Epidemiologic Profile of a New H3N2 Variant of Influenza A Mismatched to Vaccine, 2003-2004 Influenza Season

Introduction

During the 2003-2004 influenza season, A/Fujian/411/2002-like virus emerged as a new antigenic variant of influenza A(H3N2) virus⁽¹⁾, but was not included as a vaccine component⁽²⁾. Reports from the southern hemisphere and areas in Europe and North America with early influenza activity indicated that the A/Fujian strain was causing widespread morbidity with increased rates of severe complications and mortality, particularly among children. By the end of November, the United Kingdom reported five deaths in children caused by A/Fujian/411/2002-like virus⁽¹⁾, Colorado reported three pediatric influenza-related deaths⁽³⁾, and Ontario reported one childhood influenza-related death.

The British Columbia Centre for Disease Control (BCCDC) has a multifaceted influenza surveillance system designed to detect early influenza season onset and to monitor influenza virus spread, evolution and population impact. The following describes the epidemiologic profile associated with the emergence of A/Fujian/411/2002-like virus based on BCCDCs influenza surveillance system.

Methods

The BCCDC coordinates influenza surveillance activities for the province of British Columbia (BC), analyzing multiple indicators to detect and monitor influenza activity.

Sentinel physicians

BCCDC, in conjunction with the College of Family Physicians of Canada, has managed a sentinel physician network since 1977. During the 2003-2004 influenza season, 44 physicians participated in sentinel surveillance for influenza-like illness (ILI) recording the total number of patients seen every week and the number presenting with ILI. Physicians also reported the vaccination status and age of each ILI patient. ILI is defined as the acute onset of respiratory illness with fever and cough and with one or more of sore throat, arthralgia, myalgia or prostration that could be due to influenza virus. In children < 5 years, gastrointestinal symptoms may also be present. In patients < 5 years or >= 65 years, fever may not be prominent.

Each week the proportion of visits to a sentinel physician due to ILI is compared to the average proportion from the same week from historic data available in 2003-2004 for the previous 14 years.

Laboratories

The BCCDC Virology Laboratory provides weekly reports of the number of respiratory virology tests performed and the number positive for influenza A, influenza B, respiratory syncitial virus (RSV), adenovirus, and parainfluenza 1, 2, 3 and 4. The Laboratory forwards influenza isolates to the National Microbiology Laboratory (NML) for further strain characterization.

ILI outbreaks

Health regions report ILI outbreaks in long-term care facilities (LTCF), schools and workplaces to BCCDC. In LTCFs, a trigger point for possible outbreak reporting is reached when two or more cases of ILI within a 7-day period have occurred within the facility. In schools and workplaces, an outbreak is defined as > 10% absenteeism on any day, most likely due to ILI.

After outbreaks are declared over, health regions complete summary outbreak reports that include information on the institution type, number of people at risk, number ill, clinical course of illness, causative organism, outbreak management and health outcomes.

Vital statistics

The BC Vital Statistics Agency provides monthly reports on mortality rates from influenza alone (ICD-9 codes: 4870, 4871, 4878, 4879 and ICD-10 codes: J100, J101, J1010, J108, J11, J110, J111, J1110, J118) and pneumonia (P) (ICD-9 codes: 4800-4802, 4808, 4809, 4820-4824, 4828, 4829, 483, 4840-4848, 485, 486 and ICD-10 codes: J120-J122, J128, J129, J14, J150-J160, J168, J180, J181, J188, J189 ) and influenza (I) combined (P&I). The rates are expressed as the number of deaths per 100,000 population. Deaths are considered to be influenza- or pneumonia-related if either is recorded on the death certificate as an underlying or contributing cause of death. The BC vital statistics reports compare influenza- and P&I-related mortality rates to the previous month and to the average rate for the same month over the previous 17 and 14 years respectively (with 95% confidence intervals). The reports also include the number of deaths per month by age group.

Enhanced pediatric surveillance

Based on reports of severe pediatric outcomes of influenza from elsewhere, the routine influenza surveillance system in BC, described above, was further enhanced in 2003-2004 to include special reporting of severe disease among children. Electronic bulletins to physicians requested heightened awareness, laboratory testing for influenza and reports submitted to provincial authorities regarding children presenting with severe or unusual symptoms of ILI requiring hospitalization or resulting in death.

Results

Sentinel physicians

For the first 6 weeks of the 2003-2004 influenza season (weeks 40 to 46; 28 September to 8 November), the weekly proportions of sentinel physician visits due to ILI were within or below the historic ranges (Figure 1a). In week 46 this proportion (0.96%) surpassed the historic range (0.39% to 0.73%) and remained significantly higher than historic for 5 weeks (9 November to 13 December). The proportion (0.35%), then, dropped significantly below the historic range (0.85% to 1.77%) in week 1 (28 December to 4 January 2004), and remained within or below historic values for the rest of the season.

ILI visit reports peaked in week 49 (30 November to 6 December) at 1.87%. Although significantly higher than historic for week 49 (0.59% to 1.06%), this proportion was within the historic range for week 52 (1.00% to 2.31%), the week in which peak ILI activity typically occurs.

In 2003-2004, school aged children and working-aged adults comprised larger proportions of ILI cases than usual and those aged >= 65 years comprised a smaller proportion than usual (Figure 1b). Of the 955 ILI cases with known ages, 142 (15%) were < 5 years, 278 (29%) were aged 5 to 19 years, 477 (50%) were 20 to 64 years, and 58 (6%) were >= 65 years. This age distribution is significantly different (chi squared = 97.8, p < 0.001) from the 10-year historic averages of 18%, 21%, 45% and 16% respectively, for these age groups.

Of the 939 ILI cases with known vaccination status, 65 (6.9%) were vaccinated. This is within the historic range of vaccine uptake among ILI patients from the previous 10 years (mean 7.0, range 5.2% to 8.8%). Vaccine uptake among ILI cases was higher than the historic average among 20 to 64 years (9.1% vs. mean 5.9%, range 3.2% to 8.7%), lower than the historic range among those >= 65 (24.5% vs. mean 40.7%, range 26.8% to 54.6%), and was within the historic ranges < 5 and 5 to 19 years [0.7% and 3.0% vs. 0.4% (1.1% to 1.9%) and 1.3% (-0.8% to 3.4%), respectively].

Figure 1a. Proportion of sentinel physician visits due to influenza-like illness per week, British Columbia, 2003-2004

Figure 1b. Proportion of ILI cases in each age group, sentinel physicians, British Columbia, 2003-2004

Laboratories

Between 29 September, 2003 and 25 September, 2004, the BCCDC Virology Laboratory tested 4,707 respiratory specimens. Twenty percent (959) were positive for respiratory viruses (Figure 2). Of these, 754 (79%) were influenza A, 2 (< 1%) were influenza B, 59 (6%) were RSV, and 144 (15%) were adenovirus or parainfluenza 1, 2, 3 or 4.

Between 29 September, 2003 and 26 September, 2004, a subset of influenza isolates from BC was characterized by the NML. Of the 121 characterized, 115 (95.0%) were A/Fujian/411/02-like, five (4.1%) were A/Panama/2007/99-like and one (0.8%) was A/New Caledonia/20/99-like.

Figure 2. Viruses isolated and percent of specimens testing positive for respiratory viruses by week, British Columbia, 2003-2004

ILI outbreaks

BC health regions reported 176 ILI outbreaks in the 2003-2004 influenza season: 107 in schools; 60 in LTCFs; eight in acute-care hospitals; and one in a daycare (Figure 3). The first two outbreaks were reported in week 43 (19 to 25 October), both in LTCFs. The number of outbreaks reported per week peaked in week 49 with 30 outbreaks (25 in schools, four in LTCFs and one in an acute-care hospital). The majority of outbreaks reported in weeks 44 through 50 (26 October to 13 December) occurred in schools. From week 51 onward, the majority occurred in LTCFs.

The causative organism was identified in 61 (35%) of the outbreaks, including all of those in the acute-care hospitals, 48 (80%) of those in LTCF 4, (4%) of the school outbreaks and the one daycare outbreak. Laboratory testing of school-related outbreaks is not routine. Influenza A was responsible for 57 (93%) of the outbreaks for which an etiology was identified. Other etiologies included adenovirus, parainfluenza, RSV and norovirus.

Figure 3. Number of influenza-like illness (ILI) outbreaks reported and ILI rates for the past 14 years byweek, British Columbia, 2003-2004

Vital statistics

Between October 2003 and July 2004, 3,537 deaths were related to (P&I) in BC. Sixty-six of these were related to influenza. Historically, between October and July an average of 3,252 P&I-related deaths and 65 influenza-related deaths are reported in BC.

From October 2003 through January 2004, the P&I-related mortality rates were higher than historic rates (Figure 4a). This increase was statistically significant in the month of December (12.2/100,000 vs. historic 9.4/100,000). Historically, P&I-related mortality peaks in January at an average rate of 11.2/100,000. The 2003-2004 peak P&I-related mortality rate in December did not differ significantly from the historic peak rate. The overall P&I-related mortality rate for October 2003 through July 2004 was 85.4/100,000 which is within the historic range of P&I mortality for the months of October through July (85.0/100,000; 95% CI = 82.0-87.9/100,000).

When only influenza was examined, influenza-related mortality rates were higher than historic rates in November and December 2003 (Figure 4b). This difference was statistically significant in the month of December (0.7/100,000 vs. historic 0.2/100,000). Historically, this rate peaks in January at 0.5/100,000. The difference between the 2003-2004 and historic peak influenza-related mortality rates was not statistically significant in terms of rate although the 2003-2004 peak was earlier than the historic peak. The overall influenza-related mortality rate for October 2003 through July 2004 was 1.6/100,000, which is within the historic range of influenza mortality for the months of October through July (1.8/100,000; 95% CI = 4-2.2/100,000).

Although the mortality rate was within the historic range for influenza-related deaths in November 2003, nine influenza-related deaths was the second highest total observed in November in recorded BC data (since 1986-1987). The highest was 13 influenza-related deaths in November 1999. The 27 influenza-related deaths recorded in December 2003 was the highest number of deaths observed in December in recorded BC data.

Of the 66 influenza-related deaths in BC between October 2003 and July 2004, 63 were >= 65 years, two were aged 20 to 64 years, and one was aged 0 to 9. Historically, during the same months, an average of 60 individuals >= 65, two individuals aged 20 to 64, 0.4 individuals aged 10 to 19 years and 0.4 individuals aged 0 to 9 years die with influenza recorded as a related cause of death.

Figure 4a. Mortality rates for pneumonia- and influenza-related deaths, British Columbia, 2003-2004

Figure 4b. Mortality rates for influenza-related deaths, British Columbia, 2003-2004

Enhanced surveillance for morbidity and mortality in children

Three pediatric deaths were reported to the BCCDC through enhanced surveillance a 2-year-old male, a 7-year-old male, and a 7-year-old female. Follow-up investigations determined that the death of the 2-year-old was not related to influenza.

The other two children had both been previously healthy and both demonstrated influenza-like symptoms prior to death. There was also no other epidemiologic link between these two children. The girl experienced fever and vomiting that resolved on 12 December. The next day, she developed a fever and leg pain, and died in hospital on 14 December. The boy experienced fever, vomiting, anorexia and difficulty breathing commencing 12 December. On 14 December he complained of abdominal discomfort and was taken to hospital. He turned blue, became unresponsive, and died later that evening.

Autopsy determined that both of these children died from myocarditis, a known but rare complication of influenza, and other viral infections. Influenza A/Fuijian/H3N2 was isolated from tracheal aspirates from the girl and the death was attributed to influenza. Lung, liver, plasma, bronchi and brain samples from the boy all tested negative for influenza A and B, and the death was considered "suspect" influenza.

Discussion

During the 2003-2004 influenza season in BC, the predominant respiratory virus detected was influenza A. The dominant influenza A subtype was a new H3N2 drift variant, A/Fujian/411/ 2002-like that was not included as a vaccine component. According to the summary indicators from the BC surveillance system, the epidemiologic profile associated with this mismatched strain was from an earlier outbreak that had an overall impact comparable to previous seasons. As in previous seasons and despite reports of pediatric deaths elsewhere, elderly people in BC suffered the greatest consequences in terms of mortality.

The very old and very young are more likely to experience complications of influenza which may lead to their being over- represented in a sample of sentinel physician visits. The sentinel physician reports for the 2003-2004 season suggest that influenza A/Fujian may have affected persons 5 to 64 years more significantly than strains in previous years. School-aged children and working-aged adults are expected to have more extensive social networks than the very old or very young thus affecting their susceptibility disproportionately during the first season of circulation of a new variant. Despite this, the increased mortality seen in European children was not evident in BC in 2003-2004.

The overall P&I-related and influenza-related mortality rates were within the historic ranges. Individuals > 65 accounted for 95% of recorded deaths related to influenza. The number of deaths in each age group was comparable to the average number from previous years. Despite enhanced efforts to detect severe influenza disease and/or fatalities in the pediatric age group, no more than one childhood influenza-related death was confirmed in BC. With an historic average of 0.4 influenza-related deaths per year in children 0 to 9 years, one death was not unexpected. Rare but serious outcomes become important when amplified by a high attack rate such as that observed in healthy school-aged children.

The 1997-1998 influenza A epidemic in BC was also caused by an emerging influenza A (H3N2) strain that was mismatched to the vaccine-influenza A/H3N2/Sydney/05/97-like virus⁽⁴⁾. Unlike the 2003-2004 season, influenza activity was delayed in 1997-1998⁽⁴⁾. Sentinel physicians reported peak ILI activity in week 8, and the peak proportion of visits due to ILI was significantly higher than the historic peak rate⁽⁴⁾. However, similar to the 2003-2004 influenza season, more than 99% of laboratory-confirmed influenza was influenza A⁽⁴⁾. The A/Sydney outbreak had a more severe impact on mortality than did A/Fujian. The P&I- and influenza-related mortality rates from October 1997 to July 1998 were 88.7/100,000 population and 3.9/100,000 population respectively, compared to 85.4/100,0000 and 1.6/100,000 in 2003-2004.

In 1997-1998, 125 school and 62 LTCF outbreaks were reported, compared to the 107 school, 60 LTCF, eight acute-care and one daycare outbreak reported in 2003-2004⁽⁴⁾. In both years, school outbreaks provided the earliest signal of the start of the influenza season followed several weeks later by more general community activity and signals from the sentinel physician system. This is the usual pattern of influenza activity. Reporting of school outbreaks early in the influenza season should be encouraged as a way to gain lead-time in reinforcing and implementing prevention measures, such as vaccination campaigns, among high-risk populations.

The National Advisory Committee on Immunization has recently recommended routine influenza vaccination for children 6 to 23 months of age because of increased risk of hospitalization⁽⁵⁾. This surveillance summary found no deaths due to influenza in infants/toddlers, but hospitalization data was not included - an outcome of influenza particularly noteworthy in infants/toddlers. Sentinel physician data indicate that a significantly smaller proportion of ILI visits in 2003-2004 were by children < 5 years, although finer breakdowns (e.g., < 2) are not currently possible. Sentinel physician and school outbreak reports suggest school-aged children might also be an additional target group. This is supported by high attack rates in this group reported in other studies⁽⁶⁾. The only pediatric death in BC last season occurred in a school-aged child.

These surveillance data are limited as they do not measure morbidity in the entire population. Sentinel physician data measure morbidity among individuals who seek medical care, and the laboratory data reflect patients who have specimens taken. The surveillance system misses the less severe end of the disease spectrum, and thus cannot fully describe the range of illness caused by the circulating strain. However, these data do provide indicators of influenza activity in the province, as well as more detail about circumstances of greatest concern, i.e., ILI outbreaks and deaths related to influenza.

The use of vital statistics data as a measure of influenza activity has been criticized because influenza is seldom recognized as an underlying cause of death. More proximate causes, such as bacterial infections, tend to be recorded preferentially. Despite this underreporting, the mortality rates related to P&I and influenza alone mirrored influenza activity in the community, peaking in December. Although they do not represent absolute numbers of deaths, recorded P&I- and influenza-related mortality can be used to monitor trends within a given season and compare impact across seasons.

Conclusion

The A/Fujian/411/2002 (H3N2)-like strain triggered earlier influenza activity in BC in 2003-2004. Despite enhanced surveillance, the observation of increased influenza severity and mortality due to A/Fujian, particularly among children, was not experienced in BC.

Acknowledgements

The authors thank the following for their assistance: L. Panaro, Canadian Field Epidemiology Program, Public Health Agency of Canada and L. Hoogewerf, Family Practice Research, British Columbia.

References

1. WHO. Communicable disease surveillance and response: Early start of human influenza activity in the northern hemisphere. URL: <http://www.who.int/csr/don/2003_11_21a/en/>. Date of access: 15 July 2004.

2. WHO. Communicable disease surveillance and response: Influenza vaccine for the northern hemisphere 2003-2004: Additional information. URL: <http://www.who.int/csr/disease/ influenza/vaccine2003/en/>.
   Date of access: 15 July 2004.

3. Colorado Department of Public Health and Environment. Press release: State health department reports 3,957 confirmed flu cases. URL: <http://www.cdphe.state.co.us/release/ 2003/112603.html>.
   Date of access: 15 July 2004.

4. Srour L, Skowronski D, Marion S, et al. The 1997-98 influenza A epidemic in British Columbia. B C Med J 2000;42(1):18-22.

5. National Advisory Committee on Immunization. Statement of influenza vaccination for the 2004-2005 season. CCDR 2004;30(ACS-3):1-32.

6. Nicholson KG. Human influenza. In: Nicholson KG, Webster RG, Hay AJ, eds. Textbook of influenza. London: Blackwell Science, 1998;219-64.

Source: S David, MHSc, Canadian Field Epidemiology Program, Public Health Agency of Canada (PHAC) and Epidemiology Services, (BCCDC); D Skowronski, MD, FRCPC and S Tweed, MSc, Epidemiology Services, BCCDC; T Tuk and G Danderfer, British Columbia Vital Statistics Agency; Y Li, PhD, NML, PHAC; M Krajden, MD, M Petric, PhD, G McNabb, BSc, ART, and R Gillies, BSc, RT, Laboratory Services, BCCDC, British Columbia.