Main points
We present here a population-based approach to the question of whether or not colorectal MDT meetings improve survival in patients with colorectal cancer. These results reflect those achieved by a mature MDT, a group of clinicians who had been working together for over eight years. We have included comorbidity and socio-economic deprivation as well as other more widely reported demographic and clinico-pathological variables in our analysis. We found no evidence that patients with potentially curable tumours suffer harm as a result of failure in the MDT process. There is some apparent benefit from MDT discussion in patients with advanced or metastatic disease, but the evidence is insufficient to determine whether this is an artefact arising from selection bias or whether the advantage is genuine. The statistical treatment of deaths within six weeks of diagnosis had an important effect on the estimate of the magnitude of the effect associated with the MDT process. Censoring deaths occurring within six weeks of diagnosis attenuated the estimated benefit: mainly because only a small proportion (<10 %) of patients who died within 6 weeks had evidence of an implemented MDT recommendation. This is important for two reasons. Firstly, it is unlikely that the MDT process could, per se, have had any influence over the occurrence of these early deaths. Secondly, including these patients, whose fates were largely sealed by the time of diagnosis, introduces a selection bias which will exaggerate the perceived benefits associated with the MDT process.
Limitations
This is an observational study and, as such, is subject to bias. The group of patients (MDT-) who were not discussed at an MDT meeting, or where MDT recommendations were not implemented, was not derived through random allocation, and it is highly likely that membership of this group was influenced by confounding variables not considered in the analysis. Although nearly 600 patients are included in this study, there are only 175 patients in the MDT- group and so statistical comparisons may be relatively underpowered.
We adopted a loose definition of what constituted a “recommendation”. We did not stipulate that the MDT had to define a specific plan for treatment and accepted that, bearing in mind that patients themselves may not have been adequately represented at the MDT [13], it is reasonable to consider that further discussion with an oncologist could constitute an outcome.
By using cause-specific survival as the outcome of interest for the study, we excluded consideration of whether MDT discussion might have had an impact on the morbidity of treatment by, for example, recommending rectum-conserving surgery rather than abdomino-perineal excision. However, the prime purpose behind the introduction of MDT meetings was to improve survival and this is therefore the standard by which the process should be judged.
These patients were assessed and managed in the pre-biological era of treatment for colorectal cancer. All patients had access to standard chemotherapeutic agents and to conformally-planned radiotherapy. However, at the time of initial decision-making, biological agents (such as cetuximab and bevacizumab) had not been approved for use in Scotland. Pathological specimens were not routinely assessed for molecular markers – primarily because no targeted agents were available for treatment.
Strengths
This is a population-based study and the results and conclusions may therefore have more general relevance than those from studies based on data from a single hospital. We have complete follow-up, including details of vital status, for all patients. We have been able to document not just whether the patient was discussed at an MDT meeting, or whether a recommendation was made, but also whether or not any recommendation was implemented. The analysis covers a short time period, only two years, and all patients were cared for by the same team of oncologists and surgeons. There is therefore consistency of decision-making and clinical management.
General discussion
There is a dramatic difference in long-term survival when the experience of patients with early disease is compared with that of patients who present with advanced or metastatic disease: the 5 year survival for the 200 patients with advanced disease was 14.3 %, the corresponding figure for the 386 patients with early disease was 81.8 %. The magnitude of this difference may go some way to explaining the observed differences in colorectal survival when comparisons are made between institutions, or amongst nations [14]. Any underreporting or exclusion of patients with advanced or metastatic disease will significantly inflate the overall estimates of survival after a diagnosis of colorectal cancer.
We have presented results for all patients, and for that subset of patients who survived for at least six weeks after diagnosis. By excluding patients who died within six weeks of diagnosis we eliminate from consideration patients who presented as emergencies and who died within a few weeks of surgery, as well as those patients who presented in the terminal phase of their illness. It is unreasonable to expect that MDT discussion would improve outcomes for such patients, their fates are determined by events that are usually beyond the control of any MDT.
When survival analysis was restricted to patients who survived for at least six weeks after diagnosis any benefits associated with MDT discussion were less evident. This suggests that, in a population-based series such as this one, MDT discussion is to some extent an indicator of longer-term survival. Patients who die at and around the time of diagnosis are less likely to be discussed at an MDT meeting. This is consistent with the finding that MDT discussion and implementation of recommendations was less likely in patients with higher levels of comorbidity (Table 2).
Our results are consistent with the published literature [4–9] in patients with colorectal cancer: the MDT process is associated with improved survival. However, we clearly demonstrate that in patients with non-metastatic disease the MDT process contributes little to cancer-specific survival (Fig. 3 and Table 3). The apparent benefit of the MDT discussion is most marked in patients with advanced disease. This benefit is still apparent, although not statistically significant, when those patients who die within six weeks of diagnosis are excluded from analysis. This suggests that the main contribution of the MDT may be to co-ordinate the management of patients with complex clinical problems – potentially resectable liver metastases, tumours of borderline operability. Clinicians are not always aware of what their colleagues in other specialties might have to offer [15]. A recent Australian study [16] looking at the effect of MDT discussion upon management decisions drew similar conclusions. MDT discussions were of more value for patients with more complex clinical problems. Of course it is entirely possible that the observation of benefit in this group is due to hidden confounding – only those patients who, on a priori grounds, are expected to benefit are discussed, the others are not. The “beneficial effect” of discussion might simply be a self-fulfilling prophecy.
Two papers [17, 18] have looked at implementation rates for MDT recommendations. In the study from Plymouth [17] the implementation rate was 44/47 (93.6 %), the rate in the study from Bristol [18] was 137/157 (87.3 %). The rate in our study, confining analysis to patients surviving for at least six weeks, was 407/490 (83.1 %).
Our results suggest that much of the workload of MDT meetings, as they are currently constituted, may have little impact upon cancer-specific survival in patients with colorectal cancer. For the group of 386 patients with early disease, 66 % of all patients discussed, the presence or absence of adequate MDT process had no significant effect on survival. MDT discussion of all new patients is an instrument of cultural change and has helped to establish an environment in which equality of access to a uniform standard of care is now considered the norm. There may be more cost-effective ways to maintain this new culture. Coordination of care is important [19] but does not necessarily require the full majesty of an MDT meeting. Decision-support systems [20–22] could easily be used outwith a formal MDT meeting. For more complex problems, an asynchronous virtual MDT [23] might offer a more flexible and less labour-intensive approach than weekly face-to-face meetings.
Population-based screening may bring with it a new set of problems – primarily related to pathological interpretation of early disease [24]. In the future there may be an increase in the number of patients with early disease who present complex problems requiring MDT discussion [25]. This indicates the need for a flexible approach to the role, remit and constitution of the MDT. It will not be easy to modify the role of the traditional MDT. Assumptions concerning the value of this practice are now firmly embedded in the procedures for assessing the quality of cancer services: the Scottish Quality Performance Indicator (QPI) sets a target of 95 % of new patients with colorectal cancer being discussed at an MDT [26]; the colorectal MDTs in England are assessed against a set of 43 measures, all of which concern process rather than outcome [27].