Methodological Standards When Reporting From National Databases

To the Editors:

I applaud the efforts by Naseem et al¹ to explain COVID-19 outcomes in inflammatory bowel disease (IBD) patients. Their attempt to provide a comprehensive dataset combining the National Inpatient Sample (NIS) and Nationwide Readmissions Database (NRD) is ambitious and informative. However, this approach raises significant methodological concerns that could impact the study’s conclusions.

First, while innovative, the combination of the NIS and NRD may lead to overlapping data and, thus, misinterpretation of hospitalization patterns and outcomes. This overlap is particularly problematic, as it risks double-counting patients, leading to inflated or skewed results. Accurate data representation is vital in informing clinical practices and policymaking, especially in a pandemic.

Another key issue is using Student t tests for means, rather than nonparametric analysis, and reporting medians with interquartile range, as the NIS and NRD are essentially nonparametric data.² This methodological choice may not accurately reflect the underlying data distribution, potentially leading to incorrect conclusions about the statistical significance of the findings. Given the critical nature of COVID-19 research, ensuring statistical accuracy is paramount for the reliability of the conclusions drawn.²

Furthermore, the study’s approach to handling cohort imbalances could be improved. Propensity score matching, for instance, would have offered a more robust method to ensure comparability between groups. This is crucial in observational studies to minimize bias and increase the credibility of the findings.² Clarity regarding the presentation of baseline characteristics and tables is also needed. If these represent combined data from both databases, the implications for variables like race, in which one database lacks this information, must be addressed. Such clarity is essential for readers to understand and interpret the study’s findings appropriately.

While the study provides valuable insights, addressing these methodological concerns could significantly enhance its scientific rigor and applicability. Future research would benefit from considering these aspects to ensure accurate and reliable results.

Author Contribution

H.A.: ideas, writing, analyzing, approval, supervision.

Funding

None declared.

Conflicts of Interest

None declared.

Data Availability

There were no new data generated.

References