Psychometric properties of a brief measure of autonomy support in breast cancer patients

The primary finding from these analyses is that perceptions of autonomy supportive communication in breast cancer patients are consistently comprised of six dimensions for surgeons, medical oncologists, and radiation oncologists, as well as for the aggregate treatment experience. For patient perceptions of provider communication assessed using the mHCCQ, EFA for each clinician type yielded a single factor solution. Each of the 6 items loaded uniformly onto the single factor and demonstrated high internal consistency. These results increase our confidence that the six items on the mHCCQ are measuring a single underlying dimension for how breast cancer patients perceive the autonomy supportive communication style of their physicians.

A second finding from these analyses is that ratings of the overall treatment experience account for only half of the variance, suggesting that patients perceive and report differences in communication across provider types. When the three treatment-specific scales were grouped together for hierarchical EFA (6 items from each specialty for a total of 18 items), it is interesting to note that the three-factor solution loaded nearly equally onto a second-order factor, and that each of the three factors held together according to each specialty, rather than by individual questions. This suggests that answers to each question are more highly correlated with other questions on the same physician, rather than for the same question across provider types. Analysis of inter-item correlation across the 3 provider types yielded similar results, with higher correlation with other questions about each provider type than to the identical question asked of other specialties (data not shown). While it is possible that this may partially be related to grouping of provider questions together in sequence on the survey, it appears that patients clearly tend to differentiate their ratings of autonomy support according to each provider, rather than lumping their experience into an overall rating. Therefore, while assessment of only the aggregate treatment experience provides a general sense of how patients perceive providers’ autonomy supportive communication practices, assessing specialist-specific ratings appears to yield a more granular picture of patient experience. In situations where questionnaire length is limited or respondent burden is a particular concern, measuring only the overall score may be appropriate.

The correlation of specialty-specific scores with overall scores was slightly stronger for surgeons than for medical oncologists and radiation oncologists. This may indicate that the surgeon contributes more to patients’ overall treatment appraisal than the medical oncologist or radiation oncologist. However, on hierarchical factor analysis, we found that the second order factor, interpreted as a correlate of the overall treatment experience, was driven nearly equally by ratings of the surgeon and medical oncologist, with a lesser contribution from the radiation oncologist. It is also possible that because surgeon items were placed before medical oncology and radiation oncology items on the survey, an ordering effect contributed to the observed correlation. We previously observed a small ordering effect for when the overall experience and provider-specific experiences were rated (Resnicow et al., under review). Studies using random ordering of all items may be useful to determine the full impact of ordering on subject responses.

The overall means of the four scales were high at 6.0 or higher (out of 7). This positively skewed distribution is slightly higher than what has been reported in other settings using the mHCCQ [13], and is consistent with numerous prior studies of communication and decision-related ratings in cancer patients demonstrating high patient-reported satisfaction [21–24]. These high scores may in part be inflated by patient response bias related to the desire to feel good about their treatment experience in order to minimize dissonance [25]. With an average time from diagnosis of 12.6 months, recall bias and forgetting may have influenced responses. Both survey sites are renowned cancer centers, which could have biased scores upward as well. Given our study sample, we are unable to determine if surveying patients from less specialized treatment centers would have resulted in lower scores. As a measure of subjective perceptions of autonomy support, the values observed in our data may be considered inherently valid. Further external validity would be demonstrated if higher scores predicted increase in autonomous self-regulation (i.e. internalization), relative intrinsic aspirations, perceived competence, vitality, and compliance with lifestyle and medication recommendations. In addition, efforts to understand how positive reporting bias and social desirability may contribute to these scores remains an important area of future investigation. Strategies to obtain a wider spread of responses are recommended, such as adjusting the wording or response categories and range.

When we analyzed scale mean scores according to demographic criteria, we observed that black patients reported higher levels of perceived autonomy support in two of the four mean scale scores. This may suggest that either physicians communicate more effectively with black patients, or that black patients have different expectations regarding their physicians’ communication. There also appeared to be a trend toward an inverse relationship between the level of education and perceived autonomy support in three of the four scales, with more highly educated patients reporting lower mean scores. This suggests that patients with more education might be more assertive and expect more from their physicians. Overall, the study sample had relatively high levels of education, with 88 % having completed at least some college. How the scale may perform among a lower educated or lower literacy population should be examined.

We also observed that patients who underwent lumpectomy rated the autonomy supportiveness of their medical oncologist significantly higher than patients who underwent mastectomy, suggesting that medical oncologists might have a tendency to be more prescriptive in their conversations with patients who have undergone mastectomy. Studies have suggested that most patients who undergo mastectomy have the procedure due to either surgeon recommendation or an unsuccessful attempt at breast conserving surgery [26], both of which represent situations in which a patient would reasonably feel they had little decision-making autonomy. In the remaining patients who choose mastectomy due to preference, it is plausible that physicians may express disagreement with the patient’s choice, which would make those who chose mastectomy feel like they were in a less autonomy-supportive environment. Perceived autonomy support in patients who undergo mastectomy represents an area that merits additional examination in future work.

The mHCCQ assesses six dimensions of perceived autonomy support. It is possible that other domains not measured here may also play an important role in encouraging autonomous motivation. In previous studies, the mHCCQ has often been paired with other measures that assess autonomous self-regulation, such as the General Causality Orientation Scale, the Treatment Self-Regulation Questionnaire, and the Perceived Competence Scale [9, 11, 17, 27]. Inclusion of these scales may provide a more comprehensive and perhaps multi-dimensional measure of autonomy support and patient motivation for self-regulated behavior, though at the cost of a significantly longer survey. We chose to limit our evaluation to these six dimensions to keep our scales brief.