Quality of life and depression 3 months after intracerebral hemorrhage

Abstract Objectives Quality of life (QoL) after intracerebral hemorrhage (ICH) is poorly known. This study investigated factors affecting QoL and depression after spontaneous ICH. Materials and Methods This prospective study included patients admitted to Helsinki University Hospital between May 2014 and December 2016. Health‐related QoL (HRQoL) at 3 months after ICH was measured using the European Quality of Life Scale (EQ‐5D‐5L), and the 15D scale. Logistic regression analyses were used to test factors affecting HRQoL. EQ‐5D‐5L anxiety/depression dimension was used to analyze factors associated with anxiety/depression. Results Of 277 patients, 220 were alive, and sent QoL questionnaire. The questionnaire was returned by 124 patients. Nonreturners had more severe strokes with admission National Institutes of Health Stroke Scale (NIHSS) 7.8 (IQR 3.0–14.8) versus 5.0 (IQR 2.3–11.0); p = 0.018, and worse outcome assessed as modified Rankin Scale 3–5 at 3 months 59.4% versus 44.4% (p = 0.030). Predictors for lower HRQoL by both scales were higher NIHSS with OR 1.28 (95% CI 1.13–1.46) for EQ‐5D‐5L, and OR 1.28 (1.15–1.44) for 15D, older age (OR 1.10 [1.03–1.16], and OR 1.09 [1.03–1.15]), and chronic heart failure (OR 18.12 [1.73–189.27], and OR 12.84 [1.31–126.32]), respectively. Feeling sad/depressed for more than 2 weeks during the year prior to ICH was predictor for lower EQ‐5D‐5L (OR 10.64 [2.39–47.28]), and history of ICH for lower 15D utility indexes (OR 11.85 [1.01–138.90]). Prior feelings of sadness/depression were associated with depression/anxiety at 3 months after ICH with OR 3.62 (1.14–11.45). Conclusions In this cohort of ICH patients with milder deficits, HRQoL was affected by stroke severity, comorbidities and age. Feelings of depression before ICH had stronger influence on reporting depression/anxiety after ICH than stroke severity‐related and outcome parameters. Thus, simple questions on patient's premorbid feelings of sadness/depression could be used to identify patients at risk of depression after ICH for focusing follow‐up and treatment.


| INTRODUC TI ON
Intracerebral hemorrhage (ICH) is the second most common type of stroke (van Asch et al., 2010). As many as 40.4% of the patients die within the first month, and many of the survivors are left with disabilities (van Asch et al., 2010), having a major effect on the life of the patients and caregivers. Stroke increases risk of depression; depression is also a risk factor for stroke (Dong, 2012). ICH patients have been shown to have less favorable outcome compared to ischemic stroke, and they are left with more severe deficits (Sennfält, Norrving, Petersson, & Ullberg, 2018). In addition to risk for recurrent ICH, survivors are at risk for thrombotic events, such as ischemic stroke, raising concerns for the benefits and risks of secondary prevention, such as antiplatelet therapy (Poon, Fonville, & Salman, 2014;Wani, Nga, & Navaratnasingham, 2005). After ischemic stroke, there are many modifiable factors to prevent recurrence (Diener, Ntaios, O'Donnell, & Easton, 2018;Laloux, 2014). The most important etiological factor to control after ICH is hypertension; however, no ICH specific secondary preventive medications exist to date (Veltkamp & Purrucker, 2017). Seizures are more common after ICH than ischemic stroke (Christensen, Mayer, Ferran, & Kissela, 2009;Faught, Peters, Bartolucci, Moore, & Miller, 1989;Lahti et al., 2017). Such stroke sequelae and lack of effective secondary prevention can feel burdensome to the patient and caregivers, and affect their quality of life.
There are several studies on health-related quality of life (HRQoL) after stroke (Chang et al., 2016;Delcourt et al., 2010;Hackett, Duncan, Anderson, Broad, & Bonita, 2000), but few focus on ICH patients. Age, length of hospital stay, and motor function at discharge have been shown to predict HRQoL after stroke at 6 months (Chang et al., 2016). Another study reported baseline National Institutes of Health Stroke Scale (NIHSS) ≥14 versus <14 to have the strongest association with poor HRQoL 3 months after ICH. Also older age, use of antithrombotics, larger and deep ICH, intraventricular hemorrhage (IVH), early worsening of the neurological deficit, and completion of the questionnaire by proxy responder associated with worse HRQoL Faught et al., 1989;Lahti et al., 2017). Hence, these parameters are associated mostly with the severity of the index ICH.
In this study, we assessed the quality of life and depression after ICH in a prospective cohort, and addressed such factors affecting the subjective quality of life that are not included in traditional outcome measures.

| MATERIAL S AND ME THODS
This study is part of a prospective study on genetic and environmental risk factors of ICH. We aimed to recruit all consecutive patients admitted to Helsinki University Hospital, Finland, between May 2014 and December 2016 with findings of primarily appearing ICH on admission imaging. We subsequently excluded hemorrhages related to tumor, trauma, ischemic stroke, vascular malformations, and other structural abnormalities.
To participate, the patient or his/her proxy had to give an informed signed consent. If the patient was unable to write, the consent could be given orally in the presence of a witness unrelated to the study. We used structured questionnaires on profession, feelings of depression, smoking, and alcohol consumption before ICH.
The questionnaire was given upon recruitment in hospital, that is, in the acute/subacute phase, and if not returned, they were sent to the patient's home address. The return time of questionnaires were not registered systematically. Medical history was obtained by using the combination of questionnaires and medical records. Admission parameters such as blood pressure, NIHSS, and Glascow Coma Scale (GCS) were obtained from the electronic medical charts. Hematoma size was measured from first head CT after arrival, and the size was calculated using ABC/2 method, where A, B, and C are the largest perpendicular measures of the hematoma on CT images.
Living and working status, modified Rankin Scale (mRS), and Barthel Index (BI) were evaluated at 3 months (±1 week) after ICH by a combination of revisiting the electronic medical charts and telephone interview (HS). The patients were interviewed via telephone, or alternatively, if the corresponding information on the outcome parameters were recorded in their electronic medical charts, these were collected. If the patient was reached later, he/she was asked to recall the situation at 3 months. As the medical files in Finland are very comprehensive (including physiotherapists', occupational therapists', and neuropsychologists' evaluations), the information is found to a vast extent in the patients' files. At the same time point, the patients were sent 15D and EQ-5D-5L questionnaires. EQ-5D-5L is a validated measure developed by EuroQol Group (www.euroqol.org), defining health in terms of five dimensions: Mobility, Self-Care, Usual Activities, Pain/Discomfort, and Anxiety/Depression, each of them having five response levels (no problems, slight, moderate, severe, and extreme problems). The EQ-5D-5L utility score is an index, which combines the five dimensions into one score, which is calculated using population-based preference weights. One represents perfect health, 0 death, and negative values health states considered worse than death. As there was no population-based utility index dataset for EQ-5D-5L for the Finnish population, the utility index was calculated using a crosswalk dataset from the EQ-5D-3L, provided by the EuroQol Research Foundation. 15D is a 15-dimension (Mobility, Vision, Hearing, Breathing, Sleeping, Eating, Speech, TA B L E 1 Comparison of the patients with low and high EQ-5D-5L utility index (n = 117) a  Baseline ICH volume, mL 10.9 (4.5-23.0) 7.  The patients were divided into two utility index groups using the median utility indexes (low or high). For EQ-5D-5L, the HRQoL was considered low, if the index was ≤0.590 (median), and high, if the index was >0.590, and for 15D ≤0.817 and >0.817, accordingly. For the EQ-5D-5L depression/anxiety question, the patients were divided into two groups: reporting symptoms of depression/ anxiety or not. We compared the baseline and outcome characteristics between each two groups using univariate regression analysis.
Variables with p < 0.2 in the univariate regression model were chosen for the multivariable binary regression model for HRQoL, and variables with p < 0.1 for depression/anxiety. For the multivariable models on EQ-5D-5L and 15D utility indexes, we tested associations of baseline characteristics with the quality of life, and thus the outcome parameters were not included in the analyses. We also compared the characteristics of the patient groups with and without (missing) data on quality of life to test for selection bias.
Statistical analyses were made using SPSS v.24 (IBM, Armonk, NY); the multivariable analysis for depression/anxiety was made using R Package "brglm" to be able to handle the situation with no  Other parameters showed no significant differences.

| RE SULTS
In the general Finnish population, the EQ-5D index values range from 0.909 (age group 25-34) to 0.583 (age group 75+), the average value being 0.800 (Self-Reported Population Health: An International Perspective based on EQ-5D, 2014). Median EQ-5D-5L utility index was 0.590 in the present ICH cohort. For 7 (5.6%) patients the index could not be calculated due to incomplete answers. Median 15D utility index was 0.817. For 15D scores there were 16 (12.9%) patients with incomplete data. Supplemental Figure S1 shows the distribution of the EQ-5D-5L and 15D utility index values. Note. Data are n (%), mean (SE), or median (IQR). mRS: modified Rankin Scale; ICH: intracerebral hemorrhage; COPD: chronic obstructive pulmonary disease; NIHSS: the National Institutes of Health Stroke Scale; GCS: Glasgow Coma Scale; DNR: do not resuscitate. a n = 7 exclusion due to incomplete answers. b Missing data for sad or depressed for more than 2 weeks in 17 patients (14.5%), alcohol use in three patients (2.6%), no or little stress at home in 11 patients (9.4%), marital status in three patients (2.6%), high education in four patients (3.4%).
The bold values indicates P < 0.05.

TA B L E 1 (Continued)
TA B L E 2 Comparison of the patients with low and high 15D score (n = 108) a Supplemental Figure S2 shows the EQ-5D-5L and 15D utility indexes in different mRS groups.
In multivariable regression analyses, chronic heart failure, higher NIHSS on arrival, and older age were predictors for lower HRQoL in both EQ-5D-5L and 15D. Additionally, feeling sad or depressed for more than 2 weeks during the year prior to ICH was an independent predictor for lower EQ-5D-5L utility index, and a history of any ICH for lower 15D score ( As seen in Table 4, feelings of depression/anxiety at 3 months after ICH were more common among patients with mRS 3-5 at 3 months after ICH, patients with dementia before ICH, patients with DNR orders during hospital stay, and patients having been feeling sad or depressed during the year prior to ICH. However, in multivariable analysis, the only independent predictor for feelings of depression and/or anxiety at 3 months after ICH was patient reported depression in the year prior to ICH (Table 5). Only four (3.7%) patients had been treated with antidepressants, and 1 (0.9%) patient with psychotherapy.

| D ISCUSS I ON
As previously shown Delcourt et al., 2017), and also verified in our study, baseline stroke severity measured by NIHSS and ICH volume have a strong impact on the HRQoL after ICH. Of the comorbidities, asthma/COPD and chronic heart failure were associated with worse HRQoL, as earlier studies also imply (Chung & Han, 2017;Iqbal, Francis, Reid, Murray, & Denvir, 2010). As the consent given by proxy-the patient not being able to consent him/herself-indicates more severe stroke, it Note. NIHSS: the National Institutes of Health Stroke Scale; ICH: intracerebral hemorrhage. a 95 patients included in analysis due to missing data (EQ-5D-5L); 104 patients included in analysis due to missing data (15D). b Adjusted for: agreement by proxy, mRS 3-5 prior to ICH, COPD/asthma, coronary artery disease, no or little stress at home, living alone, high education, GCS on admission, baseline ICH volume, ICH location, DNR orders in hospital. c Adjusted for: agreement by proxy, antihypertensive medication, living alone, occupational status, high education, GCS on admission, baseline ICH volume, any neurosurgery, DNR orders in hospital.
TA B L E 3 Multivariable binary logistic regression analysis for predictors of decreased a) EQ-5D-5L utility index, and b) 15D score at 3 months a TA B L E 4 Comparison of the patients with reported depression and/or anxiety in EQ-5D-5L 3 months after ICH (n = 122) a is not surprising that it associates with worse HRQoL. As also seen in general population, QoL has an inverse association with older age (Self-Reported Population Health: An International Perspective based on EQ-5D, 2014). Measured by EQ-5D-5L, patients with history of cancer reported better HRQoL. This is likely to be biased due to including and not separating also benign and curatively treated cancers.
Feelings of depression during the year before ICH were shown to be associated with mood problems also after ICH. This is in line with a study on subarachnoid hemorrhage (SAH) patients, where pre-SAH depression and premorbid psychosocial stress were associated with depression at 3 months (Kreiter et al., 2013). Depression is known to be common among dementia patients (Meyers, 1998) Working at 3 months 2 (2.9) 2 (3.7) 1.000 Note. Data are n (%), mean (SE), or median (IQR). mRS: modified Rankin Scale; ICH: intracerebral hemorrhage; COPD: chronic obstructive pulmonary disease; NIHSS: the National Institutes of Health Stroke Scale; GCS: Glasgow Coma Scale; DNR: do not resuscitate. a n = 2 exclusion due to incomplete answers. b Missing data for BMI in five patients (4.1%), sad or depressed for more than 2 weeks in 17 patients (13.9%), alcohol use in three patients (2.5%), no or little stress at home in 11 patients (9.0%), marital status in three patients (2.5%), high education in four patients (3.3%), EQ-5D index in five patients (4.1%), 15D index in 14 patients (11.5%).
The bold values indicate P < 0.05. TA B L E 5 Multivariable binary logistic regression analysis for associations with depression/anxiety in EQ-5D-5L at 3 months (n = 122) a individual level, it is not surprising, that the models generated were able to explain only 50% of HRQoL. As the past feelings prior to index-ICH are questioned, it cannot be excluded, that depressive feelings at the time of evaluation affect the results, even though the patients are asked to recall the situation before ICH. The patients were sent the QoL questionnaires at 3 months, and even though the exact time point was not recorded, the analysis represents situation after the acute and subacute phase of the illness. The response rate among patients alive was 56.4%. Taken into consideration the nature of the illness this can be considered quite good, as many of the survivors are not expected to be able to answer due to their neurologic deficits. Also, as expected, the patients not participating in QoL reporting had more severe deficits, and thus our study represents HRQoL of patients with milder deficits. The response rate is in line with survey response rates published in medical journals (Asch et al., 1997).
In conclusion, higher age and parameters that associate with larger strokes, as shown also previously affect the quality of life after ICH. Of comorbidities, chronic heart failure had the strongest association with poor quality of life. Patients reporting feelings of sadness or depression before ICH reported more frequently anxiety and/or depression also after ICH. That had stronger influence on development of depression and/or anxiety than stroke severity-related and outcome parameters. Our study suggests that a questionnaire on premorbid self-reported sadness or depression could help to identify patients at risk of depression and/or anxiety after ICH. Obviously, larger prospective studies are needed to confirm our results.

ACK N OWLED G M ENTS
None.

CO N FLI C T O F I NTE R E S T S
T.S. has received grants from Orion Corporation (Finland), TEVA Pharmaceutical Industries Ltd., Allergan plc, personal fees from Boehringer Ingelheim Inc., outside the submitted work. D.S. and H.S.
report no conflicts of interests.