Predictive Factors Associated with Declining Psycho-Oncological Support in Patients with Cancer

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Thanks to the authors for the revision of the article.

The Table 1 "Study group characteristics" is a bit confused. A sample characteristic should have a different form and include important characteristics such as education and medical variables (e.g., dioagnosis). Please refer to other articles for information on which form is adequate.

What do the authors show in Figure 2 ?? What means 0.0 thru 0.8?? Please improve the presentation, in this form it is incomprehensible

Table 4 is uncleare. What numbers and sums are shown here?

Appendix A: Distribution of Missing Values: that is superfluous from my point of view.

Author Response

Dear Reviewer 1,

Thank you for your feedback and valuable input. As Reviewer 2 also found Table 1 unclear, we have revised the table based on the suggestions of reviewer 2. We hope you will now find the presentation clearer and more appropriate. Of course, we could include more variables from appendix A (e.g. F-diagnoses, sub-types of cancer) to better describe the sample, but it seems questionable, if this would add value to the manuscript? You have the last word on this, but we believe it is sufficient to mention that all types of cancer were considered. The C3Z, at which the study was conducted, is the largest treatment center for cancer in Switzerland treating all types of cancer. In addition, prior research has shown that the type of cancer has no significant influence on whether or not patients decline psycho-oncological support, experience distress or not, etc.   

Figure 2: In the revision, we have attempted to provide a more detailed explanation of Table 2. It displays the weights of variables relative to each other in the final model. In other words, figure 2 indicates the extent (0 to 1.0) to which each of the variables in the final model influence patients’ decisions to accept or decline psycho-oncological support. We hope this is clearer now.

Table 4 displays the mean value/ the relative distribution of each variable of the final model for patients wishing psycho-oncological support versus those declining psycho-oncological support. We have revised the table once again. Thank you for your input.

Appendix: The authors considered it important to list the variables considered for statistical analysis. However, we have further simplified the table according to your suggestion and limited it to listing these variables without including the distribution of missing values. 

Reviewer 2 Report

Comments and Suggestions for Authors

First of all I want to thank the editors and the authors for the opportunity to review this interesting paper! I hope that I have received the final version of the manuscript as there were many markups. Despite this, I have only a few comments and appreciate the clarity of content and language.

Methods:

1.       I am confused as in line 125 the group that wished support is called “the negative group”, while in l. 143 it is called “the smaller ‘positive wish’ group”. Personally, I would find the second definition more logical.

2.       I have to admit that I am not experienced in machine learning but the description sounds understandable and seems to be a very interesting approach. I just wondered whether the imputation/ upsampling of the “positive wish” group might introduce some bias as this will make the group more coherent than the other group?

Results:

3.       Table 1 seems a bit messy at first sight. I would recommend to show mean age, age group, distress, physical score and wish for the three groups male, female, and total. Then I would show mean age, age group, distress, physical score and sex for the three groups wish, no wish, and total (so that the subgroups to be compared will be the consistently columns instead of rows).

4.       Is there any reason why only the physical aspect of the PL is shown in Table 1? Was it because it had the highest impact?

5.       Figure 2 is not readable in the draft.

6.       Table 4: I think the results in line 2, 3, and 4 are scores, 1 is mean age, and 5, 6, and 7 are %? Are these based on a cut/off (then it should be stated in the methods which cut-off it was)? Please use the correct column headings.

Discussion:

7.       Is there an explanation why the social aspects were not important in the model, as one could imagine that lack of social support might be relevant, too?

8.       L. 283: Maybe those patients with a diagnosis of a mental disorder are already in psychotherapy and don’t need additional support?

Author Response

Dear Reviewer 2,

Thank you for your feedback and input.

 

1. Yes, we understand that this can be somewhat confusing. Since we were looking for variables that lead a patient to decline psycho-oncological support, "No Wish" was considered the "positive group" for our model. However, we have made the necessary adjustment in that section and hope it is now clearer.
2. On the topic of imputation, some researchers have argued that failure to impute data may introduce more bias than doing so (Cummings P. Missing Data and Multiple Imputation. JAMA Pediatr.2013;167(7):656–661. doi:10.1001/jamapediatrics.2013.1329). As described in reference 25, previous studies have shown that the method we have employed is the most suitable for handling such incomplete datasets.
3. We have revised Table 1 according to your suggestion. We hope it is now more understandable.
4. Yes, it seemed beyond the scope of table 1 to include all variables from the PL. We did include the physical score of the PL, because it turned out to be most relevant in the results section.
5. It appears that the editors compared the old version of our paper with the new one and presented the changes as corrections. We hope it is now more readable. Figure 2 was clearly presented in our upload.
6. We have also revised Table 4. It should now be clearer. Regardless, your interpretation was correct. Cut-off values for the DT had no relevance. There are no cut-off values defined in prior research for the PL.
7. Yes, indeed, social support could also play a role in the decision. However, we did not have available data to incorporate this aspect. We attempted to capture the social background with variables such as relationship status, employment, religion, or language proficiency. However, as mentioned in our paper, it appears that these factors do not have an impact.
8. It would indeed be interesting to investigate the influence of existing psychotherapy. Unfortunately, we did not have data available on this aspect. However, many psycho-oncologists (psychiatrists and psychotherapists specializing in psycho-oncology), especially in Switzerland and Germany, would argue that “conventional” psychotherapy cannot substitute for psycho-oncological support, which includes more cancer specific psychotherapy, counselling on social and practical aspects of relevance to patients with cancer and also includes spiritual aspects. Other research provided evidence that the assumption that a patient receiving “conventional” psychotherapy does not need psycho-oncological support is dangerous because it leads to not even asking such patients whether or not they feel distressed (see Günther, M. P., Schulze, J. B., Jellestad, L., Mehnert‐Theuerkauf, A., von Känel, R., & Euler, S. (2021). Mental disorders, length of hospitalization, and psychopharmacy–New approaches to identify barriers to psychological support for patients with cancer. (Psycho‐Oncology, 30(10), 1773-1781.). If you feel this would be appropriate, we could include this argument in the discussion section. What psycho-oncological support includes at the C3Z, where the presented study was conducted is described under “Measures” in the methods section.