Prevalence and Factors Associated with Opioid Prescription in Swiss Chronic Hemodialysis Patients

Round 1

Reviewer 1 Report

In my opinion, the authors have adequately addressed all issues.

Reviewer 2 Report

I have no further comments.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Major comments

1. This is a retrospective study in a low number of patients. My main concern is due to the small number of patients this study has not enough power to make a reliable comparison between the 2 centers and also lacks power to make reliable conclusion form the results of the multivariable analysis. Therefore, true differences / determinants may have been missed due to insufficient power.
2. Data collection: the authors state that “handwritten prescriptions of opioids were also recorded”. Please specify the policy with regard to handwritten prescriptions for opioids. How long were handwritten prescriptions stored? How do we know that this gives a complete picture? Was the policy with regards to handwritten prescriptions identical in the two centers?
3. In my opinion, it is not surprising that opioid prescriptions were higher in the academic hospital than in the private hospital. E.g. , patients in the academic center more frequently had cancer and amputations. In other words, there may be bias by indication.
4. Cancer and amputations are not mentioned among the causes of pain in Table 1. Is this correct? Or is pain after amputation categorized as neuropathic pain?
5. With only 2 centers studied (and considering the differences between these centers), we do not know whether these results are representative for the Swiss situation.
6. I assume that some patients with difficult to treat pain were prescribed a combination of drugs to relieve pain, e.g. opioids in combination with NSAIDs. This information on co-medication is missing in the manuscript.
7. The authors rightly state in the discussion that gabapentin and pregabaline are also used in neuropathic pain / polyneuropathy. However, I miss information on the use of these drugs in Table 1. The same question for drugs like amitriptyline.
8. Discussion, line 241-242: the authors state that “our study illustrates that the view on pain management of the physician in charge plays a major role in opioid prescription patters ..”. In my opinion, this conclusion cannot be based on the results of this study.
9. I miss the perspective of the patient in this study. An important question is whether the pain was adequately controlled (with or without opioids).

 

Minor comments

1. In the abstract information on the observation period – 1 year – is missing.

 

Author Response

We would like to thank the reviewers for their thoughtful reading of our study, their comments and suggestions that helped us to rewrite and improve our paper.

Review report 1:

1. This is a retrospective study in a low number of patients. My main concern is due to the small number of patients this study has not enough power to make a reliable comparison between the 2 centers and also lacks power to make reliable conclusion form the results of the multivariable analysis. Therefore, true differences / determinants may have been missed due to insufficient power.

Reply: We agree that the included number of patients is small. This is partly due to the lack of a central database of drug prescription in Switzerland. Therefore, data acquisition on drug intake is a time-consuming process. In order to obtain the data, one person (CH) spent two weeks in each center. Nevertheless, we believe it is important to publish data from as many countries as possible worldwide, also from countries without centralized databases such as Switzerland. This sort of publications will help to raise local awareness, and underline that this is not just a problem in the United States, the country that has published most studies thus far.

As suggested by the reviewer, we have downsized the comparison between the two centres in the abstract and the discussion. We also mention in this revised version that this is merely a hypothesis-generating study, and that larger studies are needed to confirm or reject the associations described in our study (page 9, last paragraph; page 10, first paragraph).

2. Data collection: the authors state that “handwritten prescriptions of opioids were also recorded”. Please specify the policy with regard to handwritten prescriptions for opioids. How long were handwritten prescriptions stored? How do we know that this gives a complete picture? Was the policy with regards to handwritten prescriptions identical in the two centers?

Reply: The policy for handwritten prescriptions for opioids is identical for all centers, as this is defined in the Swiss law (article 47 de l’ordonnance sur le contrôle des stupéfiants OCStup). Article 47 of this law states that every prescription of opioids must be written in a so called ‘carnet à souches’, a booklet with pre-printed prescriptions provided by the health authorities of each canton. Opioid prescriptions cover at most one month; one copy must be kept in the medical file of the patient, and medical files are kept at least 10 years; another copy is kept in the specific booklet. We therefore checked electronic records of each patient, but also these handwritten copies. This has been stated more clearly in the revised version (page 2, paragraph under ‘data collection and analysis’).

3. In my opinion, it is not surprising that opioid prescriptions were higher in the academic hospital than in the private hospital. E.g. , patients in the academic center more frequently had cancer and amputations. In other words, there may be bias by indication.

Reply: Indeed, patients in the academic center suffered more often from cancer and amputation, and although these variables were not associated with a greater prevalence of opioid prescription in the multivariate analysis, it remains a possibility that this-at least partly- explains the observed difference in opioid prescription between the academic and private hospital. As outlined above, we have downsized the conclusions of the multivariable analysis in this revised version, and focus more on the widespread use of opioids, also in Switzerland. (page 9, last paragraph; page 10, second paragraph).

4. Cancer and amputations are not mentioned among the causes of pain in Table 1. Is this correct? Or is pain after amputation categorized as neuropathic pain?

Reply: This is correct. In fact, we differentiated the types of pain by type and mechanism, not cause. Hence, cancer could cause both neuropathic pain or osteo-articular pain, depending on the patients. Amputation was indeed categorized as neuropathic pain. This is clarified in the revised version (page 3-4, Table 1).

5. With only 2 centers studied (and considering the differences between these centers), we do not know whether these results are representative for the Swiss situation.

Reply: This comment is correct. As mentioned in point 1), we downsized our discussion and did not draw the conclusion that our results are representative of the whole Swiss situation.

6. I assume that some patients with difficult to treat pain were prescribed a combination of drugs to relieve pain, e.g. opioids in combination with NSAIDs. This information on co-medication is missing in the manuscript.

Reply: We agree with that. We added the information on co-medication in the manuscript (p.5, first paragraph).

7. The authors rightly state in the discussion that gabapentin and pregabaline are also used in neuropathic pain / polyneuropathy. However, I miss information on the use of these drugs in Table 1. The same question for drugs like amitriptyline.

Reply: We thank you for this right comment. We added the use of anti-depressants in Table 1. Please note that we only recorded the global use of anti-depressants, not the specific type of molecules. Unfortunately, we did not collect data on the use of gabapentin or pregabaline and we can therefore not mention it. We however have information about patients suffering from polyneuropathy and patient having neuropathic pain. We will certainly remember to add the use of medication such as pregabaline in further studies.

8. Discussion, line 241-242: the authors state that “our study illustrates that the view on pain management of the physician in charge plays a major role in opioid prescription patters ..”. In my opinion, this conclusion cannot be based on the results of this study.

Reply: We agree that no definite conclusions can be made based on our results. We therefore modified the discussion, acknowledging the fact that we cannot prove the hypothesis we made. However it is important for us to show that this study opens doors for further studies.

9. I miss the perspective of the patient in this study. An important question is whether the pain was adequately controlled (with or without opioids).

Reply: That is a good point. We unfortunately did not collect data about the patients’ point of view on the control of their pain. This is definitely something that would be very interesting to do in further studies. However, we had many discussions with the patients of our study. We could see that there are two main groups of patients: those with pain and those without. Some of the patients with pain admitted that it was not well controlled by medication, which caused a huge impact on their daily activities. Prescribed medication was often accompanied by side effects, illustrating that this medical field merit more attention. We hope this publication will raise the awareness of nephrologists on this problem.

Reviewer 2 Report

To assess the prevalence of opioid prescription in HD patients from a European region and to compare prescription patterns between an academic and a private dialysis center in Switzerland, the authors conducted a retrospective study in patients on hemodialysis since at least six months, treated at Lausanne University Hospital (academic center) and Clinique Cecil (private center). A total of 117 patients were included; 29.1% received at least one opioid prescription during the study period. Center was an independent predictive factor of opioid prescription in the multivariate logistic analysis. According to these findings, the authors concluded that center was the best predictor of opioid use, underlining the importance of the prescribing physician on pain management, independently of patient-related factors.

 

The theme of this study is intriguing, but there are some concerns to be addressed.

 

1. It would be better if all the characteristics shown in Table 1 were presented in Table 2, and vice versa.

 

2. There is a possibility that the tendency to prescribe opioids differed among physicians, which might significantly affect the results in this study. Thus, the characteristics of physicians (sex, age, position, etc.) should be included in the adjustment factors.

 

3. This study was conducted at only 2 centers, one of which was an academic center (AC) and another was a private center (PC). That is, the results of this study might not be due to the difference between AC and PC, but merely due to the difference between the two centers studied. Thus, it would be desirable to conduct study in multiple centers to each of AC and PC.

 

4. As a sensitivity analysis, I would like to know the result of multivariable analysis when the factors with 0.05≤ P <0.1 (age 20-44 and osteoarticular pain) are added as adjustment factors.

 

5. The tests for variables that were not normally distributed are not informed in the Statistical analysis of the Materials and Methods section (page 2, line 80).

Author Response

We would like to thank the reviewers for their thoughtful reading of our study, their comments and suggestions that helped us to rewrite and improve our paper.

Review report 2:

1. It would be better if all the characteristics shown in Table 1 were presented in Table 2, and vice versa.

Reply: We agree, and we apologize for this. In fact, the only difference between both tables is the characteristic of cancer; we decided not to include it in Table 2 because no patients with cancer actually had an opioid medication. But we agree with your comment that it might not be clear why, and so we mentioned it in the text (p.6, first paragraph).

2. There is a possibility that the tendency to prescribe opioids differed among physicians, which might significantly affect the results in this study. Thus, the characteristics of physicians (sex, age, position, etc.) should be included in the adjustment factors.

Reply: We agree that the characteristics of physicians may influence their tendency to prescribe opioids. In our study, the two nephrologists from the private centre were one man and one woman, both aged close to 50. In the academic centre, opioids were prescribed by the head of chronic dialysis (aged 48), two certificated house officers between 35-40 years old, and seven residents between 30-35 years old. The head of dialysis and house officers were men. Among the residents, there were five women and two men. This information has been added to the revised version (page 5). We did not add the characteristics of the physicians in the multivariable model, as our study was too small to assess the influence of each physician characteristic on prescription behaviour. Instead, we used ‘center’ as the solely adjustment factor.

3. This study was conducted at only 2 centers, one of which was an academic center (AC) and another was a private center (PC). That is, the results of this study might not be due to the difference between AC and PC, but merely due to the difference between the two centers studied. Thus, it would be desirable to conduct study in multiple centers to each of AC and PC.

Reply: It is indeed possible that the differences are not due to the academic or private nature of the centers, but merely explained by local policies. We acknowledge the fact that this is a small study and hope that the publication of this study will allow us to obtain the necessary funding for a nation-wide survey. As explained in our answers to reviewer 1, obtaining data on drug prescriptions is a time-consuming and labour-intensive process in Switzerland, due to the lack of centralized drug registries. 

4. As a sensitivity analysis, I would like to know the result of multivariable analysis when the factors with 0.05≤ P <0.1 (age 20-44 and osteoarticular pain) are added as adjustment factors.

Reply: As requested by reviewer, we performed a sensitivity analysis that included all factors with 0.05≤p<0.1. Results remained unchanged (data not shown), and centre remained the only factor that was significantly associated with opioid prescription. We have added this result (but not the Table) to the text (p.6, 2^nd paragraph).

5. The tests for variables that were not normally distributed are not informed in the Statistical analysis of the Materials and Methods section (page 2, line 80).

Reply: We apologize for this omission. We have added this information to the text (p.2, statistical analysis section). These variables were log transformed.