This is the first Australian study to focus on factors predicting somatic mutation test ordering amongst cancer physicians. Oncologists were more likely to respond than non-oncologists and cash incentive improved overall response rate. The response rate (RR) in this study was consistent with United States physician surveys where questionnaires without incentives have a response rate of < 30% as compared to ≥ 50% when incentives are included.11 Whilst cash incentives have been shown to increase the RR among North American clinicians,12 and Australian pharmacists,13 to our knowledge, this is the first study to report similar findings in Australian cancer specialists.
Physicians order more somatic mutation tests than either cancer germline or pharmacogenomic tests with two thirds ordering at ≥ 5 somatic tests in the past year. This is similar to a large study of United States colorectal and thoracic oncologist survey where 31–68% ordered ≥ 5 somatic mutation tests annually.14 That study evaluated test ordering for specific somatic variants and found that ordering rates varied significantly depending on the variant, and whether there were endorsed guidelines. Overall, fewer than 3% of Queensland cancer patients were offered somatic mutation testing, which is lower than reported in one sub-specialty study. Specifically, 13% (20/153) of lung or colorectal cancer patients from a small group of United States thoracic and gastrointestinal oncologists (n = 27).15 In contrast, an international study of sequencing behaviours in breast cancer specialists found that only 38% ordered any somatic mutation testing,9 which is lower than the 83.3% of Queensland specialists. Furthermore, within the cohort of specialists who did order tests, 68% reporting doing so on ≤ 5% of their patients, which more closely reflects our findings. Physicians’ qualitative comments, in our study, suggest that the low uptake may be, in part, attributable to uncertainty regarding the actionability of the results, which is consistent with qualitative data reported previously.15 However, one question in the present study specifically assessed physicians’ perceptions on the percentage of patients who would have access to molecularly indicated agents, and the mean was 40%. This is considerably higher than the somatic mutation test ordering rate would imply.
The median germline testing was similar to previous reports with thoracic and gastrointestinal oncologists also ordering a median of two tests per annum.15 In fact, physicians in this study ordered a mean of seven germline tests in the preceding twelve months, which is higher than the average of three ordered by United States gastrointestinal oncologists.14 Similarly, in this study, a quarter of cancer physicians reported ordering pharmacogenomic tests which is consistent with a study which found that 35% of lung and colorectal oncologists ordered at least one pharmacogenomic test in the previous year.15
In order to more fully understand attitudes towards genomic testing, clinicians were asked to rate the relative utility of tumour pathology compared to TMP in informing treatment decisions. They were not equally valued, as initially predicted; rather, tumour pathology was weighted more heavily. Qualitative comments stated that the relative value of TMP is dependent on the tumour type, as has been clinically proven.2 To our knowledge, this is first study to evaluate the perceived, relative utility of each. It would be meaningful to readminister the survey in five years’ time to capture any change over time.
Cancer physicians were, understandably, more inclined to disclose Tier 1 or Tier 2 results than Tier 3. Consistent with our results, previous research has shown general support for the disclosure of Tier 1 and 2 results.8,15 Furthermore, in parallel to our findings, United States physicians were more likely to report a Tier 2 result which conferred eligibility to a Phase II clinical trial than to disclose a Tier 2 variant associated with off-label drug use.15 However, less than a quarter of respondents in this study supported disclosing Tier 3 results (disclosing as many sequencing results as the patient wanted, including raw sequencing data) as compared to 50–96% in previous studies.8,15 Qualitative comments from one study clarified that willingness to consider disclosure of Tier 3 variants stemmed from the desire to optimise treatment choices.15
Almost half of physicians reported low to negligible confidence in interpreting, explaining and making treatment recommendations relating to somatic TMP results (‘genomic confidence’). Although oncologists have greater confidence in interpreting variants than other cancer specialists, a substantial portion still report low levels of genomic confidence, consistent with previous medical oncology studies.8,9,15,16 Overall, the portion reporting low confidence in this study is considerably higher than the ~ 20% of cancer physicians internationally.8,9 This is significant as research has shown that low confidence negatively affects clinical practice and ordering behaviours.17 Of relevance, a systematic review article exploring the integration of genetics into healthcare found a lack of access to genetics services was a significant barrier.18 In this study, the majority reported confidence in their ability to identify genetics consultants, suggesting that this is less of a barrier in Queensland.
In regression modelling, the only two predictors of somatic mutation test ordering were being an oncologist, and having greater confidence in interpreting somatic variants. Previous research has shown that oncologists are more likely to order genetic testing than other cancer specialists,14,19 possibly offering insight into their increased likelihood to participate in the study in the first place. Genomic confidence generally, has been associated with a greater intention to request genetic tests8 and a higher uptake in practice.19,20 However, in this study, it is unclear whether confidence predicts ordering behaviour or whether those who order more tests become more confident in interpreting results.
Under the COM-B theory,5 test ordering behaviour should be predicted by a combination of Capability, Opportunity and Motivation, and the results of this study are consistent with this theory. Specifically, specialty (capability), and greater confidence in interpreting results (motivation) are associated with ordering behaviour. Importantly, it has been repeatedly shown that capability and motivation can be moderated through education as evidenced by increases in confidence and perceived competence.21–23 Furthermore, educational interventions have been shown to increase genomic confidence amongst cancer specialists.24 Therefore, these findings imply that educational programs which increase capability and motivation could modify physicians’ somatic ordering behaviours.
The limitations of this study include a 52% response rate, capturing just 58% of oncologists, so results are not representative of all Queensland cancer specialists. In addition, this is a heterogeneous sample and the practices of the oncologists will not usually be consistent with the practices of non-oncologists. Of note, this survey assessed physicians’ perceptions of ordering behaviour rather than their actual somatic mutation testing rate. It is possible that the actual ordering behaviour was higher. A printing error for the question assessing actionability means that only 67 responses to this question could be analysed.