[R] Incidence estimated from Kaplan-Meier

[Ravi Varadhan]

The 1-Pr(disease free survival) estimate from KM is not appropriate if
competing risk of mortality (from causes other than the disease of interest)
are present.  In that case, 1-Pr(disease free survival) over-estimates the
cumulative incidence of disease.  The larger the hazard of mortality, the
larger the over-estimation.  This is a well-known phenomenon in the
competing risks literature.  See, for example, Gooley et al. (Stats in Med
1999).

Ravi. 

----------------------------------------------------------------------------
-------

Ravi Varadhan, Ph.D.

Assistant Professor, The Center on Aging and Health

Division of Geriatric Medicine and Gerontology 

Johns Hopkins University

Ph: (410) 502-2619

Fax: (410) 614-9625

Email: rvaradhan at jhmi.edu

Webpage:  http://www.jhsph.edu/agingandhealth/People/Faculty/Varadhan.html

 

----------------------------------------------------------------------------
--------


-----Original Message-----
From: r-help-bounces at stat.math.ethz.ch
[mailto:r-help-bounces at stat.math.ethz.ch] On Behalf Of Frank E Harrell Jr
Sent: Thursday, July 05, 2007 8:48 AM
To: Nguyen Dinh Nguyen
Cc: r-help at stat.math.ethz.ch
Subject: Re: [R] Incidence estimated from Kaplan-Meier

Nguyen Dinh Nguyen wrote:
> Dear all,
> 
> I have a stat question that may not be related to R, but I would like to
> have your advice.
> 
>  
> 
> I have just read a medical paper in which the authors report the 1-p
(where
> p is the cumulative survival probability from the Kaplan Meier curve) as
> incidence of disease.  
> 
>  
> 
> Specifically, the study followed ~12000 women on drug A and ~20000 women
on
> drug B for 12 months.  During that period 29 women on drug A and 80 on
drug
> B had the disease.  The incidence of disease for A and B was 0.24% and
0.30%
> respectively.  However, instead of reporting these numbers, they report
the
> 1-p figure which was 0.3% for A and 0.6% for B. 
> 
>  
> 
> So, the incidence from 1-p was substantially higher than the actual
> incidence.  My question is: is it appropriate to use 1-p estimated from
> Kaplan-Meier as the incidence of disease?  If not, why not? 
> 
>  
> 
> Regards,
> 
> Nguyen

Yes it's appropriate, and it makes you state the cumulative incidence by 
time t rather than leaving time unspecified.  In your example it is 
likely that all women weren't followed completely, so simple incidences 
are not appropriate to compute because the denominator is not constant.

Frank

> 
>  
> 
> ____________________________ 
> Nguyen Dinh Nguyen, 
> 
> Bone and Mineral Research Program 
> Garvan Institute of Medical Research 
> St Vincent's Hospital 
> 384 Victoria Street, Darlinghurst 
> Sydney, NSW 2010 
> Australia 
> Tel; 61-2-9295 8274 
> Fax: 61-2-9295 8241 
> E-mail: n.nguyen at garvan.org.au 
> 
>  
> 
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
> 


-- 
Frank E Harrell Jr   Professor and Chair           School of Medicine
                      Department of Biostatistics   Vanderbilt University

______________________________________________
R-help at stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.