Helping you to practise evidence-based medicine

Guest Editorial

Influenza surveillance, the swine-flu pandemic, and the importance of virology

Influenza surveillance has traditionally been given priority over surveillance of other acute respiratory infections. The virus has strong epidemic potency, affects all age groups, spreads rapidly, and appears regularly mid-winter, when deaths and hospital admissions are greatest.[1][2] The type of virus responsible for influenza-like illnesses has been difficult to define because virological investigation is rarely done, the range of organisms investigated is limited, and, historically, the quality of investigation methods has been poor. Surveillance data collected during the first wave of the swine-flu pandemic in the integrated clinical and virological surveillance programme of the Royal College of General Practitioners and HPA (as part of the Weekly Returns Service [WRS; see Figure 1]) suggest the incidence of swine flu was considerably higher than estimates based exclusively on laboratory virus detection.

A clinical diagnosis of influenza is not a virus-specific diagnosis, and complementary virological data are needed for interpretation of incidence data.[3] Although clinical influenza-like illness (ILI) is common, only a fraction of people with ILI consult a doctor, and of these only a small proportion is investigated virologically. Of those tested, only a minority are confirmed as having influenza virus infection. Conversely, after most epidemics, many people have immunity (as shown by the presence of antibodies in a serum specimen), indicating exposure to the influenza virus, but not necessarily clinical illness. In community-based studies carried out in the USA in the 1960s and 1970s, seroconversion rates in some years were recorded as over 30% of the population.[4]

Influenza virus strain H1N1 of novel swine origin was the cause of the outbreak of acute respiratory infection in Mexico in March 2009. The first confirmed case in the UK (in a "non-traveller" from the Americas) was reported in week 19 of 2009. Figure 2 shows the clinical incidence of ILI per 100,000 population in the summer weeks of 2009 as reported in the WRS, along with equivalent average rates from the past 10 years, the numbers of nose/throat swabs submitted for virological investigation, and those positive for A H1N1 swine variant. These data demonstrate an epidemic wave that peaked around week 29 and lasted roughly from week 25 to 36: they also illustrate a similarity in the trends of ILI incidence, swabs submitted, and numbers of positive tests.

The increased rates of ILI that were observed from weeks 18 to 21 were not sustained, nor were they accompanied by increased reporting of other respiratory infections. These increases probably reflect the media hype surrounding swine flu. The peak incidence estimate of 153 per 100,000 people exceeds all weekly estimates reported since the winter of 2000, and is more than twice the highest rate recorded during the summer weeks since recording began in 1967. Around the time of peak incidence, there was also a small increase in the incidence of common cold, but not in incidence of acute bronchitis or acute otitis media, both of which increase concurrently during periods of influenza activity in winter. The stability of rates of incidence of these conditions is indicative of a relatively mild illness.

A total of 1902 swabs were submitted for virological investigation between week 18 and week 37: 379 (20%) were positive for the H1N1 influenza swine variant, 5 influenza B, 7 influenza H3, 2 RSV, and 5 hMPV; 79% of swabs tested were negative for all viruses tested. The weekly trends of ILI incidence and of influenza virus detections have been similar over many winters.[5]

Increased use of and improvements in investigation methods have led to the identification of many viruses that cause ILI and other respiratory infections in mid-winter,[6][7] including respiratory syncytial virus (RSV) and human metapneumovirus (hMPV). However, other viruses (e.g., rhinovirus, para-influenza, and adeno-virus coronavirus) that generally circulate over longer periods do not cause typical epidemic situations with many people infected simultaneously. In most winters, about 35% of WRS swabs yield virus and influenza positivity rates that are higher than those reported during this pandemic wave. For example, in winter 2008, 1783 swabs were submitted, 622 (35%) of which were positive for influenza; in the peak week, 129 out of 238 (54%) were positive for influenza. Clinical trials of influenza treatment have achieved positivity rates exceeding 60%; however, these trials were carried out when influenza virus was known to be circulating in the community, and were based on strict diagnostic criteria with recruitment restricted to people within 48 hours of symptom onset.[8][9] Positivity is less likely if the viral load is low, as in people with less severe illness, or in older people who typically present later.[10] In recent years, influenza H1N1 and B infections have been generally mild in older people.

So, how many people consulted for swine flu during the first wave? The summed clinical incidence of ILI between week 18 and week 37 equates to 0.64% of the population. In addition, there were patients consulting who were labelled as 'suspect swine flu' or 'swine flu contact', and others consulting elsewhere (e.g., national flu line, out-of-hours primary care centres) which, if added, increases the estimate to around 1%. The diagnosis was confirmed in only 20% of those providing swabs to the WRS, which exceeds the proportion reported from other healthcare agencies, who were generally more selective in case recruitment.[11] Extrapolation of the virology data to the consulting population might suggest that only 0.2% had swine flu.

Four-fifths (79%) of samples tested were negative for all viruses examined. Possible explanations for such a high negative rate include:

• Most people tested were not ill
  It seems unlikely that people presenting to a clinician (especially those selected for virological investigation) were well: they would not be labelled by the GP as having ILI and would be even less likely to be swabbed. The absence of any appreciable increase in reports of minor upper respiratory infections opposes the notion that people who normally self-manage consulted because of specific concerns about swine flu, or to get anti-virals. An initial period of increased consultation, which was probably attributable to media hype, was not sustained.

• The community epidemic experience was caused by an unidentified virus/other pathogen
  A few additional viral pathogens were detected, almost randomly, through the summer, and can be dismissed as major contributors to the epidemic. Whilst viruses such as rhinovirus undoubtedly contribute to the toll of acute respiratory infections, they are more commonly associated with other respiratory diagnoses (common cold, otitis media, and asthma attacks), none of which were reported with unusual frequency over the summer period. Furthermore, the impact of most (excluding para-influenza) non-influenza viral pathogens does not have the epidemic character typical of influenza as was seen this summer, nor do they have such a broad impact across all ages.[1] It is not credible that an unknown pathogen, occurring at exactly the same time as the H1N1 swine flu virus, could have caused many more people to be ill than were caused by the H1N1 virus.

• False-negative tests (i.e., true swine flu but no virus detected)
  The reliability of virological detection depends on the quality of the swab when taken, on arrival at the laboratory, and on the investigation methods. Polymerase chain reaction (PCR) methods of investigation are used throughout Europe and have acceptable levels of specificity and sensitivity when used in community-based surveillance programmes.[6][12] To date, swine flu has been a mild illness in most cases, and, since symptoms correlate with viral load, is perhaps associated with an increased likelihood of false-negative tests. It is reasonable to question the quality of the swab specimen and the postal transfer in summer temperature conditions, but harder to challenge the quality of detection methods.[13]

It is difficult to escape the conclusion that most people diagnosed with ILI did indeed have swine flu, and our ability to detect swine influenza viruses (due to poor sampling, deterioration of samples before testing, or limitations in test sensitivity specific to this virus) poses a greater problem than appreciated hitherto. In most winters, two-thirds of swabs submitted do not lead to diagnosis. Without a satisfactory explanation for this large burden of illness in the community, there can be no progress towards prevention and treatment, and optimum management regimes cannot be defined. Reliable estimation of the relative impact of individual virus pathogens is essential for studying the cost effectiveness of interventions and defining policy.

Douglas M Fleming, OBE F.Med Sci
Research and Surveillance Centre,
Royal College of General Practitioners
Birmingham
UK

• Subscribe: Formats | Institutional access | Monthly email alert
• Resources: Commissioned papers | EBM Tools | EBM links | Drug information | EBM training | Articles and Research
• About us: What we do | A guide to the text | Legal
• Contact us: By email | By post | By telephone
• Contribute: Introduction | Become a contributor | Become a peer reviewer

• Website Terms and Conditions |
• Terms and Conditions for purchase |
• Privacy policy |
• Home |
• Top