Confidence interval reporting in randomised controlled trials: A critical analysis of IJA publications

Statistical methods are integral to the design, analysis, and interpretation of clinical studies. They enable researchers to organise complex data, draw valid inferences, and quantify the reliability and clinical significance of their findings. The use of descriptive statistics, such as mean [standard deviation (SD)] and median [interquartile range (IQR)], helps summarise data, while inferential statistics, including P values and confidence intervals (CIs), allow for hypothesis testing and estimation of effect size and precision.[1] Of these, the reporting of CIs has gathered increasing attention in recent years.

CIs give important information regarding the accuracy and clinical significance of study results by defining a range that is likely to contain the true impact. CIs indicate the consistency or variability of the results. Suppose a study reports a 95% CI; in that case, it means that if the study were repeated multiple times with samples from the population, the calculated CIs from those studies would contain the true value 95% of the time.[2] It is now required to report CIs alongside P values to enhance the transparency and reproducibility of research findings.[3,4] In the context of anaesthesiology, pain management, and critical care, the Indian Journal of Anaesthesia (IJA) has emerged as a leading platform for disseminating original research and shaping clinical standards in India and beyond. The IJA has increasingly prioritised methodological rigour, particularly in the adoption of CIs and advanced statistical methods.

The present analysis included 83 randomised controlled trials (RCTs) published in the IJA between January and December 2024, with a focus on CI usage. Each article was systematically reviewed for statistical methods used, focusing on the presence and presentation of CIs for primary or secondary outcomes. As the study design represented a cross-sectional evaluation of a single publication year, temporal trends or year-wise comparison of CI usage were not assessed. The analysis revealed that 70 (84.3%) articles included 95% CIs for all primary or secondary outcomes [Table 1, Figure 1]. Descriptive statistics, particularly mean (SD), remained the most frequently used summary measure (68 of 83 articles, 82%), followed by percentages (62 articles, 75%) and median (IQR) (16 articles, 19%). P values continued to be widely reported (65 articles, 78%), often without complementary effect-size metrics or CIs.[5,6,7,8,9,10,11,12,13,14,15,16,17] Advanced methods, such as logistic regression and ROC analysis, were used in 12% and 7% of articles, respectively. While such analyses can enhance understanding of binary outcomes and diagnostic accuracy, their underutilisation may reflect constraints in researcher training, resource availability, or editorial policies. Our study did not explore these potential causes due to its retrospective bibliometric design and narrow focus on CI reporting.

Table 1.

Analysis of statistical methods employed in original research articles published in the Indian Journal of Anaesthesia in 2024

Statistical Method	Articles Using 95% CI (n=70)	Articles not Using 95% CI (n=13)	Typical Usage in IJA (context)
Confidence intervals (CIs)	70	0	Reporting effect size and precision for primary/secondary outcomes, especially in RCTs and meta-analyses
Descriptive statistics (mean±SD)	68	12	Central tendency/variability for continuous, normally distributed data; most common summary statistic
Descriptive statistics (median [IQR])	16	3	For skewed/non-normal data, underutilised despite appropriateness
Percentages/proportions	62	10	Summarising categorical variables (e.g., gender, ASA class); nearly universal in demographic tables.
P-values (hypothesis testing)	65	11	Indicating statistical significance in group comparisons, often reported without CIs.
t-test/ANOVA	54	9	Comparing means between two or more groups; the standard for parametric data.
Chi-square/Fisher’s exact test	47	8	Comparing categorical variables between groups, common in baseline and outcome tables.
Mann–Whitney/Wilcoxon tests	14	2	Non-parametric comparison of medians for skewed data; less frequently used than t-tests
Logistic regression	9	1	Multivariable analysis for binary outcomes (e.g., event/no event); rare, but increasing with more complex studies
ROC analysis	6	0	Diagnostic accuracy, cut-off determination, and predictive model evaluation; limited use
Other advanced methods	3	0	Includes linear regression, survival analysis, etc.; very rarely reported

CI=Confidence Interval; SD=Standard Deviation; IQR=Interquartile Range; IJA=Indian Journal of Anaesthesiology; RCT=Randomised Controlled Trial; ASA=American Society of Anesthesiologists; ANOVA=Analysis of Variance; ROC=Receiver Operating Characteristic

Figure 1.

Statistical Methods employed in the Indian Journal of Anaesthesia in 2024. CI = Confidence Interval; SD = Standard Deviation; IQR = Interquartile Range; ANOVA = Analysis of Variance; ROC = Receiver Operating Characteristic

Articles not reporting CIs relied on basic descriptive statistics and hypothesis testing, with minimal use of multivariable analyses or advanced diagnostic tools. In contrast, articles that reported CIs used a wider variety of statistical methods, demonstrating more thorough research analytics. These findings highlight that a significant proportion of IJA articles reported CI (84.3%) in 2024, signalling a growing methodological awareness. This is an encouraging trend, as the reporting of CIs is now recognised as essential for interpreting the clinical relevance and precision of study results, rather than relying solely on statistical significance as indicated by P values.[18]

Reporting of descriptive statistics, including mean (SD) and median (IQR), provides critical context for understanding CI usage. Although this commentary prioritised CI reporting, we included descriptive statistics and frequencies to highlight broader statistical reporting patterns in published RCTs. While SD describes the spread of data and P value test hypotheses, CIs provide the most comprehensive information by combining effect size, precision, and interpretability, making them superior for reporting and understanding study results. This preference for simpler methods is consistent with previous large-scale surveys of Indian anaesthesia journals, which found that descriptive or elementary methods were used in over 90% of studies, while more advanced techniques remained rare.[19]

The underuse of CIs and advanced methods has important implications for the quality and interpretability of published research. CIs provide essential information about the range within which the true effect is likely to lie, helping to distinguish between statistical and clinical significance. As noted in recent reviews of RCT reporting in IJA, merely reporting P values without CIs can obscure the practical importance of findings and limit their applicability in clinical decision-making.[4] CIs not only provide valuable information about effect size along with its width but also suggest possible clinical significance. Thus, in an ideal scenario, effect size should preferably be associated with a CI to assess precision.[20]

CIs are frequently misconstrued. It is essential to recognise that the CI does not represent the range of effects observed in 95% of individuals within the population. Moreover, it would be erroneous to state that there is a 95% likelihood that the CI incorporates the genuine population effect. Instead, the correct interpretation is that if the same study were repeated many times, 95% of the calculated CIs from those studies would be expected to contain the true population parameter. Interpretation of CIs is thus highly context-dependent. Emphasis during postgraduate training and dissertation writing, a basic biostatistics course, and reviewer or editorial fellowship programmes could further improve CI reporting.[21]

Presentation at conferences/CMEs and abstract publication

None.

Study data availability

De-identified data may be requested with reasonable justification from the authors (email to the corresponding author) and shall be shared upon request.

Disclosure of use of artificial intelligence (AI)-assistive or generative tools

The authors confirm that no AI tools or language models (LLMs) were used in the writing or editing of the manuscript, and no images were manipulated using AI.

Declaration of use of permitted tools

The scales, scores are freely available and not copyrighted.

Authors contributions

LG: Concept, Design, Data Analysis, Manuscript Preparation, Editing, Review, Approval. KC: Design, Editing, Review, Approval. DB: Literature Search, Data Acquisition, Editing, Review.

Conflicts of interest

There are no conflicts of interest.

Supplementary material

None.

Funding Statement

Nil.