In this population-based study performed in the Netherlands, we found that the level of knowledge regarding Influenza A (H1N1) increased between May and August 2009. At the same time, perceived severity of the new flu, perceived self-efficacy, and intention to comply with preventive measures decreased. The perceived reliability of information from the government also decreased from May to August. Feelings of anxiety decreased from May to June, and remained stable afterwards. From June to August 2009, perceived vulnerability increased and more respondents took preventive measures. Factors associated with taking preventive measures included no children in the household, high anxiety, high self-efficacy, agreeing with avoidance statements, and paying much attention to media information regarding Influenza A (H1N1). Having a strong intention to comply with government-advised preventive measures in the future was associated with older age, high perceived severity, high anxiety, high perceived efficacy of measures, high self-efficacy, and finding governmental information to be reliable.
A clear strength of this study is that data collection took place during the 2009 Influenza A (H1N1) pandemic, in contrast to other studies performed at times when pandemic influenza was not regarded as a high threat and scenarios were based on hypothetical situations [12–19]. Another strength is that this study consists of three repeated survey rounds, enabling analysis of trends over time. This is in contrast to other recent studies, which consisted of a single cross-sectional survey [20–22]. Moreover, we followed-up individuals, guaranteeing that differences between survey rounds were not due to differences between study populations, but represent real trends over time . Finally, we used an online questionnaire, which creates less social desirability bias than personal telephone interviews. The use of an Internet panel led to high response rates: 59%, 63%, and 79% in survey 1, survey 2, and survey 3, respectively. Our study also has several limitations. First, the Internet panel members who responded to our online questionnaire were not fully representative of the general Dutch population. In our study, participants were more likely to be in the older age group (> 50 years) (52% versus 44%), of Dutch ethnicity (92% versus 80%), and unemployed/retired (43% versus 24%). We cross tabulated all the measured constructs by age group (18-49 years/> 50 years), employment status (employed/unemployed) and ethnicity (Dutch/non-Dutch) (data not shown). For these constructs, there were no differences between the Dutch and non-Dutch participants. Among both the older and unemployed the perceived efficacy, self efficacy and intention to comply with measures were significantly higher, and they more agreed with statements on avoidance. Perceived vulnerability and reliability of governmental information were lower among both the older and unemployed. Furthermore, the older age group paid more attention to the information of the government. Among the unemployed the perceived severity was higher and they less agreed with the underestimation statements. This population difference may have led to a substantial bias in the absolute outcomes of Table 2, but only to a small bias in the trends over time or in the predictors of behavioural responses. Second, in the logistic regression analyses we may have lost some power, because we used dichotomized summary scales as predictors. However, we have performed additional regression analyses with the predictors as continuous variables, and found minimal differences (data not shown). Third, the validity of the questionnaire used in this study was not tested through a test-retest design, because the Influenza pandemic was ongoing and thus perceptions were not stable over time. Fourth, no data were obtained from non-responders.
This is the first national study to evaluate perceived risk, feelings of anxiety, and behavioural responses regarding Influenza A (H1N1) among the general public in the Netherlands. There was a decrease over time in perceived severity, anxiety and intention to comply with preventive measures. Initially, representatives of (inter)national health institutes predicted a worse-case scenario with large numbers of fatal cases, based on influenza pandemics in the past and early reports concerning the new Influenza virus . In the following months, media attention decreased considerably, local viral transmission remained relatively limited in the Netherlands, and the Dutch government announced that the pandemic appeared to be mild [37, 38]. Decreasing trends over time in perceived severity and anxiety are consistent with the reality: the clinical picture of influenza turned out to be mild in course of time. The decrease in perceived reliability of information from the government was not surprisingly; in the beginning the general public believed the pandemic would be severe as pronounced by the government, but this turned out to be mild. This decrease in perceived reliability of governmental information was not alarming and did not result in more feelings of anxiety or in a lower intention to comply with measures. The increase in perceived vulnerability and number of individuals taking preventive measures may be an effect of the increasing number of Influenza A (H1N1) infected cases, including the first fatal case in The Netherlands in August 2009. Previous studies showed a similar effect. For instance with the inclining phase of the SARS outbreak in 2003, the prevalence of wearing a face mask and adopting better hand hygiene increased dramatically when the number of SARS cases increased . During the current study period, there was no official recommendation from the Dutch government to take preventive measures; the government was in the process of preparing a national information campaign called 'Fight the flu'. This campaign was launched at the end of August 2009, and included announcements on television and a leaflet which was sent to every home in the country providing information about what people can do to prevent themselves and others. So, at the moment of the third data collection period the government had not yet actively informed the general public about preventive measures. For this reason, respondents were not only asked about preventive measures they had taken, but also about their intention to comply with government-advised preventive measures in the near future. People who took preventive measures during this 'pre-phase' of governmental advice were very alert to media information and seemed to be practicing preventive measures based on emotions such as anxiety. This is in line with results of the study conducted by Jones et al.  concluding that affective variables, such as self-reported anxiety over the epidemic, mediate the likelihood that respondents engage in protective behaviour. Rubin et al.  also found a significant association between anxiety and carrying out recommended behaviours. Similarly, studies on outbreaks of SARS found that anxiety was associated with taking preventive measures [39, 40]. To date, there are only few published studies assessing factors that might explain compliance with preventive behaviours in case of an Influenza pandemic. Comparison with these studies is difficult because of differences in phrasing of questionnaire items and methods of analysis. Barr et al.  collected baseline data about willingness to comply with vaccination, isolation, and wearing a face mask among Australians during a hypothetical influenza pandemic, and found a higher level of willingness to comply among people with higher levels of threat perception and among those of older age. This is in agreement with our findings, where intention to comply with measures was also associated with older age and high perceived severity.
This is one of the first studies conducted during the course of the Influenza pandemic. Additional studies on risk perception among the public are needed to further understand the field of preventive behaviour as related to control of infectious diseases. Furthermore, these studies need to address emotional aspects such as anxiety, uncertainty, or embarrassment that play a role in decision making. Finally, research regarding the translation of results from the above-suggested studies into risk communication is of utmost importance.
Our study has several implications for health authorities and public health policy. In case of an emerging infectious disease, as Influenza A (H1N1), it is very difficult to predict the further course of the outbreak. It is important that health authorities present a range of scenarios, not only worst-case but also other, more positive, scenarios. In the beginning of an outbreak, there are many uncertainties about the infectiousness and case fatality rate of the disease. Health authorities should not only communicate with the public about 'what is known' (the certainties), but they should also communicate about 'what is not known' (the uncertainties). In course of the outbreak, when more information becomes available, public health authorities should update their messages to achieve effective risk communication. This is essential not only to instruct and motivate the public to take preventive measures, but also to build trust in public health authorities and prevent misconceptions. Besides rational arguments (such as perceived severity and efficacy of measures), emotional aspects like anxiety play a role in decision making concerning preventive behaviour. Health authorities should acknowledge these emotional aspects and take these arguments into account in their risk communication with the general public.