Tweet
Epidemiology of WHO-confirmed human cases of avian influenza A(H5N1) infection
June 30, 2006
http://www.who.int/wer/wer8126.pdf [French version also available here.]
Introduction
In 1997, the first cases of human infection with the avian influenza A(H5N1) virus were reported in China, Hong Kong Special Administrative Region (Hong Kong SAR). These 18 cases included 6 deaths and coincided with outbreaks of highly pathogenic H5N1 in poultry on farms and in markets selling live poultry. Human cases ceased following the rapid destruction of the entire chicken population in Hong Kong SAR. In February 2003, 2 further human cases, with 1 death, were confirmed in a family in Hong Kong SAR who had recently travelled to Fujian Province in mainland China.
In mid-2003, the highly pathogenic H5N1 virus began to circulate widely in poultry in parts of south-east Asia, spreading within months to affect 8 countries in an outbreak unprecedented in its geographical extent. In December 2003, the first human cases associated with this outbreak occurred in VietNam. The disease remained confined to animals and humans in South-East Asia until mid-2005, when the virus expanded its geographical range through parts of central Asia to Europe, Africa and the Middle East. Between 1 December 2003 and 30 April 2006, 205 laboratory-confirmed cases and 113 deaths were reported to WHO from 9 countries. During that same period, the World Organisation for Animal Health reported outbreaks of H5N1 infection in domestic or wild birds in approximately 50 countries (Map 1). The objective of this analysis is to describe the epidemiology of laboratory-confirmed cases of H5N1 infection in humans.
Methods
This analysis included all laboratory-confirmed human cases of H5N1 infection as reported by onset date on the WHO web site1 from 1 December 2003 to 30 April 2006. Asymptomatic cases, confirmed retrospectively by testing serum samples during contact-tracing studies, were excluded.
All positive cases included in the study were confirmed by polymerase chain reaction on one or more respiratory tract specimens and/or by microneutralization assay on serum specimens.
Confirmatory testing was carried out by WHO reference laboratories for diagnosis of A/H5 infection2 for countries without reliable A/H5 influenza diagnostic capacity or without experience of diagnosis. Diagnostic results from other laboratories with strong diagnostic capacity and experience, including the national influenza centres in China and Thailand, the Pasteur Institute in Cambodia and the United States Naval Medical Research Unit Number 3 in Cairo, Egypt, were also accepted by WHO.
Data collection and analysis
Data were extracted from reports compiled by ministries of health, WHO epidemiologists, and partners in the WHO Global Outbreak Alert and Response Network.3 The original data were collected for surveillance activities rather than research and therefore the quality, reliability and format were not consistent across data from different countries. Exposure data were incomplete at the time of this publication and therefore have not been included in this analysis.
Where the precise date of onset of symptoms was missing, the month of onset was approximated using information on the date of reporting or the date of death and/or hospitalization, to allow those cases to be represented on the epidemiological curve. Such cases were, however, excluded from calculations made of the time from onset until the time of hospitalization and/or death.
The data were entered into the field information management system, designed by WHO to manage data on outbreaks, which was customized for this data set. Distributions were compared using the P2 test, and medians were compared using the Kruskal?Wallis test.
Results
Number and incidence of cases
From 1 December 2003 to 30 April 2006, 9 countries reported a total of 205 laboratory-confirmed human cases of H5N1 avian influenza to WHO (Table 1). Two asymptomatic cases in Viet Nam (an 81-year-old male and a 67-year-old female), included in the WHO count and identified during contact screening, were excluded from the analysis, giving a total of 203 cases.
Information on place of residence (to at least the first administrative level) and sex was available for all cases. Age was missing for 1 case. Information on date of onset was missing for 6% (13/203) of cases, and information on date of hospitalization was missing for 22% (45/203). For 9 of the cases, where the date of onset was missing, the date of hospitalization was used to approximate the month of onset. For the remaining 4 cases, for which the date of hospitalization was also missing, the month when the case was reported was used to assign the month of onset.
The curve of incident human H5N1 cases shows 3 peaks during the period from December 2003 to April 2006, roughly corresponding to winter and spring in the northern hemisphere (Fig. 1).
The number of new countries reporting human H5N1 cases has increased dramatically after October 2005, following the geographical extension of outbreaks among avian populations. Prior to mid-2005, only 4 countries had officially reported cases of H5N1 infection in humans. From October 2005 until the end of April 2006, 5 new countries reported cases of H5N1 infection in humans (Table 1).
Demographic characteristics
The median age of confirmed cases was 20 years. The age of cases ranged from 3 months to 75 years (n = 202). Half of the cases occurred among people aged <20 years; 90% occurred among those aged <40 years (Fig. 2). Among cases aged <10 years, 21 children were aged <5 years and 32 children were aged between 5 years and 9 years.
The overall sex ratio of males (n = 97) to females (n = 106) was 0.9, but this varied across age groups, with a ratio of 0.6?0.7 among those aged 10?19 years and 20?29 years and a ratio of 1.5 in those aged <10 years (Table 2). However, no statistically significant difference was found between the groups (P2 = 7.3, df = 6, P = 0.3).
Time from onset of symptoms until hospitalization
The number of days from onset of symptoms until hospitalization could be calculated for 73% (150/203) of cases. For all age groups, the median duration from onset of illness until hospitalization was 4 days (range = 0?18) (Fig. 3). Most patients presented for care within the first week after onset of illness. In 2004, the median interval was 5 days; in 2005 it was 4 days; and in 2006 it was 5 days. No significant statistical difference was found between these medians (Kruskal?Wallis H = 1.8, df = 2, P = 0.4).
Mortality
The overall case-fatality rate was 56%. The highest casefatality rate reported was 73%; this occurred among those aged 10?19 years (n = 49). The lowest case-fatality rate was 18%; this occurred among those aged $50 years (n = 11) (Table 3). A statistical difference was found between the age groups (P2 = 18.47, df = 6, P = 0.005). The overall case-fatality rate was lower in 2005 (43%) than in 2004 (73%) or in 2006 to date (63%).
The number of days from date of onset of symptoms until death was calculated for 97% of all fatal cases of infection (110/113). For the period from December 2003 to April 2006, the median duration from onset of symptoms until death was 9 days (range = 2?31 days); in 2004, the median duration was 11 days; in both 2005 and 2006, median duration was 8 days. No significant statistical difference was found between the medians (Kruskal?Wallis H = 4.7, df = 2, P = 0.1). In 95% of fatal cases, death occurred <23 days after onset of symptoms (Fig. 4).
Discussion
This description of cases is limited to those reported to WHO that were laboratory-confirmed and in which the patient had symptoms. Thus it cannot be inferred to what extent these cases are representative of all human infections with H5N1. Multiple selection biases may have occurred because some patients may have died before being tested or diagnosed, mildly symptomatic people may not have sought medical care, and false-positive or falsenegative test results may have occurred. However, recent serological surveys have detected only very low frequencies4, 5, 6 of non-symptomatic seropositivity to H5N1 virus among health-care contacts of patients with documented H5N1 infection. The frequency of subclinical infection or mild illness remains uncertain; therefore, more studies should be conducted among well-defined populations at risk of infection.
Although cases have occurred all year round, the epidemiological curve of H5N1 cases peaked during the cooler periods in the northern hemisphere for each of the years studied. If this pattern continues, an upsurge in cases could be anticipated starting in late 2006 or early 2007. Further studies are needed to assess the relationship between climatic conditions, poultry outbreaks due to H5N1 infection and associated human cases.
The highest proportion of cases occurred among those aged 10?29 years. Since most of these cases occurred in countries with a young population (for example, in 2005, 34% of the population in Egypt and 28% in Indonesia were aged <15 years),7 this might primarily reflect the age distribution within the countries affected, although age-related behaviours that increase risk of exposure are clearly important. The increased number of cases among females aged 10?29 years could indicate higher risk-exposure patterns (for example, by taking part in culling, defeathering or food preparation practices that are often carried out by specific population groups, such as young females). However, the incomplete nature of the data on exposure make it difficult to infer a link between age and exposure, and further studies are needed, especially to assess whether younger people or other groups (such as pregnant women) have an increased risk of contracting the infection.
A statistical difference was found in case-fatality rates across the age groups. These rates were highest among those aged 10?39 years, lowest among those aged >50 years and intermediate among children <10 years. This age profile differs from that for seasonal influenza, where the highest mortality rates are found among the very old age.8 The differences in the age-related case-fatality distribution among H5N1 cases are reminiscent of those observed during previous influenza pandemics, particularly in 1918, where case-fatality rates were higher among young adults.9
The median duration from onset of symptoms until hospitalization was 4?5 days across all years studied. This is compatible with that observed during the 1997 outbreak of 18 cases of H5N1 infection in Hong Kong SAR (median duration = 3 days).4, 10 Recall bias about the date of onset and/or hospitalization may have occurred when data were collected through interviews, especially among those cases where the investigation was initiated long after the onset of symptoms. Moreover, the date of hospitalization was missing for 22% (45/203) of cases, which hampers the drawing of conclusions. Similarly, the overall median number of days from onset of symptoms until death was 9 days, and this is broadly comparable across years.
The time-sequence observations in these data (that is, the interval from onset of symptoms until hospitalization and the interval from onset of symptoms until death) and the generally similar mortality rates suggest that illness patterns have not changed substantially across the years studied. However, further research is needed to assess the importance of other factors, including access to and quality of care, the use of antiviral drugs and complications.
Despite its limitations, this analysis may help generate hypotheses for in-depth studies that aim to identify exposure risks; it may also provide a foundation for future data collection that will lead to improvements in intervention strategies. Better standardization of the collection, validation and analysis of epidemiological and clinical data will greatly improve the ability to detect specific exposure patterns and will also enable better identification of risk groups, which in turn will help researchers to adapt and target preventive measures. Monitoring changes in the epidemiology of human cases and the severity and characteristics of the disease may help to identify changes in the virus? ability to pass from human to human or cause different patterns of illness. Collecting more detailed information on antiviral treatments and outcomes, and particularly linking this information to sequential virological sampling, could inform future management decisions.
In conclusion, this analysis describes the current epidemiology of human H5N1 cases but also highlights important gap in collection of essential data needed to understand this disease better and refine case management. As the virus is now considered endemic in poultry in some parts of the world and continuing to spread to birds in new areas, sporadic human cases will continue to occur. Moreover, the widespread distribution of the H5N1 virus in poultry and the continued exposure of humans suggest that the risk of virus evolving into a more transmissible agent in humans remains high. Therefore, the sharing of data may be seen as part of an early warning system that will collectively defend all countries against a common threat. In May 2006, the World Health Assembly adopted resolution WHA59.211 calling for immediate voluntary compliance with provisions in the International Health Regulations (2005) relevant to the threat of an influenza pandemic. If countries comply with these provisions, they will greatly assist themselves, the international community and WHO in monitoring evolving situations and supporting adequate responses as well as enabling reliable risk assessments to be made.
Footnotes
1 See http://www.who.int/csr/disease/avian.../en/index.html
2 See http://www.who.int/csr/disease/avian.../en/index.html
3 See http://www.who.int/csr/outbreaknetwork/en/
4 The Writing Committee of the World Health Organization Consultation on Human Influenza A/H5N1 Avian influenza A (H5N1) infection in humans. New England Journal of Medicine, 2005, 353:1374?1385.
5 Liem NT, Lim W, World Health Organization, International Avian Influenza Investigation Team, Viet Nam. Lack of H5N1 avian influenza transmission to hospital employees, Hanoi, 2004. Emerging Infectious Diseases, 2005, 11:210?215.
6 Apisarnthanarak A et al. Seroprevalence of anti-H5 antibody among Thai health care workers after exposure to Avian influenza (H5N1) in a tertiary care center. Clinical Infectious Diseases [online journal] , 2005, 40:e16-8 (http://www.journals.uchicago.edu/CID.../34804.web.pdf).
7 See http://unstats.un.org/unsd/demograph...2005/tab1b.htm (accessed 19 June 2006).
8 WHO. Avian influenza: assessing the pandemic threat, 2005 (http://www.who.int/csr/disease/influ...en/index.html; accessed 22 May 2006).
9 Simonsen L et al. Pandemic versus epidemic influenza mortality: a pattern of changing age distribution. Journal of Infectious Diseases, 1998, 178:53?60.
10 Yuen KY et al. Clinical features and rapid viral diagnosis of human disease associated with avian influenza A H5N1 virus. Lancet, 1998, 351:467?471.
11 See http://www.who.int/gb/ebwha/pdf_file...WHA59_2-en.pdf
June 30, 2006
http://www.who.int/wer/wer8126.pdf [French version also available here.]
Introduction
In 1997, the first cases of human infection with the avian influenza A(H5N1) virus were reported in China, Hong Kong Special Administrative Region (Hong Kong SAR). These 18 cases included 6 deaths and coincided with outbreaks of highly pathogenic H5N1 in poultry on farms and in markets selling live poultry. Human cases ceased following the rapid destruction of the entire chicken population in Hong Kong SAR. In February 2003, 2 further human cases, with 1 death, were confirmed in a family in Hong Kong SAR who had recently travelled to Fujian Province in mainland China.
In mid-2003, the highly pathogenic H5N1 virus began to circulate widely in poultry in parts of south-east Asia, spreading within months to affect 8 countries in an outbreak unprecedented in its geographical extent. In December 2003, the first human cases associated with this outbreak occurred in VietNam. The disease remained confined to animals and humans in South-East Asia until mid-2005, when the virus expanded its geographical range through parts of central Asia to Europe, Africa and the Middle East. Between 1 December 2003 and 30 April 2006, 205 laboratory-confirmed cases and 113 deaths were reported to WHO from 9 countries. During that same period, the World Organisation for Animal Health reported outbreaks of H5N1 infection in domestic or wild birds in approximately 50 countries (Map 1). The objective of this analysis is to describe the epidemiology of laboratory-confirmed cases of H5N1 infection in humans.
Methods
This analysis included all laboratory-confirmed human cases of H5N1 infection as reported by onset date on the WHO web site1 from 1 December 2003 to 30 April 2006. Asymptomatic cases, confirmed retrospectively by testing serum samples during contact-tracing studies, were excluded.
All positive cases included in the study were confirmed by polymerase chain reaction on one or more respiratory tract specimens and/or by microneutralization assay on serum specimens.
Confirmatory testing was carried out by WHO reference laboratories for diagnosis of A/H5 infection2 for countries without reliable A/H5 influenza diagnostic capacity or without experience of diagnosis. Diagnostic results from other laboratories with strong diagnostic capacity and experience, including the national influenza centres in China and Thailand, the Pasteur Institute in Cambodia and the United States Naval Medical Research Unit Number 3 in Cairo, Egypt, were also accepted by WHO.
Data collection and analysis
Data were extracted from reports compiled by ministries of health, WHO epidemiologists, and partners in the WHO Global Outbreak Alert and Response Network.3 The original data were collected for surveillance activities rather than research and therefore the quality, reliability and format were not consistent across data from different countries. Exposure data were incomplete at the time of this publication and therefore have not been included in this analysis.
Where the precise date of onset of symptoms was missing, the month of onset was approximated using information on the date of reporting or the date of death and/or hospitalization, to allow those cases to be represented on the epidemiological curve. Such cases were, however, excluded from calculations made of the time from onset until the time of hospitalization and/or death.
The data were entered into the field information management system, designed by WHO to manage data on outbreaks, which was customized for this data set. Distributions were compared using the P2 test, and medians were compared using the Kruskal?Wallis test.
Results
Number and incidence of cases
From 1 December 2003 to 30 April 2006, 9 countries reported a total of 205 laboratory-confirmed human cases of H5N1 avian influenza to WHO (Table 1). Two asymptomatic cases in Viet Nam (an 81-year-old male and a 67-year-old female), included in the WHO count and identified during contact screening, were excluded from the analysis, giving a total of 203 cases.
Information on place of residence (to at least the first administrative level) and sex was available for all cases. Age was missing for 1 case. Information on date of onset was missing for 6% (13/203) of cases, and information on date of hospitalization was missing for 22% (45/203). For 9 of the cases, where the date of onset was missing, the date of hospitalization was used to approximate the month of onset. For the remaining 4 cases, for which the date of hospitalization was also missing, the month when the case was reported was used to assign the month of onset.
The curve of incident human H5N1 cases shows 3 peaks during the period from December 2003 to April 2006, roughly corresponding to winter and spring in the northern hemisphere (Fig. 1).
The number of new countries reporting human H5N1 cases has increased dramatically after October 2005, following the geographical extension of outbreaks among avian populations. Prior to mid-2005, only 4 countries had officially reported cases of H5N1 infection in humans. From October 2005 until the end of April 2006, 5 new countries reported cases of H5N1 infection in humans (Table 1).
Demographic characteristics
The median age of confirmed cases was 20 years. The age of cases ranged from 3 months to 75 years (n = 202). Half of the cases occurred among people aged <20 years; 90% occurred among those aged <40 years (Fig. 2). Among cases aged <10 years, 21 children were aged <5 years and 32 children were aged between 5 years and 9 years.
The overall sex ratio of males (n = 97) to females (n = 106) was 0.9, but this varied across age groups, with a ratio of 0.6?0.7 among those aged 10?19 years and 20?29 years and a ratio of 1.5 in those aged <10 years (Table 2). However, no statistically significant difference was found between the groups (P2 = 7.3, df = 6, P = 0.3).
Time from onset of symptoms until hospitalization
The number of days from onset of symptoms until hospitalization could be calculated for 73% (150/203) of cases. For all age groups, the median duration from onset of illness until hospitalization was 4 days (range = 0?18) (Fig. 3). Most patients presented for care within the first week after onset of illness. In 2004, the median interval was 5 days; in 2005 it was 4 days; and in 2006 it was 5 days. No significant statistical difference was found between these medians (Kruskal?Wallis H = 1.8, df = 2, P = 0.4).
Mortality
The overall case-fatality rate was 56%. The highest casefatality rate reported was 73%; this occurred among those aged 10?19 years (n = 49). The lowest case-fatality rate was 18%; this occurred among those aged $50 years (n = 11) (Table 3). A statistical difference was found between the age groups (P2 = 18.47, df = 6, P = 0.005). The overall case-fatality rate was lower in 2005 (43%) than in 2004 (73%) or in 2006 to date (63%).
The number of days from date of onset of symptoms until death was calculated for 97% of all fatal cases of infection (110/113). For the period from December 2003 to April 2006, the median duration from onset of symptoms until death was 9 days (range = 2?31 days); in 2004, the median duration was 11 days; in both 2005 and 2006, median duration was 8 days. No significant statistical difference was found between the medians (Kruskal?Wallis H = 4.7, df = 2, P = 0.1). In 95% of fatal cases, death occurred <23 days after onset of symptoms (Fig. 4).
Discussion
This description of cases is limited to those reported to WHO that were laboratory-confirmed and in which the patient had symptoms. Thus it cannot be inferred to what extent these cases are representative of all human infections with H5N1. Multiple selection biases may have occurred because some patients may have died before being tested or diagnosed, mildly symptomatic people may not have sought medical care, and false-positive or falsenegative test results may have occurred. However, recent serological surveys have detected only very low frequencies4, 5, 6 of non-symptomatic seropositivity to H5N1 virus among health-care contacts of patients with documented H5N1 infection. The frequency of subclinical infection or mild illness remains uncertain; therefore, more studies should be conducted among well-defined populations at risk of infection.
Although cases have occurred all year round, the epidemiological curve of H5N1 cases peaked during the cooler periods in the northern hemisphere for each of the years studied. If this pattern continues, an upsurge in cases could be anticipated starting in late 2006 or early 2007. Further studies are needed to assess the relationship between climatic conditions, poultry outbreaks due to H5N1 infection and associated human cases.
The highest proportion of cases occurred among those aged 10?29 years. Since most of these cases occurred in countries with a young population (for example, in 2005, 34% of the population in Egypt and 28% in Indonesia were aged <15 years),7 this might primarily reflect the age distribution within the countries affected, although age-related behaviours that increase risk of exposure are clearly important. The increased number of cases among females aged 10?29 years could indicate higher risk-exposure patterns (for example, by taking part in culling, defeathering or food preparation practices that are often carried out by specific population groups, such as young females). However, the incomplete nature of the data on exposure make it difficult to infer a link between age and exposure, and further studies are needed, especially to assess whether younger people or other groups (such as pregnant women) have an increased risk of contracting the infection.
A statistical difference was found in case-fatality rates across the age groups. These rates were highest among those aged 10?39 years, lowest among those aged >50 years and intermediate among children <10 years. This age profile differs from that for seasonal influenza, where the highest mortality rates are found among the very old age.8 The differences in the age-related case-fatality distribution among H5N1 cases are reminiscent of those observed during previous influenza pandemics, particularly in 1918, where case-fatality rates were higher among young adults.9
The median duration from onset of symptoms until hospitalization was 4?5 days across all years studied. This is compatible with that observed during the 1997 outbreak of 18 cases of H5N1 infection in Hong Kong SAR (median duration = 3 days).4, 10 Recall bias about the date of onset and/or hospitalization may have occurred when data were collected through interviews, especially among those cases where the investigation was initiated long after the onset of symptoms. Moreover, the date of hospitalization was missing for 22% (45/203) of cases, which hampers the drawing of conclusions. Similarly, the overall median number of days from onset of symptoms until death was 9 days, and this is broadly comparable across years.
The time-sequence observations in these data (that is, the interval from onset of symptoms until hospitalization and the interval from onset of symptoms until death) and the generally similar mortality rates suggest that illness patterns have not changed substantially across the years studied. However, further research is needed to assess the importance of other factors, including access to and quality of care, the use of antiviral drugs and complications.
Despite its limitations, this analysis may help generate hypotheses for in-depth studies that aim to identify exposure risks; it may also provide a foundation for future data collection that will lead to improvements in intervention strategies. Better standardization of the collection, validation and analysis of epidemiological and clinical data will greatly improve the ability to detect specific exposure patterns and will also enable better identification of risk groups, which in turn will help researchers to adapt and target preventive measures. Monitoring changes in the epidemiology of human cases and the severity and characteristics of the disease may help to identify changes in the virus? ability to pass from human to human or cause different patterns of illness. Collecting more detailed information on antiviral treatments and outcomes, and particularly linking this information to sequential virological sampling, could inform future management decisions.
In conclusion, this analysis describes the current epidemiology of human H5N1 cases but also highlights important gap in collection of essential data needed to understand this disease better and refine case management. As the virus is now considered endemic in poultry in some parts of the world and continuing to spread to birds in new areas, sporadic human cases will continue to occur. Moreover, the widespread distribution of the H5N1 virus in poultry and the continued exposure of humans suggest that the risk of virus evolving into a more transmissible agent in humans remains high. Therefore, the sharing of data may be seen as part of an early warning system that will collectively defend all countries against a common threat. In May 2006, the World Health Assembly adopted resolution WHA59.211 calling for immediate voluntary compliance with provisions in the International Health Regulations (2005) relevant to the threat of an influenza pandemic. If countries comply with these provisions, they will greatly assist themselves, the international community and WHO in monitoring evolving situations and supporting adequate responses as well as enabling reliable risk assessments to be made.
Footnotes
1 See http://www.who.int/csr/disease/avian.../en/index.html
2 See http://www.who.int/csr/disease/avian.../en/index.html
3 See http://www.who.int/csr/outbreaknetwork/en/
4 The Writing Committee of the World Health Organization Consultation on Human Influenza A/H5N1 Avian influenza A (H5N1) infection in humans. New England Journal of Medicine, 2005, 353:1374?1385.
5 Liem NT, Lim W, World Health Organization, International Avian Influenza Investigation Team, Viet Nam. Lack of H5N1 avian influenza transmission to hospital employees, Hanoi, 2004. Emerging Infectious Diseases, 2005, 11:210?215.
6 Apisarnthanarak A et al. Seroprevalence of anti-H5 antibody among Thai health care workers after exposure to Avian influenza (H5N1) in a tertiary care center. Clinical Infectious Diseases [online journal] , 2005, 40:e16-8 (http://www.journals.uchicago.edu/CID.../34804.web.pdf).
7 See http://unstats.un.org/unsd/demograph...2005/tab1b.htm (accessed 19 June 2006).
8 WHO. Avian influenza: assessing the pandemic threat, 2005 (http://www.who.int/csr/disease/influ...en/index.html; accessed 22 May 2006).
9 Simonsen L et al. Pandemic versus epidemic influenza mortality: a pattern of changing age distribution. Journal of Infectious Diseases, 1998, 178:53?60.
10 Yuen KY et al. Clinical features and rapid viral diagnosis of human disease associated with avian influenza A H5N1 virus. Lancet, 1998, 351:467?471.
11 See http://www.who.int/gb/ebwha/pdf_file...WHA59_2-en.pdf
Comment