Studies of nursing home care in patients (aged 55 years or older) suffering from somatic illness or dementia combined with psychiatric disorders or severe behavioural problems were eligible for inclusion. Eligible inpatient interventions had to combine psychiatric care and nursing home care. Studies had to report quantitative data on the effects of the intervention.
All included studies assessed the effects of comprehensive interventions on severe problem behaviour in double care demanding nursing home patients. To be included in the intervention programmes, the patients' problem behaviour had to be threatening and require continuous observation and be unable to be successfully treated within the nursing home or on an outpatient basis. Most studies included patients with dementia, although some studies included patients with a somatic condition, as well as patients with dementia. On average, patients had three comorbid conditions, most commonly diabetes, cerebrovascular diseases and cardiovascular diseases; their mean age ranged from 70.6 to 82.9 years. The proportion of women participants ranged from 33 to 86%; one study was restricted to men.
Most interventions were administered in nursing homes, although three interventions took place in psychiatric or other hospitals. All of the included studies used individualised treatment plans with integrated tailored psycho-social nursing, medical and pharmacological interventions. Disciplines involved in the interventions included combinations of the following: (geriatric) psychiatrist, (Giro) psychologist, certified (psychiatric) nurses, activity therapist, social worker, neurologist, physician (geriatrician, internist, general), certified behaviour technician, registrar in psycho-geriatrics, and a pharmacist. Most studies were conducted in the USA; two studies were conducted in Australia and one in the UK.
Two reviewers independently assessed studies for inclusion, disagreements were resolved through discussion. One reviewer checked the final batch of papers with a second reviewer checking a random sample.