Outbreaks of Pandemic (H1N1) 2009 and Seasonal Influenza A (H3N2) on Cruise Ship

Although pandemic virus spread rapidly, especially among children, intensive control measures successfully contained these outbreaks.

D uring April 2009, pandemic (H1N1) 2009 (pandemic infl uenza) virus began to circulate worldwide. In Australia, public health efforts were initially focused on delaying the entry of the virus into the country. By May 24, 2009, a total of 14 cases had been identifi ed nationally, 2 in New South Wales (NSW), and all were associated with international travel.
On May 24, the Australian Quarantine Inspection Service reported that 6 passengers of a cruise ship had respiratory symptoms, and a point-of-care test showed positive infl uenza A virus results for all. The ship had departed from Sydney on a 10-day cruise in the Pacifi c Ocean on May 16 (cruise A) and stopped at 2 islands, neither of which had reported circulation of pandemic infl uenza virus. None of the sick passengers had been in countries known to be affected by this infl uenza strain in the week before boarding. Thus, with no reason to suspect that the pandemic strain was circulating on board, passengers were allowed to disembark in Sydney on May 25.
On May 25, the 4 available respiratory samples taken from sick passengers were quickly couriered to the South Eastern Area Laboratory Service (the major public health viral laboratory serving eastern Sydney) for infl uenza virus nucleic acid testing (NAT) by real-time reverse transcription -PCR (RT-PCR). Of these 4 samples, 2 were positive for pandemic infl uenza virus and 2 were positive for infl uenza A (H3N2) (seasonal infl uenza) virus.
In response, NSW Health requested that all passengers (1,963 from Australia and 7 from elsewhere) who were experiencing infl uenza-like illness (ILI) isolate themselves from healthy persons and that all asymptomatic passengers quarantine themselves for 7 days after disembarkation (or 7 days after onset of symptoms if they developed). This advice was communicated to passengers on the day of disembarkation through media alerts, the NSW Health website, and telephone information lines. Subsequently, passengers were contacted by telephone to ensure that they understood containment measures (how to prevent virus spread). Oseltamivir treatment (75 mg 2×/d for 5 days) was recommended for passengers or crew members with ILI (defi ned as >2 of the following: cough, fever, runny nose, or blocked nose) within 48 hours of onset and oseltamivir prophylaxis Outbreaks of Pandemic (H1N1) 2009 and Seasonal Infl uenza A (H3N2) on Cruise Ship (75 mg 1×/d for 10 days) for those in close contact with patients with laboratory-confi rmed cases. On May 25, all crew members were assessed for illness. Symptomatic members were isolated on shore, and the rest were given oseltamivir prophylaxis and continued to serve on the ship's next voyage (cruise B), which departed later the same day. Cruise B traveled along the northern coast of Australia for 7 days and made a short stop at Brisbane before returning to Sydney on June 1. To minimize the risk for infection, enhanced cleaning regimens were conducted before cruise B, and NSW Health sent a public health doctor on the cruise to conduct intense surveillance for symptomatic passengers and crew.
Outbreaks of infl uenza have previously been reported on cruise ships (1-6), but the circumstances and extent of transmission have not been well documented. The cocirculation of pandemic and seasonal infl uenza viruses on cruise ship A provided a unique opportunity to compare symptoms, severity, and attack rates of pandemic and seasonal strains. We describe our outbreak investigation, compare the epidemiology of the 2 infl uenza virus subtypes, and explore effectiveness of control measures.

Case Defi nition
We defi ned a confi rmed infl uenza A case as illness in a cruise A passenger in whom infl uenza A virus was detected by NAT during the cruise or within 7 days after disembarkation (regardless of symptoms). A case of pandemic infl uenza was defi ned as illness in a person with positive RT-PCR results for that virus. Further subtyping was conducted for 44 of 100 patients with positive infl uenza A but negative pandemic infl uenza virus results by NAT; all had positive results for seasonal infl uenza virus. Consequently, we defi ned a case of seasonal infl uenza as illness in a person with positive infl uenza A virus results by NAT but negative pandemic infl uenza virus results and in whom infl uenza subtyping for seasonal infl uenza virus by RT-PCR either produced positive results or was not conducted. A primary case was defi ned as illness in the fi rst person in a cabin to report ILI symptoms; a co-primary case, as illness in a person who reported symptom onset within 24 hours after a primary case; and a secondary case, as illness in a person whose symptoms developed >24 hours after symptom onset in the primary case-patient. Case-patients were considered infectious for 24 hours before and 7 days after symptom onset. For the childcare center investigation, children who remained asymptomatic throughout the cruise were considered susceptible to infl uenza infection at each childcare session attended. Children in whom ILI developed were considered susceptible before the infectious period began.

Case Detection
We obtained a list of the names, sex, dates of birth, nationality, contact details, and cabin numbers of all passengers and crew members on cruise A. We reviewed the cruise ship's medical records to fi nd passengers who had sought treatment for ILI during cruises A and B. Isolated symptomatic passengers from cruise A were referred to nearby hospitals for testing. Quarantined asymptomatic passengers were asked to report if symptoms developed; if so, laboratory testing was conducted. Crew members and passengers on cruise B were asked to immediately report fever or respiratory symptoms to medical staff and were tested for infl uenza by at least 2 point-of-care tests taken >24 hours apart. In all 8 Australian states and territories, public health legislation requires diagnostic laboratories to report confi rmed infl uenza cases to the jurisdictional health department (7). The names of infl uenza case-patients reported after completion of cruise A were checked against the ship's manifest.

Data Collection
Because the investigation was part of a public health control initiative, formal ethics committee review was not required. Experienced public health staff interviewed casepatients at the time of diagnosis and used a standardized questionnaire to determine symptoms, hospitalization status, and oseltamivir use. This information was entered into a statewide database. Passengers who shared a cabin with case-patients who had pandemic infl uenza were also interviewed about respiratory symptoms. Laboratory testing initially focused on identifying pandemic infl uenza cases by using the specifi c RT-PCR; samples determined negative for pandemic infl uenza virus by NAT were tested for infl uenza A (including seasonal infl uenza virus) several weeks after passengers had disembarked.
Approximately 6 weeks after disembarking, all 50 passengers who had had pandemic infl uenza were reinterviewed about the duration and severity of their illness. These passengers included 3 interstate residents who had been treated in NSW (and excluded 28 non-NSW casepatients as a convenience sample) and the 45 NSW casepatients who had seasonal infl uenza (excluding 55 non-NSW case-patients and 17 NSW case-patients for whom test results were not available at the time of interview). Ultimately, 62 cases of seasonal infl uenza were identifi ed among NSW passengers; complete symptom data from 50 passengers who were interviewed at the time of diagnosis were recorded in the statewide database.

Childcare Center Investigation
On-board childcare activities were provided in 3 daily sessions (9:00 AM -10:00 PM) in 3 areas of the ship for 3 age groups: 3-6, 7-12, and >13 years of age. Because the pandemic outbreak appeared to begin in and primarily affect children 3-6 years of age, the epidemiologic investigation focused on this group. Most childcare activities for this group took place in 1 room. We examined childcare attendance records for this group and, ≈6 weeks after disembarkation, interviewed the parents of all children in this group about symptoms, vaccination history, and composition of the traveling group. All specimens collected from childcare attendees were tested for pandemic and seasonal infl uenza subtypes.

Compliance Assessment
To assess compliance with isolation and quarantine recommendations, we interviewed all 66 households in which at least 1 person with pandemic infl uenza was isolated, 32 NSW households with at least 1 person with seasonal infl uenza, and 45 randomly selected quarantined NSW passengers. (NSW passengers were selected as a convenience sample.) Interviews were conducted by experienced public health interviewers who used a standardized questionnaire.

Laboratory Investigation
NAT detection of pandemic infl uenza virus was performed by using real-time RT-PCR with primers targeting the hemagglutinin gene of the pandemic infl uenza virus provided by the Centers for Disease Control and Prevention and following recommended protocol or by using an in-house pandemic infl uenza virus-specifi c real-time RT-PCR. Seasonal infl uenza virus was identifi ed by using a 2-target RT-PCR containing primers targeting pandemic and seasonal infl uenza virus strains (Unité de Génétique Moléculaire des Virus Respiratoires, Institut Pasteur, Paris, France) or a commercial infl uenza A subtyping assay (Easy-Plex Infl uenza profi le 6; AusDiagnostics, Sydney, NSW, Australia).

Statistical Analyses
We analyzed data by using Epi Info version 3.5.1 (www.cdc.gov/epiinfo). Relative risks were used to compare age (as a categorical variable split into 7 groups), sex, and place of residence. Fisher exact test results were used for cell sizes <5. A Mantel-Haenszel value of p<0.05 was considered signifi cant. χ 2 tests were used to compare proportions. To compare the rates of pandemic and seasonal infl uenza infection in childcare attendees, the number of sessions a child attended while susceptible were summed, and cases per child-sessions at risk and exact Poisson confi dence intervals were calculated. Nonoverlapping confidence intervals were considered signifi cantly different.

Results
A total of 1,970 passengers and 734 crew members were on cruise A. Median age of passengers was 46 years (range 1-94 years), 57% were female, and most were from Australia ( Table 1). Median age of crew members was 31 years (range 19-62 years), and most were born overseas (not in Australia). ILI developed in 13 (0.7%) passengers who sought medical attention during the cruise; and infl uenza A results from point-of-care testing were positive for 6. NAT of samples from persons who were sick during the cruise or during the 7 days after disembarkation showed positive pandemic infl uenza virus results for 76 (3.9%), positive seasonal infl uenza results for 98 (5.0%), and positive co-infection results for 2 (0.1%). ILI in the 7 days before disembarkation was reported by 15 (2.0%) crew members; NAT showed positive pandemic infl uenza results for 3 crew members and positive seasonal infl uenza results for none. These crew members were isolated on shore. The remaining 719 crew members were given oseltamivir prophylaxis and continued to work during cruise B; among these, 5 reported ILI (all within 24 hours of cruise B departing), and 3 had positive NAT results for pandemic infl uenza virus. Therefore, 20 (2.7%) crew members from cruise A reported ILI, and 6 (0.8%) of these had positive pandemic infl uenza test results; none had positive seasonal infl uenza test results. Given the relatively low attack rate for the crew, we focused further investigation on the passengers, among whom the attack rate for pandemic infl uenza was highest for children 3-6 years of age, followed by children 7-12 years of age. For seasonal infl uenza, the attack rate was similar among children in all age groups ( Table 1).

Symptoms and Severity of Illness
In total, 2 (3%) patients with pandemic infl uenza and 8 (8%) patients with seasonal infl uenza were hospitalized (p = 0.16); none died. Among the 50 passengers with the pandemic strain and 50 with the seasonal strain who were interviewed, symptoms were similar, although coryza was reported signifi cantly more often by those with pandemic infl uenza ( Table 2). Duration of illness was similar for passengers with either strain, but a higher proportion of seasonal infl uenza patients reported that illness was severe enough to limit their activities.

Epidemiologic Investigation
According to date of symptom onset, the pandemic infl uenza outbreak began in the childcare center on May 18, which was 2 days after embarkation, and peaked on May 25, the fi nal day of cruise A. The fi rst reported seasonal infl uenza case was in an adult whose symptoms began on May 17; the second, seemingly unrelated, infection developed in a childcare attendee on May 21. The number of seasonal infl uenza cases also peaked on May 25 (Figure).
One child received seasonal infl uenza vaccine in 2008 and no child received the vaccine in 2009. The fi rst case of pandemic infl uenza was in a child from Victoria, Australia, in whom symptoms developed on the third day of cruise A. The child attended childcare for 4 sessions while infectious. After the index case was identifi ed, 19 additional cases of pandemic infl uenza were identifi ed (including in 2 sets of siblings: 2 related children in whom symptoms developed at the same time and in 1 child in whom ILI developed 2 days after symptom onset in her sibling); all but 1 had attended the childcare while a known infectious case-patient was present. The fi rst child for whom seasonal infl uenza was diagnosed had attended childcare the afternoon and evening before symptom onset on May 21 and for 6 sessions while symptomatic. Subsequently, an additional 3 unrelated cases of seasonal infl uenza were identifi ed among childcare attendees. The 3 children who did not attend childcare remained healthy.
Among the 45 childcare attendees, NAT results for pandemic infl uenza were positive for 18, for seasonal infl uenza were positive for 2, and for both were positive for 2. ILI developed in an additional 10 children, but these children had negative infl uenza results by NAT; ILI developed in another 6 children who were not tested. Of these 16 children, 8 had traveling companions with positive pandemic infl uenza virus results and 2 had travelling companions with positive seasonal infl uenza virus results. The remain-ing 7 children remained asymptomatic. Of the 45 children who attended childcare, 44 attended concurrently with an infectious pandemic infl uenza case-patient and 43 attended concurrently with an infectious seasonal infl uenza case-patient. Considering the number of sessions attended by susceptible children, we determined that the risk for pandemic infl uenza infection was signifi cantly higher (19 cases from 344 sessions = 0.055 child-sessions at risk, 95% confi dence interval [CI] 0.033-0.086) than was the risk for seasonal infl uenza (3 cases from 279 sessions = 0.011 child-sessions at risk, 95% CI 0.002-0.031).

Secondary Attack Rates for Pandemic Infl uenza
A total of 66 pandemic infl uenza case-patients in 53 cabins were infectious while on cruise A. Excluding the coprimary case-patients, 91 passengers shared a cabin with an infectious primary case-patient. Of these 91 passengers, symptoms developed in 50 (55%). Of these 50 case-patients, 34 were tested and 12 (35%) had positive pandemic infl uenza results. The secondary attack rate for those <12 years of age (16/21) was signifi cantly higher than for those >12 years of age (34/70) (76% vs. 49%; p = 0.03). Of the 66 case-patients, 1 received oseltamivir treatment within 48 hours of symptom onset. Information about provision of oseltamivir prophylaxis was available for 34 (83%) of 41 asymptomatic contacts. Of these, 3  of age began receiving antiviral drug prophylaxis within 7 days of their fi rst exposure to pandemic infl uenza virus. Of these 20, only 1 received prophylactic drug within 3 days of fi rst exposure to the pandemic strain. Despite being asymptomatic, 11 (27%) of 41 passengers underwent laboratory testing and were negative for pandemic infl uenza virus by NAT.

Isolation and Quarantine
After disembarking, patients with pandemic and seasonal infl uenza were isolated in 149 discrete (family or household-like) groups. Of the 98 (66%) interviewed, 37% reported that they were fi rst made aware of the need for isolation through media reports, 27% by their treating doctor, 26% by public health staff, 6% by the ship's staff, and 5% by fellow passengers. Of the 45 quarantined passengers interviewed, 52% were initially informed of the need for quarantine through media reports, 25% by work or school colleagues, 11% from the ship's staff, 7% from a friend or relative, and 5% from public health staff. All infl uenza case-patients reported that they had obeyed isolation requirements, and 43 of 45 quarantined passengers reported that they had remained in quarantine for 7 days after disembarkation. Of the 2 passengers who did not follow quarantine requirements, 1 reportedly attended work by private vehicle and cancelled all other outings; the other denied knowledge of the requirements.

Further Virus Transmission
Three secondary pandemic infl uenza infections among family contacts of case-patients from cruise A were identifi ed; a subsequent case-patient was identifi ed as a contact of 1 person who had secondary infection. Other than these cases, no evidence of transmission to the community or to passengers of cruise B was found.

Discussion
We identifi ed dual outbreaks of pandemic and seasonal infl uenza among passengers on a cruise ship. Cruise ships provide ideal conditions for rapid spread of respiratory viral illnesses (e.g., many persons living closely together, frequently interacting in enclosed and partially enclosed environments, and often originating from both hemispheres). Although infections spread rapidly among passengers and to some crew members during the cruise, further spread to the community and the next cruise was avoided through intensive disease control measures.
After identifi cation of the outbreak, it became apparent that undetected local transmission of pandemic infl uenza virus was occurring in Victoria before cruise A (8) and that the virus was probably introduced to the ship by the index case-patient from Victoria. The pandemic virus spread rapidly among other childcare attendees and their close contacts and to other passengers and crew. Seasonal infl uenza virus was the predominant infl uenza virus circulating in NSW before the appearance of pandemic infl uenza virus (NSW Health, unpub. data).
The cocirculation of both strains in the childcare center provided a unique opportunity to compare attack rates. The pandemic strain seems to have spread among children more readily than the seasonal strain. This difference in transmissibility could have resulted from innate differences in the viruses themselves or from a level of immunity from past infection with the seasonal strain. Consistent with fi ndings in other studies, the symptoms of pandemic and seasonal infl uenza were similar (9-12). After adjusting for underlying medical conditions, we found that hospitalization rates and activity-limiting effects were higher for case-patients with seasonal than with pandemic infl uenza; however, this fi nding may be explained in part by differences in the agespecifi c attack rates. The secondary attack rate for pandemic infl uenza among cabin contacts of 55% was higher than that reported for household contacts (13,14), despite a small proportion of these persons having received antiviral drug prophylaxis, and may refl ect the close living arrangements in a ship's cabin. The intense passenger follow-up enabled us to assess the sensitivity of the ship's medical clinic for identifying infl uenza cases. Before this outbreak, ships had active containment measures in place to minimize the spread of seasonal infl uenza, including use of point-of-care infl uenza testing for patients seeking treatment for ILI and oseltamivir treatment and isolation to reduce further spread. Our active case-fi nding efforts identifi ed 79 infl uenza cases on cruise A, yet the ship's clinic identifi ed only 6 (8%) of these. Despite enhanced community awareness of the emerging pandemic, the ship's medical clinic staff underestimated the case count by 13-fold. The number of passengers who sought treatment at the ship's medical clinic does not accurately refl ect the extent of the infl uenza outbreak on board, possibly because the decision to seek treatment may have been infl uenced by a number of factors including cost, severity of symptoms, and unwillingness to be isolated while on holiday.
Our investigation had several limitations. First, the case defi nition depended on NAT detection of virus in clinical samples, which may have resulted in misclassifi cation of cases. Second, although the epidemiology is consistent with the fi rst cases of pandemic infl uenza appearing in the childcare center, undetected or asymptomatic infected passengers or crew could have carried the viruses onto the ship. However, this scenario is unlikely because the symptoms developed in the index case-patient 2 days after embarkation. Third, although most ill passengers were interviewed within 2 days after onset of illness, interviews about severity, length of illness, and the experience in isolation and quarantine were conducted some weeks later, introducing possible recall bias. Fourth, although 2 cases of co-infection were detected, only the fi rst 2 pandemic infl uenza-positive specimens from childcare attendees were subtyped for other infl uenza A subtypes; it is possible that some of the remaining pandemic infl uenza case-patients were also infected with seasonal infl uenza. Fifth, some of the remaining pandemic infl uenza case-patients may have been co-infected. Sixth, the secondary attack rate for cabin contacts may be an overestimate because passengers with negative NAT results were not tested for other respiratory infections, and passengers with onset of symptoms >24 hours after symptoms developed in a cabinmate were assumed to be secondary, rather than co-primary cases.
Mathematical modeling suggests that containment of infl uenza is possible if appropriate resources are devoted. In some countries, isolation and quarantine measures have been used in response to severe acute respiratory syndrome (15)(16)(17)(18), but these measures have rarely been used for infl uenza control. In the infl uenza outbreaks reported here, direct follow-up of passengers in isolation and quarantine, supported by intense media coverage, resulted in a high degree of compliance and successful outbreak containment. Additionally, providing oseltamivir prophylaxis for crew members may have contributed to the successful containment of the infection during cruise B. Although the robust application of containment measures can stop the spread of novel infl uenza viruses, public health resource requirements are labor-intensive and expensive and may not be sustainable except for the most virulent of pandemic viruses. Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 16 Date of onset of fi rst symptoms for cruise ship passengers, by infl uenza subtype. Excludes 1 infl uenza A (H3N2) case-patient for whom onset date was unavailable and 1 pandemic (H1N1) 2009 case-patient and 2 infl uenza A (H3N2) case-patients who were asymptomatic but whose laboratory test results were positive.