Will healthcare be a byproduct of our math illiteracy?

OK, so when I was in practice I pretty much worked every day, including weekends, even when not on call, and can attest to the burn-out potential of this life-style. My "day off" meant that I only spent eight hours in the hospital, instead of the customary twelve to fourteen. My infant son cried when I picked him up because he did not know who I was, there was little time for such mundane pleasures as doing laundry or going grocery shopping, and bedtime reading consisted of two-to-three lines of printed text swimming on a page until the book collapsed on my chest to the accompaniment of my snoring. But let's assume that I am just a wimp, and that most of my peers can take such schedule in stride.

A propos this, I came upon this column by Peter Orszag in the Sunday Times regarding the need to increase hospital efficiencies by keeping hospitals operating 100% 24/7, including the weekends. He cites not only the financial upsides of this arrangement -- basically taking advantage of spreading many of a center's fixed costs over more days of operation -- but also the potential for improving patient safety and quality of care. What I am curious about is what it means for clinicians. Doing the mental math, when the pie size increases without a commensurate rise in the pie consumers, each pie eater has to then eat a bigger piece of the pie. This is certainly a welcome quandary  when it comes to my mother's-in-law apple pie, but for the healthcare system it is a horse of an entirely different color. The growth of the healthcare delivery pie must mean an increase in physician and staff work hours, correct? By about 40%, right? Unless of course we can increase the ranks by the equivalent 40% to share the burden.

So up until now I am pretty clear on the math. Here is where confusion begins to creep in. Being a healthcare researcher, I am keenly aware of the projections of shortages for many medical specialties. But most contradictory for the Orszag plan is the monumental shorfall of hospital nursing staff, estimated to reach over 1 million by year 2020! And similar sitution holds for pharmacists, respiratory therapists, etc., just look at some of the annual reports from the American Hospital Association. Given these dire prediction, how do we implement the Orzsag plan? Do we just ask the present personnel to give up all of their free time entirely? What do we think will happen then? Especially as they watch their professions get increasingly more commoditized and their incomes shrink?

And if these conflicting ideas are not enough, there is this headline in today's amednews: "2010 could end up as worst year for hospital layoffs". OK, I will pause for a moment to let it sink in. The story goes on

As of August, 8,233 employees had been affected by mass layoffs at hospitals. Continuing at this rate, 12,349 hospital employees would be cut in mass layoffs by the end of 2010. That would be the second-highest loss recorded by the BLS since 2000, second only to 13,282 in 2005 -- with 8,687 of those jobs lost because of hospitals shut down because of damage from Hurricane Katrina.

In 2009, 11,757 hospital employees lost their jobs through mass layoffs, according to the BLS.

OK, so what am I not seeing? Orzsag is a smart economist, but if even he cannot do this math, how can our embattled public with poor math literacy and even less understanding of the intricacies of our Byzantine healthcare bureaucracy get it?

Why medicine is not like a parachute

At the recent meeting of the American College of Chest Physicians I went to a session on hospital rapid response teams. This was a talk given by a senior and respected member of the Pulmonary and Critical Care academic community, well published and known. Now, if you are wondering what the heck is a rapid response team, I will tell you: it is a team composed variously of an ICU physician (or a hospitalist, depending on local patterns), a nurse or two, a pharmacist and sometimes other ancillary personnel. The role of this team is to be at the beck and call of the ward nursing staff to evaluate urgently any floor patient whose clinical condition may appear to be deteriorating. The ultimate goal, of course, is to avert a catasprophic emergency called a code, where the patient deteriorates to the point of needing emergent resuscitation.

While such a team makes a lot of sense, no study has convincingly shown that it does what it is meant to do -- either prevent codes or save lives. The presenter, a man quite invested in the sensibility of such teams, was making the point that perhaps we just have not focused on the correct outcomes to measure for this intervention. Ultimately, he said, we should continue to support such teams because of our gestalt that they work. After all, he went on, nobody has tested the effectiveness of the parachute in a randomized controlled trial.

When I heard this, I realized that I had been hearing this comparison a lot lately, of medical interventions to parachutes. And since his statement seemed somehow wrong to me, I started to think about why such an analogy is completely fallacious. And here is the startlingly obvious answer: the number needed to treat (NNT) for a parachute is 1! What this means is that one only needs to "treat" one person with a parachute (or strap it on before jumping out of a plane) to save one life. Now, is there really anything that we do in medicine whose margin of effectiveness comes even close? Last week we argued about a screening test with the NNT of 2,000. And we also talked about a therapy with a NNT of 16 which is not being utilized because of high costs.

So, let's not be intellectually lazy and let's not resort to comparing anything we do in medicine to a parachute. The parachute is made to mitigate the laws of physics, which are inherently more predictable, stable and generalizable than the laws of biology. Instituting a rapid response team may feel good to us, but, in addition to lending no measurable improvements to patient care, it may have the "unintended consequences" of abandoning patients who really need our attention at that moment to better their outcomes.

A similar argument can be made about many interventions whose mere face validity we take as God's own word. We and others have cautioned against such gullibility, particularly in a world that runs with half-truths and misinformation under the banner of quality measures. None of us has the cause to be so self-congratulatory as to compare our wares to the parachute. Now more than ever, experts need to think and present critically, soberly and objectively, lest we continue the trend of parochialism in our out-of-control disease care system.    

Notes from a pandemic

Informative report in Eurosurveillance yesterday from New Zealand summarizing their H1N1 experience to date. A few encouraging signs are that a). the pandemic seems to be slowing down, b). proportion of people needing hospitalization is low, on the order of 0.3%, of whom ~12% require an ICU, and c). the case fatality rate is about 0.005%. All this in the face of an estimated attack rate of 11%. Very encouraging!

So, given what I said yesterday in my post, perhaps we can revise down our estimates for the US. If we were to do that on the back of a napkin, we would see a much less concerning picture of around 33 million infections, ~100,000 hospitalizations, ~12,000 ICU admissions and a little over 1,500 fatalities. Whew! Much better than what either our group or the White House advisers predicted. But is this the right time to recalculate? Maybe not.

Here is why. First, the Southern hemisphere is still in the middle of its flu season, and flu epidemics are known to come in multiple waves, separated by intervals from months to years. So, while the slow-down is encouraging, the fat lady has not yet sung. Second, the New Zealand report notes that the native Maori and Pacific peoples, two socio-economically disadvantaged ethnic groups with marked health disparities, exhibited triple to quintuple the attack and case fatality rates seen in their European counterparts. Given that 47 million people in the US are without routine access to the healthcare system, and racial and ethnic healthcare inequalities are rampant, we remain at high risk for seeing a lot of severe cases. Finally, and most importantly, New Zealand has a comprehensive nation-wide plan for mitigating a pandemic (and it is a much smaller nation than ours). Some of the phases include extensive pre-pandemic planning, improved surveillance, border control aimed at containment, and quarantining active cases if necessary. Our plan, which is actually many different plans, is still evolving and is unlikely to be this comprehensive, uniform or enforceable.

So, while I am encouraged by what is being reported, I am not willing to hang my hat on it yet, and neither should the US Department of Health and Human Services. It is true that we may get lucky this time. But this should serve as our national wake-up call, since in our global village it is only a matter of time until we are faced with a devastating health emergency. Do we really want to rely on luck next time?

H1N1 predictions demystified

The President's Council of Advisors on Science and Technology report on the US preparations for the 2009 H1N1 influenza estimate of 90,000 fatalities is getting a lot of heat from an unlikely source: the Centers for Disease Control and Prevention. According to a New York Times story, there is disagreement with this prediction. Anne Schuchat, the director of the National Center for Immunization and Respiratory Diseases is quoted as saying "We don't necessarily see this as a likely scenario". A CDC press officer apparently confirmed that no one at the CDC "expects anything like 90,000 deaths".

If you look at the report itself, even the authors hedge on this point: they prefer to present is as a "planning scenario", rather than a prediction. OK, I'll buy this but what is a prediction but a planning scenario? True, it could be fewer. But if we sit around and simply argue about the number, it could be a lot worse.

Let me tell you a little bit about why and how we create these predictions (or planning scenarios, your choice). The why is pretty self-evident: we want to be prepared. Preparation entails having an idea, vague though it may be, of what's to come. Why do you care about tomorrow's weather forecast? So that you can plan what to wear, how long to allocate to your commute, or whether or not to cancel that trip to Florida during the hurricane season. Similarly, policy makers need to have some idea of how to allocate resources in the future. So, we make predictions, create forecasts or planning scenarios.

The irony is that all this future planning is built on past events as we know them in the present. (So you can see how a rapidly-evolving pandemic can throw a monkey-wrench into any estimates.) What makes us think that there are likely to be 30,000-90,000 fatalities associated with H1N1 in the US this flu season? Its past behavior, of course -- we look at what has happened to date, both in the US, and, in this case, in the Southern hemisphere, where the flu season is in full swing, and we apply those numbers to our population. The result is a mathematical formula which, when populated with these carefully evaluated assumptions, spits out the desired estimate. Now, it is also possible to include some other assumptions into the model, such as what happens at different levels of vaccine availability, efficacy and penetration. However, since all of these factors are to date unknown, putting them into the model would create a tremendous amount of uncertainty, and the model would be useless.

But even without introducing conjecture around potential modifiers, these estimates are prone to a large degree of uncertainty. This is why whenever you hear a report of a single number as the estimate derived from a prediction model, ask for more: there is usually a confidence interval calculated around that number based on varying the assumptions across some justifiable range. Generally, the tighter the confidence interval, the more useful the prediction. For example, in our recent study we estimated that on average we can expect ~300,000 cases of acute respiratory failure associated with H1N1 in the US, with the confidence interval ranging from ~225,000 to ~450,000. Now, this is a wide range, and it betrays our uncertainty about assumptions that went into the model.

One final note. The estimates provided by the White House advisors and the ones derived by us do not necessarily account for the effect of an all-out effort at prevention. So, if we sit by and wait for the virus to hit, and if it does not get more or less virulent than its current incarnation, we will have the predicted number of fatalities. That is a lot of "ifs", some of which are certainly within our control -- aggressive education, prevention, containment, to name a few.

So, with all of the pitfalls of planning scenarios, they are necessary for us to appreciate the potential magnitude of the problem and to plan for it rationally. Sweeping them under the rug and throwing their creators under the bus is just playing politics with an uncertain situation: these numbers evolve, and we need to acknowledge that. After all, as that populist philosopher Yogi Berra once pointed out: "Predictions are very hard, especially about the future".

Caught with our pants down by H1N1

I am still on vacation and my brain is too mushy to do a good new post. So, am recycling an old one. By the way, I modeled the potential number of respiratory failures we can expect in association with this winter's flu season, and now I am more sure than ever that we may be in deep doo-doo. This is made even more relevant by the story in the WaPo yesterday discussing the challenges of the vaccine development and distribution.

Got an e-mail today from a Bloomberg reporter asking about my thoughts on the state of US preparedness for H1N1. Specifically he was interested in the ICU capacity as it applies to dealing with the more severe respiratory failure that has been reported in association with this infection. You might wonder why he called me of all people. Well, this is because I published a bunch of papers recently (that have been dutifully ignored by policy makers) that quantify the expected stress to the healthcare system in the US posed by the sky-rocketing increase in the population of patients requiring ventilators to assist them with breathing. How naïve I was to think that our numbers might be used for planning purposes, so that we could be better prepared for this growth!

A year since we first started to come out with the numbers, there are still no reliable estimates, or even attempts to derive them, as far as I can tell, of the US ICU surge capacity. And this information is more critical now than ever, as we are faced with the potential for untold numbers of H1N1-related hospitalizations accompanied by severe respiratory failure. It is shocking to me that we are leaving these infrastructure issues to chance.

But wait, why am I so shocked? This is just another byproduct of our fragmented healthcare "system". There is no one either to take or to designate responsibility for planning. While our professional organizations are doing the best they can, their approach is usually siloed and somewhat distracted: in case you have not heard, we are also tackling the most rampant pandemic of antibiotic resistant infections, which affects ICUs disproportionately.

So, who will take charge of this hot potato? The CDC? Homeland Security? FEMA (chuckle-chuckle)? Press your representatives to tackle this thorny issue, because if you are worried about planned rationing now, wait until you see what unplanned bedlam looks like. Condi Rice said that the US intelligence community just did not have the imagination necessary to anticipate the 9/11 terrorist attack. Let us not have to resurrect this tired excuse this flu season.