Association between patient outcomes and key performance indicators of stroke care quality: A systematic review and meta-analysis

Abstract

Purpose

Translating research evidence into clinical practice often uses key performance indicators to monitor quality of care. We conducted a systematic review to identify the stroke key performance indicators used in large registries, and to estimate their association with patient outcomes.

Method

We sought publications of recent (January 2000–May 2017) national or regional stroke registers reporting the association of key performance indicators with patient outcome (adjusting for age and stroke severity). We searched Ovid Medline, EMBASE and PubMed and screened references from bibliographies. We used an inverse variance random effects meta-analysis to estimate associations (odds ratio; 95% confidence interval) with death or poor outcome (death or disability) at the end of follow-up.

Findings

We identified 30 eligible studies (324,409 patients). The commonest key performance indicators were swallowing/nutritional assessment, stroke unit admission, antiplatelet use for ischaemic stroke, brain imaging and anticoagulant use for ischaemic stroke with atrial fibrillation, lipid management, deep vein thrombosis prophylaxis and early physiotherapy/mobilisation. Lower case fatality was associated with stroke unit admission (odds ratio 0.79; 0.72–0.87), swallow/nutritional assessment (odds ratio 0.78; 0.66–0.92) and antiplatelet use for ischaemic stroke (odds ratio 0.61; 0.50–0.74) or anticoagulant use for ischaemic stroke with atrial fibrillation (odds ratio 0.51; 0.43–0.64), lipid management (odds ratio 0.52; 0.38–0.71) and early physiotherapy or mobilisation (odds ratio 0.78; 0.67–0.91). Reduced poor outcome was associated with adherence to swallowing/nutritional assessment (odds ratio 0.58; 0.43–0.78) and stroke unit admission (odds ratio 0.83; 0.77–0.89). Adherence with several key performance indicators appeared to have an additive benefit.

Discussion

Adherence with common key performance indicators was consistently associated with a lower risk of death or disability after stroke.

Conclusion

Policy makers and health care professionals should implement and monitor those key performance indicators supported by good evidence.

Keywords

Stroke indicator care quality patient outcome

Introduction

In recent years, there have been concerted efforts to develop and implement clinical practice guidelines for the management of patients with acute stroke.¹ Clinical guidelines are written to promote diagnostic or therapeutic interventions applicable to the majority of patients in most circumstances. However, the use of guideline recommendations for individual patients has traditionally been left to the discretion of individual clinicians.² A recognised approach to assist the translation of research evidence into clinical practice is to monitor the quality of care using standardised performance indicators³ also called quality indicators, process of care measures or key performance indicators (KPIs). Performance indicators are standards of care which imply that health care professionals are providing inadequate care if eligible patients do not receive that standard of care. Performance indicators can be used to monitor the adherence to current guidelines, and support the transfer of new evidence into everyday clinical practice.⁴

There are now numerous stroke interventions that have been shown to improve patient outcomes in research trials; admission to specialised stroke units, use of intravenous thrombolysis, mechanical thrombectomy, antiplatelet drugs, anticoagulants and management of fever, hyperglycaemia and swallowing dysfunction for selected patient groups.^5–9 However, application into routine practice is challenging and regular monitoring is important.¹⁰ Ideally, implementation of clinical evidence can be demonstrated using a range of stroke KPIs, which offer proxy measures for ideal care being delivered. In turn, this would lead to evidence of better patient outcomes.¹¹

In a previous systematic review of the association between stroke quality (performance) indicators and patient-centred outcomes, out of 14 studies that met the eligibility criteria; 9 had mostly positive associations, whereas 5 reported little or no association with a lower risk for mortality, disability, medical complications, stroke recurrence or patient dissatisfaction.¹² A limitation of this review was the exclusion of stroke unit care as a performance indicator. With the ongoing developments in clinical guidelines and quality indicators for monitoring the application of these guidelines,^10,13 we believe that there is a need for up-to date comprehensive information on KPIs for stroke care.

We aimed to conduct a systematic literature review to identify the KPIs that have been described in stroke care and to summarise their association with patient outcomes. We intend that information gathered from this review will provide decision makers and health care professionals with information on reliable and meaningful KPIs that can be implemented to improve outcomes post stroke.

Methods

This review was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.¹⁴ This review was registered in Prospero Database (CRD42016050798).

Search strategy

Searching sources were Ovid Medline, Embase and PubMed databases, and relevant references from screening the bibliographies of the initial articles included in the search.

We used Medical Subject Headings (MeSH) and all subheading terms including ‘stroke’, ‘cerebrovascular accident’, ‘cerebrovascular disease’, ‘cerebrovascular disorders’, ‘brain hemorrhage’, ‘intracranial hemorrhages’, ‘brain infarction’, ‘subarachnoid hemorrhage’, ‘health care quality’, ‘quality of health care’, ‘quality indicators, health care’, ‘quality assurance, health care’, ‘quality control’, ‘quality indicator’, ‘performance indicator’, ‘register’, ‘registries’, ‘clinical audit’, ‘treatment outcome’, ‘case fatality rate’, ‘mortality’, ‘survival’, ‘disability’, ‘functional status’, ‘hospitalization’, ‘cost’, ‘quality of life’, ‘complication’, ‘hospital discharge’ and ‘stroke recurrence’. Our search was restricted to full-text manuscripts published in English from 1 January 2000 to 24 May 2017.

The search strategies for different databases are detailed in the Online Supplement.

Inclusion criteria

We included national or regional registers that recorded the independent association (after adjusting for at least age and a measure of stroke severity) between the KPIs and stroke patient outcomes, and involved patients from at least three hospitals.

Exclusion criteria

We excluded reports that were reviews or did not provide odds ratio (OR), hazard ratio (HR) or rate ratio (RR) data.

Screening and quality assessment

One author (GU) reviewed each title and excluded obviously irrelevant studies. Articles identified as potentially relevant underwent a full review by two authors (GU and PL) to determine if they met the inclusion criteria. In cases of disagreement, final determination was by discussion and consensus.

Data extraction

We used a standardised form to record information on country, main inclusion or exclusion criteria for the recruitment of participants, sample size, stroke severity measure, KPIs and outcome(s) reported, and reported results (and 95% confidence interval (CI)).

Data analysis

Initially, the identified KPIs and their association with the patient outcomes were categorised on whether the authors reported a significant association between the KPI and patient outcome. There was then a further quantitative analysis (meta-analysis) of the relationship (adjusted for at least age and stroke severity) between the KPIs identified and patient outcomes. Some checking of the consistency of KPIs and outcomes was required with grouping of similar KPIs. For the meta-analysis, we sought information on case fatality and poor outcome (death and disability or requiring support) after stroke.

The meta-analysis was done using the Review Manager (version 5.3) software. Log ORs were combined using an inverse variance analysis (random effects model). First, we assumed that HRs and RRs approximate the ORs and performed the primary meta-analysis including all studies reporting on association of KPIs with case fatality and poor outcome. Second, we performed sensitivity meta-analysis by excluding studies that used HR or RR as measures of association.

Results

The review profile is shown in Figure 1. We identified 3606 references from which 30 studies^15–44 were eligible for the qualitative review. Among these, only 22 were eligible for the meta-analysis.

Figure 1.

Review profile showing selection of studies. HR: hazard ratio; KPI: key performance indicator; OR: odds ratio.

Included studies

Table 1 shows the studies considered for our systematic literature review. Most of the included studies^15–29 were conducted in Europe: One European study¹⁵ was multinational (across ten countries), the rest were conducted in Denmark (n = 6), Sweden (n = 2), United Kingdom (n = 3) (one in England and two in Scotland), Italy (n = 1), Spain (n = 1) and Greece (n = 1). The non-European studies were conducted in the USA, Canada, Chile, Australia, New Zealand, China, Thailand and Taiwan. Two reports from Denmark^17,18 and two from Scotland^24,25 were based on the same datasets but since they provided associations with different outcomes, they were all included in this systematic review.

Table 1.

Studies eligible for the systematic literature review.

Study	Stroke type	Stroke severity measure	Sample size	Performance indicator	Patient outcome	OR/HR/RR	95% CI
Europe¹⁵	SAH excluded	Level of consciousness, incontinence, dysphagia, dysphasia and paralysis	1847 (10 countries)	Brain imaging	3-month case fatality	0.7	0.4–1.3
				Brain imaging	3-month disability (BI ≤ 18)	1.45	0.39–7.4
				Organised stroke care^a	3-month case fatality	0.5	0.3–0.8
				Organised stroke care^a	3-month disability (BI ≤ 18)	1.3	0.6–1.76
Denmark¹⁶	Ischaemic stroke	SSS scale (SSS)	22,179 (all Danish hospitals)	Anticoagulants for IS with AF	4-year survival	1.91	1.44–2.52
Denmark¹⁷	SAH excluded	SSS	29,573 (40 hospitals)	Specialised stroke unit by 2nd day	90-day case fatality	0.76	0.69–0.83
				Antiplatelet therapy by 2nd day		0.71	0.62–0.81
				Anticoagulants for IS with AF by 14th day		0.41	0.31–0.52
				CT/MRI scan by first day		1.35	1.24–1.46
				Assessment by a PT by 2nd day		0.81	0.73–0.88
				Assessment by an OT by 2nd day		0.83	0.75–0.91
				Nutritional risk assessment by 2nd d.		0.69	0.61–0.76
				Number of criteria fulfilled
				1 vs 0		0.94	0.65–1.49
				2 vs 0		0.78	0.54–1.02
				3 vs 0		0.60	0.42–0.78
				4 vs 0		0.61	0.42–0.79
				5 vs 0		0.45	0.31–0.60
				6 vs 0		0.48	0.31–0.65
Denmark¹⁸	SAH excluded	SSS	2636 (7 stroke units)	Stroke unit (2nd day)	Prolonged LoS	0.71	0.65–0.77
				Antiplatelet for IS (2nd day)		0.80	0.73–0.87
				Anticoagulant for IS with AF (14th day)		0.78	0.62–0.98
				CT/MRI scan (2nd day)		0.82	0.74–0.91
				PT assessment (2nd day)		0.87	0.81–0.93
				OT assessment (2nd day)		0.85	0.80–0.91
				Nutritional risk assessment (2nd day)		0.83	0.77–0.90
				Swallowing assessment (2nd day)		0.78	0.69–0.87
				Constipation risk assessment (2nd day)		0.70	0.63–0.78
				Mobilisation (2nd day)		0.67	0.61–0.73
				Intermittent catheterisation (2nd day)		0.77	0.64–0.92
				DVT prophylaxis (2nd day)		0.82	0.71–0.95
				Percentage of criteria fulfilled
				25%–49% vs 0%–24%		0.77	0.69–0.86
				50%–74% vs 0%–24%		0.67	0.60–0.75
				75%–100% vs 0%–24%		0.53	0.48–0.59
Denmark ¹⁹	All ischaemic strokes	SSS	4292 (all Danish hospitals)	Thrombolysis	1.4 years mortality	0.66	0.49–0.88
	All ischaemic strokes				1.4 years recurrent stroke	1.05	0.68–1.64
					1.4 years major bleeding	0.59	0.24–1.47
Denmark²⁰	First ever ischaemic strokes	SSS	5070 (Aarhus County)	Antidepressants during hospitalisation	30-day case fatality	0.28	0.18–0.43
Denmark²¹	All stroke types	SSS	11,757 (10 stroke units in two counties)	Early admission to a stroke unit	Any medical complication^b during admission (LoS = 13 days)	0.79	0.68–0.92
				Antiplatelet therapy for IS		0.95	0.79–1.15
				Anticoagulant therapy for IS with AF		0.59	0.45–0.76
				CT/MRI scan		1.52	1.35–1.72
				Assessment by a PT		1.10	0.94–1.28
				Assessment by an OT		1.10	0.94–1.27
				Assessment of nutritional risk		0.87	0.70–1.07
				Swallowing assessment		0.97	0.84–1.11
				Early mobilisation		0.43	0.35–0.53
				Percentage of criteria fulfilled
				25%–49% vs 0%–24%		0.77	0.67–0.88
				50%–74% vs 0%–24%		0.57	0.46–0.70
				75%–100% vs 0%–24%		0.50	0.36–0.68
Sweden²²	SAH excluded	Level of consciousness	8194 (all hospitals in Sweden)	Stroke unit (Independent before stroke)	2-year case fatality	0.81	0.72–0.92
Sweden²²	SAH excluded	Level of consciousness	8194 (all hospitals in Sweden)	Stroke unit (Independent before stroke)	2-year functional dependency	0.79	0.66–0.94
Sweden²³	First ever ischaemic strokes	ADLs function	14,529 (all hospitals in Sweden)	Antiplatelet therapy for IS	3-month case fatality	0.83	0.68–1.01
				ACE inhibitors therapy		1.00	0.87–1.14
				Statins therapy		0.78	0.67–0.91
				Anticoagulants therapy for IS with AF		0.58	0.44–0.76
UK (England)²⁴	Ischaemic stroke	Level of consciousness and neurological deficit	36,197 (106 hospitals)	Seen by a stroke consultant or associate specialist within 24 h	30-day case fatality	0.88	0.80–0.97
				Brain scan within 24 h		0.96	0.86–1.07
				Bundle 1: seen by nurse and one therapist within 24 h and all relevant therapists within 72 h		0.90	0.82–0.99
				Bundle 2: nutrition screening and formal swallow assessment within 72 h where appropriate		0.76	0.67–0.87
				Bundle 3: patient’s first ward of admission was stroke unit and they arrived there within 4 h of hospital admission		0.99	0.90–1.08
				Bundle 4: patient given antiplatelet therapy where appropriate and had adequate fluid and nutrition for first 72 h		0.46	0.42–0.50
				Number of criteria fulfilled
				Quality score 5 or 6 v 0–4		0.74	0.66–0.83
UK (Scotland)²⁵	All stroke types	SSV	36,055 (36 hospitals)	Stroke unit on day 0 or 1	6-month case fatality	0.79	0.74–0.85
				Swallow screen on day 0		0.95	0.86–1.04
				Brain scan on day 0		0.95	0.88–1.03
				Aspirin on day 0 or 1		0.54	0.49–0.58
				Number of criteria fulfilled	6-month case fatality
				0 vs 4		2.26	1.60–3.21
				1 vs 4		1.67	1.45–1.93
				2 vs 4		1.44	1.31–1.59
				3 vs 4		1.17	1.08–1.27
				Number of criteria fulfilled	Discharge to home/usual residence at 6 months
				0 vs 4		0.70	0.50–0.98
				1 vs 4		0.74	0.65–0.84
				2 vs 4		0.84	0.76–0.91
				3 vs 4		0.91	0.85–0.98
UK (Scotland)²⁶	All stroke types	SSV	41,692 (36 hospitals)	Admission to stroke unit	1-year survival	1.43	2.71–3.56
UK (Scotland)²⁶	All stroke types	SSV	41,692 (36 hospitals)		6-month discharged home	1.19	1.11–1.28
Italy²⁷	All stroke types	Level of consciousness	11,572 (424 stroke units and 260 hospitals)	Stroke unit	2-year case fatality	0.79	0.68–0.91
					2-year death or disability (mRS > 2)	0.81	0.72–0.91
					2 year not living at home	0.85	0.74–0.97
Spain²⁸	SAH excluded	NIHSS	1767 (47 hospitals)	Brain imaging < 24 h	1-year case fatality risk for noncompliance	1.4	0.71–2.76
				Screening of dysphagia		1.23	0.88–1.71
				Antiplatelets < 48 h for IS		1.3	0.84–2.02
				Early mobilisation		1.54	1.05–2.24
				Assessment of rehabilitation needs		1.48	1.06–2.07
				DVT prevention		0.98	0.60–1.60
				Management of hyperthermia		0.67	0.25–1.79
				Management of hypertension		1.87	1.22–2.86
				Management of dyslipidemia		1.29	0.86–1.93
				Anticoagulants for IS with AF		1.70	0.95–3.05
				Antithrombotics at discharge (IS)		2.79	1.41–5.54
Greece²⁹	First-ever acute ischaemic stroke	SSS	794 (different Athenian hospitals)	Statin at discharge	10-year case fatality	0.43	0.29–0.61
Greece²⁹	First-ever acute ischaemic stroke	SSS	794 (different Athenian hospitals)		10-year stroke recurrence	0.65	0.39–0.97
USA³⁰	Ischaemic stroke	NIHSS	1363 (5 hospitals)	Neurology assessment	In-hospital mortality, discharge to hospice, or discharge to a skilled nursing facility	1.13	0.59–2.17
				Swallowing evaluation		0.64	0.43–0.94
				DVT prophylaxis		0.60	0.37–0.96
				Early mobilisation		0.69	0.42–1.14
				Blood pressure management		1.00	0.67–1.50
				Fever management		0.71	0.35–1.41
				Hypoxia management		0.26	0.09–0.73
USA³¹	All stroke types	Weakness and altered level of consciousness	18,017(222 hospitals from 6 States)	Dysphagia screening	Higher risk of pneumonia for no screening	2.15	1.74–2.66
Canada³²	Ischaemic stroke	CNS	3631 (11 hospitals)	OCI^c 1 vs 0	1-year case fatality	0.69	0.44–1.09
				OCI 2 vs 0		0.39	0.25–0.62
				OCI 3 vs 0		0.40	0.25–0.64
				Antithrombotic therapy		0.33	0.22–0.50
Canada³³	Intracerebral haemorrhage stroke	CNS	2466 (11 hospitals)	Statin use in hospital	6 months case fatality	0.2	0.1–0.3
	Intracerebral haemorrhage stroke	CNS			Poor outcome (mRS4–6.) at discharge	0.6	0.4–0.9
Canada³⁴	Ischaemic stroke	CNS	6223 (12 centres)	OCI^c 2–3 vs 0–1	30-day case fatality	0.23	0.19–0.28
Chile³⁵	Ischaemic stroke	Aphasia, hemiplegia, reduced level of consciousness, and speech disturbance	677 (7 hospitals)	Neurological evaluation on admission	30-day case fatality	2.02	0.77–5.30
					In-hospital pneumonia	1.07	0.79–1.44
				Dysphagia screening within 48 h	30-day case fatality	0.52	0.26–1.04
					In-hospital pneumonia	1.58	0.60–4.15
Australia³⁶	SAH excluded	GCS, ability to lift both arms, ability to walk, and urinary incontinence	468 (8 hospitals)	Thorough (n-l ≤ 1)^d adherence to 15 processes of stroke care^e	Independent at 28 weeks	1.78	0.93–3.38
					Being at home at 28 weeks	1.69	0.86–3.32
					Alive at 28 weeks	2.10	0.92–4.82

				Complete (n-l ≤ 0) adherence to 15 processes of stroke care	Independent at 28 weeks	2.61	0.96–7.10
					Being at home at 28 weeks	3.09	0.96–9.87
					Alive at 28 weeks	3.22	0.66–15.86

Australia³⁷	All stroke types	FIM	2119 (108 rehabilitations units)	ADLs rehabilitation	Discharged home (Median LoS = 26 days)	1.01	0.33–3.13
				DVT prevention		0.58	0.41–0.81
				Home assessment		6.15	3.70–10.22
				Balance rehabilitation		0.54	0.35–0.83
				Secondary prevention on discharge^f		1.99	1.12–3.53
				Education to patients^g		2.37	1.30–4.29
				Discussing post-discharge needs with patients		1.27	0.66–2.43
Australia³⁸	All stroke types	Ability to walk on admission	16,665 (42 hospitals)	1 process received vs 0	180-day case fatality	0.63	0.41–0.97
				2 processes received vs 0		0.46	0.31–0.68
				3 processes received vs 0		0.30	0.18–0.47
				1 process received vs 0	90 - to 180-day Quality of Life (QoL)	12.53	−2.22 to 27.28
				2 processes received vs 0		16.67	0.30–33.05
				3 processes received vs 0		18.70	1.86–35.55
New Zealand³⁹	All stroke types	Age, initial FIM, pre-stroke FIM, and being European	181 (3 hospitals)	Swallowing assessment	1-year poor outcome (death or moved from home) for swallowing recorded ‘no’	3.2	0.97–10.7
China⁴⁰	Ischaemic stroke	NIHSS	1951 (23 hospitals in 11 major cities of China)	Antiplatelet therapy for IS	1-year case fatality	0.42	0.21–0.86
					Recurrent cerebrovascular event	0.58	0.36–0.92
					Functional improvement	1.25	1.02–1.52

China⁴¹	First ever ischaemic stroke	NIHSS	7455 (132 hospitals)	Stain use during hospitalisation	3-month case fatality	0.51	0.38–0.67
China⁴¹	First ever ischaemic stroke	NIHSS			3-month dependency	0.95	0.81–1.11
China⁴²	Intracerebral haemorrhage stroke	NIHSS	3218 (132 hospitals)	Stain use during hospitalisation	1-year case fatality	0.49	0.27–0.86
China⁴²	Intracerebral haemorrhage stroke	NIHSS	3218 (132 hospitals)		1-year good functional outcome	2.04	1.37–3.06
Thailand⁴³	Ischaemic stroke	NIHSS	1222 (76 hospitals)	Stroke unit admission	Poor outcome (mRS 5–6 at discharge) (LoS = 4 days)	0.54	0.33–0.87
				Thrombolysis		0.09	0.03–0.23
				Aspirin within 48 h		1.25	0.73–2.15
Taiwan⁴⁴	All stroke types	NIHSS	30,599 (39 hospitals)	IV tPA for 2 h	6-month functional dependency (mRS ≥ 2)	0.52	0.35–0.76
				Antithrombotics at discharge	6-month risk of cardiovascular events and death	0.41	0.35–0.47
				Anticoagulation for IS with AF at discharge		0.59	0.44–0.80
				Lipid-lowering agents at discharge		0.94	0.78–1.13

Organised stroke care included wards which encompassed multidisciplinary team-working, a physician with an interest in stroke, as well as taking into account the proportion of time spent (>50% of their length of stay) in such an environment. The wards that encompassed organised stroke care included neurology, elderly care, stroke specific unit and intensive care unit.

The complications that were considered in the analysis included pneumonia, urinary tract infection, pressure ulcer, falls, venous thromboembolism and constipation.

OCI is a summary score based on the presence of occupational therapy or physiotherapy, stroke team assessment and admission to a stroke unit. A score of zero indicates that stroke patients received none of these services, and higher scores indicate access to more services. The ‘organised care’ index was classified as having received 0, 1, 2 or 3 services.

n indicates number of applicable PoC; i, number of PoC adhered to.

The 15 processes of care consisted of CT scan < 24 h since admission, swallow < 24 h since admission, allied health < 24 h since admission, incontinence addressed, discharged on antiplatelet agent, fever > 38.5 managed, documented premorbid function, documented discharge needs, regular neurology observations for the first 24 h of admission, physiotherapist within 24 h, occupational therapist within 24 h, speech pathologist within 24 h, enteric feeding if nil by mouth > 48 h, aspiration avoidance and DVT prophylaxis if not ambulant.

Secondary prevention included deep vein thrombosis prophylaxis, discharged on lipid-lowering medication, discharged on blood-pressure–lowering medication and discharged on antithrombotics.

Education to patients consisted of lifestyle advice, information on sexuality poststroke, information about peer support, information on self-management programs, carer training and providing contact to patient.

ACE: angiotensin-converting enzyme; ADLs: Activities of Daily Living; AF: atrial fibrillation; BI: Barthel Index; CNS: Canadian neurological scale; CT: computerised tomography; MRI: magnetic resonance imaging; CI: confidence interval; DVT: deep vein thrombosis; FIM: functional independence measure; GCS: Glasgow coma scale; HR, hazard ratio; IS: ischaemic stroke; LoS: length of hospital stay; mRS: modified Rankin Scale; NIHSS: national institute of health stroke scale; PT: physiotherapy; OR: odds ratio; OT: occupational therapy; RR: rate ratio; SAH: subarachnoid haemorrhage; SSS: Scandinavian stroke scale; SSV: six simple variable; OCI: organised care index; PoC: processes of care.

The majority (23/30) of the included studies used prospective recruitment while the rest^{25,26,28–30,35,37} consisted of retrospective audits. Thirteen^{16,19,20,23,24,29,30,32,34,35,40,41,43} included only patients with ischaemic stroke, and two^33,42 included only patients with intracerebral haemorrhage. The remainder included both ischaemic and haemorrhagic stroke. Among those studies that included both types of stroke, six^{15,17,18,22,28,36} excluded patients with subarachnoid haemorrhage.

For the association between KPIs and patient outcomes, the majority²² of the included studies used OR, six studies^{16,19,23,29,38,40} used HR while the remaining two^17,18 used rate RR. The included studies also used different measures for stroke severity as a case mix variable for adjustment to estimate the independent association between a KPI and a patient outcome. Twenty of the included studies used validated tools including National Institute of Health Stroke Scale,^{28,30,40–44} Scandinavian Stroke Scale,^16–21,29 Canadian Neurological Scale,^32–34 Six Simple Variables^25,26 and Glasgow Coma Scale,³⁶ while the remainder used stroke severity proxies such as level of consciousness, incontinence, dysphagia, dysphasia, paralysis and disability.

Reporting of published KPIs

As there were some variations in data definitions and analysis methods, several assumptions were made to allow easy comparison between the studies:

Swallow/nutritional assessment – This single KPI comprised an assessment of swallowing, dysphagia and/or nutritional risk. If separate data for both swallow and nutritional risk assessment^18,21 were reported, we preferentially included data for swallow assessment.

Antiplatelet drugs for ischaemic stroke (IS) – Aspirin administration reported in two studies^25,43 was combined with a KPI for antiplatelet drugs for ischaemic stroke reported in seven studies.^{17,18,21,23,24,28,40}

Early nurse/rehabilitation assessment – This combined indicator of early assessment by a nurse²⁴ and early assessment of rehabilitation needs.²⁸

Early physiotherapy/mobilisation – This combined five reports of early mobilisation^18,21,28,30 with one¹⁷ about early physiotherapy assessment.

Selection of outcome measures

As there were minor variations in the approach to outcome analysis adjustments were made to the reported OR, HR, RR and CI to allow comparisons between the studies. The online supplement Table S1 provides a summary of the adjustments made.

Data reported in terms of poor outcome,^33,39,43 disability,^15,22,41,44 death or disability^17,30 or not returning home²⁷ post stroke were all combined as a ‘poor outcome’ post stroke. Finally, the results on the association between KPIs and stroke case fatality were categorised at the end of scheduled follow-up although the timing of follow-up was included in sensitivity analyses.

KPIs

There were 25 reported KPIs in total. The KPIs that were reported by at least a quarter of the eligible studies were swallow/nutritional assessment, stroke unit admission and antiplatelets for ischaemic stroke.

Stroke unit admission was variably defined across the related studies.^{15,17,18,21,22,24–27,43} Two Danish studies^17,18 defined a ‘stroke unit’ as a hospital department/unit that exclusively or primarily is dedicated to patients with stroke and which is characterised by multidisciplinary teams, a staff with a specific interest in stroke, involvement of relatives and continuous education of the staff. In the Italian study,²⁷ stroke unit was defined as a hospital ward with dedicated beds (at least 80% stroke admission) and with a dedicated stroke staff (at least one physician and one nurse) who work exclusively in the care of stroke patients.

The online supplement Table S2 provides a list of reported KPIs and their frequencies out of the 30 studies. Table 2 indicates the reported KPIs and their association with patient outcomes.

Table 2.

Reported KPIs and their association patient outcomes.

KPI		Study	Treatment time	End follow-up period	OR/HR/RR	95% CI
1. Reported KPIs and their association with case fatality
CT/MRI brain imaging		Europe¹⁵		3 months	0.70	0.40–1.30
		Denmark¹⁷	1st day of LoS	3 months	1.35	1.24–1.46
		UK (England)²⁴	≤24 h	1 month	0.96	0.86–1.07
		UK (Scotland)²⁵	day 0	6 months	0.95	0.88–1.03
		Spain²⁸	<24 h	1 year	0.71	0.36–1.41
Neurological assessment		Chile³⁵	On admission	1 month	2.02	0.77–5.30
Thrombolysis		Denmark¹⁹		1.4 year	0.66	0.49–0.88
Stroke unit admission		Europe¹⁵		3 months	0.50	0.30–0.80
		Denmark¹⁷	2nd day of LoS	3 months	0.76	0.69–0.83
		Sweden²²		2 years	0.81	0.72–0.92
		UK (England)²⁴	≤4 h	1 month	0.99	0.90–1.08
		UK (Scotland)²⁵	day 0 or 1	6 months	0.79	0.74–0.85
		UK (Scotland)²⁶		1 year	0.70	0.65–0.75
		Italy²⁷		2 years	0.79	0.68–0.91
Swallow/nutritional assessment		Denmark¹⁷	2nd day of LoS	3 months	0.69	0.61–0.76
		UK (England)²⁴	≤72 h	1 month	0.76	0.67–0.87
		UK (Scotland)²⁵	day 0	6 months	0.95	0.86–1.04
		Spain²⁸		1 year	0.81	0.58–1.14
		Chile³⁵	≤48 h	1 month	0.52	0.26–1.04
Antiplatelets for IS		Denmark¹⁷	2nd day of LoS	3 months	0.71	0.62–0.81
		Sweden²³	–	3 months	0.83	0.68–1.01
		UK (England)²⁴	≤72 h	1 month	0.46	0.42–0.50
		UK (Scotland)²⁵	day 0 or 1	6 months	0.54	0.49–0.58
		Spain²⁸	< 48 h	1 year	0.77	0.50–1.19
		China⁴⁰	LoS	1 year	0.42	0.21–0.86
Anticoagulants for IS with AF		Denmark¹⁶	Acute LoS	4 years	0.52	0.40–0.69
		Denmark¹⁷	By 14th day	3 months	0.41	0.31–0.52
		Sweden²³	–	3 months	0.58	0.44–0.76
		Spain²⁸		1 year	0.59	0.33–1.05
Blood pressure lowering therapy		Sweden²³	–	3 months	1.00	0.87–1.14
		Spain²⁸		1 year	0.53	0.35–0.82
Hyperthermia management		Spain²⁸		1 year	1.50	0.56–4.00
Lipid management		Sweden²³	–	3 months	0.78	0.67–0.91
		Spain²⁸		1 year	0.78	0.52–1.16
		Greece²⁹	At discharge	10 years	0.43	0.29–0.61
		Canada³³	Acute LoS	6 months	0.2	0.1–0.3
		China⁴¹	LoS	3months	0.51	0.38–0.67
		China⁴²	Acute LoS	1 year	0.49	0.27–0.86
DVT prophylaxis		Spain²⁸		1 year	1.02	0.63–1.67
		Canada³²	Acute LoS	1 year	0.33	0.22–0.50
Early medical assessment		UK (England)²⁴	≤24 h	1 month	0.88	0.80–0.97
Early nurse/rehabilitation assessment		UK (England)²⁴	≤24 h	1 month	0.90	0.82–0.99
		Spain²⁸		1 year	0.68	0.48–0.94
Early physiotherapy/mobilisation		Denmark¹⁷	2nd day of LoS	3 months	0.81	0.73–0.88
		Spain²⁸		1 year	0.65	0.45–0.95
Occupational therapy assessment		Denmark¹⁷	2nd day of LoS	3 months	0.83	0.75–0.91
Antidepressant therapy		Denmark²⁰	LoS	1 month	0.28	0.18–0.43
2. Reported KPIs and their association with poor outcome
CT/MRI brain imaging	Europe¹⁵			3 months	1.45	0.39–7.4
Thrombolysis	Thailand⁴³			LoS = 4 days	0.09	0.03–0.23
	Taiwan⁴⁴		3 h of onset	6 months	0.52	0.35–0.76
Neurological Assessment	USA³⁰			LoS	1.13	0.59–2.17
Stroke unit admission	Thailand⁴³			LoS = 4 days	0.54	0.33–0.87
	Europe¹⁵			3 months	1.3	0.6–1.76
	Italy²⁷			2 years	0.85	0.74–0.97
	UK (Scotland)²⁶			6 months	0.84	0.78–0.90
	Sweden²²			2 years	0.79	0.66–0.94
Swallow/nutritional assessment	Thailand⁴³			LoS = 4 days	0.54	0.33–0.87
	New Zealand³⁹			1 year	0.31	0.09–1.03
	USA³⁰			LoS	0.64	0.43–0.94
Antiplatelets for IS	China⁴⁰		LoS	1 year	0.80	0.66–0.98
	Thailand⁴³		48 h	LoS = 4 days	1.25	0.73–2.15
Blood pressure lowering therapy	USA³⁰			LoS	1.00	0.67–1.50
Hyperthermia management	USA³⁰		All episodes	LoS	0.71	0.35–1.41
Hypoxia management	USA³⁰		All episodes	LoS	0.26	0.09–0.73
DVT Prophylaxis	USA³⁰			LoS	0.60	0.37–0.96
	Australia³⁷			26 days	1.72	1.23–2.44
Early physiotherapy/mobilisation	USA³⁰			LoS	0.69	0.42–1.14
ADLs rehabilitation	Australia³⁷			26 days	0.99	0.32–3.03
Home assessment	Australia³⁷			26 days	0.16	0.10–0.27
Balance rehabilitation	Australia³⁷			26 days	1.85	1.20–2.86
Secondary prevention on discharge	Australia³⁷			26 days	0.50	0.28–0.89
Education to patients	Australia³⁷			26 days	0.42	0.23–0.77
Discussing post-discharge needs with patients	Australia³⁷			26 days	0.79	0.41–1.52
Lipid management	China⁴¹		Acute LoS	3 months	0.95	0.81–1.11
	China⁴²		Acute LoS	1 year	0.49	0.33–0.73
	Canada³³		Acute LoS	At discharge	0.6	0.4–0.9
3. Reported KPIs and their association with prolonged length of hospital stay
Stroke unit admission	Denmark¹⁸		2nd day		0.71	0.65–0.77
Antiplatelets for IS	Denmark¹⁸		2nd day		0.80	0.73–0.87
Anticoagulants for IS with AF	Denmark¹⁸		14th day		0.78	0.62–0.98
CT/MRI brain imaging	Denmark¹⁸		2nd day		0.82	0.74–0.91
Swallow/nutritional assessment	Denmark¹⁸		2nd day		0.78	0.69–0.87
Constipation risk assessment	Denmark¹⁸		2nd day		0.70	0.63–0.78
Early physiotherapy/mobilisation	Denmark¹⁸		2nd day		0.67	0.61–0.73
Occupational therapy assessment	Denmark¹⁸		2nd day		0.85	0.80–0.91
Intermittent catheterisation	Denmark¹⁸		2nd day		0.77	0.64–0.92
DVT prophylaxis	Denmark¹⁸		2nd day		0.82	0.71–0.95
4. Reported KPIs and their association with medical complications
CT/MRI brain imaging	Denmark²¹			LoS = 13 days	1.52	1.35–1.72
Neurological assessment	Chile³⁵		On admission	30 days	1.07	0.79–1.44
Stroke unit admission	Denmark²¹			LoS = 13 days	0.79	0.68–0.92
Swallow/nutritional assessment	Chile³⁵		≤48 h	30 days	1.58	0.60–4.15
	Denmark²¹			LoS = 13 days	0.97	0.84–1.11
	USA³¹			LoS = 5 days	0.47	0.38–0.57
Antiplatelets for IS	Denmark²¹			LoS = 13 days	0.95	0.79–1.15
Anticoagulants for IS with AF	Denmark²¹			LoS = 13 days	0.59	0.45–0.76
Early physiotherapy/mobilisation	Denmark²¹			LoS = 13 days	0.43	0.35–0.53
Occupational therapy assessment	Denmark²¹			LoS = 13 days	1.10	0.94–1.27
Thrombolysis	Denmark¹⁹			1.4 years	0.59	0.24–1.47
5. Reported KPIs and their association with stroke recurrence
Antiplatelets for IS	China⁴⁰		LoS	12 months	0.58	0.36–0.92
Anticoagulants for IS with AF	Taiwan⁴⁴		At discharge	6 months	0.59	0.44–0.80
Lipid management	Taiwan⁴⁴		At discharge	6 months	0.94	0.78–1.13
	Greece²⁹		At discharge	10 years	0.65	0.39–0.97
DVT prophylaxis	Taiwan⁴⁴		At discharge	6 months	0.41	0.35–0.47
Thrombolysis	Denmark¹⁹			1.4 year	1.05	0.68–1.64

ADLs: activities of daily living; AF: atrial fibrillation; CT: computerised tomography; MRI: magnetic resonance imaging; CI: confidence interval; DVT: deep vein thrombosis; HR: hazard ratio; IS: ischaemic stroke; KPI: key performance indicator; LoS: length of hospital stay; OR: odds ratio; RR: rate ratio.

Association between individual KPIs and lower risk for case fatality at the end of scheduled follow-up

The median time of scheduled follow-up for the studies reporting on the association between individual KPIs and case fatality was 1 year; range from 1 month to 10 years. Significant reductions in case fatality were observed across multiple studies for stroke unit admission,^{15,17,22,25–27} swallow/nutritional assessment,^17,24 antiplatelets for ischaemic stroke,^17,40,24,25 anticoagulants for ischaemic stroke with atrial fibrillation,^16,17,23 lipid management,^{23,29,33,41,42} early nurse/rehabilitation assessment^24,28 and early physiotherapy/mobilisation.^17,28 In addition, significant associations within single studies were observed for DVT prophylaxis³² and blood pressure lowering therapy.²⁸

In contrast, several studies reported wide CIs and no statistically significant association between the reported KPIs and stroke case fatality; stroke unit admission,²⁴ swallow/nutritional assessment,^25,28,35 antiplatelets for ischaemic stroke,^23,28 anticoagulants for ischaemic stroke with atrial fibrillation, lipid management,²⁸ DVT prophylaxis²⁸ and blood pressure lowering therapy.²³ In one study,¹⁷ the CT/MRI brain imaging was associated with increased risk of early case fatality (RR: 1.35, 95% CI: 1.24–1.46), while in other studies,^15,24,25,28 no evidence for an association of CT/MRI brain imaging and stroke case fatality was found.

Figure 2 summarises the primary meta-analysis results regarding the associations between individual KPIs and stroke case fatality at the end of follow-up. The KPIs that were associated with lower risk for case fatality include stroke unit admission (OR: 0.79, 95% CI: 0.72–0.87), swallow/nutritional assessment (OR: 0.78, 95% CI: 0.66–0.92), antiplatelets for ischaemic stroke (OR: 0.61, 95% CI: 0.50–0.74), anticoagulants for ischaemic stroke with atrial fibrillation (OR: 0.51, 95% CI: 0.43–0.61), lipid management (OR: 0.52, 95% CI: 0.38–0.71) and early physiotherapy/mobilisation (OR: 0.78, 95% CI: 0.67–0.91). However, the significant associations of stroke unit admission, swallow/nutritional assessment, antiplatelets for ischaemic stroke and lipid management were complicated by substantial heterogeneity (I²> 50%). When analysed at a fixed time point, swallow/nutritional assessment (OR: 0.72, 95% CI: 0.66–0.79), antiplatelets for ischaemic stroke (OR: 0.64, 95% CI: 0.44–0.93) and lipid management (OR: 0.64, 95% CI: 0.42–0.97) were associated with a lower risk for early case fatality (up to 3 months post stroke), but the heterogeneity was reduced for swallow/nutritional assessment (I²= 1%) only. Stroke unit admission (OR: 0.77, 95% CI: 0.71–0.82), antiplatelets for ischaemic stroke (OR: 0.57, 95% CI: 0.45–0.72) and lipid management (OR: 0.45, 95% CI: 0.27–0.74) were associated with lower risk for late case fatality (beyond 3 months post stroke), but the heterogeneity was reduced for antiplatelets for ischaemic stroke (I²= 34%) only.

Figure 2.

Association between individual KPIs and lower risk for case fatality. AF: atrial fibrillation; CT: computerised tomography; MRI: magnetic resonance imaging; CI: confidence interval; DVT: deep vein thrombosis; I²: heterogeneity; IS: ischaemic stroke; IV: inverse variance; KPI: key performance indicator; N: number of.

The meta-analysis showed no evidence for the association between the stroke case fatality and DVT prophylaxis, blood pressure lowering therapy, early nurse/rehabilitation assessment and CT/MRI brain imaging.

The sensitivity analysis excluding those that used HR or RR produced results that were similar to those in Figure 2 (data not shown): stroke unit admission (OR: 0.79, 95% CI: 0.71–0.89), swallow/nutritional assessment (OR: 0.82, 95% CI: 0.69–0.98), antiplatelets for ischaemic stroke (OR: 0.53, 95% CI: 0.44–0.63) and lipid management (OR: 0.47, 95% CI: 0.30–0.74) remained associated with lower risk for case fatality, and there was no evidence for the association between the stroke case fatality and DVT prophylaxis, early nurse/rehabilitation assessment and CT/MRI brain imaging.

Significant associations within single studies were observed for thrombolysis,¹⁹ early medical assessment,²⁴ OT assessment¹⁷ and antidepressant therapy,²⁰ but there was no evidence for the association between stroke case fatality and hyperthermia management,²⁸ and neurological assessment.³⁵

Association between individual KPIs and the risk for poor outcome

For studies reporting on the association between individual KPIs and poor outcome, the available follow-up periods were between 4 days and 2 years, with a mean of 282 days.

KPIs that were reported to be associated with the lower risk for poor outcome included thrombolysis,^43,44 stroke unit admission,^22,26,27,43 swallowing/nutritional assessment,^30,43 antiplatelets for ischaemic stroke,⁴⁰ DVT prophylaxis,³⁰ and lipid management management.^33,42 However, some studies found no evidence of an association with poor outcome and stroke unit admission¹⁵; swallowing/nutritional assessment,³⁹ antiplatelets for ischaemic stroke,⁴³ DVT prophylaxis³⁷ and lipid management.⁴¹

As summarised in Figure 3, the meta-analysis showed that the KPIs associated with the lower risk for poor outcome were stroke unit admission (OR: 0.83, 95% CI: 0.77–0.89) and swallowing/nutritional assessment (OR: 0.58, 95% CI: 0.43–0.78), while there was no evidence for the association with poor outcome for thrombolysis, antiplatelets for ischaemic stroke, DVT prophylaxis and lipid management.

Figure 3.

Association between individual KPIs and lower risk for poor outcome. ADLs: activities of daily living; CT: computerised tomography; MRI: magnetic resonance imaging; CI: confidence interval; DVT: deep vein thrombosis; I²: heterogeneity; IS: ischaemic stroke; IV: inverse variance; KPI: key performance indicator; N: number of.

Several individual studies reported significant associations between lower risk for poor outcome and hypoxia management³⁰; home assessment, secondary prevention on discharge and education to patients.³⁷ No association with poor outcome was found for CT/MRI brain imaging¹⁵; neurological assessment, blood pressure lowering therapy, hyperthermia management and early physiotherapy/mobilisation³⁰; ADLs rehabilitation, balance rehabilitation and discussing post-discharge needs with patients.³⁷

All the studies included for the primary meta-analysis about the association of KPIs with poor outcome used ORs, except one Chinese study.⁴⁰ After excluding that study, antiplatelets for ischaemic stroke remained with a single study,⁴³ which showed no association with poor outcome (OR: 1.25, 95% CI: 0.73–2.14).

Association between individual KPIs and relative length of hospital stay

A single Danish study¹⁸ reported that a shorter relative length of hospital stay was associated with stroke unit admission, antiplatelets and anticoagulants for ischaemic stroke with atrial fibrillation, CT/MRI brain imaging, early physiotherapy/mobilisation, occupational therapy assessment, swallowing/nutritional assessment and DVT prophylaxis, with rate ratio ranging from 0.67 (0.61–0.73) for early physiotherapy/mobilisation to 0.85 (0.80–0.91) for occupational therapy assessment.

Association between individual KPIs and the risk for medical complications and stroke recurrence

Stroke unit admission, anticoagulants for ischaemic stroke with atrial fibrillation and early physiotherapy/mobilisation,²¹ as well as swallow/nutritional assessment³¹ were found to be associated with lower risk for medical complications (OR: 0.79; 0.68–0.92; 0.59, 0.45–0.76 and 0.43, 0.35–0.53; 0.47, 0.38–0.57, respectively). By contrast, CT/MRI brain imaging was associated with a greater risk for medical complications with (1.52, 1.35–1.72).²¹ Other studies with wide CIs did not show evidence for the association between the occurrence of medical complications and neurological assessment³⁵; swallow/nutritional assessment^21,35; antiplatelets for ischaemic stroke, occupational therapy assessment²¹; and thrombolysis.¹⁹

KPIs that were reported to be associated with lower recurrence rate for stroke included antiplatelets for ischaemic stroke,⁴⁰ anticoagulants for ischaemic stroke with atrial fibrillation and DVT prophylaxis⁴⁴ and lipid management.²⁹ However, in one study,²¹ evidence for the association between lipid management and stroke recurrence was not found,⁴⁴ and there was no evidence of an association with thrombolysis.¹⁹

Association between adherence to groups of KPIs and the risk for case fatality

Seven studies^{17,24,25,32,34,36,38} had consistent findings whereby adherence to a combination of several KPIs (‘bundle’) was associated with a greater decrease in stroke mortality. A lower risk for poor outcome was also reported when full stroke care bundle was achieved.^25,36 An Australian study³⁸ also showed that achieving full care bundle was associated with better quality of life at 3 to 6 months post stroke. Increased adherence to stroke care KPIs¹⁸ was associated with shorter length of hospital stay (data are not shown in Table 3).

Table 3.

Association between the number of KPIs achieved and patient outcomes.

	Case fatality				Poor outcome				Quality of life
Study	FU period	Number of processes	HR	95% CI	FU period	Number of processes	OR/HR/ RR	95% CI	FU period	Number of processes	OR/HR/ RR	95% CI
Denmark¹⁷	3 months	1 vs 0	0.94	0.65–1.49
		2 vs 0	0.78	0.54–1.02
		3 vs 0	0.60	0.42–0.78
		4 vs 0	0.61	0.42–0.79
		5 vs 0	0.45	0.31–0.60
		6 vs 0	0.48	0.31–0.65

UK (England)²⁴	1 month	5–6 vs 0–4	0.74	0.66–0.83

UK (Scotland)²⁵	6 months	0 vs 4	2.26	1.60–3.21	6 months	0 vs 4	1.43	1.02–2.00
		1 vs 4	1.67	1.45–1.93		1 vs 4	1.35	1.19–1.54
		2 vs 4	1.44	1.31–1.59		2 vs 4	1.19	1.10–1.32
		3 vs 4	1.17	1.08–1.27		3 vs 4	1.10	1.02–1.18

Canada³²	12 months	OCI 1 vs 0	0.69	0.44–1.09
		OCI 2 vs 0	0.39	0.25–0.62
		OCI 3 vs 0	0.40	0.25–0.64

Australia³⁶	6 months	All or n-1	0.48	0.21–1.09	6 months	All or n-1	0.59	0.30–1.16
		All	0.31	0.06–1.52		All	0.32	0.10–1.04

Australia³⁸	6 months	1 vs 0	0.63	0.41–0.97					3–6 months	1 vs 0	12.53	−2.22 to 27.28
		2 vs 0	0.46	0.31–0.68						2 vs 0	16.67	0.30–33.05
		3 vs 0	0.30	0.18–0.47						3 vs 0	18.70	1.86–35.55

Canada³⁴	1 month	2–3 vs 0–1	0.23	0.19–0.28

CI: confidence interval; FU: follow-up; HR: hazard ratio; n: number of applicable processes of care; OCI: organised care index; OR: odds ratio; RR: rate ratio.

Overall (see Online Supplement Table 3), only stroke unit admission, swallow/nutritional assessment, antiplatelets for ischaemic stroke, anticoagulants for ischaemic stroke with atrial fibrillation, lipid management and early physiotherapy/mobilisation were found to be significantly associated with improved outcomes after a meta-analysis of two or more studies. Thrombolysis results were associated with reduced poor outcome in two studies, but the combined analysis was not significant due to substantial heterogeneity. Data were very limited for the outcomes of length of stay, stroke recurrence or medical complications.

Discussion

The publications we have reviewed provide a large and diverse body of evidence on whether quality of care, as measured by adherence with a KPI, is associated with improved clinical outcomes in patients hospitalised with stroke. Our primary meta-analysis indicated that several KPIs including stroke unit admission, swallowing/nutritional risk assessment, antiplatelets for ischaemic stroke, anticoagulants for ischaemic stroke with atrial fibrillation, lipid management and early physiotherapy/mobilisation were associated with a reduction in case fatality or poor outcome. However, although our meta-analysis showed significant associations between lower risk for case fatality and several individual KPs at the end of scheduled follow-up, there was substantial heterogeneity (I²> 50%) for stroke unit admission, swallowing/nutritional risk assessment, antiplatelets for ischaemic stroke and lipid management. Therefore, our meta-analysis results need to be interpreted with caution.

The strong association of stroke unit care with improved outcomes could be anticipated from a substantial number of RCTs.⁴⁵ Our review confirms this across a range of studies in routine care. Given the evidence for specialised multidisciplinary stroke unit care in stroke,⁴⁵ one might also expect to see benefits associated with early nurse or rehabilitation assessment and early medical assessment,²⁴ as well as occupational therapy assessment.¹⁷ These indicators lack direct evidence from randomised trials, but may possibly be markers for admission to a stroke unit and multidisciplinary stroke care. However, there were no comparable data from many studies about early medical assessment, early nurse or rehabilitation assessment or early occupational therapy assessment for our review.

Our finding of a reduced risk of case fatality after early physiotherapy/mobilisation was in accordance with the literature about stroke unit care,⁴⁵ and some small RCTs⁴⁶ but not consistent with recent RCTs of very early mobilisation.⁴⁷ However, the recent AVERT trial tested mobilisation at an earlier stage than in routine care, so the optimal timing of mobilisation remains unclear, and very early intensive mobilisation within 24 h may carry some hazard.⁴⁷

Our meta-analysis showed that swallow or nutritional assessment was associated with lower risk for both mortality and disability post stroke. This finding was consistent with a randomised controlled trial,⁹ which found that reinforcement of multidisciplinary management of swallowing dysfunction was significantly associated with lower risk for death or dependency. Thus, swallowing or nutritional assessment may be of paramount importance. The current meta-analysis also showed that early antiplatelet use for ischaemic stroke was associated with reduction in case fatality, and this was consistent with the results from a previous systematic review⁴⁸ of eight randomised trials. It showed that early antiplatelet therapy was associated with mortality reduction at a final follow-up between 1 and 6 months. However, our review showed greater apparent benefit than the 8% reduction in case fatality that was reported in the review of randomised trials.⁴⁸ However, a recent individual patient data meta-analysis of aspirin trials⁴⁹ confirms an important short-term benefit of aspirin therapy in preventing recurrent cerebral ischemia and that benefits may be greater than previously estimated. Our meta-analysis finding of a reduced risk of stroke case fatality associated with lipid management was consistent with the results from a meta-analysis⁵⁰ of 42 randomised trials.

One major disagreement with the RCTs is that our meta-analysis also showed that early anticoagulant use for ischaemic stroke with atrial fibrillation was associated with a reduction in early and late case fatality. However, this finding was not supported by a review⁵¹ of 24 randomised clinical trials. This disagreement may be explained by the participants’ inclusion criteria. In fact, while the randomised trials included in the review⁵¹ recruited patients with presumed or confirmed ischaemic stroke, the studies included in our review considered only patients with ischaemic stroke and atrial fibrillation. Additionally, as the studies included in our review were neither randomised nor blinded, the apparent effects of antiplatelets and anticoagulants for ischaemic stroke may have been overestimated due to selection bias and incomplete adjustment for confounders. Alternatively, KPIs may also reflect other important and unmeasured aspects of care, which would not be tested in a well-designed RCT. Additionally, the duration of follow-up for the studies included in our meta-analysis varied between 3 and 48 months (mean: 16.5 ± 21.4 months) while the duration of follow-up in the trials was generally shorter. This short-term follow-up may lead to missing a significant proportion of deaths that occur after 1 month, and disability is best assessed between 3 to 6 months when most of the recovery has taken place.⁵¹

Our review has also identified some areas with inconsistent evidence of the association of KPIs with outcome. Deep vein thrombosis (DVT) prophylaxis was found to be associated with significant benefits in studies in Canada³² and the USA³⁰ but not in Spain.²⁸ However, a meta-analysis of RCTs has failed to show improvements in survival or independence.⁵¹

Regarding thrombolysis, in two studies included for our review,^43,44 thrombolysis was associated with a lower risk for poor functional outcome, and this was consistent with the systematic review of the RCTs.⁵² However, because of high heterogeneity (I²= 88%) between the two studies reviewed, the summary effect was not statistically significant.

Our review showed that CT/MRI brain imaging and neurological assessment were not associated with any reported patient outcomes. This may be due to several reasons. First, the assessment itself, if not combined with adequate care, is unlikely to show any difference in outcome. For instance, once ischaemic stroke is diagnosed with brain imaging, further management by intravenous tissue plasminogen activator was found to be effective. It was however recently reported that only 3% of low-income, 19% of lower-middle-income, 33% of upper-middle-income and 50% of high-income-countries use it.⁵³ Second, the increased risk of early case fatality¹⁷ and medical complications²¹ that were reported after early CT/MRI brain imaging in two Danish studies was most likely due to reverse causality; patients who deteriorated during the first hours after hospitalisation were more likely to receive an early CT/MRI brain imaging, and also had a greater risk of death or medical complications.²¹ Third, some of the analyses of data may have been hampered by small sample sizes, and lack of statistical power to show the differential benefit.

Adherence to an individual measure in isolation may not have a clinically detectable impact on outcomes, making determination of an effect more difficult.⁵⁴ However, adherence to several KPIs was always associated with improved outcomes.

Strengths and weaknesses

Our systematic review has several strengths including searching a wide range of databases using standardised methodology. Furthermore, the review report was based on the PRISMA guidelines. The studies that were included in our review involved large sample sizes in general, allowing sufficient statistical power and enhancing the external validity of the results. One study¹⁵ was multinational, and 12 studies^{16,17,19,22,23,27,37,38,41–44} involved nationwide datasets. The remaining studies were conducted regionally with the recruitment of participants from between three³⁹ to 222 hospitals.³¹ Additionally, we only conducted analyses using data from studies that corrected for patient casemix (age and stroke severity). Our approach to meta-analysis has used a conservative random-effects approach to acknowledge the diversity of studies identified. Finally, we performed a sensitivity analysis to evaluate the robustness of our findings.

We must acknowledge some weaknesses. We did not use any scoring system to assess risk of bias in included studies, but simply included large register studies reporting independent association of KPIs with patient outcomes after adjusting at least two variables including age and stroke severity. Second, the review was based on data from observational studies with different follow-up time periods and designs. Third, although we have only included data that used a multivariable analysis to correct for patient casemix, there remains the possibility that the patient outcomes were influenced by unmeasured or residual confounding factors such as indication bias or factors related to the nonrandomised study design rather than the reported KPIs themselves. Fourth, our review could be subject to publication bias because our search strategy was limited to electronic databases and references known to the authors, and manuscripts published in English only. Fifth, there is a potential concern about combining results from studies from different settings and using different research methodologies. For instance, there were different measures for stroke severity for case mix adjustment, different models of stroke unit and different models of implementing or measuring the KPIs. Finally, we were limited to a few studies reporting data on important outcomes such as the length of hospital stay and quality of life, and none of the studies considered the cost of care, which is clearly important in a disabling condition such as stroke.

Conclusion

Our review found that the most frequently reported KPIs for stroke care were swallow/nutritional assessment, stroke unit admission, antiplatelets for ischaemic stroke, CT/MRI brain imaging, anticoagulants for ischaemic stroke with atrial fibrillation, lipid management, DVT prophylaxis and early physiotherapy/mobilisation. Stroke unit admission and early interventions including swallowing/nutritional risk assessment, antiplatelets for ischaemic stroke, anticoagulants for ischaemic stroke with atrial fibrillation, lipid management and early physiotherapy/mobilisation were all associated with better patient outcomes. Achieving a combination of several KPIs was always associated with a better outcome. Both policy makers and health care professionals should be encouraged to implement the KPIs for stroke management that are reliable and meaningful for regularly monitoring the quality of stroke care. Future research could focus on novel stroke care quality indicators, particularly in the post-acute period.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: DAC holds a Research Fellowship from the National Health and Medical Research Council (1063761 co-funded Heart Foundation), and has received educational grants from Boehringer Ingeheim for unrelated work.

Ethical approval

Not applicable.

Informed consent

Not applicable.

Guarantor

GU and PL.

Contributorship

GU and PL conceived the study, researched literature, analysed data and wrote the first draft of the manuscript. All authors reviewed and edited the manuscript and approved the final version of the manuscript.

References

Adams

Brott

et al.

Guidelines for the early management of patients with ischemic stroke. A scientific statement from the stroke council of the American Stroke Association. Stroke 2003; 34: 1056–1083.

American Heart Association (AHA). Measuring and improving quality of care: A report from the American Heart Association/American College of Cardiology First Scientific Forum on Assessment of Healthcare Quality in Cardiovascular Disease and Stroke. Circulation 2000; 101: 1483–1493.

Grol

Grimshaw

. From best evidence to best practice: effective implementation of change in patients’ care. Lancet 2003; 362: 1225–1230.

Grube

Dohle

Djouchadar

et al.

Evidence-based quality indicators for stroke rehabilitation. Stroke 2012; 43: 142–146.

Sandercock

Gubitz

Foley

et al.

Antiplatelet therapy for acute ischaemic stroke. Cochrane Database Syst Rev 2003; 2: CD000029.

Kwan

Sandercock

. In-hospital care pathways for stroke. Cochrane Database Syst Rev 2004; 4: CD002924.

Saxena

Koudstaal

. Anticoagulants for preventing stroke in patients with nonrheumatic atrial fibrilation and a history of stroke or transient ischemic attack. Cochrane Database Syst Rev 2004; 4: CD000187.

Goyal

Menon

van Zwam

et al.

Endovascular thrombectomy after large-vessel ischaemic stroke: a meta-analysis of individual patient data from five randomised trial. Lancet 2016; 387: 1723–1731.

Middleton

McElduff

Ward

et al.

Implementation of evidence-based treatment protocols to manage fever, hyperglycaemia, and swallowing dysfunction in acute stroke (QASC): a cluster randomised controlled trial. Lancet 2011; 378: 1699–1706.

10.

Cadilhac

Kim

Lannin

et al.

National stroke registries for monitoring and improving the quality of hospital care: A systematic review. Int J Stroke 2016; 11: 28–40.

11.

Quality of Care and Outcomes Research in CVD and Stroke Working Groups. Measuring and improving quality of care: A report from the American Heart Association/American College of Cardiology first scientific forum on assessment of healthcare quality in cardiovascular disease and stroke. Circulation 2000; 101: 1483–1493.

12.

Parker

Schwamm

Fonarow

et al.

Stroke quality metrics systematic reviews of the relationships to patient-centered outcomes and impact of public reporting. Stroke 2012; 43: 155–162.

13.

Reeves

Parker

Fonarow

et al.

Development of stroke performance measures: definitions, methods, and current measures. Stroke 2010; 41: 1573–1578.

14.

Moher

Shamseer

Clarke

et al.

Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015 statement. Syst Rev 2015; 4: 1–9.

15.

Bhalla

Grieve

Tilling

et al.

Older stroke patients in Europe: stroke care and determinants of outcome. Age Ageing 2004; 33: 618–624.

16.

Andersen

Olsen

. Reduced poststroke mortality in patients with stroke and atrial fibrillation treated with anticoagulants: results from a Danish quality-control registry of 22179 patients with ischemic stroke. Stroke 2007; 38: 259–263.

17.

Ingeman

Pedersen

Hundborg

et al.

Quality of care and mortality among patients with stroke: A nationwide follow-up study. Med Care 2008; 46: 63–69.

18.

Svendsen

Ehlers

Andersen

et al.

Quality of care and length of hospital stay among patients with stroke. Med Care 2009; 47: 575–582.

19.

Schmitz

Simonsen

Hundborg

et al.

Acute ischemic stroke and long-term outcome after thrombolysis: Nationwide propensity score-matched follow-up study. Stroke 2014; 45: 3070–3072.

20.

Mortensen

Johnsen

Larsson

et al.

Early antidepressant treatment and all-cause 30-day mortality in patients with ischemic stroke. Cerebrovasc Dis 2015; 40: 81–90.

21.

Ingeman

Andersen

Hundborg

et al.

Processes of care and medical complications in patients with stroke. Stroke 2011; 42: 167–172.

22.

Glader

Stegmayr

Johansson

et al.

Differences in long term outcome between patients treated in stroke units and in general wards: A 2-year follow-up of stroke patients in Sweden. Stroke 2001; 32: 2124–2130.

23.

Åsberg

Henriksson

Farahmand

et al.

Ischemic stroke and secondary prevention in clinical practice: A cohort study of 14 529 patients in the Swedish Stroke Register. Stroke 2010; 41: 1338–1342.

24.

Bray

Ayis

Campbell

et al.

Associations between the organisation of stroke services, process of care, and mortality in England: prospective cohort study. BMJ 2013; 346: f2827.

25.

Turner

Barber

Dodds

et al.

Implementing a simple care bundle is associated with improved outcomes in a national cohort of patients with ischemic stroke. Stroke 2015; 46: 1065–1070.

26.

Turner

Barber

Dodds

et al.

The impact of stroke unit care on outcome in a Scottish stroke population, taking into account case mix and selection bias. J Neurol Neurosurg Psychiatry 2015; 86: 314–318.

27.

Candelise

Gattinoni

Bersano

et al.

Stroke-unit care for acute stroke patients: an observational follow-up study. Lancet 2007; 369: 299–305.

28.

Abilleira

Ribera

Permanyer-Miralda

et al.

Noncompliance with certain quality indicators is associated with risk-adjusted mortality after stroke. Stroke 2012; 43: 1094–1100.

29.

Milionis

Giannopoulos

Kosmidou

et al.

Statin therapy after first stroke reduces 10-year stroke recurrence and improves survival. Neurology 2009; 72: 1816–1822.

30.

Bravata

Wells

et al.

Processes of care associated with acute stroke outcomes. Arch Intern Med 2010; 170: 804–810.

31.

Lakshminarayan

Tsai

Tong

et al.

Utility of dysphagia screening results in predicting poststroke pneumonia. Stroke 2010; 41: 2849–2854.

32.

Saposnik

Fang

O’Donnell

et al.

Escalating levels of access to in-hospital care and stroke mortality. Stroke 2008; 39: 2522–2530.

33.

Dowlatshahi

Demchuk

Fang

et al.

Association of statins and statin discontinuation with poor outcome and survival after intracerebral hemorrhage. Stroke 2012; 43: 1518–1523.

34.

Smith

Hassan

Fang

et al.

Do all ischemic stroke subtypes benefit from organized inpatient stroke care?

Neurology 2010; 75: 456–462.

35.

Hoffmeister

Lavados

Comas

et al.

Performance measures for in-hospital care of acute ischemic stroke in public hospitals in Chile. BMC Neurol 2013; 13: 23–32.

36.

Cadilhac

Ibrahim

Pearce

et al.

Multicenter comparison of processes of care between stroke units and conventional care wards in Australia. Stroke 2004; 35: 1035–1040.

37.

Hubbard

Harris

Kilkenny

et al.

Adherence to clinical guidelines improves patient outcomes in Australian audit of stroke rehabilitation practice. Archiv Phys Med Rehabil 2012; 93: 965–971.

38.

Cadilhac

Andrew

Lannin

et al.

Quality of acute care and long-term quality of life and survival: the Australian Stroke Clinical Registry. Stroke 2017; 48: 1026–1032.

39.

McNaughton

McPherson

Taylor

et al.

Relationship between process and outcome in stroke care. Stroke 2003; 34: 713–717.

40.

Ding

et al.

Association of antiplatelet therapy with lower risk of death and recurrent cerebrovascular events after ischemic stroke results from the China Ischemic Stroke Registry Study. Circ J 2009; 73: 2342–2347.

41.

Song

Wang

Zhao

et al.

Association between statin use and short-term outcome based on severity of ischemic stroke: A cohort study. PloS One 2014; 9: e84389.

42.

Pan

Jing

Wang

et al.

Use of statin during hospitalization improves the outcome after intracerebral hemorrhage. CNS Neurosci Therap 2014; 20: 548–555.

43.

Nilanont

Nidhinandana

Suwanwela

et al.

Quality of acute ischemic stroke care in Thailand: A prospective multicenter countrywide cohort study. J Stroke Cerebrovasc Dis 2014; 23: 213–219.

44.

Hsieh

Lien

Chen

et al.

Get with the guidelines-stroke performance indicators: Surveillance of stroke care in the Taiwan Stroke Registry-Get With The Guidelines-Stroke in Taiwan. Circulation 2010; 122: 1116–1123.

45.

Stroke Unit Trialists’ Collaboration. Organised inpatient (stroke unit) care for stroke. Cochrane Database Syst Rev 2013; 9: CD000197.

46.

Langhorne

Pollock

. What are the components of effective stroke unit care? Age Ageing 2002; 31: 365–371.

47.

Bernhardt

Langhorne

Lindley

et al.

Efficacy and safety of very early mobilisation within 24 h of stroke onset (AVERT): a randomised controlled trial. Lancet 2015; 386: 46–55.

48.

Sandercock

PAG

Counsell

Tseng

et al.

Oral antiplatelet therapy for acute ischemic stroke (Review). Cochrane Database Syst Rev 2014; 3: CD000029.

49.

Rothwell

Algra

Chen

et al.

Effects of aspirin on risk and severity of early recurrent stroke after transient ischemic attack and ischemic stroke: time-course analysis of randomised trials. Lancet 2016; 388: 365–375.

50.

O'Regan

Arora

et al.

Statin therapy in stroke prevention: a meta-analysis involving 121,000 patients. Am J Med 2008; 121: 24–33.

51.

Sandercock

PAG

Counsell

Kane

. Anticoagulants for acute ischemic stroke. Cochrane Database Syst Rev 2015; 3: CD000024.

52.

Wardlaw

Murray

Berge

et al.

Thrombolysis for acute ischemic stroke. Cochrane Database Syst Rev 2014; 7: CD000213.

53.

Berkowitz

Mittal

McLane

et al.

Worldwide reported use of IV tissue plasminogen activator for acute ischemic stroke. Int J Stroke 2014; 9: 349–355.

54.

Katzan

. Improvement in stroke performance measures: are we moving forward or in circles? Circ Cardiovasc Qual Outcomes 2011; 4: 493–495.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.27 MB