Skip to main content
. Author manuscript; available in PMC: 2017 Apr 1.
Published in final edited form as: J Intern Med. 2015 Dec 20;279(4):376–387. doi: 10.1111/joim.12453

Fig. 1.

Fig. 1

Non-linear dose–response relationship between alcohol consumption and subsequent physical component summary (PCS) scores. Data are derived from a restricted cubic spline generalized estimating equation model (lifestyle-adjusted model) in alcohol consumers. The model is adjusted for previous PCS score, age, physical activity, energy intake, body mass index, smoking, marital status, employment status, night shift work, race, region, living arrangement, and parity. Alcohol consumption is winsorized at the 99.5th percentile. The 95% confidence interval is indicated by the gray area. P < 0.001, test for non-linear relationship.