Improving management of comorbidity in patients with colorectal cancer using comprehensive medical assessment: a pilot study

Comorbidity is the occurrence of other diseases alongside the disease of interest and is common among patients with cancer [1‐4]. At least one comorbid condition is recorded in approximately half of all patients with cancer, and two or more conditions are present in a third of patients [5, 6]. This is partly because both cancer and comorbid conditions increase in incidence with ageing, and because cancer and some comorbid conditions share common risk factors [7]. The presence of comorbidity has significant implications for cancer treatment decisions and outcomes [1, 3, 8‐12]. Comorbidity substantially reduces the likelihood that an individual will receive cancer treatment [3, 8, 13‐16], and adversely impacts cancer survival [1, 8, 10, 17, 18]. However, this adverse impact can be moderated through receipt of definitive treatment [10, 19‐22], Despite the importance of comorbidity, comorbid cancer patients are often excluded from clinical trials [7, 8] and most treatment guidelines focus on the management of one disease only [9]. Thus, there is little evidence relating to cancer treatment that includes comorbid patients.

Comprehensive geriatric assessments (CGA) are increasingly used to assess patient functional and nutritional status, psychological and social wellbeing, presence of geriatric syndromes and polypharmacy, and to better manage these factors in the geriatric cancer patient [23‐28]. While few studies have assessed the extent to which CGA or similar broad-based medical review at the time of cancer diagnosis actually alter cancer outcomes, those that have, without exception, found that such interventions have a positive impact [29‐32]. However, comorbidity can occur independent of ageing and can impact on treatment or outcome independently [7]. To date, no studies have determined the value of such an intervention for cancer patients of all ages – a comprehensive medical assessment (CMA) - nor have any studies reported qualitative data pertaining to such an approach.

New Zealand (NZ) has a publicly funded national health system with six cancer treatment centres delivering the majority of cancer treatment to its five million population. Prior to embarking on a large multi-centre intervention study we wished to address several questions regarding feasibility. Therefore, this study assessed the feasibility of the comorbidity screening tools and of collecting key data including proposed outcome measures, the acceptability and potential effect of the CMA, and the feasibility of a larger study.

The aim of this pilot study was to assess the feasibility and acceptability of comprehensive medical assessment in comorbid patients of all ages with colorectal cancer, and the feasibility of conducting a large multi-centre study in NZ. Although the results of this study can only be considered indicative, the higher rate of chemotherapy completion in the intervention group is promising and feedback from patients and staff was largely positive, indicating acceptability of the approach. The information generated by this study has provided a solid foundation of learnings on which to further develop the coordinated approach to management of comorbid cancer patients proposed.

Screening for and active management of comorbidity soon after cancer diagnosis shows promise in positively altering cancer outcomes, such as improved surgical outcomes [26, 30], reduced risk of chemotherapy toxicity, and increased likelihood of completing chemotherapy as planned [23‐26, 29, 39]. Similarly, we found that a higher proportion of patients undergoing our intervention completed chemotherapy without modification than those who did not (6/11 and 0/14 respectively). Although, as a pilot study, our patient numbers were small, this finding indicates some potential in this approach. Interpretation of the results is complicated by the fact that the intervention group was healthier than the observation group in most variables measured at baseline. Due to informal feedback, we suspect that research nurses were less inclined to approach sicker patients during the intervention phase, because they felt those patients may be already over-burdened. A randomised controlled trial (RCT) would reduce the variability observed between the patient groups. However, the effectiveness of service interventions are difficult to evaluate using an individual-level RCT framework due to potential contamination between control and intervention groups [40]. A cluster RCT would avoid these issues, but there may be scientific and practical problems with implementing such a study in NZ - primarily that there may be too few cancer centres in NZ to provide sufficient numbers of clusters. Other approaches such as a step-wedge trial may be useful but have limitations including additional data burden and complexity of analysis [40]. A non-randomised evaluation using a before and after approach, with mechanisms in place to ensure balanced invitation processes, may still be the best way to continue this work within the NZ context.

In order to progress to a larger-scale study, several changes are required to the study tools and processes. First, a screening tool with greater specificity is required. Although we found high levels of comorbidity in the study population (consistent with published literature, [7, 39]), it does appear that the threshold which triggered the referral to CMA was set too low within our study; both geriatricians indicated that the patients referred to them had few comorbidities that were amenable to intervention. Ideally, a screening tool should have not only high sensitivity but also high specificity to identify those comorbid conditions which are amenable to intervention, and so minimise patients being referred unnecessarily [41, 42]. To date, no one gold standard tool has been developed for the identification of important comorbidity [23, 24, 41, 42], although the G-8 tool has promise for correctly identifying older cancer patients who might benefit from a CGA [43]. The G-8 tool includes questions on nutritional status/food intake, weight loss, body mass index, motor skills, psychological status, number of medications, and patient-perception of health, along with age > 80 years as criteria [43]. The G-8 tool is designed to be completed by trained nurses, whereas our study utilised a mix of patient self-report and clinical note review by nurses to determine eligibility for CMA. However, a self-administered screening tool consisting of 14 measures has also been shown to be feasible to administer in the hospital setting [42], and helpful in selecting patients who will most benefit from a more formal comorbidity management process [42, 44]. The design of an effective screening tool that correctly identifies which cancer patients will most benefit from additional management of their comorbid conditions prior to cancer treatment is vital to any future approach in this area.

Secondly, the CMA itself requires refinement. While the findings of our qualitative work indicate acceptability of the CMA intervention, the geriatricians each questioned their mandate to change patient management. Geriatricians typically receive referrals with a presenting problem clearly identified, whereas, in this study, referral occurred if patients met certain study criteria. A number of management recommendations were made to the patients’ primary care team within clinic letters rather than enacted within the CMA itself. It is possible that the CMA not be delivered by geriatricians. A revised CMA using a different health professional - such as cancer specialist nursing and/or the wider oncology or surgical teams [42] - to act on the results of a comorbidity screen is plausible. If the geriatrician-led CMA model continues with a more select group of patients, consideration should be given to providing patients with more information about the intervention at point-of-referral and, conversely, providing more information about the patients to the geriatricians. Further engagement with other health professionals, including primary care practitioners [45], will also be integral to future refinement of the study.

Thirdly, findings from this study suggest that the CMA should be introduced early in the cancer treatment pathway. The recruitment focus in this study was changed from soon after diagnosis to post-operatively on feedback from research nurses indicating that they felt that patients would be unable or unwilling to participate in research at this early point in their cancer journey. While it is plausible that those patients with the greatest and most complex health needs - who might benefit most from a comprehensive comorbidity-management intervention - may be least likely to participate in research, information from the qualitative phase of this feasibility study suggests that earlier patient recruitment and thus earlier CMA for those eligible would have been acceptable. All 13 patients interviewed – including four patients enrolled pre-operatively - indicated that that research requests were reasonable, and the geriatricians stated that they believed a pre-operative CMA would have been more useful. As emphasised by a cancer nurse coordinator ‘patients will be overwhelmed soon after diagnosis but this should not stop us enacting things that will improve their care’. Previous studies have identified patients for CMA once they had already been selected for a treatment such as surgery or chemotherapy, but before treatment was commenced [29]. Pre-operative CMA has shown to be useful to identify and better manage cancer patients likely to experience post-operative complications, including delirium [42]. It is likely pre-operative intervention would have been equally useful in this study and should be considered in the future.

Finally, whatever screening tool and comorbidity management model are utilised in the future, it is important that they be well accepted and effectively used in practice; participatory research approaches with key hospital staff may best enable this [46‐48]. Participatory research is compatible with current implementation science [49, 50], with increasing attention being given to researchers, end-users and decision-makers working together to design and subsequently evaluate health-care interventions [46, 51‐53]. In addition to a participatory research approach, a combination of strategies such as those we enacted to gain buy-in to this health service intervention research will be critical to any future study that intends to transform clinical practice or modify a complex cancer treatment pathway. Process evaluation to better understand how and why the study succeeds along with economic evaluation of any future study should also occur (Table 5).

The key strengths of this study are that - to our knowledge - no studies to date have investigated a comorbidity optimisation intervention for cancer patients of all ages. Additionally, this study utilised a mixed methodological approach involving key study staff and patients to assess on-going feasibility of the approach. This study also had several important limitations. First, small study numbers prevent reporting anything other than descriptive statistics. However, as this was a pilot study no formal hypothesis testing was planned. Rather, this study was designed to provide critical information to inform a larger study. Secondly, this study investigated colorectal cancer only and the findings are not directly generalisable to other cancers. Thirdly, the study is specific to the NZ health system. However, the key outcome finding that patients undergoing comprehensive management of their comorbid conditions appeared more likely to complete chemotherapy treatment as planned supports findings in previous comparable studies investigating a range of cancers in the United Kingdom [29] and United States of America [30].

This pilot study showed promise in its key outcome measure and the modified model of care – geriatrician-led CMA – was acceptable to comorbid cancer patients. Importantly, the study has provided useful insight into how the study can be implemented in the future.