This study examined the proportion of Danish medical PhD-graduates continuing research production measured through published papers. Data from all Danish universities were used and had a median follow-up of 7 years. Four main findings were found. First, a larger proportion of the PhD graduates were women, but characteristics of the individuals were similar across the three universities. Second, the majority of PhDs (two thirds) were inactive 2 years after graduation. Third, women were more likely to stop research than men. Fourth, last-author proportion increased after 4 to 5 years suggesting that being a research-supervisor/in chair of research projects took long time to develop.
Elsewhere, gender distribution among PhD graduates has been investigated. Two thirds of PhD-graduates at The University of Aarhus, Denmark were males in the period from 1993 to 1998 . In contrast, contemporary analysis of inclusion within PhD programs from all PhD schools in Denmark suggested women to be in slight favour among the laureates. Furthermore, our results demonstrate a change in the gender distribution among medical PhD-graduates as well as a change in the gender distribution among physicians. The problem was also assessed by Kuehnle et al. who found that men constituted 2/3 of the PhD-graduates and students in the Swiss National MD-PhD program from 1992 to 2007 .
Approximately two thirds of the PhD-graduates became research inactive within two years after graduation in this study. Importantly, our results show that the majority of this measured activity is comprised by co-authorships and not first- or last-authorships. In addition, we relate last-author papers with a person who has evolved his research and is now in the role of supervisor/mentor. Our results show that such persons are few in numbers and that transition to last-author happens roughly 5–10 years post PhD-graduation. Furthermore the last-author proportion increased after 4 to 5 years suggesting that being a research-supervisor/in chair of research projects took long time to develop. Yet, still only 10–15 % of the MDs had last/first author publications at 8–12 years post-PhD graduation. We see this as a relatively low number and it probably represents those MDs who went on to take a fulltime or a part time research position (professors). It is indeed encouraging that we see this increase in senior authorships and this represents the long-term yield of the PhD education program in Denmark—yet, we also believe that this is a very low number. One could ask if it is rational to educate this many PhDs if only 10–15 % to some extend go on to build their own research groups/communities. Maybe a shorter program could be sufficient and create more researcher friendly programs for those with a special interest. Overall, our results contrast the findings of Jørgensen et al., who concluded that Danish PhD graduates in the years 1995–1997 remained productive after obtaining the degree . In accordance with our results, Brass et al. investigated the career-path of several MD-PhDs over 40 years. They found that only one third of the questioned devoted more than 3/4 of their time to research .
The low degree of researchers having both a MD and PhD continuing to be research productive may also be explained by local factors. For instance, since February 2008 physicians from Denmark who have initiated their residency have a time limit of 5 years to achieve internship in either surgery or intern medicine. If the physician is unsuccessful, specialisation in Denmark is no longer possible according to the Danish health authority . Stronger competition to achieve a specialist position over a constrained time horizon is implied by the increment in flow of PhD-graduates in Denmark. PhD-students from 2003 to 2013 has doubled demonstrated through annual increments . Second, a substantial part of the PhDs will start clinical training following PhD-graduation reducing the probability of research continuation. In accordance; Whitcomb raised this as a central issue with the PhD-program and suggested restructuring of the PhD program . Although reported elsewhere  that PhDs wish research as an important part of their future job description, our data could not confirm such a trend. However, a comparison between non-PhDs and PhDs was undertaken by Merani et al. and showed that volume and impact of research activity were greater for the PhD-graduates .
Factors associated with research inactivity suggested gender differences, as women were more likely than men to stop researching. This issue has been assessed in several studies. Lelièvre et al. concluded that male gender predicts publishing among pharmacists . The gender gap has further been analysed by Kaufman et al. among physical therapy faculty members. They found similar results; male gender was a positive predictor for publication .
Our results suggest that future scientific yield from researchers with a medical background and the PhD is far from optimal. In order to improve productivity in the future, Whitcomb  suggested that the PhD program should be reformed through demands of higher scientific yields; also suggested by Olesen .
The external validity of our results is hard to say since differences between PhD schools worldwide are evident. However, the clinical work profile for most specialists are similar in western medicine, which could indicate that our results can be extrapolated to other countries if underlying reasons for not continuing research are lack of time or simply reduced competition as soon the PhD laureate achieve a consultant post. In Denmark, at least, the clinician does not have a strong incentive nor allocated time to continue doing research when he or she has secured a specialty training position (typically 5 years as fellow-in-training). Hence, research is conducted on their spare time or by taking leave from the fellowship (which very few trainees do). Nevertheless, the results here deserve to be replicated in other national based investigations, which may enhance the validity found here.
The current study had several limitations. The scientific yield of the PhD was evaluated retrospectively through manual searches in the PubMed database. Furthermore, this chosen search strategy was not validated. Another caveat was that searching in other databases such as Embase was not performed. Individual benefits from the PhD education, which were immeasurable, were not evaluated and may have intangible effects on these future medical doctors work as clinicians and their approach to medicine. However, a better understanding of clinical studies and statistical insight for the individual PhD graduate may conversely not contribute to better treatment for the overall population. Lack of detailed demographic data about the PhD-graduates was also a downside of our study. This means for example that we did not have access to personal information of PhD laureates being married after graduation. For instance, we had no information of the education level of the parents to the PhD graduates including a PhD or their social class. In addition we did not have access to personal covariates and career choice, which may have been explanatory for research activity. For instance, leave due to pregnancy and career choice of an academic or clinical pathway were all, unavailable. Accessing the latter may even have made a sharper statement of inactivity among clinicians with a PhD. We did not examine sub-groups of research according to type of research as these data were not readily available to us. An extended margin for research activity equating 3 or 4 years may also have increased the proportion of active researchers, but we still think that achieving funding and academic skills after 3 to 4 years of inactivity is difficult. During these years after PhD-graduates, the MD has many competing interests (especially clinical fellowship), but we believe that the system should be better at incentivising the graduates in continuing their research. Our results especially point to women as being more susceptible to this early decline in research productivity, which may be expected as child birth, clinical training, maternity leave and other competing interests are presenting themselves. We believe that funding mechanisms and research groups could be much better at holding the PhD graduates engaged during these straining years if the goal is to keep MDs research-active and hence help make patients and the future of medicine better. Eventually, we acknowledge that the MD PhD combination in Denmark may differ from countries elsewhere underscoring that the generalisability of the results remains elusive.