Clinical Neurology and Neurosurgery 126 (2014) 24–29
Contents lists available at ScienceDirect
Clinical Neurology and Neurosurgery journal homepage: www.elsevier.com/locate/clineuro
Comparing clinician- and patient-reported outcome measures after hemicraniectomy for ischemic stroke Michael L. Kelly a,∗ , Benjamin P. Rosenbaum a , Varun R. Kshettry a , Robert J. Weil a,b,c a b c
Department of Neurosurgery, Neurological Institute, Cleveland Clinic, Cleveland, USA Rose Ella Burkhardt Brain Tumor and Neuro-Oncology Center, Cleveland Clinic, Cleveland, USA Department of Neurosurgery, Geisinger Health System, Danville, USA
a r t i c l e
i n f o
Article history: Received 29 March 2014 Received in revised form 26 July 2014 Accepted 2 August 2014 Available online 15 August 2014 Keywords: Hemicraniectomy Stroke Quality of life Outcomes Health status measures
a b s t r a c t Background: The association between clinician- and patient-reported health status measures (HSM) after hemicraniectomy for ischemic stroke is understudied. We compared HSMs to determine how HSM type and follow-up affect the interpretation of outcomes. Methods: We identiﬁed patients that underwent hemicraniectomy for ischemic stroke at the Cleveland Clinic (CC) from January 2009 through May 2013. HSMs were obtained from the CC Knowledge Program Data Registry. Outpatient follow-up was divided into “Early” (3 ± 2 months (standard deviation)) and “Late” (9 ± 3 months) time periods. Clinician-reported HSMs (National Institutes of Health Stroke Scale (NIHSS) and Modiﬁed Rankin Scale (mRS)) were compared to patient-reported HSMs (EuroQol quality of life index (EQ-5D), Patient Health Questionnaire-9 (PHQ-9), and the Stroke Impact Scale-16 (SIS-16)). Results: 11 of 32 patients completed all HSMs during both follow-up periods. Clinician-reported median NIHSS scores improved from 12 to 7 (p = 0.003). Median mRS scores demonstrated little improvement from 4 to 3 (p = 0.2). Patient-reported median EQ-5D scores improved from 0.33 to 0.69 (p = 0.03). Among EQ-5D sub-scores, “usual activity” improved from a median score of 3 (extreme problems) to 2 (some problems) (p = 0.008). Median PHQ-9 scores improved from 9 to 1 (p = 0.06) as did SIS-16 scores from 23 to 57 (p = 0.01). EQ-5D and mRS score differences between periods were correlated (r = −0.65, p = 0.03), but only the EQ-5D showed signiﬁcant improvement over time. Conclusions: Both HSM types, clinician- and patient-reported outcome measures, improved over time. The structure of clinical trials, and, in particular, deﬁning clinical endpoints and framing outcomes, has a profound impact on the interpretation of what a “favorable” outcome means. © 2014 Elsevier B.V. All rights reserved.
1. Introduction The debate over outcomes in hemicraniectomy for patients with ischemic stroke is ongoing. Three randomized controlled trials (RCTs) examined the beneﬁt of hemicraniectomy for patients with large ischemic middle cerebral artery (MCA) stroke [1–3]. A subsequent meta-analysis of the results from these trials showed that surgery reduced the mortality rate from 78% to 29% and increased the percentage of those with a “favorable” outcome (Modiﬁed Rankin Scale (mRS) ≤ 3) from 21% to 43% . Similar results were reported in the recent publication of DESTINY II (Decompressive Surgery for the Treatment of Malignant Infarction of the Middle
∗ Corresponding author at: Department of Neurosurgery, Neurological Institute, S-40, Cleveland Clinic, 9500 Euclid Avenue, Cleveland, OH 44195, USA. Tel.: +1 216 444 5539; fax: +1 216 636 0454. E-mail address: [email protected]
(M.L. Kelly). http://dx.doi.org/10.1016/j.clineuro.2014.08.007 0303-8467/© 2014 Elsevier B.V. All rights reserved.
Cerebral Artery trial II), which reported a beneﬁt in mortality and functional outcomes for patients over the age of 60 years . However, the dichotomization of a “favorable” versus “unfavorable” outcome in hemicraniectomy research is controversial . Critics argue that the distinction is arbitrary and fails to accommodate the patient’s perspective . Clinician-reported outcome measures traditionally used in stroke clinical trials, such as the mRS, Barthel Index (BI), and the National Institutes of Health Stroke Scale (NIHSS), do not include the patient’s experience of the disease, the treatment, or the outcome. Hospitals, insurance agencies, and researchers have begun to collect metrics related to patient-and clinician-reported outcome measures, also termed health status measures (HSMs) . In stroke patients, examples of clinician-reported HSMs include the mRS, BI, and the NIHSS [4,9]. Examples of patient-reported measures include the Euroqol quality of life index score (EQ-5D), Stroke Impact Scale-16 (SIS-16), and the Patient Health Questionnaire (PHQ-9) for depression [10–13]. Several observational studies in
M.L. Kelly et al. / Clinical Neurology and Neurosurgery 126 (2014) 24–29
hemicraniectomy for stroke have included these clinician- and patient-reported HSMs as well as others [14–16]. However, a recent systematic review of the hemicraniectomy literature demonstrated that no studies have directly compared patient- and clinicianreported HSMs in order to link clinical impressions to patient experience . Since 2009, the Cleveland Clinic Knowledge Program (KP) has been systematically collecting both clinician- and patientreported HSMs in ischemic stroke patients who have undergone hemicraniectomy . In this study, we compare clinician- and patient-reported outcome measures after hemicraniectomy for ischemic stroke to assess how differing perspectives affect the reporting of outcome over time.
2. Methods 2.1. Data source We retrospectively identiﬁed all patients admitted to the Cleveland Clinic (CC) main campus between January 2009 and May 2013 who underwent hemicraniectomy for ischemic stroke using the CC KP Data Registry. The KP program was designed to aggregate patient speciﬁc research data by collecting standardized patient-reported HSMs and clinician-reported HSMs at each outpatient follow-up. The KP contains a compilation of validated questionnaires that allow patients and clinicians to report HSMs, which are integrated into the electronic medical record (EMR). Completed questionnaires are automatically scored and made available in the KP database for use in outcomes research. Additional information is available at: http://my.clevelandclinic. org/neurological institute/research/knowledgeprogram.aspx. This study was approved by the CC Institutional Review Board.
2.2. Patients Ischemic stroke patients who had undergone hemicraniectomy were identiﬁed in the KP database using International Classiﬁcation of Diseases, Ninth Revision, Clinical Modiﬁcation (ICD-9-CM) diagnostic codes for ischemic stroke (433.11, 433.31, 433.81, 433.91, 434.01, 434.11, 434.91, 436) and American Medical Association Current Procedural Terminology (CPT® ) codes for decompressive craniotomy/craniectomy (61322/61323). Individual EMRs were reviewed for each patient to conﬁrm coding accuracy for hemicraniectomy. The combination of diagnostic and procedural codes produced an ischemic stroke cohort that underwent hemicraniectomy. Identiﬁed hemicraniectomy patients were included in this study only if they completed both patient- and clinician-reported HSMs at follow-up. Patients were excluded if the stroke was hemorrhagic, resulted from a recent craniotomy, or if the patient died or did not complete HSM measures during follow-up. Additional clinical information was collected from the EMR including the patient’s age, gender, advance directive status, race/ethnicity, admission time, time to surgery, medical comorbidities, clinical and radiographic characteristics related to stroke (including Glasgow Coma Scale (GCS), electrolyte levels, laterality, vascular distribution, midline shift, and the use of hyperosmolar therapy in the preoperative period). Midline shift, using the preoperative computed tomography (CT) scan, was determined at the level of the foramen of Monroe or the septum pellucidum, and both CT scans and magnetic resonance imaging (MRI) scans of the brain conﬁrmed ischemic stroke location and distribution in all patients. Stroke etiology was determined by a stroke neurologist according to the modiﬁed Trial of Org 10172 in Acute Stroke Treatment (TOAST) criteria .
2.3. Outcomes Outpatient follow-up was divided into “Early” and “Late” time periods, which were deﬁned as the earliest and latest follow-up visits available in the KP database for each patient. The mean follow-up for patients in Early period was 3 ± 2 months (standard deviation) and for patients in the Late period was 9 ± 3 months. Outcome data obtained from the KP registry were clinicianreported HSMs, including the mRS and NIHSS, and patient-reported HSMs, including the EQ-5D index and sub-scores, PHQ-9, and SIS-16 as summarized in Fig. 1. All clinician-reported HSMs were documented in the EMR by an attending physician, nursepractitioner, or clinical fellow. All clinician-reported HSMs were then conﬁrmed in the EMR by the attending neurologist or neurosurgeon during the follow-up visit. Patient-reported HSMs were completed either by the patient without assistance or by the patient with the assistance of a family member/caretaker at each follow-up visit. All patient-reported HSM questionnaires are structured to reﬂect only the patient’s perspective on outcome and health status. Patient-reported HSMs were completed on a tablet computer separate from the computer used by the clinician. 2.4. Surgical procedure and medical therapy Patients with large, space occupying ischemic strokes were placed under a “hemicraniectomy watch” upon admission to the Neurological ICU for approximately 72 h, which included frequent neurological exams, daily CT scans, and the administration of hypertonic therapy. Surgical intervention was under the discretion of the consulting neurosurgeon and neurointensivist. A standard hemicraniectomy with duraplasty was performed in all cases and conﬁrmed by the operative note. No patients underwent invasive intracranial pressure monitoring in the preoperative period and all patients received hyperosmolar therapy in the intensive care unit prior to surgery. Serum sodium levels were routinely collected in the preoperative period. All preoperative patient characteristics and clinical information are listed in Table 1. 2.5. Statistical analysis The difference in HSM scores between time periods was calculated and deﬁned as “Median difference.” Score differences were also calculated as a percentage of each HSM score range and reported as “Percentage (%) median difference.” The median difference for each HSM was compared using the Wilcoxon-signed rank test. We then compared patient- and clinician-reported HSM median differences using Spearman (non-parametric) correlation coefﬁcients. Statistical signiﬁcance was set at p ≤ 0.05. Statistical analyses were conducted using JMP® Pro 10.0.2 (SAS Institute Inc., Cary, NC). 3. Results Thirty-two patients underwent hemicraniectomy for ischemic stroke and were present in the KP database since 2009. Patientand clinician-reported HSMs were available for 11 patients (33%) at both follow-up time periods. Seven patients (22%) were excluded due to mortality either during hospitalization or during follow-up, and 14 patients (44%) were excluded due to incomplete follow-up after discharge (Fig. 2). Baseline admission characteristics reported in Table 1 were similar between included and excluded patient groups except for signiﬁcantly lower admission GCS in the excluded patient group. The median age was 55 years. Patients were predominately female, white, and had multiple medical comorbidities (Table 1).
M.L. Kelly et al. / Clinical Neurology and Neurosurgery 126 (2014) 24–29
Clinician report: functional ability
0 - 42
Clinician report: functional ability
Patient report: global quality of life
-0.11 - 1.0
Patient report: - mobility - self-care - usual activities, - pain/discomfort - anxiety/depression
Patient report: depression
0 - 27
Patient report: functional ability
0 - 100
Interpretation minor stroke (0-5) moderate stroke (6-13) major stroke (≥14) Functional independence (≤ 3) worse than death (