Author's response to reviews Title: Socioeconomic conditions and number of pain sites in women Authors: Finn E Skjeldestad (fisk@fhi.no) Toril Rannestad (Toril.Rannestad@hist.no) Version: 2 Date: 17 January 2012 Author's response to reviews: see over
Oslo, January 17, 2012 Dear Editorial Team We have with great interest read the editorial response and the comments from the referees to our ms Socioeconomic conditions and number of pain sites in women (VS: 1475014289611094). We have copyedited the ms according to the instructions. E-mail addresses and abstract are uniformly displayed. We declare that we have no competing interests. We appreciate the referees comments and have changed the manuscript accordingly. We have responded to the issues raised by the referees in the same chronological order as the objections were raised. Kind regards, Finn Egil Skjeldestad (corresp. author). Reviewer Haitze Janko de Vries Minor Essential Revisions: 1. Abstract Use the abbreviation NPS consequently. In results and conclusion: SCI "score". REPLY: The sentence with number of pain sites is deleted, fourth line. We have changed SCI score to SCI scores in the 17 th line 2. Introduction: 1st line: prevalence of health complaints like pain. Which pain do the authors mean? Acute pain, chronic pain, nonspecific pain, all kinds of pain? REPLY: We have specified pain and revised the first sentence to:.health complaints (1,2), like generalized (3), musculoskeletal (4), chronic nonmalignant (5) and complex/frequent/intensive pain (6). 3. 2nd line: number of pain conditions is considered a better parameter in pain assessment. I think the authors have to mention here why NPS is a better parameter for pain assessment (better predictor?) Only referring to the literature is not sufficient.
Perhaps it is better to state that (moderate) evidence is available that NPS might be a better parameter. REPLY: We have included a specification of better parameter : NPS is regarded as a better parameter in pain assessment (7) and may be more important than actual pain sites in determining the impact on health (8) and functioning (9). 4. is considered as (or: is regarded as) REPLY: Changed from considered to regarded 5. 2nd paragraph, 2nd line: We discovered general population. Does this mean that poor socioeconomic conditions is a predictor for cancer surviving? REPLY: One of our results in a previous publication (17) was that cancer survivors are living under poorer socioeconomic conditions. Since cancer survivorship is not the main focus in the present paper, we will not explore this issue any further. 6. Results I missed the descriptive information of the subjects. If this information has been presented in a former study, than may be referred to it. REPLY: We have added a sentence in the last line of Material and methods/study population: More detailed information on the study population is provided elsewhere (16-18). 7. Discussion The first paragraph of the discussion contained much repetition of results. It reads more comfortable if this paragraph is summarized. REPLY: The paragraph is shortened in the revised ms. 8. last paragraph: I wonder why a similar result compared to another Norwegian study is a strength of this study. REPLY: We have revised the second line of the last paragraph of the discussion to: The prevalence of women reporting no, one, two, three, or 4-7 pain sites (table 1) is similar to what is reported in another Norwegian study (12), supporting the external validity of the study.
9. Please explain why the cross-sectional design was a limitation. REPLY: We have rewritten the text in the revised ms, last paragraph before conclusion. 10. In the conclusion is stated that there is a threshold in the NPS count. It would be convenient for the reader if the authors explain somewhere in the discussion what the (clinical) consequences of this finding may be. REPLY: We have revised the discussion and added a new paragraph, 4 th paragraph. Major Compulsory Revisions: 1. Abstract: I believe two research questions interfere with each other. a) what is the association between socioeconomic conditions and NPS in women? b) is being a gynecological cancer survivor related to more NPS? The second question probably could have been answered in a previous study of the authors. To answer the first question, it would be more credible to exclude the women who are cancer survivors. REPLY: We have changed the abstract and Introduction to prevent misunderstandings. However, we disagree with the referee that we should exclude gynecological cancer survivors from the study population. In order to answer the second question raised by the referee we need a control group (women from the general population). We can study gynecological cancer survivor as an explanatory variable. Regardless of association, we can adjust for any finding in our models. However, we found that gynecological cancer survivorship was not associated with NPS. This is important information for clinicians, and their patients. That s why we have included gynecological cancer survivorship as an explanatory variable in our study. 2. Introduction The research question should have been stated clearly. In addition, some information about the relevance of knowing the association between NPS and socioeconomic conditions is desirable. REPLY: We have shortened the introduction and focused the aim of the study. 3a. Methods For me there is a major concerning point in this study: Cancer survivors usually are very sensitive to pain for the obvious reason that recurrence of cancer may occur. So one might expect that prevalence and NPS are higher compared to the general population. However, this was not observed in this study, so the authors conclude that they can consider the cancer survival group as a general group with poor socioeconomic conditions. I wonder if this is an appropriate sample to answer the research question, because the results can in my opinion not be generalized to a whole population. For the purpose of this research
question it may have been better to collect data from a poor socioeconomic area in a city or country. REPLY: In order to study poor socioeconomic position you need the whole range of socioeconomic status otherwise there will be no comparison with low and higher socioeconomic categories. Accordingly we disagree with the referee on this issue. Regarding expectation on NPS among cancer survivors, we have discussed this issue under Major Revision, paragr. 1. 3b. A solution may be to change the title into Socioeconomic conditions and number of pain sites in female long-term cancer survivors, and then carefully make suggestions of generizability to the general population. Or use only the women from the general population, the 493 are probably enough to do the analysis. The authors did not make it plausible that the cancer survivor group was really needed in this study. REPLY: We don t think this is a good solution. Se response to Major revision; paragr. 1 and 3 (above). 4. I understand that the authors used a case control design for their former study (Rannestad and Skjeldestad, 2007). However, in the study under review, this design is not needed. There were no cases to be compared with controls, so it is better to not mention the case control design. REPLY: We agree with the referee on this issue and has accordingly rephrased/deleted the design information several places in the ms. 5. Logistic regression analysis was used to explore the relation between NPS and socioeconomic conditions, with NPS as dependent variable. I recommend to apply linear regression (ENTER and not Forward) with socioeconomic conditions as (continuous) dependent variable. You can enter NPS as dummy variables into the model. In that case, table 4 can be reduced to 1 analysis, and should also present the standardized Beta and p- value. REPLY: We have tried different linear models. However, we intuitively don t like the Beta values, and find them less interpretable than aor. Thus we have remained our log. regression approach. 6. A high Odds ratio of 16.9 was found in model C. However, the confidence interval indicated that this finding was not very precise. At least the authors should mention this in the discussion. REPLY: We have in the discussion not mentioned the estimate (aor 16.9; CI: 4.6-61.7), just given information that NPS is significantly higher in the lowest SCI score. We think this is an appropriate way of handling a very high aor with large confidence intervals.
Level of interest: An article whose findings are important to those with closely related research interests Quality of written English: Acceptable Statistical review: No, the manuscript does not need to be seen by a statistician. Declaration of competing: I declare that I have no competing interests
Reviewer: Rebeca Rios Reviewer's report: Overall the manuscript represents an interesting contribution to the knowledge concerning social determinants of novel pain outcome. The cross-sectional survey design is appropriate, the manuscript overall is well-written, and the data is appropriately analyzed. Major concerns are in relation to points of clarity in describing the rationale of the study and in presentation of the results, and minor concerns are related to specific grammatical and wording issues. Major compulsory revisions: 1) I recommend further clarification of the rationale provided for investigating number of pain sites as the pain outcome of choice in relation to socioeconomic conditions. In the first paragraph of the first section, authors state that NPS is established as a better measure of pain, but do not explain how it is better or why this particular dimension of pain is particularly of interest in relation to socioeconomic conditions. The authors provide relevant citations that establish an association between socioeconomic conditions and pain prevalence, severity, but some further expansion on the different aspects of pain as they relate to socioeconomic conditions would strengthen the paper. REPLY: The same issues were raised by referee 1; se Minor revision paragr. 2, and Major revision paragr. 2. 2) Comments on results. Presentation of the results is overall logical and well-described. A) The statistical analyses section states that chi square tests and logistic regression were inferential methods used, but it would be helpful to make it clear in the first paragraph 2 and 3 of the results that chi square tests were used, and how many. REPLY: We have added statistical method as legend to table 2 and 3. Could experimentwise error be an issue here? If so, this should be commented on in the discussion. REPLY: We have not tested for multiple comparisons. Since we used stratified analysis (Table 2 and 3) for variable selection for log. regression, we don t see the relevance of this objection, as we rely upon the log. reg. results.. B) In the first sentence of the fourth paragraph of the results, it would help to clarify the description of the logistic regression models. It reads somewhat awkwardly. It states that cofactors from table 3 as well as age, BMI and history of gyn CA were entered, but age and CA history were included and significant variables in that table.
REPLY: We have included the approach of regression analysis; forward stepwise We have rewritten the sentence about variable selection for regression analysis. We agree with the referee that this issue was presented a little awkwardly. What is meant by co-factor, and how are the variables referred to as co-factors treated differently from BMI, age, gyn CA history (explanatory vs. control variables)? REPLY: In our understanding a co-factor is the same as an explanatory variable. We distinguish major exposure, SCI, from all other variables, which we label co-factors. C) Finally, tests of co-morbidity as a moderator are mentioned in the last sentence of the results. It would be helpful to describe here or in the statistical analysis section why they were conducted. REPLY: Several co-morbidities comprise pain. That is why we adjusted for comorbidities. We will submit a separate publication on this issue if the present ms is accepted for publication. 3) Comments on discussion section. The discussion section thoroughly describes the findings in the context of relevant literature, but just a few things seem missing. Particularly due to the cross-sectional design, the authors should consider reverse causality. A) It is possible that greater number of pain sites, through the impact of disability on ability to work, influences her socioeconomic standing. REPLY: We have added a sentence in the last paragraph of the discussion B) The measuree of socioeconomic conditions used here includes a measure of social support or social embeddedness, satisfaction with number of close friends. There are well-established linkages between social support and embeddedness with health. The measure of SCI includes social with more economic indicators; how do the findings relate to research on social support or embeddedness? REPLY: We are not aware of literature with NPS as outcome including social support or embeddedness. Minor essential revisions: 1) The first sentence in the abstract under methods uses the term study or studied three times. Please reword to reduce repetition. REPLY: Revised
2) In the second sentence of the introduction, is prevalence the correct term? I wonder if pain frequency is the intended term, since prevalence is a population characteristic and not a self-reported characteristic. REPLY: Revised 3) In the sentence in the study population paragraph under methods beginning In total, 653 responses.. I believe the term eligible is the wrong word. This appears to refer to the number in the final sample rather than those who were eligible. REPLY: We agree with the referee and have rephrased. 4) Results section, third paragraph, last sentence. This sentence reads somewhat awkwardly and could be more clearly stated. REPLY: Revised Discretionary Revisions 1) In the methods section, study population subheading, if data is available it would be informative to provide a statement commenting on how similar the study sample s demographic characteristics were to the general population. This then could be mentioned in the discussion as a comment on generalizability of the findings. REPLY: We selected controls (women from the general population) matched by age to gynecological cancer survivors in a 4:1 ratio. Therefore our sample will have a skewed age distribution towards higher age. We have compared employment status among women with 0-3 NPS with employment status of women in the general population and did not find any difference (discussion, 1st paragraph, reference 22). Regarding non-responders to the study, there were no differences in age and marital status when compared to responders (last paragraph of discussion). Level of interest: An article of importance in its field Quality of written English: Needs some language corrections before being published Statistical review: No, the manuscript does not need to be seen by a statistician. Declaration of competing interests: I declare that I have no competing interests.