Overview
1. The paper is original, it reads well and is of importance to clinicians, patients, teachers and policymakers. The BMJ is the appropriate place to publish because it can reach each of these target groups. The scientific reliability is further discussed below with reference to the consort statement for reporting of randomised controlled trials1 . I would recommend publication in the BMJ following revisions by the authors.
Title and Abstract
2. The title identifies the study as a randomised controlled trial and the abstract is presented in a structured format. Subjects: states that 251 students were recruited to the study when 272 were actually recruited from a cohort of 1515 students with 251 (92%) completing the study.
Introduction
3. The aim of the study and outcome measures are clearly defined with appropriate reference to the literature.
Methods- protocol
4. The planned study population is clearly defined together with inclusion criteria. The planned intervention (Triple A Program) was outlined together with timescales. Primary and secondary outcome measures were clearly stated with clinical relevance linked to minimum stated difference in Quality of Life (0.5 unit change in QoL total score). No target sample size was projected which is acceptable since the unit of randomisation was school rather than individual. Rationale and methods for statistical analysis were entirely appropriate.
Methods- Assignment
5. Cluster randomisation was used to assign the schools to the study group. The allocation schedule was generated using a random number generator and concealed by the closed envelope technique. The generator and executor of the assignment were independent.
Methods- Masking
6. At baseline, measurements were made prior to implementation of the allocation schedule therefore blinding was possible. However, during the intervention the study team conducted a workshop to train Asthma Peer Leaders in the Intervention group however it is not clear whether the outcome assessors and data analysts were actually blinded for the follow up.
Reference:
1Begg C, Cho M, Eastwood S, Horton R, Moher D, Olkin I, et al. Improving the quality of reporting of randomised controlled trials:the CONSORT statement. JAMA 1996;276:637-9.
Results
7. Figure 1 summarises participant flow throughout the study. However, the numbers reported in the central box (control=167, intervention= 158) add up to 325 and not 317 as stated. These 8 patients are not referred to in the boxes that follow. Other missing data was adequately explained and a description provided.
8. Table 1 provides baseline data for the 251(113 intervention and 138 control) participants who completed the study. Paragraph 2 of the results section states " There were more females in the intervention group (n=80, 66%) than in the control group (n=67,44%)." The percentage values should be 71% (80/113) and 49%(67/138) respectively and not as stated. Alternatively the text may be correct and the table values of n understated. The content and/ or the title of the table should be revised to confirm whether the data relates to baseline data for all students recruited to the study (n=272) or baseline data for all students who completed the study (n=251).
9. It would improve the paper if the baseline data for years 7 and 10 were presented separately in table 1. In contrast all follow up data are presented separately for year 7 and 10 as proportions, what numbers were in each of these groups? Provision of detailed baseline data would assist interpretation of these results and permit alternative analysis.
10. The results section (paragraph 2) states that the mean (SD) QoL was 5.6 (1.3) in the control group at baseline. In contrast table 1 states the QoL Total Score (SD) was 5.7 (1.3) in the control group at baseline. Once again is this an error or do the figures in the text relate to the data for all students recruited and the data in the table relate to baseline data for all students who completed the study? This seems unlikely as the text cross references to the table. In addition the terminology is inconsistent with mean QoL used in the text and QoL Total Score used in the table, the latter has quoted a standard deviation therefore I assume it refers to the mean also.
11. The primary outcome measure was QoL score. However only the mean, difference and CI were presented following the intervention. Data presented in Table 2 indicates the proportion of students with a clinical improvement in QoL Score (defined as >0.5units). No results are provided relating to the status of the remaining students. It would be interesting to quantify the proportion of students with a decrease in QoL score of >0.5 units.
12. The results section (paragraph5) states that "…no change in the control group (8.5days Vs 8.2 days)." In contrast figure 3 illustrates the values (5.5 Vs 4.0 days) in the control group. The text for the histogram data (provided separately) supports the latter.
13. Figure 1 illustrates that 251 students completed the study. The results (paragraph 1) state " matched baseline and post intervention data were only available for 251 students." However, table 3 provides results for 253 students – the title of the table includes " matched for pre and post n=253." Where did the data for the additional students come from? In table 3 the controls (n=142) is 4 higher than expected and for the intervention group (n=111) is 2 less than expected.
14. For all statistical tests, significance was accepted when p<0.05. This is not consistent with the results (paragraph 4) which refers to"… females improving in the activities domain (p=0.06)." The number of outcome measures in this study increases the risk of a Type 1 error. The P value could have been adjusted using the Bonferoni or equivalent method to allow for multiple comparisons.
Discussion
15. The study aims have been answered. The findings of the study were discussed critically in comparison with other studies. The validity of the QoL instrument and reliability of the methodology was discussed in relation to other published literature.
16. Discussion paragraph 3 "Hawthorne effect"- reference 23 is stated however it relates to the paper by Sackett, which is reference 24.
17.Limitations of the study are discussed however I would have expected some discussion on the finding that there were no significant intervention effects on QoL, school absenteeism and asthma attacks for year 7 students in contrast to the findings for year 10 students. This is particularly interesting considering the design of the Triple A Program, which involves the cascading of health lessons. In this paper the cascade included year 11 (Asthma Peer leaders) providing lessons to year 10, who in turn develop and present key messages to year 7. Why did the improvements in QoL and asthma morbidity fail to cascade beyond year 10? This requires further investigation.
Conclusion
18. The paper reads well despite the errors in the presentation of the results. The methodology is robust and the subject significantly original to merit publication following revision. In the past decade the move from clinician management to patient self- management has improved the clinical outcomes for a number of chronic diseases including asthma. Peer led education provides a novel strategy for empowering adolescents with asthma and this paper (following revision) will make a useful contribution to the literature.
……………………………………………
Professor Clare A Mackie
The School of Pharmacy
The Robert Gordon University
Aberdeen
AB10 1FR
16th August 2000
Note: Major points include 2,7,8,10,12,13,14 and 17.