Michael D. Lockshin, MD

Atul Gawande, the surgeon-writer, is outraged (A Lifesaving Checklist 12/30/07.) Apparently an ongoing medical study, that he endorsed, was stopped because of a bureaucratic rule. The study was trying to this question:  Are there fewer hospital infections if doctors follow a checklist, as do airline pilots on take-off (sample: Have you washed your hands? Have you draped the patient with a sterile cover) than if treat patients according to their own rules? This, Dr. Gawande fomented, is bureaucracy at its worst. Stopping the study is outrageous, because the benefit of checklists is so obvious. (He did not ask why, if the benefit is obvious, the study should be done at all.) 

The study asked the doctors for half the patients to use checklists to prepare for a procedure while the doctors for the other half would prepare as they saw fit. The reason the bureaucracy—in this case the Office for Human Research Protections (OHRP)—gave to cancel the study was that the investigators had not asked patients to consent to participate. Some might have refused. After all, if you were the patient, would you want to be under the care of the doctor who did not use a checklist?  It is, of course, possible that using a checklist makes no difference at all, that doctors do the right thing all the time, but, if you are the patient, do you really want to take that chance? Shouldn’t you be asked? 

Dr. Gawande thought that insensitive bureaucracy had stupidly ended a useful study. He missed a larger point. First, we do not know whether the checklist does or does not make a difference (preliminary studies suggests it does), so it is perfectly appropriate to do the test—improving our knowledge is clearly worthwhile. Second, an experiment on a patient is not just good or bad. Human experiments engage several value systems not all of which are easily seen. The conflict between Dr. Gawande’s wish to continue and OHRP’s command to stop was a conflict between, and a different weighting of, these different sets of values. 

In research using patients, probably the highest value is accorded to patient autonomy.  A research patient has a fundamental right: to refuse to participate. This right trumps any perceived benefit of the research. OHRP clearly focused on this value; patients should have been given the right to refuse. 

The second most highly ranked value is that of privacy. This right was emphasized in first days of AIDS, to protect against discrimination or the public release of personal information.  To protect this right, medical researchers now are burdened with cumbersome regulations; all patient-identifying information must be removed from all documents.  Research data must be kept in locked cabinets. Physicians may not transmit information about patients to anyone, including family, except another treating physician, unless a patient expressly consents (a ruling that applies to all aspects of patient care).  This is not only a matter of ethics; in many states it is a matter of law. Researchers and treating physicians find it exhausting to comply with the paperwork engendered by these rules.  Privacy rules delay research; burdensome regulations cause some potential researchers not to enter the field at all.  Since not very much that is private is a risk in the checklist research, Dr. Gawande seems to rate the privacy value low; he would dispense with the bureaucratic needs. 

A third value can be summarized by this sentence: Medical research should provide the greatest good to the greatest number.  The checklist study would do this. The investigators may have assumed that checklists and observation of doctor behavior does not count as research. They may be correct. Audits of practice, whether or not one collects additional data, are usually exempt from research rules.  Furthermore, the agencies that rule over human research exempt some research from consent—studies of population behavior, for instance—and provide expedited approvals for low risk research.    The researchers should have asked for an exemption, but they certainly know that local human research monitoring boards are inconsistent.  When the same research is done in several different sites, as the checklist study was done, one local board may consider it exempt from consent rules, while another may require full consent.  Such inconsistencies exasperate researchers and delay research if not force it to a complete halt. Had OHRP been asked before the experiment began, it might have ruled that checklists are practice audits exempt from consent, or it might not. One does not know until one asks. 

Society values innovation and rapid dissemination of research results.  These values contradict the values of autonomy and privacy. Innovation and dissemination demand flexibility of experiment rules and open and ready access by the public; autonomy and privacy restrict the rate of flow and narrow the volume of information that may be shared.  Even the creation of a widely desired a national medical database is hobbled by privacy concerns. 

The value “do no harm” has a very high priority. Pragmatically, however, the fact of the study is closer to “do less harm” or “do more good” than it is to “do no harm.” Consider the comparison of a new drug with an old; the new one is thought to be safer, but it may not be. If the new drug is better, those receiving the old will be (in one sense) harmed by the experiment; if the new drug turns out to be, in fact, more toxic, those receiving it will be harmed. Or consider the circumstance in which early surgery is thought to give a better cure rate than a medical treatment. Either the surgery or the medical-treated patient group could be harmed or helped, depending what the answer turns out to be. In some cases an experiment demonstrating good yields a very small, one might say trivial, improvement; in another case the difference might be as great as that between death and cure. The amount of potential good or potential harm can be weighed. Committees that judge the overall value of a research study on humans use common sense rules but apply no formal measurement scales to these values. Absent a metric, OHRP’s and Dr. Gawande’s opinions collide. 

So what to do?  Protect the privacy of the research subject?  Facilitate the rapid conduct of research? Protect at all costs against possible harm? Or make possible a public good? 

One cannot reconcile the conflicting demands.  One can try to make the committees more uniform, the rules more efficient, but these are minor fixes. Instead, it would be best to concede that our contradictory rules derive from different societal values, and that no one rule is absolute.  We can assign point systems—degrees of risk—that have positive (for benefit) and negative (for risk of harm) numbers instead of ruling by yes/no criteria. There are different levels of privacy and autonomy, and different degrees of good and harm. Scientific importance can be quantitated. The potential public gain can be as well.  The numbers can be summed, and the potential worth of the research scored. 

The table below gives examples. How the scoring system is actually constructed is not an important point. What is important is that the investigators should document how their proposals address each of the values; what is important is that the committee should systematically weigh each value when it decides to approve or disallow a specific project of human research. When that happens, Dr. Gawande, instead of railing against a faceless bureaucracy, will see its decision from another point of view, and OHRP will understand that some quantitatively small risks can be waived.

The New York Times today had an article written by Stephanie Saul entitled THE COVERAGE GAP Avoiding Medicare’s Big Hole. The last three paragraphs discuss the ‘fact’ that many patients are reluctant to discuss the cost of their prescriptions with their doctors. The claim that patients and their doctors do not want to discuss cost of medicines is both incomplete and wrong. Several factors inform both my choice of what to write on a prescription pad and the out-of-pocket cost to the patient: * Different insurers choose as "preferred" prescriptions different medications of the same class, presumably because the insurer has purchased the chosen drug in bulk at discount. Unfortunately, when a physician works with six or more different insurers, keeping in mind which is preferred for which insurer–they are never the same–is difficult and tedious to find out. * Insurers that do have "preferred" prescriptions change their preferences at random, unpredictable intervals, so that a "preferred" prescription in October may have changed when the patient returns in April. * The same problem occurs when a patient changes insurers, as many annually do. * Because of co-payment rules, patients often ask me to write a three-month supply for a medication they need to take for 10 days, the former having no co-payment, the latter requiring out-of-pocket payment. * A patient may speak of the cost of medication at a convenient pharmacy, but the cost for the same medication will be quite different at, for instance, a Wal-Mart on the other side of town. * Some patients clearly have adverse or beneficial responses to one brand of "equivalent" drug but not to another–and are quite explicit in saying so. * New York State has a generic drug prescription law, requiring me to certify by a separate signature that a specific brand is required. If I write a prescription using the easier-to-remember proprietary name and do not sign the box certifying the brand, the pharmacist is required to supply the cheaper generic–or equivalent–instead. Most pharmacists do, so I have to ask my patients on return visit what drug they have been given in lieu of the one I prescribed. * Years of practice have given me a comfort level with some medications compared to others, an instinct, if you will, regarding good and bad things to expect. To use a usual doctor's double-negative: this experience is not irrelevant. The problem is not so much that patients and doctors are embarrassed or unwilling to discuss drug prices as it is that, when all factors are considered, the facts of cost are hard to ascertain. If there is any simple fix, it is to standardize costs and preferences across all insurers and suppliers, so that when I write a prescription I can reasonably predict what will happen next.

A patient with APS recently emailed me with a question about vaccinations for her two small children. Her concern was that her children may be genetically susceptible to having a reaction or could develop an autoimmune condition from their vaccinations. She like the rest of the general public has gotten conflicting information about vaccinations, not only for autoimmune diseases, but other illnesses like Autism as well. And she wanted to know where she could get information about the question of vaccinations and autoimmune diseases. Here is what I wrote her:

There is a Nobel Prize available for the person who can answer your question(s). Current thinking is that lupus (and, by extension, APS) is triggered by a common virus in a susceptible person. The leading suspicious virus is cytomegalovirus (CMV), which causes a mononucleosis-like disease; Epstein-Barr virus (EBV), which causes mononucleosis, is another candidate. There is an extensive medical literature on this topic. John Harley and Judith James, in Oklahoma, have published several papers indicating that the infection likely occurs a decade or more before the disease becomes obvious. You can hunt for these papers at the National Library of Medicine site (PubMed) and search on the terms “Harley J AND lupus” or “lupus AND CMV” or such to find the papers, or, given a day or two, I can compile a list for you.

None of us in my world think of vaccinations as likely virus triggers; on the other hand, if you have no concept of how devastating polio, tetanus, diphtheria, measles, etc. can be (lethal, brain-damaging, permanently disabling) you may underestimate the value of vaccination. If you choose not to vaccinate your children, you rely on “herd immunity”, that is, the luck that others they encounter will be immune and protect them. Loss of herd immunity has resulted in epidemics as recently as this year of these common diseases in fundamentalist communities, Native American communities, and “hippie” colonies in the United States this year, so these diseases are by no means gone.

Bottom line: although your children may have inherited (a degree of) susceptibility from you, they are somewhat protected by their father’s genes; vaccinations are not the things to fear. Rather, common infections, some of which can be prevented by vaccination, are more likely triggers.

Three years ago, at a journal club presentation at our hospital and on the hospital web site, I expressed my opinion that the data then available suggesting an excess cardiovascular risk from use of Vioxx (rofecoxib) were inconclusive. I based this opinion on the following. The data compared a large dose of Vioxx (50 mg daily) with a common dose of naproxen. Moreover, the added protective effect the naproxen provided that Vioxx did not provide was not duly taken into account . I believed it possible that the cardiac events reported were not due to the drug but rather to possible misuse of the drug. Several subsequent research papers have pointed in the same direction but have also been flawed. One example was a recent large-scale epidemiological study suggesting that most nonsteroidal anti-inflammatory drugs increase cardiovascular risk above normal, but that Vioxx, as used in the community (dose not specified, concomitant use of aspirin not specified), increased the risk of heart attack somewhat more than did the other drugs. The reported risk was not great and was maximal within the first 90 days of taking the drug. Results from the most recent study, the one which led Merck to withdraw Vioxx from the market, were not available until this week. These too were reported to show an excess cardiac risk, approximately double the normal risk, and, inexplicably, occurring not within 90 days of first taking Vioxx, but after 18 months of continuous use. This timeframe introduces more questions. The study, however, was prospective and the data appear to be the most valid of the data available to date. Based on the results of this study as reported, I now must conclude it likely that Vioxx presents increased cardiovascular risk as compared with some of the other nonsteroidals. Clearly, I no longer endorse use of Vioxx. When the story of the withdrawal of Vioxx from the market broke, I was interviewed by Ms. Gina Kolata of the New York Times, whose article was published on October 1, 2004. The following is a partial quote from her article. It accurately reflects my current opinion.

But others, including Dr. Michael Lockshin, an arthritis specialist at the Hospital for Special Surgery in New York, were less certain. Dr. Lockshin said that, until now, "the data were quite minor" to indicate that Vioxx was causing a problem. The original study's findings had an alternative explanation, he said, and the other studies were complicated by the fact that they lacked enough rigor to make the case. The new data, however, convinced him. Even if Merck had not withdrawn the drug, he would no longer prescribe it, Dr. Lockshin said. "Although the numbers are not huge in terms of risk, these are the things that I, looking at it as a practitioner, couldn't say, 'Oh, well. This drug is as safe as another.' "

He added, however, that some of his patients might have difficulty finding an alternative to Vioxx. Pain relief, he said, is idiosyncratic and unpredictable. "Some patients respond beautifully to one drug and not to another."

See the complete article on the Hospital for Special Surgery's web site