Fig 2. Population-average and province-specific segmented linear trends in hospital admission rate ratios for chronic obstructive pulmonary disease and asthma in selected Spanish provinces, 2003–2012.
Province-specific segmented linear trends in hospital admission rate ratios (gray lines) were obtained from overdispersed Poisson additive models with distinct linear segments within the 2003–2005 pre-ban, 2006–2010 partial ban, and 2011–2012 comprehensive ban periods, and adjusted for seasonality, day of the week, temperature, influenza epidemics, acute respiratory infections, and pollen counts for asthma-related admissions (only available in Asturias, Barcelona, Madrid, Murcia, and Zaragoza). Pooled segmented linear trends (bold black lines) were obtained from random-effects multivariate meta-analyses on province-specific estimates of segmented regression coefficients. The reference time point (rate ratio = 1) was set at the midpoint pre-ban period (July 1st, 2004). Vertical dashed lines represent the dates at which the partial and comprehensive smoking bans entered into force on January 1st, 2006, and January 2nd, 2011, respectively.