Hospital presentations for self-poisoning during COVID-19 in Sri Lanka: an interrupted time-series analysis

This paper examines the impact of the COVID-19 pandemic on cases of self-poisoning in low and middle income countries.

Authors: Duleeka Knipe, Tharuka Silva, Azra Aroos, Lalith Senarathna, Nirosha Madhuwanthi Hettiarachchi, Sampath R Galappaththi, Matthew J Spittal, David Gunnell, Chris Metcalfe, Thilini Rajapakse
Published in: The Lancet



There is widespread concern over the impact of public health measures, such as lockdowns, associated with COVID-19 on mental health, including suicide. High-quality evidence from low-income and middle-income countries, where the burden of suicide and self-harm is greatest, is scarce. We aimed to determine the effect of the pandemic on hospital presentations for self-poisoning.


In this interrupted time-series analysis, we established a new self-poisoning register at the tertiary care Teaching Hospital Peradeniya in Sri Lanka, a lower-middle-income country. Using a standard extraction sheet, data were gathered for all patients admitted to the Toxicology Unit with self-poisoning between Jan 1, 2019, and Aug 31, 2020. Only patients classified by the treating clinician as having intentionally self-poisoned were included. Data on date of admission, age or date of birth, sex, and poisoning method were collected. No data on ethnicity were available. We used interrupted time-series analysis to calculate weekly hospital admissions for self-poisoning before (Jan 1, 2019–March 19, 2020) and during (March 20–Aug 31, 2020) the pandemic, overall and by age (age <25 years vs ≥25 years) and sex. Individuals with missing date of admission were excluded from the main analysis.


Between Jan 1, 2019, and Aug 31, 2020, 1401 individuals (584 [41·7%] males, 761 [54·3%] females, and 56 [4·0%] of unknown sex) presented to the hospital with self-poisoning and had date of admission data. A 32% (95% CI 12–48) reduction in hospital presentations for self-poisoning in the pandemic period compared with pre-pandemic trends was observed (rate ratio 0·68, 95% CI 0·52–0·88; p=0·0032). We found no evidence that the impact of the pandemic differed by sex (rate ratio 0·64, 95% CI 0·44–0·94, for females vs 0·85, 0·57–1·26, for males; pinteraction=0·43) or age (0·64, 0·44–0·93, for patients aged <25 years vs 0·81, 0·57–1·16, for patients aged ≥25 years; pinteraction=0·077).


This is the first study from a lower-middle-income country to estimate the impact of the pandemic on self-harm (non-fatal) accounting for underlying trends. If the fall in hospital presentations during the pandemic reflects a reduction in the medical treatment of people who have self-poisoned, rather than a true fall in incidence, then public health messages should emphasise the importance of seeking help early.