Prescription of blood pressure lowering treatment after intracerebral haemorrhage: Prospective, population-based cohort study

Introduction Blood pressure (BP) lowering reduces the risk of recurrent stroke after intracerebral haemorrhage (ICH). However, implementation of BP lowering in clinical practice in the UK is unknown. Patients and methods We identified adults with first-ever incident ICH to quantify the proportion who survived >14 days after hospital discharge and were prescribed BP-lowering medication in a prospective, population-based, inception cohort study in the Lothian region of Scotland during June 2010–May 2012 and January–December 2019. After the first cohort, we analysed reasons for avoiding BP-lowering medication in a sample from the Lothian region of the Scottish Stroke Care Audit during January 2017–November 2017, which informed a quality improvement intervention that was implemented in the second cohort. Results After efforts to improve monitoring and lowering of BP amongst ICH survivors, there was an increase in the proportion of patients prescribed BP-lowering medication at hospital discharge between the first and second population-based cohorts (81/130 [62%] vs. 68/89 [76%]; P = 0.028). Compared with patients not prescribed BP-lowering medication at hospital discharge, patients prescribed BP-lowering medication presented with higher systolic BP (177 vs. 156 mm Hg, P = 0.002 and 180 vs. 149 mm Hg, P < 0.001, in the first and second population-based cohorts, respectively), and were more likely to have pre-morbid hypertension (85% vs. 33%, P < 0.001 and 72% vs. 29%, P < 0.001) and atrial fibrillation (35% vs. 4%, P < 0.001 and 26% vs. 5%, P < 0.034). Conclusion In this population-based study, the proportion of patients with ICH who were prescribed BP-lowering medication at hospital discharge increased after a quality improvement intervention.


Introduction
Elevated blood pressure (BP) is the strongest modifiable risk factor for intracerebral haemorrhage (ICH) and accounts for approximately half of the population-attributable risk. 1,2 In patients who survive beyond 90 days after ICH, failure to achieve adequate BP control is common, 3 and is associated with a higher risk of recurrent lobar and non-lobar ICH, 4 and other adverse cardiovascular outcomes. 5 In 2004, the perindopril protection against recurrent stroke study (PROGRESS) randomised controlled trial showed that BP lowering reduced the relative risk of recurrent stroke by 50% in ICH survivors. 6 The United Kingdom national stroke guidelines were updated in 2012 to recommend lowering systolic BP to <130 mm Hg in stroke survivors where there are no contraindications. 7 We sought to determine the frequency of prescription of BP-lowering medication and reasons for avoidance in ICH survivors at hospital discharge in the Lothian region of Scotland, United Kingdom. Our findings informed a quality improvement intervention that aimed to increase the use of BP monitoring and BP-lowering medication in Lothian. We compared the proportion of ICH survivors who were prescribed BPlowering medication at hospital discharge in a population-based cohort study before and after the quality improvement intervention to identify a temporal change in practice, and assessed patient characteristics associated with the prescription of BP-lowering medication to identify areas for further improvement.

Setting
Prospective, population-based, longitudinal cohort study of ICH used for audit, research and quality improvement. We identified patients with ICH (presumed due to cerebral small vessel disease) in The Lothian Audit of the Treatment of Cerebral Haemorrhage (LATCH) from 1 June 2010 to 31 May 2012, and 1 January 2019 to 31 December 2019. LATCH ascertained cases from the Lothian Health board region of Scotland (including 3 hospitals) in residents aged !16 years diagnosed with first-ever or recurrent ICH (mid-2010 population aged !16 years: 695,335) through multiple routes: prospective hot pursuit sources through a trust-wide network of physicians, neurologists, neurosurgeons, radiologists, pathologists, specialist nurses and audit personnel, daily neuroradiology meetings and review of all computed tomographic (CT) head scans; and retrospective searches of electronic patient notes, International Classification of Diseases 10th Revision (ICD-10) coded hospital discharge records (http://www.isdscot land.org), the Scottish Stroke Care Audit (http:// www.strokeaudit.scot.nhs.uk), and records at the Office of the Procurator Fiscal. 8 Retrospective search of the Scottish Stroke Care Audit (SSCA) database. In order to inform a quality improvement intervention that could be evaluated during the second population-based study, we did a retrospective search of the Scottish Stroke Care Audit (SSCA) database to identify patients with ICH in Lothian from 2 January 2017 to 9 November 2017, and examined their hospital records to identify reasons why BPlowering medication was not prescribed at hospital discharge.
Study participants. We defined ICH as a symptomatic event (new headache, focal neurological symptoms, or altered consciousness), referable to a discrete parenchymal bleed (confirmed by temporally consistent radiology or pathology findings). We included patients with incident first-ever ICH presumed due to cerebral small vessel disease who survived >14 days after hospital discharge. We reviewed patients' medical records and investigations to exclude cases of recurrent ICH, isolated extra-axial intracranial haemorrhage, and ICH definitely attributable to a macrovascular cause, tumour, trauma, and haemorrhagic transformation of ischaemic stroke. Standard clinical practice in NHS Lothian involves targeted use of CT angiography informed by clinical predictors of the risk of underlying macrovascular pathology. 9 LATCH audit standard. The LATCH audit standard relating to long-term BP lowering after ICH is as follows: 'blood pressure-lowering therapy should be prescribed by hospital discharge if systolic BP is >130 mm Hg, unless there is a contra-indication to the use of these drugs (e.g. end of life care, hypotension etc).' Therefore, the primary outcome was the prescription of BP-lowering medication at hospital discharge, irrespective of prior BP-lowering medication use or prior hypertension diagnosis.
Quality improvement intervention. The 2017 audit indicated that there was a persisting unexplained deficit in proportion of ICH survivors that received a prescription for BP-lowering medication at hospital discharge, so we sought to improve the availability and use of BP monitoring and BP-lowering medication where it was indicated by clinical guidelines 7 To promote the quality improvement intervention, we organised a series of education sessions about ICH-ACT and the various BP monitoring and lowering activities for key stakeholders in secondary care (colleagues in neurology, stroke, acute internal medicine and emergency departments). We worked closely with clinical colleagues to maintain engagement with the intervention: this included weekly departmental meetings attended by stroke clinicians, and monthly email reminders to LATCH collaborators (supplementary material). We anticipated that these activities would improve the uptake of guideline-recommended BP management by raising awareness of this important intervention amongst patients and their hospitalbased healthcare practitioners, and by providing several options for their long-term BP management. 7,10 Data collection. We retrospectively collected prospectively recorded data on BP-lowering medication use after ICH from paper and electronic patient records associated with the corresponding hospital admission including initiation of, or changes to, BP-lowering medication during admission and BP-lowering prescription on discharge. In the population-based cohorts, we extracted additional data from patients' medical records: medical history; use of BP-lowering, antiplatelet and anticoagulant medication at ICH diagnosis; clinical data on admission (BP and Glasgow Coma Scale score); and haematoma characteristics on baseline CT brain scan (anonymised scans were prospectively collected and held on a secure server). We rated ICH location on baseline CT brain scans according to the CHARTS rating system. 11 In view of prescribing patterns in the first population-based cohort, we assessed reasons for the avoidance of BP-lowering medication in the audit and the second population cohort (these data were not available for the first population-based cohort). To assess participation in several BP-monitoring activities available from 2018 onwards, we collected information about the long-term BP monitoring plan from the electronic discharge summary in the second population cohort.
Statistical analysis. To identify a temporal change in practice, we compared the proportion of ICH survivors who were prescribed BP-lowering medication at hospital discharge before (first population-based cohort) and after (second population-based cohort) the quality improvement intervention using the chi-squared test. We assessed the associated change in the proportion of ICH survivors who were prescribed BP-lowering at hospital discharge in a multivariable logistic regression model adjusted for clinically important baseline confounders, which included age (!80 years vs. <80 years), Glasgow coma scale (GCS) score ( 12 vs. >12), ICH location (infratentorial vs. other), presence of intraventricular haemorrhage, presence of any cardiovascular comorbidity, and prior use of anticoagulation. We also compared the two population-based cohorts by their patient characteristics, stratified by prescription of BP-lowering medication at hospital discharge (yes vs. no), using the Kruskal-Wallis test for continuous variables, and the chi-squared test for categorical variables. The significance level was set at P ¼ 0.05.
The subsequent audit that informed the quality improvement intervention (2 January 2017 to 9 November 2017, inclusive) revealed that of 61 patients with ICH who survived >14 days after hospital discharge, 33 (54%) were prescribed BP-lowering medication, 19 (31%) were not prescribed BP-lowering medication and had a reason for avoiding them and 9 (15%) were not prescribed BP-lowering medication and did not have a clear reason for avoiding them (Figure 2).
Following the introduction of the quality improvement intervention, the second population-based cohort (1 January 2019 to 31 December 2019, inclusive) revealed that of 136 cases of incident first-ever ICH, 89 (65%) survived >14 days after hospital discharge. A comparison of the baseline characteristics of the first and second population-based cohorts is shown in Supplementary Table 1: compared to the first population-based cohort, anticoagulant use was more common, and subarachnoid extension less common in the second population-based cohort. Among 89 patients with ICH who survived >14 after from hospital discharge, 68 (76%) patients were prescribed BPlowering medication, 15 (17%) patients were not prescribed BP-lowering medication and had a reason for avoiding them, and 6 (7%) patients were not prescribed BP-lowering medication and did not have a clear reason for avoiding them (Figure 2). Compared with the proportion of patients who received BP-lowering medication at hospital discharge in the first population-based cohort, the proportion of patients who received BPlowering medication in the second population-based cohort represented a statistically significant increase (76% vs. 62%, P ¼ 0.028). In adjusted analyses, the  association between the quality improvement intervention and the proportion of patients prescribed BPlowering medication at hospital discharge persisted (adjusted odds ratio 2.26 [95% confidence interval 1.08 to 4.73], p ¼ 0.030, area under ROC curve 0.81).
The distribution of reasons for avoiding BPlowering medication was similar in the audit and second population-based cohort: among patients with a reason for avoiding BP-lowering, there were 11/19 (58%) and 9/15 (60%) with systolic BP levels <130 mm Hg; 5/19 (26%) and 3/15 (20%) with frailty/ poor function; and 3/19 (16%) and 3/15 (20%) with a contraindication to BP-lowering medication in the audit and second population-based cohort, respectively. Table 1 shows the baseline characteristics of ICH patients who survived >14 days after hospital discharge, stratified by population-based cohort epoch and prescription of BP-lowering medication at hospital discharge. Compared with patients not prescribed BP-lowering medication at hospital discharge, patients prescribed them were similar in age and sex, and the proportion with deep/infratentorial haematoma location was also similar (57% vs. 46%, P ¼ 0.241 and; 57% vs. 48%, P ¼ 0.433, in the first and second population-based cohorts, respectively). Compared with patients not prescribed BP-lowering medication at hospital discharge, patients prescribed them presented with higher systolic BP (177 vs. 156 mm Hg, P ¼ 0.002 and; 180 vs. 149 mm Hg, P < 0.001, in the first and second population-based cohorts, respectively), and were more likely to have pre-morbid hypertension (85% vs. 33%, P < 0.001 and; 72% vs. 29%, P < 0.001) and atrial fibrillation (35% vs. 4%, P < 0.001 and; 26% vs. 5%, P < 0.034), and be prescribed BP-lowering medication (none vs. one vs. many: P < 0.001 and P < 0.001) and anticoagulant medication (none vs. one: P ¼ 0.003 and P ¼ 0.041) at symptom onset. Overall, there were 48/130 (37%) and 19/89 (21%) prescribed antiplatelet medication, and 13/130 (10%) and 18/89 (20%) prescribed anticoagulant medication at the time of ICH onset in the first and second population-based cohorts, respectively. Table 2 shows the BP management of ICH patients: 3 or medications were prescribed in 26/128 (20%) and 21/89 (24%) of patients in the first and second population-based cohorts, respectively; 8/128 (6%), and 7/89 (8%) of patients received the combination of a renin-angiotensin system blocker, calcium channel blocker and diuretic. In the first population-based cohort and the audit, renin-angiotensin system blockers were the most frequently prescribed agent, compared with the second population-based cohort where calcium channel blockers were used most frequently. In the second population-based cohort, long-term BP monitoring was attributed to the patient's general practitioner in 41/89 (46%) patients, home tele-monitoring in 10/89 (11%) patients, home self-monitoring in 3/89 (3%) patients and a clinical trial in 3/89 (3%) patients; no plan was documented in 32/89 (36%) patients.

Discussion
In two epochs of this prospective, population-based, longitudinal cohort study and an interim audit, between half and three quarters of ICH survivors were prescribed BP-lowering medication at hospital discharge; around a third of patients who did not receive BP-lowering medication did not have a clear  reason for avoiding them. In 2019 (the second population-based cohort), there was a significant increase in the proportion of patients who received BP-lowering medication to 76% from 62% in the first population-based cohort; 24% were prescribed 3 or more agents, and the use of calcium channel blockers overtook renin-angiotensin system blockers. Premorbid hypertension and atrial fibrillation were more common among patients prescribed BPlowering treatment at hospital discharge compared with patients who were not; age, sex and the proportion of patients with deep/infratentorial haematoma location were similar. Despite evidence of benefit emerging from the PROGRESS trial nearly two decades ago, 6,12 and updates to the UK National Clinical Guidelines for Stroke in 2012 7 and European Stroke Organisation guidelines in 2014, 10 prescription of BP-lowering medication for the secondary prevention of recurrent stroke after ICH could be further improved. Yet, while little change occurred between 2010-2012 (first populationbased cohort) and 2017 (audit), we found an increase in the proportion of ICH patients that received BPlowering at hospital discharge in 2019 (second population-based cohort), which was associated with the introduction of a quality improvement intervention involving Lothian-wide guidance on ICH management and opportunities to recruit patients to two randomised controlled trials (TRIDENT and PROHIBIT-ICH) and a home BP tele-monitoring initiative (Florence) from 2018 onwards. These findings are encouraging, and may reflect increasing awareness of the benefits of BP-lowering after ICH among clinicians due to embedded quality improvement.
Our findings are similar to those from other observational studies of prescribing practice after ICH: 74% of patients were prescribed BP-lowering medication after ICH in a recent, large Australian study, 13 61% in a large US study 4 ; and when considered alongside all-cause strokes, varying but substantial underprescribing is reported in Canada 14 and the UK. 15 To understand patient factors that influence prescribing habits, we stratified patients by population-based study epoch and BP-lowering medication prescription at hospital discharge. We found pre-morbid hypertension and atrial fibrillation were more common among patients who received BP lowering in first and second population-based cohorts, and hyperlipidaemia was more common in patients who received BP lowering in the second population-based cohort; differences in prior myocardial infarction, transient ischaemic attack, and diabetes mellitus did not meet statistical significance. These findings are consistent with those from an Australian registry where detailed comorbidity data were collected in 1,498 ICH patients, 13 and when considered together, probably reflect the continued use of pre-morbid BP-lowering medication in patients with a diagnosis of hypertension or other cardiovascular comorbidity, and b-blockers prescribed for atrial fibrillation. The PROGRESS trial results showed that benefits from BP-lowering treatment are likely to exist irrespective of baseline BP and other cardiovascular comorbidities. 6,12 Therefore, it is important to reach all patients without a contraindication to BP-lowering medication.
By the same criteria, we assessed differences in admission BP and haematoma characteristics on diagnostic CT scan: compared with patients who were not prescribed BP-lowering medication at hospital discharge, patients prescribed them had significantly higher BP on admission to hospital. We also identified that a systolic BP level <130 mm Hg was a common reason for avoiding BP-lowering medication. This indicates some degree of concordance with clinical guidelines, 10,16 where BP lowering to a systolic BP target <140 mm Hg is recommended during the first 7 days due the potential for modest benefits on function and health-related quality of life, 17 followed by continued treatment to achieve systolic BP levels <130 mm Hg in the long term. However, haematoma location, and presence of extension into intraventricular, subarachnoid and subdural spaces did not significantly differ according to discharge BP lowering. These findings are encouraging in view of evidence to suggest benefits from long-term BP lowering exist in both lobar and non-lobar ICH, 4,18 although our small sample size may have precluded the detection of any true differences.
A key methodological strength of this study is its population-based design: to our knowledge, all incident ICH cases in the Lothian population were captured through multiple sources of ascertainment. Patient characteristics associated with the prescription of BP-lowering medication at hospital discharge were internally consistent in two population-based cohorts separated by nearly a decade. These similarities suggest that changes in the proportion of patients who were prescribed BP-lowering medication may have resulted from the introduction of quality improvement measures rather than changes in patient characteristics. Other temporal changes in prescribing appeared to reflect emerging evidence, including: more frequent use of calcium channel blockers at hospital discharge where data suggest greater efficacy for stroke prevention 19,20 ; and more frequent use of anticoagulants in patients with atrial fibrillation, where reductions in the risk of ischaemic stroke are greater than in patients taking antiplatelets. 21 However, our small sample and multiple comparisons increase the likelihood of chance findings.
We acknowledge there are limitations to our study. First, the increase in the proportion of patients with ICH who were prescribed BP-lowering at hospital discharge associated with the quality improvement intervention is prone to confounding from unmeasured factors, including temporal changes in practice related to early intensive BP lowering. Second, in the absence of long-term follow up data from patients identified by our population-based study in 2019, we were unable to assess temporal changes in the level of concordance with treatment in the community, nor the proportion achieving optimal systolic BP control in the longer term. In view of our finding that multi-agent treatment regimens are common among ICH survivors, and recent evidence to suggest a fixed low-dose polypill in addition to standard care is safe, tolerable and lowers BP more than standard care alone in middle-aged adults with hypertension, 22 randomised evidence from the TRIDENT randomised controlled trial is awaited to determine the effects of this strategy on recurrent stroke and vascular events in ICH survivors.
In summary, the prescription of BP-lowering medication at hospital discharge for the prevention of recurrent stroke after ICH over the last decade could be improved upon. A quality improvement intervention that involved Lothian-wide guidance on the management of ICH, and the opportunity for ICH survivors to participate in several BP monitoring and BP-lowering activities, was associated with an increase in the proportion of patients who received BP-lowering medication at hospital discharge. Further improvements may be possible if BP lowering is considered in patients without a previous diagnosis of hypertension or other cardiovascular comorbidities.