Jan 15, 2004 (CIDRAP News) – A retrospective study indicates that this year's injected influenza vaccine did little to prevent influenza-like illnesses (ILIs) among workers at a Denver hospital, but the study had many limitations and the vaccine is still recommended, according to the Centers for Disease Control and Prevention (CDC).
The study offers the first scientific assessment of this year's vaccine, which does not precisely match the predominant flu strain circulating so far this season. Because of the mismatch, health officials have been concerned that the vaccine would offer less protection than usual, though the CDC has said it should provide some.
The survey of more than 1,800 workers at Children's Hospital in Denver suggests that the flu vaccine had "no or low effectiveness" against flu-like illnesses between Nov 1 and Dec 11, 2003, according to the report in the Jan 16 Morbidity and Mortality Weekly Report.
But the CDC says more studies of the vaccine's efficacy are needed because this one did not use a random sample and did not determine how many ILIs were actual flu, among other limitations. "Influenza vaccine continues to be recommended, particularly for persons at increased risk for influenza-related complications, their household contacts, and health-care personnel," the report states.
In the study, a questionnaire was sent to all hospital staff members the week of Dec 11 through 17. Workers were asked whether they had had a flu shot, had any condition that could increase the risk of flu complications, had had a flu-like illness on or after Nov 1, or had contact with patients. Of the 1,818 workers who completed the survey, 78% had been vaccinated and 16% reported having had an ILI. Of the 289 people who reported an ILI, only 28 had been tested for flu, and 13 tested positive.
The investigators analyzed the surveys two different ways. In a categorical analysis, workers who had been vaccinated on or after Nov 1 were excluded from the results. In a "person-time" analysis, investigators included workers who were vaccinated on or after Nov 1, but they used vaccination dates to calculate the total person-months of unvaccinated and vaccinated time for all the workers between Nov 1 and Dec 11. In each of these two analyses, the researchers used two different ways of adjusting for the fact that it takes about 2 weeks for the immune system to respond fully to the vaccine.
The categorical analysis indicated the vaccine protected only 14% (95% confidence interval [CI], -12% to 34%) or 3% (95% CI, -28% to 27%) of the workers, depending on the method of adjusting for the delay in immune response. The person-time analysis showed the vaccine had no effect; the actual results were negative numbers that "were not statistically different from zero." The report says the negative numbers "suggest that the study had unknown or uncorrected disparities between the vaccinated and unvaccinated groups in terms of risk for disease or other factors."
Among many qualifiers concerning the study, the CDC says that flu vaccines generally perform better against laboratory-confirmed influenza than against ILIs. For example, in 1997-98, when the vaccine was not well matched to circulating strains, the vaccine had no effect on risk of ILIs but was 50% effective against confirmed flu.
Further, the study did not measure the vaccine's efficacy against more severe illness outcomes or against influenza B or influenza A(H1N1). Even with a "suboptimal" match between vaccine and viral strain, flu vaccines can still prevent flu complications, the report says. In a 2001 study involving patients 65 years and older, the vaccine prevented 61% of flu-related deaths when the vaccine and circulating strains were well matched and 35% when they were not well matched.
Also, because study participants were not selected at random, workers with more exposure to patients and possibly to flu were more likely to be vaccinated, the article says. Other possible differences between the vaccinated and unvaccinated groups probably biased the results as well, the article says.
An important limitation of the study is that illnesses the participants had were not identified in most cases, according to Michael T. Osterholm, PhD, MPH, director of the University of Minnesota Center for Infectious Disease Research and Policy (CIDRAP), publisher of this Web site.
"There are a number of respiratory viruses that could've been causing illness at the time," Osterholm said. "A second similar illness can significantly dampen the efficacy [findings] if that isn't broken out and the effect actually measured." If the vaccine actually was very effective, but sizable shares of both vaccinated and unvaccinated groups fell ill with another virus that looked like flu, the vaccine would appear much less effective than it really was, he said.
"I'm not sure the design of this particular study is rigorous enough to really provide us with accurate estimates of vaccine efficacy," Osterholm said. However, he said the mismatch of vaccine to viral subtype this year gives good reason to believe the vaccine may be less effective than usual.
"The take-home message is that we need new and better ways to make influenza vaccine," he said. It takes about 5 months to mass-produce vaccine with the current method of growing it in chicken eggs. The predominant virus strain this season—the Fujian subtype of influenza A(H3N2)—was seen in the Southern Hemisphere last year, but it was recognized too late for inclusion in this year's vaccine, CDC officials have said.
CDC. Preliminary assessment of the effectiveness of the 2003-04 inactivated influenza vaccine—Colorado, December 2003. MMWR 2004;53(1):8-11 [Full text]
CDC fact sheet on flu vaccine study
CDC question-and-answer bulletin on 2003-04 flu season, with section on vaccine effectiveness study