PLoS One. 2014 Jan 8;9(1):e84239.

The temporal trend of influenza-associated morbidity and the impact of early appearance of antigenic drifted strains in a Southeast Asian country.

Lian IeB1, Wu HD2, Chang WT1, Chao DY3.


1 Graduate Institute of Statistics and Information Science, National Changhua University of Education, Changhua, Taiwan.

2 Department of Applied Math, National Chung-Hsing university, Taichung, Taiwan ; Institute of Statistics, National Chung-Hsing university, Taichung, Taiwan.

3 Graduate Institute of Microbiology and Public Health, College of Veterinary Medicine, National Chung-Hsing university, Taichung, Taiwan.



Globally, influenza infection is a major cause of morbidity and mortality in the elderly, who are suggested to be the major target group for trivalent influenza vaccine (TIV) vaccination by World Health Organization. In spite of an increasing trend in vaccine coverage rates in many countries, the effect of vaccination among the elderly in reducing hospitalization and mortality remains controversial. In this study, we conducted a historical cohort study to evaluate the temporal pattern of influenza-associated morbidity among persons older than 64 years over a decade. The temporal patterns of influenza-associated morbidity rates among the elderly older than 64 years indicated that Taiwan’s elderly P&I outpatient visits have been decreasing since the beginning of the 1999-2000 influenza season; however, hospitalization has been increasing despite significant increases in vaccine coverage. The propensity score logistic regression model was implemented to evaluate the source of bias and it was found that the TIV-receiving group had a higher propensity score than the non-receiving group (P<0.0001). In order to investigate the major factors affecting the temporal pattern of influenza-associated morbidity, we then used the propensity score as a summary confounder in a multivariate Poisson regression model based on the trimmed data. Our final models suggested that the factors affected the temporal pattern of morbidity differently. The variables including co-morbidity, vaccination rate, influenza virus type A and B isolation rate were associated with increased outpatient visits and hospitalization (p<0.05). In contrast, variables including high propensity score, increased 1°C in temperature, matching vaccine strains of type A/H1N1 and type B were associated with decreased outpatient visits and hospitalization (p<0.05). Finally, we assessed the impact of early appearance of antigenic-drifted strains and concluded that an excess influenza-associated morbidity substantial trends toward higher P&I hospitalization, but not outpatient visits, during the influenza season with early appearance of antigenic-drifted strains.

PMID: 24416205



Infection by influenza viruses is a major cause of morbidity and mortality among all age groups globally, particularly in the group of 65 years and older. The way that influenza viruses can persistently circulated among human population involves two mechanisms: antigenic drift or antigenic shift. Antigenic drift is a mechanism for variation in viruses that involves the accumulation of mutations within the genes that code for antibody-binding sites. It results in a new strain of virus particles which cannot be inhibited as effectively by the antibodies generated by either previous infection or vaccination targeted against previous strains, making it easier for the virus to spread throughout a partially immune population. Antigenic shift is the process by which two or more different strains of influenza viruses, combine to form a new subtype having a mixture of the surface antigens of the two or more original strains. The example of antigenic shift will be 2009 pandemic influenza virus or recent discovered H7N9 influenza virus. As of antigenic shift, which occurred historically every 10 years, antigenic drift occurs every 2-3 years and the vaccines currently used against seasonal influenza contain antigens against three influenza strains (A/H1N1, A/H3N2 and B), which are altered yearly to target the strains that are predicted to circulate in the upcoming season by WHO. However, in the region of Southeast Asia and Recent surveys of influenza viruses in Taiwan by hemagglutination (HA) sequence comparisons have indicated that a high rate of vaccine mismatch and found that epidemic strains in this region often became the vaccine strains 2–3 years later for some unknown reasons. Generally, antigenic alterations of local strains that make existing antibody levels in the population no longer protective are considered to be related to disease severity, increased medical utility and vaccine efficacy.

The most effective way to combat against influenza virus infection is through vaccination. The yearly influenza vaccination of at-risk individuals became common practice worldwide after the Second World War. After 2000, more than 40 developed or rapidly developing countries recommended vaccination to prevent influenza and its complications for “high risk” groups, such as the elderly (65 years or older), patients with chronic conditions, and institutionalized populations. Nevertheless, vaccination of the elderly remains controversial. In spite of an increasing trend in vaccine coverage rate, pneumonia and influenza-associated hospitalization and mortality among the elderly have continued to rise in Italy, the United States and South Korea.

Similar results were also found in our temporal trend analysis of influenza-associated hospitalization. During year 2000-2009, the influenza outpatient visit rate gradually decreased for all age groups; however, the hospitalization rate gradually increased for only the older age groups (Fig 1). During the same time period, the yearly vaccination rates among the elderly older than 64 years of age increased. Although we don’t know the vaccination rate for age under 65 years old as it is relatively difficult to obtain, the increase of hospitalization rate instead of outpatient visit rates could be due to the change of the medical behaviors as the physicians tend to admit the elder patients into hospitals because of the insurance program or the increasing frailty/comorbidity of the elderly group.

Assessment of the long-term influenza vaccine effectiveness, defined by medical uses including outpatient visits and hospitalization, is complicated by the degree of match between vaccine and virus, vaccine delivery policy, individual vaccination history, gradual decline in immune competence with age, antibody quality and quantity providing cross-reactivity, medical accessibility, co-morbidity among the elderly and personal behaviors. In Taiwan, several policies implemented here make it advantageous to assess the long-term influenza vaccine effectiveness in the elderly including (1) the high coverage rate (over 97% at the end of 2003) of National Health Insurance (NHI) program implemented in Taiwan since 1995; (2) the Taiwan Centers for Disease Control (Taiwan-CDC) has coordinated a laboratory-based influenza virological surveillance network (Lab-ISN) starting in 2000 for providing up-to-date information on viral characteristics and activities; (3) The Taiwan program of targeted free influenza vaccination for people with underlying medical disorders was implemented since 1996 and further expanded in 1998 to include all people older than 64 years. Vaccines are delivered in health care settings by nurses or physicians, or in community settings through public health departments.

Although the vaccine clinical trial studies are most powerful in evaluating efficacy or effectiveness through randomization, it is expensive, smaller population and complicated by the circulation of influenza viruses matched with the vaccine strains. Using NHI research database could avoid the above problems but careful adjusting confounding factors will be crucial since the comparison groups are not randomized. Therefore, when we analyzed further using the poisson regression model, a propensity score was calculated to understand the unknown source of bias between vaccinated and un-vaccinated groups among the elderly group. We extracted as many as possible variables from the NHI research database regarding medical use behavior, medical condition, residential places and determine the most appropriate variables as surrogates to differentiate two groups.

After we finish the poisson regression model and use the model to predict the temporal trend of influenza morbidity, including outpatient visits and hospitalization, the graph was plotted against the observed numbers from the database, we could actually observe an unusual discrepancy during each epidemic and it was correlated with the virus isolation data obtained through Lab-ISN (Fig 2). This gave us an idea to calculate the excess morbidity by subtracting predicted morbidity numbers from the observed ones and the excess morbidity correlated with the appearance of antigenic-shifted influenza viruses. We found that an excess influenza-associated morbidity substantial trends toward higher P&I hospitalization, but not outpatient visits, during the influenza season with early appearance of antigenic-drifted strains.

In summary, our study concluded that the early appearance of antigenic-drifted strains of influenza viruses was correlated with the increased risk of P&I-associated hospitalization among the elderly. The inclusion of local strains which appeared antigenic-drifted based on the surveillance system in the composition of TIV among the elderly warrants careful consideration particularly in East or Southeast Asian countries.

fig-1Fig 1. The influenza outpatient visit rate and the hospitalization rate During year 2000-2009

Fig 2. Excess peak of influenza cases during 2000-2009 in Taiwan


Multiselect Ultimate Query Plugin by InoPlugs Web Design Vienna | Webdesign Wien and Juwelier SchönmannMultiselect Ultimate Query Plugin by InoPlugs Web Design Vienna | Webdesign Wien and Juwelier Schönmann