unknown unknowns

Heard in a Coursera class: research should address the gap between the known and the unknown. I was under the impression that research should look at the unknown, at least because tertium non datur and studying the known is really low among funding priorities. I kept thinking for a few minutes what could be that gap, when it struck me: this question has no answer, precisely because that gap is more esoteric than the unknown. There one finds (or rather cannot find) the unknown unknowns.

Mammography quackerism

While on the hallways of BU School of Medicine (waiting is one of the privileges of a PhD candidate), I happened upon handouts from the med school DRx class. BU students are told that mammograms are recommended yearly, starting from the age of 40. Now, US Preventive Task Force recommends starting at 50, similar to Canadian, German, and Italian guidelines. England & Wales NICE is lowering the start time from 50 at 47 (not sure why). Pretty much every country suggests or pays for bi-annual tests. Where did yearly from 40 come from?

Stats 101

From a paper that seems shoddier than I thought: “Abc neutralizing antibody does not reduce the number of XYZ-positive colonies in the assays shown in G; n = 3.” The only thing can be proven after as few as three measurements is the sainthood in the Catholic Church.

“price effective” fun

Ophtalmology just published a cost-effectiveness comparison for treating diabetic macular edema comparing {ranibizumab + laser}, {triamcinolone + laser} and laser alone. Authors’ mathematical simulation shows that eyesight is better preserved for the same money when ranibizumab (brand name Lucentis) is used rather than the cheap, generic triamcinolone.

The elephant in the room is of course bevacizumab (brand name Avastin) which was not part of this comparison. Avastin is 30 times cheaper than Lucentis and the two have been found to be equally effective in treating diabetic macular edema on the relevant outcome, eyesight, used in the current analysis. Manufacturer-sponsored studies show that Avastin carries a higher risk of death than Lucentis, but the company has essentially failed to push the more expensive option in UK, where costs are better controlled than elsewhere.

Unsurprisingly, the disclosure sounds as follows:  “This project was completed by Dr. Dewan as part of his research elective as a fourth-year medical student at Washington University School of Medicine (WUSM). No funding was received from any source for support of this project. [...] Dr. Kymes receives research support from Genentech and Pfizer. He acts as a consultant for Pfizer, Allergan, Genentech, and TreeAge (maker of the software used in this project). Dr. Apte acts as a consultant for Genentech, Eyetech, Allergan, Ophthotech, Alimera, Regeneron, and Baxter.” Roche owns Genentech and sells both Avastin and Lucentis.

More unsurprisingly, the Medscape story title is “Diabetic Macular Edema: Pricey Option May Be Cost-Effective”.

PSA is resuscitated briefly

A study in today’s NEJM shows that early cancer may benefit from surgery. However, PSA makes an unwelcome addition to the study. Somehow, the fact that the study participants had to be “diagnosed” through PSA screening may be interpreted as evidence that this study supports PSA screening.

After reading this and the preceding paper, I am still unclear where is PSA screening coming into the picture. The participants were required to be previously diagnosed with prostate cancer by any means. I could not find anywhere a sign that they were required to be asymptomatic before their PSA was first measured. But after this diagnostic was made and after they agreed to participate in the study, their PSA was measured again, quite opposite to the order of action in real-life screening, where PSA measurement come first, diagnosis – later. Because the study does not start at PSA screening time, the number of unnecessary biopsies to get to this sample size is not available. Unsurprisingly, in the study population, the PSA had a (pardon the pun) healthy increase to a median of 7, whereas in-real life screening, difficult questions are asked at 4.

If the hypothesis was, as the title says, surgery helps well-diagnosed patients, then yes, the authors have provided evidence, although I am not sure whether this is novel. The novelty, PSA screening helps, is not supported. Eighteen years in the making, PIVOT remains an informative trial, but does not answer today’s questions.

One more nail into the coffin of indiscriminate mammography

Comparing the incidence of breast cancer deaths among Swedish counties, Autier and colleagues found that there was no advantage for those who got on the bandwagon of mammography screening early on, sometimes as early as 20 years before the sluggish. Swedish women have one of the best rates of taking the test, 85-95%. But from here, it is only downhill. The study is almost identical to the Danish study.

evidence for Pap smear screening

Reading Should I Be Tested for Cancer?: Maybe Not and Here’s Why by HG Welch, I am surprised to learn that nobody bothered with a randomized trial of the Pap test as screening.

Paul Nurse on class

Melvyn Bragg did his best to carry his charms from BC Radio 4 to the other medium, the TV, and for the most part, he succeeded. When the program started (must be in UK to watch), I was apprehensive, thinking that I will be offered the same images from the BBC archives, documenting how UK moved from food rations to decency up to the late sixties, and back to inequality and insecurity ever since. Indeed, there are many cliches, such as the Queen’s coronation, which is by now more worn out than the Geico ads.

One extraordinary appearance makes up for all that. Paul Nurse, the Nobel Prize winner who discovered the enzymatic mechanisms of cell division, points out that the science field is up for grabs for lower classes. He says he did not go to the opera, did not go to the theatre, so he never considered an arts career. On the other hand, science was open for all. It did not matter what your class was. It was a more leveled field. Maybe that is what he felt during the late sixties, when he was in school.

ActRIIB – and cancer?

An interesting factoid from a talk I have seen today: an ActRIIB antagonist slows down cancer cachexia and increases survival.

Something else puzzles me. The two consequences have been linked for long. Cancer-induced muscle loss is a major cause of death, or so the generally-accepted wisdom goes. By this logic, anabolic steroids would be also useful. Some (rare) trials do show that. But even an increase in burger intake would help. For some reason, the medical community prefers to extend patients’ lives by three weeks, only if they suffer and pay as much as they would in three years. What is ActRIIB adding? If survival is increased by muscle mass increase, why not go for burgers?

I am intrigued about what ActRIIB blockade does to normal bodies. Can it be used by bodybuilders? Can it be used in cattle? If  not, does the muscle mass have a set point, along the lines of the body mass set point maintained by leptin?

Hospital weekends

Another article (behind paywall) in the series “get sick only during regular business hours” show that in England admission over the weekend increases patient’s chance to die by about 10% compared with those admitted during business days. Data was validated on a set of 250 US hospitals, where they found a 20% increase in risk of dying associated with weekend admission. In England, but not in US, Sunday was even worse than Saturday.

On the other hand, should one make it, the chances of dying during the weekend shift is about 10% lower than on Wednesday. It is now unclear if, over the weekend, the puerperal fever carriers must be called to work or rather sent home.