In the last 15 years, the quality movement has addressed the variation between medical practices at all levels: between physicians within the same department, between departments, between hospitals and between countries. Variation is something like the smoke in the heurism: “Where there is smoke, there is fire”: if medical decision making is rational, how can there be variation in practice? However, the issue of medical practice variation has long drawn attention from researchers from outside the quality movement. Researchers assuming social conditions, financial incentives or training are interested in variation. Of course, policy makers are stakeholders in this variation debate. Judith de Jong (2007) studies variation following Freidson (1975) and Westert and Groenewegen (1999), assuming that especially social conditions influence physicians’ medical behavior. The prediction is that physicians' ways of working are more similar if they share their work environment (a practice or hospital). Social circumstances do more to shape their behavior than their individual preferences. These ‘shaping’ processes are considered to be institutional. In describing these, she uses Scott’s (2001) distinction among three institutional mechanisms: the regulative, normative and cultural-cognitive mechanisms. De Jong's idea is that once the influence of a certain mechanism has been determined, it should be possible to influence variation. To test her hypotheses, De Jong uses secondary data from three different databases: two from national surveys held among general practitioners in The Netherlands and one from hospitals in the United States.

Working in the same circumstances implies sharing resources. In combination with visibility of behavior, this will lead to less variation among physicians. This is what she concluded from her analysis of the variation among general practitioners. There could be several reasons for this, such as adaptation of physicians to an existing way of working or selection processes of new physicians. It seems that circumstances themselves are most dominant in explaining homogeneity of behavior. This not only downplays the role of adaptation and selection, but also the importance of individual differences. In Chap. 4, this aspect is further elaborated. Here, the hypothesis is tested that physicians working in two hospitals act differently in those hospitals. It appears that where two hospitals differ, e.g., in length of stay, physicians work differently, even if they work in both hospitals. Again, it was confirmed that circumstances are important in understanding medical practice variation.

In Chaps. 5 and 6, the behavior of physicians under restrictive (regulative) circumstances is examined both for physicians in the managed care system in the USA as well as for general practitioners. Central here is the effect of peer expectation and professional control translated into guidelines. The effect of the introduction on variation is moderate. The author suggests that guidelines may have a more substantial effect if combined with cultural-cognitive tools, such as computerized decision aids. In Chap. 7, the influence on variation of cultural-cognitive processes by using a decision support system when prescribing was examined for general practitioners in The Netherlands. It appears that the use of a decision support system reduces variation only when it does not allow any variation, for example, when only one piece of advice is given for prescribing.

The implications for health policy are not so clear. The author has some difficulty being precise about this. I think the main reason is that she carefully avoids the distinction between desired variation and avoidable variation. There are good reasons for this if there are mechanisms that explain variation. From a quality or economic point of view, variation is only a signal of something that maybe(!) is not desirable. From a policy point of view, the issue is: how do we avoid the wrong kind of variation? De Jong discusses several strategies to reduce variation: payment-for-performance and uniform performance indicators. According to the author, this will lead to more uniformity among practices and increasing dependencies within practices. This will reduces variation. However, the results of her study show that using institutional mechanisms will influence medical decision making, but not reduce variation. The author, however, expects (hopes?) that in the long run, cognitive-culture tools will press physicians towards homogeneity. An example is that she expects that performance measurement and reimbursement systems will lead to the separation of standardized medical care in high volume clinics. It is not clear, however, what such a process would look like, and especially when this would be desirable to prevent the wrong kind of variation. The problem is that it is not clear how institutional processes work when variation should be avoided and when it should not. Probably often this is not clear. For these reasons, it is difficult to predict how institutional mechanisms will promote effective and efficient care. Variation is often an important signal of non-effective and non-efficient care, but not always.

To conclude, De Jong's work is highly original in explaining variation from an institutional perspective. It provides new and important insights into the mechanisms at work in situations where physicians work and how these pattern their actions.