When to Boost: How Dose Timing Determines the Epidemic Threshold

1. Problem Statement

The paper addresses a critical public health challenge: optimizing vaccination schedules for diseases requiring multiple doses (e.g., mRNA vaccines for SARS-CoV-2). While standard protocols often recommend fixed intervals between the first (primer) and second (booster) doses, resource constraints and varying vaccine efficacies necessitate strategic prioritization.

• The Dilemma: Should limited vaccine supplies be used to maximize the number of individuals receiving a first dose (partial protection, wide coverage), or should they be used to ensure individuals receive both doses quickly (higher individual protection, lower coverage)?
• The Gap: Previous studies have touched on timing, but this paper aims to mathematically derive the epidemic threshold ($\lambda_c$) as a function of the prioritization strategy between first and second doses, determining when a specific scheduling strategy shifts the system from an endemic state to a disease-free state.

2. Methodology

The authors employ a combination of analytical compartmental modeling and agent-based simulations on complex networks.

A. The S3I2 Compartmental Model

The authors introduce a generalized S3I2 model (Susceptible-Infected-Susceptible with 3 Susceptible and 2 Infected states) to account for hybrid immunization (vaccination and natural infection) and waning immunity.

• States:
  • $S_0$: Naive (no immunity).
  • $S_1$: Partially immune (1st dose or natural infection).
  • $S_2$: Highly immune (2nd dose or reinfection).
  • $I_1, I_2$: Infected states corresponding to $S_1$ and $S_2$ susceptibility levels.
• Key Parameters:
  • $\alpha \in [0, 1]$: The priority index.
    • $\alpha = 1$: Absolute priority to first doses (maximizing $S_1$).
    • $\alpha = 0.5$: First-come, first-served (standard approach).
    • $\alpha < 0.5$: Priority to second doses (boosters).
  • $c$: Overall vaccination rate.
  • $\tau$: Inter-dose interval, defined as $\tau = 1/(c(1-\alpha))$.
  • $\sigma_1, \sigma_2$: Reduction factors in transmission rate for $S_1$ and $S_2$ respectively ($\sigma_2 \le \sigma_1 \le 1$).
  • $\eta_1, \eta_2$: Waning rates for partial and maximal immunity.
• Dynamics: The model uses a system of differential equations (Eq. 1) to track the flow between states. The transition from $S_0 \to S_1$ occurs at rate $c\alpha$, and $S_1 \to S_2$ at rate $c(1-\alpha)$.

B. Analytical Derivation

• Stationary Analysis: The authors analyze the system at stationarity to find the epidemic threshold ($\lambda_c$), the critical transmission rate above which the disease becomes endemic ($\rho > 0$).
• Optimization: They derive an explicit expression for $\lambda_c(\alpha)$ and solve for the optimal priority index $\alpha^*$ that maximizes $\lambda_c$. A higher $\lambda_c$ implies a more robust system capable of withstanding higher transmission rates without an outbreak.
• Critical Threshold ($c^*$): They identify a critical vaccination rate $c^*$ that separates two distinct regimes of optimal strategy.

C. Agent-Based Simulations

To validate the mean-field analytical results, the authors performed simulations on three network topologies:

1. Erdős–Rényi (ER) random graphs.
2. The Enron email network (real-world social structure).
3. Geometric random graphs (S1 model).
   They utilized a quasi-stationary state sampling method to accurately estimate the critical infection rate $\beta_c$ in finite-size systems.

3. Key Contributions

1. Mathematical Proof of Regime Shift: The paper proves that the optimal vaccination strategy is not static; it depends on the vaccination rate ($c$).
   • Low Vaccination Rate ($c \le c^*$): The optimal strategy is absolute priority to first doses ($\alpha^* = 1$). This maximizes the number of partially protected individuals, effectively reducing the susceptible pool ($S_0$) the most.
   • High Vaccination Rate ($c > c^*$): The optimal strategy shifts to a non-trivial prioritization ($\alpha^* < 1$). In this regime, a specific balance between first and second doses (shortening the inter-dose interval) yields a higher epidemic threshold than simply rushing first doses.
2. Explicit Threshold Formula: The authors provide a closed-form expression for the critical vaccination rate $c^*$ and the optimal priority index $\alpha^*$, allowing public health officials to calculate the exact scheduling needed based on available supply and vaccine efficacy parameters.
3. Hybrid Immunization Framework: The model uniquely integrates natural infection and vaccination, acknowledging that prior infection acts as a "first dose," which is crucial for modeling real-world scenarios where immunity is heterogeneous.

4. Results

• Epidemic Threshold Maximization: The analytical results (Fig. 3 and Fig. 4) show that $\lambda_c$ is a non-monotonic function of $\alpha$. There exists a peak $\lambda_c$ at $\alpha^*$.
  • For low $c$, the peak is at $\alpha=1$.
  • For high $c$, the peak shifts to $\alpha < 1$, indicating that delaying the first dose slightly to ensure faster booster delivery improves population-level protection.
• Validation: Agent-based simulations on ER, Enron, and S1 networks (Fig. 5) confirmed the theoretical predictions. The simulated critical infection rates ($\beta_c$) peaked at the predicted $\alpha^*_{sim}$, validating the mean-field approximation even on structured networks.
• Gain Analysis: The "gain" $R$ (the improvement in threshold over a first-come-first-served approach) is significant. The paper demonstrates that suboptimal scheduling (e.g., rigidly following manufacturer intervals regardless of supply) can lead to preventable endemic states.

5. Significance and Implications

• Policy Guidance: The study provides a quantitative framework for health authorities to adjust vaccination campaigns dynamically. It suggests that during severe supply shortages, "one dose for everyone" is mathematically superior. However, as supply increases, the strategy should pivot to optimizing the booster interval to maximize herd immunity.
• Resource Allocation: It resolves the ethical and practical trade-off between individual protection (two doses) and collective protection (one dose for many) by showing that the "best" choice is mathematically determined by the vaccination rate.
• Future Directions: The authors note limitations, such as the assumption of homogeneous mixing and the lack of vaccine hesitancy modeling. They suggest future work should incorporate heterogeneous populations (e.g., prioritizing vulnerable groups) and the risk of vaccine-resistant strains emerging from prolonged partial immunity.

In conclusion, the paper demonstrates that timing is a critical control parameter in epidemic management. By mathematically tuning the inter-dose interval and prioritization scheme based on the current vaccination rate, it is possible to shift the epidemic threshold and potentially eliminate the disease, whereas a static approach may fail.