Seasonality of viral infections is ingrained in our vocabulary (‘influenza’: influence of the cold) and in our understanding of infectious diseases. However, the reasons underlying seasonality are poorly understood. Our ability to predict outbreaks or other deviations from the normal seasonal patterns is probably no better than that of Hippocrates in 400 BC . Understanding seasonality is highly relevant for public health planning. In this issue, Fisman  reviews the mechanisms underlying the seasonality of viral and vector-borne diseases. While noting the drivers for seasonality for different types of infection, Fisman cautions that the internal validity of the studies reporting on seasonality is limited, because of confounding by seasonality of risk factors for infection and other methodological issues. Examples include seasonal behavioural phenomena (e.g. school vacations), aggregation of incidence data, true geographical variation, and failure to account for person-to-person transmission of infections. Moorthy et al.  tackle the difficult issue of deviations in influenza seasonality. The authors argue that deviations from normal seasonality stem from the nature of the disease, i.e. relate to the emergence of a novel influenza virus strain, differences in behaviour of specific strains, or cross-protection from previously observed strains. These are, indeed, poorly predictable. A problem specific to the study of influenza seasonality is the variability in surveillance disease definitions, including influenza-like illness, documented influenza, hospital claims, pneumonia, and deaths.
Although we accept as common knowledge that influenza and many other viral infections are seasonal, we are less aware of seasonal effects in common bacterial infections. In this issue, we review the accumulating and surprising evidence from empirical studies on the seasonality of Gram-negative and Gram-positive bacterial infections [4,5]. Seasonal variation has been shown for bloodstream infections (BSIs) caused by Acinetobacter spp., Escherichia coli, Enterobacter cloacae, Klebsiella spp., and Pseudomonas aeruginosa, with higher rates of infection during the summer months. The strongest evidence exists for Acinetobacter spp., with summer peaks even in healthcare-associated infections. Few studies have examined specific climatic factors. Higher mean temperature has been associated with rates of BSIs caused by Acinetobacter spp., P. aeruginosa, and the Enterobacteriaceae. Higher humidity has been associated with P. aeruginosa BSIs . Lower urinary tract infections have been associated with higher mean temperature and lower humidity. Whether Staphylococcus aureus BSI has a seasonality pattern is unclear. There is, perhaps, a weak association with higher temperature. S. aureus skin and soft tissue infections peak in the summer or autumn. This has also been described for community-acquired methicillin-resistant S. aureus. S. aureus colonization might be more frequent in warm seasons, but these data are confounded by the seasonality of upper respiratory tract viral infections and pneumococcal colonization .
It is intriguing to consider whether these observations relate to real effects of environmental temperature, precipitation, humidity or other seasonal factors on bacteria and their pathogenicity. The alternative explanation is that the above findings represent confounding by unidentified factors or spurious trends related to inappropriate statistical methods being used for assessment of seasonal effects. The consistency of several different studies showing increased incidence of Gram-negative infections in summer and with increasing temperature point to a real bacterial effect. If this is true, globally, the population incidence of Gram-negative infections should be higher in countries with warmer climates. For Acinetobacter spp. this might be true; community-acquired Acinetobacter spp. infections are reported mainly in tropical and subtropical countries . Although we might believe in the seasonality of community-acquired bacterial infections, such an association would not be plausible in hospitals where climatic factors are artificial and stable. Kaier et al.  examine whether the incidence of hospital-acquired infections is related to seasonality through bed occupancy and crowding. The authors review empirical studies assessing such an association, and show that, although the evidence on this is limited, most studies do indeed report a significant association of bed occupancy rate and understaffing with hospital-acquired infections. Nearly all studies examined methicillin-resistant S. aureus infections. Thus, outside and inside hospitals, different mechanisms might explain the seasonality of bacterial infections, and studies examining seasonality should distinguish between hospital-acquired and community-acquired infections.
Common to all of the reviews included in this issue is a concern with the methods of analysis of seasonal trends in the primary studies describing the seasonality of infectious diseases. Christiansen et al.  provide helpful guidance on the conduct and appropriate analysis of such studies. The bias introduced by simplistic analyses of aggregate data (e.g. before–after studies) is difficult to predict. Such analyses might mask or exaggerate seasonal effects. Future studies should adhere to appropriate methods for analysis of seasonality in infectious diseases, to correctly elucidate this intriguing phenomenon.