library("survatr")
data(nhefs_surv)
nhefs_surv[, .N, by = "seqn"][, range(N)] # all 120 periods per id[1] 120 120sum(nhefs_surv$event) # 318 deaths[1] 318NHEFS survival data in monthly person-period format
Description
The NHEFS (National Health and Nutrition Examination Survey Epidemiologic Follow-up Study) cohort in rectangular monthly person-period format. Each of the 1629 participants contributes exactly 120 rows (one per calendar month from January 1983 to December 1992), forming the rectangular grid required by surv_fit(). The event indicator is 1 only in the month of death for the 318 participants who died before the end of follow-up; it is 0 elsewhere.
The primary use of this dataset is to reproduce the Hernán & Robins Chapter 17 causal survival analysis: the standardised 120-month survival under smoking cessation vs. no cessation, and the corresponding risk difference (≈ 0.2 pp, essentially null).
Usage
nhefs_surv
Format
A data.table with 195 480 rows and 13 columns:
- seqn
- Participant identifier (integer).
- time
- Calendar month of follow-up, 1 = January 1983, 120 = December 1992 (integer).
- event
- Binary event indicator: 1 = death in this month, 0 = alive or censored (integer).
- qsmk
- Smoking cessation between the baseline and follow-up interviews: 1 = quit, 0 = did not quit (numeric).
- sex
- Sex: factor with levels 0 (male) and 1 (female).
- race
- Race: factor with levels 0 (white) and 1 (non-white).
- age
- Age at baseline interview in years (numeric).
- education
- Highest education level: factor with levels 1 (8th grade or less), 2 (HS dropout), 3 (HS graduate), 4 (some college), 5 (college or more).
- smokeintensity
- Average number of cigarettes per day at baseline (numeric).
- smokeyrs
- Number of years smoked at baseline (numeric).
- exercise
- Recreational exercise: factor with levels 0 (much), 1 (moderate), 2 (little or none).
- active
- Activity level at usual occupation: factor with levels 0 (very active), 1 (moderately active), 2 (inactive).
- wt71
- Body weight at the 1971 baseline interview in kg (numeric).
Source
Assembled from causaldata::nhefs (Huntington-Klein 2021) by the data-raw/nhefs_surv.R script. The NHEFS data are publicly distributed by Hernán & Robins at https://www.hsph.harvard.edu/miguel-hernan/causal-inference-book/.
References
Hernán MA, Robins JM (2025). Causal Inference: What If. Chapman & Hall / CRC. Chapter 17: Causal survival analysis.
Huntington-Klein N et al. (2021). causaldata: Example Datasets for Causal Inference Textbooks. R package.
See Also
surv_fit(), contrast()