Prevalence, patterns, and predictors of patient-reported non-motor outcomes at 30 days after acute stroke: Prospective observational hospital cohort study

Background: Adverse non-motor outcomes are common after acute stroke and likely to substantially affect quality of life, yet few studies have comprehensively assessed their prevalence, patterns, and predictors across multiple health domains. Aims: We aimed to identify the prevalence, patterns, and the factors associated with non-motor outcomes 30 days after stroke. Methods: This prospective observational hospital cohort study—Stroke Investigation in North and Central London (SIGNAL)—identified patients with acute ischemic stroke or intracerebral hemorrhage (ICH) admitted to the Hyperacute Stroke Unit (HASU) at University College Hospital (UCH), London, between August 1, 2018 and August 31, 2019. We assessed non-motor outcomes (anxiety, depression, fatigue, sleep, participation in social roles and activities, pain, bowel function, and bladder function) at 30-day follow-up using the Patient-Reported Outcome Measurement Information System-Version 29 (PROMIS-29) scale and Barthel Index scale. Results: We obtained follow-up data for 605/719 (84.1%) eligible patients (mean age 72.0 years; 48.3% female; 521 with ischemic stroke, 84 with ICH). Anxiety (57.0%), fatigue (52.7%), bladder dysfunction (50.2%), reduced social participation (49.2%), and pain (47.9%) were the commonest adverse non-motor outcomes. The rates of adverse non-motor outcomes in ⩾ 1, ⩾ 2 and ⩾ 3 domains were 89%, 66.3%, and 45.8%, respectively; in adjusted analyses, stroke due to ICH (compared to ischemic stroke) and admission stroke severity were the strongest and most consistent predictors. There were significant correlations between bowel dysfunction and bladder dysfunction (κ = 0.908); reduced social participation and bladder dysfunction (κ = 0.844); and anxiety and fatigue (κ = 0.613). We did not identify correlations for other pairs of non-motor domains. Conclusion: Adverse non-motor outcomes were very common at 30 days after stroke, affecting nearly 90% of evaluated patients in at least one health domain, about two-thirds in two or more domains, and almost 50% in three or more domains. Stroke due to ICH and admission stroke severity were the strongest and most consistent predictors. Adverse outcomes occurred in pairs of domains, such as with anxiety and fatigue. Our findings emphasize the importance of a multi-domain approach to effectively identify adverse non-motor outcomes after stroke to inform the development of more holistic patient care pathways after stroke.


Introduction
Globally, stroke is the second most common cause of death and the third leading cause of disability-adjusted life years. 13][4][5][6][7] Furthermore, impairments not included in the mRS, such as post-stroke anxiety, depression, and fatigue, are associated with functional disability and premature death. 3,4[10][11][12][13][14][15][16] We therefore aimed to (1) assess a full range of patientreported non-motor outcome domains at 30 days after acute stroke; (2) identify the burden of adverse outcomes in multiple domains; (3) investigate correlations between nonmotor outcome domains; and (4) determine baseline independent predictors (including ischemic stroke vs intracerebral hemorrhage, ICH) for each adverse non-motor outcome.

Study design and data source
The Stroke Investigation in North and Central London (SIGNAL) prospective hospital-based cohort study is based at the University College London Hospitals (UCLH) hyperacute stroke unit (HASU) which serves an ethnically diverse population of 1.6 million adults' resident within five North Central London boroughs (Camden, Islington, Enfield, Harringay, and Barnet).

Study population
We assessed all patients admitted consecutively with acute stroke between August 1, 2018 and August 31, 2019, for eligibility.Patients were included if they were aged ⩾ 18 years; resident in North Central London; had a clinical diagnosis of acute stroke (ischemic or hemorrhagic) validated by a consultant stroke physician and confirmed on brain imaging (computed tomography (CT), magnetic resonance imaging (MRI), or both) by a consultant neuroradiologist; and were able to provide complete data from two or more domains of the Patient-Reported Outcome Measurement Information System-Version 29 (PROMIS-29) patient-reported health outcome scale at 30-day follow-up (further details below).Sociodemographic and clinical characteristics including age, sex, ethnic origin, admission stroke severity (defined using the NIH Stroke Scale, NIHSS), previous medical history, medication history, discharge destination, and functional outcome at hospital discharge (mRS) were obtained from electronic health records.
We made every effort to include all eligible participants by providing additional supporting measures.We reduced patient burden for individuals who reported moderate to severe communication problems, cognitive impairment, or language difficulties by providing options to be given extra time, complete a postal questionnaire, or have the outcome measures translated or completed by a proxy responder.Proxy responders were eligible to assist individuals in completing the questionnaires if they were listed as a registered carer or next of kin (NOK) in the patients' health records.

Outcomes at 30-day follow-up
Patient-reported non-motor outcome scales were administered as part of routine care by clinically trained practitioners via telephone appointment.We used the PROMIS-29 v2.0 scale, which assesses seven health domains (physical function, anxiety, depression, fatigue, sleep disturbance, social participation, and pain) using four items per domain.Domains including anxiety, depression, fatigue, sleep disturbance, and pain capture individuals' status in real time, asking about their experiences in the past 7 days.By contrast, the physical function and social participation domains capture status at the time of the initial stroke.Each PROMIS-29 domain score is standardized to the US general population on the T scale, with a mean of 50 and a standard deviation of 10; higher mean scores indicate worse outcomes.We considered an adverse non-motor domain to be present if the standardized domain score was ⩾ 55 (i.e. a score half a standard deviation worse than the general population). 11,15e measured bowel and bladder function using the Barthel Index.We considered individuals to have bowel or bladder dysfunction if they scored between 0 and 1, or if they had a urinary catheter at 30-day follow-up.

Protocol approvals, registrations, and patient consent
The SIGNAL registry was approved by the UCL Hospitals NHS Foundation Trust Governance Review Board as a Service Evaluation (code: 5-201920-SE).Since the study data were collected as part of routine clinical care, the requirement for individual patient consent was waived.

Statistical analysis
We summarized patient demographics and clinical characteristics with descriptive statistics.We compared continuous data using the Wilcoxon rank sum test, and categorical data with Pearson's chi-squared tests.We recorded the clinical and sociodemographic characteristics of those patients who did not meet the inclusion criteria (Supplemental Material Table 1).We used histograms and q-q plots to understand the distribution of the data.We calculated the prevalence of adverse non-motor outcomes for each domain.We used the Pearson's chi-squared test to compare the differences in prevalence of non-motor outcomes between ischemic stroke and ICH.We used the Benjamini-Hochberg false discovery rate procedure to guard against potential falsepositive discoveries. 17We performed multivariable logistic regression analysis for individual non-motor outcome domains controlling for clinically relevant variables (age, sex, previous history of stroke) and characteristics significantly (p < 0.10) associated with each non-motor outcome in univariable analysis; The statistical tests were two-sided, and statistical significance was set at p < 0.05.
The characteristics of those reporting more than two or three adverse outcomes were compared to the total study sample (see Supplemental Material Table 3) using chisquared test and t-test or the Kruskal-Wallis test as appropriate.We calculated Kappa statistics to quantify co-occurrence of non-motor outcome pairs.All statistical analysis were performed using Stata statistical software version 16.1.

Patient characteristics
We included 605/719 (84%) of potentially eligible survivors (mean age 72.0 years; 48.3% female; 521 with ischemic stroke, 84 with ICH) (see Figure 1, and Table 1).The reasons for excluding 114 individuals were that: they declined follow-up; did not attend follow-up; were uncontactable; or were unable to provide the required non-motor data.Supplemental Material Table 3 provides detailed baseline characteristics of individuals excluded from our analysis.We provided additional supporting measures to facilitate data completion in 51/605 (8.4%).Of these 51 patients who needed additional support, 9 had a recorded dementia diagnosis, 17 had moderate-severe memory problems, and 13 had severe communication problems, while 12 were non-English speakers.Regarding the additional support provided, 27/51 (52.9%) needed direct proxy assistance (from a registered carer or next of kin) to complete outcomes, 4 (7.8%)required proxy assistance to translate, and 20 (39.2%) required extra time either during the assessment or by being sent a postal pack.

Prevalence of adverse non-motor outcomes in multiple domains
The prevalence of ⩾1, ⩾2, and ⩾3 adverse non-motor outcomes were 88.4%, 66.3%, and 45.8%, respectively.Figure 3 shows how frequently each non-motor health domain outcome was associated with one, two, or three cooccurring outcomes.All non-motor domains were frequently associated with other co-occurring adverse outcomes.
Correlations between each non-motor domain on PROMIS-29 and Barthel Index are summarized in Figure 4.There was substantial correlation for anxiety with fatigue (κ = 0.613); reduced social participation with bladder dysfunction (κ = 0.844); and bowel dysfunction with bladder dysfunction (κ = 0.908).

Discussion
We found that adverse patient-reported non-motor outcomes were extremely common in people following acute stroke at 30 days: 88.4% of the evaluated stroke survivors reported at least one adverse non-motor outcome, 66.3% two or more, and 45.8% three or more.The health domains most affected were anxiety (57.0%), fatigue (52.7%), impaired bladder function (50.2%), and reduced participation in social roles and activities (49.3%).Most health domains were more likely to be affected after ICH compared to ischemic stroke, even after adjusting for confounding factors including stroke severity.We identified pairs of co-occurring domains, such as bowel dysfunction and bladder dysfunction, and reduced social participation and bladder dysfunction.Our findings have potential clinical implications: first, they show high burden of patientreported non-motor outcomes; second, they underline a substantial unmet patient-reported healthcare need requiring the development of appropriate multidisciplinary specialist comprehensive multi-domain screening and care delivery pathways; third, they highlight those at highest risk, for example, people with stroke due to ICH; fourth, the design of future care pathways to address adverse patient-reported non-motor outcomes may be informed by our findings of associations between these outcomes (e.g.anxiety and fatigue) .
Our findings are in line with other, predominantly single-domain studies, [5][6][7][8][9]12,15,[17][18][19][20][21][22][23][24][25] including some which suggest that even after complete motor recovery, deficits in non-motor domains, such as anxiety, depression, fatigue, pain, bowel, and bladder dysfunction, may prevent a return to independent living. 12,13,15 Ou analysis extends these findings by providing new data for all key domains, and predictive factors, in the same cohort; few previous studies investigated all these domains in the same population.14,15,17,25 We found that anxiety affected 57% of patients and was associated with ICH (compared to ischemic stroke), age < 50 years and worse admission stroke severity.A previous meta-analysis found persistent anxiety in 38-76% of patients but did not investigate predictive factors.3,12 Small cohort studies suggest that ICH volume, lobar location, and MRI findings of cerebral amyloid angiopathy (CAA) predict anxiety after stroke, 26 but further larger studies are needed in ICH subgroups.International Journal of Stroke, 19 (4)   Our findings that fatigue and depression are common (affecting 52.7% and 36.5% of patients, respectively) are consistent with previous reports describing fatigue in 42-53% (between 1 and 6 months after stroke) and depression in 24-39% (between 3 months and 5 years).3,12,21,26 We found that fatigue is more prevalent after ICH than ischemic stroke (65.8% vs 50% of patients, respectively), in line with a meta-analysis of small studies that found that ICH stroke survivors had nearly double the prevalence of fatigue compared to ischemic stroke (66% vs 36%, respectively).12,[20][21][22] Our findings of reduced social roles and participation after stroke are consistent with some, [23][24][25]27 but not all previous studies.28,29 In agreement with previous studies, [23][24][25]27 patients with previous stroke or TIA, and poor functional outcome at discharge, more often reported reduced social participation, but we additionally found associations with admission stroke severity and ICH which may be important to better target interventions, such as training carers, improved social support, dedicated rehabilitation, and treatment of depression.30 Our findings confirm that pain is commonly reported after stroke (overall prevalence 47.9%), consistent with previous estimates of between 11% and 55%.26,[31][32][33] We found that pain is more common after ICH than ischemic stroke (60.3% vs 46%) consistent with a small cohort study which reported pain to be more frequent after thalamic ICH than ischemic stroke.34 We identified sleep disturbance in 40.9% of patients at 30-day follow-up, in agreement with previous studies reporting prevalence of 20-67% at 1 month to 5 years after stroke; [35][36][37][38][39][40] black ethnic origin was associated with sleep disturbance, consistent with previous large cohort studies. 39,40 Wefound that bladder dysfunction was more prevalent than bowel dysfunction and that bowel dysfunction was more prevalent after ICH than ischemic stroke, in line with a previous meta-analysis.41 We found a lower prevalence of bowel dysfunction than some previous reports, which could be attributed to not sub-classifying fecal incontinence and s constipation, underreporting, the screening tool we used (Barthel Index), or selection or ascertainment bias in other studies.[41][42][43]  International Journal of Stroke, 19 (4)   Few previous studies have investigated patterns of cooccurrence and correlations between adverse non-motor outcomes after stroke.We found a high burden of multiple adverse non-motor outcomes, with about two-thirds of participants reporting two or more, and nearly on-half reporting three or more.We also found that all adverse non-motor outcomes were associated with the co-occurrence of one, two, or three others (i.e. none occurr in isolation), but that some co-occurred more often than expected by chance (anxiety and fatigue, social participation and bladder function, and bladder and bowel function).
Our study has strengths.We included consecutive patients with acute stroke from a defined ethnically diverse large and defined North London population, and systematically assessed adverse non-motor outcomes in multiple health domains.Stroke diagnosis was confirmed by brain imaging ensuring accurate diagnosis and classification.We used false discovery rate analysis to avoid false-positive findings and were able to assess other independent baseline predictors (including ICH) while adjusting for potential confounding factors.
We also acknowledge potential limitations including potential selection bias, though this was minimized by including sequential patients with a high rate of follow-up (84.1%).Compared to those included, there was a higher proportion of people of non-White ethnicity in those excluded, emphasizing the importance for future studies including all ethnic groups.We also acknowledge that for anxiety and depression PROMIS-29 measures symptom burden over the 7 days and does not provide a formal diagnosis of these conditions.Our data describe the burden on non-motor outcomes at 1 month, but more data are needed at longer-term follow-up.Although 47.8% of participants were discharged to an acute stroke unit (ASU), where they received rehabilitation input, including physiotherapy, speech and language, and occupational therapy, we do not have detailed data on the exact degree of rehabilitation input received for each participant.Furthermore, we did not have access to neuropsychological interventions provided after HASU discharge.
Our findings regarding the prevalence, patterns, and predictors of adverse patient-reported non-motor outcomes should help stroke services to plan pathways to first ascertain and then address these patient-reported adverse outcomes to improve post-stroke quality of life and provide patient-centered stroke care pathways.However, further long-term studies-including information on functional impact-are needed to fully establish the clinical relevance of these patient-reported non-motor outcomes.

Figure 2 .
Figure 2. Prevalence of adverse non-motor outcomes at 30-day follow-up for all patients and according to stroke type (ischemic stroke or ICH).

Figure 4 .
Figure 4. Correlations between pairs of non-motor outcome domains.

Table 1 .
Demographic and clinical characteristic data for all patients and according to stroke type (ischemic stroke or ICH).

Table 2 .
Multivariable analysis of baseline factors associated with adverse non-motor outcomes.