Among 780 patients admitted to the PCU, 35.5% were excluded since they did not satisfy the inclusion criteria; one-third were excluded because of delirium in progress.
Table 1 shows the general characteristics of 503 patients enrolled in the study at the moment of admission to PCU. After a median follow-up time of 16 days (interquartile range, IQR, 6–40), 95 (18.9%) patients developed delirium. The characteristics of 95 patients who developed delirium and 408 patients who did not develop it are listed separately. Fifty-six percent of patients were male, mean age was 76 years; 49.8% of them had primary education or less, 54.7% of patients were married, 90.3% had a diagnosis of cancer (of whom about 87% metastasized). Over 64% of patients were initially cared for at home and 35.8% in hospice. The distribution of characteristics was similar in patients who did and did not develop delirium, although significant differences were observed in relation to age and setting of care. Patients who developed delirium were on average almost 3 years older than those who did not develop it (mean age 78.2 and 75.4, respectively), and were less frequently treated at home (49.5% and 67.6%, respectively). Median survival time was 9 days (IQR 2–22) in patients with delirium and 19 days (IQR 8–42) in those without delirium (data not shown).
Table 2 presents the distribution of comorbidities included in the CIRS and the KPS, overall and according to the presence of delirium. Prevalence of comorbidities was not significantly different between patients with and without delirium; moreover, no significant difference was found according to levels of the CIRS score and KPS, although values of CIRS ≥ 8 were found more frequently in patients who developed delirium (20.0%) than in those who did not develop it (13.5%), and general conditions were more severe in patients with delirium than in those without delirium (KPS ≤ 30 in 33.7% and in 24.5% of patients, respectively).
The prevalence of clinical factors in all patients and in the two sub-groups of patients who developed and did not develop delirium is given in Table 3. No significant differences were found for most clinical factors; however, the presence of hypoxia and the total number of simultaneously present clinical factors were significantly more frequent in patients who developed delirium than in those who did not develop it (24.2% versus 14.7% respectively with hypoxia, and 58.9% and 47.5% respectively with at least two clinical factors). Only 17.7% of patients (12.6% of those with delirium and 18.9% of those without delirium) had no clinical factors (data not shown).
In relation to symptoms, the presence of breathlessness and poor well-being was significantly higher in patients who developed delirium (79.0% and 63.2%, respectively) than in those who did not develop it (64.5% and 46.1%; Table 4). Conversely, for other symptoms, such as pain, fatigue, anxiety, and depression the prevalence was similar in patients with and without delirium.
The relationship between the severity of symptoms (measured by ESAS) and risk of developing delirium is shown in Table 5. For most symptoms, the severity was similar in patients who developed and in those who did not develop delirium. Only for drowsiness, poor well-being, and breathlessness, was the presence of moderate/severe degree symptoms higher in the former (17.9%, 26.3%, and 17.9%, respectively) than the latter group (9.6%, 18.4%, and 12.0%, respectively).
Table 6 shows the distribution of the main classes of drugs prescribed as “around the clock” therapy in all patients, and separately according to the presence of delirium. Use of haloperidol and other drugs acting on the central nervous systems (CNS; tricyclic and SSRI antidepressants, antiepileptics, antiparkinsonians, antipsychotics, barbiturates, and benzodiazepines) was more frequent in patients who developed delirium (24.2% and 31.6%, respectively) than in those who did not develop it (14.5% and 26.2%, respectively). For other drugs considered, the prevalence of use was similar in the two groups of patients.
The univariate and multivariate analyses of the 18 factors with a p value < 0.1 in univariate analysis are shown in Table 7. Factors that were significantly related to delirium in univariate analyses were care in hospice, compromised performance status, kidney disease, fever, renal failure, hypoxia, dehydration, drowsiness, poor well-being, breathlessness, “around the clock” treatment with haloperidol and other drugs acting on the CNS, cardiovascular drugs, anticoagulants, gastroprotective drugs, and morphine. After adjustment for each of these factors, setting of care, presence of breathlessness, and administration of CNS active drugs, particularly haloperidol were significantly associated with the development of delirium: the HR was 2.28 for hospice versus home care (95% CI 1.45–3.60, p < 0.001), 1.71 for presence versus no presence of breathlessness (95% CI 1.03–2.831.74), and 2.17 for haloperidol administration versus no administration of any CNS drugs (95% CI 1.11–4.22, p = 0.0248).