PEER REVIEW HISTORY BMJ Open: first published as 10.1136/bmjopen-2016-014603 on 3 April 2017. Downloaded from

BMJ Open publishes all reviews undertaken for accepted manuscripts. Reviewers are asked to complete a checklist review form (http://bmjopen.bmj.com/site/about/resources/checklist.pdf) and are provided with free text boxes to elaborate on their assessment. These free text comments are reproduced below.

ARTICLE DETAILS

TITLE (PROVISIONAL) Patient perceptions of side effects: a cross-sectional survey of users in an online health community. AUTHORS Costello, Ruth; Patel, Rikesh; Humphreys, Jennifer; McBeth, John; Dixon, Will

VERSION 1 - REVIEW

REVIEWER Russell, Anthony Univ Alberta,Canada REVIEW RETURNED 14-Oct-2016

GENERAL COMMENTS An interesting approach with unexpected answers.

REVIEWER Michele Iudici Past address:

Rheumatology Unit Second University of Naples, Italy

REVIEW RETURNED 15-Nov-2016 http://bmjopen.bmj.com/

GENERAL COMMENTS The authors report an online cross-sectional survey aiming to identify the GC-related adverse events most important to patients. They analyzed data from 604 patients and reported the ranks for each side effect, without providing any additional analysis that can make the paper more interesting and appealing for the readers. For

example they do have data on the number and time of pills intake on October 1, 2021 by guest. Protected copyright. per day. It would be interesting to investigate if patient's perception of some AEs could be related to these parameters (evening intake and sleep disturbances...intake during meals and digestive symptoms...). Moreover, authors could also in deepen analyze how the experience of an AEs impacts on patient's perception. I wuold be prone to think that the experience of a vertebral fracture could absolutely change my point of view on this AE and the same can happens for the others. So, I would suggest to make a subanalysis taking into account the patient's history of GC-AEs occurrence. Another point I think the authors should underline is that this kind of exercise could be flawed by difference in patient's educational levels. They should acknowledge this in the text.

REVIEWER Adrian Loerbroks BMJ Open: first published as 10.1136/bmjopen-2016-014603 on 3 April 2017. Downloaded from University of Düsseldorf, Germany REVIEW RETURNED 21-Nov-2016

GENERAL COMMENTS Costello and colleagues examined which potential side effects of glucocorticoid (GC) use may matter most to patients. Doing so, they conducted an online survey among individuals reporting current or recent GC use. The most important side effects from patients’ perspectives were weight gain, and a “moon face”.

Overall, this study provides interesting findings as it illustrates how patient perceptions of the salience of GC side effects may differ from those of clinicians. Further, the authors adopted a novel approach to data collection (i.e. recruitment through a social online network).

The study suffers from a number of limitations though, which the authors partly acknowledged, but which should be presented more explicitly in my view. Also, additional analyses may be of interest. Those points and additional suggestions are detailed below.

Specific comments: 1. I recommend adding to the introduction why your research aim matters. You mention this in the discussion (e.g. patient perceptions of side effects will likely affect adherence), but I would do so early on. 2. Does you study make a novel contribution in light of what is currently known (study 8-10)? 3. In my view your findings could be affected by selection bias in at least two ways: 1) selection due to imperfect participation (as you acknowledged and as reflected by the very low response rate) and 2) selection due to your sampling choice (e.g. even if 100% had participated, those http://bmjopen.bmj.com/ using such social networks may differ from patients who do not). I suggest acknowledging in particular the latter selection process more explicitly. 4. It would be of interest to learn how exactly you obtained and used clinicians’ and patients’ feedback in the construction of the questionnaire measuring potential GC side effects. Especially, the approach to account for patient views needs to be presented in

greater detail, because I wondered whether you had been able to on October 1, 2021 by guest. Protected copyright. identify and consider potential patient misconceptions about GC side effects. In the field of asthma research, it has been observed for instance that patients may be concerned about developing „tolerance“ /addiction with regard to some drugs (e.g. Hyland et al., Clnical Therapeutics 2004). Possibly, the same holds true for GC, but has not been measured. 5. What was the distribution of responses to the items in block 6? 6. As the authors stated, the ranking of the importance of side effects may partly reflect their prevalence. It would be of interest, in this context, to re-run analyses while excluding those individuals who report to actually suffer from the respective symptom (item set #6 in your questionnaire). In my view, such analyses would be a more valid representation of real life scenarios related to the prescription of GC: Patients may decide not to use GC because of concerns about potential and anticipated side-effects (that are thus not yet experienced). 7. Another shortcoming of your study is that you only inquired after conditions and symptoms that are attributed to GC use. As a result, it remains elusive what proportion of the sample suffered from BMJ Open: first published as 10.1136/bmjopen-2016-014603 on 3 April 2017. Downloaded from particular complaints irrespectively of GC treatment. For instance, a reasonable proportion of your respondents (aged 50+) may have suffered for and may therefore have ranked this side effect unimportant. However, these cases are probably only partly captured by the items in block #6, which related hypertension to GC treatment. I suggest mentioning this limitation in the discussion section.

VERSION 1 – AUTHOR RESPONSE

Response to reviewers

Thank you to the reviewers for their helpful comments. We have responded to each of the reviewer’s comments below:

Reviewer 2: The authors report an online cross-sectional survey aiming to identify the GC-related adverse events most important to patients. They analyzed data from 604 patients and reported the ranks for each side effect, without providing any additional analysis that can make the paper more interesting and appealing for the readers. For example they do have data on the number and time of pills intake per day. It would be interesting to investigate if patient's perception of some AEs could be related to these parameters (evening intake and sleep disturbances...intake during meals and digestive symptoms...). Moreover, authors could also in deepen analyze how the experience of an AEs impacts on patient's perception. I wuold be prone to think that the experience of a vertebral fracture could absolutely change my point of view on this AE and the same can happens for the others. So, I would suggest to make a subanalysis taking into account the patient's history of GC-AEs occurrence. Another point I think the authors should underline is that this kind of exercise could be flawed by difference in

patient's educational levels. They should acknowledge this in the text. http://bmjopen.bmj.com/

Response: Thank you for your helpful suggestions. Although we have the usual time per day of administration, we do not have the number of pills per day so we did not feel we could conduct any additional analysis around this, though it is an interesting area. We have added stratification of scores by both experience and community group. Thank you for the suggestion to look at the impact of prior experience of GC-associated side effects and how this influenced scoring. We had in fact done this analysis for a conference presentation post-submission and it revealed some interesting on October 1, 2021 by guest. Protected copyright. findings. Prior experience of a side effect did increase the importance of that side effect to respondents. The results are tabulated in table 4, with the following text summarising the results on page 11: “When stratified by prior experience, participants who had previously experienced the side effect of interest reported higher median scores, with smaller IQRs. The side effects most important to those who had experienced them were diabetes, eye disease and CVD, all scoring a median of 10. The side effects most important to those who had not experienced them were reduced bone strength, CVD and eye disease (Table 4, figure S2). Although weight gain had the highest rank overall, it was ranked only fourth in those who had and who hadn’t experienced it prior to completing the survey, with median scores and IQRs of 9 (7-10) and 6 (2-9), respectively. The most commonly experienced side effects were, in order, weight gain, round face, insomnia, changes in mood, skin changes and indigestion, all of which were experienced by over half of the 604 respondents.”

We added stratification by community to give an indication of the effect of dose on the perception of side effects and this showed community rankings were mainly similar except for the PMR/GCA community. The results are tabulated in Table 3, with the following text at the bottom of page 9: “When stratified by community group the rankings remained similar to the overall rankings for all BMJ Open: first published as 10.1136/bmjopen-2016-014603 on 3 April 2017. Downloaded from communities except the PMRGCAUK community group, where the side effects most important to respondents were eye disease, CVD and insomnia, with weight gain fourth (Table 3).”

The reasons for the difference by prior experience and by community are discussed in the first paragraph of the discussion (page 12): “It is known that oral GCs have many side effects, but few studies have investigated which matter the most to patients. This survey found that overall weight gain, insomnia, and moon face were the side effects ranked highest by patients, despite them being less clinically serious. The importance of side effects to respondents was different depending on whether they had been experienced, with clinically serious side effects (diabetes, eye disease and CVD) being most important to respondents who had experienced them. As these clinically serious side effects had not been experienced by the majority of respondents they dropped in the rankings overall. Weight gain, scored at 9 out of 10 for those who had experienced it and 6 out of 10 for those who hadn’t, ranking at fourth position in both groups, but rose to the top ranking overall because of its high prevalence having been experienced by 442/604 (73%) participants. Participants from the PMRGCAUK community rated eye disease as most important, with CVD second and insomnia and weight gain joint third. This contrasted to all other communities where weight gain was the most important side effect overall. This group may be taking a higher dose of GC, compared to the other communities, which may explain the difference. Alternatively, respondents from this community may be older, and thus could be more concerned about diseases more prevalent at this higher age. Awareness of potential ocular involvement of GCA may also make the possible occurrence of further eye disease particularly concerning.”

We have also discussed how education could have influence scores in the discussion (page 14, paragraph 1), although didn’t have any such data to allow an analysis exploring whether educational level influenced beliefs.

Reviewer 3: Specific comments: 1. I recommend adding to the introduction why your research aim matters. You mention this in the http://bmjopen.bmj.com/ discussion (e.g. patient perceptions of side effects will likely affect adherence), but I would do so early on.

Response: Thanks for this recommendation, this has been added to the end of the introduction (page 4, paragraph 2), to state the following: “Furthermore, patients may elect not to take GC therapy because of concerns about possible side effects.” on October 1, 2021 by guest. Protected copyright.

2. Does you study make a novel contribution in light of what is currently known (study 8-10)?

Response: This study looks at side effect perceptions across disease groups, which is novel in that the previous studies have either looked at specific disease groups: rheumatic diseases, ITP or adrenal insufficiency. This has been added to the latter part of the introduction (page 4, paragraph 2).

3. In my view your findings could be affected by selection bias in at least two ways: 1) selection due to imperfect participation (as you acknowledged and as reflected by the very low response rate) and 2) selection due to your sampling choice (e.g. even if 100% had participated, those using such social networks may differ from patients who do not). I suggest acknowledging in particular the latter selection process more explicitly.

Response: Thank you, we have now acknowledged this in the discussion (page 13, paragraph 2). Further details are given in the response to the editor above. BMJ Open: first published as 10.1136/bmjopen-2016-014603 on 3 April 2017. Downloaded from 4. It would be of interest to learn how exactly you obtained and used clinicians’ and patients’ feedback in the construction of the questionnaire measuring potential GC side effects. Especially, the approach to account for patient views needs to be presented in greater detail, because I wondered whether you had been able to identify and consider potential patient misconceptions about GC side effects. In the field of asthma research, it has been observed for instance that patients may be concerned about developing „tolerance“ /addiction with regard to some drugs (e.g. Hyland et al., Clnical Therapeutics 2004). Possibly, the same holds true for GC, but has not been measured.

Response: The methods section has been expanded to describe further how the questionnaire was developed (page 5 & 6) (see details in Response to Editor above). This specifically notes that no additional GC-associated side effects were suggested by the patient review process. We had not considered patients misconceptions of GC side effects, but have added to the discussion that patients’ education and understanding of their own condition (e.g. risk of blindness in GCA) may have influenced the scores (page 14, paragraph 1).

5. What was the distribution of responses to the items in block 6?

Response: This has been added, see table 4, as well as the comment below.

6. As the authors stated, the ranking of the importance of side effects may partly reflect their prevalence. It would be of interest, in this context, to re-run analyses while excluding those individuals who report to actually suffer from the respective symptom (item set #6 in your questionnaire). In my view, such analyses would be a more valid representation of real life scenarios related to the prescription of GC: Patients may decide not to use GC because of concerns about potential and anticipated side-effects (that are thus not yet experienced).

Response: We have now added an analysis stratified by patients’ prior experiences of GC-associated side effects, presented in Table 4 and also in the results text on page 11 (further details are given in response to reviewer 2). These interesting results are discussed further in the first paragraph of the http://bmjopen.bmj.com/ discussion (page 12): “The importance of side effects to respondents was different depending on whether they had been experienced, with clinically serious side effects (diabetes, eye disease and CVD) being most important to respondents who had experienced them. As these clinically serious side effects had not been experienced by the majority of respondents they dropped in the rankings overall. Weight gain, scored at 9 out of 10 for those who had experienced it and 6 out of 10 for those who hadn’t, ranking at fourth position in both groups, but rose to the top ranking overall because of its high prevalence on October 1, 2021 by guest. Protected copyright. having been experienced by 442/604 (73%) participants.”

7. Another shortcoming of your study is that you only inquired after conditions and symptoms that are attributed to GC use. As a result, it remains elusive what proportion of the sample suffered from particular complaints irrespectively of GC treatment. For instance, a reasonable proportion of your respondents (aged 50+) may have suffered for hypertension and may therefore have ranked this side effect unimportant. However, these cases are probably only partly captured by the items in block #6, which related hypertension to GC treatment. I suggest mentioning this limitation in the discussion section.

Response: This limitation has been added to the discussion as follows on page 14, paragraph 1: “We did not collect information about comorbidities in participants and were thus unable to examine how this may have influenced beliefs. For example, a patient with prevalent hypertension may have considered high or cardiovascular disease to be particularly important to them as a GC-associated side effect.” VERSION 2 – REVIEW BMJ Open: first published as 10.1136/bmjopen-2016-014603 on 3 April 2017. Downloaded from

REVIEWER Michele Iudici University of Geneve, Suisse REVIEW RETURNED 04-Jan-2017

GENERAL COMMENTS The authors substantially improved the text, as requested.

REVIEWER Adrian Loerbroks University of Düsseldorf, Germany REVIEW RETURNED 10-Jan-2017

GENERAL COMMENTS The paper hase improved a lot. The authors may want to revise the abstract based on the revised paper (e.g. mentioning the knowledge gap, the additional analyses), if the word count permits. This decision is entirely at the authors' discretion.

http://bmjopen.bmj.com/ on October 1, 2021 by guest. Protected copyright.