Chest -- eLetters for Bach et al., 126 (6) 1742-1749

Author:

Keyword(s):

Year:

Vol:

Page:

To the editor:

We described a model for predicting the number of lung cancer deaths that will occur in a cohort of current and former smokers over a six year time period¹. In that study, we showed that the model predicted very accurately, virtually paralleling the observed number of lung cancer deaths in three ‘validation’ cohorts. In our study, we noted that these validation cohorts had been assembled decades earlier, and were comprised of mostly men, raising concerns that the model might not predict as accurately when applied either to more contemporary cohorts or to women. We now report on an additional validation analysis, conducted on women enrolled in the Nurses Health Study I, a large epidemiologic study that has been extensively described in other publications and used to evaluate risk factors for lung cancer and other diseases².

For our analyses, we focused on women who responded to the 1994 wave of the survey. As in our other validations, risk factor data were ascertained at the beginning of the analytic time period (from the survey responses). Only those women with risk factors within specified ranges were included, and women with a history of lung cancer were excluded. The model’s risk attenuation factor, used to estimate the decrease over the first year associated with being sufficiently healthy to enter a study, was not included, as the women in our analyses had entered the study many years earlier. All subjects were assumed to not have occupational asbestos exposure, and all deaths due to lung cancer were those validated by medical record review.

There were 12,257 nurses who responded to the 1994 survey and had risk factors in the range for inclusion in the model. These women had an average age of 63, had smoked an average of 22 cigarettes per day for an average of 36 years. Over six years of follow-up (70,550 person-years), the number of observed and expected lung cancer deaths were very similar (166 vs. 168, p = 0.82), Table 1. These results further validate our lung cancer mortality prediction model, and mitigate the concern that the model may not work as well in women, or in more contemporary cohorts.

Peter B. Bach, MD, MAPP

Colin B. Begg, PhD

Health Outcomes Research Group

Memorial Sloan-Kettering Cancer Center

New York, NY 10021

bachp{at}mskcc.org

1. Bach PB, Elkin EB, Pastorino U, et al. Benchmarking lung cancer mortality rates in current and former smokers. Chest 2004;126(6):1742-9.

2. Bain C, Feskanich D, Speizer FE, et al. Lung cancer rates in men and women with comparable histories of smoking. J Natl Cancer Inst 2004;96(11):826-34.

Table 1. Comparison of expected and observed lung cancer deaths among women responding to the 1994 round of the Nurses Health Study who meet the criteria for inclusion in the model.

Year of follow-up	Expected	Observed
1	28	36
2	56	64
3	84	99
4	113	124
5	141	143
6	169	166