Assessment and prediction of affective disorders in patients after cerebral stroke using modern measurement scales

Abstract

Objective — to analyze the results of scale‑based assessments of post‑stroke depression (PSD) and post‑stroke anxiety disorders (PSAD) in different phases of cerebral stroke (CS) as well as to determine independent predictors of PSD at discharge from the Stroke Center (StC), and to evaluate the characteristics of the respective predictive models. Methods and subjects. Two hundred patients, including 92 (46.0 %) women and 108 (54.0 %) men with the median age of 65.6 years (IQR 58.2 — 75.1) were enrolled. The health status of all patients was assessed after hospitalization using the National Institutes of Health Stroke Scale (NIHSS), Barthel Index, Modified Rankin Scale, Mini‑Mental State Examination (MMSE), and Montreal Cognitive Assessment (MoCA). 172 (86.0 %) patients were diagnosed with ischemic stroke (IS), 28 (14.0 %) — intracerebral hemorrhage. Among patients with IS, 58 (33.7 %) had an atherothrombotic subtype, 85 (49.4 %) had a cardioembolic subtype, 16 (9.3 %) had a lacunar subtype, 13 (7.6 %) had another or unknown subtype. The PSD and PSAD were assessed using the Hospital Anxiety and Depression Scale (HADS) and the Patient Health Questionnaire (PHQ‑9) before discharge from the Stroke Center. The impact of factors was assessed by odds ratio (OR) and its 95 % confidence interval (95 % CI). The method of constructing and analyzing logistic regression models was used to determine independent predictors of PSD at discharge. Results. The baseline NIHSS score ranged from 1 to 29. The mRS scores upon admission were from 1 to 5, and the BI scores from 0 to 100. Forty‑one (20.5 %) patients were admitted in the hyperacute period, 55 (27.5 %) in the acute period, 68 (34.0 %) in the early subacute period, 13 (6.5 %) in the late subacute period, and 23 (11.5 %) in the chronic phase of stroke. The HADS‑D score ranged from 0 to 18, and the HADS‑A score from 0 to 15. PHQ‑9 scores ranged from 0 to 21. Based on the HADS score, 19 (9.5 %) of the patients had clinically significant PSD and 16 (8.0 %) of the participants had clinically significant PSAD. According to the total HADS score, 22 (11.0 %) of the patients had clinically significant affective disorders. With PHQ‑9 showed that clinically significant PSD was detected in 45 (22.5 %) patients. The HADS and PHQ‑9 scores had a strong positive significant correlation, but neither of them correlated with the age or sex of the patients, the subtype or severity of CS. However, univariate analysis showed that the risk of clinically significant PSD at discharge (according to HADS‑D) was significantly directly related to age and atrial fibrillation in addition to inverse relationship with the BI, MMSE and MoCA scores, LA subtype of IS and ICH. The risk of moderate to severe PSD (according to PHQ‑9) had a statistically significant direct corelation with the initial NIHSS score, as well as an inverse corelation with the baseline BI, MMSE, and MoCA scores. In multivariate analysis, 4 features were independently associated with PSD (HADS‑D > 10) at discharge: initial MMSE score (OR 0.93; 95 % CI 0.88 — 0.98, on average, for each additional point, p = 0.006), arterial hypertension (OR 8.5; 95 % CI 0.9 — 76.3; p = 0.057) or obesity (OR 0.23; 95 % CI 0.05 — 1.14; p = 0.072) as well as hospitalization after 30 days from CS onset. The predictive model based on these 4 variables had excellent sensitivity (94.7 %) and satisfactory specificity (73.3 %) and could assess the risk of developing PSD with good accuracy (AUC = 0.847). Furthermore, three factors were independent predictors of moderate or severe PSD (PHQ‑9 > 9) at discharge: age (OR 1.04; 95 % CI 1.00 — 1.08, on average, for each additional year, p = 0.028), the baseline MoCA score (OR 0.94; 95 % CI 0.91 — 0.98, on average, with an increase in the score for each additional point, p = 0.005) and UN subtype of IS. The prognostic model based on the latter 3 variables had satisfactory sensitivity (65.1 %) and specificity (75.5 %), but good accuracy of PSD prediction (AUC = 0.735). Conclusions. The HADS and PHQ‑9 scores in CS patients varied widely, and indicated high prevalence of clinically significant PSD and PSAD. HADS and PHQ‑9 scores correlated with each other, but not with age, sex, subtype, or severity of stroke. Elderly patients with significant cognitive impairment on admission were at a higher risk of affective disorders. The prognostic models allow accurate PSD prediction, which can contribute to the timely detection and initiation of PSD treatment in patients at risk.