1. Seasonality of influenza in humans: a conundrum across latitudes
Wladimir J. Alonso
Fogarty International Center
NIH

2. This talk: Influenza and …
The conundrum of human influenza seasonality
Causative explanations
Examples of contributions from Brazil in:
Describing the phenomenon
Understanding the dynamics of influenza pandemics
Public health policy implications 2

3. Current understanding of influenza seasonality
“Infectious disease dynamics offer a wide variety of intriguing and unexplained phenomena, yet none is as consistently observed while still remaining so poorly understood as the seasonality of influenza”
Lofgren et al (2007)
Influenza Seasonality: Underlying Causes and Modeling Theories
J Virol 81: 3

4. Maps of influenza peak timing
Based on the literature search in Pubmed between May 2009 and February 2010 for influenza and RSV articles published between 1990 and 2009.
all hospital and community-based studies using one or more laboratory tests, with a minimum average of 2 virus-positive specimens per month
Bloom-Feshbach et al. (2013) . PLoS ONE 8(2)

5. Maps of influenza peak timing
Based on the literature search in Pubmed between May 2009 and February 2010 for influenza and RSV articles published between 1990 and 2009.
all hospital and community-based studies using one or more laboratory tests, with a minimum average of 2 virus-positive specimens per month
Bloom-Feshbach et al. (2013) . PLoS ONE 8(2)

6. So, how do astronomical seasons trigger
influenza epidemics in different latitudes

7. Influenza and … The conundrum of influenza seasonality
Causative explanations
Examples of contributions from Brazil in:
Describing the phenomenon
Understanding the dynamics of influenza pandemics
Public health policy implications 7

8. See also from http://www.sercc.com/FuhrmannGeogCompassFlu.pdf :
There has been much debate as to which mode of transmission is most significant with respect to the epidemiology of influenza. For example, Lemieux et al. (2007) found that respiratory particles are most likely to settle in the upper respiratory tract, implicating
droplet and contact modes (i.e., large particles) as the primary means of transmission. This is further supported by Brankston et al. (2007) who suggested that most viral infections occur over short distances through physical or direct contact. However, a later study by
Lowen et al. (2008) on guinea pigs found that the relative role of contact and airborne transmission may be sensitive to changes in ambient temperature. Therefore, the dominant mode of transmission likely varies according to environmental conditions (Hall 2007). Earlier work by Schulman (1967) on virus transmission between mice corroborates this statement. While no study was found that quantified the size distribution of respiratory particles expelled during a typical cough or sneeze, it may be safe to assume that
both small and large droplets, as defined above, are expelled simultaneously. If this is the case, then smaller respiratory particles may disperse and affect susceptible populations across a broader geographical area (i.e., airborne mode), while larger particles may settle
directly onto an individual or other surface in a more confined environment (i.e., contact or droplet mode) (Hall 2007). Over the course of a full seasonal epidemic, it is likely that all three modes of transmission contribute to infection across varying space and time scales
(Weber and Stilianakis 2008).
Diana Marques 8

9. Practical application in recommendations for preventing the transmission in a pandemic9

10. A framework for addressing seasonality of influenza in humans
Global influenza seasonality: Reconciling patterns across temperate and tropical regions J Tamerius, MI Nelson, SZ Zhou, C Viboud, MA Miller, WJ Alonso Environmental health perspectives 119 (4), 439

11. Putative relationship and causal connections among key seasonal stimuli, mediating mechanisms, and influenza epidemics.

12. In the proposed framework, seven key “seasonal factors”—solar radiation, temperature, humidity, precipitation, viral interference, socio-behavior and intrinsic dynamics—can modulate disease incidence through three “mediating mechanisms”: host contact rate, virus survival, and/or host immunity (Figure 1). Below we define these three mediating mechanisms and describe how each can drive influenza virus seasonality.
Contact Rate
Increased proximity between susceptible and infected hosts is frequently identified as being an important driver of influenza seasonality. Observations of the rapid dissemination of influenza at schools (Cauchemez et al. 2008), cultural events (Ahmed et al. 2006), and on public transportation (Moser et al. 1978) suggests that crowding enhances transmission. In this review, we examine the effects of socio-behavior, temperature and precipitation on contact rates and influenza seasonality.
Virus Survival
Although there are several known modes of influenza transmission, the importance of each mechanism is still debated (Brankston et al. 2007, Weber and Stilianakis 2008, Tellier 2009). Airborne transmission occurs when virus laden droplets and/or aerosols are expelled from the respiratory tract and are inhaled by a susceptible host (Tellier 2009). In addition, an individual can become infected through digital inoculation on the mouth, nose and possibly through the eyes after contact with the external surface of an infected host, or inanimate objects (Weber and Stilianakis 2008). In any case, during transport between the infected and susceptible host, the virus is vulnerable to the ambient environment. Thus, seasonal factors which alter virus survival during transport such as temperature, humidity, and solar radiation can potentially explain influenza seasonality.
Immunity
Immunity refers to the capacity of a host to avoid or mitigate infection after exposure to influenza viruses, and is dependent upon the condition of several aspects of the host’s immune system. The notion that seasonal variation in immunity explains influenza seasonality is bolstered by observations that humans are more resistant to influenza during inter-epidemic periods (Shadrin et al. 1977), and influenza transmission varies by season in laboratory experiments with mice despite constant temperature and humidity (Shulman and Kilbourne 1963). We discuss the effects of viral interference, temperature, precipitation, solar radiation and intrinsic dynamics on immunity.
Mediating mechanisms

13. Seasonal Factors
In the proposed framework, seven key “seasonal factors”—solar radiation, temperature, humidity, precipitation, viral interference, socio-behavior and intrinsic dynamics—can modulate disease incidence through three “mediating mechanisms”: host contact rate, virus survival, and/or host immunity (Figure 1). Below we define these three mediating mechanisms and describe how each can drive influenza virus seasonality.
Contact Rate
Increased proximity between susceptible and infected hosts is frequently identified as being an important driver of influenza seasonality. Observations of the rapid dissemination of influenza at schools (Cauchemez et al. 2008), cultural events (Ahmed et al. 2006), and on public transportation (Moser et al. 1978) suggests that crowding enhances transmission. In this review, we examine the effects of socio-behavior, temperature and precipitation on contact rates and influenza seasonality.
Virus Survival
Although there are several known modes of influenza transmission, the importance of each mechanism is still debated (Brankston et al. 2007, Weber and Stilianakis 2008, Tellier 2009). Airborne transmission occurs when virus laden droplets and/or aerosols are expelled from the respiratory tract and are inhaled by a susceptible host (Tellier 2009). In addition, an individual can become infected through digital inoculation on the mouth, nose and possibly through the eyes after contact with the external surface of an infected host, or inanimate objects (Weber and Stilianakis 2008). In any case, during transport between the infected and susceptible host, the virus is vulnerable to the ambient environment. Thus, seasonal factors which alter virus survival during transport such as temperature, humidity, and solar radiation can potentially explain influenza seasonality.
Immunity
Immunity refers to the capacity of a host to avoid or mitigate infection after exposure to influenza viruses, and is dependent upon the condition of several aspects of the host’s immune system. The notion that seasonal variation in immunity explains influenza seasonality is bolstered by observations that humans are more resistant to influenza during inter-epidemic periods (Shadrin et al. 1977), and influenza transmission varies by season in laboratory experiments with mice despite constant temperature and humidity (Shulman and Kilbourne 1963). We discuss the effects of viral interference, temperature, precipitation, solar radiation and intrinsic dynamics on immunity.

14. RelativeHumidity Virus survival Absolute Humidity Virus survival
Hemmers et al Harper G 1961
Hood 1963
Loosli et al. 1943
RelativeHumidity
Virus survival
would explain better current data better
Shaman & Kohn 2009
anomalously ▼AH associated with the onset of wintertime influenza in US
Shaman et al 2010
Absolute
Humidity
Virus survival
RelativeHumidity
Virus survival
Template
Tropical
Shechmeister 1950, Schaffer et al. 1976

15. Does not explain circulation in the tropics
RelativeHumidity
Virus survival
Absolute
Humidity
Virus survival
RelativeHumidity
Virus survival
Template
Tropical
Shechmeister 1950, Schaffer et al. 1976
Problems:
Does not explain circulation in the tropics

16. Sensitivity of influenza viruses to UV radiation
(Tamm and Fluke 1950, Powell and Setlow 1956, Jensen 1964)
Low inactivation rates during low sun seasons
Sagripanti and Lytle 2007
Problems:
- the amount of UV indoors (where a large proportion of influenza transmission possibly occurs) is constant

17. Clinical trial (Urashima et al 2009):
Tested if vitamin D supplements affected the incidence of influenza in school-aged children.
Yes for Influenza A (and Asthma)
but No for Influenza B
Vitamin D (dependent upon exposure to UVB) acts as an immune system modulator
(Cannel et al 2006)
Vitamin D levels have a strong effect on immunity by promoting CD4 T-cell and mucosal antibody responses (Hayes et al 2003)
The fact that the body can produce vitamin D by itself has changed our understanding of this substance. It is no longer believed to be a true vitamin, but rather a steroid hormone. However, the direct exposure of the skin to sunlight is critical, and any limitation in sunlight exposure may result in vitamin D deficiency.
Vitamin D3 is synthesized in the skin from 7-dehydrocholesterol under the influence of UV-B (wavelength 290–315 nm) radiation and temperature [ 4–6]. UV-B radiation depends on many factors. First of all, the clothing style is very important [ 7]. Black clothes exclude 100% UV-B. Glass and plastic also exclude 100% UV-B [ 8]. The use of sunscreen factor 8 excludes 95% [ 9, 10]. In studies performed in Boston (42°N) Holick and coworkers [ 1, 2, 8] demonstrated the importance of latitude, season and time of the day of sunlight exposure. The maximal production of vitamin D3 was seen around noon in July, with decreasing production in spring and autumn and no production between 1 November and 15 March. In countries around the equator, the production is constant throughout the year.
It is obvious that moving people from sun-rich countries to countries with more northern or southern latitudes without changing their habit of avoiding direct sunlight exposure may cause vitamin D deficiency. In agreement with this, 'Asian osteomalacia' has been reported, especially in the UK, during the last three to four decades [ 11–21], and recently also in the Netherlands [ 22], Norway [ 23, 24] and Denmark [ 25].
Individuals exposed to very limited amounts of sunlight become dependent on an oral intake of vitamin D, as if it were a true vitamin. We do not know, however, what doses of vitamin D should be given to individuals deprived of sunlight to secure a normal serum level of 25-hydroxyvitamin D (25-OHD, the storage form of vitamin D). Recently, the recommendations for vitamin D intake have been increased [ 26–29], especially for elderly individuals, who also display reduced capacity for cutaneous vitamin D production [ 30, 31]. In a study where submariners were sunlight-deprived for 3 months, Holick [ 8] found that a daily dose of 600 IU (15 µg) vitamin D was not enough to keep a normal serum level of 25-OHD.
The aim of this study was to determine the degree of vitamin D deficiency amongst sunlight-deprived veiled Moslems living in Denmark. Additionally our aim was, through food intake analysis, to get an impression of what doses of oral vitamin D intake were needed to maintain normal 25-OHD levels.
(Glerup et al 2000)
See also:

18. Problems: Still poor evidence
It might act as a facilitator, but would still need additional mechanisms to explain why changes in weather trigger epidemics
Vitamin D levels have a strong effect on immunity by promoting CD4 T-cell and mucosal antibody responses (Hayes et al 2003)
The fact that the body can produce vitamin D by itself has changed our understanding of this substance. It is no longer believed to be a true vitamin, but rather a steroid hormone. However, the direct exposure of the skin to sunlight is critical, and any limitation in sunlight exposure may result in vitamin D deficiency.
Vitamin D3 is synthesized in the skin from 7-dehydrocholesterol under the influence of UV-B (wavelength 290–315 nm) radiation and temperature [ 4–6]. UV-B radiation depends on many factors. First of all, the clothing style is very important [ 7]. Black clothes exclude 100% UV-B. Glass and plastic also exclude 100% UV-B [ 8]. The use of sunscreen factor 8 excludes 95% [ 9, 10]. In studies performed in Boston (42°N) Holick and coworkers [ 1, 2, 8] demonstrated the importance of latitude, season and time of the day of sunlight exposure. The maximal production of vitamin D3 was seen around noon in July, with decreasing production in spring and autumn and no production between 1 November and 15 March. In countries around the equator, the production is constant throughout the year.
It is obvious that moving people from sun-rich countries to countries with more northern or southern latitudes without changing their habit of avoiding direct sunlight exposure may cause vitamin D deficiency. In agreement with this, 'Asian osteomalacia' has been reported, especially in the UK, during the last three to four decades [ 11–21], and recently also in the Netherlands [ 22], Norway [ 23, 24] and Denmark [ 25].
Individuals exposed to very limited amounts of sunlight become dependent on an oral intake of vitamin D, as if it were a true vitamin. We do not know, however, what doses of vitamin D should be given to individuals deprived of sunlight to secure a normal serum level of 25-hydroxyvitamin D (25-OHD, the storage form of vitamin D). Recently, the recommendations for vitamin D intake have been increased [ 26–29], especially for elderly individuals, who also display reduced capacity for cutaneous vitamin D production [ 30, 31]. In a study where submariners were sunlight-deprived for 3 months, Holick [ 8] found that a daily dose of 600 IU (15 µg) vitamin D was not enough to keep a normal serum level of 25-OHD.
The aim of this study was to determine the degree of vitamin D deficiency amongst sunlight-deprived veiled Moslems living in Denmark. Additionally our aim was, through food intake analysis, to get an impression of what doses of oral vitamin D intake were needed to maintain normal 25-OHD levels.
(Glerup et al 2000)
See also:

19. (influence of the cold)
Influenza di freddo
(influence of the cold)
Italy circa 1530
Goulart 2003

20. Experience with guinea-pigs
Virus stability
Experience with guinea-pigs
Aerosol transmission
Ordering of
phospolipids
5oC oC
Temperature
(<21oC)
Temperature
Polozov et al 2008
Lowen et al 2007

21. Problems:
Hard to explain circulation in the tropics

22. Cold winter weather causes people to crowd indoors
(one of the most accepted hypothesis)
“This idea of crowding indoors facilitating the spread of infection was generally accepted by scientists because there was no competing hypothesis to explain the seasonality of respiratory infections. In his textbook published in 1965, Andrewes (16) reluctantly accepts the idea that crowding indoors may explain the seasonality of colds but also raises the criticism that our cities are just as crowded in summer as in winter and states ‘‘Many people regard this (crowding) as the likeliest ‘winter factor’ to explain the facts. I have always had doubts about this. Indoor workers in towns spend their working hours in much the same way winter and summer; they are cheek-by-jowl in their of. ces or at the factory bench or canteen all through the year. There may be rather better ventilation in summer, but that is the only likely difference. If close contact were all, one would think that London Transport would ensure an all-the-year round epidemic’’. Andrewes goes on to conclude his discussion on colds and cold weather as follows: ‘‘So, as to why we get more colds in the winter, we must at present admit that we just do not know’’.
Our understanding of the seasonality of URTI has not progressed very far during the 20th century and in a textbook published in 1996 (17) the crowding theory is still put forward as the major explanation of seasonality: ‘‘There is no evidence to date to indicate that cold weather per se, chilling, wet feet, or draughts play any role in the susceptibility of people to colds. Epidemiological data suggest that school attendance and other forms of crowding populations (particularly children) are the major factors in uencing common cold virus transmission rates’’. However, the authors do acknowledge that ‘‘The underlying biological explanation for seasonal differences has remained elusive’’. At the start of the 21st century the crowding theory persists in textbooks of human virology as an explanation for the seasonality of respiratory infection and there are some signs that any link with climate has now been lost. In a summary on the seasonality of viral respiratory infections Collier and Oxford (2) state that ‘‘The precise conditions that result in seasonal spread are not known with certainty but may be attributed more to changes in social behaviour with the seasons e.g. overcrowding in cold weather, than with variations in humidity and temperature’’”
Eccles, R An explanation for the seasonality of acute upper respiratory tract viral infections. Acta Otolaryngol 122:

23. Individuals spend on average 1–2 hours more indoors during cold weather in the USA
(Graham and McCurdy 2004)
“This idea of crowding indoors facilitating the spread of infection was generally accepted by scientists because there was no competing hypothesis to explain the seasonality of respiratory infections. In his textbook published in 1965, Andrewes (16) reluctantly accepts the idea that crowding indoors may explain the seasonality of colds but also raises the criticism that our cities are just as crowded in summer as in winter and states ‘‘Many people regard this (crowding) as the likeliest ‘winter factor’ to explain the facts. I have always had doubts about this. Indoor workers in towns spend their working hours in much the same way winter and summer; they are cheek-by-jowl in their of. ces or at the factory bench or canteen all through the year. There may be rather better ventilation in summer, but that is the only likely difference. If close contact were all, one would think that London Transport would ensure an all-the-year round epidemic’’. Andrewes goes on to conclude his discussion on colds and cold weather as follows: ‘‘So, as to why we get more colds in the winter, we must at present admit that we just do not know’’.
Our understanding of the seasonality of URTI has not progressed very far during the 20th century and in a textbook published in 1996 (17) the crowding theory is still put forward as the major explanation of seasonality: ‘‘There is no evidence to date to indicate that cold weather per se, chilling, wet feet, or draughts play any role in the susceptibility of people to colds. Epidemiological data suggest that school attendance and other forms of crowding populations (particularly children) are the major factors in uencing common cold virus transmission rates’’. However, the authors do acknowledge that ‘‘The underlying biological explanation for seasonal differences has remained elusive’’. At the start of the 21st century the crowding theory persists in textbooks of human virology as an explanation for the seasonality of respiratory infection and there are some signs that any link with climate has now been lost. In a summary on the seasonality of viral respiratory infections Collier and Oxford (2) state that ‘‘The precise conditions that result in seasonal spread are not known with certainty but may be attributed more to changes in social behaviour with the seasons e.g. overcrowding in cold weather, than with variations in humidity and temperature’’”
Eccles, R An explanation for the seasonality of acute upper respiratory tract viral infections. Acta Otolaryngol 122:

24. It would also be valid for precipitation, as people spend about 0
It would also be valid for precipitation, as people spend about 0.5 hours more indoors during rainy weather conditions.
“This idea of crowding indoors facilitating the spread of infection was generally accepted by scientists because there was no competing hypothesis to explain the seasonality of respiratory infections. In his textbook published in 1965, Andrewes (16) reluctantly accepts the idea that crowding indoors may explain the seasonality of colds but also raises the criticism that our cities are just as crowded in summer as in winter and states ‘‘Many people regard this (crowding) as the likeliest ‘winter factor’ to explain the facts. I have always had doubts about this. Indoor workers in towns spend their working hours in much the same way winter and summer; they are cheek-by-jowl in their of. ces or at the factory bench or canteen all through the year. There may be rather better ventilation in summer, but that is the only likely difference. If close contact were all, one would think that London Transport would ensure an all-the-year round epidemic’’. Andrewes goes on to conclude his discussion on colds and cold weather as follows: ‘‘So, as to why we get more colds in the winter, we must at present admit that we just do not know’’.
Our understanding of the seasonality of URTI has not progressed very far during the 20th century and in a textbook published in 1996 (17) the crowding theory is still put forward as the major explanation of seasonality: ‘‘There is no evidence to date to indicate that cold weather per se, chilling, wet feet, or draughts play any role in the susceptibility of people to colds. Epidemiological data suggest that school attendance and other forms of crowding populations (particularly children) are the major factors in uencing common cold virus transmission rates’’. However, the authors do acknowledge that ‘‘The underlying biological explanation for seasonal differences has remained elusive’’. At the start of the 21st century the crowding theory persists in textbooks of human virology as an explanation for the seasonality of respiratory infection and there are some signs that any link with climate has now been lost. In a summary on the seasonality of viral respiratory infections Collier and Oxford (2) state that ‘‘The precise conditions that result in seasonal spread are not known with certainty but may be attributed more to changes in social behaviour with the seasons e.g. overcrowding in cold weather, than with variations in humidity and temperature’’”
Eccles, R An explanation for the seasonality of acute upper respiratory tract viral infections. Acta Otolaryngol 122:
Photo: Bart Pogoda

25. Problems:
These differences seem minimal in the 24 hours of the day (mainly in the current crowded conditions of urban areas)
No empirical data has shown an association between increased contact rates due to weather conditions and increases in influenza transmission (Lofgren et al. 2007)
“This idea of crowding indoors facilitating the spread of infection was generally accepted by scientists because there was no competing hypothesis to explain the seasonality of respiratory infections. In his textbook published in 1965, Andrewes (16) reluctantly accepts the idea that crowding indoors may explain the seasonality of colds but also raises the criticism that our cities are just as crowded in summer as in winter and states ‘‘Many people regard this (crowding) as the likeliest ‘winter factor’ to explain the facts. I have always had doubts about this. Indoor workers in towns spend their working hours in much the same way winter and summer; they are cheek-by-jowl in their of. ces or at the factory bench or canteen all through the year. There may be rather better ventilation in summer, but that is the only likely difference. If close contact were all, one would think that London Transport would ensure an all-the-year round epidemic’’. Andrewes goes on to conclude his discussion on colds and cold weather as follows: ‘‘So, as to why we get more colds in the winter, we must at present admit that we just do not know’’.
Our understanding of the seasonality of URTI has not progressed very far during the 20th century and in a textbook published in 1996 (17) the crowding theory is still put forward as the major explanation of seasonality: ‘‘There is no evidence to date to indicate that cold weather per se, chilling, wet feet, or draughts play any role in the susceptibility of people to colds. Epidemiological data suggest that school attendance and other forms of crowding populations (particularly children) are the major factors in uencing common cold virus transmission rates’’. However, the authors do acknowledge that ‘‘The underlying biological explanation for seasonal differences has remained elusive’’. At the start of the 21st century the crowding theory persists in textbooks of human virology as an explanation for the seasonality of respiratory infection and there are some signs that any link with climate has now been lost. In a summary on the seasonality of viral respiratory infections Collier and Oxford (2) state that ‘‘The precise conditions that result in seasonal spread are not known with certainty but may be attributed more to changes in social behaviour with the seasons e.g. overcrowding in cold weather, than with variations in humidity and temperature’’”
Eccles, R An explanation for the seasonality of acute upper respiratory tract viral infections. Acta Otolaryngol 122:

26. School schedules, calendar festivities, etc could drive influenza
Holidays reduced transmission among children in France by 20–29% (Cauchemez et al. 2008)
Template
Tropical

27. tropical epidemics do not overlap with school calendar
Problems:
- Influenza peaks during the winter in temperate locations, and not during the fall or spring (when children are also in school)
tropical epidemics do not overlap with school calendar
crowding also occurs year-round at festivals, sporting events, and conferences without consistent outbreaks of infection
Template
Tropical

28. Exposure to cool (and dry) temperatures can affect host immunity:
vasoconstriction in the nose and respiratory tract (Le Merre et al. 1996)
mucociliary function (Salah et al. 1988)
increases energetic demand (Lochmiller and Deerenberg 2000)
abrupt changes in temperature are also implied in a broad range of diseases, including influenza (Bull and Morton, 1978)
Several experimental studies indicate that exposure to cool temperatures can affect host immunity through a number of processes. For example, the inhalation of cold air causes vasoconstriction in the nose and respiratory tract, resulting in reduced blood flow (Le Merre et al. 1996), diminishing the supply of leukocytes and phagocytic activity in these areas (Eccles 2002, Mourtzoukou and Falagas 2007). Mucociliary function, which helps to clear the bronchi of foreign materials such as viruses, is inhibited by the inhalation of cool air (Proctor 1983). Also, exposure to cool temperatures increases the energy required for thermoregulation, possibly limiting resources below which is required for proper immune function (Lochmiller and Deerenberg 2000). An interesting aspect is that not only low temperatures, but also abrupt changes in temperatures, can be implied in a broad range of diseases, including influenza (Bull and Morton, 1978). In this case, a role of immunity seems more plausible than regarding to virus survival.
“The seasonality of URTI is related to long-term changes in the incidence of URTI measured on a monthly basis, but there is also a long-standing folkloric belief that rapid changes in air temperature associated with a short period of very cold weather are associated with a sudden epidemic of URTI. This belief is so widespread and has persisted for so long that it is dif. cult to dismiss the idea as having no credibility. The occurrence of an epidemic of URTI immediately
following a spell of cold weather may be explained by the conversion of many asymptomatic subclinical infections in the population into symptomatic clinical infections.” (Eccles 2002)

29. Problems:
Hard to explain circulation in the tropics
(although the hypothesis of increases in energetic demand is consistent with the requirement for thermoregulation during damp conditions of rainy season in the tropics)

30. Brazil 1) good epidemiological data is available
2) large territorial extension ranging from equatorial to semi-temperate climates 30

31. Problems in describing influenza seasonality
Estimating the circulation of influenza based on hospitalizations and deaths is not straightforward in temperate countries, and it is even less so in tropical countries.
Severe clinical outcomes are often caused by secondary bacterial infections, and a primary influenza infection may be unrecognized. Additionally, laboratory confirmation of influenza infection is rarely conducted.
As a result, most influenza-related hospitalizations and deaths are not attributed to influenza on discharge forms and death certificates 31

32. Study in Brazil
1) We investigated monthly pneumonia and influenza mortality data, from all Brazilian states, in a period comprising 22 years
2) We obtained independent confirmation of our findings with virus surveillance data from available years ( ) 32

33. Phase of the major peak (months of the year)
Peak timing was found to be structured spatio-temporally, with annual peaks being earlier in the north, and gradually later in the south of Brazil 5 -5
-10
-15
-20
-25
-30
-35
Latitude (degrees)
Colors indicate the same regions in both figures and sizes correspond to (log) of population of each state
J F M A M J J A S O N D
Phase of the major peak (months of the year)

34. Confirmation with virus surveillance data34

35. Climatologic factors
Monthly climatic data obtained from worldwide climate maps generated by the interpolation of climatic information from ground-based meteorological stations
Mitchell TD, Jones PD. An improved method of constructing a database of monthly climate observations and associated high-resolution grids. International Journal of Climatology 2005;25: (data at: 35

36. While temperature seems important in the South…
… precipitation discriminates much better influenza seasons in the equatorial regions
Plots of periods (averages of 3 months) of high- (circles) and low-influenza incidence (triangles)
While temperature seems important in the South… 36

37. This talk: Influenza and …
The conundrum of influenza seasonality
Causative explanations
Examples of contributions from Brazil in:
Describing the phenomenon
Understanding the dynamics of influenza pandemics
Public health policy implications 37

38. Influenza vaccination in Brazil
Previous weeks of southern cold were not necessarily in the right timing of vaccination for the entire country
da Cunha, Camacho et al 2005 Influenza vaccination in Brazil: rationale and caveats. Revista de Saúde Pública 39(1):129-36 38

39. Recommendation of optimization of influenza vaccination in Brazil as a by-product of the studies of influenza seasonality 39

40. Month of detection at IEC Month of detection at IAL
Virus of influenza detected at the Evandro Chagas Institute (IEC) and Adolfo Lutz Institute (IAL) in the period from I991 to 2007.
São Paulo’s data from Adolfo Lutz Institute and Belém data from Evandro Chagas Institute
Inferring time-dependent connections between influenza strains detected in Belém and São Paulo and vaccination timing
year
Month of detection at IEC
Detected vírus at IEC
Month of detection at IAL
Detected vírus at IAL
Vírus vacinais
1991
June
A/England/427/88 (H3N2)
A/Beijing/353/89 (H3N2)
October
July
September
B/Gingdao/102/91
A/Washinghton/15/91 (H3N2)
A/Beijing/358/89 (H3N2)
1992
Sem detecção
1993
April
A/Beijing/32/92 (H3N2)
1994
A/Guan Dong/25/93 (H3N2)
B/Guan Dong/8/93
Februrary
B/Brazil/216/94
1995
May
A/Shang Dong/9/93 (H3N2)
March
A/Johannesburg/33/94 (H3N2)
1996
January
A/Wuhan/359/95 (H3N2)
A/Taiwan/1/86 (H1N1)
A/Bayern/07/95(H1N1)
B/Beijing/184/93
1997
1998
August
A/Sydney/05/97 (H3N2)
1999
A/Sydney/5/97 (H3N2)
A/Beijing/262/95 (H1N1)
2000
A/NewCaledonia/20/99(H1N1)
A/Panamá/2007/99 (H3N2)
2001
A/NewCaledonia/20/99 (H1N1)
A/Panama/2007/99 (H3N2)
B/Sichuan/379/99
2002
B/Hong Kong/330/2001
2003
December
A/Newcaledonia/20/99 (H1N2)
A/Fujian/411/2002 (H3N2)
A/Panama/2007/99 (H3N2
2004
B/Hong Kong/1434
B/Brisbane/
2005
B/Shanghai/361/2002 H3
A/Wellington/1/2004 (H3N2)
2006
A/California/7/2004 (H3N2)
A/Wisconsin/67/2005 (H3N2)
B/Malaysia/2506/2004
2007
Belém
São Paulo

41. Southern Hemisphere Flu vaccine not good for Brazil?
influenza viruses isolated monthly from 1999 to 2007 in Belém and São Paulo
Belém
São Paulo
Mello et al (2010)

42. Southern Hemisphere Flu vaccine not good for Southern tropics?
Immunity based on hypothetical scenario where the Northern Hemisphere vaccination recommendations and schedule are used in both cities.
Immunity based on historic vaccination strategy adopted by Brazilian authorities, relying on the Southern Hemisphere vaccine recommendations and schedule.

43. Southern and Northern Hemisphere recommendation

44. Matches with Southern Hemisphere recommendations

45. Matches with Northern Hemisphere recommendations

46. Influenza vaccination in Brazil
Previous weeks of southern cold were not necessarily in the right timing of vaccination for the entire country
da Cunha, Camacho et al 2005 Influenza vaccination in Brazil: rationale and caveats. Revista de Saúde Pública 39(1):129-36 46

47. This talk: Influenza and …
The conundrum of influenza seasonality
Causative explanations
Examples of contributions from Brazil in:
Describing the phenomenon
Understanding the dynamics of influenza pandemics
Public health policy implications 47

48. Neste sentido, adquirem relevância especial para este simpósio aqueles parasitas que se utilizam de um vetor para “pegar uma carona” de um organismo para outro - os quais podem estar muito distantes entre si. Tais vetores são geralmente artrópodes (mosquitos, barbeiros, carrapatos, etc.) que necessitam perfurar mecanicamente nossa pele e injetar saliva para obter sangue, desta forma não somente servindo como veículos para os parasitas, mas também como canais de acesso para o interior do organismo dos hospedeiros. A eficiência desta forma de transmissão de parasitas pode ser notada, por exemplo, através do fato de que doenças humanas transmitidas por vetores geralmente alcançam valores de R0 muito superiores àqueles de doenças de transmissão direta (no caso da malária este pode superar a centena, enquanto no sarampo está em torno de 13; Spielman, 1999; Spielman e D'Antonio, 2001).

49. Data Mortality from two publicly available and independent sources:
Records from the Mortality Information System of the vital statistics agency of the Brazilian Ministry of Health ( )
Laboratory-confirmed H1N1pdm deaths from the National Surveillance Information System of Notifiable Diseases (SINAN) as of March 12 (2012)
Vital statistics collection in Brazil is uniform throughout the year and covered approximately 90% of the population in 2007 [14]. Mortality records from SIM were aggregated by state (27 administrative units: 26 states and the Federal District), age (<5, 5–14, 15–24, 25-44, 45-64, and >64 years), month and cause of death (as coded by ICD10; [15]). Considering previous evidence showing that pneumonia and influenza (hereafter P&I) deaths are the most specific endpoint for studying influenza mortality [16, 17], monthly- and age-specific P&I deaths (ICD10: J09- J18.9) were used. We also analyzed deaths from respiratory (ICD10: J) and circulatory (ICD10: I) causes, as these outcomes have been linked to influenza mortality [18]. We additionally investigated the impact of pandemic influenza on mortality from renal diseases (ICD10: N00-N39), as it has been associated with an increased risk of mortality in H1N1pdm patients [19-21].
In Brazil, notification of suspected pandemic influenza cases and deaths was mandatory [22, 23]. Initially, case-definition was restricted to fever higher than 38oC, cough, and close contact with an infected person or recent travel to countries with confirmed cases. However, as transmission became widespread (after epidemiological week 28, July 16), mandatory notification and laboratory investigation were restricted to cases of severe acute respiratory infection (SARI), including fever, cough, and dyspnoea or death [23]. Nationwide notification of cases and deaths was made through a national web-based reporting system and respiratory specimen collection and diagnosis was performed by the National Influenza Surveillance System, a network of 62 sentinel units established in 2000 to monitor virus circulation systematically in all Brazilian states. Specimen collection was standardized throughout the country and testing of respiratory specimens for H1N1pdm by real-time RT-PCR was centralized at three reference laboratories (Instituto Adolfo Lutz; Instituto Evandro Chagas; Fundação Oswaldo Cruz) [23].
Population estimates for each state and age group, demographic density and the proportion of the population in urban areas were obtained from the Brazilian Institute of Geography Statistics (IBGE; censuses 1993, 2001 and 2010). Annual population data were calculated by spline interpolation of census data.

50. Laboratory-confirmed H1N1pdm deaths
High diagnostic specificity
Relatively straightforward to obtain the parameters
However, it does not capture deaths caused by secondary complications
The burden and geographical patterns of pandemic mortality were analyzed using both laboratory-confirmed deaths and monthly vital statistics data. Due to the higher preciseness of laboratory-confirmed data, it was used to define the period of analysis. However, laboratory-confirmed data does not capture deaths caused by secondary complications, when viral detection is no longer possible.

51. Pneumonia and influenza coded records from the Mortality Information System
No attempt to estimate all deaths caused by H1N1pdm strain.
Instead, we estimate mortality in the pandemic period above that expected in a non-pandemic year (and attributed causes)

52. Geographical patterns in the timing of pandemic mortality: week of the first lab-confirmed death by state.
Neste sentido, adquirem relevância especial para este simpósio aqueles parasitas que se utilizam de um vetor para “pegar uma carona” de um organismo para outro - os quais podem estar muito distantes entre si. Tais vetores são geralmente artrópodes (mosquitos, barbeiros, carrapatos, etc.) que necessitam perfurar mecanicamente nossa pele e injetar saliva para obter sangue, desta forma não somente servindo como veículos para os parasitas, mas também como canais de acesso para o interior do organismo dos hospedeiros. A eficiência desta forma de transmissão de parasitas pode ser notada, por exemplo, através do fato de que doenças humanas transmitidas por vetores geralmente alcançam valores de R0 muito superiores àqueles de doenças de transmissão direta (no caso da malária este pode superar a centena, enquanto no sarampo está em torno de 13; Spielman, 1999; Spielman e D'Antonio, 2001).
The size of the data points is proportional to the population of each state.
The same colors are used to show each state on the graph and map.

53. Overall, there was an excess of 2,273 P&I associated deaths and 2,787respiratory-associated deaths No excess of circulatory deaths
95% confidence threshold was obtained
Monthly time series of deaths from various diseases, Brazil, The grey line represents observed data and the bold central line the estimated seasonal-trend mortality model calculated from 1996 to 2008, and extrapolated up to Thin grey lines represent the 95% CI. Positive and negative residuals (grey bars) represent positive and negative differences between the data and the 95% CI. Excess mortality was calculated as the sum of all residuals within the pandemic period (shaded area).
The onset of excess mortality from P&I and respiratory causes (Fig.1) coincided with the period when laboratory-confirmed A/H1N1 deaths were increasingly reported (shaded area; Fig.1), supporting the association between excess P&I and respiratory mortality and pandemic virus activity. Conversely, there was no excess mortality in 2009 or 2010 from circulatory (Fig.1), or from cerebrovascular and ischemic heart disease (data not shown)causes. Mortality from renal causes was also higher in 2009 and 2010 than in previous years (Fig.1). However, there was substantial excess mortality from renal causes well before the estimated start of pandemic virus circulation in Brazil and until the end of 2010, well after the notification of pandemic cases and deaths had nearly ceased in Brazil.

54. Geographical patterns in the severity of pandemic mortality
Laboratory-confirmed pandemic death rates (a) and P&I mortality in excess of that in pre-pandemic years (b) from June 01, 2009 to May 30, 2010 in each state (all age groups). The size of the data points is proportional to the population of each state. The same colors are used to show each state on the graph and map.

55. Lessons learned from the 2009 Pandemic in Brazil
Timing more similar to seasonal influenza (therefore putativelly constrained by climate) – different therefore from 1918 pandemic
Gradient of impact from the south to the equator, where pandemic had nearly no impact
(confirmed by recent studies in Africa)
Neste sentido, adquirem relevância especial para este simpósio aqueles parasitas que se utilizam de um vetor para “pegar uma carona” de um organismo para outro - os quais podem estar muito distantes entre si. Tais vetores são geralmente artrópodes (mosquitos, barbeiros, carrapatos, etc.) que necessitam perfurar mecanicamente nossa pele e injetar saliva para obter sangue, desta forma não somente servindo como veículos para os parasitas, mas também como canais de acesso para o interior do organismo dos hospedeiros. A eficiência desta forma de transmissão de parasitas pode ser notada, por exemplo, através do fato de que doenças humanas transmitidas por vetores geralmente alcançam valores de R0 muito superiores àqueles de doenças de transmissão direta (no caso da malária este pode superar a centena, enquanto no sarampo está em torno de 13; Spielman, 1999; Spielman e D'Antonio, 2001).
No excess mortality from secondary causes in Brazil (other than respiratory, pneumonia or influenza)

56. Importance of external drivers in influenza
Totally seasonal
Importance of climate
No seasonal at all
experience less pronounced seasonal differences
in humidity and rainfall-e .g., Indonesia
and Zaire-likewise exhibited much less
pronounced seasonal differences in the incidence
of smallpox (Fig . 4.6) . But where there
were clearly distinguishable hot and cool
seasons, with high and low absolute humidities
respectively, the incidence was always
much higher in the cool, dry season . Data on
importations of smallpox into Europe during
the period support the findings in
the endemic countries (see Chapter 23) . There
were 3 times as many importations during the
months December-May as in the following 6
months, which is explained by the higher
incidence of smallpox in the main "exporting"
countries in the Indian subcontinent at
that time. Further, each case imported during
December-May gave rise to an average of 24
subsequent cases (median 4 .5), whereas each
case imported in the period June-November
gave rise to an average of 1 .6 cases (median
1 .0) (Henderson, 1974) .
countries in the Northern and Southern
Hemispheres and in the humid tropics. In
Bangladesh the maximum reported incidence
was in the period from January to the end of
April ; in Brazil, in the Southern Hemisphere,
the maximum reported incidence occurred in
the spring, from August to the end of
October, with well-marked minima in September
to the end of December and in
February to the end of May . The maximum
incidence in Indonesia occurred in January
but thereafter the monthly incidence showed
little variation . Bearing in mind the delays in
reporting (see box), the maximum transmission
rate probably occurred in December in
Bangladesh and in June in Brazil-i .e ., during
the early winter. Old records for European
countries (Low, 1918 ; Mielke et al., 1984)
show that smallpox there was also mainly a
winter disease.
Within India, Rogers (1928) observed that
the seasonal fluctuation was the least marked
and the incidence the most uniform from year
to year in the State of Tamil Nadu (formerly
Seasonal Variations in Incidence
Most infectious diseases show characteristic
seasonal variations in incidence . In
temperate climates, where there are pronounced
seasonal differences in temperature,
arbovirus infections usually occur in summer,
enteroviral infections in summer and autumn,
influenza and other infections of the respiratory
tract mainly in winter, and measles,
chickenpox and mumps mainly in winter and
spring. Smallpox showed a seasonal incidence
similar to that of measles and chickenpox ; it
was mainly a disease of winter and spring . For
several of these diseases the seasonal variations
are blurred in tropical climates, where
the seasonal changes in temperature and
humidity are often much less marked . However,
smallpox showed a response to seasonal
effects in many tropical regions ; in Bangladesh
the seasonality was so pronounced that
smallpox was called, in Bengali, guti bashunto,
the spring rash (joarder et al., 1980) .
The most detailed studies of the seasonality
of smallpox were those reported by Sir
Leonard Rogers, using mortality figures from
British India (Rogers, 1926,1948) and data on
reported cases from England and Wales
(Rogers, 1928) and certain parts of Africa
(Rogers, 1948) . Although many of his data
were poor, comprising deaths rather than
cases, both being grossly underreported,
Rogers' observations of the seasonal incidence
of smallpox have been confirmed by the more
accurate data on case incidence obtained
during the global smallpox eradication
campaign.
Fig. 4.6 shows the monthly incidence of
reported cases of smallpox in representative
Madras), in South India, which experienced
little seasonal variation in temperature and
humidity ; these observations were confirmed
for the city of Madras by Rao et al. (1960) .
Other countries in the tropics, which also
A variety of factors can be envisaged that
could have contributed to the seasonal incidence
of smallpox : viability of the virus in an
infectious state, social factors and possibly the
physiological susceptibility of the host.
Immunity of the population

57. Prediction: more lethal and transmissible -> less seasonal
No seasonal at all
(pigs as
Mentioned by Martha -
Pigs in a constant pandemic?)
No seasonal at all
(pigs as
Mentioned by Martha -
Pigs in a constant pandemic?)
Totally seasonal
Totally seasonal
Interesting articles for this theory:
Ver esta alternativa, que eu não acredito:
Totally seasonal
Reed et al 2013
Novel Framework for Assessing Epidemiologic Effects of Influenza Epidemics and Pandemics, Emerg Infect Dis

58. Alonso et al 2011 The 1918 influenza pandemic in Florianopolis, a subtropical city in Brazil. Vaccine
1918 pandemic
Schuck-Paim et al. (2012) Exceptionally high mortality rate of the 1918 influenza pandemic in the Brazilian naval fleet. Influenza and Other Respiratory Viruses

59. www. epipoi. info. Alonso & McCormick 2012
.Alonso & McCormick BMC public health 12 (1), 1-9
“A user friendly analytical tool for extraction of temporal and spatial parameters from epidemiological time-series”.
The accessibility to analytical methods is often restricted to those who are familiar with computer programming. In an effort to redress this imbalance and make our methods, which have contributed to several papers, available to a wider audience we have developed a user-friendly tool. We believe that the BMC Public Health journal is an ideal platform to address the practitioners who may have data sets, but lack either the software or statistical background to fully utilize their data.
Our software, called EPIPOI is a stand-alone Windows application that we release freely to public health practitioners and researchers. We have included the basic program in the submitted materials, though we note it might be necessary for reviewers who do not have a Matlab license to download an additional free file that is too large to upload to the BMC server, though can be found on the accompanying website:

60. Thanks! alonsow@mail.nih.gov Collaboration on the studies presented:
Cécile Viboud
Mark A Miller
Rodolfo Acuña-Soto
Martha Nelson
Fernanda E Moura
Francielle Nascimento
Lone Simonsen
James Tamerius
Mirleide C dos Santos
Terezinha Paiva
Maria Akiko Ishida
Margarete A Benega
Roberto M. Fernandes
Marcia L. Carvalho
Luciane Z Daufenbach
Gerardo Chowell
Cynthia Schuck-Paim
Ministry of Health (Brazil)
Adolfo Lutz Institute (Brazil)
Evandro Chagas Institute (Brazil)
Oswaldo Cruz Foundation (Brazil)
Federal University of Ceara (Brazil)
National Autonomous University of Mexico (Mexico)
Arizona State University (USA)
Fogarty International Center/NIH (USA) 60