1. The hypothesis or research question is not stated early in the paper. This is typically done at the end of the "Introduction" section. I had to dig for the hypothesis, which in fact showed up on pg.6, I think. There appears to be two hypotheses one on line 5 and another on the last line "Our main outcome…".
   
2. The term leaflet needs better definition. According to the authors the methods of intervention were a booklet, summary card, and "…a one page leaflet…". On page 5, line 7 "…also enclosed with the leaflet and the leaflet." About which leaflet are they talking? The word leaflet appears to be used to designate all 3 intervention methods. The word leaflet is also used in the description of the "Summary card". Certainly all three interventions could be called "leaflets", however, once the interventions have been differentially defined those definitions must be carried through out the paper else confusion will result.

In addition on Pg. 3. "…two kinds of general patient information by post…" According to Pg. 4 in fact three kinds of patient information were sent. See line 1 in the "intervention" section.

3. Pg. 4, line 7 might break the line into two sentences. "…systematic reviews." The content initially…"
   
4. I believe a section heading "Baseline questionnaire" would be helpful above the paragraph dealing with the baseline questionnaire.
   
5. The authors speak of the baseline questionnaire containing a measure of "attitudes". Attitudes about what?
   
6. They do not state their justification for "Notes review". Why is it important to the study to do this? Rationale is not presented.
   
7. Pg. 6 Should state the N used in the factor analysis, along with the methods of factor extraction and rotation. Reporting factor communalities (h²) would also be helpful as an indicator of factorial validity.
   
8. Although Cronbach’s alpha is acceptable, the test-retest reliability is not. Any basic measurement book will note a minimal value of about .70 as acceptable. The question is whether test-retest reliability is even necessary to report here. I doubt it.
   
9. Pg. 7, If using logistic regression they should state what was used as the independent variables, and the N.
   
10. Pg. 8, lines 4-6 are difficult to understand and interpret. Are these results presented as odds ratios with confidence intervals and do they refer to Table 2? The format of these data is inconsistent. If these are odd ratios with confidence intervals then the convention for reporting such results is as follows:

odds ratio (LL – UL) example: 1.37 (1.25 – 1.17)

11. Pg. 10, Table 1. The placement of superscripts is inconsistent.
    
12. Pg. 11, Table 2. The placement of superscripts is inconsistent. I was unable to find an explanation for superscript "2". In addition there are no differential definitions of "high attendance" and "not high attendance". As they appear in the table are the column headings "high attendance" and "not high attendance" reversed? The table title mentions "common attendance", yet common attendance doesn’t appear in the table. Table column capitalization is not consistent. Finally, the method of reporting odds ratios and confidence interval may not be consistent with convention (see item #10).
    
13. Pg. 12, Table 3. Chi square p-values reported as 0.000 adds nothing to the concept of significance. It’s ok to pick a p-value as a cut off. For these data, .01, .05, .005 or 0.001 are fine. If they are going to include ANOVA results in the table, then the table column heading "Chi square (p)" is inappropriate. Use some more general label such as "Statistical Test" then sub-label the table with "Chi Square Results" and "ANOVA Results", or some such thing. The table title needs to better describe what’s in the table.
    
14. Unless there is something specific to BMJ format I don’t understand why there are two sections labelled "Main results".
    
15. Finally Figure 1 is difficult to follow. The choice of words in the labels are hard to understand, i.e. what does "received as allocated" mean? Difficult to follow their subject brake down. Do they mean that a participant must have returned a questionnaire AND there must be attendance notes in order for that participant to be included in the study?
    
16. The entire article needs to be carefully proof read for punctuation, accuracy of reporting results in a conventional journalistic structure.

I believe the authors had a good idea. It can enhance medical cost savings and be of importance in managed care situations. Unfortunately, no one may ever know how such a good idea might have impact because of the 16 items stated above.

I believe that in its current form the manuscript is not acceptable for publication.

I would suggest that the authors rework the article and then resubmit.