During 2021 and 2022 Starboard Maritime Intelligence included a Covid-19 relative risk model to support those working at borders and completing vessel inspections to manage the risk of coming into contact with Covid-19.
Prior to the Omnicron strain of Covid-19, Aotearoa and several of the Pacific island’s management of Covid-19 included an elimination strategy and border closures.
The relative Covid-19 risk model was removed from Starboard in November 2022 reflecting that these strategies and restrictions have largely been removed for most countries in the Pacific.
How Covid-19 risk assessments were reflected in the app
Every vessel had a comparative risk rating, which was an output of our Covid-19 model. The epidemiological model took in several factors about a vessel’s history and daily incidence rates of Covid-19 in order to mathematically simulate contagion scenarios.
The colour of the vessel on the map represented the estimated comparative risk of crew onboard being infected with Covid-19. Red meant higher risk, orange and yellow lower. The risk rating could also be found in the vessel report and on the vessel details panel.
Due to the unpredictable nature of real-world outbreaks, the rating was not meant to be definitive, but rather representative of the most-likely scenario. Note that a vessel with comparatively high risk rating still had low absolute probability of infection (i.e. <1%).
Covid-19 information was updated daily.
Figure 1. Epidemiology risk model based on vessel history
How the Covid-19 model worked
The Covid-19 risk was assessed using an epidemiological model that mathematically estimated likely contagion scenarios based on a vessel’s travel history.
Epidemiological models are widely used to inform policy development¹, predict outbreaks², and implement controls measures³ in response to Covid-19 both in Aotearoa New Zealand⁴𝄒⁵ and abroad.
Our model was based on the work of public health experts Wilson et al.² and was derived from a stochastic version of the compartmental CovidSIM model⁶, which assigned people to compartments based on their infection status (susceptible, exposed, infectious, and recovered/removed) and accounted for the unpredictable nature of real-world outbreaks.
The model was populated with parameters for SARS-CoV-2 transmission, historical infection rates across the globe, and shipping characteristics for each vessel. A list of parameters is found in Table 1.
For each vessel, we considered an initially uninfected crew of 20 at a time 30 days ago. At each subsequent port visit, there is the possibility that crew may become infected as a result of interaction with the community, due to either shore leave or routine contact with port staff, stevedores, maintenance workers, etc. Contagiousness for the duration a vessel was in port, as defined by the effective reproduction number, is 2.5². Historical infection rates at the time the vessel was in port for each port country are obtained from Johns Hopkins University⁷.
During the subsequent voyage, any infected crew members can potentially infect others on board. Contagiousness aboard the vessel is set at 3.0². We assumed that 71% of infected Covid-19 cases develop clearly detectable symptoms².
This process was repeated sequentially until the present time, at which the number of infected crew members was counted. Due to the very high probability of zero infected crew members, the simulation was repeated half a million times for each vessel. This results in a distribution, or histogram, showing the likelihood of different outcomes, i.e. the number of infected crew members. The very large number of simulations with zero infected crew is beyond the axis range of the plot and is therefore not shown.
The uncertainty (spread) around possible contagion scenarios is high², but by taking the average outcome over half a million simulations, we can approximate the most likely scenario, i.e. the average number of infected crew (Figure 2).
Figure 2. The most likely contagion scenario is found by calculating the average number of infected crew over half a million simulations
The final step is to classify each vessel as high, medium or low risk. This is done by considering the vessel’s risk relative to all others calculated in the past month. High risk is assigned to vessels having a likelihood of infected crew in the top 10% relative to all others. Medium risk is assigned to those in the top 75–90%, and low risk is assigned to those vessels in the 0–75% bracket of risk. This is done by calculating a cumulative distribution function based on all vessel risks from the past month and identifying the corresponding brackets of risk.
Figure 3. High risk is assigned to vessels having a likelihood of infected crew in the top 10% relative to all others. Medium risk is assigned to those in the top 75–90%, and low risk is assigned to those vessels in the 0–75% bracket of risk.
Vessels and their estimated comparative risk were then visualised in Starboard and updated on a daily basis.
We included all commercial (cargo/tanker) vessels within the vicinity of Aotearoa New Zealand.
Incidence of SARS-CoV-2 infection
Based on daily incidence rates from Johns Hopkins University⁷, adjusted for under-estimation by using a 10-fold difference between reported cases and infections²,⁸
Percent of infections that are asymptomatic
See 2, 9
See 2, 10
10 days (split into 2 periods of 5 days each)
See 2, 11
Relative contagiousness in the prodromal period
See 2, 10
Contagiousness after the prodromal period
100% (first 5 days), 50% (second 5 days)
See 2, 12
Effective reproduction number on board the ship
Effective reproduction number in the port
See 2, 10
Duration in port
Based on vessel transponder data
Based on vessel transponder data.
See 2, 13.
Thompson, R.N. Epidemiological models are important tools for guiding Covid-19 interventions. BMC Med 18, 152 (2020). https://doi.org/10.1186/s12916-020-01628-4.
Wilson N, Blakely T, Baker M, Eichner M. Estimating the Risk of Outbreaks of Covid-19 Associated with Shore Leave by Merchant Ship Crews: Simulation Studies for New Zealand. N Z Med J (in press).
Wilson, N., Baker, M. G., & Eichner, M. (2020). Estimating the Impact of Control Measures to Prevent Outbreaks of Covid-19 Associated with Air Travel into a Covid-19-free country: A Simulation Modelling Study. medRxiv.
Jefferies, S., French, N., Gilkison, C., Graham, G., Hope, V., Marshall, J., … & Prasad, N. (2020). Covid-19 in New Zealand and the impact of the national response: a descriptive epidemiological study. The Lancet Public Health, 5(11), e612–e623.
Wilson, N., Telfar Barnard, L., Kvalsvig, A., & Baker, M. (2020). Potential health impacts from the Covid-19 pandemic for New Zealand if eradication fails: report to the NZ ministry of health. Wellington: University of Otago Wellington.
Schneider, K., Ngwa, G. A., Schwehm, M., Eichner, L., & Eichner, M. (2020). The Covid-19 Pandemic Preparedness Simulation Tool: CovidSIM. Available at SSRN 3578789.
Dong E, Du H, Gardner L. An interactive web-based dashboard to track Covid-19 in real time. Lancet Inf Dis. 20(5):533-534. doi: 10.1016/S1473-3099(20)30120-1.
Havers FP, Reed C, Lim T, Montgomery JM, Klena JD, Hall AJ, et a. Seroprevalence of Antibodies to SARS-CoV-2 in 10 Sites in the United States, March 23-May 12, 2020. JAMA Intern Med 2020.
Pollan M, Perez-Gomez B, Pastor-Barriuso R, Oteo J, Hernan MA, Perez-Olmeda M, Sanmartin JL, Fernandez-Garcia A, Cruz I, Fernandez de Larrea N, Molina M, Rodriguez-Cabrera F, Martin M, Merino-Amador P, Leon Paniagua J, Munoz-Montalvo JF, Blanco F, Yotti R, Group E-CS. Prevalence of SARS-CoV-2 in Spain (ENE-COVID): a nationwide, population- based seroepidemiological study. Lancet 2020.
Centers for Disease Control and Prevention. Covid-19 pandemic planning scenarios. 2020; (20 May). https://www.cdc.gov/coronavirus/2019-ncov/hcp/planning-scenarios.html.
WHO-China Joint Mission. Report of the WHO-China Joint Mission on Coronavirus Disease 2019 (Covid-19). 2020; (16–24 February). https://www.who.int/docs/default-source/coronavirus/who-china-joint-mission-on-covid-19-final-report.pdf.
Woelfel R, Corman V, Guggemos W, Seilmaier M, Zange S, Mueller M, Niemeyer D, Vollmar P, Rothe C, Hoelscher M, Bleicker T, Bruenink S, Schneider J, Ehmann R, Zwirglmaier K, Drosten C, Wendtner C. Clinical presentation and virological assessment of hospitalized cases of coronavirus disease 2019 in a travel-associated transmission cluster. medRxiv 2020; (8 March). https://www.medrxiv.org/content/10.1101/2020.03.05.20030502v1.
Wikipedia. Maritime transport. (Accessed 20 August 2020). https://en.wikipedia.org/wiki/Maritime_transport.