The June issue of McKinsey Chart Focus has a chart showing that the number of cardiothoracic surgeons in England increased by 60 percent from 1998 to 2007. Meanwhile, with the growing popularity of angioplasty (a nonsurgical treatment for blocked arteries), the number of surgeries has fallen by more than 10 percent. This means that the number of surgeries performed per surgeon has fallen from 175 per year to 93 per year, a decline of almost 50 percent. This data comes from a longer article in McKinsey Quarterly Dec 2009.

McKinsey uses this data to support the argument that it is difficult to forecast labor demand for specialists who take a long time to train and that there is a glut of cardiothoracic surgeons in England. Both statements may be true, and the longer article is quite convincing. But shortening the argument by just using a couple of charts to prove the point is wrong as I will explain.


Who wants to be a cardiothoracic surgeon? Image from McKinsey

First, we have no evidence as to whether the number of surgeries performed per surgeon in 1998 is optimal, too many, or too few. If it is too many, and was causing poor outcomes, then it would be good to reduce each surgeon’s caseload. We also don’t know the distribution of workload among the surgeons and how that changed.

Second, labor markets are not static, even for surgeons that require lots of training. Prior to deciding to become surgeons, newly graduated medical students use salary as one of the factors when deciding which specialty to enter. If salaries are not falling and there are no stories of unemployed cardiothoracic surgeons, then there is no surplus (in the economic sense). Yes, they may be entering a field that may be unfulfilling because there isn’t enough work. But that is the result of policies of the NHS which hopefully can be fixed. Also, once young surgeons enter a specialty like cardiothoracic surgery, they are not stuck there forever. It is possible to switch to another specialty. This should allow some balancing, if the labor demand changes.

Third, hopefully, hospitals don’t have surgeons just sitting around if they are not performing surgery. In particular, we can presume they are performing the angioplasty treatments. The number of these procedures has been rising rapidly. This isn’t shown in the graphs above. In fact, since angioplasty is considered lower risk than open surgery, they probably perform lots more of them than surgeries. Surgeons may actually be busier (treating more patients) than before.

Thus, we can conclude that on its own the McKinsey chart doesn’t provide enough evidence to support the argument that improper labor demand forecasting is a common problem or that there are too many cardiothoracic surgeons in England. Again, reading the full article is more convincing.


Incidentally, this is a good place to illustrate a problem that can occur when comparing the benefits and costs of a new medical procedure with an existing one. If angioplasty (or some other noninvasive intervention) is being performed on more patients, that must mean more marginal cases are being treated. That is, there are patient cases where surgery was not indicated because benefit did not outweigh the risk. There are two types of patients that were previously left untreated. First are patients who are healthier than the treated patients and whose cases were not considered serious enough for surgery. Since the risks have dropped with angioplasty, some of them can now be treated.

The second group consists of patient who had other comorbidities that made them poor candidates for open surgery. Again, since the risk has dropped, some of them can now be treated using angioplasty.

Now look what happens to the outcome statistics before and after the introduction of the new technology. Let’s divide patients into three groups, Treatment provided, No treatment required (case not considered serious enough for intervention), and No treatment advised (case too risky for surgery). Let’s take a look at the hypothetical 5-year survival rates for the three groups before and after the replacement of open surgery with angioplasty.

Survival rates for two treatments (hypothetical data)

5-year survival rate

Treatment provided

No treatment required

No treatment advised

Open surgery








Switching from open surgery to angioplasty has a small negative effect on survival of patients receiving treatment. But that doesn’t mean it is less effective than surgery. Because the composition of patients has changed, there is no way to know if it better, worse, or the same. As a weird side effect, the switch in treatment has effects on the groups that didn’t get treatment. It will have a positive effect on the group that doesn’t require treatment (because the sickest of them now receive treatment, leaving only healthier patients). And it causes lower survival of the sickest patients who don’t receive treatment (because the healthiest of them now receive treatment, leaving only the patients with the worst prognoses in the pool).

This is a common problem when evaluating new medical technology. Introducing a newer, safer technology increases the number of interventions and apparently reduces good outcomes due to changes in the treated patient population. Further, new treatments are often more expensive than existing ones. This drives total healthcare costs upward as both the cost per patient increases and the total number of patients receiving treatments increases. Healthcare gets more expensive and without complex multivariate analysis it looks like nothing improved.