Reporting bias: Missing the point of HCP influenza vaccine mandates

We invited our good friends Hilary Babcock and Tom Talbot, distinguished epidemiologists and co-authors on the SHEA statement regarding mandatory influenza vaccination of healthcare personnel (HCP), to respond to our previous posts on this topic.  Thanks, Tom and Hilary!

Disclaimer: This commentary solely represents the opinions of the authors and does not represent an official statement from SHEA

Nothing seems to ignite the fires of the Controversies blog like a few key infection prevention topics: CAUTI as an HAI metric, bare below the elbows, and the topic of a recent blog post, mandatory HCP influenza vaccination and the SHEA 2010 Position Paper. As authors on the 2005 and 2010 SHEA papers, we provide a different perspective on the issue.

A key misconception of opponents of mandatory influenza vaccination programs is the claim that the 4 cluster RCTs referenced in the recent post are the only evidence underpinning the position of SHEA (and the numerous other professional societies also endorsing such policies). The 2005 SHEA paper has 100 references while the 2010 paper has 63, a majority of which are peer-reviewed publications in the medical literature. The 4 RCTs noted are far from the only “papers on which SHEA based its recommendation,” as Dr. Edmond states and the De Serres paper implies.

These studies are not perfect by a long shot and the challenges of accurately studying the impact of HCP vaccination on patient outcomes have been nicely highlighted in analyses of these trials, of which there are several – the De Serres paper highlighted in the blog, a CDC analysis that came to different conclusions (and surprisingly was never featured as a blog post), and three separate Cochrane analyses (are there not enough topics out there for them to review??). We also encourage readers to closely review the detailed editorial that accompanied the De Serres paper for an excellent response to the many assumptions in their analysis. If only other infection prevention proposals were scrutinized as heavily as these 4 studies (cough, cough, we’re looking at you bare below the elbows . . . ). Of note, the limitations of these 4 trials were noted in the 2010 SHEA paper, which also cites the first Cochrane analysis in that discussion (despite claims to the counter).

What is missed in the arguments against mandatory programs (including the recent blog post) is the biologic rationale and additional supporting evidence for this strategy as part of a comprehensive infection prevention program. We refer readers to the 2005 and 2010 papers for more detail but summarize briefly here:
  1. HCP (like our patients) can become infected with influenza.
  2. Persons infected with influenza may not present with classical ILI symptoms. This is often missed in criticisms of vaccine effectiveness. Babcock et al demonstrated how poorly the ILI definition captures hospitalized adults with laboratory-confirmed influenza. Thus studies that look just at vaccine impact on ILI will likely miss true influenza outcomes, while also capturing infections due to other viruses not covered by influenza vaccination. Since clinical testing is usually prompted by stereotypical ILI symptoms, lab-confirmed influenza outcomes are also likely under captured. In addition, since hospital length of stay is shorter and most facilities do not have post-discharge surveillance for influenza, the impact in acute care facilities is more challenging to assess than in long term care settings. 
  3. Persons infected with influenza can shed virus even with minimal or no symptoms. While not to the same degree as a coughing, febrile person, the role of asymptomatic infection has been noted in numerous studies of households and experimental challenges. A 6-year study from Hong Kong of a cohort of 824 households with an identified 224 cases of secondary influenza infection examined the relationship between symptoms and viral shedding (as detected on nasal and throat swabs). Shedding was detected before onset of respiratory symptoms in influenza A-infected persons but peaked on the first 2 days of clinical illness, while influenza B shedding peaked up to 2 days prior to symptom onset. So relying on HCP to stay home when ill (which they don’t anyway, see below) will not protect patients. 
  4. HCP work while ill. Sadly, this is a huge issue, as nicely noted in several posts on this blog. Studies consistently note ~75% of HCP with febrile ILI admit to working while ill, a startling yet unsurprising fact that does not capture those with atypical, mild, or even asymptomatic infection. We completely agree that this is a major infection prevention issue. The SHEA 2010 paper notes the importance of “restriction of ill HCP from working in the facility” as part of a “comprehensive infection control program” to prevent healthcare-associated influenza (and other respiratory infections). We definitely support a stronger stance on the infection risk of HCP working while ill and the various disincentives for staying at home (e.g. leave policies where sick days and vacations days are in one lump “bucket”). Perhaps we need a new chapter of the Compendium focused on the prevention of healthcare-associated respiratory infections?
  5. HCP have contact with patients at higher risk for complications from influenza than the general public. The privilege of that close relationship carries an obligation to do all we can to protect those patients. 
  6. Many patients won’t be adequately protected by receiving the vaccine themselves. Proponents of mandatory vaccination frequently note the moderate effectiveness of the vaccine, and we completely agree on the need for a better vaccine. The suboptimal effectiveness only emphasizes the need to optimize immunity among those who are more likely to respond in order to prevent transmission of influenza in healthcare settings (to protect the “herd”). Most HCP are healthy and therefore more likely to have a robust immune response to vaccine than already ill patients, many of whom are elderly and/or immunosuppressed. 
  7. Most agree that HCP should be vaccinated against influenza, though they may disagree with a ‘mandate.’ Extensive literature now demonstrates that a mandate is the most effective way to increase HCP vaccination rates. If it is an outcome (HCP influenza vaccination) that we all support, should we not encourage its use through the most effective method for high vaccination rates? 
Fortunately, as more institutions have employed a mandatory program, new evidence that further supports the need for HCP immunization that was not available when we wrote the 2010 paper has emerged:
Even more important than the impact on HCP illness is the impact on patient outcomes:
  • MD Anderson Cancer Center implemented a mandatory vaccine with masking policy and examined the impact of increasing HCP influenza immunization over the course of 8 years. The proportion of influenza infections that were healthcare-associated among patients significantly decreased and was significantly associated with increased HCP vaccination rates. 
  • A cluster randomized trial in the Netherlands of HCP at six medical centers, where the intervention arms offered vaccination to HCP vs. no vaccination at control facilities, noted a significantly lower rate of healthcare-associated influenza among internal medicine patients at the facilities with the higher rates of HCP influenza vaccination (3.9% vs. 9.7% of patients). 
  • In a study encompassing 7 influenza seasons and over 62,000 hospitalized patients, a significant association was noted between increasing influenza vaccine coverage among HCP and decreasing healthcare-associated ILI among patients at an Italian acute care hospital.
  • Finally, a nested case-control study in France noted a significant association between lower rates of laboratory-confirmed healthcare-associated influenza among patients and higher vaccination rates among HCP.
We don’t have time to go into the ethical arguments for HCP vaccination or the need to broaden these programs to include all recommended immunizations for HCP, but we close with a noteworthy pronouncement made by the Board of the National Patient Safety Foundation’s Lucian Leape Institute: there are two “must do’s” for HCP to ensure patient safety, hand washing and HCP influenza vaccination. 

Time to stop re-analyzing those 4 poor cluster RCTs – that horse has been beaten to death.


  1. Of the "new evidence" only the Netherlands study was randomized--a non-blinded cluster design. If you read that study carefully, you will note that the secondary outcome combined Influenza AND Pneumonia together as a composite outcome. Luckily Table 4 provides enough raw numbers to post hoc look at just Influenza. And here we find that the Intervention (influenza vaccine program) hospitals had a nosocomial influenza risk of 33/1387 or roughly 2%, whereas the Control hospitals had a nosocomial influenza risk of 20/1980 or roughly 1%. Someone please check my math. I combined the Internal Medicine and Paediatrics data. So, If one excludes Pneumonia from the composite outcome, one gets the exact opposite result that the paper purports to find. Tricky business.
    --Former Hospital Epidemiologist (couldn't take it anymore). Now I teach epidemiology and statistics, including bias to students. Hope for the future, not the present
    P.S. Composite outcomes often/always? lead to information bias.

  2. Agree. I come to similar conclusions here:!/2017/02/when-your-dead-horse-is-your-only-horse.html

  3. Just read this again and would note this quote: "In an analysis of the province-wide vaccination with masking policy in British Columbia, researchers noted a significantly reduced rate of HCP absenteeism due to all-cause illness in vaccinated vs. unvaccinated HCP." I would want to exclude the possibility that HCW feel protected since they were vaccinated and would work while sick, decreasing absenteeism, but increasing presenteeism. This would be particularly concerning in years with poor vaccine efficacy.


Post a Comment

Thanks for submitting your comment to the Controversies blog. To reduce spam, all comments will be reviewed by the blog moderator prior to publishing. However, all legitimate comments will be published, whether they agree with or oppose the content of the post.

Most Read Posts (Last 30 Days)