To examine predictors of suicidal behavior (SB) in adults aged 75 years and above with dementia.

Longitudinal national register-based study.

Swedish residents aged ≥75 years with dementia identified in the Swedish Dementia Registry (SveDem) between 1 January 2007 and 31 December 2017 (N = 59 042) and followed until 31 December 2018. Data were linked with numerous national registers using personal identity numbers.

Outcomes were nonfatal self-harm and suicide. Fine and Gray regression models were used to investigate demographics, comorbidities, and psychoactive medications associated with fatal and nonfatal SB.

Suicidal behavior was observed in 160 persons after dementia diagnosis; 29 of these died by suicide. Adjusted sub-hazard ratio (aSHRs) for SB was increased in those who had a previous episode of self-harm (aSHR = 14.42; 95% confidence interval [CI] = 7.06–29.46), those with serious depression (aSHR = 4.33, 95%CI = 2.94–6.4), and in those born outside Sweden (aSHR = 1.53; 95% CI = 1.03–2.27). Use of hypnotics or anxiolytics was also associated with a higher risk of SB; use of antidepressants was not. Milder dementia and higher frailty score also increased risk of SB. Risk was decreased in those who received home care (aSHR = 0.52; 95%CI = 0.38–0.71) and in the oldest group (aSHR = 0.35; 95%CI = 0.25–0.49).

In addition to established targets for suicidal behavior prevention (improved identification and treatment of depression and previous self-harm), several new risk factors were suggested. There is a need for innovative public health strategies to meet the needs of older dementia patients with a foreign background. Home care may have a potential positive effect to prevent SB in people with dementia, but this needs to be further explored.