Background
Self-poisoning is one of the most common methods of suicide worldwide. The intentional ingestion of pesticides is the main contributor to such deaths and in many parts of rural Asia pesticide self-poisoning is a major public health problem. To inform the development of preventive measures in these settings, this study investigates small-area variation in self-poisoning incidence and its association with area-based socioeconomic and agricultural factors.

Methods
Ecological analysis of intentional self-poisoning in a rural area (population 267,613) of Sri Lanka in 2002. The geographic distribution of cases was mapped to place of residence. Using administrative division (GN), median population size 1416, as unit of analysis, associations with socioeconomic and agricultural indicators were explored using negative binomial regression models.

Results
The overall incidence of intentional self-poisoning in the study area was 315 per 100,000 (range: 0 – 2168 per 100,000 across GNs). Socioeconomic disadvantage, as indexed by poor housing quality (p = 0.003) and low levels of education (p < 0.001) but not unemployment (p = 0.147), was associated with a low self-poisoning incidence. Areas where a high proportion of the population worked in agriculture had low overall levels of self-poisoning (p = 0.002), but a greater proportion of episodes in these areas involved pesticides (p = 0.01). An association with extent of cultivated land was found only for non-pesticide poisoning (p = 0.01).

Conclusion
Considerable small-area variation in incidence rates of intentional self-poisoning was found. The noteworthy concentration of cases in certain areas and the inverse association with socioeconomic deprivation merit attention and should be investigated using individual-level exposure data.