RESUMEN
<p><b>OBJECTIVE</b>To investigate the prevalence and natural outcome of late-life depression in the community and to analyze the risk-prediction models.</p><p><b>METHODS</b>A community in Hang Zhou was selected as a trial. A total of 1 275 persons aged 60 or more in this community were screened by PHQ-9 questionnaire; SCID was used for interviewer to diagnostic interview the people whose PHQ-9 was more than 10 points, 50 % of those whose PHQ-9 was from 5 to 9 points and 5 % of those whose PHQ-9 was less than 5 points, then all those who accepted diagnostically interview were interviewed by PHQ-9 every 3 months in one year, and were diagnostic interviewed by SCID in the last month. Logistic regression analysis was used to explore depressive risk factors in 12 months.</p><p><b>RESULTS</b>There were 141 (11.1%) persons whose PHQ-9 score was more than 10 points, 298 (23.4%) whose PHQ-9 score were 5-9 points, and 836 (65.5%) whose PHQ-9 score were 0 to 4 points in the preliminary survey, 93 were major depressive disorder (MDD). The prevalence of late-life depression was 7.3%. Compared with the PHQ-9 score in one year, 17.6% of those with no depressive symptoms emerged depression; 50% of those who had depressive symptoms declined, 9% developed to significant depressive symptoms, and 41% did not change; 12% of those with significant depressive symptoms were found no depression, 24% reduced, and 64% still had depression. The significant predictors were the accumulation of disease, social support, educational level, daily capacity and baseline of depression.</p><p><b>CONCLUSION</b>The prevalence of late-life depression was high. The rates of recognition, diagnosis and treatment were low. The natural outcome after a year did not relieve apparently. Specialist-community health partnership management model is one of the important ways to prevent and treat late-life depression.</p>