Achilles happens upon his friend Tortoise in the lobby of his doctor’s office building.

Achilles: Hi ho, Mr. T! Fancy meeting you here. I hope you’re not under the weather. There’s a new flu going around, you know.

Tortoise: Isn’t there always? In fact, I’m not feeling very well, but it has nothing to do with the flu. You see, I’ve been to see my doctor, and what he told me has given me the worst headache of my life.

Achilles: How so?

Tortoise: Well, as you know, I’ve become rather sedentary lately—although I believe I could still take you in a foot race—and have started thinking about doing more to improve my health. My mum and dad lived well into their hundreds, and I’d like to do the same.

Achilles: So?

Tortoise: I was searching the Internet (always a dangerous thing) and came across a recent clinical trial called JOVE—one of those maddening pseudoacronyms, the meaning of which I can’t, for the life of me, recall. Anyway, this trial described the benefits of some new cholesterol drug.

Achilles: Interesting!

Tortoise: I thought so too. So I showed the report to my doctor and asked him if I should be taking this drug.

Achilles: What did he say?

Tortoise: That’s when all the trouble started. He knew of the study but didn’t think it applied to me.

Achilles: Why not? Was it flawed in some way?

Tortoise: Not at all. In fact, he said it was well designed and meticulously conducted. But according to him, although the study subjects are much like me— on average —I differ from them in several important ways. Despite my sedentary lifestyle, I’m younger and have a better family history. Consequently, Dr. C doesn’t think treatment would benefit me as much as the group. As my doctor, he says he’s responsible to me as an individual and is not about to treat me out of some cookbook. He actually got rather worked up over this. “Damn it, Tim,” he said, “I’m a clinician, not an epidemiologist.” According to Dr. C, clinical problems should be approached from a clinical perspective .

Achilles: That sounds like a reasonable guideline.

Tortoise: Oh, it’s much better than a guideline.

Achilles: Why?

Tortoise: Because it’s always true. Guidelines aren’t! Dr. C calls them quasi-religious opiates for the masses. He says I’m no more likely to benefit from your average guideline than from the next toss of the dice.

Achilles: Then what good are they to us?

Tortoise: Not much! Dr. C thinks guidelines are simply a way for the academic priesthood to exercise its influence over the clinical laity. But he also knows that many clinicians just do not have the time or training to critically evaluate the large volume of scientific information crossing their desks every day. Guidelines help these clinicians distill that information into pragmatic, ready-to-use rules of thumb. Just don’t expect them to be right all the time.

Achilles: Hmm. I’m just playing devil’s advocate here, but suppose this trial had been negative. Couldn’t Dr. C still point to some adverse predictor (such as your sedentary lifestyle) and say that you are personally likely to benefit even though the group did not? In treating you as an individual, it seems he might be playing fast and loose with the scientific evidence.

Tortoise: Now you know why I have a headache. I do want to be treated as an individual, but I also want to be treated according to the best scientific evidence.

Achilles: But doesn’t that evidence come from the study of groups?

Tortoise: Certainly! And the larger and more random these groups, the better the evidence. “The more, the merrier,” as they say.

Achilles: Are there no rigorous scientific studies of individuals?

Tortoise: Goodness no! Individual case studies are decidedly anecdotal and unscientific. As Dr. C explained it, each individual comprises a veritable infinity of observations. Some predict benefit from treatment; others do not. In this context, the individual is the sum of a group of predictors, and the trial is the sum of a group of individuals. The average of the group, however, doesn’t necessarily represent any particular individual. Clinical decision making rests more on one’s expectation of benefit to the individual than to the group. Real-world effectiveness trumps trial efficacy.

Achilles: But on what does Dr. C rely when making these individual clinical decisions?

Tortoise: On randomized clinical trials, of course. They’re the apotheosis of evidence-based medicine, you know.

Achilles: But didn’t you just say the average of the group doesn’t necessarily represent a particular member of the group—you, for example?

Tortoise: Touché! Evidence seems a curious thing. The closer you look, the curiouser it gets. What’s a turtle to do?

Achilles: Perhaps you should get a second opinion from someone with a firmer grasp on the rules of evidence.

Tortoise: I did! Dr. C’s partner happens to be formally trained in the arts of epidemiology and biostatistics and was a co-investigator on JOVE. She calls herself an interventional epistemologist—although that specialty will not be certified for several millennia. I just came from her office.

Achilles: What did she say?

Tortoise: Dr. E said she analyzed a variety of subgroups in JOVE—including those based on age, lifestyle, and family history—and found none of them to differ significantly from the overall group. As she put it, a difference that makes no difference is not a difference.

Achilles: So that settles it!

Tortoise: Not quite. According to her, none of these additional analyses were conclusive, because the number of individuals in each subgroup was rather small. Something about statistical power. I didn’t quite get it. Anyway, she said there’s no reason to expect I would do any differently than the group. Even so, she too wouldn’t treat me.

Achilles: Why not?

Tortoise: She said her principal allegiance was to clinical investigation and health policy, not to patient care. She acknowledged that investigators wear many hats but didn’t want to treat a clinical problem from the perspective of an epidemiologist.

Achilles: Well, what good is it, then, to expend all this effort on these clinical trials, if the doctors just wind up doing what they damn well please? It all smacks of intellectual gerrymandering.

Tortoise: I thought so too, but Dr. E sees it differently.

Achilles: She would. Doesn’t she get paid to do these trials?

Tortoise: Yes, of course. But it’s more complex than that. Turns out she and Dr. C both serve as consultants to the agency that regulates drugs and devices based on the results of these trials.

Achilles: Why should that matter?

Tortoise: Well, according to Dr. E, when acting in this so-called regulatory capacity, their focus is more on the group than on the individual. Here, efficacy trumps effectiveness—at least in principle.

Achilles: But not in practice?

Tortoise: It seems that Dr. C just can’t help thinking like a clinician at these meetings.

Achilles: You sound critical. Is that a bad thing?

Tortoise: Not necessarily. Clinicians often help trialists identify important new questions deserving of attention. In this particular case, however, it caused some problems. According to Dr. E, 2 different trials were being discussed at this meeting. I don’t remember the details, but one compared drug A to drug Z, and the other compared drug B to drug Z. Although there are no trials directly comparing drug A to drug B, the consultants were asked their views of A versus B based on the indirect comparisons of A versus Z and B versus Z.

Achilles: You’re losing me.

Tortoise: I’m not surprised! Here’s how Dr. E explained it. If the difference in benefit between A and Z is (a − z), and the difference between B and Z is (b − z), then the difference between A and B is (a − z) − (b − z) or (a − b)—assuming, of course, that Z is the same for both. It’s a simple matter of logic. She calls it the transitory property or something. Things that are unequal to the same are unequal to each other.

Achilles: That make sense…I guess. So, what was the problem?

Tortoise: Dr. C acknowledged the logical validity of the method but doubted that the 2 populations treated with drug Z were sufficiently similar to justify the analysis. In effect, he weighted their differences more heavily than their similarities. As he saw it, Z was not the same for both.

Achilles: He was treating individual trials like individual patients.

Tortoise: Precisely! Dr. C was analyzing an epidemiologic problem from the perspective of a clinician. Dr. E instead considered the similarities more important than the differences.

Achilles: A difference that makes no difference is not a difference.

Tortoise: Spot on! According to Dr. E, epidemiologic problems should be approached from an epidemiologic perspective .

Achilles: So, just as Dr. E shouldn’t tell a clinician how to treat an individual patient, Dr. C shouldn’t tell an epidemiologist how to treat an individual trial. But what if there’s more than 1 trial?

Tortoise: Ahh! That’s a horse of an entirely different color. There’s something called meta-analysis that’s specifically designed to analyze groups of groups.

Achilles: How’s that work?

Tortoise: Just as a trial consists of a group of individuals, a meta-analysis consists of a group of trials. As we’ve seen, trials can’t tell us anything about treating a particular individual, but only about a particular group of individuals. That’s why Dr. E didn’t want to tell Dr. C how to treat me. Likewise, meta-analysis can’t tell us anything about a particular trial, but only about a particular group of trials.

Achilles: I’m getting a little dizzy…

Tortoise: Look: Dr. E was right in not telling Dr. C how I should be treated. Only a clinician can do that. But for similar reasons, Dr. C—excellent though he might be as a clinician—was wrong in telling the regulators how they should treat a group of trials. Even Dr. E—excellent though she might be as an analyst—can’t do that. That’s the job of a meta-analyst. Sometime in the future, we’ll have so many meta-analyses, we’ll begin doing meta-meta-analyses to make sense of them all. There’s no end to it, I fear.

Achilles: Now I’ve got a headache!

Tortoise: Jolly good! The cure begins by recognizing you have a problem. Meanwhile, I’ll just plan to watch my weight and exercise more. Wanna race me home?

A short time thereafter, Dr. C and Dr. E meet over lunch to discuss the issues raised by Tortoise and agree to seek the counsel of a controversial, but widely regarded, epistemologic adviser, the Reverend Thomas Bayes. With his guidance, they explore a broad range of arcane topics bridging the art and science of medicine and thereby begin to resolve the philosophical differences deriving from their alternative perspectives. They continue to practice together—and are all the better for it.