I'm still in my minimalist mode for postings as I struggle with a website redesign, but last week provided me with a relatively unique opportunity to comment on a big success of recent years, prevention of RSV infection and complications in infants. A trio of articles appeared that are great examples of different approaches to post-marketing surveillance of vaccine and monoclonal antibody products. All the articles are open access, no journal subscription required.
Systematic Review and Meta-analysis
Canadian investigators looked at postlicensure observational studies of nirsevimab effectiveness in preventing hospitalizations and emergency department visits for infants with lower respiratory tract infection. Like any systematic review worth its salt, the search methodology and all aspects of analysis are clearly reported and summarized. They ended up with 15 studies that could be analyzed in the systematic review, 11 of which could be analyzed with meta-analysis. (The discrepancy is that 4 studies did not have a concurrent control group and thus were inappropriate for meta-analysis.)
As seen in virtually every other study of nirsevimab, effectiveness was very high. Here's a sample of the results; note that effectiveness was higher against RSV-related disease compared to all-cause disease that included other infections in addition to RSV.
As the authors mention, looking at studies from real-world settings can introduce all sorts of confounding factors, including underlying medical conditions, household crowding, prior exposure to RSV or other respiratory pathogens, and access to care. Also, in this particular group of studies, high income countries were over-represented, and the studies were very heterogeneous likely due to differences in test ordering strategies and other clinical practices. So, combining these studies into a single number estimate of effectiveness comes with caveats.
I've often said that meta-analysis is the most dangerous study design in medicine, primarily because the statistical methods are complex and often draw criticism from statisticians promoting one approach versus another. Also, systematic reviews and meta-analyses probably are better suited to defining areas for future research rather than settling a clinical management question once and for all.
Test-negative Case Control Study
This study design is very commonly used to assess vaccine effectiveness - we see it every year to look at influenza vaccine effectiveness. This report combined data from 7 pediatric academic medical centers to compare effectiveness of nirsevimab administration to infants to maternal RSV immunization during pregnancy. The nice addition to this type of study is that a "number needed to immunize" (NNI) can be calculated. NNI can be defined as the number of infants needed to be immunized to prevent 1 additional event (e.g. hospitalization due to RSV) from occurring. Broadly speaking, nirsevimab was more effective that maternal immunization, and younger infants benefitted more than older.
A nice feature was that the authors included both observed effectiveness (comparing the 2024-2025 RSV season to prepandemic seasons) as well as a predicted ("counterfactual") analysis using predictive numbers for just the 2024-2025 season. The table shows aggregate results for RSV-associated hospitalizations.
You may have noticed a confounding factor in this study: parents seeking care for their infants at academic pediatric medical centers may not be reflective of the population as a whole. Maybe we would have seen different results from a group of urgent care centers or in a rural population.
Cohort Study
The third article is a cohort study from France also comparing nirsevimab to maternal vaccination. Again, nirsevimab was superior for most outcomes.
The authors didn't supply us with an NNI, but the 12-point absolute risk reduction in the primary outcome in the top row above translates to an NNI of about 8. This means that for using nirsevimab instead of maternal vaccination to prevent hospitalization for RSV-associated respiratory tract infection, 8 infants would need to receive the monoclonal antibody product to prevent 1 additional event of hospitalization for RSV. That's a very low number in the world of NNI.
Note that in this study and the test-negative trial mentioned above, another confounding factor is that families choosing maternal vaccination may differ significantly from those who chose infant monoclonal antibody administration. For example, it is likely that the nirsevimab group included mothers who received less, or no, prenatal care.
Take-away points from today:
- Different study designs produce different results, driven by appropriate use of statistical methods.
- Postlicensure assessment of medications and vaccines often use real-world data This is much more likely to show how the product will perform once it's commonly used, but it also creates many confounding factors not seen in randomized, prospective trials where strict research rules and monitoring lessen individual physician and patient variations. These confounding variables need to recognized, but they are difficult to measure objectively.
- Both RSV maternal vaccination and RSV monoclonal antibody administration to infants are highly effective in preventing bad outcomes from RSV infection in infants.
- Nirsevimab likely is more effective than maternal immunization for improving infant outcomes. It's too early for real-world data for clesrovimab, the other monoclonal antibody product approved for use in the US.
- Very effective systems are in place to track postlicensure vaccine safety and effectiveness.
A Poem for Your Thoughts
I never thought I'd see a poem entitled, "National Vaccine Injury Compensation Program," but hey, we're nearing the end of an unprecedented year on many fronts. Take a few moments to read it, and maybe a few more moments to look up the unknown (to me) allusions. A quick literature search yielded nothing about "woke mind virus."