Assessing delayed effects of a multi-site system intervention for homeless persons with serious mental illness.

ACCESS demonstration sites were followed for an additional two years beyond the scheduled four-year evaluation to assess whether any delayed effects had occurred in system and project integration. For system integration, findings indicate that there was a sharp increase between Wave 3 (1998) and Wave 4 (2000), but experimental and comparison sites had identical trends. For project integration, experimental sites at Wave 4 sustained the high level of integration achieved at Wave 3, but the comparison sites achieved the same level as the experimental sites at Wave 4, through an abrupt increase in their scores. The absence of delayed effects is likely due to diffusion of the interventions to comparison sites both in the latter stages of the demonstration and immediately afterwards. Further, aggressive lobbying on the part of ACCESS program managers to generate local and state support to sustain their services following the termination of federal funding, had an integrating effect thereby creating linkages among comparison site agencies. Implications of these findings for policy and further research are highlighted.