We aimed to explore the psychometric properties of the DSE in a cohort of patients undergoing VATS resection or SABR therapy for NSCLC. Our findings suggest the DSE is valid: the 11-item measure has good internal consistency (α of 0.96), and is one scale explaining 81% of the variance. The developers’ recommendation for one scale is confirmed by the observed high internal consistency in our study, however, we would recommend further exploration of the DSE in other cancer populations.
The observed significant ceiling effect (over 15% of responses) should be noted. No data about ceiling or floor effect have been published for the DSE. However, a recent systematic review of existing measures of self-efficacy in cancer patients, reported that these psychometric properties were not often assessed(32). The timing of assessment in our population (after the decision) may have affected the results.
Almost 70% of the sample completed the decision self-efficacy scale demonstrating that the collection of this data is possible in this population. Furthermore, the overall proportion of missing data was low (1.5%), indicating that DSE was acceptable to patients.
There was no difference in efficacy with decision making by treatment type in this study. Patients with poor clinical performance status were more likely to be less confident in making their decision for treatment. We know patients who have a worse PS and limited functional capacity tend to have more difficulty tolerating rigorous NSCLC cancer treatments, i.e. they have less favourable outcomes than fitter patients regardless of treatment type (33, 34). One explanation can be that regardless of the treatment type when patients are less independent physically (as in those with higher PS score), they have more conflict or difficulty deciding about the best treatment to meet their needs. In addition, performance status and NSCLC cancer stage were significantly more influential than biological age when recommending treatments(35). In this sense, physicians may tend to involve patients with a higher PS score less in the decision-making process, presumably with concerns about higher expected morbidity and mortality. In those cases, patients may perceive similarly less confidence in making that decision which is more “physician-driven”. Another possible explanation can be related to the fact that patients more compromised were never involved in discussions about possible treatment alternative i.e. surgery: this may have influenced their efficacy in making the treatment decision as their opinion may have not played a role at all in all the course of the decision-making process.
The decision efficacy scale has previously been evaluated in patient populations referring to patient’s making decision regarding immunizations, screening, hormone replacement therapy, blood pression medications adherence(36-38) suggesting a good understanding and applicability of this questionnaire in field where difficult decisions need to be taken(17). For NSCLC patients it would be useful to investigate if they, not only have the self-efficacy but the ability to ask questions and clearly express their values and prediction of outcomes.
O’Connor developed within the same conceptual framework of the DSE a 16-item Decisional Conflict Scale rated on a Likert scale measuring the uncertainty in choosing options, modifiable factors contributing to uncertainty (information, values and social support)(17).
In situations where the evidence available is not clearly defined and the long-term benefits are still undetermined (as with SABR Vs VATS), the understanding of conflict in difficult decisions may be more relevant to help identifying patient’s needs and possibly develop tailored decision aids.
The implementation of a decision aid in the field of early stage NSCLC has the potential also to streamline the pre-treatment pathway and reducing the referral to a second speciality opinion in these patients care. The decision conflict scale could be clinically more applicable to the conflictual choice between surgery and SABR especially for those borderline patients, where there is a clear equipoise in terms of risks and benefit, an observation highlighted in the SABRTooth trial(39). In high risk patients where the surgery has not been completely excluded by objective parameters, the decision should be tailored and supported by the medical team but also with the use of validated decision aids, as successfully demonstrated in other specialities(40). It would also be important to investigate and measure the involvement in decision and the ability to access and understand information(41). However, we must be aware that a good decision often doesn’t correspond to a good outcome: indicators of good decision making may include reduced uncertainty, improved knowledge, more realistic expectations, improved clarity of values and value congruence with the decision; reduced decision delay; improved adherence to the decision, and efficacy (17, 42). Understanding the latter, can have crucial clinical implications in the development of a decision aid and ultimately help people considering their options and making the best decision for themselves.
Limitations
The return rate of a 67% is reasonable for self-report questionnaires, but there may be a possibility those who did not complete the questionnaire had different experiences which could impact on the findings. Further investigation of its psychometric properties in samples which include a wider group of patients is advised, and methods to enable further validation data.
The study had a relatively small sample size, and was performed in a single centre. Certainly, it seems that the processes to choose between treatment types in NSCLC cancer are similar, suggesting the same type of information about the risks and benefits and long-term consequences of both treatments should be presented equally by clinicians to support informed choice. Alternatively, it may be that this questionnaire is not as sensitive to the differences between the different types of choices as other measures, e.g. decisional conflict scale. Certainly, patients found it difficult to make a treatment choice, regardless of treatment type, suggesting a decision aid might be helpful for patients to reach a decision with greater confidence. However, the questionnaire has not up till now been tested in a cancer population. Previously the DSE had been utilised with menopausal woman and psychiatric patients (16, 19), thus limiting the comparability of our findings.
Another limiting factor of our study design is that we collected the questionnaires after patients made their final decision; it may be that there are more patients who do not feel involved in their decision making earlier in the treatment pathways. This data collection method may, in part, explain the high ceiling effects as people’s views and exposure to further information will change from diagnosis to after treatment.