Recall bias, or how I learned to stop worrying about measles and love my vaccination beliefs

The multi-state outbreak of measles in the US, which originated at Disneyland theme parks in California, has received much attention in the press recently. Sadly, the event isn't terribly surprising: many people, including children, lack immunity. The reasons for this are varied, and include the MMR vaccine being recommended for children over 12 months of age; persons not completing the recommended vaccination sequence; waning immunity in older persons; and lack of vaccination.

This latter issue is related to the anti-vaccination movement and this dimension is receiving attention in the media. Some of the discourse is revealing. One example appeared in the New York Times recently, which quoted a Santa Monica pediatrician who has cautioned against the way vaccines are used in the past (but does administer the vaccine in his practice):

“I think whatever risk there is -- and I can’t prove a risk -- is, I think, caused by the timing,” he said, referring to when the shot is administered. “It’s given at a time when kids are more susceptible to environmental impact. Don’t get me wrong; I have no proof that this vaccine causes harm. I just have anecdotal reports from parents who are convinced that their children were harmed by the vaccine.”

Perhaps such comments could be rephrased more bluntly: I believe that the vaccine could be dangerous to kids because people who know less about medicine and epidemiology than I do tell me, with great conviction, that they think it's dangerous. Even blunter still might be: The idea that vaccines can be dangerous has a high truthiness, so count me in. Or even, After talking to some patients I've decided to stop worrying about measles and love my unfounded beliefs about vaccination.

I've written before about the importance of understanding those who subscribe to anti-vaccination notions. Statements like this from a physician illustrate how much work has to be done. Several years ago Delgado-Rodríguez and Llorca wrote a very nice continuing professional education paper on bias in epidemiology, which physicians with similar beliefs should read. The following passage is especially important:

Recall bias: if the presence of disease influences the perception of its causes (rumination bias) or the search for exposure to the putative cause (exposure suspicion bias), or in a trial if the patient knows what they receive may influence their answers (participant expectation bias). This bias is more common in case-control studies, in which participants know their diseases, although it can occur in cohort studies (for example, workers who known their exposure to hazardous substances may show a trend to report more the effects related to them), and trials without participants’ blinding.

If someone wants to find a cause for a medical event, they will. The plural of "anecdote" is not "data", no matter how convincing the anecdotes may seem.

(image source: Wikipedia)

The epidemiology of Fearbola

In the mid-19th Century, a newspaper could reach several thousand people daily or weekly. By the mid-20th Century, TV and radio reached 10s to 100s of millions of people instantaneously and possessed a multinational reach. Today, with the Internet, and satellite TV and radio, it is possible to reach 100s of millions of people or more across the globe within minutes. This vast and practically instantaneous reach of technology feeds a seemingly insatiable, 24/7 appetite for news and information. What are the implications of this for fighting epidemics? 

We've seen some of the consequences in the Ebola outbreak this year. On the one hand, the ability of aid groups to spread information broadly has been helpful for raising awareness and bringing additional resources to bear on the epidemic. On the other hand, news headlines resulted in near-hysteria and much counterproductive behavior in the US and other developed nations. Examples were highlighted in a previous post, and many, many others have offered similar observations and commentary.

Perhaps the reactions observed in the US have been understandable, as many ingredients were present for an epidemic of sensationalism and fear: An active public imagination rooted in previous popular books and movies, a government that addressed the issue late and with almost Pollyannaish credibility at first, and a wealth of news outlets offering non-expert commentary while playing to the continuous news cycle.

The resulting epidemic of "Fearbola" should thus not be surprising. It is sobering, however, for it provides a warning for domestic public health agencies: Understand how to administer effective public health messages that are relevant and appealing to the constant clamor of CNN, Fox, and the like, or else risk being drowned out by noise and hype. What if this epidemic had been of a pathogen possessing a short serial interval and high virulence, transmissibility, and R₀? The medical system may or may not be prepared, but it seems clear that our risk communication strategies are not. Would the news coverage we saw during the Ebola hysteria in recent months have served the public well if this had been a bona fide threat to US public health?

It's important to understand how the epidemic of fear and hype came into being and propagated so well. I tend to think of messages as themselves being infectious. From that perspective, the ideas that resulted in the hysteria surely had R₀ > 1. For ideas related to "dread threats", such as virulent infections with no known cure, is this unavoidable given the high contact rate (e.g., frequency of checking for news and rumors combined with near-constant coverage), short serial internal (e.g., rush to post on social media), broad coverage, and rapid dissemination of modern communications?

If so, we must learn how to craft public health messaging strategies so that authoritative messages will out-compete hype and fear in our hyper-connected world. If we don't learn how to do so, we run the risk that important messages will be drowned out by high-incidence, fearful messages in future outbreaks of international public health importance.

(image source: David Hartley)

Epidemiology and behavior in the time of Ebola

This week the WHO Ebola Response Team published a paper raising the notion that Ebola could become endemic in the human population of West Africa. The idea hadn't occurred to me previously, and it struck me as very unlikely. After all, this is a directly transmissible disease that, as many have told us, we know how to control.

After reflecting on the possibility, however, I don't think it can be discounted out of hand. On the one hand, breaking the chain of transmission can be achieved theoretically with careful attention to infection control and prevention practice, which is well defined in the healthcare environment. On the other hand, this isn't a nosocomial outbreak. Community transmission is the major driver of incident cases, so changing human behavior in the community must occur if this epidemic is to be stopped. In general behavior is hard to affect, and in this case it may be even harder, given recent descriptions of distrust between healthcare providers and the community.

As I've mentioned before, one of the uses of mathematical modeling is to support clear and careful thinking. In this case, epidemiologists have applied models to estimate the basic reproduction ratio, R₀, and have found it to be greater than 1, consistent with estimates from past outbreaks. Such an R₀ suggests that the virus has the potential to circulate permanently in the human population at some non-zero endemic prevalence. Endemic prevalence levels could be, relatively speaking, high or low (or intermediate). If low enough, the disease could fade out stochastically on its own, but at higher prevalences the continual danger of sporadic cases could persist indefinitely. Models can help us gain a sense of the relative likelihood of such outcomes.

The risk factors for acquiring Ebola virus infection are well known. If effective interventions reducing risky behavior are instituted widely and adhered to, they may reduce the effective reproduction ratio, R_eff, to less than 1, thereby breaking the chain of transmission. Achieving that must entail not only nosocomial infection control but also infection prevention through behavioral change in the community.

Changing behavior surely involves building and rebuilding trust between healthcare providers and local people. I suspect and hope that the recent massive pledges of, and plans for, assistance will help build the necessary rapport and trust. Maybe the construction of clinic facilities that better support effective care will help advance such endeavors. One thing is certain, however: the longer those pledges take to become reality, the more likely the worst scenarios for the course of this epidemic become.

(image source: Wikipedia)

Legionella on ice

"Legionnaires' disease" was the name originally given to an illness observed at a 1976 American Legion convention. Today, we call the illness associated with Legionella pneumophila infection, which can range from mild respiratory illness to severe pneumonia, "legionellosis". Legionella bacteria exist naturally in water and moist soil and colonies tend to grow in warm water, pools of which often form in improperly operated or maintained HVAC systems, hot tubs, and hot water systems. Legionella is an important cause of both hospital- and community-acquired pneumonia in both immunocompetent and immunosuppressed patients. Hospital-acquired cases are often associated with potable water systems colonised with Legionella.

A story in the Pittsburgh Tribune-Review recently described an unusual outbreak at a Pittsburgh hospital. Although the epidemiology showed an association with ice chips, investigators were unable to find Legionella in the hospital water system. How could Legionella appear in ice from machines supplied by cold (not hot) water lines free from the bacterium? It was ultimately determined reservoirs within hospital ice machines were warmed by internal compressors, thus allowing Legionella colonies to grow.

Previous outbreaks involving ice makers and Legionella have been described in the literature (see, e.g., Schuetz et al, 2009, Graman et al, 1997, and Stout et al, 1985), but I doubt that many would immediately respond "ice machine" when asked about likely sources of Legionella infection in a hospital. Though anecdotal, this story illustrates how counter-intuitive outbreak investigation can be: One wouldn't necessarily think to look in a freezer for a bug that needs warm water to grow. But there it was, and hospital investigators figured it out when the ice chips were implicated. Bravo! 

(image source: Wikipedia)