To the best of our knowledge this is the first study to explore the relationship between spatio-temporal variation in the dynamics of the 2009 A/H1N1 pandemic and demographic and climatic factors, and international travel patterns across a large range of latitudes. We found that the 2009 A/H1N1 pandemic influenza in Chile was characterized by a South to North gradient of increasingly late peak activity and decreasing disease transmissibility. Geographical variation in pandemic peak timing was associated with differences in latitude and climatic conditions, with latitude, maximum temperature and specific humidity accounting for 69-80% of the variability. Our findings could have important public health implications. In particular, intensified surveillance strategies in southern regions could lead to earlier detection of novel influenza viruses and improved pandemic control outcomes.
The latitudinal gradient evidenced in laboratory–confirmed A/H1N1 SARI hospitalizations in this study is confirmed by geographical patterns in weekly incidence of influenza-like-illnesses reported to the Chilean Ministry of Health (Additional file
1: Figures S2–S3). In addition, our data are consistent with previous studies of laboratory-confirmed A/H1N1 influenza reporting peak pandemic activity on June 22, 2009 in the central region of Santiago
, and on June 7–13, 2009, in the southern city of Puerto Montt
The south–north gradient in peak timing observed in our study is consistent with a decreasing trend in the reproduction number in the same direction and is statistically associated with maximum temperature and specific humidity. These findings are in agreement with experimental studies suggesting that influenza transmission is more efficient under dry and cold conditions
[24–29]. In particular, Chile experienced a south–north gradient in climatic conditions during May 2009 as pandemic activity was building up, ranging from 3.3 to 9.3 g/kg for specific humidity and 6.4°C to 22°C for maximum temperature.
The northward gradient of 2009 pandemic activity evidenced in Chile is reminiscent of the spread of the 2009 pandemic in Brazil, with the Southernmost regions of this country being hit earlier and experiencing greater severity than the Northernmost regions
. In contrast, seasonal influenza originates from low-population regions in the equatorial North of Brazil and travels to highly populous regions in the subtropical South over a 3-month period
, together with a weak transmissibility gradient
. In light of the intriguing Brazilian experience, it would be interesting to contrast the spread of the 2009 pandemic with that of seasonal influenza in Chile. Unfortunately, SARI surveillance data was limited to the pandemic period in Chile and no comparable information exists for prior years.
Although the main period of pandemic activity in Chile seems primarily correlated with local climatic conditions, we note that the timing of introduction of the first A/H1N1 cases (pandemic ‘onset’) was weakly associated with population size, with larger population centers experiencing earlier introductions than less populous regions. This hierarchical pattern of spread is in agreement with seasonal influenza epidemics in the United States
, the 2009 A/H1N1 influenza pandemic in Peru
 and Mexico
, and the 1918 influenza pandemic in England and Wales
[53, 54]. The Chilean experience also suggests that despite early introduction of the A/H1N1 virus in a large population center like Santiago in May 2009, located in the center of the country, local climatic conditions were not favorable for immediate full-scale transmission of the pandemic virus.
Winter school vacation period started on July 11th and coincided with the beginning of the downward phase of the pandemic in northern regions, similar to the 2009 pandemic experience in Peru
. In contrast, pandemic activity was well into the declining phase by the time winter vacations started in southern and central regions. Accordingly, the shift in the age distribution of cases associated with the winter vacation period was most pronounced in the northern area. A similar pattern has been reported in previous studies
[1, 2]. Of note, a widespread teachers strike involving public schools across Chile (approximately from May 18th to June 8th, 2009) could have slowed down the initial growth rate of the pandemic, although we were not able to quantify the effect of the strike here.
Our reproduction number estimates are consistent with previous studies reporting estimates in the range 1.2-2.1 for Chile, using national ILI and laboratory-confirmed influenza A/H1N1 cases
[5, 20, 55]. Further, R was estimated to be 1.8 (95% CI: 1.6, 2.0) for the southern region of Puerto Montt
, in line with our R estimate for Chile’s southern region (R~ 1.6-2.0). Overall, our transmissibility estimates for Chile are consistent with the range of global estimates reported for community-based settings at 1.2-2.4
[1, 2, 5, 18, 43, 48, 56–63], while higher estimates have been obtained in school settings
[2, 44, 64, 65]. The surprising gradient of pandemic A/H1N1 transmissibility observed from South to North Chile could be associated with differences in climatic conditions, prior immunity, or baseline differences in influenza transmission between regions, similarly to those reported for Brazil
. Unfortunately, the resolution of our data did not allow for more detailed analysis.
Chile experienced a single pandemic wave in 2009 as did other Southern Hemisphere countries including Argentina
[12, 13], New Zealand
 and Peru
. Other countries experienced multiple pandemic waves, including Mexico, the United States, and the United Kingdom
[9, 10, 66]. The timing of pandemic waves has been correlated with population density, school cycles and the season in which the novel influenza virus is introduced into local populations
[1, 2]. Our 2009 pandemic study set in Chile adds to our current understanding of the role of climatic conditions in modulating the transmission dynamics of pandemic influenza.
Several strengths and caveats of our study are worth noting. We used data on all SARI hospitalizations as well as a subset of hospitalizations that were laboratory-confirmed for A/H1N1 infections, representing all public and private hospitals in Chile. Testing rates for A/H1N1 influenza remained at 35% throughout the pandemic and were comparable to those of other countries
. Moreover, national respiratory virus surveillance for influenza-like-illness demonstrated that A/H1N1 predominated among other respiratory viruses in 96% of individuals aged 5 years and over
. In contrast, RSV co-circulated with A/H1N1 influenza among children < 5 years, although influenza remained the dominant virus in this age group
[18, 47]. Because a total of 365 (5.7%) of the SARI hospital records were missing the date of symptoms onset, we used the date of notification for 197 of these records for which the date of notification was available. Differences in reporting across regions cannot be ruled out, although there was no evidence of weaker reporting rates in less populous regions. On the contrary, regions with lower population sizes reported more SARI hospitalizations proportionally than larger regions in Chile. We note that our estimates of the reproduction number were derived using simple methodology relying on the initial growth rate of the pandemic across geographic areas of Chile
[41, 67]. More detailed epidemiological data providing information on the number of imported cases could have allowed the use of more refined estimation methods, but were not available to us (see e.g.,