HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Permissions Request Permissions Services Email this article to a friend Similar articles in this journal Similar articles in Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Braunstein, G.L. Search for Related Content PubMed PubMed Citation Articles by Braunstein, G.L.

Merck & Co. Inc., Rahway, NJ, USA

From the authors:

We are in agreement with G.L. Braunstein's comment that it is important to discuss end­points that are of clinical relevance to patients and to describe not just average values but results accounting for the entire population. His letter also provides us with an opportunity to rechallenge some of the commonly held assumptions about the clinical interpretation of asthma data and re­emphasise the clinical advantages of our approach to describing the response of patients to asthma therapy. We particularly wish to focus on the methodological issues that he has raised.

The best definition available of asthma control is from theconsensus guidelines; unfortunately, this description is neither population­based nor validated. Difficulties have been encountered in efforts to identify and accurately validate the important clinical measures of asthma control, and it remains unclear which outcome or combination of outcomes will define and serve to measure control of the disease. For example, a number of recent publications have focused on novel approaches to asthma control 1–3. In our study, the asthma control day was chosen as the primary end­point because this end­point incorporates patient­oriented measures, such as symptom scores, the need for rescue medication, and incidence of asthma exacerbations, measures that are directly and clinically relevant to patients.

We believe that the analyses of clinical trials that rely solely on comparisons of mean values frequently result in loss of important clinical information that is available in the distribution of response in the population. The overlap in response distributions can often provide a clearer determination of clinical comparability because the entire range of responses of individuals in the study population is taken into account. The plot and overlap statistics give the reader a more comprehensive interpretation of the study data than means and p­values, conveying to physicians a better understanding of the range of expectations for patient outcomes.

In addition, comparisons based on means are questionable when using data that show a clear deviation from normality. The overlap statistic is based on the well­known nonparametric Mann­Whitney U-test statistic. This approach is strictly rank­based and therefore does not rely on the assumption of normally distributed data. Using these criteria, there was a clinically important degree of overlap (>85%) in the response of patients treated with either active therapy in the asthma control day end­point despite the mean differences. In addition, both active therapies produced an overlap response (>95%) for forced expiratory volume in one second with no important difference in mean values between both treatments.

The question of equivalence bands with response distributions requires a different consideration from that noted by G.L. Braunstein. The approach here has to be similar to that described for pharmacokinetic bioequivalence. For example, our paper prespecified a criterion for the lower boundary of a 95% confidence interval estimate of overlap being >80%. Values of 80% were suggested by ROM and HWANG 4 and correspond roughly to differences between means of approximately half a SD, which is generally considered small­to­moderate. This is consistent with the bioequivalence literature and its use in other areas 5.

We agree with G.L. Braunstein that crossover designs have the theoretical advantage of providing within­patient comparisons. However, our extensive experience with these studies leads us to the conclusion that the day­to­day biological variability of asthma (at least in moderate­to­severe asthmatic patients) is so large that it confounds the ability to make clear within­patient conclusions (for example, a patient "responds" to one therapy but not to another).

Finally, we call for an accelerated effort in additional methodological work in the field of asthma, including the development of population­based and validated definitions of severity, control and response (all three measuring different dimensions of disease in patients and of treatment) as well as clearly defined and validated definitions of asthma control outcomes and surrogate clinical end­points. It is only through the development of these methodologies and definitions that we will be able to make the best judgments for asthma patients.

REFERENCES

1. Sont JK, Willems LN, Bel EH, et al. Clinical control and histopathologic outcome of asthma when using airway hyperresponsiveness as an additional guide to long­term treatment. Am J Respir Crit Care Med 1999;159:1043–1051.[Abstract/Free Full Text]
2. Reddel H, Ware S, Marks G, Salome C, Jenkins C, Woolcock A. Differences between asthma exacerbations and poor asthma control. Lancet 1999;353:364–369.[Abstract/Free Full Text]
3. Green RH, Brightling CE, McKenna S, et al. Asthma exacerbations and sputum eosinophil counts: a randomised controlled trial. Lancet 2002;360:1715–1721.[Abstract/Free Full Text]
4. Rom DM, Hwang E. Testing for individual and population equivalence based on the proportion of similar responses. Stat Med 1996;15:1489–1505.[Abstract]
5. United States Food and Drug Administration, Centre for Drug Evaluation and Research. Guidance for Industry: Statistical approaches to establishing bioequivalence. Rockville, United States Food and Drug Administration, 2001.

This Article Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Permissions Request Permissions Services Email this article to a friend Similar articles in this journal Similar articles in Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Braunstein, G.L. Search for Related Content PubMed PubMed Citation Articles by Braunstein, G.L.