Response to ‘Early view: an opportunity for enhanced peer review’

We welcome the anonymous readers’ interest in our advance-access manuscript, and appreciate the opportunity to respond to the Editor’s comments on our article as it reaches the Journal’s pages in print. First, the Editor suggests ‘it may be incorrect to state that the intervention led to a reduction in hypoglycaemia rates’ because the study did not report the change in hypoglycaemia rates from baseline. The aim of our study was to compare rates of adverse events between intervention and control groups during the course of the trial. Comparison with a control group, rather than comparison over time, is a critical feature of a clinical trial, because this controls for temporal trends unrelated to the intervention. In a sample assessed over a time of known developmental change – the transition into and across adolescence – this is particularly crucial. In an adequately powered trial, prerandomization differences between groups are not expected due to stratified randomization (and none were found on any demographic or disease-related characteristic). Therefore, our finding lower hypoglycaemia rates in the intervention arm than in the control arm during the trial supports the article’s conclusion that the intervention led to a reduction in hypoglycaemia rates.

Second, the Editor questions our decision to analyse orally and parenterally treated hypoglycaemia events separately, and comments on differences in events within the intervention group between study years (without comparison to the controls). For clarity, we summarize the absolute number of events and incidence rates for each year in Table 1. The decision to analyse orally and parenterally treated hypogly-caemia events separately was made a priori, and neither the former Editor nor any of the reviewers raised any questions or concerns with this approach during the peer review process. Based on the Editor’s newly raised questions outlined in this letter, we have conducted additional analyses using total events as an outcome. If examining the incidence of orally and parenterally treated hypoglycaemia combined, the control group demonstrated a modest increase, and the intervention group a modest decrease, from year 1 to year 2. However, the difference between the intervention and control groups in total hypoglycaemic events during year 2 was not statistically significant, likely due to two outliers, both in the intervention group, who reported five and six hypoglycaemia events requiring parenteral therapy during year 2.

Table 1.

Hypoglycaemic events by study year and treatment group in a randomized controlled trial of a behavioural intervention

	Year 1		Year 2
	Control	Intervention	Control	Intervention
Hypoglycaemic events treated by oral ingestion
Number of events	43	42	54	27
Incidence per 100 person-years	23.2	22.1	27.3	13.6
Hypoglycaemic events requiring parenteral therapy
Number of events	14	12	11	23
Incidence per 100 person-years	7.5	6.3	5.6	11.6
All hypoglycaemic events
Number of events	57	54	65	50
Incidence per 100 person-years	30.7	28.4	32.8	25.1

The Editor suggests that the current findings on hypogly-caemia events ought to have been included in our 2012 publication [1] reporting on the study’s primary outcomes, including HbA_1c, so that readers would not have to look elsewhere for those results. We note that the most relevant findings from that 2012 publication are summarized in the Introduction (last paragraph) and Discussion (first paragraph and second-to-last paragraph) of the current article. This study was a registered trial with pre-specified primary outcomes delineated in the study protocol, specifically, HbA_1c and adherence. As such, the primary outcomes article was limited to examination of these two outcomes. Notably, hypoglycaemia rates are infrequently examined in studies of behavioural interventions; a research gap this paper sought to address.

The Editor’s concluding comments allude to other issues without any substantiation that in our view inappropriately cast a negative cloud on our publication. The Editor speculates that the number of citations within the 5 years preceding submission of our manuscript provides insight into the validity of our conclusions. We simply do not see how the two are related. The Editor also suggests that there were too few authors on our article. As we did when we originally submitted our article for publication (in accordance with the Guidelines and Policies for the Conduct of Research in the Intramural Research Programs at the NIH), we affirm that all those who qualify for authorship on this article are named, and all those named qualify for authorship.

All that said, we agree with Dr Holt on the importance of scientific dialogue (yes, early and also often), and appreciate the opportunity to engage the Journal’s readers in such.