Varicella is a common childhood disease. The seropositivity increases sharply with age from 18% for both 0- and 1-year-olds, to 49%, 59%, 76% and 93% for 2-, 3-, 4- and 5-year-olds, respectively [1]. Only about 5% of first varicella-zoster-virus (VZV) infections occur asymptomatically [2]. The contagious period lasts from 1-2 days before up to 5-7 days after appearance of the rash [3]. Varicella usually is an uncomplicated, self-limiting disease. However, serious complications leading to hospitalisation and, sometimes, death do occur. In the Netherlands, the general practitioner (GP) consultation rate (all ages) for varicella is 253 per 100,000 populations leading to an estimated 40,000 visits to the GP. The hospital admission rate is 1.3 per 100,000 leading to 200 hospital admissions, and there are about two deaths each year on a population of 16 million [1].

Universal vaccination for varicella has been introduced in several developed countries (e.g. Germany, Australia, Japan, Canada and the United States), showing dramatic decline of disease incidence in the latter [4]. In the Netherlands the decision to include varicella in the national immunisation vaccination programme is pending. The main points of uncertainty relate to the impact of varicella vaccination on the incidence of zoster infections later in life [5], and uncertainty related to the burden of complications following varicella infection, and the indirect non-medical costs associated with varicella [2]. Most data we have on the burden of varicella stem from hospital data and GP-databases. However, the Netherlands is known for more conservative health seeking behaviour as compared to other European countries. The GP contact rate is only lower in Finland and Greece [6].

Hence we will miss a considerable burden of varicella in the population when we look at health service based data, because not all of the parents will seek medical help. We acknowledge that those cases can be expected to be mild, but they still might have an impact on society through work loss by the parents.

The aim of this study is to provide insight in the burden of varicella from a parent's perspective. We will describe the frequency of symptoms, including complications, the use of medical resources, and the societal impact including cases, which do not enter the healthcare system.

This study provides insight in the medical burden of varicella from a parent's perspective and in addition also on the impact on work absenteeism and productivity loss at work.

Only 38% of the parents visit a medical doctor due to symptom or complication. The whole range of symptoms associated with varicella was more present among children of parents who visited a medical doctor. On average there were about 4 hours of work lost per varicella episode when parents had visited a medical doctor, and about 1.6 hours when they did not. Parents report more symptoms than what we know from GP databases.

The results of this study were based on an Internet panel survey on varicella, with a 'response rate' of 37%. Although this seems rather low, it should be kept in mind that this is not a classic response rate as it depends on how well the parents with young children could be targeted by the market research agency. This depends on the background data available at the market research agency. When this background is incomplete, a larger, less targeted wave of invitations is sent out, which will result in relatively low response rates.

The mean age of the parents who filled out the questionnaire was 36 years (SD 5 years) - in line with expected ages of parents with children under six years old. This age group is very Internet-literate in the Netherlands with 86% of 25-54 year olds using the Internet at least once a week [12]. In most cases the mother was the respondent (82%), but we explicitly asked the respondent to fill out the data on productivity loss for other possible caregivers (e.g. father or grandparents) as well. The demographics of the sample showed that the geographical distribution of the respondents was representative of the population in the Netherlands. The educational level of the respondents was somewhat higher than reference data from the national statistics (highest education 35% versus 24%). The educational and income level of the participants can influence the outcome, as there might be a difference in use of medical resources between educational levels [13]. Socioeconomic status (measured as educational level and income) might also affect employment rate or type of employment, which could affect the results of absence from a paid job.

Comparison of results of an Internet access panel with a probability sample (random digit dialling (RDD)) has shown that internet responses were significantly more likely to agree with RDD responses when the question asked was about the respondent's personal health (nine times more likely than when not about their personal health) and when the question was factual (nine times more likely than non-factual) [14]. The questions in our questionnaire complied largely with these characteristics.

This Internet survey showed that 88% of the respondents' children aged younger than six years had a history of varicella. De Melker et al. reported a seroprevalence of varicella for children aged five years of 93%, including asymptomatic cases of varicella [1]. About 5% of first VZV infections occur asymptomatically and this internet survey did not include asymptomatic cases of varicella. Thus, the prevalence of symptomatic cases in our survey is comparable with the seroprevalence as reported by De Melker et al. [1].

The prevalence of symptoms reported in this study gives an overview of the prevalence of the most common varicella symptoms from a parent perspective. It shows, not surprisingly, that almost all children have skin vesicles and show itching. Fever occurs in about three quarters of the children, two thirds have problems with sleeping, and about half of the children have problems with eating or drinking. 11% of the parents reported an episode with complications for which they visited a medical doctor.

Parents are capable of reporting a history of varicella in their children. Several studies have shown high positive predictive value of 95% or more for parent reports [15, 16]. The accuracy of self-reported data diminishes with time elapsed. We asked parents to report on a varicella episode of their children aged less than six years. We analysed whether there were differences in the report of complications for a history of varicella up and including 12 months and 12 months or more. There were no significant differences (Chi-square test; p = 0.964).

It is probably difficult for parents to distinguish between a more serious course of the varicella episode and complications. Therefore we just listed mild and more serious symptoms in the questionnaire (see Table 1) and we labelled 'skin infection or inflammation', 'balance or movement disorder', 'fits/convulsions', and 'unconsciousness' as complications afterwards in the analyses. Parents reported 'complications' in 18% of all varicella episodes irrespective of physician visit. We used lay-men vocabulary to ask about the more serious symptoms. As a consequence of that, within the 18%, symptoms like 'balance and movement disorder', 'fits/convulsions', and 'unconsciousness' will probably include mild presentations of these symptoms.

The varicella episodes with complications for which a medical doctor had been visited probably represent the more serious complications. This was the case in 11% of all varicella episodes (278/2564); in 6% of all varicella episodes drugs were prescribed in an episode with complications (149/2564). Those percentages fit in the range of data on complications of varicella published elsewhere. The complication rate in varicella in primary care ranges from 2% to 15% in other published data [2, 17–23]. Although those other studies are primary care based and our study is population based, we believe that this comparison is justified. Health care use is much higher in other European countries in mild childhood diseases as compared with the Netherlands. In the Netherlands, for any given disease-episode in young children 47% of the parents visit a medical doctor, whereas this is 79% in Germany, 89% in Spain, 86% in France and 87% in Italy [24]. In other words, it is likely that in other countries less serious cases present at primary care as well, contrary to the Netherlands.

Another factor that should be taken into account to appreciate the results of the Internet survey is that the results are based on parent reports which might result in a higher prevalence estimate as compared with primary care based studies. GP's or paediatricians might only record what they consider the primary symptom for which parents came for a consultation, while parents have likely reported all symptoms they remembered about the disease episode. This might, for example, include relatively minor skin infections that a GP would normally not record. In addition, children are infected with varicella-zoster-virus early in life in the Netherlands (50% seroconversion by age two), as compared to for example Germany (50% seroconversion by age four). This might also affect the complication rate and explain differences in complication rate between countries.

Other factors that explain differences in complication rates among publications are the complications included e.g. dehydration from vomiting or diarrhoea, exacerbation of asthma, pneumonia, bacterial super infections, acute neurological disorders, bronchitis and otitis media, conjunctivitis and corneal infections, central nervous system injuries and stomatitis. Studies differ also in age groups included (e.g. up to 14 years or under 4 years), and whether or not immunocompromised children are included [2].

In case of hospitalizations, we asked the symptoms that led to admission of the child. Although the number of hospitalized children was low (n = 17), the distribution of symptoms seems more comparable with published data. For example, in Germany, febrile convulsions were present in 30% of the hospitalized children < 16 years in the pre-immunization era [25], and skin infections in 23% [26]. Super infections of the skin occurred in 37% of hospitalized children in a French study [27]. Neurological complications were reported in about one quarter of hospitalized cases in a French and German study [25, 28].

Data on health care use of mild childhood diseases, and specifically the percentage of children not seeking medical care are scarce. Our results on health care use could be compared with a postal survey among parents of children aged 2-6 years in the Netherlands in 1998 [29]. IThe postal survey relates to all kinds of childhood diseases. In our study, parents did not seek medical help for 62% of the disease episodes, compared to 50% in the postal survey. There had only been telephone contact with a GP in 17% versus 23% of the disease episodes in respectively the Internet and postal survey. Use of OTC drugs was reported in the Internet survey for 72% and in the postal survey for 64% of the episodes. Results from another Internet survey indicated that 53% of the children with symptoms of otitis media did not consult a medical doctor in the Netherlands [24].

The hospitalization rate of 0.7% in children with varicella is comparable with hospitalization rates reported in other studies. The hospitalization rate reported by De Melker et al for children presenting with varicella at the GP was 0.4% for 5-9 year olds, 1.0% in 1-4 year olds and 2.7% in infants [1]. Banz et al. (2004) reported a hospitalization rate for cases with varicella or an associated complication of 0.8% in Germany [11]. He referred in the same article to other published data indicating that in the United States between 0.1% and 0.6% of children are admitted to hospital as a result of varicella or an associated complication, while data from Canada suggest a somewhat higher hospitalization rate for children during an acute episode of varicella, ranging between 0.2% and 1.5%, and in France about 0.5% of all varicella patients need hospital care due to severe complications [11].

To reduce recall bias, questions on societal impact (e.g. work absenteeism) were only asked about varicella episodes that had occurred up to and including 12 months ago; half of the varicella episodes had occurred up to 6 months ago. Results for work absenteeism (Table 4) are comparable with other studies. Seventeen per cent of the parents needed to take time off from work to care for their child. It has previously been estimated from a Dutch community study on costs of gastroenteritis that 15% of the parents had to miss work for a child's episode of rotavirus diarrhoea [30, 31]. A review study on economic evaluations of varicella vaccination programmes has reported numbers of working days lost for a common case of varicella of 1.2 in Spain, 0.6 in France and 2.6 in Germany (adjusted for caregivers outside the labour market) [32]. Other studies have reported 0.6 to 3.7 working days missed for any uncomplicated case of varicella [11, 33, 34]. Results for the Netherlands (0.5 days for episodes with and 0.2 days for episodes without consultation of a medical doctor) are at the low end of this range, which can probably be explained by a high-proportion of part-time work (80% among women) in the Netherlands [35].

In sum, this study provides insight in the full spectrum of varicella-episodes and associated healthcare use, which is of interest for an infectious disease that infects virtually all young children. In addition, insight in the proportion of cases not seeking medical care and the societal impact associated with those cases gives valuable information for the cost-effectiveness analyses on vaccine introduction in the Netherlands. This study shows the otherwise not visible burden of disease outside the healthcare system, which is of particular importance in countries with conservative health seeking behaviour.