Comparing Competing COVID-19 Vaccination Strategies Using Google Gemini Deep Research
Using generative artificial intelligence to sample the online knowledge space
On April 22, 2025, Drs. Shira Doron and Monica Gandhi published an opinion piece in MedPage Today on COVID-19 vaccination policy titled “Risk-Based Vaccination Strategies Can Build Back Public Trust”1 On May 9, 2025, the Food and Drug Administration (FDA) announced the appointment of Dr. Vinayak Prasad as the director of the FDA’s Center for Biologics Evaluation and Research (CBER). Then, on May 15, 2025, the New York Times (NYT) published “Federal Officials May Limit Recommendations for COVID-19 Vaccine“2 and the Wall Street Journal (WSJ) published “HHS to Stop Recommending Routine Covid Shots for Children, Pregnant Women.“3
Drs. Doron and Gandhi summarized the two competing COVID-19 vaccine approaches:
“Maintain the current universal vaccine policy, which recommends initial and seasonal COVID-19 vaccination for all individuals 6 months of age and older; or
Narrow vaccine recommendations to target only high-risk individuals for the initial and/or annual vaccines.”
They promote a “risk-based”4 recommendations for initial and annual vaccination that prioritize these higher risk groups:
older age,
severe immunocompromise,
chronic heart and lung disease, and
pregnancy (to protect infants).
They also support a permissive recommendation allowing those who want to receive it.
The WSJ summarized the possible changes to COVID-19 vaccine policy:
“HHS, led by Robert Kennedy Jr., plans to drop CDC recommendations for routine Covid-19 vaccines for pregnant women, teens and children.
FDA plans a new vaccine approval framework, possibly requiring more data from companies, including placebo testing for all new vaccines.
The Covid change could reduce vaccinations, impacting insurance coverage and reversing a key Trump-era health initiative, Operation Warp Speed.”
Currently, the Centers for Disease Control and Prevention (CDC) recommends that everyone six months and older, including pregnant women, receive Covid vaccines (Figure 3 in Appendix).
Dr. Marty Makary, the new FDA commissioner, and Dr. Vinay Prasad, new CBER director, have been highly critical of the FDA and CDC approach to COVID-19 vaccine policy and will likely move away from current FDA and CDC guidance.5 In contrast, Dr. Paul Offit, director of the Vaccine Education Center at Children’s Hospital of Philadelphia and an adviser to the FDA is very concerned that “… we are in the midst of watching the vaccine infrastructure being torn down bit by bit.”6
“Where you stand depends are where you sit”
How do we make sense of these competing strategies?
Vaccine guidance development is making decisions on how to influence and promote vaccine decisions by clinical providers and patients. The final guidance recommendations (and how they are received and adopted or not adopted) depend on our values, goals, prior beliefs, and cognitive biases.7
My personal biases align with Dr. Paul Offit.8 I have also observed (anecdotally) that pediatricians and pediatric infectious disease specialists (eg, Dr. Paul Offit) view vaccine recommendations for children differently than general infectious disease specialists (eg, Drs. Shira Doron and Monica Gandhi).
Instead of posting a blog entry that just promotes my biases, I wanted to mitigate my confirmation bias and sample from the information and knowledge space not readily available to me (Figure 1). Because of confirmation bias we gather information from our “Known Knowns” and the “Known Unknowns” (eg, call a colleague who is an expert). Even better, a diverse, trans-disciplinary advisory committee can help me gather information from the “Unknown Knowns.” The problem is that we do not have the time, resources, and capabilities to gather information effectively and efficiently using traditional methods! Even if we could deploy an advisory committee, we would miss the information and knowledge from the “Unknown Unknowns” because our committee’s knowledge base is finite.
Accessing knowledge spaces outside of our awareness
Artificial intelligence can help us mitigate our confirmation bias by accessing and summarizing the available online information and knowledge spaces outside of our awareness. Figure 2 displays Figure 1 in more detail.
To understand this issue more deeply and to explore my perceptions I enlisted Google Gemini Deep Research to prepare two reports.
Prompt for Report 1:
“The Centers for Disease Control and Prevention (CDC) recommend COVID-19 for all age groups, including children who have lower risks than adults for hospitalization and death from COVID-19. Pediatricians do not compare children to adults when developing vaccine recommendations. Pediatricians ask, given a child's age and
risk for becoming infected with SARS-CoV-2,
risk of hospitalizations if infected,
risk of death if infected, and
risk of a severe vaccine adverse reaction,
will the total benefits vs total risks favor getting vaccinated vs not getting vaccinated? This type of analysis led the CDC to recommend that vaccine for children. In contrast, Drs Vinay Prasad, Monica Gandhi, and Shira Doron do not recommend COVID-19 vaccine for children based on a "risk-based" strategy described in this article: https://www.medpagetoday.com/opinion/second-opinions/115232 . Summarize the scientific rationale for both approaches, including if there is a consensus approach. Include mathematical modeling studies that address this research question.”
Report 1 (public Google Doc): Scientific Rationales for Universal versus Risk-Stratified COVID-19 Vaccination in Children: A Comparative Analysis
Executive Summary
“The COVID-19 pandemic prompted the rapid development and deployment of vaccines, which eventually included pediatric populations. This expansion led to significant debate over the optimal vaccination strategy for children, who generally experience milder disease than adults but are still susceptible to infection, severe outcomes, and long-term complications. Two primary strategies have emerged: a universal approach and a risk-stratified approach.
The universal COVID-19 vaccination strategy, advocated by organizations like the U.S. Centers for Disease Control and Prevention (CDC) and the American Academy of Pediatrics (AAP), recommends vaccination for all children within eligible age groups. This rationale is based on:
Disease Burden: Children can experience SARS-CoV-2 infection, severe illness (including hospitalization and death, albeit less frequently than adults), and specific complications like Multisystem Inflammatory Syndrome in Children (MIS-C) and post-COVID conditions ("long COVID"), even if previously healthy.
Vaccine Efficacy and Safety: Vaccines are considered effective in preventing infection and severe outcomes, with a safety profile where benefits outweigh the rare risks, such as myocarditis.
Public Health Goals: Beyond individual protection, this strategy aims to reduce viral transmission, minimize disruptions to education and societal functions, and protect vulnerable community members.
The risk-stratified COVID-19 vaccination strategy, supported by clinicians and researchers such as Drs. Vinay Prasad, Monica Gandhi, and Shira Doron, and reflected in World Health Organization (WHO) guidance for healthy children, questions the net benefit of universal vaccination for healthy children. Key arguments include:
Lower Absolute Risk: Healthy children have a very low absolute risk of severe COVID-19 outcomes.
Vaccine Adverse Events: For low-risk children, potential rare vaccine adverse events (e.g., myocarditis) might not be clearly outweighed by the benefits.
Natural Immunity: Significant emphasis is placed on the protective role of immunity acquired from prior SARS-CoV-2 infection.
Transparency and Honesty: Proponents call for greater transparency in public health messaging, acknowledging uncertainties and tailoring recommendations more narrowly, suggesting a "campaign of honesty" to maintain public trust.
A global consensus on pediatric COVID-19 vaccination is lacking. While the CDC and AAP recommend universal vaccination, the WHO/SAGE categorizes healthy children as a 'low priority group,' advising context-specific decisions. The European Medicines Agency (EMA) authorizes vaccines based on a favorable benefit-risk profile but leaves programmatic decisions to member states.
Mathematical modeling studies have been used to conduct benefit-risk analyses, evaluate cost-effectiveness, and project impacts of different strategies. Findings vary based on model assumptions (e.g., viral transmissibility, vaccine efficacy, population contact patterns). Some models suggest substantial benefits from pediatric vaccination, including herd immunity effects, while others, particularly for specific low-risk subgroups or booster doses, indicate that potential harms might outweigh direct benefits.
Synthesis and Conclusion: The debate highlights a complex interplay of evolving scientific evidence, ethical considerations (individual vs. community benefit), and differing interpretations of risk. Key areas of ongoing uncertainty include the long-term impacts of long COVID in children, long-term outcomes of vaccine-associated myocarditis, and the duration of protection from vaccines and natural infection against new variants. Continued research, transparency in data and modeling, and clear, nuanced public health communication are crucial. Ultimately, pediatric COVID-19 vaccination policy must navigate these complexities, aiming to protect children's health while considering broader public health and societal values.”
Prompt for Report 2:
“Based on this analysis by Google Gemini Deep Research (<LINK to Report 1>), there are competing approaches to recommending COVID-19 vaccines for children in the United States. Among infectious disease specialists, I perceive that pediatric infectious disease specialists (and pediatricians) tend to side with the current CDC and AAP guidance compared to adult infectious disease specialists such as Drs. Shira Doron and Monica Gandhi. Is my perception correct?”
Report 2 (public Google Doc): Divergent Perspectives on Pediatric COVID-19 Vaccination in the United States: An Analysis of Official Guidance and Expert Commentary (2024-2025)
Executive Summary
“This report examines the landscape of COVID-19 vaccine recommendations for children in the United States for the 2024-2025 period, addressing the perception of competing approaches among infectious disease specialists. The Centers for Disease Control and Prevention (CDC) and the American Academy of Pediatrics (AAP) advocate for universal COVID-19 vaccination for all individuals aged 6 months and older with the updated 2024-2025 vaccine formula. This guidance is supported by the Pediatric Infectious Diseases Society (PIDS) through its members' active participation in the Advisory Committee on Immunization Practices (ACIP), which formulates these recommendations. Pediatric vaccine experts, such as Dr. Paul Offit, generally align with this approach, emphasizing established scientific processes and public education.
However, a number of prominent adult infectious disease specialists, including Drs. Shira Doron, Monica Gandhi, and Vinay Prasad, have expressed more nuanced, risk-stratified, or critical perspectives regarding the universal application of COVID-19 vaccines to all healthy children. Dr. Doron, while publicly aligning with current CDC recommendations, has previously published on the lower risk profile for children and shares concerns about eroding public trust in vaccines. Dr. Gandhi more explicitly advocates for a risk-based approach, particularly for healthy children with prior infection, suggesting they may not need the annual updated vaccine. Dr. Prasad offers a more fundamental critique, questioning the adequacy of the evidence supporting universal vaccination for healthy children and opposing mandates.
The analysis confirms the user's perception: pediatric infectious disease specialists and pediatricians, as represented by their leading organizations and prominent voices, tend to support the current universal CDC and AAP guidance. In contrast, some adult infectious disease specialists, notably Drs. Gandhi and Prasad, lean towards more selective or critical approaches for healthy children. Significantly, recent ACIP discussions indicate a potential future shift towards a more risk-based recommendation for the 2025-2026 season for individuals under 65, suggesting that the debate is internal to public health advisory bodies as well as external. This evolving landscape underscores the complexity of public health messaging and the importance of ongoing evaluation of vaccine strategies.”
Final thoughts
Here is a summary of my goals
Understand these COVID-19 strategic approaches and trade-offs
Mitigate my confirmation bias
Sample information and knowledge spaces beyond my awareness
Identify mathematical modeling studies that tend to take a more comprehensive, objective approach
Identify sources for further study: This is an emerging area and we need to clarify goals and trade-offs.
What did I learn?
I learned a lot reading these reports.
These report are a place to start
There is no overwhelming consensus (eg, varies by country).
There are trade-offs.
AI can help us mitigate our biases when we study important topics. In turn, this can accelerate our learning.
These reports can be of good quality.
To be more humble!!!
What are my concerns?
AI-bot-generated research reports will pollute the internet and reduce the quality and utility of these reports.
Public health professionals will not take advantage of these disruptive technologies to reduce their biases, improve their learning, and making better decisions and recommendations.
Appendix - Current CDC guidance for COVID-19 vaccines
Shira Doron and Monica Gandhi. “Opinion: Risk-Based Vaccination Strategies Can Build Back Public Trust.” MedPage Today, April 22, 2022. https://www.medpagetoday.com/opinion/second-opinions/115232.
Apoorva Mandavilli. “Federal Officials May Limit Recommendations for Covid Vaccine.” The New York Times, May 15, 2025. https://www.nytimes.com/2025/05/15/health/covid-vaccine-recommendations.html.
Liz Essley Whyte. “HHS to Stop Recommending Routine Covid Shots for Children, Pregnant Women.” The Wall Street Journal, May 15, 2025. https://www.wsj.com/health/healthcare/trump-hhs-rfk-jr-covid-vaccine-31923718.
I used quotes for “risk-based” because because both strategies are based on risk analyses. Using the term “risk-based” might confuse readers to believe that the alternative approach is not risk-based.
Also see Dr. Vinay Prasad’s Substack blog: Vinay Prasad's Observations and Thoughts.
Also see Dr. Paul Offit’s Substack blog: Beyond the Noise.
Oeberst, Aileen, and Roland Imhoff. “Toward Parsimony in Bias Research: A Proposed Common Framework of Belief-Consistent Information Processing for a Set of Biases.” Perspectives on Psychological Science 18, no. 6 (November 2023): 1464–87. https://doi.org/10.1177/17456916221148147. (highly recommended!)
I also tend to align the Center for Infectious Disease Research and Policy.