Transmission of SARS-CoV-2 Associated with Cruise Ship Travel: A Systematic Review

Background: Maritime and river travel may be associated with respiratory viral spread via infected passengers and/or crew and potentially through other transmission routes. The transmission models of SARS-CoV-2 associated with cruise ship travel are based on transmission dynamics of other respiratory viruses. We aimed to provide a summary and evaluation of relevant data on SARS-CoV-2 transmission aboard cruise ships, report policy implications, and highlight research gaps. Methods: We searched four electronic databases (up to 26 May 2022) and included studies on SARS-CoV-2 transmission aboard cruise ships. The quality of the studies was assessed based on five criteria, and relevant findings were reported. Results: We included 23 papers on onboard SARS-CoV-2 transmission (with 15 reports on different aspects of the outbreak on Diamond Princess and nine reports on other international cruises), 2 environmental studies, and 1 systematic review. Three articles presented data on both international cruises and the Diamond Princess. The quality of evidence from most studies was low to very low. Index case definitions were heterogeneous. The proportion of traced contacts ranged from 0.19 to 100%. Studies that followed up >80% of passengers and crew reported attack rates (AR) up to 59%. The presence of a distinct dose–response relationship was demonstrated by findings of increased ARs in multi-person cabins. Two studies performed viral cultures with eight positive results. Genomic sequencing and phylogenetic analyses were performed in individuals from three cruises. Two environmental studies reported PCR-positive samples (cycle threshold range 26.21–39.00). In one study, no infectious virus was isolated from any of the 76 environmental samples. Conclusion: Our review suggests that crowding and multiple persons per cabin were associated with an increased risk of transmission on cruise ships. Variations in design, methodology, and case ascertainment limit comparisons across studies and quantification of transmission risk. Standardized guidelines for conducting and reporting studies on cruise ships of acute respiratory infection transmission should be developed.


Introduction
The emergence of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was declared a public health emergency by the World Health Organization (WHO) in 1.
If so, what is/are the predominant mode(s) of transmission? 3.
Are there particular circumstances that facilitate transmission (practices, ship layout, or populations)?

Materials and Methods
The present work is an open evidence review on SARS-CoV-2 transmission associated with cruise ship travel. We developed the present protocol [13] based on a previous protocol used for a series of systematic reviews on the transmission dynamics of COVID-19 (Supplementary Materials File S1).
We searched four electronic databases: LitCovid, medRxiv, Google Scholar, and WHO COVID-19 database, up to 26 May 2022. Search terms were COVID-19, SARS-CoV-2, transmission, ship, and appropriate synonyms (Supplementary Materials File S2). We also searched for additional studies through checking reference lists of relevant articles, including reviews. We did not set any language restrictions.
We included studies reporting on SARS-CoV-2 transmission aboard cruise ships, from passengers and crew to passengers or crew. We considered any potential modes of transmission, including long-range airborne, close contact, droplet, fomite, fecal-oral, and/or mixed routes. We included studies of any design except predictive or modeling studies. If two or more papers presented the same data, we included only the most comprehensive report. Articles were excluded if they did not report primary data.
One reviewer (ECR) extracted data from included studies, and these were independently checked by a second reviewer (EAS). Disagreements were resolved via consensus.
The quality of included human and environmental studies was assessed based on a modified QUADAS-2 tool using five criteria. A detailed presentation of the quality assessment is presented in the review protocol [13].
As the included studies were not primarily designed as diagnostic accuracy studies, we adapted the QUADAS-2 tool. Two reviewers (ECR, EAS) independently assessed the quality of included studies. We resolved disagreements via consensus. We did not formally assess the quality of included systematic reviews but summarized their findings.
For studies generating a hypothesis of onboard SARS-CoV-2 transmission, we also assessed the strength of evidence of each study depending on the methods used to investigate the virus transmission [14]. The results are presented in tabular format. Where relevant, we reported results of specific subgroups. Meta-analysis was only considered appropriate if minimal heterogeneity was found among included studies.
Six studies reporting on the DP outbreak presented data focusing on repatriated citizens from the US [7,26,29], Hong Kong [21], Israel [17], and Australia [35]. Although there were similarities between results of the studies on the US repatriated citizens [7,26,29], there were also some notable discrepancies, including the number of repatriated citizens, the number of SARS-CoV-2 cases, and importantly, the number of symptomatic and asymptomatic cases (Supplementary Materials File S6).
Our search identified one systematic review on SARS-CoV-2 outbreaks on ships [38]. The authors searched four databases up to 31 July 2020. They included 37 studies: 33 reported several aspects of SARS-CoV-2 outbreaks from cruise ships, three reported outbreaks of SARS-CoV-2 on navy vessels, and one study presented an outbreak on a cargo ship [38]. The review did not assess the risk of bias in the included studies.   Abbreviations: px-passengers; Ct-cycle threshold; RT-PCR-real time reverse transcription-polymerase chain reaction; GS-genome sequencing. Wastewater from cruise ship sanitation system; two wastewater grab samples (1 L) were collected from the influent and effluent of the membrane bioreactor of a cruise ship.
For the untreated wastewater collected from the cruise ship, all six replicate samples prepared using both virus concentration methods yielded a positive signal for SARS-CoV-2 RNA using the CDC N1 assay. The CDC N2 and NIID_2019-nCoV N assays detected SARS-CoV-2 RNA in four replicate samples. The E_Sarbeco assay appeared to be less analytically-sensitive (i.e., greater ALOD); only one of six replicates were RT-qPCR positive. The N_Sarbeco assay did not produce any amplification for these samples in two consecutive RT-qPCR runs. The CDC N1 and CDC N2 assays were consistently positive in replicate RT-qPCR reactions. The results showed positive SARS-CoV-2 signals, though concentrations were close to the limit of detection For the adsorption-extraction method, the mean Cq value (Cq = 33.5) of the CDC N1 assay was much lower than the mean Cq value (Cq = 38) of CDC N2, E_Sarbeco, and NIID_2019-nCoV N. For ultrafiltration with the Amicon ® Ultra-15, the mean Cq value (Cq = 36.5) of the CDC N1 assay was slightly lower than the mean Cq value (Cq = 37.15) of CDC N2, E_Sarbeco, and NIID_2019-nCoV N assays. Among the replicate cruise ship untreated wastewater samples, four of six replicate samples were positive according to the CDC N1 RT-ddPCR assay. Of the five replicate cruise ship effluent wastewater samples (after treatment), Cq values ranged from 36.0 to 38.7. Greater concentrations were observed in the influent from the cruise ship in comparison with the effluent of the cruise ship. The frequency of SARS-CoV-2 RNA detection in treated cruise ship effluent wastewater was low in replicate RT-qPCR reactions compared with the cruise ship influent sample; this indicates that SARS-CoV-2 removal occurred in the wastewater treatment process.  29.79) days after cabins were vacated, both obtained from the floor around the toilet. These findings suggest that environmental surfaces may have played a role in transmission of the virus. SARS-CoV-2 RNA was detected on multiple surfaces of case cabins, most often on bed pillows and the floor around the toilet in the bathroom, for up to 17 days, longer than previously reported. There was no difference in surface contamination between cabins of cases who were symptomatic and asymptomatic.
It was evident that surface contamination occurred in rooms occupied by persons who were classified as being asymptomatic at the time they vacated their cabins. The high Cq values in most of the positive samples suggested low-level contamination of the environment after the COVID-19 cases vacated the cabins, potentially explaining why no virus was isolated.

Quality of Included Studies
None of the studies reported a published protocol. The risk of bias assessment of the included studies is presented in Tables 3 and 4. Abbreviations: Ct-cycle threshold; Cq-quantification cycle; RT-PCR-real time reverse transcriptionpolymerase chain reaction.   For the environmental studies [15,16], we found an adequate description of their methods, with sufficient detail to replicate the findings. Sample sources were clearly reported, and the analysis and reporting were considered appropriate; there were no concerns about their applicability. However, we found that neither study adequately addressed potential biases (Table 4).
In studies reporting outbreaks on international ships, 1/11 (9%) reports clearly described the setting [34]. Demographic characteristics and sampling procedures were unclear in all studies. A comprehensive follow-up strategy was presented only in 3/11 (27%) studies [22,32,36]. Transmission outcomes were not adequately assessed by any authors (0/11), and only 1/11 (9%) studies took into consideration the validity of the data [22] ( Table 4). The overall reporting quality across studies was considered low (Figure 1).
For the environmental studies [15,16], we found an adequate description of their methods, with sufficient detail to replicate the findings. Sample sources were clearly reported, and the analysis and reporting were considered appropriate; there were no concerns about their applicability. However, we found that neither study adequately addressed potential biases (Table 4).
In studies reporting outbreaks on international ships, 1/11 (9%) reports clearly described the setting [34]. Demographic characteristics and sampling procedures were unclear in all studies. A comprehensive follow-up strategy was presented only in 3/11 (27%) studies [22,32,36]. Transmission outcomes were not adequately assessed by any authors (0/11), and only 1/11 (9%) studies took into consideration the validity of the data [22] ( Table 4). The overall reporting quality across studies was considered low (Figure 1).

Environmental Studies
One study collected wastewater samples from a docked cruise ship [15] over a month after passengers disembarked, with only crew on board. Unconfirmed reports suggested that 24 infected individuals may have been on board in the days before sample collection. The wastewater samples were positive for SARS-CoV-2 RNA, but concentrations were near the assay limit of detection (Cycle quantification [Cq] 36.0-38.7). Greater concentrations were observed in the influent from the cruise ship than the effluent ( Table  2).

Environmental Studies
One study collected wastewater samples from a docked cruise ship [15] over a month after passengers disembarked, with only crew on board. Unconfirmed reports suggested that 24 infected individuals may have been on board in the days before sample collection. The wastewater samples were positive for SARS-CoV-2 RNA, but concentrations were near the assay limit of detection (Cycle quantification [Cq] 36.0-38.7). Greater concentrations were observed in the influent from the cruise ship than the effluent ( Table 2).
The DP departed from Yokohama on 20 January 2020 and visited six ports in three countries (Japan, Hong Kong, and Vietnam). The timeline of the cruise is presented in Supplementary Materials File S7. However, only one report presented a detailed itinerary [30], and group activities were investigated by three studies [26,30,35]. On 3 February, the ship returned to Yokohama. On 5 February, passengers were quarantined in their cabins for 14 days; the crew continued to maintain the ship's functions and assist passengers with food, clothing, and shelter-related needs.
On 19 February, the disembarkation of uninfected passengers began. Forty-one days after the ship started the voyage, on 1 March, disembarkation was completed, with the last crew members leaving the ship [7].
Nine studies reported SARS-CoV-2 outbreaks at the international level [6,7,18,22,28,[32][33][34]36]. Nonetheless, the total number of the voyages could not be calculated, as some authors did not report specific data about the cruise (i.e., names of the ships or origin and destinations of vessels). Therefore, some voyages may have been included in several different reports. The total number of passengers and crew members in each study is presented in Table 1.
One study presented data on 89 voyages, including 70 ships from US waters or carrying US citizens; 16 ships had recurrent outbreaks [18]. Other authors [6] reported an analysis of 36 ships with COVID-19 globally and an analysis of data from 10 cruise liners with SARS-CoV-2 cases from Australia.
One study reported COVID-19 cases from two Nile River cruises [28], but the total number of passengers and crew was not mentioned. Two studies reported on single cruise ship outbreaks (Greg Mortimer ship [22] and Grand Princess [7]). Two studies reported on the outbreak on the Costa Atlantica cruise ship, docked at Nagasaki City since January 2020 for complete maintenance, with only crew on board and no passengers [32,36]. Another report investigated the SARS-CoV-2 outbreak on a Norwegian cruise vessel [34], with two one-week voyages, and another study analyzed data on five cruises with 80 itineraries between the different Canary Islands [33].
The technical specifications of the ship were provided in one study [33]. Four studies reported the voyage duration [22,28,34,36], and the itinerary was detailed by one study [22]. All studies were conducted in the first half of 2020, except the Norwegian cruises (July-August 2020) [34] and the cruises in the Canary Islands (November 2020-May 2021) [33]. The latter study was performed after the implementation of vaccination programs. However, none of the COVID-19 cases were vaccinated, and information re-garding the vaccination statuses of other passengers and the crew was unavailable to the authors [33].

Case Definitions: Index Cases, Contacts, and Secondary Cases
The definition of index cases, contacts, and secondary cases varied across studies (Supplementary Materials File S8).
Eight studies reporting the DP outbreak [17,[19][20][21]24,29,35,37], and 10/11 of the studies reporting other international outbreaks [6,7,18,22,28,32,33,36] did not provide a clear definition of the index case. In addition, close contacts were considered to be cabinmates of confirmed case-patients [25,30] or suspected case-patients [30], individuals who had been in a room with someone for more than 15 min without wearing a mask [26,33], or passengers who joined the Kagoshima tour with the index case from the DP [30]. Case definitions for secondary infections included asymptomatic and symptomatic passengers or crew.
On the DP, initially, only individuals with fever or respiratory symptoms and their close contacts were tested by RT-PCR for SARS-CoV-2. Subsequently, to support phased disembarkation of passengers, testing was expanded, prioritizing older individuals, people with comorbidities, and people accommodated in interior cabins without access to the outdoors [7,25,31].
Eight studies followed up only a sub-population: repatriated cases [17,21,26,29,35], crew [23], or individuals admitted to a single center [20,24]. Considering all the studies, the proportion of traced contacts ranged from 0.19 to 100% in these studies. The total numbers of identified passengers and crew and the total number of successfully traced individuals could not be calculated as several studies did not report specific data on the number of passengers, crew, or medical staff on board [6,18,28] (Table 1).

Reporting of Index Cases, Secondary Cases, Contact Tracing, and Follow-Up
Four studies on the DP outbreak did not provide any information on the index case; the remaining eleven studies report inconsistent data. Several authors considered the index case to be an 80-year-old man who boarded the ship on 20 January and disembarked in Hong Kong on 25 January [7,19,21,26,27,30,35]. However, another study reports that a different passenger with laboratory-confirmed COVID-19 developed symptoms on 22 January [23]. Other authors considered the case reported by Hong Kong authorities merely an indicator case, i.e., the first detected individual among many infected persons [25]. Another study suggested that the outbreak most likely originated from either a single infected person, or simultaneously with another primary case [37]. Additionally, data on both the nature and, most importantly, the onset of symptoms was highly variable. Yamagishi et al. reported that the Hong Kong passenger had a cough starting one day before embarkation [30]. Other authors reported that the index case presented with a mild cough on 23 January [19,27], and developed a fever on 29 January [19] or 30 January [31]. Many cases had onset dates before the ship arrived in Yokohama and, by that time, infection had already spread across several decks without any spatial clustering [25].
Five studies reporting SARS-CoV-2 outbreaks on over 100 international ships did not present any data on index cases [6,18,28]. In five studies, the number of index cases could not be determined [7,22,33,34,36], as multiple passengers or crew presented with respiratory symptoms around the same period. In 11 studies with data on the index case, the laboratory diagnosis was based on binary RT-PCR results, without data on Ct (Supplementary Material S8). Two studies provided the Ct cutoff value [32,34]. The Costa Atlantica crew was initially assessed with a reverse transcription loop-mediated isothermal amplification (RT-LAMP) assay; RT-qPCR confirmed the results of the COVID-19-positive samples [32,36]. The timing of RT-PCR is clearly stated for the Hong Kong passenger from the DP and for passengers and crew from two other reports [7,22]. On the Greg Mortimer ship, the tests were performed on the 20th day of the voyage, 12 days after the first case presented symptoms [22]; on the Grand Princess, the RT-PCR was performed on the 23rd day of the voyage [7].
The secondary cases on the DP were reported as asymptomatic, pre-symptomatic, and symptomatic. At that time, only respiratory symptoms and fever were considered indicative of SARS-CoV-2 infection. The studies with a comprehensive tracing strategy found an attack rate (AR) of 18.5 [25], 18.8 [31], or 19.2% [7]. The number of asymptomatic cases at the time of testing, based on the limited definitions for symptomatic illness in use at the time, was reported to be 46.5 [7], 55.0 [25], and 58.9% [31]. Up to 24 February, among 687/3711 individuals with SARS-CoV-2 infection, 544 (20.4%) were passengers, and 314 (57.7%) of those were reported to be asymptomatic. In addition, 143 (13.7%) were crew members, with 64 (44.8%) reported as being asymptomatic. Symptomatic cases among passengers peaked on 7 February. Thirty-five passenger cases had symptom onset before 5 February. Cases among crew members peaked on 11 and 13 February [25].
In crew members, the earliest case of infection was detected in a food service worker who developed a fever on 2 February and had a positive RT-PCR the following day [23]. By 9 February, 31 crew members reported with a fever, but only 20 had a positive RT-PCR test; 15/20 confirmed cases were food service workers, and 16/20 cases occurred among persons with cabins on deck 3, where food service workers lived [23].
One study investigating SARS-CoV-2-positive crew members from the DP found that 4/7 cases presented with throat swab cultures and sputum samples that were positive for bacterial infections (Haemophilus influenzae, Klebsiella pneumoniae, and Staphylococcus aureus). Three cases were asymptomatic and four presented with a cough [20]. Also, among Australian citizens, five individuals were confirmed positive for Influenza A [35].
In Hong Kong repatriated citizens, eight cases that tested negative in Japan had a positive RT-qPCR upon arrival, five days later [21]. The median viral load in nasopharyngeal samples at baseline was 4.31 log10 copies/mL (IQR 3.79-6.65). One case presented anti-RBD IgG with undetectable viral load. Six patients remained asymptomatic during quarantine [21]. In Israeli repatriated citizens, 4/6 confirmed cases were initially diagnosed in Japan and 2/6 upon arrival to Israel (with Ct values of 24-40) [17]. Three patients were reported as being asymptomatic, and others were paucisymptomatic [17]. In the US repatriated citizens [7,26,29] (Supplementary Material Files S6 and S8), the most comprehensive study [26] reported that 114/437 (26%) individuals tested positive for SARS-CoV-2; 98 cases had a positive RT-PCR in Japan, 10 cases tested negative in Japan but had a positive result in the US, and 6 cases were never tested in Japan, with a positive result in the US [26]. Of 66 travelers with positive tests and complete symptom information, 14 (21%) were asymptomatic while on the ship [26]. Among Australian citizens, 46/56 were diagnosed in Japan and 10/56 after repatriation, with 29% asymptomatic cases [35].
On other international voyages, the number of secondary cases were not reported [6,18,36] or were unclear [7,22,33,34,36] because the number of index cases were unknown. The authors reported only the total number of passengers or crew infected with SARS-CoV-2. One GS study, reporting on two Nile River cruises, found three passengers on one cruise and two on another, all symptomatic, who were most probably infected during their voyage [28]. The authors traced only these passengers, without any data on other passengers or crew [28]. The AR varied between 7.67 and 75.12% [6,7,18,22]. The analysis of 36 cruise ships with reported SARS-CoV-2 infections worldwide showed an overall AR of 8.66% (0.03-75.10%) [6].

Spatial Distribution
Spatial distribution of passengers and crew was investigated in eight studies on the DP [7,23,[25][26][27]30,31,35] and seven studies on international ships [6,7,22,[32][33][34]36]. On the DP, >80% of crew cabins were on decks 2-4. Initially, most cases occurred among persons with cabins on deck 3 [23]. By the end of quarantine, the distribution of infections from the crew decks produced a more generally distributed pattern, but a large number of cases were noted on deck 3 [25]. The mean number of persons per cabin was 1.73 (1 to 3) for crew and 1.98 (1 to 4) for passengers [7].
Another study from the DP conducted between 3-9 February reported that the ARs among passengers were similar across decks. Among confirmed cases, 144 (84%) met the definition of a suspected case before testing, whereas 19 (11%) shared a cabin with a confirmed person. There were 24 asymptomatic cases (14%); most of them were passengers who joined the Kagoshima bus tour (with the assumed Hong Kong index case) [30].
In US citizens [26], the AR for passengers in single cabins or without infected cabin mates was 18% (58/329), 63% (27/43) for passengers that shared the cabin with an asymptomatic infected person, and 81% (25/31) for those with a symptomatic infected cabinmate, again suggesting a dose-response relationship (p < 0.01). Genome sequences from persons sharing cabins clustered together [26]. In addition, the association between attending some events, such as the bus excursion in Cai Rang, and group activities on 3-4 February, suggested several everyday mass exposure events [26].
In Australian citizens, before the quarantine, exposure to a close contact or cabinmate confirmed later as SARS-CoV-2 positive was associated with a 3.78-fold (95%CI, 2.24-6.37) higher risk of infection. Exposure to a positive cabinmate during the quarantine period resulted in a significantly increased risk of infection RR 6.18 (95% CI, 1.96-19.46) [35]. The authors found no statistically significant association between participation in shore trips, tours, or social events before quarantine or visiting public areas during quarantine and subsequent infection [35].
Other authors also noted that during quarantine on the DP, the elevator hall situated in front of the medical center may have presented a higher infection risk because infected and uninfected people could not use the elevator separately [31].
Studies on other international ships also reported on the spatial distribution of people on board. On the Greg Mortimer, in 10 instances, two passengers sharing a cabin presented positive and negative results [22]. On the Grand Princess, the median number of persons per cabin was 1.95 (range of 1-4) for passengers and 1.75 (range of 1-4) for the crew; among 469 persons with available results, 78 (16.6%) had a positive RT-PCR [7].
A study investigating 36 cruise ships [6] found that the number of available cabins presented a moderate inverse correlation with the AR; as the number of available cabins per ship increased, the AR decreased. There was a fair inverse correlation of decks with cabins; the more spread out the cabins were through several decks, the lower the AR. Also, the authors found a moderate positive correlation with the passenger-to-space ratio; as the passenger-to-space ratio increased (i.e., the ship became more crowded), the AR increased [6], suggesting a dose-response relationship. The AR was predicted by all spatial distribution variables (i.e., number of cabins per ship and decks with cabins), but not by cruise duration [6].
In the Canary Islands cruises, with a preventive protocol, the AR between the close contacts during quarantine was 3%, and 21% of confirmed cases were casual contacts of a case at onboard food and beverage venues [33].
Another study reported on a ship with updated but not fully implemented prevention protocols. The first symptomatic cases worked in catering and mechanical operations. Nine days later, staff working in administrative and passenger service areas reported symptoms. There were no cases among participants working in electricity, carpentry, or medical and spa services. The authors note that the risk of SARS-CoV-2 infection was associated with working in mechanical operations (OR 8.26, 95% CI 1. 54-44.16) and catering services (6.06, 1.78-20.67). Sharing a cabin with an infected case was significantly associated with an increased risk of infection in crude analysis (7.20, 2.48-20.41), but was found to be non-significant in the full model (3.27, 0.97-11.07) [34].
On the Costa Atlantica ship, the infection was probably introduced by a crew member from the entertainment occupation group, which then spread widely inside the vessel, regardless of occupational group or location of the crew cabin. The cases were similarly distributed across occupation types. Also, the cabin rooms of crewmembers presenting with a fever were widely distributed throughout the ship [36].

Use of Masks
Three studies reported the use of masks on the DP after quarantine started [23,25,30]. The crew used surgical/N95 masks [30]. None of the authors specified if a "fit test" was performed to evaluate if the mask fit and sealed properly.
Organized by deck and section, passengers were allowed a 60-min daily period on an exterior deck. During this time, they were required to wear masks, not touch anything, and maintain a one-meter distance from other people. Meanwhile, they were observed by monitors. After each group, the areas were disinfected [25]. Ten studies do not provide any information on masking [7,17,[19][20][21]24,26,27,29,31]. On international ships, one study reported using masks after the quarantine was issued [22]. Also, masks were required for all people onboard indoors, in spaces of shared use (including excursions), except when eating, drinking, or staying in the cabins [33].

Genome Sequencing (GS) and Phylogenetic Analysis
Five studies on the DP outbreak [19,24,26,27,37] and one on the passengers from the Nile River cruises [28] performed GS and phylogenetic analysis. The methods used for performing these investigations were essentially similar across studies (Supplementary Material File S9).
Sekizuka et al. [27] analyzed 70 whole-genome sequences obtained from RT-qPCR positive samples. These sequences and three additional DP isolates from the Global Initiative on Sharing All Influenza Data (GISAID) were compared with the Wuhan-Hu-1 genome sequence. The frequencies of single nucleotide variations (SNVs) indicated that all 73 isolates shared an SNV (52 SNVs in DP isolates vs. 449 SNVs in all the isolates, including GISAID entries) [27] (Supplementary Material File S9).
The DP-A cluster was predominant (29 isolates), suggesting that it was the ancestral haplotype for subsequent transmission. Although further spreading may have been prevented by quarantine, some of the subsequent progeny clusters, as well as DP-B (five isolates) and DP-C (six isolates), were probably formed via transmission through other links, such as mass gatherings in recreational areas and direct transmission among cabinmates. In addition, 33 patients (45%) not included in the DP-A, -B, or -C clusters had unique SARS-CoV-2 haplotypes and patient-specific unique SNVs and/or deletions [27], suggesting there may have been multiple introductions of different strains on the DP. All whole-genome sequence data were deposited to GISAID [27].
Murata et al. [24] performed GS of four sequential specimens collected from one infected person (Carrier_1) and nine specimens obtained from his cabinmate and six others. All SARS-CoV-2 strains belonged to clade 19A, with a single nucleotide mutation (G11083T transversion), as previously described [27], suggesting this strain was transmitted between these cabinmates and the six others. The GS analysis of consecutive samples of Carrier_1, who shed infectious virus for 15 days, identified the emergence of two novel SNVs (C8626T transition and C18452T transition) in the sample collected on day 15. None of these mutations were found in samples collected from the cabinmate of Carrier_1 and other cases [24].
Plucinski et al. [26] reported data on GS from samples obtained from 28 individuals that tested positive after repatriation. All genome sequences clustered in the B group of the global phylogenetic tree, containing all the genome sequences reported previously on the DP. [26]. All genome sequences presented the same mutation reported in the assumed Hong Kong index case [26].
Twelve sequences were from six pairs of close contacts. In all instances, pairs of linked genomes grouped closely within the haplotype network. The linked sequences were separated by zero to two SNVs compared with zero to nine SNVs among all sequences from the DP [26].
Another study [19] retrieved all publicly available SARS-CoV-2 genome sequences with clinical information from the GISAID database up to 7 August 2020 (n = 78,448). The phylodynamic analysis of 67 sequences collected between 10-17 February 2020 estimated that the outbreak originated on 21 January, coinciding with the boarding of the presumed index case from Hong Kong. The affected population size increased around 30 January and exponentially surged from 2 February, before the quarantine. After quarantine, the transmission of the virus slowly continued [19]. Although branch bootstrap values were low, all sequences from the DP clustered with some isolates reported in other countries [19].
Yeh et al. [37] analyzed the evolution dynamics of SARS-CoV-2 in 28 cases from the DP. They identified 24 new viral mutations across 64.2% (18/28) of samples; the virus evolved into at least five subgroups. Based on their findings on the limited number of cases they analyzed, these authors suggested that the outbreak most likely originated from either a single person or simultaneously with another primary case infected with a virus containing the G11083T mutation [37].
On the Costa Atlantica ship, the authors analyzed all samples with high viral titers (Ct < 30 by RT-qPCR), obtaining 94 complete GS [32]. The strains showed three main clusters; the CA-A cluster was genetically closest to the reference strain (Wuhan-Hu-1), possibly indicating that it was the haplotype initially introduced into the CA cluster. CA-A was not a large cluster; only two infected persons perhaps had a central role in spreading SARS-CoV-2 in the CA-A cluster. The core populations of clusters CA-B and CA-C comprised of more than ten individuals, indicating that superspreading event-like infections caused these clusters [32].
Sekizuka et al. [28] reported data on GS from five SARS-CoV-2-positive passengers from two Nile River cruises. Three passengers aboard the same ship presented identical SARS-CoV-2 genome sequences, with a close lineage to isolates from Europe. In addition, a couple of passengers boarded a different cruise ship, but had the same Tokyo to Cairo flight as one of the travelers reported above. These two SARS-CoV-2 isolates showed identical GS, but differed from the genome sequences of the first three travelers by only one SNV [28]. The authors compared the GS of the passengers with the only two genome sequences of SARS-CoV-2 isolates in Egypt available in GISAID at the time of the study. The haplotype network exhibited that one of the first cruise passengers and the two passengers from the second cruise was closely related to isolates from Egypt, with only two or three differences in SNVs [28].

Viral Cultures
Two studies performed viral cultures [17,24] (Table 1; Supplementary Material Files S8 and S9). Murata et al. [24] analyzed 116 PCR-positive samples and 50 PCR-negative samples. The median Ct value of culture-positive samples was 24.6 (IQR, 20.4-25.8; range, 17.9-30.3) vs. culture-negative samples (Ct 35.9). SARS-CoV-2 was successfully cultured from nine (7.8%) PCR-positive samples obtained from seven carriers; none of the PCRnegative samples presented cytopathic effects (CPE) [24]. A specimen from a 70-year-old woman with a medical history of diabetes mellitus and hypertension, who had prolonged RT-PCR positivity for >21 days, was found to have CPE on culture after 15 days. This was confirmed using PCR, following the initial positive PCR test [24], but the result may be questionable given the Ct values and passage of time.
The second study [17] reported on cell cultures of a 62-year-old woman with comorbidities. She presented with two negative RT-PCR tests before repatriation, but tested positive upon arrival to Israel (Ct = 24). The nasal and throat swabs sampled four days later showed a notable CPE on Vero E6 cell culture, but no data was provided on the methods used for cultures [17].

Summary of Main Findings
We identified 23 studies assessing SARS-CoV-2 transmission aboard cruise ships, two environmental studies, and one systematic review. The findings suggested lower ARs in ships with a higher number of available cabins, with cabins more spread out over various decks, and in settings with a lower passenger-to-space ratio. However, the duration at sea did not appear to influence the AR [6]. In addition, a consistent dose-response relationship was found in multiple studies, demonstrating that as the number of passengers in a cabin decreased, ARs decreased [25,26,35], with the highest ARs in individuals staying in four-person cabins (30-63%) and the lowest (8-18%) in singleperson cabins [25] or without infected cabinmates [26]. The AR for those sharing a cabin with an asymptomatic infected cabinmate was lower compared with passengers with a symptomatic infected cabinmate [26]. The risk of infection was higher if an individual had close contact with a confirmed case [35]. Environmental samples found no difference in detection proportion between cabins of symptomatic and asymptomatic cases [16], but there were only a limited number of samples and they were taken after the ship was vacated, making the interpretation of the environmental positives uncertain. A potential common exposure area, with higher infection risk, was suggested to be the elevator hall [31].
Participation in events such as excursions or other group activities was associated with an increased risk of infection [26,30], with the exception of one study that found no statistically significant association in this regard [35].
Epidemiological studies from the DP suggested that passengers and crew members presented symptoms from the beginning of the cruise; the number of suspected cases (with symptoms or close contacts) remained low for two weeks, followed by a notable increase [30]. The infected passengers were distributed across different decks, without any identifiable aggregation or large-scale clustering by deck or zone [25]. The effective population size began to increase around the 10th day of the voyage, surging exponentially from the 13th day. After quarantine, the infection transmission continued more slowly, based on one interpretation [19]. The first crew member that tested positive from the DP presented symptoms ten days after the first passenger [23,30]. Most crew infections were among food service workers [23], with the highest AR on the deck where most food service workers lived, suggesting that infection spread during ship activities such as parties [23,30].
The role of masks for preventing SARS-CoV-2 transmission aboard cruise ships remains unclear. On the DP, masks were used after the start of quarantine [23,25,30], with ARs reported between 18.8 [31] and 19.18% [26]. On international ships, one study reported that use of masks was associated with an AR of 59% [22]. With the implementation of mitigating measures and pre-disembarkation screening, ARs between the identified close contacts during quarantine was 3% [33]. Lower ARs (i.e., 0.03%) and higher ARs (i.e., 75.12%) were also reported on some ships [6], but the role of masks as a mitigating measure was not assessed. On the Costa Atlantica, with only crew on board, the AR was 24% [36]. When probable cases were included (with symptoms indicative of COVID-19 but a negative test result), the AR was 41% [36].
Researchers reported the possibility of virus transmission from asymptomatic, presymptomatic, or symptomatic individuals. Nonetheless, a significant limitation of all studies was the possibility of "asymptomatic" index cases transmitting the infection and "asymptomatic" secondary cases not being investigated due to a lack of fever or respiratory symptoms. Although these are part of the COVID-19 symptom complex, using these symptoms alone would have decreased sensitivity and grossly overestimated the number of asymptomatic cases. In addition, failure to identify a common starting time for exposure or illness may have led to systematic misclassification and failure to identify the existence of multiple index cases leading to biased AR estimates. Furthermore, the number of studies reporting Ct of RT-PCR was limited; therefore, case ascertainments are likely biased [14]. Also, the timeline of sample collections after disembarking may have led to bias if there were contacts with others after leaving the ship.
The GS studies [19,24,[26][27][28]32] reported high-quality evidence supporting transmission of SARS-CoV-2 on cruise ships. Some studies suggested that, on the DP, there was a single starting source of infection with the virus bearing a G11083T transversion mutation. They also provided evidence of zero to two mutations per genome being the norm when comparing the virus in the putative source with those in the target cases. Although Hoshino et al. [19] found most isolates were very similar, based on only a two to three SNV difference, some strains clustered with strains from other countries, so it was difficult to rule out introductions of other strains from other passengers. In addition, although Sekizuka et al. [27] reported all their analyzed strains had the G11083T mutation, there were up to 449 SNV differences between all isolates, and 33 patients had uniquely different sequences despite having the mutation, making it difficult to rule out differing virus strain introductions. Whether there was a single or multiple isolate entries to the DP from multiple passengers remains uncertain, but the weight of all the evidence supports multiple entries [23,37]. Also, the number of mutations in the DP cluster was remarkably lower than that of the Costa Atlantica cluster, indicating that there were some environmental differences between the two cruise ships [32]. In addition, although GS methods may provide reliable phylogenetic insights into the relationship between the putative index and secondary cases, using GS databases to ascertain transmission may induce bias if the number of published sequences is limited [39]. Missing data may also induce bias in phylogenetic analyses, and substantial gaps in global sequencing data may hinder the accurate recognition of an infection source.
The positive results of viral cultures [17,24] provide further evidence of transmission of SARS-CoV-2 aboard cruise ships, indicating an infectious virus with the potential for transmission to other individuals was present. The chain of transmission to secondary cases was well-documented by evidence confirming that the index case was contaminated (i.e., low Ct values) with an infectious virus and confirmed by GS. In the environmental study, no viable virus could be isolated [16]. However, the high Cq values in most positive samples likely explain the negative viral cultures [16]. Viral cultures were not performed on the index cases [17,24], but they provided important insights on SARS-CoV-2 transmission. Murata et al. [24] demonstrated "asymptomatic" carriers on the DP [40].
The possibility of alternative exposures deserves attention. Common sites of alternative exposures include sites before embarkation (i.e., waiting lines), during the cruise (i.e., tours), and after disembarkation (i.e., lining up to exit the ship, checking documents, and traveling to the final destination).
One study found that 4/7 crew members from the DP presented with bacterial coinfections [20]. Also, another study investigating 896 RT-PCR samples from the DP reported that 43/65 (66.15%) symptomatic cases had a negative RT-PCR [27]. Nonetheless, only two studies [20,35] presented data on investigations aiming to detect a possible co-infection in SARS-CoV-2 positive cases or ran any additional laboratory tests to investigate the cause of acute respiratory illness (ARI) in individuals with a negative RT-PCR. In Australian citizens [35], ARI was reported as 62/196 (31.6%) for those with a negative SARS-CoV-2 test, and 52/62 (84%) of these individuals could identify their onset date; 37 individuals reported symptom-onset dates on board whereas 15 reported symptom-onset dates after quarantine in Australia. Five individuals were Influenza A positive, including one case who tested Influenza-A-positive and a fortnight later SARS-CoV-2-positive. ARI symptoms were reported starting from the first day on the DP in three passengers. Seventy-one percent (25/35) of symptomatic COVID-19 cases reported experiencing fever (≥37.5 • C), in contrast to 23% (14/62) of individuals with non-specific ARI (p < 0.001) [35]. ARI symptoms may be present in COVID-19 cases, but they are also found in other viral or bacterial infections, including Legionellosis. Furthermore, Legionella co-infection has been reported in cruise ship passengers with COVID-19 [41]. ARI accounted for up to 29% of recorded illnesses on cruise ships [3,42], and in many cases, it was due to respiratory viral infections [43,44]. One surveillance project reported that 83% of crew members and passengers with ARI tested positive for at least one respiratory virus; 71% had Influenza A or B virus. Over three years, 13 respiratory viruses were identified, including influenza, human rhinovirus, human metapneumovirus, parainfluenza, and adenovirus C, with nine different co-infections [44]. In COVID-19 patients, an undetected co-infection could lead to biased results in fatality rate or hospitalization. Also, excluding other pathogens as a cause of symptoms is important. An incomplete investigation may cause mimicry bias, leading to false conclusions about the causes of the disease of interest.
Further doubts about the validity of the overall findings are raised by the variability in contact tracing strategies, contact tracing timelines, proportion of passengers and crew that were traced successfully, use of distinct case definitions, testing strategy, and case ascertainment.
Only one study in the present review showed evidence of both positive virus cultures as well as genomic evidence [24]. Definitive route(s) of transmission on cruise ships need further investigation. On the DP, the possibility of long-range airborne transmission could not be ruled out at the time of the initial outbreak, but is now considered unlikely given the dose-response gradient with higher numbers among cabinmates and the fact that internal air recirculation was stopped [30]. Evidence from the available studies suggests that transmission on the DP was associated with close proximity [35] and potentially with common source exposure events [25,30]. More recent viral culture research reports significant amounts of infectious SARS-CoV-2 in the environment [45][46][47]. Although no live virus was found on cruise ships on fomites, a recent systematic review focusing on high-quality studies found that replication-competent SARS-CoV-2 was present on fomites [47]. Replication-competent SARS-CoV-2 is significantly more likely when the PCR Ct for clinical specimens and fomite samples is <30. However, the timing of sample collection from symptom onset markedly influences the probability of obtaining positive viral culture results [47].
Our review did not compare risks between cruise ships and other similar settings (e.g., other types of ships). However, because cruise ships often have high-risk senior passengers and may offer comprehensive medical services (e.g., oxygen therapy and dialysis units), they could present infection control problems similar with those in nursing homes. Multiple studies have demonstrated very high transmission rates of other viruses on cruise ships including norovirus and other respiratory viruses [48,49].
To date, only one systematic review assessed the evidence for transmission of SARS-CoV-2 aboard ships [38]. However, the literature search went up until July 2020. The authors included 37 studies on cruise, navy, and cargo ships, but many of the included studies were not relevant for SARS-CoV-2 transmission (i.e., case reports on conjunctivitis or clinical aspects of pneumonia among ship passengers or crew). Furthermore, they did not include environmental studies and did not formally assess study quality.

Strengths and Limitations of the Review
We performed an extensive literature search, accounted for the quality of included studies, and reported all relevant outcomes, including GS and viral cultures. We included results from one non-peer-reviewed study [29], which may affect the reliability of the results. However, due to the ongoing pandemic, such studies could potentially be of research benefit. Our review is one of the most comprehensive ones to date. By focusing on higher-quality studies, we were able to draw meaningful interpretations, which facilitates our understanding of the transmission aboard cruise ships.
Limitations of the present review are mostly related to the low quality of included studies and the fact that different studies provided different data on the same outbreak with inconsistent results. In addition, data extraction was challenging due to missing, incomplete, or unclear descriptions of investigations. Also, most primary studies investigated only some aspects of the outbreak (e.g., spatial distribution, a subgroup of individuals from a cruise, or the early days of the outbreak).
Various reasons may explain the low quality of the published literature. There is a lack of standardized methodology and clear reporting criteria, with substantial methodological variation in SARS-CoV-2 transmission studies [40]. Nonetheless, similar to other studies on SARS-CoV-2 transmission in other closed/semi-closed settings, in times of a pandemic, the opportunities for rigorous studies that trace, interview, and test hundreds or thousands of individuals are challenging and often lacking [50,51]. The likelihood of case ascertainment bias based on symptoms was likely higher at the time of the initial cruise ship outbreaks, given that the full symptom complex of SARS-CoV-2 was underappreciated. Other types of bias were also discussed in previous paragraphs.
That there was a risk of SARS-CoV-2 transmission on cruise ships was evident a few weeks after the DP outbreak. Nonetheless, 2.5 years later, we still lack definitive information on cruise ship transmission modes. Furthermore, only one study was performed after implementing vaccination programs; however, none of the COVID-19 cases were vaccinated [33]. Information about the vaccination status of the rest of the passengers and crew were not available to the authors [33]. We found no studies on more recent variants such as Delta or Omicron.
Therefore, besides its historical value, the present review raises awareness of the paucity of data regarding this topic and the necessity for high-quality research on future cruise ship outbreaks. Otherwise, we will be unable to understand and prevent any similar disease outbreaks with Omicron variants, other future SARS-CoV-2 variants, or other respiratory pathogens.
We did not include lists of public health authorities on SARS-CoV-2 transmission aboard cruise ships. However, we included studies reporting analyses of public lists with retrospectively known cases of SARS-CoV-2 infection aboard ships.

Implications for Practice and Research
Our findings highlight the need for a standardized approach to investigate and report SARS-CoV-2 transmission aboard cruise ships, with possibly a standard international protocol to investigate ship-borne outbreaks. Future research should aim for a thorough epidemiological investigation, a comprehensive evaluation of passengers and crew, with a comprehensive symptom and signs assessment, a rigorous follow-up strategy, and a more robust testing strategy. Factors that may influence transmission should be consistently assessed: pre-embarkation screening strategies; technical specifications of the ship; voyage duration and the number of ports of call; movement and activities (i.e., tours, social activities, drinking or eating, contact with contaminated surfaces, and use of elevators or lavatory in common areas); passenger and crew spacing; onboard screening/surveillance procedures; infectivity of the index case (asymptomatic, pre-symptomatic, or symptomatic, vaccination and immunological status, and mask-wearing or not); the susceptibility of passengers and crew (previous SARS-CoV-2 infection or vaccination and compliance with masks and distancing); and effectiveness of exposure (proximity to the index case and exposure duration).
Future studies should provide Ct values when reporting RT-PCR results and present data on timing and methods of sample collection. Further research, including virus isolation, GS, and phylogenetic analysis, should be conducted to strengthen the current evidence. Consequently, standardization of research reporting should be a priority.
Cruise ships accommodate large numbers of passengers and crew members originating from different countries. In addition, passengers are often older with multiple comorbidities and an increased risk for severe disease and complications. Also, the close quarters, partially enclosed settings, and prolonged contact among individuals increase the risk of infectious disease transmission. The presence of other viral and bacterial causes of ARI in the context of a SARS-CoV-2 outbreak adds additional complexity because COVID-19 may present with similar symptoms. Methods for assessing disease conditions that mimic the inciting cause should be clearly defined, including testing strategies, procedures of isolation, notifications of the authorities, and criteria for returning to work [43]. An integrated syndromic and virologic surveillance could improve the detection and characterization of SARS-CoV-2 outbreaks and other respiratory pathogens [44]. Although expensive, adding a laboratory component to routine cruise ship respiratory surveillance could inform better resource allocation and anticipate needs for cruise ship populations. In addition, it should be considered that some passengers may not have international or travel healthcare insurance coverage. Therefore, costs could restrain passengers from seeking medical care until they present with severe illness. Surveillance of only symptomatic passengers or crew reporting to the ship's infirmary may bias detection of cases to those with more severe symptomatology or those more likely to seek healthcare [44].
Moreover, a SARS-CoV-2 outbreak on a cruise ship has substantial economic and human resource costs for both public health agencies and cruise ships. Additional staff may be needed to implement active and passive surveillance and organize testing; during illness or isolation, crew time is lost, and treatment and hospitalization costs can be expensive. Therefore, the potential benefits should be weighed against the operational limitations of a thorough surveillance program.
Several mitigating measures were introduced to prevent the transmission of SARS-CoV-2 aboard cruise ships. However, the zero-COVID-19 countermeasures presented serious fiscal consequences, and a more practical near zero-risk approach was recently proposed [52]. The latest point of view advocates a holistic mitigating perspective, including behavioral (i.e., social distancing), procedural (i.e., different boarding times), and technical (i.e., testing procedures) measures.
The measures taken at the beginning of the pandemic reflect the difficulties of implementing zero-risk countermeasures, including high costs and logistics. In addition, to implement efficient mitigating measures (e.g., respiratory isolation processes PPE, airflow. Etc.), we first need a thorough understanding of transmission routes and risk factors. Therefore, high-quality research, with at least one study per type of setting and intervention, is required.

Conclusions
Current evidence indicates a definite risk of transmission of SARS-CoV-2 aboard cruise ships, with crowding and multiple persons per cabin being associated with an increased transmission risk. The highest ARs were found in individuals staying in four-person cabins and the lowest in single-person cabins or those without infected cabinmates. However, the currently published data do not allow a conclusive assessment of the risk factors and extent of the transmission. Nonetheless, valuable information may be gleaned from the highest quality studies. We found that the quality of evidence from most published studies was low. The analysis of findings across studies was restricted by variations in study design and methodology. Standardized guidelines for performing and reporting future cruise ship outbreaks should be developed.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/tropicalmed7100290/s1, File S1; Updated Generic Protocol [53][54][55][56][57], File S2; Search strategy, File S3; Flow chart showing the process for inclusion of studies investigating cruise ships transmission, File S4; List of excluded studies, File S5; List of included studies, File S6; USA citizens repatriated from Diamond Princess, File S7: Timeline of events on Diamond Princess cruise ship [58,59], File S8; Characteristics of included studies, File S9; Genomic sequencing studies, File S10; Viral cultures. Conflicts of Interest: T.J. Competing interests are available at https://restoringtrials.org/competinginterests-tom-jefferson/ (accessed on 9 July 2022). C.H. holds grant funding from the NIHR, the NIHR School of Primary Care Research, the NIHR BRC Oxford, and the World Health Organization for a series of Living rapid reviews on the modes of transmission of SARS-CoV-2 (WHO Registration 2020/1077093). He has received financial remuneration from an asbestos case and given legal advice on mesh and hormone pregnancy tests cases. He has received expenses and fees for his media work, including occasional payments from BBC Radio 4 Inside Health and The Spectator. He receives expenses for teaching EBM and is also paid for his GP work in NHS out of hours (contract Oxford Health NHS Foundation Trust). He has also received income from the publication of a series of toolkit books and for appraising treatment recommendations in non-NHS settings. He is the Director of CEBM and an NIHR Senior Investigator. D.E. holds grant funding from the Canadian Institutes for Health Research and Li Ka Shing Institute of Virology relating to the development of COVID-19 vaccines, as well as from the Canadian Natural Science and Engineering Research Council concerning COVID-19 aerosol transmission. He is a recipient of funding from the World Health Organization and Province of Alberta, which supports the provision of BSL3-based SARS-CoV-2 culture services to regional investigators. He also holds public and private sector contract funding relating to the development of poxvirus-based COVID-19 vaccines, SARS-CoV-2-inactivation technologies, and serum neutralization testing. J.M.C. holds grants from the Canadian Institutes for Health Research on acute and primary care preparedness for COVID-19 in Alberta, Canada and was the primary Local Investigator for a Staphylococcus aureus vaccine study funded by Pfizer for which all funding was provided only to the University of Calgary. He is co-investigator on a WHO-funded study using integrated human factors and ethnography approaches to identify and scale innovative IPC guidance implementation supports in primary care with a focus on low-resource settings and using drone aerial systems to deliver medical supplies and PPE to remote First Nations communities during the COVID-19 pandemic. He also received support from the Centers for Disease Control and Prevention (CDC) to attend an Infection Control Think Tank Meeting. He is a member and Chair of the WHO Infection Prevention and Control Research and Development Expert Group for COVID-19 and a member of the WHO Health Emergencies Programme (WHE) Ad-hoc COVID-19 IPC Guidance Development Group, both of which provide multidisciplinary advice to the WHO, for which no funding was received and from which no funding recommendations were made for any WHO contracts or grants. He is also a member of the Cochrane Acute Respiratory Infections Working Group. J.B. is a major shareholder in the Trip Database search engine (www.tripdatabase.com), as well as an employee there. In relation to this work, Trip has worked with a large number of organizations over the years, but none have any links with this work. Their main current projects are with AXA and Collateral Global. He worked on Living rapid literature review on the modes of transmission of SARS-CoV-2 (WHO Registration 2020/1077093-0) and is part of the review group carrying out a scoping review of systematic reviews and meta-analyses of interventions designed to improve vaccination uptake (WHO Registration 2021/1138353-0). A.P. holds grants from the NIHR School of Primary Care Research. E.C.R., I.J.O., and E.A.S. have no conflicting interests to disclose.