Marcel Slockers

65 The effects of the Action Plan for Social Support on the health of homeless people ( P = 0.96). Distinguishing three periods (before 2003, 2003–08 and 2008) did not show a significant difference in mortality between the periods ( P = 0.07). Using 10, 1-year periods showed that the mortality rates fluctuated from year to year, but also in this analysis there was no significant effect ( P = 0.61). Table 2. Mortality rate ratios comparing mortality within cohort of homeless, Rotterdam, 2001–10 Time Rate ratio (95% confidence interval), adjusted for age and sex P value Main analyses: 2006–10 vs. 2001–05 (=ref) 0.97 2001–05 (reference) 1.00 2006–10 1.05 (0.79–1.28) 3 periods 0.71 2001–02 (reference) 1 2003–08 0.91 (0.65–1.28) 2009–10 1.01 (0.68–1.51) 10 periods 0.61 2001 1.00 2002 0.88 (0.47–1.66) 2003 0.83 (0.44–1.57) 2004 1.10 (0.60–2.02) 2005 0.65 (0.34–1.27) 2006 0.67 (0.34–1.29) 2007 1.03 (0.56–1.91) 2008 0.83 (0.44–1.57) 2009 0.88 (0.47–1.65) 2010 1.01 (0.54–1.88) Discussion Mortality among homeless men and women in Rotterdam was not lower in the period after 2006 than in the period 2001–05, despite important policy efforts that improved the living situation of homeless people in this city. Other recent studies covering the 21st century also suggest that the high mortality among homeless men and women is rather persistent. A Dutch study among socially marginal- ized, including homeless, persons found a 3 times higher risk of mortality in this group