| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Guest Access | Sign In via User Name/Password
|
|
|
|||||||
Guest Access | Sign In via User Name/Password |
|||||||||
Firestone Institute for Respiratory Health, Hamilton, ON, Canada
Correspondence to: Malcolm R. Sears, MBChB, Firestone Institute for Respiratory Health, McMaster University, Hamilton, ON, Canada L8N 4A6; e-mail: searsm{at}mcmaster.ca
To the Editor:
Nelson et al1 discuss an apparent lower rate of deaths associated with salmeterol in the Salmeterol Multicenter Asthma Research Trial (SMART) [1.22 per 1,000 person-years] compared with the previous UK surveillance study (2.32 per 1,000 person-years).2 However, this low rate cannot be verified from their data.
There were 13 deaths in 13,176 salmeterol-treated patients, with median treatment duration of 197 days. Hence, the death rate is (13/13.176) x (365/197), namely 1.83 deaths per 1,000 person-years. Given 22% discontinued prematurely, mean rather than median treatment duration should be the denominator, and so even 1.83 deaths per 1,000 person-years is an underestimate of the true risk.
It is unclear how Nelson et al1 calculated the risk of 1.22 per 1,000 person-years (see page 25 of their report). Does this rate refer to white patients only? Six deaths among 9,281 white patients gives (6/9.281) x (365/197) or 1.20 deaths per 1,000 person-years. If so, the calculation for African-American patients should also be reported, namely (7/2.366) x (365/197) or 5.48 deaths per 1,000 person-years.
Ages at death in the salmeterol-treated patients were 14, 37, 41, 46, 46, 47, 47, 51, 56, 56, 60, 62, and 67 years, similar to those deaths in the UK study.3 The true comparators for these study populations are death rates for their age group, not for the total population, as most asthma deaths occur in the elderly.4 In the United States from 1990 to 2001, mortality from asthma as the underlying cause of death in the population aged 45 to 64 years (the majority of those dying in the SMART) was 2.4 per 100,000.5 Using a conservative 5% prevalence for adult asthma, this translates to a rate of 0.48 per 1,000 person-years. Hence, the death rate in the SMART is approximately fourfold higher than would be expected in the US asthmatic population of that age.
Footnotes
The Firestone Institute for Respiratory Health has received research funding from Altana, AstraZeneca, Aventis, Bayer, Boehringer-Ingelheim, GlaxoSmithKline, Merck Sharp Dohme, Pfizer, Schering-Plough. The author holds the AstraZeneca-endowed Chair in Respiratory Epidemiology at McMaster University.
Dr. Nelson is a consultant, speaker, and recipient of research grants from GlaxoSmithKline, and was also a member of the Morbidity and Mortality Review Committee. Dr. Dorinsky is an employee of GlaxoSmithKline.
References
National Jewish Research and Medical Center, Denver, CO GlaxoSmithKline, Research Triangle Park, NC
Correspondence to: Paul M. Dorinsky, MD, FCCP, GlaxoSmithKline, 5 Moore Dr, Research Triangle Park, NC 27709; e-mail: paul.m.dorinsky{at}gsk.com
To the Editor:
We appreciate Dr. Sears’ comment regarding the asthma-related death rate for salmeterol recipients in the Salmeterol Multicenter Asthma Research Trial (SMART). In the "Discussion" section, we inadvertently listed the asthma death rate for the total population (1.22 per 1,000 person-years). The asthma death rate for subjects exposed to salmeterol in SMART was 1.98 per 1,000 person-years. However, the conclusions regarding the asthma death rate with regard to Castle et al1 and Martin and Shakir2 remain unchanged. We have submitted a correction to the journal with this information.
As the number of asthma-related deaths in SMART for subjects receiving salmeterol was extremely low (approximately 10 deaths in 10,000 subjects), extrapolation of the death rate to larger populations, such as the US population of subjects with asthma, should be interpreted with caution. In addition, applying rates obtained from subpopulations, such as certain age groups, can further decrease the accuracy of the results due to an even smaller number of events, and without appropriate context such extrapolations can be misleading.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
