- Email: [email protected]
A population-based study on incidence and economic burden of influenza-like illness in south China, 2007
p u b l i c h e a l t h 1 2 5 ( 2 0 1 1 ) 3 8 9 e3 9 5
available at www.sciencedirect.com
Public Health journal homepage: www.elsevier.com/puhe
Original Research
A population-based study on incidence and economic burden of influenza-like illness in south China, 2007 R.N. Guo, H.Z. Zheng*, J.S. Li, L.M. Sun, L.H. Li, J.Y. Lin, J.F. He Institute of Epidemic Disease Prevention and Control, Centre for Disease Control and Prevention of Guangdong Province, No. 176 Xin Gang Xi Lu, Guangzhou, 510300, People’s Republic of China
article info
summary
Article history:
Objectives: The disease burden of influenza-like illness (ILI) in most tropical and subtropical
Received 3 March 2010
countries has not been described adequately to date. The aim of this study was to deter-
Received in revised form
mine the epidemiology and incidence of ILI, and to assess the economic burden in south
21 December 2010
China.
Accepted 5 March 2011
Study design: Prospective study.
Available online 25 May 2011
Methods: A population-based household survey was conducted quarterly in 2007 in Guangdong Province.
Keywords:
Results: The average number of subjects in each quarterly survey was 13,687. In total, 1002
Population study
cases of self-reported ILI were identified in all four surveys, indicating an annual incidence
Influenza
of 7.23 cases of ILI per 100 persons. The second quarter had the highest incidence of ILI
Incidence
(2.83 cases per 100 persons). Children aged 1e4 years, <1 year and 5e9 years had the
Disease burden
highest annual incidence rates of ILI (49.87, 35.19 and 21.24 cases per 100 persons, respectively). The incidence of ILI was significantly higher in males than females (P < 0.001), and significantly higher in rural residents than urban residents (P < 0.001). The individual cost per episode of ILI represented approximately 20% of monthly per-capita income of residents. Conclusions: The results of this large-scale household study confirm that ILI places a substantial health and economic burden on south China. Ultimately, the results of this study will provide further information for understanding the disease burden of influenza in subtropical areas. ª 2011 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved.
Introduction
loss of productivity that illness and lost work days place on employers and society as a whole.1,5e8
Substantial morbidity, hospitalization and mortality rates of influenza have been well documented in North America, Europe and certain districts in subtropical areas, such as Hong Kong.1e4 As well as the severe health impact, influenza has an extensive socio-economic burden, mainly due to the indirect
Global influenza surveillance is, to a large extent, based on sentinel surveillance of influenza-like illness (ILI). It has been reported that factors such as age, season, degree of latitude and population density could affect the magnitude of influenza activity. However, whole-population data are still scarce,
* Corresponding author. Tel.: þ86 20 84181985; fax: þ86 20 84193323. E-mail address: [email protected] (H.Z. Zheng). 0033-3506/$ e see front matter ª 2011 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.puhe.2011.03.004
390
p u b l i c h e a l t h 1 2 5 ( 2 0 1 1 ) 3 8 9 e3 9 5
with population-based ILI surveillance practiced in certain age groups.9e14 It has been hypothesized that south China is an epicentre for influenza,15,16 close to Hong Kong and Macao Special Administrative Region, Guangdong Province have large transient populations. As such, a collaboration centre was established in 2006 between Guangdong Province and the World Health Organization (WHO) for emerging infectious disease surveillance, research and training. The aim of this collaboration was to share Guangdong’s experience in emerging infectious disease surveillance with south China as a whole and neighbouring subtropical countries. Guangdong Province has conducted routine influenza surveillance since the 1970s. In 1998, it joined the People’s Republic of China influenza surveillance network, and has subsequently been conducting year-round influenza surveillance. For years, Guangdong Province has sent influenzavirus-positive specimens to the Chinese National Influenza Centre, which sends viral isolates to WHO. Specimens from Guangdong Province represent a significant proportion of the viral isolates that China shared with the WHO Global Influenza Surveillance Network. Guangdong Province currently maintains a surveillance network that includes 13 national and provincial level cities, 29 sentinel hospitals of all levels and 23 community/school clinics.17,18 This sentinel surveillance is useful for monitoring the patterns and circulating strains in Guangdong Province, but is limited in its ability to accurately estimate the burden or cost of disease. This is due, in part, to the fact that a certain proportion of influenza cases may not see a doctor during their illness, and a population-based study is needed to estimate the true incidence in the population. As such, a population-based door-to-door survey of self-reported ILI was conducted in 2007 to determine the incidence and epidemiological features of ILI in the community, and to gather background data to assess of the burden of influenza. Furthermore, investigations were made into medical care costs and days lost due to illness in order to understand the economic burden of ILI on residents in the community.
Methods Study population and sampling method Guangdong Province is one of the largest provinces in south China, with a total area of 177,900 km2. It has a typical tropical humid monsoon climate, with average annual temperature of 22 C and average rainfall of 1500e2000 mm. Guangdong Province has a population of 91,940,000 according to a population sampling survey in 2005. Based on multiple factors including geographic location, percapita gross national product and compliance with influenza surveillance work in recent years, Zhuhai and Zhaoqing Prefectures were selected as the two survey cities. The multistage stratified cluster sampling method19 was used to select one urban district and one rural county at random from each city. Two community residents committees were selected for each selected urban district. For each selected rural county, three towns were chosen, from which two administrative villages were
drawn. In total, eight urban community committees and 12 rural villages were selected at random. The door-to-door survey was conducted with each household as a survey unit.
Case definitions In this study, self-reported ILI was defined as fever with a measured body temperature >38.0 C and symptoms of either cough or sore throat. As a small number of subjects did not use thermometers, those who believed that they had a fever with a cough or sore throat were also considered as cases. New onset of ILI was defined as absence of the symptoms for at least 7 days prior to the onset of ILI.
Sample size calculation and sample assignment The sample size was calculated using a simple random sampling formula. The urban and rural groups were stratified. With a predicted ILI incidence rate per quarter (P) of 5% and allowable error (d) of 15%*P,20 a sample size of 3400 persons for each group was calculated. The sample size was increased by a factor of 1.1 to allow for lack of follow-up and non-response. The sample size was further enlarged by a factor of 1.5, to allow for case clustering. As such, a sample size of 5600 for each group (urban and rural) was determined. In total, it was anticipated that 11,200 persons would be surveyed each quarter. In accordance with the reported mean household population of 3.2 persons per household in the urban areas and 3.7 persons per household in the rural areas,21 220 households per urban community committee and 130 households per rural village were surveyed.
Data collection method Standardized questionnaires were designed. The household survey was administered by a researcher with a health background who received training. For each selected household, the head of the household was interviewed on behalf of all family members; if the head of the household was absent, another family member aged >18 years was interviewed. All subjects were enrolled in December 2006, and completed a registration form with demographic information. A mercurial thermometer for measuring axillary temperature and a notification letter was left for each family. The household survey was subsequently undertaken among the same cohort between 15 and 30 March, June, September and December of 2007. The effect of loss to follow-up (ranged from 2.4% to 4.1% in this study) was reduced by recruiting a number of subjects with the same demographic information to guarantee stability of the survey cohort. Interviews were conducted with each detected case of ILI by a trained surveyor, collecting information such as clinical characteristics, onset of illness within 3 months, medical care costs and days lost due to illness.
Calculation for incidence of ILI and epidemiological analysis This population-based year-round household survey investigated the incidence of ILI for the entire population (different genders, ages, occupations and education levels) living in
391
p u b l i c h e a l t h 1 2 5 ( 2 0 1 1 ) 3 8 9 e3 9 5
urban and rural areas in different seasons. Quarterly incidence rates of ILI were calculated. The annual incidence rate was obtained by calculating the cumulative incidence of ILI, with the total number of new ILI cases identified during the four quarters as the numerator, and the mean surveyed population over the four quarters as the denominator. Morbidity rates for ILI by person, season and area (urban/rural) were analysed. An estimate of the annual number of ILI patients in the whole province was obtained by multiplying the incidence among the surveyed population by the mid-2007 population size. The annual numbers of urban and rural ILI patients were calculated in the same way.
Medical care costs for ILI Medical costs included registration fee, prescribed drug fee, examination and treatment fee etc. In total, 811 self-reported ILI cases with valid medical cost data were detected during the four quarterly surveys. Mean cost per ILI episode, proportion of different costs and the medical cost difference between urban and rural areas were calculated. The exchange rate for Chinese Yuan to U.S. Dollars was 0.146. The currency units described in this article have been converted to U.S. Dollars.
Days lost due to illness In the face-to-face interviews, each identified ILI patient was asked about days lost due to illness. For employed persons, days lost represented lost work days. For those without a job, days lost were calculated as lost ‘normal’ days, and for students or children at kindergarten/childcare centre, days lost were lost school days. In total, 922 cases had valid data for lost time. For these patients, the mean number of days lost, the proportion of different types of days lost, and the difference in days lost between urban and rural areas were calculated.
Statistical analysis Frequencies were used for descriptive analyses, and Chisquared test was used to compare the proportions of different
Table 1 e Demographics of the surveyed subjects, household survey, 2007, Guangdong Province, China. No. surveyed First quarter
Second quarter
Third quarter
Fourth quarter
7054 7082 14,136
6814 6838 13,652
6772 6803 13,575
103 676 928 2084 8314 2031 14,136
103 661 899 2002 7997 1990 13,652
108 660 899 1985 7951 1972 13,575
7152 6984 14,136
6906 6746 13,652
6811 6764 13,575
Gender Male 7050 Female 7054 Subtotal 14,104 Age (years) <1 106 1～4 674 5～9 928 10～19 2075 20～59 8286 60 2035 Subtotal 14,104 Area of residence Urban 7045 Rural 7059 Subtotal 14,104
variables. ManneWhitney’s non-parametric test was adopted to compare medical costs between urban and rural areas. P < 0.05 was considered to be statistically significant for twotailed testing.
Results Demographics of surveyed subjects In total, 14,104 persons were surveyed in the first quarter of 2007, 14,136 were surveyed in the second quarter, 13,652 were surveyed in the third quarter, and 13,575 were surveyed in the fourth quarter. Of the average of 13,867 person-years of exposure, 1002 cases of self-reported ILI were identified over the four quarters, resulting in an annual incidence rate of 7.23 per 100 persons (1002/13,867). Gender, age and area of residence (urban/rural) of the surveyed subjects are shown in Table 1.
Table 2 e Gender-specific incidence of self-reported influenza-like illness (ILI) in the four quarterly surveys, 2007, Guangdong Province, China. Male
First quarter Second quarter Third quarter Fourth quarter Annual totalb
Female
Total
c2
Pvalue
No. surveyed
No. ILI
Incidence (%)a
No. surveyed
No. ILI
Incidence (%)a
No. surveyed
No. ILI
Incidence (%)a
7050
157
2.23
7054
116
1.64
14,104
273
1.94
6.303
0.012
7054
223
3.16
7082
177
2.50
14,136
400
2.83
5.633
0.018
6814
122
1.79
6838
96
1.40
13,652
218
1.60
3.245
0.072
6772
51
0.75
6803
60
0.88
13,575
111
0.82
0.695
0.405
6923
553
7.99b
6944
449
6.47b
13,867
1002
7.23b
11.977
0.001
a Unit of ‘%’ is the shortened expression of cases of ILI per 100 persons per quarter. Annual incidence is cases of ILI per 100 persons per year. b Annual incidence rate was calculated based on number surveyed in all four quarters.
392
7.23 1002 13,867 2.99 60 2007 2.83
c Chi-squared test found a significant difference in the incidence of ILI between age groups (c2 ¼ 2497.860, P ¼ 0.000).
b Annual incidence rate was calculated based on number surveyed in all four quarters.
a Unit of ‘%’ is the shortened expression of cases of ILI per 100 persons per quarter. Annual incidence is cases of ILI per 100 persons per year.
230 8137 7.22 147 21.24 Annual totalb
105
38
36.19
668
333
49.87
914
194
2037
0.82 111 13,575 0.51 10 1972 0.18 14 7951 0.76 15 3.67 3 108
2.78
660
36
5.45
899
33
1985
1.6 218 13,652 0.7 14 1990 0.53 42 7997 1.55 31 5.12 8 103
7.77
661
77
11.65
899
46
2002
2.83 400 14,136 0.84 17 2031 1.41 117 8314 3.07 64 7.87 15 103
Second quarter Third quarter Fourth quarter
14.56
676
114
16.86
928
73
2084
1.94 273 14,104 0.93 19 2035 0.69 57 8286 1.78 37 4.53 42 928 15.73 106 674 11.32 12 106 First quarter
No. surveyed
No. ILI
Incidence (%)a
No. surveyed
No. ILI
Incidence (%)a
No. surveyed
Incidence (%)a
2075
No. ILI No. ILI Incidence (%)a No. ILI
Incidence of self-reported ILI in different seasons
No. ILI
No. surveyed
No. ILI
Incidence (%)a
No. surveyed
20～59 10～19 5～9 1～4 <1
Table 3 e Age-specific incidence rate of influenza-like illness (ILI), household survey, 2007, Guangdong Province, China.
No. surveyed
60
Incidence (%)a
No. surveyed
Totalc
Incidence (%)a
p u b l i c h e a l t h 1 2 5 ( 2 0 1 1 ) 3 8 9 e3 9 5
Annual incidence rates of ILI for males and females were 7.99 and 6.47 cases per 100 persons, respectively; this difference was significant (c2 ¼ 11.977, P ¼ 0.001). Chi-squared tests were performed to determine the difference in incidence between the genders for each quarter. A difference was found in the first and second quarters, but not in the third and fourth quarters. A difference in incidence was also found between quarters (c2 ¼ 162.96, P ¼ 0.000). The highest incidence (2.83 per 100 persons per quarter) was seen in the second quarter of 2007 (Table 2). There was a significant difference in the incidence of ILI between age groups (c2 ¼ 2497.860, P ¼ 0.000). The three highest annual incidence rates occurred in children aged 1e4 years, <1 year and 5e9 years (49.87, 36.19 and 21.24 cases of ILI per 100 persons, respectively). Annual incidence of ILI was 7.22 per 100 persons for those aged 10e19 years, and 2.99 per 100 persons for those aged >60 years. The lowest annual incidence of ILI was found among adults aged 20e59 years (2.83 per 100 persons; Table 3). Age-stratified incidence analysis by area of residence was undertaken. The annual incidence of ILI was 5.66 and 6.91 cases per 100 persons in urban and rural areas, respectively; this difference was significant(c2 ¼ 18.653, P ¼ 0.000). Occupation analysis for ILI patients showed that the most affected population was preschool children, with an annual incidence rate of 42.08 cases per 100 persons, followed by students (10.73 cases per 100 persons), peasants (3.85 cases per 100 persons) and health workers (3.81 cases per 100 persons). Education level was also a contributing factor for ILI. The annual incidence of ILI showed a decreasing trend as education level increased, with the highest incidence (44.2 cases per 100 persons) seen in preschool children and the lowest incidence (3.18 cases per 100 persons) seen in persons who had education level above college degree.
Medical care costs and days lost due to ILI For the 1002 cases of self-reported ILI, 898 cases (89.6%) sought medical care and 10.4% chose to self-treat. For the 898 cases who sought medical care, only 811 cases had complete data regarding the consultation fee. Medical care cost analysis for these 811 self-reported ILI cases showed that one episode of illness cost approximately 172.5 Yuan (US$25.2). The mean cost for each urban inhabitant was 274.8 Yuan [US$40.12; 95% confidence interval (CI) 223.8e325.8 Yuan (US$32.7e47.6)], and the mean cost for each rural inhabitant was 107.2 Yuan [(US$15.65); 95% CI 88.6e125.8 Yuan (US$12.9e18.4)]. The Mann-Whitney test revealed that the medical care cost for each urban inhabitant was significantly higher than that for each rural inhabitant (Z ¼ 13.440, P ¼ 0.000). Nine hundred and twenty-two cases of self-reported ILI with complete data on days lost due to illness were also analysed. Among the 542 rural ILI patients, 97 cases (17.9%) lost a median of 2 work or school days. Among the urban ILI patients, 127 cases (33.7%) lost a median of 1 work or school day (Table 4).
p u b l i c h e a l t h 1 2 5 ( 2 0 1 1 ) 3 8 9 e3 9 5
Table 4 e Medical care costs and days lost due to selfreported influenza-like illness (ILI), household survey, 2007, Guangdong Province, China. Urban area No. ILI
%
Medical cost (REN MIN BI) <10 1 0.3 10～49 32 10.1 50～99 53 16.8 100～199 93 29.4 200～299 52 16.5 300～399 34 10.8 400～499 13 4.1 500～999 21 6.7 1000～1999 12 3.8 2000 5 1.6 Total 316 100.0 Lost time (days) 0 253 66.6 0.5 14 3.7 1 54 14.2 2 26 6.8 3 10 2.6 4 6 1.6 5 6 1.6 6 4 1.1 7 6 1.6 8 0 0.0 14 1 0.3 15 0 0.0 Total 380 100.0
Rural area
Total
No. ILI
%
No. ILI
%
7 225 125 75 27 13 5 9 7 2 495
1.4 45.4 25.3 15.2 5.5 2.6 1.0 1.8 1.4 0.4 100.0
8 257 178 168 79 47 18 30 19 7 811
1.0 31.7 22.0 20.7 9.7 5.8 2.2 3.7 2.3 0.9 100.0
445 7 28 24 13 11 3 7 2 1 0 1 542
82.1 1.3 5.2 4.4 2.4 2.0 0.6 1.3 0.4 0.2 0.0 0.2 100.0
698 21 82 50 23 17 9 11 8 1 1 1 922
75.7 2.3 8.9 5.4 2.5 1.8 1.0 1.2 0.9 0.1 0.1 0.1 100.0
Discussion This household survey was conducted quarterly reflecting ILI throughout 2007. The data showed that the annual incidence of ILI was 7.23 per 100 persons, with the highest quarterly incidence (2.83 per 100 persons) seen in the second quarter; this was slightly lower that that found by Claque et al.22 and Malone et al.23 The ILI activity pattern found in the present survey was consistent with local surveillance data for 2007, which revealed a peak in influenza in June. Virology monitoring identified a predominance of subtype H3 of A type and B type influenza viruses. This study clearly demonstrated that ILI activity in Guangdong Province occurs year-round, but with an evident peak in summer.17,18 This single-peak feature differs slightly from the findings of Nguyen et al., who reported two influenza peaks in south-east Asia.24 The present survey found that children <10 years of age were most affected by ILI. Annual ILI incidence among children aged 1e4 years was 50 cases per 100 persons. Primary schools and childcare centres were high-risk places for outbreaks of influenza.15,17,18 As a result, influenza prevention and control measures should focus on school-aged children, and the vaccination policy should prioritize this population. This study also found that males had a higher incidence of ILI than females, especially in the first and second quarters of the year. The reason for this difference is not yet clear and further investigation is needed.
393
Another interesting finding was that rural residents had a higher incidence of ILI than urban residents. Given a higher population density and worse air quality in urban areas, infection and transmission of influenza virus is presumably more feasible in urban areas than rural areas; accordingly, higher morbidity was expected in the urban areas.25 However, better compliance was observed among rural residents during this population-based household survey, so it is possible that this may have contributed to the higher incidence rate among rural inhabitants. The individual cost per ILI episode was US$40.12 for each urban inhabitant and US$15.65 for each rural inhabitant. This represents approximately 20% of monthly per-capita disposable income of urban residents, and 20% of monthly percapita total income of rural residents based on the Guangdong Statistic Year Book for 2006. ILI places a considerable economic burden on Guangdong communities regardless of whether they are urban or rural. Furthermore, given lost productivity due to absence from work, and other direct and indirect expenditures during the illness, such as extra nutrition costs, transportation fees and payment for nursing care, the actual economic burden per ILI episode could be much higher. Medical expenditure for urban residents is much higher than that for rural residents; this may be due to multiple aspects of healthcare-seeking behaviour, such as consultation rate, preferred prescribed drugs, preferred hospitals and ILI awareness. Among the urban ILI patients, 33.3% had lost work or school days due to illness, compared with 17.9% of rural patients. This shows that ILI had more severe consequences for urban residents than their rural counterparts. In terms of days lost, the median amount of lost time was 2 days for rural patients and 1 day for urban patients. As well as severity of illness, the position, seniority, institution type and production mode were related to the length of absence.26 Previous studies have documented the incidence, hospitalization rate, risk factors and economic burden of ILI or influenza. It has been reported that factors such as age, season, degree of latitude and population density could affect the magnitude of influenza activity.1e8 However, data targeting total populations are scarce, with population-based ILI surveillance practiced in certain age groups.9e11 In China, no systematic research has been undertaken to date on the health and economic burden of influenza or ILI on the community. The present study, based on a large-scale household survey, reports gender, age and location (urban/ rural) differences in the incidence of ILI; explores the incidence and epidemiological features of ILI in a community in south China; and estimates, for the first time, the burden that ILI places on South China. This population-based household study will provide background data for further assessments of the disease burden of influenza, and ultimately, the results of this study will provide further information for understanding the disease burden of influenza in subtropical areas. This study has several limitations. First, self-reported ILI was used as the case definition of ILI. As no virology validation was performed, the definition may capture a wide range of febrile illnesses, which may have led to to an overestimation of the incidence of ILI. Second, this large-scale survey lasted for 1 year and was administered four times. Decreased compliance
394
p u b l i c h e a l t h 1 2 5 ( 2 0 1 1 ) 3 8 9 e3 9 5
by the last stage of the survey is possible. Thirdly, only medical care costs and days lost due to illness were calculated in the evaluation of economic burden; this may have led to underestimation of the true burden of influenza as a variety of direct and indirect costs were not taken into consideration. Finally, the cohort involved a large population of approximately 14,000 persons, and loss to follow-up of a small number of subjects was inevitable. However, as the percentage of loss to follow-up was small (<5%), this is unlikely to have had a significant impact on the study results. Despite the limitations, the results of this populationbased household survey could supplement the provincial sentinel influenza surveillance system. Based on the primary results of this household survey study, it is estimated that at least 6.83 million people may have ILI in Guangdong Province. The estimated medical cost for the treatment of ILI patients would be US$155 million, given that 89% of patients seek medical care; this would constitute a significant public health problem. An appropriate estimate of the burden of influenza in Guangdong Province can be obtained by combining the population-based ILI survey and the sentinel hospital-based influenza surveillance.
Ethical approval This survey was approved by the review board of the Centre for Disease Control and Prevention of Guangdong Province, China, and oral consent was obtained from all subjects.
Funding World Health Organization.
Competing interests None declared.
Acknowledgements The authors thank Zhou Weigong for his technical support in preparing the methods and discussion sections of the text. Thanks also go to staff members of the Zhuhai and Zhaoqing Centres for Disease Prevention and Control of all levels, the local health bureau, and the trainees of the second phase of Guangdong Field Epidemiology Training Programme for their assistance and cooperation in implementation in the field survey.
references
1. Molinari NAM, Ortega-Sanchez IR, Messonnier ML, Thompson WW, Wortley PM, Weintraub E, et al. The annual impact of seasonal influenza in the US: measuring disease burden and costs. Vaccine 2007;25:5086e96. 2. Schanzer DL, Langley JM, Tam TWS. Hospitalization attributable to influenza and other viral respiratory illnesses in Canadian children. Pediatr Infect Dis J 2006;25:795e800.
3. Lenglet AD, Hernando V, Rodrigo P, Larrauri A, Donado JDM, de Mateo S. Impact of flu on hospital admissions during 4 flu seasons in Spain, 2000e2004. BMC Public Health 2007;7:197. 4. Li CK, Choi BCK, Wong TW. Influenza-related deaths and hospitalizations in Hong Kong: a subtropical area. Public Health 2006;120:517e24. 5. Espositol S. Clinical and socio-economic impact of influenza and respiratory syncytial virus infection on healthy children and their households. Clin Microbiol Infect 2005;11:933e6. 6. Li S, Leader S. Economic burden and absenteeism from influenza-like illness in healthy households with children (5e17 years) in the US. Respir Med 2007;101:1244e50. 7. Keren R, Zaoutis TE, Saddlemire S, Luan XQ, Coffin SE. Direct medical cost of influenza related hospitalizations in children. Pediatrics 2006;118:1321e7. 8. Nichol KL, D’Heilly SJ, Greenberg ME, Ehlinger E. Burden of influenza-like illness and effectiveness of influenza vaccination among working adults aged 50e64 years. Clin Infect Dis 2009;48:292e8. 9. Iwane MK, Edwards KM, Szilagyi PG, Walker FJ, Griffin MR, Weinberg GA, et al. Population-based surveillance for hospitalizations associated with respiratory syncytial virus, influenza virus, and parainfluenza viruses among young. Pediatrics 2004;113:1758e64. 10. Meier CR, Napalkov PN, Wegmu¨ller Y, Jefferson T, Jick H. Population-based study on incidence, risk factors, clinical complications and drug utilization associated with influenza in the United Kingdom. Eur J Clin Microbiol Infect Dis 2000;19:834e42. 11. Immerman JM, Lertiendumrong J, Dowell SF, Uyeki T, Olsen SJ, Chittaganpitch M, et al. The cost of influenza in Thailand. Vaccine 2006;24:4417e26. 12. Zhang J, Shi G, Shu Y, Yang W. Strategies on enhancing capacities of influenza surveillance in China. Chin Plan Immun 2005;11:339e42 [in Chinese]. 13. Duan P, Zhang R, Lin J, Zhang L, Zhao H, Yan B. Survey on incidence, absenteeism and drug use of influenza-like illness of barrack solders in 8 districts. Liber Army Prev Med 2001;19: 194e5 [in Chinese]. 14. Sun S, Yang S, Zhang H, Zhang Y, Zhang L. Impact of influenza on health of middle school and primary school student in 1998e1999 winter season of Tianjin urban district. Occup Health 2000;16:40e1 [in Chinese]. 15. Huang P. Study on epidemic, outbreak and molecular variation of influenza in Guangdong. Chin J Dis Control Prev 2004;8:144e7. 16. Shortridge KF. Pandemic influenza: a zoonosis? Semin Respir Infect 1992;7:11e25. 17. Huang P, Lin J, Ni H, Zhou H, Du Z, Liu S. Influenza analysis from 1998 to 2001 in Guangdong district. S Chin J Prev Med 2002;28:21e2. 18. Huang P, Chen W, Ni H, Zhou H, Liu S. Analysis on influenza epidemic and surveillance in Guangdong between 1988 and 1998. Strait J Prev Med 2001;7:15e7. 19. Du Z. Sampling method and its application. Beijing: Qinghua University Publishing Company; 2005. p. 8. 20. Guo R, Tang H, Kang M, Luo H, Zheng H. Telephone survey on incidence of influenza-like illness and vaccination status in community residents. Chin J Public Health 2008;24:843e5. 21. Department of Population and Employment Statistics National Bureau of Statistics of China. Communique of 1% population sampling survey in 2005 of Guangdong province. Beijing: China Population Statistics Yearbook; 2006. 22. Clague B, Chamany S, Burapat C, Wannachaiwong Y, Simmerman JM, Dowell SF, et al. A household survey to assess the burden of influenza in rural Thailand. SE Asian J Trop Med Public Health 2006;37:488e93. 23. Malone JL, Madjid M, Ward Casscells S. Telephone survey to assess influenza-like illness, United States. Emerg Infect Dis 2006;2008(14):129e35.
p u b l i c h e a l t h 1 2 5 ( 2 0 1 1 ) 3 8 9 e3 9 5
24. Nguyen HL, Saito R, Ngiem HK, Nishikawa M, Shobugawa Y, Nguyen DC, et al. Epidemiology of influenza in Hanoi, Vietnam, from 2001 to 2003. J Infect 2007;55:58e63. 25. Lofgren E, Fefferman NH, Naumov YN, Gorski J. Naumova1 EN. Influenza seasonality: underlying causes and modeling
395
theories: pathogen survival and transmission: social and environmental causes. J Virol 2007;81:5429e36. 26. Keech M, Scott AJ, Ryan PJJ. The impact of influenza and influenza-like illness on productivity and healthcare resource utilization in a working population. Occup Med 1998;48:85e90.
available at www.sciencedirect.com
Public Health journal homepage: www.elsevier.com/puhe
Original Research
A population-based study on incidence and economic burden of influenza-like illness in south China, 2007 R.N. Guo, H.Z. Zheng*, J.S. Li, L.M. Sun, L.H. Li, J.Y. Lin, J.F. He Institute of Epidemic Disease Prevention and Control, Centre for Disease Control and Prevention of Guangdong Province, No. 176 Xin Gang Xi Lu, Guangzhou, 510300, People’s Republic of China
article info
summary
Article history:
Objectives: The disease burden of influenza-like illness (ILI) in most tropical and subtropical
Received 3 March 2010
countries has not been described adequately to date. The aim of this study was to deter-
Received in revised form
mine the epidemiology and incidence of ILI, and to assess the economic burden in south
21 December 2010
China.
Accepted 5 March 2011
Study design: Prospective study.
Available online 25 May 2011
Methods: A population-based household survey was conducted quarterly in 2007 in Guangdong Province.
Keywords:
Results: The average number of subjects in each quarterly survey was 13,687. In total, 1002
Population study
cases of self-reported ILI were identified in all four surveys, indicating an annual incidence
Influenza
of 7.23 cases of ILI per 100 persons. The second quarter had the highest incidence of ILI
Incidence
(2.83 cases per 100 persons). Children aged 1e4 years, <1 year and 5e9 years had the
Disease burden
highest annual incidence rates of ILI (49.87, 35.19 and 21.24 cases per 100 persons, respectively). The incidence of ILI was significantly higher in males than females (P < 0.001), and significantly higher in rural residents than urban residents (P < 0.001). The individual cost per episode of ILI represented approximately 20% of monthly per-capita income of residents. Conclusions: The results of this large-scale household study confirm that ILI places a substantial health and economic burden on south China. Ultimately, the results of this study will provide further information for understanding the disease burden of influenza in subtropical areas. ª 2011 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved.
Introduction
loss of productivity that illness and lost work days place on employers and society as a whole.1,5e8
Substantial morbidity, hospitalization and mortality rates of influenza have been well documented in North America, Europe and certain districts in subtropical areas, such as Hong Kong.1e4 As well as the severe health impact, influenza has an extensive socio-economic burden, mainly due to the indirect
Global influenza surveillance is, to a large extent, based on sentinel surveillance of influenza-like illness (ILI). It has been reported that factors such as age, season, degree of latitude and population density could affect the magnitude of influenza activity. However, whole-population data are still scarce,
* Corresponding author. Tel.: þ86 20 84181985; fax: þ86 20 84193323. E-mail address: [email protected] (H.Z. Zheng). 0033-3506/$ e see front matter ª 2011 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.puhe.2011.03.004
390
p u b l i c h e a l t h 1 2 5 ( 2 0 1 1 ) 3 8 9 e3 9 5
with population-based ILI surveillance practiced in certain age groups.9e14 It has been hypothesized that south China is an epicentre for influenza,15,16 close to Hong Kong and Macao Special Administrative Region, Guangdong Province have large transient populations. As such, a collaboration centre was established in 2006 between Guangdong Province and the World Health Organization (WHO) for emerging infectious disease surveillance, research and training. The aim of this collaboration was to share Guangdong’s experience in emerging infectious disease surveillance with south China as a whole and neighbouring subtropical countries. Guangdong Province has conducted routine influenza surveillance since the 1970s. In 1998, it joined the People’s Republic of China influenza surveillance network, and has subsequently been conducting year-round influenza surveillance. For years, Guangdong Province has sent influenzavirus-positive specimens to the Chinese National Influenza Centre, which sends viral isolates to WHO. Specimens from Guangdong Province represent a significant proportion of the viral isolates that China shared with the WHO Global Influenza Surveillance Network. Guangdong Province currently maintains a surveillance network that includes 13 national and provincial level cities, 29 sentinel hospitals of all levels and 23 community/school clinics.17,18 This sentinel surveillance is useful for monitoring the patterns and circulating strains in Guangdong Province, but is limited in its ability to accurately estimate the burden or cost of disease. This is due, in part, to the fact that a certain proportion of influenza cases may not see a doctor during their illness, and a population-based study is needed to estimate the true incidence in the population. As such, a population-based door-to-door survey of self-reported ILI was conducted in 2007 to determine the incidence and epidemiological features of ILI in the community, and to gather background data to assess of the burden of influenza. Furthermore, investigations were made into medical care costs and days lost due to illness in order to understand the economic burden of ILI on residents in the community.
Methods Study population and sampling method Guangdong Province is one of the largest provinces in south China, with a total area of 177,900 km2. It has a typical tropical humid monsoon climate, with average annual temperature of 22 C and average rainfall of 1500e2000 mm. Guangdong Province has a population of 91,940,000 according to a population sampling survey in 2005. Based on multiple factors including geographic location, percapita gross national product and compliance with influenza surveillance work in recent years, Zhuhai and Zhaoqing Prefectures were selected as the two survey cities. The multistage stratified cluster sampling method19 was used to select one urban district and one rural county at random from each city. Two community residents committees were selected for each selected urban district. For each selected rural county, three towns were chosen, from which two administrative villages were
drawn. In total, eight urban community committees and 12 rural villages were selected at random. The door-to-door survey was conducted with each household as a survey unit.
Case definitions In this study, self-reported ILI was defined as fever with a measured body temperature >38.0 C and symptoms of either cough or sore throat. As a small number of subjects did not use thermometers, those who believed that they had a fever with a cough or sore throat were also considered as cases. New onset of ILI was defined as absence of the symptoms for at least 7 days prior to the onset of ILI.
Sample size calculation and sample assignment The sample size was calculated using a simple random sampling formula. The urban and rural groups were stratified. With a predicted ILI incidence rate per quarter (P) of 5% and allowable error (d) of 15%*P,20 a sample size of 3400 persons for each group was calculated. The sample size was increased by a factor of 1.1 to allow for lack of follow-up and non-response. The sample size was further enlarged by a factor of 1.5, to allow for case clustering. As such, a sample size of 5600 for each group (urban and rural) was determined. In total, it was anticipated that 11,200 persons would be surveyed each quarter. In accordance with the reported mean household population of 3.2 persons per household in the urban areas and 3.7 persons per household in the rural areas,21 220 households per urban community committee and 130 households per rural village were surveyed.
Data collection method Standardized questionnaires were designed. The household survey was administered by a researcher with a health background who received training. For each selected household, the head of the household was interviewed on behalf of all family members; if the head of the household was absent, another family member aged >18 years was interviewed. All subjects were enrolled in December 2006, and completed a registration form with demographic information. A mercurial thermometer for measuring axillary temperature and a notification letter was left for each family. The household survey was subsequently undertaken among the same cohort between 15 and 30 March, June, September and December of 2007. The effect of loss to follow-up (ranged from 2.4% to 4.1% in this study) was reduced by recruiting a number of subjects with the same demographic information to guarantee stability of the survey cohort. Interviews were conducted with each detected case of ILI by a trained surveyor, collecting information such as clinical characteristics, onset of illness within 3 months, medical care costs and days lost due to illness.
Calculation for incidence of ILI and epidemiological analysis This population-based year-round household survey investigated the incidence of ILI for the entire population (different genders, ages, occupations and education levels) living in
391
p u b l i c h e a l t h 1 2 5 ( 2 0 1 1 ) 3 8 9 e3 9 5
urban and rural areas in different seasons. Quarterly incidence rates of ILI were calculated. The annual incidence rate was obtained by calculating the cumulative incidence of ILI, with the total number of new ILI cases identified during the four quarters as the numerator, and the mean surveyed population over the four quarters as the denominator. Morbidity rates for ILI by person, season and area (urban/rural) were analysed. An estimate of the annual number of ILI patients in the whole province was obtained by multiplying the incidence among the surveyed population by the mid-2007 population size. The annual numbers of urban and rural ILI patients were calculated in the same way.
Medical care costs for ILI Medical costs included registration fee, prescribed drug fee, examination and treatment fee etc. In total, 811 self-reported ILI cases with valid medical cost data were detected during the four quarterly surveys. Mean cost per ILI episode, proportion of different costs and the medical cost difference between urban and rural areas were calculated. The exchange rate for Chinese Yuan to U.S. Dollars was 0.146. The currency units described in this article have been converted to U.S. Dollars.
Days lost due to illness In the face-to-face interviews, each identified ILI patient was asked about days lost due to illness. For employed persons, days lost represented lost work days. For those without a job, days lost were calculated as lost ‘normal’ days, and for students or children at kindergarten/childcare centre, days lost were lost school days. In total, 922 cases had valid data for lost time. For these patients, the mean number of days lost, the proportion of different types of days lost, and the difference in days lost between urban and rural areas were calculated.
Statistical analysis Frequencies were used for descriptive analyses, and Chisquared test was used to compare the proportions of different
Table 1 e Demographics of the surveyed subjects, household survey, 2007, Guangdong Province, China. No. surveyed First quarter
Second quarter
Third quarter
Fourth quarter
7054 7082 14,136
6814 6838 13,652
6772 6803 13,575
103 676 928 2084 8314 2031 14,136
103 661 899 2002 7997 1990 13,652
108 660 899 1985 7951 1972 13,575
7152 6984 14,136
6906 6746 13,652
6811 6764 13,575
Gender Male 7050 Female 7054 Subtotal 14,104 Age (years) <1 106 1～4 674 5～9 928 10～19 2075 20～59 8286 60 2035 Subtotal 14,104 Area of residence Urban 7045 Rural 7059 Subtotal 14,104
variables. ManneWhitney’s non-parametric test was adopted to compare medical costs between urban and rural areas. P < 0.05 was considered to be statistically significant for twotailed testing.
Results Demographics of surveyed subjects In total, 14,104 persons were surveyed in the first quarter of 2007, 14,136 were surveyed in the second quarter, 13,652 were surveyed in the third quarter, and 13,575 were surveyed in the fourth quarter. Of the average of 13,867 person-years of exposure, 1002 cases of self-reported ILI were identified over the four quarters, resulting in an annual incidence rate of 7.23 per 100 persons (1002/13,867). Gender, age and area of residence (urban/rural) of the surveyed subjects are shown in Table 1.
Table 2 e Gender-specific incidence of self-reported influenza-like illness (ILI) in the four quarterly surveys, 2007, Guangdong Province, China. Male
First quarter Second quarter Third quarter Fourth quarter Annual totalb
Female
Total
c2
Pvalue
No. surveyed
No. ILI
Incidence (%)a
No. surveyed
No. ILI
Incidence (%)a
No. surveyed
No. ILI
Incidence (%)a
7050
157
2.23
7054
116
1.64
14,104
273
1.94
6.303
0.012
7054
223
3.16
7082
177
2.50
14,136
400
2.83
5.633
0.018
6814
122
1.79
6838
96
1.40
13,652
218
1.60
3.245
0.072
6772
51
0.75
6803
60
0.88
13,575
111
0.82
0.695
0.405
6923
553
7.99b
6944
449
6.47b
13,867
1002
7.23b
11.977
0.001
a Unit of ‘%’ is the shortened expression of cases of ILI per 100 persons per quarter. Annual incidence is cases of ILI per 100 persons per year. b Annual incidence rate was calculated based on number surveyed in all four quarters.
392
7.23 1002 13,867 2.99 60 2007 2.83
c Chi-squared test found a significant difference in the incidence of ILI between age groups (c2 ¼ 2497.860, P ¼ 0.000).
b Annual incidence rate was calculated based on number surveyed in all four quarters.
a Unit of ‘%’ is the shortened expression of cases of ILI per 100 persons per quarter. Annual incidence is cases of ILI per 100 persons per year.
230 8137 7.22 147 21.24 Annual totalb
105
38
36.19
668
333
49.87
914
194
2037
0.82 111 13,575 0.51 10 1972 0.18 14 7951 0.76 15 3.67 3 108
2.78
660
36
5.45
899
33
1985
1.6 218 13,652 0.7 14 1990 0.53 42 7997 1.55 31 5.12 8 103
7.77
661
77
11.65
899
46
2002
2.83 400 14,136 0.84 17 2031 1.41 117 8314 3.07 64 7.87 15 103
Second quarter Third quarter Fourth quarter
14.56
676
114
16.86
928
73
2084
1.94 273 14,104 0.93 19 2035 0.69 57 8286 1.78 37 4.53 42 928 15.73 106 674 11.32 12 106 First quarter
No. surveyed
No. ILI
Incidence (%)a
No. surveyed
No. ILI
Incidence (%)a
No. surveyed
Incidence (%)a
2075
No. ILI No. ILI Incidence (%)a No. ILI
Incidence of self-reported ILI in different seasons
No. ILI
No. surveyed
No. ILI
Incidence (%)a
No. surveyed
20～59 10～19 5～9 1～4 <1
Table 3 e Age-specific incidence rate of influenza-like illness (ILI), household survey, 2007, Guangdong Province, China.
No. surveyed
60
Incidence (%)a
No. surveyed
Totalc
Incidence (%)a
p u b l i c h e a l t h 1 2 5 ( 2 0 1 1 ) 3 8 9 e3 9 5
Annual incidence rates of ILI for males and females were 7.99 and 6.47 cases per 100 persons, respectively; this difference was significant (c2 ¼ 11.977, P ¼ 0.001). Chi-squared tests were performed to determine the difference in incidence between the genders for each quarter. A difference was found in the first and second quarters, but not in the third and fourth quarters. A difference in incidence was also found between quarters (c2 ¼ 162.96, P ¼ 0.000). The highest incidence (2.83 per 100 persons per quarter) was seen in the second quarter of 2007 (Table 2). There was a significant difference in the incidence of ILI between age groups (c2 ¼ 2497.860, P ¼ 0.000). The three highest annual incidence rates occurred in children aged 1e4 years, <1 year and 5e9 years (49.87, 36.19 and 21.24 cases of ILI per 100 persons, respectively). Annual incidence of ILI was 7.22 per 100 persons for those aged 10e19 years, and 2.99 per 100 persons for those aged >60 years. The lowest annual incidence of ILI was found among adults aged 20e59 years (2.83 per 100 persons; Table 3). Age-stratified incidence analysis by area of residence was undertaken. The annual incidence of ILI was 5.66 and 6.91 cases per 100 persons in urban and rural areas, respectively; this difference was significant(c2 ¼ 18.653, P ¼ 0.000). Occupation analysis for ILI patients showed that the most affected population was preschool children, with an annual incidence rate of 42.08 cases per 100 persons, followed by students (10.73 cases per 100 persons), peasants (3.85 cases per 100 persons) and health workers (3.81 cases per 100 persons). Education level was also a contributing factor for ILI. The annual incidence of ILI showed a decreasing trend as education level increased, with the highest incidence (44.2 cases per 100 persons) seen in preschool children and the lowest incidence (3.18 cases per 100 persons) seen in persons who had education level above college degree.
Medical care costs and days lost due to ILI For the 1002 cases of self-reported ILI, 898 cases (89.6%) sought medical care and 10.4% chose to self-treat. For the 898 cases who sought medical care, only 811 cases had complete data regarding the consultation fee. Medical care cost analysis for these 811 self-reported ILI cases showed that one episode of illness cost approximately 172.5 Yuan (US$25.2). The mean cost for each urban inhabitant was 274.8 Yuan [US$40.12; 95% confidence interval (CI) 223.8e325.8 Yuan (US$32.7e47.6)], and the mean cost for each rural inhabitant was 107.2 Yuan [(US$15.65); 95% CI 88.6e125.8 Yuan (US$12.9e18.4)]. The Mann-Whitney test revealed that the medical care cost for each urban inhabitant was significantly higher than that for each rural inhabitant (Z ¼ 13.440, P ¼ 0.000). Nine hundred and twenty-two cases of self-reported ILI with complete data on days lost due to illness were also analysed. Among the 542 rural ILI patients, 97 cases (17.9%) lost a median of 2 work or school days. Among the urban ILI patients, 127 cases (33.7%) lost a median of 1 work or school day (Table 4).
p u b l i c h e a l t h 1 2 5 ( 2 0 1 1 ) 3 8 9 e3 9 5
Table 4 e Medical care costs and days lost due to selfreported influenza-like illness (ILI), household survey, 2007, Guangdong Province, China. Urban area No. ILI
%
Medical cost (REN MIN BI) <10 1 0.3 10～49 32 10.1 50～99 53 16.8 100～199 93 29.4 200～299 52 16.5 300～399 34 10.8 400～499 13 4.1 500～999 21 6.7 1000～1999 12 3.8 2000 5 1.6 Total 316 100.0 Lost time (days) 0 253 66.6 0.5 14 3.7 1 54 14.2 2 26 6.8 3 10 2.6 4 6 1.6 5 6 1.6 6 4 1.1 7 6 1.6 8 0 0.0 14 1 0.3 15 0 0.0 Total 380 100.0
Rural area
Total
No. ILI
%
No. ILI
%
7 225 125 75 27 13 5 9 7 2 495
1.4 45.4 25.3 15.2 5.5 2.6 1.0 1.8 1.4 0.4 100.0
8 257 178 168 79 47 18 30 19 7 811
1.0 31.7 22.0 20.7 9.7 5.8 2.2 3.7 2.3 0.9 100.0
445 7 28 24 13 11 3 7 2 1 0 1 542
82.1 1.3 5.2 4.4 2.4 2.0 0.6 1.3 0.4 0.2 0.0 0.2 100.0
698 21 82 50 23 17 9 11 8 1 1 1 922
75.7 2.3 8.9 5.4 2.5 1.8 1.0 1.2 0.9 0.1 0.1 0.1 100.0
Discussion This household survey was conducted quarterly reflecting ILI throughout 2007. The data showed that the annual incidence of ILI was 7.23 per 100 persons, with the highest quarterly incidence (2.83 per 100 persons) seen in the second quarter; this was slightly lower that that found by Claque et al.22 and Malone et al.23 The ILI activity pattern found in the present survey was consistent with local surveillance data for 2007, which revealed a peak in influenza in June. Virology monitoring identified a predominance of subtype H3 of A type and B type influenza viruses. This study clearly demonstrated that ILI activity in Guangdong Province occurs year-round, but with an evident peak in summer.17,18 This single-peak feature differs slightly from the findings of Nguyen et al., who reported two influenza peaks in south-east Asia.24 The present survey found that children <10 years of age were most affected by ILI. Annual ILI incidence among children aged 1e4 years was 50 cases per 100 persons. Primary schools and childcare centres were high-risk places for outbreaks of influenza.15,17,18 As a result, influenza prevention and control measures should focus on school-aged children, and the vaccination policy should prioritize this population. This study also found that males had a higher incidence of ILI than females, especially in the first and second quarters of the year. The reason for this difference is not yet clear and further investigation is needed.
393
Another interesting finding was that rural residents had a higher incidence of ILI than urban residents. Given a higher population density and worse air quality in urban areas, infection and transmission of influenza virus is presumably more feasible in urban areas than rural areas; accordingly, higher morbidity was expected in the urban areas.25 However, better compliance was observed among rural residents during this population-based household survey, so it is possible that this may have contributed to the higher incidence rate among rural inhabitants. The individual cost per ILI episode was US$40.12 for each urban inhabitant and US$15.65 for each rural inhabitant. This represents approximately 20% of monthly per-capita disposable income of urban residents, and 20% of monthly percapita total income of rural residents based on the Guangdong Statistic Year Book for 2006. ILI places a considerable economic burden on Guangdong communities regardless of whether they are urban or rural. Furthermore, given lost productivity due to absence from work, and other direct and indirect expenditures during the illness, such as extra nutrition costs, transportation fees and payment for nursing care, the actual economic burden per ILI episode could be much higher. Medical expenditure for urban residents is much higher than that for rural residents; this may be due to multiple aspects of healthcare-seeking behaviour, such as consultation rate, preferred prescribed drugs, preferred hospitals and ILI awareness. Among the urban ILI patients, 33.3% had lost work or school days due to illness, compared with 17.9% of rural patients. This shows that ILI had more severe consequences for urban residents than their rural counterparts. In terms of days lost, the median amount of lost time was 2 days for rural patients and 1 day for urban patients. As well as severity of illness, the position, seniority, institution type and production mode were related to the length of absence.26 Previous studies have documented the incidence, hospitalization rate, risk factors and economic burden of ILI or influenza. It has been reported that factors such as age, season, degree of latitude and population density could affect the magnitude of influenza activity.1e8 However, data targeting total populations are scarce, with population-based ILI surveillance practiced in certain age groups.9e11 In China, no systematic research has been undertaken to date on the health and economic burden of influenza or ILI on the community. The present study, based on a large-scale household survey, reports gender, age and location (urban/ rural) differences in the incidence of ILI; explores the incidence and epidemiological features of ILI in a community in south China; and estimates, for the first time, the burden that ILI places on South China. This population-based household study will provide background data for further assessments of the disease burden of influenza, and ultimately, the results of this study will provide further information for understanding the disease burden of influenza in subtropical areas. This study has several limitations. First, self-reported ILI was used as the case definition of ILI. As no virology validation was performed, the definition may capture a wide range of febrile illnesses, which may have led to to an overestimation of the incidence of ILI. Second, this large-scale survey lasted for 1 year and was administered four times. Decreased compliance
394
p u b l i c h e a l t h 1 2 5 ( 2 0 1 1 ) 3 8 9 e3 9 5
by the last stage of the survey is possible. Thirdly, only medical care costs and days lost due to illness were calculated in the evaluation of economic burden; this may have led to underestimation of the true burden of influenza as a variety of direct and indirect costs were not taken into consideration. Finally, the cohort involved a large population of approximately 14,000 persons, and loss to follow-up of a small number of subjects was inevitable. However, as the percentage of loss to follow-up was small (<5%), this is unlikely to have had a significant impact on the study results. Despite the limitations, the results of this populationbased household survey could supplement the provincial sentinel influenza surveillance system. Based on the primary results of this household survey study, it is estimated that at least 6.83 million people may have ILI in Guangdong Province. The estimated medical cost for the treatment of ILI patients would be US$155 million, given that 89% of patients seek medical care; this would constitute a significant public health problem. An appropriate estimate of the burden of influenza in Guangdong Province can be obtained by combining the population-based ILI survey and the sentinel hospital-based influenza surveillance.
Ethical approval This survey was approved by the review board of the Centre for Disease Control and Prevention of Guangdong Province, China, and oral consent was obtained from all subjects.
Funding World Health Organization.
Competing interests None declared.
Acknowledgements The authors thank Zhou Weigong for his technical support in preparing the methods and discussion sections of the text. Thanks also go to staff members of the Zhuhai and Zhaoqing Centres for Disease Prevention and Control of all levels, the local health bureau, and the trainees of the second phase of Guangdong Field Epidemiology Training Programme for their assistance and cooperation in implementation in the field survey.
references
1. Molinari NAM, Ortega-Sanchez IR, Messonnier ML, Thompson WW, Wortley PM, Weintraub E, et al. The annual impact of seasonal influenza in the US: measuring disease burden and costs. Vaccine 2007;25:5086e96. 2. Schanzer DL, Langley JM, Tam TWS. Hospitalization attributable to influenza and other viral respiratory illnesses in Canadian children. Pediatr Infect Dis J 2006;25:795e800.
3. Lenglet AD, Hernando V, Rodrigo P, Larrauri A, Donado JDM, de Mateo S. Impact of flu on hospital admissions during 4 flu seasons in Spain, 2000e2004. BMC Public Health 2007;7:197. 4. Li CK, Choi BCK, Wong TW. Influenza-related deaths and hospitalizations in Hong Kong: a subtropical area. Public Health 2006;120:517e24. 5. Espositol S. Clinical and socio-economic impact of influenza and respiratory syncytial virus infection on healthy children and their households. Clin Microbiol Infect 2005;11:933e6. 6. Li S, Leader S. Economic burden and absenteeism from influenza-like illness in healthy households with children (5e17 years) in the US. Respir Med 2007;101:1244e50. 7. Keren R, Zaoutis TE, Saddlemire S, Luan XQ, Coffin SE. Direct medical cost of influenza related hospitalizations in children. Pediatrics 2006;118:1321e7. 8. Nichol KL, D’Heilly SJ, Greenberg ME, Ehlinger E. Burden of influenza-like illness and effectiveness of influenza vaccination among working adults aged 50e64 years. Clin Infect Dis 2009;48:292e8. 9. Iwane MK, Edwards KM, Szilagyi PG, Walker FJ, Griffin MR, Weinberg GA, et al. Population-based surveillance for hospitalizations associated with respiratory syncytial virus, influenza virus, and parainfluenza viruses among young. Pediatrics 2004;113:1758e64. 10. Meier CR, Napalkov PN, Wegmu¨ller Y, Jefferson T, Jick H. Population-based study on incidence, risk factors, clinical complications and drug utilization associated with influenza in the United Kingdom. Eur J Clin Microbiol Infect Dis 2000;19:834e42. 11. Immerman JM, Lertiendumrong J, Dowell SF, Uyeki T, Olsen SJ, Chittaganpitch M, et al. The cost of influenza in Thailand. Vaccine 2006;24:4417e26. 12. Zhang J, Shi G, Shu Y, Yang W. Strategies on enhancing capacities of influenza surveillance in China. Chin Plan Immun 2005;11:339e42 [in Chinese]. 13. Duan P, Zhang R, Lin J, Zhang L, Zhao H, Yan B. Survey on incidence, absenteeism and drug use of influenza-like illness of barrack solders in 8 districts. Liber Army Prev Med 2001;19: 194e5 [in Chinese]. 14. Sun S, Yang S, Zhang H, Zhang Y, Zhang L. Impact of influenza on health of middle school and primary school student in 1998e1999 winter season of Tianjin urban district. Occup Health 2000;16:40e1 [in Chinese]. 15. Huang P. Study on epidemic, outbreak and molecular variation of influenza in Guangdong. Chin J Dis Control Prev 2004;8:144e7. 16. Shortridge KF. Pandemic influenza: a zoonosis? Semin Respir Infect 1992;7:11e25. 17. Huang P, Lin J, Ni H, Zhou H, Du Z, Liu S. Influenza analysis from 1998 to 2001 in Guangdong district. S Chin J Prev Med 2002;28:21e2. 18. Huang P, Chen W, Ni H, Zhou H, Liu S. Analysis on influenza epidemic and surveillance in Guangdong between 1988 and 1998. Strait J Prev Med 2001;7:15e7. 19. Du Z. Sampling method and its application. Beijing: Qinghua University Publishing Company; 2005. p. 8. 20. Guo R, Tang H, Kang M, Luo H, Zheng H. Telephone survey on incidence of influenza-like illness and vaccination status in community residents. Chin J Public Health 2008;24:843e5. 21. Department of Population and Employment Statistics National Bureau of Statistics of China. Communique of 1% population sampling survey in 2005 of Guangdong province. Beijing: China Population Statistics Yearbook; 2006. 22. Clague B, Chamany S, Burapat C, Wannachaiwong Y, Simmerman JM, Dowell SF, et al. A household survey to assess the burden of influenza in rural Thailand. SE Asian J Trop Med Public Health 2006;37:488e93. 23. Malone JL, Madjid M, Ward Casscells S. Telephone survey to assess influenza-like illness, United States. Emerg Infect Dis 2006;2008(14):129e35.
p u b l i c h e a l t h 1 2 5 ( 2 0 1 1 ) 3 8 9 e3 9 5
24. Nguyen HL, Saito R, Ngiem HK, Nishikawa M, Shobugawa Y, Nguyen DC, et al. Epidemiology of influenza in Hanoi, Vietnam, from 2001 to 2003. J Infect 2007;55:58e63. 25. Lofgren E, Fefferman NH, Naumov YN, Gorski J. Naumova1 EN. Influenza seasonality: underlying causes and modeling
395
theories: pathogen survival and transmission: social and environmental causes. J Virol 2007;81:5429e36. 26. Keech M, Scott AJ, Ryan PJJ. The impact of influenza and influenza-like illness on productivity and healthcare resource utilization in a working population. Occup Med 1998;48:85e90.