Health-Related Quality of Life Among Survivors of Intracerebral Hemorrhage: A Hypothesis-Generating Study of One Year Recovery Trajectories and Health Care Decision Making

Background
Intensive Care Unit clinician-family decision making regarding prognosis and limitation of care occurs early and often in intracerebral hemorrhage (ICH). Recovery in ICH is prolonged and unpredictable, resulting in major challenges to estimating short-term mortality and long-term health-related quality of life (HRQoL). With goals of care decisions centering on prognostic estimates in ICH, further work is needed to examine the influence of early prognostication on withdrawal of life-sustaining treatment (WLST) decisions, and survivor perspectives of their long-term recovery trajectory after severe ICH. Using trial data from Clot Lysis: Evaluating Accelerated Resolution of Intraventricular Hemorrhage (CLEAR III) and Minimally Invasive Surgery Plus Alteplase in Intracerebral Hemorrhage Evacuation (MISTIE III), the purpose of this dissertation is to examine: (1) the demographic and disease-severity predictors of WLST following ICH; (2) the HRQoL of ICH survivors with dichotomized good vs. poor functional outcome over time; and (3) the long-term disposition and HRQoL outcomes of ICH survivors with similar baseline demographic and clinical characteristics of patients who had WLST.

Methods
Aim 1: A retrospective analysis compared no WLST patients (n=861) to WLST patients (n=118) enrolled in the CLEAR III and MISTIE III trials. Multivariable logistic regression was used to estimate the influence of established demographic and clinical severity of disease factors associated with good and poor outcomes, such as age, Glasgow Coma Scale (GCS), ICH location (deep or lobar), and total blood burden, defined by adding ICH and Intraventricular Hemorrhage (IVH) volumes (mL), at stability for WLST patients. Receiver operating characteristic (ROC) curves were used to evaluate the discriminatory power of the predictors of WLST and poor functional recovery of ICH patients at day 365. Rationale for WLST was assessed via standardized questions after goals of care discussions.

Aim 2: A retrospective analysis compared the EQ-5D-3L dimensions and EQ-5D visual analog scale (EQ-VAS) scores of ICH survivors (n=732) enrolled in the CLEAR III and MISTIE III trials with dichotomized “good” (modified Rankin Scale [mRS] 0–3) vs. “poor” (mRS 4–5) functional outcome at days 30, 180, and 365. The EQ-5D-3L dimensions were dichotomized by "no problems" vs. "any problems" and evaluated the percentage of participants by dichotomized mRS at days 30, 180, and 365. The proportion of ICH survivors by dichotomized mRS was evaluated at days 30, 180, and 365.

Aim 3: A matched cohort analysis using a modified severity index compared ICH survivors (n=379) enrolled in MISTIE III to patients who had WLST (n=61) after the first 72 hours. Patient disposition and EQ-VAS of matched survivors were evaluated at days 30, 180, and 365. The mean EQ-VAS at day 365 was compared to the mean EQ-VAS US population norm for persons aged 45–75 years. The rationale for WLST was examined via standardized questions after goals of care discussions.

Results
Aim 1: Of 979 participants, 118 (12%) had WLST. Nearly 73% had WLST performed within the first 30 days following diagnosis of ICH/IVH, with the highest number of WLST cases occurring within the first two weeks following initial presentation. Older age, lower GCS and greater total blood burden were significantly associated with WLST. For every year increase in age, there was a four percent increase in the odds of having WLST performed. Patients with GCS 9-12 were two times more likely to have WLST performed than patients with GCS 13-15. Patients with GCS 3-8 were four times more likely to have WLST performed than patients with GCS 13-15. Patients with total blood burden of > 55 mL were nearly five times more likely to have WLST performed. The area under the ROC curve of 79% indicated that the severity factors were moderately effective in distinguishing between patients who had WLST and those who did not. The severity factors identified 82% of patients with poor functional recovery at day 365. An anticipated dependent outcome was attributed to the rationale for WLST in 62% of WLST cases.

Aim 2: Of 732 survivors, 607 survivors or their proxies completed the EQ-5D-3L dimensions and 557 survivors had EQ-VAS data at all timepoints. At day 30, 80.6% of survivors had a mRS of 4-5. By one year, 61.6% of survivors had a mRS of 0-3. Survivors with mRS 0-3 and mRS 4-5 showed significant differences at days 30, 180, and 365 in all five EQ-5D-3L dimensions, where a significantly higher percentage of survivors with mRS 4-5 reported having “any problems” compared to survivors with mRS 0-3. Survivors with mRS 4-5 reported the highest percentage of “any problems” with mobility at days 30 (99.1%) and 365 (98.3%), usual activities (99.6%) at day 180; and the lowest percentage of “any problems” related to anxiety/depression at days 30 (55.9%), 180 (53.3%), and 365 (56.3%). Proxies of survivors with mRS 4-5 had a significantly higher percentage of those who reported “any problems” with mobility and self-care at days 30, 180, and 365, and with usual activities at days 180 and 365 compared to proxies of survivors with mRS 0-3. EQ-VAS of the survivors increased within the first six months, but not significantly beyond six months to one year. Of the survivors with severe disability in the acute phase, almost 60% achieved functional independence and reported EQ-VAS that approached the EQ-VAS US population norm by one year.

Aim 3: Of 379 survivors in MISTIE III at one year, 90 were matched to patients who had WLST (n=61). Of the 90 matched survivors, 11.1%, 65.6%, and 73.3% returned home by days 30, 180, and 365. The mean (SD) EQ-VAS of matched survivors at days 30, 180, and 365 was 41.9 (24), 62.2 (20.8), and 65.6 (21.8). At day 365, matched survivors living at home (n=66) had mean (SD) EQ-VAS of 70.6 (18.9) as compared to the mean EQ-VAS US population norm of 75. The rationale recorded for WLST was an anticipated dependent state for 38 (62%) of the 61 WLST patients.

Conclusion
These results suggest that the same disease-severity predictors of poor outcome, first described in the late 1980s, continue to influence the decision to withdraw life-sustaining treatment in the critically ill ICH patient population. Yet, the survivors of ICH with severe disability at 30 days following ICH diagnosis demonstrated a significant trend toward functional independence and HRQoL improvement at one year, and nearly half of those with severe disability at one year, reported no problems with anxiety/depression and pain/discomfort. The HRQoL of ICH survivors with clinical and demographic characteristics similar to those who had WLST, specifically those living at home, approached the US population norm of HRQoL for age-matched persons at one year. Consequently, early prognostication of pessimistic outcomes does not appear to match the potential for acceptable outcomes of ICH survivors. The results of this dissertation study challenge the current goals of care, decision-making practices of early identification of poor ICH outcomes as the solely determinative element of prognosis for the purpose of WLST.