Dear listmembers,
Could anyone help me clarify whether there is an adjustment one can use to
calculate CER, EER, ARR and NNT when control group and experimental group
data is presented in "events per 100 patient years" rather than proportion
of the group measured. The Diabetes Control and Complications Trial (NEJM
1993; Sept. 30; 329(14): 977-986) did just that in the results table (Table
2, page 981).
For example: in the primary prevention group for >3 step sustained
retinopathy:
control rate (conventional therapy): 4.7 / 100 patient-year
experimental rate (intensive therapy): 1.2 / 100 patient-year
risk reduction % (95% CI): 76 (62-85)
My calculations: 4.7 - 1.2 as .047 - .012 = .035 and NNT is 28.
Looking at p 979 under retinopathy, primary prevention cohort, it says,
"During a mean of six years of followup, retinopathy as defined above
("...a change of 3 steps or more on fundus photography that was sustained
over a 6 month period") developed in 23 patients in the intensive therapy
group and 91 patients in the conventional therapy group." N for intensive
therapy group was 348 and for conventional therapy group was 378 (table 1).
>From the text above, I calculated: 91 / 378 for CER (.2407) and 23 / 348
for EER (.0661). Then ARR = .24 - .07 = .17 and NNT is 5.88 or 6.
I would appreciate any insights on this! There must have been a good
reason to use per 100 patient years rather than the simple proportions.
best regards,
Ati Yates
______________________
Ati Yates, M.D.
Internal Medicine and Psychiatry
Michigan State University
Mailing address: 6092 Beechwood Drive,
Haslett, MI 48840
Phone: W 517 353 4362
H 517 339 5037
Fax: 517 339 5569
2nd Fax: 517 432 3603
--------------------------------------------e-mail: [log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|