It is well known that the serological response to the infection with a given influenza subtype is influenced by the extent of cross-reaction with heterologous subtypes, previous exposure to related strains, individual variability and age. In addition, the measure of antibody response is prone to variability due to the experimental procedures, such as type of red blood cells or virus preparation. Several studies have considered a HAI titer in the range of ~1
40 for their conclusions, as the scope was to establish the level of pre-existing protective immunity, while only few studies reports lower thresholds (e.g.1
10), as indicators of previous exposure to related strains of influenza virus. Considering the protective threshold, in Italy the pre-pandemic frequency of antibodies against A/H1N1pdm in the general population is in the range of 6–7% individuals aged 0–55 years, and increases to 22% in over 65 
. In UK, a protective titer has been detected in 9.8% of individuals aged 25–49 
, while 18.1% shows antibodies titered 1
8 in the pre-pandemic era. In another study, conducted in Finland, 0.8% of individuals aged 20–39 show protective antibodies, and 2.5% showed titers of 1
. The pre-pandemic prevalence of HAI antibodies in military personnel was 15% with at 1
10 and 9.4% at 1
. In residual hospital serum samples from persons aged 18–24 years the pre-pandemic seroprevalence (1
40) was 6% 
The aim of the present study was to establish retrospectively the rate of infection in a close community for which pre-infection serum samples were not available. The combined serological approach, based on the contemporary presence of HAI and of CF antibodies at a titer ≥1
10, was established on the basis of a preliminary analysis conducted in patients who seroconverted by definition (infected patient's population, group 1), and was validated in three sentinel groups, representing, respectively, age matched male individuals sampled before the pandemic (group 2), at the time of the cruise end (group 3) and after the 2009 pandemic influenza wave (group 4). In fact, the low HAI titer threshold allowed us to maximize the sensitivity, and the contemporary presence of CF was used to increase the specificity of our infection criterion, as this type of antibodies is commonly recognized as indicator of recent infection 
On the whole, the data obtained on a representative adult population matching the demographical characteristics of our study population indicates that, indeed, the rate of serological positivity against A/H1N1pdm in the pre-pandemic era is about 10%, confirming previous literature data.
In the time frame overlapping with that considered in the study, this proportion do not significantly increase in the control population.
Thus, assuming a similar background of 10% of seropositivity in the study population, we may infer that about 30% of the sailors actually had acquired the infection during the cruise. The association of the positivity rate with crowding of living quarters supports this hypothesis.
This estimate is in keeping with that from a study conducted in military personnel from Singapore (29.4%), based on seroconversion rate over a 3 months period during 2009 
In the Scirocco ship, only two of the seropositives showed clinical signs consistent with the case definition, while about half showed mild symptoms and the remaining half remained fully asymptomatic during the cruise.
These results are in apparent contradiction with those from other shipborne outbreaks, where higher proportion of febrile infections has been reported 
. The reasons for such differences are unclear, although some discrepancies in clinical surveillance systems may be at least in part accounting for. Nevertheless, our results suggest that the introduction influenza A/H1N1pdm in a close community may result in a small number of clinically relevant diseases in spite of a wide spread, detected by serological investigation.
Our data also suggest that previous vaccination against seasonal viruses was not protective against A/H1N1pdm infection, in agreement with previous reports performed in similar settings 
The present study presents some limitation. First, pre-pandemic serum samples of the crewmen were not available; therefore our estimate of the baseline seroprevalence is only speculative, leading to a possibly imprecise estimate of proportion of peoples who contracted the infection during the cruise. Second, according to the surveillance protocol, afebrile or low grade fever cases were not tested by PCR, so it is not possible to demonstrate that cases occurring in patients who tested seropositive at the end of cruise were actually attributable to influenza A/H\N1pdm infection.
The results of the present study may be relevant for planning public health strategies in the context of early pandemic. In fact, clinical surveillance criteria established during the early phases of an evolving pandemic may prove inadequate to monitor the actual spread and the severity of the phenomenon, and need timely update to provide a realistic estimate. In this context, serosurveillance data, particularly from semi-closed communities, may be crucial in order to timely define the real spectrum of clinical presentation and the possible public health impact, essential to identify and implement adequate control measures. As final consideration, biorepositories may represent a valuable resource to help define the pre-pandemic population immunity and to monitor the changes in the sero-prevalence, providing unbiased collection of samples supplied with demographic, epidemiological and clinical information.