25 Mar 2010


Our oncologist friends told us that because their field was advancing so rapidly, they seldom cited full papers. To provide updated information, oncologists mostly cited conference abstracts. They anticipated they would soon have to cite data from press conferences.

That made us look very backward. At the time when the development of therapies for chronic hepatitis B was at its height, we did cite some conference abstracts. This dampened a lot in the last one to two years after the data of the major drugs in this field had been published as full papers. After all, why should we cite the year six data of a drug presented in the latest conference when the year five data have been published in peer-reviewed journals? The early online publication of accepted papers by biomedical journals also facilitates rapid dissemination of data. Certainly, we are far from the stage of citing New York Times except for fun.

A comprehensive review of 29,729 conference abstracts found that only 53% of all abstracts were eventually published as full papers at 9 years.[Scherer RW et al. Cochrane Database Syst Rev 2007 Apr 18;(2):MR000005] The observation may have a number of reasons. There may be publication bias where negative studies are less likely to be published. Some studies may have major methodological flaws. While they look alright in abstract form, they cannot get through the scrutiny of peer reviewers when presented in full. Moreover, some conference abstracts are obviously put up by pharmaceutical companies for promotional purpose and never intended to be published as full papers.

While it is charming to cite a reference to support our view, we must remember that big message does not equal big evidence. We should scrutinize the reliability of data, whether they are from scientific papers, abstracts or New York Times.

18 Mar 2010


Shortly after the meeting, I had lunch with SW who came to Hong Kong for a job interview. Having spent three years in the PhD program, he would like to pursue his career as a clinician-scientist.

“Actually,” he remarked, “I seldom see my supervisor. He is so busy that I may meet him only once every several months. I learned most of my laboratory techniques and bioinformatics from a new postdoctoral fellow.”

After talking to quite a number of postgraduate students around the world, I am afraid this is a common phenomenon. In some countries, big professors may have one to two dozens of postgraduate students at any time. There are occasions in which a supervisor does not know the existence of a student, not to mention what he/she is working on.

It is understandable that students only want to follow big names though they are most unlikely to be available to provide supervision. Choosing among less famous supervisors can be even riskier. As such, how can the students survive?

My usual advice is not to depend too much on the supervisor. Find a big center and try the best to learn from people around you. After all, postgraduate study is about self-learning and problem solving.

11 Mar 2010


At a meeting last week, CM explained the rules of choosing postgraduate students in the department. I beamed at my mentor. “Game theory!”

It goes like this. First, each department chooses a number of postgraduate students according to the assigned quota. These students would certainly be recruited. After that, there will still be vacancies in the whole faculty. Each department would then select candidates from the remaining list for central bidding. The latter students would compete with candidates recommended by other departments. Those with the best academic results would get the remaining seats.

Let’s ignore the issue of fairness for the time being. The central bidding step is the most fascinating part of the whole process. If we place less impressive candidates in the guaranteed list and pose the best students for central bidding, we would have a higher chance of getting more students for the department but also risk losing the best candidates. Conversely, if we put the best candidates in the guaranteed list and recommend the rest for central bidding, we for sure will retain the candidates we want but would less likely get extra students.

In game theory jargons, since the total number of vacancies for central bidding is fixed each year, this is a multi-person, zero-sum game. The payoff is the number of extra students obtained. Assuming that the quality of applicants in each department is similar, we can presume that a department will have almost 100% chance of winning if it sends the best candidate for bidding while other departments send those on the waiting list. Similarly, a department will have nearly zero chance of winning if it sends a waiting list candidate for bidding while others send the best ones. When two departments send the best or the so-so candidates for bidding at the same time, we can assume that the chance of either winning would be 50-50. The probability will decrease by proportion as more departments decide to send the best candidates.

In this situation, because there are multiple players in the game, the probability of having none of them choose the best (or very good) candidates for bidding is too trivial to consider seriously. Therefore, the only chance to win is to send the best student for the competition.

Then there is no need to hesitate, right? Not quite. In the above argument, the only payoff considered was the total number of extra candidates a department obtains. In reality, the perceived value of getting an extra candidate and that of losing a superb candidate should be taken into consideration too. That weighting is subjective.

I cannot say for others, but what is the perceived value for me?

That would be better explained by Cheshire Cat.

Alice: I was just wondering if you could help me find my way.

Cheshire Cat: Well that depends on where you want to get to.

Alice: Oh, it really doesn’t matter, as long as …

Cheshire Cat: Then it really doesn’t matter which way you go.

4 Mar 2010


Last week, we received a peculiar e-mail proposing a change in our specialist on-call system. Instead of starting our call duties when we arrive at work in the morning, we are to hand over duties at midnight.

The news was received with mixed feelings. Most specialties had already anticipated the change and stopped providing on-call service. Team members who continued to provide emergency service were particularly anxious. If one is paged right after midnight and again before the next midnight, this means one may need to work through two consecutive nights. The dissociation between the specialist and general on-call system also results in lengthening of working hours to fulfill both duties.

Why all these troubles? Under the latest court verdict, doctors are entitled compensatory leave after non-residential call. If the call day spans across midnight, administrators are worried that doctors may take advantage of the system and request two days of compensatory leave.

Having never taken compensatory leave before, I felt somewhat humiliated for being considered as a professional who may take advantage of others. On second thoughts, who can you blame?

During the Zhou Dynasty, Yurang (豫讓) served Earl Zhi (智伯) in Jin (晉國). During a civil war, Earl Zhi was killed by his enemies. Viscount Xiang (趙襄子) hated Earl Zhi so much that he used his skull as a cup.

After this, Yurang repeatedly tried to assassinate Viscount Xiang for revenge. To avoid being recognized by the Viscount and his followers, he rubbed lacquer on his skin to produce ulcers and swallowed hot ashes to destroy his voice. Although all his plans failed in the end, even Viscount Xiang was moved and asked why he had to suffer all these.

“Others treated me as a commoner,” answered Yurang, “and I repaid them as a commoner. Earl Zhi treated me as a hero and I should repay him as a hero.” 《史記•刺客列傳》

Courageous as it might, Yurang’s answer reflects the natural response of most people. My only challenge is if you do not want to be treated as a commoner, do not act like one.