Access to clinical trials among oncology patients: results of a cross sectional survey

The current study was conducted as part of a larger cross-sectional study examining psychological outcomes among medical oncology patients. This study was conducted in three cancer treatment centres in Australia. Data were collected between November 2012 and August 2014. All treatment centres were located in public teaching hospitals with links to Universities and had a clinical trials unit. Treatment centres A and C were located in capital cities, while treatment centre B was located in a major regional area. Treatment centres A and B had 200–500 beds; while treatment centre C had more than 500 beds. Ethics approvals were obtained from the Human Research Ethics Committees of the University of Newcastle, Cancer Institute of New South Wales, as well as institutional ethics committees.

Patients who were aged 18 or older, diagnosed with cancer, English speaking and presenting for a medical oncology outpatient appointment were eligible to participate. Those attending the medical oncology clinic for the first time, and those unable to provide informed consent due to cognitive impairment or mental illness were excluded.

A research assistant approached eligible patients in the clinic to seek written informed consent. Consenting patients were asked to complete a paper and pencil survey either in clinic or at home. Those who elected to take the survey home were asked to return it within a week in the reply paid envelope provided. This survey included questions about sociodemographic, disease and treatment variables, as well as questions about psychological wellbeing. Approximately 1 month later, a second survey was mailed to participants. The questions on clinical trial participation reported here were included in the second survey. Up to two reminder letters were sent to non-responders after 3 and 6 weeks.

Frequencies and percentages with 95% confidence intervals were calculated for all variables of interest. Characteristics of the sample (gender, cancer type and age) were compared to national data using a one sample chi-square test. Fisher’s exact test was used to explore factors associated with being asked to participate in a trial. Those with a p-value <0.1 were included in a multivariate exact logistic regression. Odds ratios (95% confidence intervals) and exact p-values were calculated. All statistical analyses were programmed using SAS v9.4 [19].

One hundred and forty six (38%; 95% CI = 33.23–43.0) respondents reported that they had been invited to take part in a clinical trial. The number of respondents who had been invited by type of trial is shown in Table 2. Eighteen (13%; 95% CI = 7.2–18.3) respondents reported being invited to take part in more than one type of trial, with the most common combination being radiation therapy and chemotherapy trials (n = 7).

Of those who were invited to take part in a trial, 129 (93%; 95% CI =88.4–97.1) reported that they had consented. That is, overall, 33% (95% CI =28.0–37.3) of the sample reported having participated in clinical trial.

For the 17 respondents (11%; 95% CI =6.2–16.6) who were invited but declined participation, ten provided reasons for non-participation: ‘other’ reasons (n = 5) which included accessibility, feeling too unwell to participate, feeling that too much else was going on, and already being involved in a trial; wanting to choose own treatment (n = 2); concern regarding risk factors and side effects from participation (n = 2), and not liking the idea of clinical trials (n = 1).

Of those who reported that they had agreed to participate, 112 respondents (89%; 95% CI = 83.3–94.4) said that if asked to make the decision again that they would still choose to participate.

The univariate analysis showed that treatment centre, private health insurance, time since diagnosis, remission, chemotherapy and biological treatment were associated with being asked to take part in a clinical trial (p < 0.1). None of the factors examined were associated with the decision to take part in a clinical trial (data not shown). After adjusting for multiple variables, there was a significant association between treatment centre, time since diagnosis and being asked to take part in a clinical trial. Participants from treatment centre B had increased odds of being asked to participate in a trial compared to those as centre A. The odds of being asked to participate in a trial decreased with increasing time since diagnosis. All other associations were found to be non-significant after adjusting for confounders (Table 3).

Overall 33% of the sample reported having participated in a clinical trial. This rate is very high compared to other studies which have reported rates of participation in the range of 3 to 18% [4, 10, 21]. This may reflect that our sample over-represented women with breast cancer and people with colorectal cancer. A recent review found representation of breast cancer in clinical trials is proportionate to its incidence, whereas lung cancer has the highest incidence and highest mortality (27.6%), but accounts for only 9.2% of trials [22].

The overwhelming majority of those asked to participate in a trial reported that they had agreed to do so (93%). This rate is higher than other studies that have reported rates of acceptance into clinical cancer trials of 51–72% [21, 23]. Further, most reported that they would make the same decision to participate in a clinical trial again (89%). Together these results suggest that the biggest barrier to participation in a trial is not being invited to do so. Due to the study methods, however, it was not possible to determine the underlying reasons for not being invited to take part in a clinical trial. For example, we do not know whether trials were available at the treatment centres respondents attended, nor the eligibility of respondents for those trials. However, previous studies show that even after patients are deemed eligible by physicians and appropriate clinical trials for the cancer type and stage of disease are available, 40–50% of cancer patients will still not be offered participation in a trial [21].

Reasons reported for declining participation in a trial were broadly consistent with past research. For example, similar to our study, other research has indicated that being overwhelmed, unwell, and preferring to choose one’s own treatment rather than be randomised to a treatment condition [15, 21, 23‐25] as well as fear of side effects [24] act as key barriers to participation.

Those diagnosed over 12 months ago were significantly less likely to be invited to take part in a trial, with the odds of being invited to participate decreasing with increasing time since diagnosis. This may reflect that more clinical trials are likely to be available for those who are newly diagnosed. Therefore, as time since diagnosis increases, recall of invitations to participate in trials during the early post-diagnosis period may decrease. The higher proportion of recently diagnosed people who reported being asked to take part in trials may also reflect the impact of national initiatives to increase trial participation; and increasing recognition of the critical role that clinical trials play in evidence-based health care.

Treatment centre was the only other factor found to be associated with being invited to participate in a trial in the multivariable analysis. Those attending centre B had 2.79 greater odds of having been invited to participate than those at treatment centre A. These results are surprising given that treatment centre B was located in a regional area rather than a large metropolitan area, where greater research infrastructure and hence trial activity might be expected. While only a small number of treatment centres participated in the current study, results are consistent with previous reports suggesting considerable variation in trial activity between centres [5]. Previous research has reported that the number of oncologists and the presence of an approved multidisciplinary cancer program were significantly associated with higher accrual to clinical trials [26], however, we did not measure these characteristics.

In contrast to past research [21, 26‐31], we found no evidence that age, socioeconomic factors, or country of birth were associated with being invited to participate in a trial. It is possible that the lack of an association between country of birth and trial invitation was due to the exclusion of non-English speakers from the sample. There was also no association between cancer type and trial invitation. This is at odds with other Australian research which has identified a disproportionate number of trials on breast cancer [32] and the relative rarity of haematological cancer trials [4].

The treatment centre was a major determinant of being asked to take part in a clinical trial. High acceptability of trials was indicated by high reported consent rates and reported willingness to be offered the opportunity to participate in more than one trial. Together these findings suggest that one way to increase clinical trial participation rates may be to upskill and/ or resource hospitals with low rates of trial activity so that they can actively recruit for a range of clinical trials. This may have the advantage of ensuring equity of access to trials for a broader range of patients as well as enabling more rapid recruitment of the trial sample. However, if such measures are to be effectively implemented it is likely that some simplification to the complex multi-site ethics processes that researchers face may be necessary. The need to obtain multiple approvals and comply with a range of organisation-specific requirements may act as a significant barrier to expanding trial activity to a broader range of institutions.

The generalisability of our findings may be limited due to the small number of centres that participated in our study, and over-representation of women and those with breast cancer. It is also possible that non-consenters to clinical trials were under-represented in the sample. Further, translation of survey instruments into languages other than English and provision of interpreter services were cost prohibitive. Therefore, we had to exclude people who were not fluent in English from participating. Lack of representation of culturally and linguistically diverse populations within our sample may have resulted in the prevalence of being approached to participate in clinical trials being overestimated. Given these limitations, further investigation of our findings in a larger and more diverse sample of treatment centres and patients is warranted.

Data regarding trial participation was self-reported by respondents. While a clear definition was provided about what participation in a clinical trial constituted, it is possible that participants did not accurately recall these discussions with their healthcare providers.

Data on clinical and treatment variables were based on patient self-report due to the prohibitive cost of extracting this information from hospital medical records. One study comparing the accuracy of self-reported disease and treatment among 895 women with breast cancer to medical records data, showed that for general questions about whether the person had had surgery, chemotherapy, radiotherapy or hormone treatment), accuracy was very high (100%, 99%, 99 and 94% respectively) [33]. Agreement about whether a recurrence had occurred was also high at 90%; while agreement regarding stage at diagnosis was lower at 60%. This suggests that most self-reported data about basic cancer disease and treatment information is likely to be accurate.

Finally, although few cancer patients receive all of their cancer care at one treatment centre, we did not assess whether participants were invited to and participated in a trial at the treatment centre from which they were recruited. The inclusion of this information in future surveys may help to clarify the role that treatment centre plays in clinical trial access.