Optimizing one-dose and two-dose cholera vaccine allocation in outbreak settings: A modeling study

Background A global stockpile of oral cholera vaccine (OCV) was established in 2013 for use in outbreak response and are licensed as two-dose regimens. Vaccine availability, however, remains limited. Previous studies have found that a single dose of OCV may provide substantial protection against cholera. Methods Using a mathematical model with two age groups paired with optimization algorithms, we determine the optimal vaccination strategy with one and two doses of vaccine to minimize cumulative overall infections, symptomatic infections, and deaths. We explore counterfactual vaccination scenarios in three distinct settings: Maela, the largest refugee camp in Thailand, with high in- and out-migration; N’Djamena, Chad, a densely populated region; and Haiti, where departments are connected by rivers and roads. Results Over the short term under limited vaccine supply, the optimal strategies for all objectives prioritize one dose to the older age group (over five years old), irrespective of setting and level of vaccination coverage. As more vaccine becomes available, it is optimal to administer a second dose for long-term protection. With enough vaccine to cover the whole population with one dose, the optimal strategies can avert up to 30% to 90% of deaths and 36% to 92% of symptomatic infections across the three settings over one year. The one-dose optimal strategies can avert 1.2 to 1.8 times as many cases and deaths compared to the standard two-dose strategy. Conclusions In an outbreak setting, speedy vaccination campaigns with a single dose of OCV is likely to avert more cases and deaths than a two-dose pro-rata campaign under a limited vaccine supply.

Thank you very much for your positive comments about our work. In our manuscript, we considered allocating vaccine 15 and 115 days after the outbreak had started in Maela and Chad respectively, and 10 months following the devasting earthquake in 2010 (much longer waiting period) in Haiti. Our results show indeed that vaccination in Haiti, did not have as big an impact as it does in our other two other scenarios. We fully agree that delays to rapid OCV deployment should be stressed. We have expanded on this point in the discussion: However, the procurement and deployment of OCV include significant delays. There is a median delay of three months between the declaration of an emergency and the start of the first round of vaccination [35]. Delay times range widely---the time from the first laboratory confirmation of cholera (or occurrence of humanitarian emergency) to the receipt of the official OCV request spans between 12 and 206 days [35]. These delays in vaccine shipment may result in campaigns starting after the outbreak is over [43].
The authors should describe the three mechanisms for accessing OCV from the global stockpile. These include 1) emergency use to control outbreaks, 2) emergency use to prevent outbreaks during a humanitarian crisis, and 3) preventive use of vaccine for areas determined to be hot-spots. It is not clear which of these situations apply to their model and the manuscript would benefit from aligning their model with these three.
For example, if one is using the vaccine for the emergency control of an outbreak, most outbreaks occur suddenly, and are of relatively short duration so nearly all the benefit occurs in the first year. Another outbreak may not occur for several years so determining a benefit over three years would show little impact relative to a one year impact and providing a second dose to this area will increase costs but provide little additional benefit. This would change of course, if this same area was also previously determined to be a hot-spot. (Outbreaks are more likely to occur in such hot-spots.) If the campaign is intending to vaccinate a "hot-spot" where outbreaks occur frequently, the second dose could more readily demonstrate additional benefit. Both of these situations are also then different from the endemic situation in Bangladesh where cholera occurs continuously at predicable rates. Obviously, a model cannot consider every possible scenario, but the discussion could provide additional context for these different situations.
Thank you for the comments on OCV access from the global stockpile. We agree that it is beneficial to describe the three mechanisms for a request of OCV to give the reader additional context. We have added a few sentences in the introduction and in the discussion.
In the introduction: The three main contexts under which countries may request doses from the global OCV stockpile include 1) emergency use to control an active outbreak, 2) emergency use to prevent outbreaks during a humanitarian crisis, and 3) preventive use for areas deemed ``hot-spots'' or where cholera is endemic [3].
In the discussion: It is important to consider that the optimal allocation of vaccine may change under different epidemiological contexts. In the present work we showed the optimal allocation of OCV under the context of emergency use to control an active outbreak (in epidemic situations). However, cholera outbreaks range widely depending on the region (short outbreaks are commonly seen in Africa while cholera is endemic in Bangladesh), and the optimal vaccine allocation may be different if used to prevent outbreaks or in endemic areas with hot spots.
A specific queries and suggestions.
For the age group <5, did the model assume that all children under age 5 would be vaccinated or only the age group >1 year. (The vaccine is not approved for children <1 year).
We thank the reviewer for pointing out that this was not clear. Yes, the model assumed that all children under five years old were eligible to be vaccinated. We have now clarified this in the methods and added this as a limitation of the study: We also made a simplifying assumption in our models that all children under five years old were eligible to be vaccinated. In reality, the vaccine is not approved for children under one year old.

Reviewer's Responses to Questions
Reviewer #1: yes, the analysis matches the plan The figures need to better label the x and y axes.
We thank the reviewer for this suggestion. We have updated the vertical axis label for the figures of the optimal vaccine allocation strategies and the figures of reductions in the metrics of disease burden to "Vaccination coverage ("% of total population)". We hope that this new label is more clear. Additionally we increased the font size of the ticks.
Reviewer 2: --Overall comment: 1. Methods section is not clear and needs to be more specific.
We thank the reviewer for this comment. We have now modified the Methods section and hopefully improved its clarity.
We have made the following changes: i) We have added a fuller description of the SEIRS-type model, which reads: Figure 1A).

Susceptible (S) individuals are infected directly through contact with an infectious individual or indirectly through contact with contaminated water. They then become exposed but not yet infectious (E). After the latent period, they become infectious either with symptoms (I) or without symptoms (A). Eventually they recover (R) and are temporarily immune to infection until they become susceptible again. This model follows the general SEIRS framework, with two infectious compartments and a rainfall-modulated water reservoir (W) that tracks the concentration of V. cholerae (
ii) The general description of the models above now mentions waning immunity. This would reduce confusion to the reader when the durations of natural and vaccineinduced immunity are described later in the Methods. We have updated the model diagrams in Figure 1 by adding arrows for waning immunity as part of the general framework of the models. Additionally, we have elaborated on the epidemiological states in the figure caption. The Figure 1 caption now reads: Figure 1: General framework of the models. The three models share common epidemiological states: susceptible (S), 2. Inclusion and exclusion criteria needs to be better explained.

exposed but not yet infectious (E), infectious with symptoms (I), infectious without symptoms (A), and recovered (R). Infectious individuals shed bacteria into the water (W) (dashed line
We are unsure what is meant by inclusion and exclusion criteria. If it meant the populations that are included in the vaccination campaign, then we have clarified that the entire population is eligible to receive vaccination (those aged 0 to 4 years, and those aged 5 years and older). We have listed in the limitations of the study that in reality the OCV is not approved for children under 1 year old.
3. In the discussion part, some comments: A. case-control (MCC), test-negative case-control (TNCC) and case-cohort study can be compared about effectiveness of one dose of oral cholera.
We agree that comments on vaccine effectiveness from different study types should be added. We have added the following to the discussion: One-dose vaccine effectiveness from case-control; test-negative, case-control; and casecohort studies range widely from 33% to 87%, and some of these estimates were not statistically significant [52,12,53,17].
B. Besides influenza, comparison with other diarrhoeal agents including rotavirus needs to be made. In 2016, rotavirus infection caused more than 258 million episodes of diarrhea in children under the age of five.
We agree that comparison to rotavirus is important to add context to our results on cholera.
We have included the following in the discussion: Besides cholera, other diarrheal diseases pose significant burden. In 2016, rotavirus infection caused more than 258 million episodes of diarrhea among children under five years old globally [47]. Rotavirus vaccines are also delivered in multiple doses, and some studies have suggested that an incomplete series of the rotavirus vaccine may provide sufficient protection against severe disease and hospitalizations [48,49,50].
We have added a few confounders to the limitation section of the paper. They now read: Further studies are needed to truly evaluate the burden of cholera in young children. In particular, serological studies [52] might be useful to determine the burden of different diarrheal diseases in a particular region. We did not consider other possible confounders (e.g., differences in waning immunity and pre-existing immunity in endemic versus epidemic settings, the use of other interventions in tandem with vaccination) which could affect the population impact of vaccination.
Page 11: Better to add in the conclusion part the following "However, in places where there was a lot of population migration, the long time between treatments meant that many people only got one OCV dose. In the event that the risk persists, more vaccination campaigns may be required to maintain protection over time." Or this can be added in the discussion part. This is a very good point, and we thank the reviewer for this suggestion. The last sentence of the paper now reads: However, in places where there is a lot of population migration, a long interval between the first and the second dose might result in having a proportion of the population receiving only one dose. In the event that the risk of future epidemics persist, recurrent vaccination campaigns may be required to maintain protection over time. Figure 1 : Elaboration of "S,E,A,I,R,W" is needed.
The caption of Figure 1 now elaborates on the SEIARW compartments to add more clarity to the Methods section. See our response to Reviewer 1's comment above for what the caption now reads.