Medical creep happens when doctors practice “outside the box.” They perform procedures or prescribe treatments in the absence of clear evidence that patients will benefit. For example, doctors often use drugs or devices approved by the Federal Drug Administration for treatment of one condition (based on strong science evidence) to treat patients with other conditions (lacking any scientific evidence). Not surprisingly, pharmaceutical companies and medical device manufacturers promote this kind of creep, sometimes called “off-label” usage. Government regulatory agencies and insurers tend to take the opposite view; creep blows budgets and, sometimes, harms patients. But medical creep sometimes saves lives.
Cardiopulmonary resuscitation (CPR), the procedure used to treat many patients who suffer cardiac arrest, is an example of medical creep. Before 1960, the only way to treat cardiac arrest required surgical opening of the chest cavity and manual cardiac massage, whereby the surgeon holds the heart in his hands and squeezes it rhythmically to pump blood to the brain and other vital organs. Extraordinarily heroic, such “open-chest” cardiac massage was attempted rarely—and succeeded even more rarely. Then, in 1960, surgeons at Johns Hopkins reported their experience with 20 patients with whom they had used a new CPR technique called closed-chest cardiac massage. This team of surgeons, who had also developed the external cardiac defibrillator, had devised a way to pump the arrested heart without opening the chest surgically. They successfully resuscitated every one of the first 20 patients they treated, 14 of whom (70 percent) survived without brain damage or other ill effects. They concluded that “the method… can be used wherever the emergency arises, whether that is in or out of the hospital.”
And so the creep began. Six years later, the National Academy of Sciences issued its own report, including illustrative diagrams, instructing medical personnel in the proper performance of closed-chest cardiac massage and CPR. Soon thereafter, all cardiac arrests in U.S. hospitals were treated with these methods. Resurrecting the dead became medicine’s obsession—and not just inside hospitals. Many U.S. communities funded “mobile intensive care units” that enabled ambulance personnel to deliver CPR outside the hospital, too. In other words, CPR became—and remains today—the “default” treatment for every person who dies. Unless you explicitly forbid it, you will leave this world the same way: Doctors or ambulance personnel will pump on your chest, put a breathing tube down your throat, squeeze oxygen into your lungs, jab you with needles, and electrocute your heart.
Anyone who has watched these things done to a frail, demented, or terminally ill 90-year-old person understands just how crazy, and creepy, this can be. How did the CPR technique pioneered by the Hopkins doctors (which continues to save many lives today) become an accepted final rite of passage for everyone?
Three things doctors and researchers didn’t understand in 1960 contributed to CPR creep. First, the Hopkins researchers, in their initial study, treated a very narrow spectrum of patients; most were young, healthy people (including several children) whose hearts had stopped during elective surgery, victims of anesthesia mishaps. Their impressive success rate (70 percent) was much higher than the success rate for patients who suffer cardiac arrest in hospital today (5 to 15 percent), most of whom are elderly people with advanced heart disease and other serious conditions. (Recent large studies involving only elderly patients have documented CPR survival rates as low as zero and as high as 18 percent, with up to one-quarter of all survivors suffering permanent brain damage.) We now know that many new treatments, studied initially in a narrow spectrum of patients, aren't nearly as successful when used in a broader patient population. Ignorance of this “spectrum bias” potentiates creep.
Second, CPR creep reflects our failure to understand the difference between efficacy and effectiveness. The efficacy of a medical treatment refers to whether it can achieve its desired effect when studied under the ideal conditions of a research study. In contrast, the effectiveness of a medical treatment measures how well it performs in the “real world,” where conditions are far from ideal. For example, the first study of out-of-hospital CPR in 1967 found that 50 percent of cardiac arrest victims in Belfast were resuscitated successfully. But when doctors in New York tried to replicate these spectacular results on the streets of lower Manhattan, their “mobile ICU” ambulances were able to save only 6 percent of cardiac arrest victims. (Recent studies have shown that survival rates after out-of-hospital CPR range from 2 percent in urban Chicago to almost 20 percent in suburban Seattle.) Medical innovations that prove efficacious in one research setting often are ineffective elsewhere. Confusing efficacy with effectiveness promotes creep.
Mass media promote medical creep, too. Episodes of television “doctor shows” like ER, the popular weekly series based fictionally on Chicago’s Cook County Hospital (where I worked at the time), perpetuated spectrum bias about CPR and grossly inflated its effectiveness. Researchers found that two-thirds of all (fictional) cardiac arrests portrayed on ER (and other doctor shows) involved young patients who had suffered rare events like drowning or lightning strikes, rather than old people with heart disease (who account for 90 percent of cardiac arrests in real-life settings, including Cook County Hospital). Not surprisingly, most of these fictional TV patients did well, unlike the vast majority of CPR recipients in real life. In addition, as the researchers noted:
On television, the outcome of CPR was generally portrayed as either full recovery or death … [as] if CPR were a benign, risk-free procedure that offered a good hope of long-term survival in the face of otherwise certain death … But CPR can lead to prolonged suffering, severe neurologic damage, and an undignified death. In these … medical dramas and reenactments, such outcomes were never portrayed.
Do patients and their relatives who insist on receiving CPR understand these things? So much depends on whether doctors explicitly debunk CPR creep when they discuss resuscitation with patients or their surrogate decision-makers. One geriatrician, who cared for elderly patients in a long-term care facility wrote that 36 of his 40 patients (whose average age was 87) told him that they wanted CPR. This confused him until he found that they did not understand the low likelihood of benefit and the potential downside of CPR. After he had discussed the realities with his patients, 39 of the 40 opposed resuscitation. Both doctors and patients are susceptible to the notion that if a potentially beneficial treatment can be done, it should be done—even if we don’t know how often the treatment succeeds in a particular clinical situation or how its potential benefits compare with its potential harms.
Nevertheless, creep can propel medical progress, too. For example, it was not so long ago that patients with bacterial endocarditis were never offered surgery. Never. Even when their hearts began to fail due to worsening function of their infected valve, surgical treatment was not considered, because it was assumed that a new valve, surgically sewn into infected tissue, would become infected, too. But, in 1965, a surgical team replaced the infected heart valve of one such patient, saving his life. (Why did they operate? Because they felt certain that the patient would die if they didn't operate.) Ten years later, 239 endocarditis patients worldwide had undergone similar surgery, most of whom (73 percent) survived. Since then, thousands more endocarditis patients have opted to undergo surgery despite the fact that 25 to 30 percent of them will die. These (highly selected) patients are offered such risky surgery because their mortality rate is twice as high (55 to 60 percent) without it.
Nowhere is medical creep more prevalent today than in the treatment of cancer patients. For example, bladder cancer metastatic to bone has a median survival of about six months. Recent studies have shown that new chemotherapy treatments can extend this median survival to about 12 months. None of these studies included patients like my father, a blind, bedridden 91-year-old man with poor “functional status.” (Frail, elderly patients like Dad are typically excluded from research studies of new treatments.) And yet, lacking any evidence about patients like him, doctors offered Dad the new chemotherapy treatment (which he refused). A patient of mine, Mr. Mukaj, younger than Dad and very similar to patients in the relevant research trials, experienced creep of a different kind. In order to make it possible for him to receive the new chemotherapy treatment (at home in Croatia), our doctors planned to remove his bleeding bladder and replumb his kidneys, a radical operation rarely done for patients whose cancer has already spread to distant sites. Is this kind of creep bad?
Precisely because every patient is different, creep in medicine will continue, for better and for worse. Statistical analyses of large groups of patients cannot predict outcomes for individual patients.
The poster child for this idea is Stephen Jay Gould, a famous scientist himself. Quoting Mark Twain in an essay titled “The Median Isn't the Message,” Gould identified “three species of mendacity, each worse than the one before—lies, damned lies, and statistics.” A renowned paleontologist and evolutionary biologist at Harvard, Gould was diagnosed with abdominal mesothelioma, a rare and deadly cancer, at the age of 40. After a systematic search of the medical literature (in 1982), he learned that patients with this disease had a median survival of eight months. Gould realized that this prognosis was not as gloomy as many (less statistically savvy) people might think. Rather than “I will probably be dead in eight months,” this median survival means that one-half of patients die within eight months but the other half live longer. Most important, because the mesothelioma survival curve has a very long “tail,” a few lucky patients will live a lot longer.
An otherwise healthy young man, Gould thought he might be one of the lucky ones and signed up for an unproven, experimental treatment. When he died 20 years later, he left behind not only a lesson in statistics but a legacy of hope for all cancer patients. Gould wrote:
All biologists know that variation itself is nature’s only irreducible essence.
Variation is the hard reality, not a set of imperfect measures [like] means and medians … Therefore I looked at the … statistics quite differently—and not only because I am an optimist who tends to see the doughnut instead of the hole but primarily because I know that variation itself is the reality. I had to place myself amidst the variation.
This is the challenge all doctors face: trying to see each of their unique patients within the context of natural variation.
This post is adapted from Brendan Reilly's One Doctor: Close Calls, Cold Cases, and the Mysteries of Medicine.