RESPONSE TO REVIEW COMMENTS:
PART 1: From the reviewer.
General comments: We agree with the reviewer that in the study setting we cannot rule out the possibility that some of the observed benefits may be due to factors that we could not control for in the analysis. This we had already put as the main point in the discussion section.
Specific comments:
| 1. | Introduction: the statement that most children live beyond the reach of hospital has been removed as suggested. So is the statement on the large-scale implementation. This was done in order to shorten the introduction and make it more focused on treated nets evaluation. | ||
| 2. | Methods: | ||
| a. | Local income: Assessment of local income was done the frame of this project in terms of ability and willingness to pay for the nets. This has been added in the text. | ||
| b. | Observation of sleeping places: This was done in the third year because a similar exercise was done in all the 18 villages included in the demographic surveillance system (DSS) in January-April 1997. The residents of the villages were not very happy with what they saw as an intrusion of their privacy. Hence, the assessment for this study was planned only once during the third year. | ||
| c. | Slide reading: The slides were read "blind" and corrections have been included in the paper | ||
| d. | Crude measurement: We also agree that some of the measurements were crude and we had clearly pointed that out in the paper. But we do not share the hesitation on the importance of net treatment for the following points. | ||
| i. | The programme distributed only pre-treated nets. The untreated nets were sold by other commercial companies and agents. | ||
| ii. | The comparison is with those with no nets at all. The question that we were answering is whether having a treated nets makes a difference compared to not having a net at all. It should also be noted that in the randomised controlled trials (summarized in Table 5 of the paper) most studies had those without nets as the comparison group. | ||
| iii. | The study was not designed to look at the differences between those with nets and those with treated nets. If this had been the case then the issue of correctly ascertaining treatment would have been the primary concern in the design, and a more complicated assessments of net treatment would have been necessary. | ||
| iv. | Since crude measurements of net treatment were used in our study, the misclassification will tend to assign positive treatment status to nets that are effectively not treated. This will result in an estimate of efficacy reduced towards the estimate for untreated nets. For us this conservative estimate of the benefits of treated nets was adequate for the evaluation. | ||
| v. | We agree that in this study untreated nets have also been shown to be very beneficial. A statement to that effect has been added in the discussion. | ||
| 3. | Results: The last paragraph has been clarified. This was done by rewording of the text and limiting the information provided to describe only the comparison of the observation of sleeping places and claim of use. Information on the comparison of the observations and reported ownership of net is omitted in the text. | ||
| 4. | Procedure: The sentence starting with "Selected children... was corrected to "The selected children...", as suggested. | ||
Part 2: From the Statistical review:
| 1. | Survey response: Numbers of selected children and those interviewed are given in table 1. The number of those who refused consent is now included in the text. The reasons for not being able to trace 142 children was mainly that they had moved to another location either in or outside the DSS area, either temporarily (attending social functions, weddings, funerals or moved to a farm house) or permanently. These are young children so they usually accompany their mothers everywhere. It can also be argued that for a small number of these children their parents refused consent by not being there at the time of the survey, as prior information was sent through their respective local leaders. But this could not be verified. | |
| 2. | Analysis: | |
| a. | The linear regression and logistic models were fitted with a random effect at the village level. Hence clustering by village was taken into account. The text is reworded to reflect that. | |
| b. | Survey time did not feature in any of the tables because the differences between the surveys were not significant. We also explored whether the magnitude of effect changed between the surveys but this was also not significant. | |
| 3. | This is a valid point. By dichotomising we lose power in our analysis and lose a bit of information. Therefore we have added the results of regression analysis of haemoglobin values in the results section. However, it is our view that public health and malaria control implementers more easily grasp the concept of a protective effect. Hence the inclination towards presenting all of our results in dichotomous outcomes. | |
| 4. | We have added a very short definition of social marketing in the abstract and introduction as suggested. | |
PART 3: From the full editorial committee:
| 1. | Responses for the referee's comments are given in part 1 |
| 2. | Responses for the statistical review are given in part 2 |
| 3. | The introduction has been shortened as suggested |
| 4. | Splenomegally was assessed in the physical examination by palpation of the right upper quadrant with the child seated on the mother's lap. And it was recorded as present or absent. So measurement or classification (e.g. Hackett) was not done. |
| 5. | Haemoglobin measurement was done with a HemoCue, a commercial test kits that comprises of portable photometer that measures light absorbed by azide-methemoglobin in the sample. A drop of whole blood is placed in a special cuvette which has reagents that releases haemoglobin from the cells and change it to azide-methemoglobin. |
| 6. | A brief comment has been added in the discussion that the biggest changes occurred between 1997 and 1998 |
| 7. | Protective efficacy has been defined in the results section. |
| 8. | The total number of references is reduced to 26 as suggested. |