (Laing & Rankin, 2011). If the odds ratio is less than 1, the risk of contracting this disease is less in a group that is not exposed (Laing & Rankin, 2011). If the odds ratio value is one, then the data is insignificant and does not suggest an increased or decreased risk (Laing & Rankin, 2011).

Practical Use and Interpreting Confidence Intervals and Odds Ratio

Confidence intervals are used when researchers need to make inferences based on sample sizes and populations (Laing & Rankin, 2011). They are used in sample size calculations and prevalence surveys (Hazra, 2017). With a larger sample size, there is less variability, a narrower interval, indicating more precision in the estimate, and the opposite is true with contrasting parameters (Laing & Rankin, 2011).

Odds ratios are used in case-control studies, cross-sectional and cohort studies (Szumilas, 2010). These ratios are a mode of comparison between cause and effect.

There is a relationship between the two statistical figures. An odds ratio is computed based on the sample studied (Laing & Rankin, 2011). The confidence interval provides the range within which the odds ratio will fall (Tenny & Hoffman, 2020). Hence, the confidence intervals and odds ratios are correlated.

As mentioned above, the null value in an OR is 1.00, and the null value for a difference in means is zero. The P-value and the CI are intimately linked.

When interpreting CIs, it is important to consider both the statistical and clinical significance (Simon, 2001). If the null value of 1.00 for an OR, or zero for a difference in means, is not included in the 95% interval, then the P value, by definition, is statistically significant (ie. less than 0.05). An interval containing the null value means that the result is not statistically significant. Clinical significance, however, is contextual and is determined by each practitioner in each clinical setting. For example, with a 95% CI for a difference in the mean neutrophil count of 1.2, 46.8 does not include the null value of zero; however, the interval is too wide to make a meaningful clinical inference. Even in the case of a narrow CI such as 72.2 to 76.3 for a mean difference in diastolic blood pressure, the interval does not contain the null value; however, the difference may have no clinical significance. In both cases, by definition, the results are statistically significant, but how meaningful the clinical significance is requires interpretation by the health professional in that setting.

Statistical and Clinical Significance of Confidence Intervals and Odds Ratio

When interpreting confidence intervals and odds ratios, both the clinical and statistical significance is important (Laing & Rankin, 2011). Suppose the null value of 1, for an odds ratio, is not included in the confidence interval range. In that case, the value is considered to be statistically significant (where P is less than 0.05) (Laing & Rankin, 2011). In the same manner, if 0 is not included in the difference of means, then the values are statistically significant (Laing & Rankin, 2011). If a confidence interval contains the null value of the odds ratio (ie. 1), then the value is not statistically significant.

For example, in a case-control study addressed by Laing and Rankin (2011), researchers were trying to study if executive dysfunction (EF) was a predictor of smoking in childhood cancer survivors. They found that survivors of childhood cancer were two times more likely (odds ratio) to get executive dysfunction than their healthy siblings. As seen in Table 1 below, researchers here were 95% confident that the odds ratio would be between 1.04 and 3.83. So, since the confidence interval does not include the odds ratio value of 1, the values are statistically significant.

Table 1: Predictors of Smoking Within the Childhood Cancer Survivor Study (from Laing and Rankin Study)

For clinical significance, it is more subjective as the practitioner(s) in each clinical setting will have to determine this (Laing & Rankin, 2011). Despite the confidence intervals and odds ratios, clinical significance changes with the setting and circumstances and must be interpreted by professionals accordingly.

Respiratory therapists are healthcare professionals that are required to regularly find research and evidence to support their clinical practice. As their scope increases and they continue to expand their skills and clinical care, being able to interpret data and determine the clinical significance becomes essential (Laing & Rankin, 2011). Hence, the basic concepts of confidence intervals and odds ratios have been covered to help respiratory therapists in their clinical practice and help them determine the best interventions and care plans for their patients.

References

Clarke, J. (2012). What is a CI? Evidence Based Nursing, 15(3), 66–66. https://doi.org/10.1136/ebnurs-2012-100802

Hazra, A. (2017). Using the confidence interval confidently. Journal of Thoracic Disease, 9(10), 4124–4129. https://doi.org/10.21037/jtd.2017.09.14

Laing, C. M., & Rankin, J. A. (2011). Odds Ratios and Confidence Intervals: A Review for the Pediatric Oncology Clinician. Journal of Pediatric Oncology Nursing, 28(6), 363–367. https://doi.org/10.1177/1043454211426575

Szumilas, M. (2010). Explaining odds ratios. Journal of the Canadian Academy of Child and Adolescent Psychiatry = Journal De l’Academie Canadienne De Psychiatrie De L’enfant Et De L’adolescent, 19(3), 227–229.

Tenny, S., & Hoffman, M. R. (2020). Odds Ratio (OR). In StatPearls. StatPearls Publishing. http://www.ncbi.nlm.nih.gov/books/NBK431098/

[printfriendly]