Running head: CRITIQUE OF "A META-ANALYSIS OF INTERVENTIONS TO 1 Critique of "A meta-analysis of interventions to promote mammography among ethnic minority women" Brian Foster, Roxanne McMurray, Rhonda Oldham, & April Yearwood Union University NUR 710 Fall 2011 CRITIQUE OF "A META-ANALYSIS OF INTERVENTIONS TO 2 Critique of "A meta-analysis of interventions to promote mammography among ethnic minority women" Han et al. (2009) identified a lack of research related to the interventions to increase the mammography screening in the ethnic population. Based on the author’s concerns, they completed a meta-analysis to increase knowledge of the issues. The Problem or Justification Han et al. (2009) clearly stated the objective, purpose and goal of the meta-analysis; but did not utilize the terms problem, clinical question or hypothesis. However, from the stated objective, purpose and goal, one can identify the problem and determine a need for the metaanalysis. The authors clearly lay a case for the research by discussing multiple studies concerning the incidence of breast cancer, the need for preventive intervention and the lack of evaluation of intervention success where ethnic minority women are concerned. The objective, “was to determine the effectiveness of the interventions for improving mammography screening among asymptomatic ethnic minority women” (Han et al., 2009, p. 246). “The purpose of this meta-analysis was to determine the effects of intervention programs on mammography screening among ethnic minority women” (Han et al., p. 247). The authors identified that other appropriate research lacked data for minority women and noted their goal “was to fill this gap by conducting analyses on more recent studies …… that were targeted specifically to ethnic minority women” (Han et al., p. 247). The scope of the project was appropriate to address the problem or purpose of the study. Russo states “meta-analysis provides a standard approach for examining the existing literature on a specific, possibly controversial, issue to determine whether a conclusion can be reached regarding the effect of a treatment or exposure” (2007, p. 637). The authors plainly noted CRITIQUE OF "A META-ANALYSIS OF INTERVENTIONS TO 3 current research, a gap in the research and formulated a meta-analysis study to address the issue. The gap was the lack of other studies related to ethic minority women, which the purpose of the study specified. The topic of the review is important to nursing because it specifically relates to women’s health issues. With the growing number of minorities in America, nurses must be aware of the disparities and ways to overcome this issue. The results of this study concerning interventions to promote mammography use in minorities could allow nurses to re-focus their education efforts in ways that are more effective. For example, this study found “the intervention effect was bigger for studies conducted in a healthcare setting than for the community-based studies” (Han et al., 2009, p. 251). This information facilitates health care providers, including nurses, to strengthen valuable health care resources in the healthcare setting rather than the community environment. The meta-analysis did not use specific terms such concepts, variables or phenomena. Polit and Beck state phenomenon is “the abstract concept under study, often used by qualitative researchers in lieu of the term variable” (2012, p. 737). The phenomenon defined in this study is asymptomatic ethnic minority women in relation to interventions for improving mammography. Search Strategy The study clearly described reasonable criteria for selecting reference material including only English based articles from 2000 to 2008. The material included was from studies and abstracts. Studies were located using electronic and hand searches from a variety of collections with appropriateness verified by two study team members. The criteria for the material included in the study were: (a) The study aimed to increase use of mammography screening among asymptomatic women, either exclusively or in addition to other health behaviors; (b) the study included CRITIQUE OF "A META-ANALYSIS OF INTERVENTIONS TO 4 more than 40% of women with ethnic minority background (i.e., Asian American, Black or Hispanic) in the sample; (c) outcomes were based on a woman’s adherence to mammography’s screening, documented either by self-report or in a clinical database or medical record; (d) an experimental or quasi-experimental design was used in the study; and (e) the study was reported between September 2000 and August 2008. Not included were international studies because the focus was on intervention strategies to improve breast cancer screening among ethnic minority women in the United States. (Han et al., 2009, p. 247) These criteria were appropriate because they focused the literature review back to the objective, purpose and goal of the meta-analysis. Criteria (a) limited the review to asymptomatic women that is required by the objective. Criteria (b) focused the review on ethnic women, which is central to the overall purpose of the study. Criterion (c) and (d) centered the review on experimental or quasi-experimental designs to ensure high-level research techniques focused on the appropriate subject material were included. The databases used by the researchers included Medline, CINAHL, PsycInfo and Web of Science. The databases were appropriate because they offer a large number of original research articles and provided a wealth of information as evidenced from the reference list; but they are not all inclusive. Han et al. discussed by “excluding unpublished sources (e.g., dissertations) or government documents that might not be readily available” (2009, p. 251). Additional databases to consider are Ovid and EBSCO because of their unique search abilities; however these are subscription and do include a cost for use. The study included the key word phrases of “Asian, African American, Hispanic or Latino, breast cancer screening, mammography, experimental studies, interventions, and CRITIQUE OF "A META-ANALYSIS OF INTERVENTIONS TO intervention studies (Han et al, 2009, p. 247). The key words seemed to include most of the necessary items for an extensive review based upon the study’s objective, purpose and goal; but it was not exhaustive. One of the main words used throughout the meta-analysis was minority, but the authors did not include it as a key word phrase for the review. Han et al. (2009) discussed the study methods of electronic and hand searches to locate 749 titles and abstracts. However, the authors did not address any specific measures utilized to identify unpublished studies. Additional sources of work are identified by expanding the search to other databases beyond ones that focus on original, published work. The Sample The authors utilized studies based on specified inclusion criteria. The inclusion and exclusion criteria were easily elicited. They selected studies that had outcomes in common. Among these inclusion criteria were increasing mammography among “asymptomatic women”, the study must contain at least 40% of women with an “ethnic minority background”, outcome was dependent upon compliance with screening, “experimental or quasi-experimental design, and the study must have been reported between “September 2000 and August 2008” (Han et al., 2009, p. 247). The authors excluded international studies because they would not be representative of strategies implemented in the United States. The inclusion and exclusion criteria seem to express coherently the population of interest. The search strategy employed initially yielded 749 studies. The authors state that, “of these 749 studies, 607 were excluded” (Han et al., 2009, p. 247). However, the authors provide no explanation for this initial exclusion. They further exclude from the remaining 142 based on the absence of characteristics from the inclusion criteria. The authors utilized 23 studies that meet inclusion criteria (Han et al., 2009, p. 247). The population size available across all 5 CRITIQUE OF "A META-ANALYSIS OF INTERVENTIONS TO 6 included studies is 22,949 (Han et al., 2009, p. 249). The search strategy employed yielded a large sample of studies but given the exclusion of 607 studies with no description of where they fell among inclusion and exclusion criteria, it cannot be determined if this was a comprehensive analysis. There is no mention of specific strengths or weakness of the sample included in this analysis. However, the authors did test for bias in the 23 included studies by using statistical strategies such as the leave-one-out approach. Once they removed a study, they recalculated the mean weighted effect size. They also used the funnel plot and fail-safe N (Han et al., 2009, p. 249). This type of statistical analysis for bias shows the strength of the study sample. The authors did not describe a process to resolve issues with excluded studies. There were no accounts of contacting the original researchers to obtain previously unpublished data. Those studies with insufficient data were simply excluded. The authors state, “In addition, 13 did not include enough information to calculate the effect size” (Han et al., 2009, p. 249). Studies were also excluded for reasons other than insufficient data. For example, the authors did not include 43 studies because they did not include at least “40% ethnic minority women” (Han et al., 2009, p. 247). They also excluded studies based on lack of a control group, lack of clearly stated interventions, and absence of the appropriate outcome such as “a woman’s adherence to mammography screening” (Han et al., 2009, p. 247). The authors also excluded based on sample populations that were used in other included studies. Quality Appraisal The reviewers utilized a numeric scoring system to delineate the quality of the included studies. This was derived from a review of “relevant literature” (Han et al., 2009, p. 248). There is no defense of the quality determination method other than the citations provided. The quality CRITIQUE OF "A META-ANALYSIS OF INTERVENTIONS TO 7 was scored based items such as study design, outcome measure, clarity of outcome definition, and information on withdrawal. Each item had a possible score of 0 or 1 depending on the presence or absence of the desired characteristic. The authors state, “For the purpose of this analysis, studies with scores of 1-2 were considered to be low quality and those with scores of 34 were considered to be high quality” (Han et al., 2009, p. 249). While this method was clear and easy to follow, the authors made no attempt to defend their use of this quality rating tool. The authors utilized two raters to code the quality variables independently. In cases of identified inconsistencies, the team members would resolve the discrepancy by discussion. The authors did provide a significant қ for coding agreement 0.86 (Han et al., 2009, p. 249). However, the authors did not describe individual discrepancies or give specifics about their nature, scope, or significance. The appraisal information was overall adequately described. The authors allowed the reader to see the quality appraisal tool and how they used it. The results of the quality rating of each included study are listed in the supplemental hyperlink provided in the text of the study (Han et al., 2009, p. 249). Data Extraction (from individual studies reviewed) The authors adequately discussed the methods of data extraction. They stated after the literature was reviewed with limited via key words and English only articles, relevant studies were screened for inclusion criteria. Then, after being identified by two independent authors, the “titles and abstracts of all identified studies were reviewed by two study team members” (Han et al., 2009, p. 247) in which Russo (2007), states that “two or more authors of a meta-analysis should abstract information from studies independently” (p. 639). However, they did not state whether they were blinded to the authors. The authors developed a structured abstraction form CRITIQUE OF "A META-ANALYSIS OF INTERVENTIONS TO 8 they refer to as “Figure 1. Summary of the study selection process” (Han et al., 2009, p. 248) as well as a written description to define their process and reasoning for exclusion out of 749 studies and the eligibility for inclusion of the 23 studies obtained for systematic review. The reviewers coded the characteristics of the studies (variables for inclusion) as: first author, year, study design, setting, sample (percentage of ethnic minority women), unit of assignment, type of intervention, intervention period, time to outcome measure (months), method of outcome ascertainment, number of participants in the study groups, mean age of the study sample, proportion of mammography screening for the treatment and control groups, theory, control group (no intervention, minimal intervention or usual care, or other non-breast-cancer intervention), any cultural strategies used, and study quality. (Han et al., 2009, p. 248) This information was established beforehand to help create a homogenous population with the authors giving a rationale for exclusion and inclusion criteria. There is even more information provided on each study via an online link to a summary of the patient data characteristics for this meta-analysis. They further go into how they categorized the interventions as listed on page 248 of the article. The authors created and provided a table that addresses how they determined the study quality rating and defined each rating as: “studies with scores of 1 – 2 were considered to be low quality and those with scores of 3 – 4 were considered to be high quality” (Han et al., 2009, p. 248). Russo (2007), states that “a quality score for each study included in a meta-analysis may be useful to ensure that better studies receive more weight” (p. 639). Additionally, “two raters independently coded the variable. Every discrepancy was identified and resolved by discussion among team members” (Han et al., 2009, pp. 248-249). Furthermore, they conducted a CRITIQUE OF "A META-ANALYSIS OF INTERVENTIONS TO 9 sensitivity analysis on high versus low quality ratings as well as on possible influential studies to “gauge the impact on the variability of effect sizes” (Han et al., 2009, p. 250). Lastly, additional analyses were performed to examine publication bias as noted in the data analysis below. Data Analysis-overall The reviewers explained their method of pooling the results and how they integrated the data. They used the meta command in Stata to conduct the data analysis and to determine an overall mean weighted effect size (MWES) for the 23 studies (Han et al., 2009). The commonly used DerSimonian-Laird fixed results model was used to summarize the interventions effect along with using estimated confidence intervals. They used multiple approaches to compare the overall MWES along with examining publication bias via a funnel plot and fail-safe N calculations. The analysis of data seems to be thorough and credible within the studies used. Subgroup analyses and sensitivity analyses were conducted to rid the study of any extreme outliers that might skew the results. A table on page 250 shows the estimated effect sizes (p < .001) with a 95% confidence interval (0.043 to 0.113) for each of the 23 studies and is helpful to show the summary of findings. The authors also referenced an online table “which summarizes the characteristics of studies included in this meta-analysis, http://links.lww.com/A1244” (Han et al., 2009, p. 249). Data Analysis The meta-analysis performed was justifiable. The implications associated with the meta-analysis results indicate the “interventions were effective in improving mammography screening among ethnic minority women” with a 95% confidence interval from 23 studies (Han et al., 2009, p. 259). CRITIQUE OF "A META-ANALYSIS OF INTERVENTIONS TO 10 Appropriate procedures were followed for computing effect size for all relevant outcomes. Han and colleagues discussed in detail how effect size was calculated and what the sensitivity analysis included. Odds ratio was calculated for the estimated effect sizes to examine publication bias. However, the odds ratio value was not disclosed. The pooled results calculated effect size by using the random-effects model to report for significant heterogeneity among interventions, which was indicated by a significant Q (Q = 92.95, df = 22, p<.001) (Han et al., 2009, p. 249). The random-effects model was repeated to calculate MWES for various subgroups to further deal with heterogeneity. The appropriate subgroups analyses were accounted for in detail, to examine effectiveness of the different intervention methods to indicate whether effect size results varied by population subgroups. Publication bias was adequately and clearly dealt with. The authors first examined publication bias by “plotting the standard error by the natural logarithm of the logged odds ratio for the estimated effect size” (Han et al., 2009, p. 250). A funnel plot, and fail-safe N was also completed. The authors indicated that the funnel plot appeared slightly asymmetrical indicating potential publication bias. However, the authors stated that the fail-safe aided in adding rigor the study. More stated values, charts and tables would have been helpful to aid in summarizing the findings. Usability of Results-Conclusions The conclusions about the quality, quantity and consistency of evidence, are addressed in the discussion and are related to the research question. Quality and quantity are discussed in the limitations section. The results indicate that there is an overall increase of mammography use for minority women in the intervention group by 7.8% (Han et al., 2009, p. 251). Han and company acknowledged and gave a rational for why the results could not be compared directly with those of CRITIQUE OF "A META-ANALYSIS OF INTERVENTIONS TO 11 other meta-analysis, because of different study selection criteria and intervention typology, which were consistent with two other referenced studies. Inconsistency and low quality findings were addressed and accounted for in the promotoras intervention. The authors contributed the small and negative effect size to the study design, non-randomization, and large sample sizes. The authors indicated that in the future when using the network design such as promotoras, a smaller community with a well-prepared trainer that had a well-monitored plan, would be more appropriate. Interventions that were most effective in analyzing effect size were specified, along with which intervention studies, based on valid theories, were suggested for further studies. Consistency was present along with previous meta-analysis and found that intervention effect was bigger in community or healthcare settings. The significant results versus the non-significant results of the individual studies were indicated to generalize the outcomes and feasibility. Non-significant data was stated and a rationale was given in regard to implications in designing mammography-enhancing interventions for ethnic minority women. The majority of the results were nonsignificant. The access enhancing intervention, and directed approaches “yielded the biggest increase in mammography use, but the results could not be computed directly to other meta-analyses due to different study selection criteria and intervention typology” (Han et al., 2009, p. 251). The tailored, theory based intervention resulted in a bigger effect size compared to the non-tailored intervention. The authors could have eliminated some of the low-quality studies to change the results of analyses, but the study was still indicated as rigorous. Ethical implications were addressed by assessing that community research with a traditional randomized controlled trial could put the control group at an unfair advantage, if free mammograms were not available to all. CRITIQUE OF "A META-ANALYSIS OF INTERVENTIONS TO 12 The limitations of the review were noted. The limitation of lacking culturally matched materials or matching interventions deliveries were pointed out and suggested that future metaanalysis is needed, as more experimental data is available. Also, published analysis of intervention effects for Asian or Pacific Islanders or Hispanic women were lacking. The design limitations were found in examination of characteristics in the Hispanic subgroup, which was attributed to a large sample size, less controlled situation, and lack of randomization. Unpublished sources resulted in other limitations; but were justified as not affecting their inclusion and exclusion data. However, Polit and Beck, (2012) indicates that the “exclusion of grey literature in a systematic review can lead to bias, particularly the overestimation of effects” (p. 658). Another study revealed “restricting studies by language overestimates treatment effect only by 2%” (Russo, 2007, p. 640). The reviews used only from the United States pose a limitation, but the study question indicated that the study was specific to the United States and findings may not be generalized to other countries. Another limitation was the difficultly to tease out the effect of each individual component. A factorial design would solve this limitation, but was avoided because of expense. This is the only time expense was mentioned or considered in this study. The authors declared a limitation in the quality and quantity of their study. Uneven quality and quantity of meta-analysis studies is common. The authors attempted to address this by assigning estimates to low and high studies. Four to six studies were labeled as nonsignificant studies, which the authors stated that the low estimates should have been used as preliminary studies, however the studies were still used. The Quality of Reporting of Meta-analyses (QUOROM) guideline to determine validity, of the results, would have been helpful and wise to use. Implications for healthcare practice would be to increase access-enhancing strategies, increase the use of a theory-based, tailored approach and increase active community partners to CRITIQUE OF "A META-ANALYSIS OF INTERVENTIONS TO 13 improve screening outcomes. Through the above intervention, mammography use among ethnic minority women would improve. The need for future research was clearly stated. The authors suggested that future studies needed more tightly controlled trials to improve results when reviewing mammography studies among minority groups, particularly Hispanic women. They also suggested that alternative research is warranted to allow for research to benefit all ethnic minority communities. Additionally, it was suggested that future interventions are needed to promote mammography screening among ethnic minority women. Finally, the authors suggested that it would be helpful in future meta-analysis studies if the rate differences of mammography screening were consistent. The results indicate that there was “an average of 7.8% increase in mammography use in minority women in the treatment groups receiving a variety of interventions” (Han et al., 2009, p. 251). It was concluded that this “meta-analysis had a estimated intervention effect with a 95% CI of 0.043 to 0.113 indicating that the interventions were effective in improving mammography screening among ethnic minority women by using a random-effects model to account for significant heterogeneity” (Han et al., 2009, p. 249). The study size included 23 studies that dealt with variance, sensitivity analyses, quality, quantity, bias, and limitations, and the study had a robust combined result. This meta-analysis is reasonable to address this clinical problem. The results suggest that interventions are a need for increase use of a theory-based, tailored approach with enhancement of access, to promote mammography among ethnic minority women. Although it is difficult to decipher the results as a reader, the study can be a footing for future studies. With the evidence provided, it should be suggested that a change in practice is recommended. The study has external validity, and generalization of the results could be applied to minority women. Through an increase use of a theory-based, tailored approach and enhancement of CRITIQUE OF "A META-ANALYSIS OF INTERVENTIONS TO 14 access, minority women will improve in mammography screening. This will aid in detection and diagnosis of breast cancer and increase survival. Anytime survival rate can be increased, accommodations in practice should be made as it is realistic and possible. CRITIQUE OF "A META-ANALYSIS OF INTERVENTIONS TO 15 References Han, H., Lee, J., Kim, J., Hedlin, H. K., Song, H., & Kim, M. T. (2009). A meta-analysis of interventions to promote mammography among ethnic minority women [Entire issue]. Nursing Research, 58(4). Polit, D.F., & Beck, C. T. (2012). Evidence-based practice in nursing & healthcare (2nd ed.). Philadelphia, PA: Wolters Kluwer. Russo, M. W. (2007, August). How to review a meta-analysis. Gastroenterology & Hepatology, 3(8), 637-642.