Primary Care Corner with Geoffrey Modest MD: Should we delay giving flu vaccine??

There are several studies finding that influenza vaccine effectiveness may wane over time, suggesting that it might be better to delay vaccination to closer to the onset of the influenza outbreak. A new study looked at the effectiveness of vaccines from the 2011-12 through the 2014-15 seasons in the US (see DOI: 10.1093/cid/ciw816).
— the US Influenza Vaccine Effectiveness Network compiled data on adults and children seeking outpatient care for acute respiratory illnesses at 5 different locations scattered through the US, after influenza was confirmed in those communities
— eligible patients had illness with cough, or cough and/or fever of less than 7 days duration, and had nasal specimens tested for influenza/PCR done.
— patients who tested positive for influenza were compared with patients who tested negative.
— they only included patients if they were greater than 9 yo and had received the influenza vaccine at least 14 days before illness onset.
— there were 2800 cases of influenza A (H3N2); 1025 cases of influenza A (H1N1); and 1105 cases of Influenza B
— approximately 50% were aged 18-49, 20% 9-17, 22% 50-64, and 15% > 65. 8% black, 55% female, 37% with one or more high risk condition, approx 40% had influenza vaccinations (both for the season prior to the study as well as the current season). Average time between vaccination and onset of illness was about 100 days
–This study specifically assessed the association between influenza vaccine effectiveness (VE), and the time between the vaccination and the medically-attended acute respiratory illness.
–For the 4 seasons study, median influenza case onset ranged from the 1st week of January in 2014-15, to the 1st week in March in 2011- 12.
–There was decreasing VE with increasing time since vaccination for the various vaccine components:
    –for influenza A(H3N2): Maximum VE was 35% at 14 days postvaccination, reaching 0 at 158 days postvaccination, yielding approximately a 7% (absolute) decline per month, p=0.004
    –for influenza A(H1N1): Maximum VE was 80% at 14 days postvaccination, and minimum VE was 37% at 128 days postvaccination, yielding approximately a 6-11% (absolute) decline per month​, p=0.01
    –for influenza B: Maximum VE was 59% at 14 days postvaccination, and minimum VE was 23% at 180 days postvaccination, yielding approximately a 7% (absolute) decline per month, p=0.04
–for VE, there was an interaction between prior season vaccination and time since vaccination, with VE decline being more pronounced in those with prior season vaccination
— no interaction overall between age and VE, though there was a trend to higher VE in those <60yo against influenza A(H1N1) and lower against influenza B.
— the influenza season in the US typically begins in December or early January, though in the past 18 years it started in the beginning of December in 3 seasons, and in the 3rd week of January or later in another 3 seasons.
— influenza vaccine is available sooner than it used to be, typically by the end of July.
— European and Australian studies have suggested that the longer the interval from vaccination, the lower the vaccine effectiveness. As commented in a recent blog : a study of the 2011/12 season showed that those immunized <3months before the flu outbreak had 53% vaccine effectiveness, whereas those immunized more than 3 months before had only 12% effectiveness. (See Peabody RG. Euro Surveill. 2013 18(5):pii=20389)
–there are potential biases in this and other observational influenza studies (the first 2 noted in the article):
    –has there been a change in the vaccine effectiveness because of drifts in the virus structure emerging and circulating during the flu season which change the effectiveness of the vaccine (ie are people getting the vaccine closer to the onset of the flu actually having a vaccine which matched the flu better)?
    –is there a statistical issue here in that unvaccinated people who become infected with the flu are taken out of the at-risk pool (decreasing it) leading to the vaccinated population to be increasingly over-represented over the course of the flu season? (unlikely from their statistical analysis)
    –are the people getting earlier vaccine different from those getting later vaccines? Do they have more chronic diseases or other conditions which push them to getting vaccines earlier, or are they older, which might blunt their immunologic response to the vaccine? 
    –or, are those getting vaccines earlier less connected with the health care system, healthier, going to the local pharmacy to get flu shots, and more likely to come to the health center specifically for mild to moderate flu symptoms (perhaps they need a note to be excused from work)? If they in fact had a mild case of the flu, they would listed as a vaccine failure. (though it is important to note that these observational studies do not account for the observation that those who have had a flu shot but get the flu often have milder cases, which is a clear clinical benefit not reflected in VE).  And are those who are more medically complex who had flu shots late in the season coming for one of their regularly scheduled visits, having some mild respiratory symptom from a documented flu but only being coded for their difficult-to-treat diabetes, heart failure, etc.  (ie, it looks like the vaccine is working better for them, since they are not coded as having the flu)
–there were a few unexpected findings to me in the above study:
    –there was less VE in those who got the vaccine the year before, though some other studies have also found a blunted immunologic response in those getting repeat vaccination (I would have expected a boosted response from repeated immunologic challenge by the vaccine antigens)
    –there was no difference by age of patient. My reading of the literature is that there are a few studies showing significantly decreased antibody response in the elderly vs younger people, but the data comparing clinical effectiveness broken down by age is a tad murkier, though it is widely held that the vaccine works less well in elderly (eg, a study of community-dwelling elderly randomized to vaccine vs no vaccine, and their care-givers to vaccine vs no vaccine found that vaccinating the caregivers was the only significant intervention). the lack of finding of an age difference in this study may be overstated, since the population of those >65 yo was pretty small (15% of the overall population studied)
So, what makes sense (at least to me)?
–it certainly makes sense to give the vaccine early:
    –in those who are young and need 2 shots (which is why the above study excluded those younger than 9 years old)
    –and, especially in those who might not come back later for a shot just before the flu season begins
–but based on these studies, I will be holding off on giving the vaccine to:
    –my patients who come to see me regularly in the office for management of their chronic medical/psychosocial conditions
    ​–those in home care who can get the shot at home at any time