In the United States, rotavirus activity crosses the US states from south-west to north-east in a repetitive yearly pattern with a delay in peak activity observed in recent years.1 Pitzer et al2 have explained these patterns to largely be due to variation in birth rates across states and declining birth rates over time. In Europe, Koopmans and Brown3 observed a similar geographical pattern of rotavirus activity. Their analysis, based on data from 5 European countries in the late 80s and early 90s, suggested a south-west to north movement of rotavirus activity across Europe.3 Our study extends this work by examining seasonality data from 17 European countries between 1999 and 2009. We examined the association of birth rates (as identified in the United States) and weather factors with timing of rotavirus activity across Europe.
Weekly counts of laboratory-confirmed rotavirus infections between 1999 and 2009 were extracted from the European Rotavirus Network (EuroRotaNet) database. EuroRotaNet is a European collaborative network of sentinel surveillance laboratories for rotavirus that has been established to gather standardized, comprehensive information on the rotavirus types cocirculating throughout Europe.4 Crude birth rates (live births per 1000 estimated population per year) for European countries were obtained from the European Commission statistical database Eurostat.5 Data on weather variables were obtained from the MIDAS Land Surface Observation database.6 For each European country, we constructed population-weighted mean weekly series for ambient temperature, relative humidity, and total rainfall as previously described by Atchison et al.7 We used population-weighted weather series to better represent weather conditions in populated areas and improve our ability to detect the influence of weather factors on timing of rotavirus activity. Subsequently, we collapsed these weather series to obtain average weather series for the winter (December, January, and February) each year from 1999 to 2009.
Using methods similar to those described by Pitzer et al,2 we conducted univariable and multivariable regression analyses to explore the contribution of birth rates and weather factors on the spatiotemporal timing of rotavirus epidemics across Europe. For each country (c), timing of rotavirus activity was defined as the mean week (T) of the rotavirus season (r), where each week in the season was weighted by the number of rotavirus reports as follows:
where w is week of the rotavirus season running from 1 to 52, where week 1 starts July 1st and week 52 ends June 30th, and casesc,r,w is the number of reported rotavirus infections in country c in rotavirus season r and week w. We used the Bonferroni correction to adjust for multiple testing during the exploratory analysis with 4 explanatory variables (birth rates, ambient temperature, relative humidity and rainfall). Only explanatory variables in univariable analysis with a P value below 0.05/4 = 0.0125 were included in the multivariable regression model.
There was a regular annual south to north spread of rotavirus activity in Europe (Fig. 1). Rotavirus activity peaked earliest in Southern Europe (Italy, Greece, Spain) in late December and early January and latest in Northern Europe (Scotland, Denmark, Finland) in mid to late March. There was strong correlation between latitude and timing of rotavirus activity (R 2 0.59; P < 0.001) and no association with longitude (R 2 0.04; p 0.47).
In multivariable regression models earlier epidemics were found in countries with lower birth rates, higher winter temperatures and lower winter rainfall (Table, Supplemental Digital Content 1, http://links.lww.com/INF/A369). The significance of this is uncertain as neither birth rates nor weather factors were associated with temporal variation in rotavirus timing within countries.
We found a general south to north movement of rotavirus activity across Europe. This pattern of rotavirus activity is similar to the geographical pattern observed in the United States.2 Pitzer et al recently demonstrated that earlier epidemics were associated with higher birth rates, whereas weather factors could not explain the observed variability in timing of rotavirus epidemics either across states or over time in the United States.2 In contrast to what was observed in the United States, birth rates in Southern Europe tend to be lower than in Northern Europe, where epidemics occur later. We found that neither birth rates nor weather factors were associated with both spatial and temporal patterns in rotavirus timing across Europe. Spatiotemporal variations in birth rate in Europe range from 8 to13 live births per 1000 population per year and are lower (United States: 11 to 20 live births per 1000 population per year) and changing less than those in US states. This could explain, in part, why, unlike in the United States, varying birth rates are not important in determining the timing of rotavirus epidemics in Europe.
The use of rotavirus surveillance data and crude birth rates aggregated at a national level, as well as national weather series as an aggregate weather series for the whole of a country may dilute the weather and birth effects seen on timing of rotavirus activity within and across European countries. More detailed region or city-specific analysis would address this problem albeit at the loss of statistical power because some countries reported only small numbers of laboratory-confirmed rotavirus infections to the EuroRotaNet database. The Health Protection Agency has reliably collected reports of laboratory-confirmed rotavirus infections from laboratories across England and Wales since the early nineties and the data can be readily broken down by region.8 A survey of clinical laboratory practices found that most laboratories in England and Wales test for rotavirus all year round in all cases of gastroenteritis of those 5 years of age or younger and that the degree of under-reporting and the testing criteria for rotavirus has not varied significantly over time.8 Therefore, these data are very likely to be representative of patterns of rotavirus disease in the population. We explored the association between birth rates and weather factors and the timing of rotavirus activity in each region of England and Wales between 1993 and 2009. There was no evidence of any association (Table, Supplemental Digital Content 2, http://links.lww.com/INF/A370). This adds further support to our findings from our European analysis.
In summary, higher birth rates are associated with earlier timing of rotavirus activity in the United States,2 but not across Europe. Rotavirus transmission is likely to be driven by a complex interaction of population demographics, environmental factors and circulation of rotavirus strains. Under different epidemiologic conditions, different factors may be responsible for variation in local patterns.
The authors thank the members of EuroRotaNet and all collaborating institutes for contributing data from participating surveillance sites.
1. Turcios RM, Curns AT, Holman RC, et al. Temporal and geographic trends of rotavirus
activity in the United States, 1997–2004. Pediatr Infect Dis J
2. Pitzer VE, Viboud C, Simonsen L, et al. Demographic variability, vaccination, and the spatiotemporal dynamics of rotavirus
3. Koopmans M, Brown D. Seasonality and diversity of group A rotaviruses in Europe
. Acta Paediatr Suppl
4. Iturriza-Gomara M, Dallman T, Banyai K, et al. Rotavirus
surveillance in Europe
, 2005–2008: web-enabled reporting and real-time analysis of genotyping and epidemiological data. J Infect Dis
. 2009;200(suppl 1):S215–S221.
7. Atchison CJ, Tam CC, Hajat S, et al. Temperature-dependent transmission of rotavirus
in Great Britain and The Netherlands. Proc Biol Sci
. 2009 Nov 25. [Epub ahead of print.]
8. Atchison CJ, Lopman BA, Harris CJ, et al. Clinical laboratory practices for the detection of rotavirus
in England and Wales: can surveillance based on routine laboratory testing data be used to evaluate the impact of vaccination?. Euro Surveill