Recently a group of medical researchers affiliated with Case Western Reserve University School of Medicine, the Center for Clinical Informatics Research and Education, MetroHealth System and the National Institute on Drug Abuse, National Institutes of Health in Bethesda Maryland produced an interesting study showcased in a January 20, Research Letter in the medical journal JAMA. Tapping into TriNetX, a federated analytics platform that accesses and makes available real world data from 63 academic medical centers representing 89 million patient records in corresponding electronic health records, the study team conducted population-wide retrospective observational patient-level analyses exempt from Institutional Review Board requirements under the auspices of Case Western Reserve University. The study team probed breakthrough SARS-CoV-2 infections occurring between July and November 2021 during the dangerous Delta surge of the COVID-19 pandemic. They applied propensity score matching (factoring in demographics, social determinants of health, transplants and comorbidities known to impact risk and outcomes) for monthly incidence rates of breakthrough infections to those individuals who received the Moderna (mRNA-1273) and Pfizer-BioNTech (BNT162b2) COVID-19 vaccines. Applying Kaplan-Meier survival and Cox proportional hazard analyses to patients for 14 days post the index event—that is full vaccination (again accessed in underlying electronic health records thanks to the federated ability of TriNetX)—they applied Hazard ratios (HRs) and 95% confidence intervals calculations based on underlying comparisons of time-to-event rates associated with the two vaccinated cohorts. So, using the TriNetX platform statistical engine with a 2-sided P < .05,the authors applied the propensity score-matched cohorts and COVID-19 vaccines, the study authors compared hospitalization and mortality in infected patients from two months post the onset of the infection. What did they find?
The authors were able to generate statistics on the following:
Moderna (mRNA-1273)Pfizer-BioNTech (BNT162b2)# Subjects62,628574,538Age CommentsSignificantly older & more comorbidities but matching harmonizes Case per 1,000 person days1.62.8
However once applying the matching the differences were mitigated. The authors found that with the surge of Delta variant-based infections came a surge in overall breakthrough infections, noting that those incidences were greater in the Pfizer-BioNtech vaccine than Moderna’s subjects. This figure reached 2.8 and 1.6 cases per 1000 person days by November
Note that when the authors applied the matching, they found aht the hazard rate associated with breakthrough infection was significantly higher in the Pfizer-BioNTech COVID-19vaccine –or put another way mRNA-1273 was associated with a far lower hazard for breakthrough infection then BNT162b2 (n=62584) (HR, 0.85; 95% CI 0.80-0.89).
What were some other findings?
CategoryModerna (mRNA-1273)Pfizer-BioNTech (BNT162b2)# Subjects62,628574,53860-day hospitalization risk12.7% (392/3078)13.3% (2489/18737)60-day mortality1.14% (35/3078)1.10 (207/18737)
What about the risk of hospitalization?
The authors report that when comparing the matched cohorts mRNA-1273 recipients (n -3054) experienced a lower risk of 60 day hospitalization when compared to the Pfizer vaccine (n=3054)(HR, 0.80; 95% CI, 0.70-0.91). Meanwhile they found no statistical difference for mortality (HR, 0.79; 95% CI, 0.50-1.23).
Limitations
As with all observational studies, limitations can influence the final results. For this particular study the authors include the following to consider:
Based on underlying patient records, this was an observational, retrospective study which introduces known limitations
The authors are clear that the “generalizability of the results from the TriNetX platform is unknown
Confounding factors could be present
Less than perfect matching resulted from propensity score matching
Need for Transparency
Some critiques are raised with results such as these. While a snapshot of the data is made available in associated tables, the authors don’t explain how the results granularly trace to the data sources. The fact that there is no “generalizability” to the TriNetX platform output represents an example. Noteworthy the JAMA editors precluded any comments—why would this be the case?
Funding
National Institute on Alcohol Abuse and Alcoholism
The National Institute on Aging
The National Institute on Drug Abuse
The National Cancer Institute
Clinical and Translational Science Collaborative of Cleveland
What is TriNetX
Covered by TrialSite, TriNetX is a global network of healthcare organizations and life sciences companies driving real-world research to accelerate the development of new therapies. Through its self-service, HIPAA, GDPR, and LGPD-compliant platform of federated EHR, datasets, and consulting partnerships, TriNetX markets its offering as placing the power of real-world data into the hands of its worldwide community to improve protocol design, streamline trial operations, and enrich real-world evidence generation.
Lead Research/Investigator
Lindsey Wang, Case Western Reserve University School of Medicine, 1Center for Artificial Intelligence in Drug Discovery
Pamela B. Davis, PhD, MD, Case Western Reserve University School of Medicine, Center for Community Health Integration
David C. Kaelber, MD, PhD, MPH, MetroHealth System, Center for Clinical Informatics Research and Education
Noral D. Volkow, MD, National Institute on Drug Abuse, National Institutes of Health
Rong Xu, PhD, Case Western Reserve University School of Medicine, 1Center for Artificial Intelligence in Drug Discovery