70 trials, few answers: the case for better research on interventions for depression among patients with cancer

on

Maria Pertl, Katie Verling, Garret Monahan and Frank Doyle

Depression is common among people living with and beyond cancer, and can hugely impact on quality of life, as well as health outcomes. Yet, despite decades of research, we still don’t know which treatments for depression work best.

Our recent review, published in the Journal of Psychosomatic Research, set out to answer this question by comparing the effectiveness of all available intervention approaches for depression in adults with cancer. What we found was both revealing and concerning.

A comprehensive review with a twist

We reviewed 70 randomised controlled trials (RCTs), that included over 6,800 participants and a wide range of intervention approaches – from psychotherapy and pharmacotherapy to exercise, complementary therapies, and collaborative care. Our goal was to use a network meta-analysis (NMA) to compare the effectiveness and acceptability of these treatments.

The quality problem

Most of the trials we reviewed had serious methodological issues. Of the 70 studies we identified, one single trial had a low risk of bias, while the remaining had high risk (54.3%) or some concerns (44.3%). When we evaluated the trials using a research integrity assessment, we found that 80% would be excluded if we applied this assessment as a screening tool. The key reason for failing the integrity assessment was the absence of pre-registration – that is, documenting a study’s objectives, methods, key outcomes and planned analyses before it is carried out in a clinical trials registry. This is important because pre-registration helps to deter the selective reporting of trials based on the results.

We also found other concerning issues related to poor reporting of randomisation and receiving research ethics or institutional review board approval, unrealistically large effect sizes, and a suspicious absence of any loss to follow-up. Indeed, in nearly a third of the studies, no participants at all discontinued with the trial, even though all experienced both cancer and depression, and the interventions were often long and time-intensive. We also found one clear instance of plagiarism, where one trial had noteworthy overlap with another article published a year previously by different authors. >Because of these issues, we took the perhaps controversial decision not to proceed with our planned analysis. We felt that the results – based almost exclusively on data vulnerable to bias and with very questionable integrity – were very unlikely to reflect true treatment effectiveness. Instead, we highlighted the methodological shortcomings and individually considered the evidence from the most trustworthy studies.

What does the (reliable) evidence say?

Among the few high-quality trials, collaborative care stood out. This involves a multicomponent approach with active collaboration between different specialists, and considers pharmacologic and psychological treatment options within an integrated management team that includes a patient care manager. Trials evaluating collaborative care showed consistent benefits for reducing depressive symptoms. Importantly, these trials were well-designed, registered before the results were known, and transparently reported – perhaps not surprising given the resources and funding required to deliver such interventions in the first place.

Why this matters

Depression can negatively affect cancer outcomes, reduce quality of life, and increase healthcare costs. Yet, the evidence base guiding treatment decisions in oncology settings is shaky at best. This has real-world implications for patients, clinicians and policymakers.

Until more robust evidence is available, treatment decisions will need to rely on clinical expertise and findings from the general population. However, the benefits of and adverse effects associated with interventions for depression may be different among patients with cancer. For example, drug-to-drug interactions between cancer treatments and antidepressants, and the potential ongoing psychological and physical consequences that can arise from a cancer diagnosis and cancer treatment, can add an additional layer of complexity.

We can – and must – do better.

What needs to change?

It has become almost standard practice to carry out systematic reviews, report that many of the trials are at risk of bias, and analyse their results anyway. Many – though not all – reviewers couch their findings in the caveat that their results might be biased by low quality trials. More diligent reviewers investigate and report the extent to which potential bias may have impacted on their results. But are these caveats ever truly considered? Why are we continuing to include and perpetuate findings from trials that are vulnerable to bias (at best, and potentially fraudulent at worst)? To safeguard the integrity of systematic reviews and build a reliable evidence base for clinical guidelines and practice, we need to start adhering to and strictly enforcing (long) available research standards and taking concrete steps to weed out problematic studies. This includes:

But the responsibility for this does not just rely on us, the researchers. Journal editors, publishers, research institutions, funders and ethics boards also need to ensure that these standards are actually upheld.

No doubt, some will argue that ‘gold standard’ psycho-oncology research is almost impossible to attain, especially since it is often carried out by well-meaning clinicians and researchers with little or no funding, or support from trial units. Yes, the standards are incredibly high – but they are so for good reason. The potential costs of not adhering to them and relying on potentially misleading findings to inform the care of patients, who have already been through enough, is simply unethical. There are therefore also important implications for funders: behavioural trials are expensive and researchers must be adequately resourced and funded to carry them out properly.

Final thoughts

Our review highlights a paradox: while there is no shortage of studies evaluating interventions for depression among patients with cancer, the quality of evidence is often too poor to guide practice. The lack of reliable evidence does not mean that these treatments are ineffective. Indeed, in our qualitative research with patients and healthcare professionals, they have consistently reiterated that each and every intervention approach has value and helps to alleviate depressive symptoms for some patients. But they are as dismayed and disappointed with the quality of the trials as we are – not least, because nearly 7,000 patients gave up their time to participate in them. Moreover, they agree that unreliable evidence should not be included in reviews because – as one healthcare professional put it, “You can’t put the rabbit back in the hat.”

Funding

The research discussed in this blogpost was funded by the Irish Cancer Society and the National Cancer Control Programme.

Authors

  • Dr Maria Pertl is a lecturer in the Department of Health Psychology, School of Population Health, RCSI. She is the Principal Investigator on the ENHANCE (Effective maNagement of depression among patients witH cANCEr) project.
  • Katie Verling is a member of the ENHANCE Public and Patient Involvement (PPI) panel
  • Garret Monahan is a member of the ENHANCE Public and Patient Involvement (PPI) panel
  • Professor Frank Doyle is an associate professor in the Department of Health Psychology, School of Population Health, RCSI.

Comments

Post a Comment