Methods
Identification and Selection of the Literature
Online searches were conducted on the Pubmed (1950s to June 2008), Embase (1974 to June 2008), Scopus (1960 to June 2008), Amed (1985 to June 2008), and Cinahl (1982 to June 2008) databases. The Pedro database and Cochrane register of systematic reviews were also searched. The databases were searched using a combination of the following keywords and MESH terms that were generated with the help of a medical librarian: "post-operative pain AND mastectomy OR breast neoplasm," "Pain, postoperative" [Mesh] AND "Mastectomy [Mesh] OR Breast/neoplasm/[Mesh] surgery" OR morbidity [Mesh] AND "post mastectomy" AND "Pain, postoperative," "post-operative pain" AND "morbidity" AND "mastectomy" OR "breast cancer." Hand searches were conducted on the bibliographies of identified articles to check for relevant studies not identified by the search strategy. The search terms yielded 1,551 titles that included one of the search terms, and one reviewer (M.C.) read the titles and eliminated irrelevant studies. Articles were eliminated if the titles clearly identified that the content did not relate to post treatment symptoms. The abstracts of the remaining studies (n = 227) were read by the same reviewer to determine if pain was measured and results reported. The related articles links were scanned for any relevant articles not identified by the search terms. Nine further studies suitable for inclusion were identified. The process yielded 69 articles for possible inclusion (Fig. 1)
(Enlarge Image)
Figure 1.
Search strategy stages 1 and 2 of the systematic review.
Inclusion Criteria
Articles were included if they were studies on human subjects that presented epidemiologic data on pain after nonmetastatic primary breast cancer treatment. Only studies published in peer-reviewed journals, in English, and with a cohort described as prospective were included. Prospective studies are more reliable in determining whether the variable of interest is related to the event. The information on pain could be reported separately or as a subset of a larger instrument.
Exclusion Criteria
Studies that did not have pain as an outcome measure, letters/comments, or case studies, descriptive articles on postmastectomy pain syndrome, and articles detailing management, treatment strategies, or treatment techniques for posttreatment pain were excluded. Articles published in a language other than English and studies that included participants with diagnoses of metastases or recurrence of cancer were also excluded.
Methodologic Quality
In this systematic review, an 18-item checklist concerning methodologic aspects of studies was used to rate the studies (Appendix A). All of the items in the checklist have been used in previous studies. Originally used by Borgouts, Koes, and Bouter (1998) and Hudak, Cole, and Frank (1998), the list has since been modified and used by Scholten-Peeters et al. (2003). Kuijpers, van der Windt, van der Heijden, and Bouter (2004) and McLean (2007) incorporated modifications for assessing validity as described by Altman (2001). The items can be scored as positive (+), negative (−), or unclear (?). A positive score is awarded if there is sufficient information and a positive assessment of an item. A negative score is awarded if there is sufficient information but potential to bias due to inadequate design or conduct. Finally, a score of unclear is awarded if there is insufficient or no information given. In line with authors of earlier systematic reviews, we considered a study to be of good quality if it scored positive for ≥50% of the criteria (Borgouts et al., 1998; Kuijpers et al., 2004; McLean, Klaber-Moffett, Macfie Sharp, & Gardiner, 2007; Scholten-Peeters et al., 2003).
Assessment Procedure
Two authors (M.C. and E.C.Q.) reviewed each full text article to attempt to limit bias, minimize errors, and improve reliability of assessment procedure (Higgins & Green, 2008). Reviewers independently assessed the quality of the studies. Each study was given an overall score based on the number of positively rated criteria (maximum score = 18). A third reviewer was identified for use in cases where disagreement persisted.
Before the assessment of the studies, a pilot study was conducted using a sample of papers to ensure that the appraisal criteria could be applied consistently. Higgins & Green (2008) suggests that a sample of three to six papers spanning low to high risk of bias be used for this purpose. Four articles were used to pilot the quality-rating process. Assessors independently scored the articles and then met to discuss results. Piloting highlighted differences in interpretation of several items, and the criteria were amended to clarify what constituted a positive score. Amendments are in italics in Appendix A.
Paper data collection forms were used to facilitate discussing disagreements and correcting errors. Studies were rated blindly using the forms designed for the purpose. The two assessors then met and presented the scoring on a study by study basis.