national stroke registries for monitoring and improving ... · systematic review national stroke...

Systematic Review

National stroke registries for monitoringand improving the quality of hospitalcare: A systematic review

Dominique A Cadilhac1,2, Joosup Kim1,2, Natasha A Lannin3,4,Moira K Kapral5, Lee H Schwamm6, Martin S Dennis7,Bo Norrving8 and Atte Meretoja2,9,10

Abstract

Background: Routine monitoring of the quality of stroke care is becoming increasingly important since patient out-

comes could be improved with better access to proven treatments. It remains unclear how many countries have

established a national registry for monitoring stroke care.

Aims: To describe the current status of national, hospital-based stroke registries that have a focus on monitoring access

to evidence-based care and patient outcomes and to summarize the main features of these registries.

Summary of review: We undertook a systematic search of the published literature to identify the registries that are

considered in their country to represent a national standardized dataset for acute stroke care and outcomes. Our initial

keyword search yielded 5002 potential papers, of which we included 316 publications representing 28 national stroke

registries from 26 countries. Where reported, data were most commonly collected with a waiver of patient consent

(70%). Most registries used web-based systems for data collection (57%) and 25% used data linkage. Few variables were

measured consistently among the registries reflecting their different local priorities. Funding, resource requirements, and

coverage also varied.

Conclusions: This review provides an overview of the current use of national stroke registries, a description of their

common features relevant to monitoring stroke care in hospitals. Formal registration and description of registries would

facilitate better awareness of efforts in this field.

Keywords

Stroke, outcome assessment, quality assurance, health care, registries

Received: 17 June 2015; accepted: 30 August 2015

Introduction

Stroke is a leading contributor to the global burden ofdisease. Variability in the quality of stroke care maylead to avoidable deaths or disability. In countrieswhere a hospital-based national clinical quality registryof stroke care (i.e. national stroke registry) has beenimplemented in order to monitor acute stroke care,the reporting of the data collected has led to improve-ments in the quality of care, patient outcomes, andhealth policy.1,2

In general, the ideal national stroke registry shouldhave an appropriate governance structure and policesto support its operations including those for data accessand security, publication of results, and have an effect-ive method of communicating results, so participatinghospitals can develop plans to improve care.3 To be

1Stroke and Ageing Research, School of Clinical Sciences at Monash

Health, Monash University, Clayton, Victoria, Australia2Florey Institute of Neuroscience and Mental Health, Heidelberg,

Victoria, Australia3College of Science, Health and Engineering, La Trobe University,

Bundoora, Victoria, Australia4Alfred Health, Prahran, Australia5Department of Medicine, University of Toronto, Toronto, Canada6Department of Neurology, Stroke Service Division, Massachusetts

General Hospital and Harvard Medical School, Boston, MA, USA7Centre for Clinical Brain Sciences, University of Edinburgh, Scotland8Department of Clinical Sciences, Neurology, Lund University, Sweden9Department of Medicine, Royal Melbourne Hospital, University of

Melbourne, Parkville, VIC, Australia10Department of Neurology, Helsinki University Central Hospital,

Helsinki, Finland

Corresponding author:

Dominique A Cadilhac, Stroke and Ageing Research, School of Clinical

Sciences at Monash Health, Monash University, Clayton, Victoria,

Australia.

Email: [email protected]

International Journal of Stroke, 11(1)

International Journal of Stroke

2016, Vol. 11(1) 28–40

! 2016 World Stroke Organization

Reprints and permissions:

sagepub.co.uk/journalsPermissions.nav

DOI: 10.1177/1747493015607523

wso.sagepub.com

practical, affordable, and effective, it is important thatthe data are kept minimal; that variables collected areepidemiologically sound and reproducible; that all eli-gible hospitals participate; selection bias is minimized;and that if outcomes are analyzed, these should be risk(case-mix) adjusted.3

The evolution of national stroke registries

Data on stroke care in hospitals have been collectedsince at least the 1950s,4 and the term ‘‘registry’’ wasfirst used in the field of stroke in the 1970s in a studyconducted in order to investigate the relative frequen-cies of stroke subtypes.5 While local registries providedvaluable information to the participating sites, therewas a need to make comparisons between hospitalsand reliably report trends in demographics, access tointerventions, and outcomes. Further, with the growthof clinical guidelines and concern about deficiencies inthe management of acute stroke, monitoring the qualityof care became necessary. The terms ‘‘register’’ and‘‘registry’’ are commonly used interchangeably in theliterature, but the distinction is that a ‘‘register’’ is adatabase used for systematic collection of health-related information, while the system or organizationgoverning the register is known as the ‘‘registry.’’3 Thefirst national stroke registry was established in Swedenin 1994, and since then many other similar registrieshave been developed.6 In this review, we describe thenational stroke registries that have been established indifferent countries and outline their common featuresand differences.

Methods

We performed a systematic search of peer-reviewed lit-erature for national stroke registries supplemented withgrey literature (e.g. non peer-reviewed literature such asreports available on websites) available from generalinternet searches. We summarize our main findingsand provide case studies of national stroke registriesfrom the authors’ respective countries.

Peer-reviewed literature search strategy

We carried out a comprehensive search of the literaturepublished up to 21 May 2015 using PubMed and OvidMedline. Our search was restricted to manuscripts pub-lished in English on patients admitted to hospital foracute stroke. We used Medical Subject Headings(MeSH) and all field search terms including ‘‘stroke,’’‘‘transient ischemic attack,’’ ‘‘intracerebral haemor-rhage,’’ ‘‘national stroke registry,’’ and ‘‘strokeaudit.’’ For full details of the search strategy used, seethe Online Supplement.

Inclusion criteria. We defined a stroke registry as a datacollection program (using a register, databank or data-base) for monitoring standardized indicators of carequality at multiple sites, in patients hospitalized withacute stroke. Only registries with at least a full year ofprospective data collectionwere included.We considereda stroke registry to be ‘‘national’’ if it was reported as theaccepted country-wide system for data collection; carriedthe name of a country; or was titled as ‘‘national’’. Ourguiding principle was to define country as a UnitedNations (UN) member state or constituent country of aUNmember state. In the case of the registry in Taiwan, itwas decided to consider its inclusion despite ongoingdebate over Taiwan’s political status.

Exclusion criteria. Registries were excluded if they onlyregistered a narrow subgroup of patients (e.g. patientsreceiving reperfusion therapy or limited to paediatriccases); were developed for epidemiological diseasemonitoring without collection of clinical care indica-tors; or included multiple diseases.

Article identification and selection

One author (JK) first reviewed the titles and abstract ofpotential articles. If eligibility based on the title orabstract was unclear, then a second author (DC)reviewed these. Any that remained uncertain underwenta full article review. We made contact with correspond-ing authors when two registries appeared to be thesame, but the names of the registries differed betweenpublications if they were a part of the same program(e.g. for Denmark and Poland). The authors then scru-tinized the results and provided additional informationto ensure no eligible registries were missed.

We acquired additional information through Googlesearches with the name of the registry as the search term.Also, we searched all Organisations for Economic Co-operation andDevelopment (OECD)Member countriesin Google with the term ‘‘stroke registry’’. Not all regis-tries had explicit protocol papers or websites available inEnglish. Therefore, for some of the registries, the infor-mation on their datasets may not represent the completeset of variables collected.

Results

The search strategy yielded 5002 articles of which 2103were duplicates (Figure 1). Following a detailed review,there were 307 publications that met the inclusion cri-teria. Another nine eligible articles known to theauthors were added and this identified one additionalregistry.

These 316 publications described 28 national strokeregistries in 26 countries (Table 1, Figure 2). The


Cadilhac et al. 29

majority of national stroke registries commenced in theearly to mid-2000s. About half of these national strokeregistries included collection of data for TIA and themajority used web-based data entry (Table 1). Themethod for obtaining patient consent to have dataincluded was often not reported and appeared todiffer based on whether personal information was col-lected or if the registry was mandated by a governmentagency; in many cases, there was a waiver of consent.Coverage, representing the total number of eligiblestroke admissions included for a country in a year,varied, but was generally more complete where partici-pation was mandated by government rather than whenthere was voluntary participation by hospitals. Somenational stroke registries used only linked administra-tive data whilst the majority involved manual dataentry for some or all variables.

There were few publications on national stroke regis-tries in Argentina,7 Bahrain,10 Hungary,16 Ireland,17

Malaysia,19 Mexico,20,21 Russia,23 Singapore,25 andThailand.30 Sufficient information was provided to

adjudicate if these registries satisfied the inclusion cri-teria. However, details of governance, privacy mainten-ance and methods of care monitoring were unavailablein English. For registries in Hungary and Russia, indi-cators of care quality collected could not be conclu-sively determined. In Germany, the national registrywas unique in that it was a joint collaboration of regio-nal registries coordinated by the German StrokeRegisters Study Group.15 The Danish Stroke Registrywas established in 200313 as an extension of the DanishNational Indicator Project that was run in 2000–2002(Wildenschild, personal communication). There weretwo national stroke registries in South Korea26,27 andMexico.20,21 These registries and the China NationalStroke Registry12 required written informed consent.The Austrian registry9 was unique in that it was oper-ational in hospitals with stroke units only and may notbe representative of all patients with stroke admitted toAustrian hospitals. We also identified a registry inMalaysia that satisfied our selection criteria despiteappearing to involve only two hospitals.19

Figure 1. Summary of literature search strategy and included articles.


30 International Journal of Stroke 11(1)

Tab

le1.

Sum

mar

yof

nat

ional

stro

kere

gist

ries

use

dar

ound

the

worl

d.

Regi

stry

Act

ive

dat

es

Pat

ient

types

Hosp

ital

par

tici

pat

ion

aFo

llow

-up

peri

od

Dat

aco

llect

ion

meth

od

Conse

nt

Arg

entinia

nN

atio

nal

Stro

keR

egi

stry

(ReN

AC

er)

72004–

Stro

ke,T

IAVolu

nta

rypar

tici

pat

ion,w

ith

74

hosp

ital

sas

of

2008

Not

speci

fied

Web-b

ased

Info

rmed

conse

nt

not

requir

ed

Aust

ralia

nSt

roke

Clin

ical

Regi

stry

(AuSC

R)8

2009–

Stro

ke,T

IAVolu

nta

rypar

tici

pat

ion,

with

more

than

50

hosp

ital

sas

of

2015

90–180

day

sW

eb-b

ased

Lin

ked

adm

inis

trat

ive

dat

aW

aive

rof

conse

nt

for

in-

hosp

ital

deat

hs

Opt

out

conse

nt

Aust

rian

Stro

keU

nit

Regi

stry

92003–

Stro

keT

IAM

andat

ory

par

tici

pat

ion

of

hosp

ital

sw

ith

stro

keunits,

with

35

hosp

ital

sas

of

2014

(cap

ture

stw

oth

irds

of

all

acute

stro

ke)

Thre

em

onth

sW

eb-b

ased

Info

rmed

conse

nt

not

requir

ed

Bah

rain

b10

2011

Stro

keTw

oof

two

gove

rnm

enta

lhosp

ital

sN

ot

speci

fied

Usi

ng

UK

audit

form

Not

speci

fied

Regi

stry

of

the

Can

adia

nSt

roke

Netw

ork

11

(now

the

Onta

rio

Stro

keR

egi

stry

)

2001–

Stro

ke,T

IAIn

itia

llyre

gional

stro

kece

ntr

es

inm

ultip

lere

gions.

Now

man

dat

ory

for

all�

150

hosp

ital

sin

the

pro

vince

of

Onta

rio

2001–2002:si

x-m

onth

tele

-phone

inte

rvie

ws

2003

onw

ard:in

defin

ite

follo

w-u

pfo

rdeat

hs

and

read

mis

sions

thro

ugh

linkag

es

toad

min

istr

a-tive

dat

a

Web-b

ased

and

lapto

pco

mpute

rsLin

ked

adm

inis

trat

ive

dat

a

Info

rmed

conse

nt

initia

lly,

but

no

longe

rre

quir

ed.

Chin

aN

atio

nal

Stro

keR

egi

stry

12

2007–2008

Stro

ke,T

IAH

osp

ital

snom

inat

ing

for

par

-tici

pat

ion,w

ith

132

sele

cted

by

ast

eeri

ng

com

mitte

e

Thre

e,si

x,12,18

and

24

month

sFo

rms

‘‘com

pute

rdat

aentr

y’’

Info

rmed

conse

nt

Dan

ish

Stro

keR

egi

stry

13

2003–

Stro

keM

andat

ory

par

tici

pat

ion

of

all

hosp

ital

str

eat

ing

pat

ients

with

stro

ke,w

ith

28

hos-

pital

sas

of

2014

(cap

ture

s�

90%

of

allst

roke

s)

Not

speci

fied

Form

Web-b

ased

Lin

ked

adm

inis

trat

ive

dat

aIm

plie

dth

atit

isnot

requir

ed

asit

ism

anda-

tory

toin

clude

all

pat

ients

with

stro

ke

Finnis

hSt

roke

Dat

abas

e(P

ER

FEC

TSt

roke

)14

1999–

Stro

ke,T

IAA

uto

mat

icca

ptu

reof

acute

stro

keca

ses.

Curr

ently

has

dat

afr

om

more

than

300

hosp

ital

s(c

aptu

res>

85%

of

allst

roke

s)

Continuous

(10þ

year

s)Lin

ked

adm

inis

trat

ive

dat

aIn

form

ed

conse

nt

not

requir

ed

(continued)


Cadilhac et al. 31

Tab

le1.

Continued

Regi

stry

Act

ive

dat

es

Pat

ient

types

Hosp

ital

par

tici

pat

ion

aFo

llow

-up

peri

od

Dat

aco

llect

ion

meth

od

Conse

nt

Germ

anSt

roke

Regi

ster

Study

Gro

up

(AD

SR)1

51999–

Stro

ke,T

IAH

osp

ital

spar

tici

pat

ing

innin

ere

gional

stro

kere

gist

ries,

with

627

hosp

ital

sas

of

2012

Nil

Not

speci

fied

Implie

dth

atit

isnot

requir

ed

asdat

aar

ese

nt

toth

epoolin

gce

nte

ran

onym

ousl

yIn

form

ed

conse

nt

requir

ed

for

follo

w-u

p

Hunga

rian

Stro

keD

atab

ase

Pro

ject

16

1997–1998

Stro

keEle

ven

regi

onal

hosp

ital

sw

ith

stro

kew

ards

Not

speci

fied

Not

speci

fied

Not

speci

fied

Nat

ional

Stro

keR

egi

ster

(Ire

land)1

72011–

Stro

ke,T

IAA

imin

gfo

rnat

ional

roll-

out,

with

25

hosp

ital

spar

tici

pat

-in

gas

of

2012

Not

speci

fied

Uncl

ear

Not

speci

fied

Japan

ese

Stan

dar

dSt

roke

Regi

stry

18

2001–

Stro

ke,T

IA146

par

tici

pat

ing

hosp

ital

ssi

nce

2007

(cap

ture

s�

10%

of

stro

ke)

Not

speci

fied

Web-b

ased

(enco

ded)

Offlin

e(p

ers

onal

info

rmat

ion

rem

ove

d)

Info

rmed

conse

nt

not

requir

ed

Nat

ional

Stro

keR

egi

stry

(Mal

aysi

a)19

2009–

Stro

keTw

ohosp

ital

sas

of

2010

Thre

ean

d12

month

sW

eb-b

ased

Not

speci

fied

PR

EM

IER

study

(Mexic

o)2

02005–2006

IS,T

IA59

urb

anhosp

ital

sO

ne

year

Form

Info

rmed

conse

nt

REN

AM

EVA

SCst

udy

(Mexic

o)2

12002–2005

Stro

ke25

hosp

ital

s30

day

sFo

rmIn

form

ed

conse

nt

Polis

hH

osp

ital

Stro

keR

egi

stry

22

2000–2002,

2004–2005,

2007–2008

Stro

keVolu

nta

rypar

tici

pat

ion

of

123

hosp

ital

sas

of

2008

Not

speci

fied

Pre

viousl

yfo

rm-b

ased

Now

web-b

ased

Lin

ked

dat

a

Not

speci

fied

Russ

ian

Nat

ional

Stro

keR

egi

stry

23

2008–

Stro

keA

llpri

mar

yst

roke

units

and

com

pre

hensi

vest

roke

cente

rs

30

day

sW

eb-b

ased

Not

speci

fied

The

Scott

ish

Stro

keC

are

Audit

24

2002–

Stro

keM

andat

ory

par

tici

pat

ion

of

all

hosp

ital

str

eating

pat

ients

with

stro

ke(s

ince

2005)

Not

speci

fied

Web-b

ased

Lin

ked

adm

inis

trat

ive

dat

aIn

form

ed

conse

nt

not

requir

ed

Singa

pore

Stro

keR

egi

stry

25

2001–

Stro

keA

llhosp

ital

sN

ot

speci

fied

Web-b

ased

Lin

ked

adm

inis

trat

ive

dat

aIn

form

ed

conse

nt

not

requir

ed

Kore

anSt

roke

Regi

stry

(South

Kore

a)26

1999–

ISA

tle

ast

one

par

tici

pat

ing

site

ineac

hre

gion,w

ith

more

than

Not

speci

fied

Uncl

ear

Not

speci

fied

(continued)



Tab

le1.

Continued

Regi

stry

Act

ive

dat

es

Pat

ient

types

Hosp

ital

par

tici

pat

ion

aFo

llow

-up

peri

od

Dat

aco

llect

ion

meth

od

Conse

nt

30

hosp

ital

spar

tici

pat

ing

asof

2014

Clin

ical

Rese

arch

Cente

rfo

rSt

roke

–fif

thD

ivis

ion

(CR

C-5

)re

gis-

try

(South

Kore

a)27

2008–

Stro

ke,T

IA15

hosp

ital

sco

veri

ng

most

areas

of

South

Kore

aas

of

2014

Thre

ean

d12

month

sW

eb-b

ased

Lin

ked

adm

inis

trat

ive

dat

aIn

form

ed

conse

nt

Rik

sstr

oke

(Sw

eden)2

81994–

Stro

ke,T

IAIn

itia

llyvo

lunta

ry,but

full

cove

rage

of

hosp

ital

ssi

nce

1997

(cap

ture

s>

90%

of

all

stro

kes)

Thre

ean

d12

month

sFo

rmIn

form

ed

conse

nt

not

requir

ed

Tai

wan

Stro

keR

egi

stry

29

2006–

Stro

ke,T

IAVolu

nta

rypar

tici

pat

ion

of

39

hosp

ital

sas

of

2010

(cap

-tu

res�

18%

of

stro

ke)

One,

thre

ean

dsi

xm

onth

sW

eb-b

ased

Info

rmed

conse

nt

not

requir

ed

Thai

Stro

keR

egi

stry

30

2008–

ISVolu

nta

rypar

tici

pat

ion

of

76

hosp

ital

sas

of

2010

Not

speci

fied

Form

Not

speci

fied

SentinelSt

roke

Nat

ional

Audit

Pro

gram

me

(UK

–Engl

and,N

ort

hern

Irela

nd,W

ales)

31

2013–

Stro

keA

llelig

ible

trust

sin

Engl

and,

Nort

hern

Irela

nd

and

Wal

es

Six

month

sW

eb-b

ased

Not

speci

fied

Eth

os

(USA

)32

2001–2005

IS,T

IAVolu

nta

rypar

tici

pat

ion

of

86

hosp

ital

sN

ot

speci

fied

Web-b

ased

Not

speci

fied

Get

With

the

Guid

elin

es

–St

roke

(USA

)33

2001–

Stro

ke,T

IAVolu

nta

rypar

tici

pat

ion

of1991

hosp

ital

sin

2014

30

day

s,optional

Web-b

ased

Info

rmed

conse

nt

not

requir

ed

by

the

AH

A,

but

loca

lsi

tehum

anre

sear

chco

mm

itte

es

may

request

it

Pau

lC

ove

rdell

Nat

ional

Acu

teSt

roke

Regi

stry

(USA

)34

2001–

Stro

ke,T

IAVolu

nta

rypar

tici

pat

ion

ofm

ore

than

300

hosp

ital

sin

six

stat

es

asof

2012

Not

speci

fied

Not

speci

fied

Not

speci

fied

IS:Is

chae

mic

stro

ke;T

IA:Tra

nsi

ent

isch

aem

icat

tack

;U

K:U

nited

Kin

gdom

;U

SA:U

nited

Stat

es

of

Am

eri

ca.

aPar

tici

pat

ion

ofth

ehosp

ital

sre

flect

sth

eco

vera

gein

aco

untr

yan

dth

elik

elih

ood

offu

llor

par

tial

case

-asc

ert

ainm

ent

ofpat

ients

and

may

be

influ

ence

dby

wheth

er

the

regi

stry

isvo

lunta

ryor

man

dat

ed,t

he

num

ber

of

stro

kead

mis

sions

ateac

hhosp

ital

,an

dm

ayal

sobe

subje

ctto

chan

gedependin

gon

fundin

gfr

om

year

toye

ar.

bN

ost

udy

nam

e.


Cadilhac et al. 33

From the review of these national stroke registries,their main features in terms of governance, data custo-dial arrangements, maintenance of privacy, fundingsources and how feedback is generally provided to hos-pitals are summarized in Table 2. These features havenot been consistently reported in registry publications.

A further four registries in the USA, India, SaudiArabia, and the Arabian Gulf were identified, butwere not included according to our criteria based onthe limited information provided in the full publicationor grey literature (see Figure 2 and Online supplement).

Case studies from different countries

The Riksstroke from Sweden is the longest runningnational stroke registry. It started in 1994 and hashad full coverage of all hospitals treating patientswith stroke since 1997 (25). In Sweden, quality moni-toring of health services is mandated by the NationalBoard of Health and Welfare. This registry is estimatedto include >90% of all patients with stroke in Sweden(Bo Norrving, personal communication 8/6/2015). Inofficial reports of the data, individual hospitals arenamed and this open public reporting has demon-strably improved the quality of care in Sweden.

The PERFormance, Effectiveness, and Cost ofTreatment episodes in Stroke (PERFECT Stroke)

study began in Finland in 1999.14 This registry avoidsburdensome manual data collection since all people inFinland have a unique identifier allowing multiplenational and administrative health database linkagesto create the registry. For the PERFECT Stroke regis-try, patients are automatically identified using ICD-10discharge codes, coded by the treating physicians withproven high reliability. The registry is estimated toinclude >85% of all population-based strokes, with16,000 patients registered annually from >300 hospitalsand a total of >200,000 total cases included. The dataare published online with hospitals named and bench-marked, including risk-adjustment and confidenceintervals. Similar national data linkage approacheshave since been applied in other European countriesas part of the EuroHOPE register study.35

The Registry of the Canadian Stroke Network(RCSN), now known as the Ontario Stroke Registry,was launched in 2001 with data collection on consecu-tive patients seen at regional stroke centres acrossCanada, and with a six-month follow-up of functionalstatus and quality of life.11 Initially, written consentwas required and the biases introduced by thismethod were documented by Tu et al.36 In 2003, awaiver of consent was approved in the province ofOntario (population �12 million). Between 2003 and2013, the registry coordinated continuous data

Figure 2. World map with the location of national stroke registries marked with pins. Inset: Location of European national

stroke registries. Red pin: one confirmed registry, green pin: two confirmed registries, blue pin: three confirmed registries, yellow

pin: one possible stroke registry. Source: https://batchgeo.com/ (powered by Google).



https://batchgeo.com/

Tab

le2.

Sum

mar

yof

bro

adopera

tional

feat

ure

sof

nat

ional

stro

kecl

inic

alqual

ity

regi

stri

es.

Com

mon

feat

ure

/sD

eta

ilQ

ual

ity

of

report

ing

inpublic

atio

ns

Gove

rnan

ceO

vera

rchin

gSt

eeri

ng

or

Man

agem

ent

Com

mitte

e

May

hav

ea

separ

ate

group

resp

onsi

ble

for

the

day

-to-d

aym

anag

em

ent

Not

alw

ays

clear

or

report

ed

Dat

acu

stodia

nC

entr

alentity

separ

ate

toth

epar

tici

-

pat

ing

hosp

ital

sth

atre

ceiv

ean

d

colla

teth

edat

a.

Resp

onsi

ble

for

undert

akin

gdat

alo

gic

check

san

dqual

ity

of

care

monitor-

ing/

report

ing

Oft

en

implie

dfr

om

stat

em

ents

inth

e

meth

odsN

ot

clear

ifth

edat

acu

sto-

dia

nhas

ow

ners

hip

of

the

dat

aset

Pri

vacy

De-identific

atio

nof

pat

ient

dat

aan

d

aggr

ega

tere

port

ing.

Identific

atio

nof

hosp

ital

s

Genera

lly,pat

ient

dat

aar

eco

llect

ed

or

report

ed

ina

de-identifie

dfo

rmat

Not

alw

ays

clear

or

report

ed

Fundin

gG

ove

rnm

ent

Non-g

ove

rnm

ent

Indust

ry

Ongo

ing

fundin

g

Tem

pora

ry/g

rant

fundin

g

Multip

leso

urc

es

com

mon

Report

ed

inm

eth

ods

or

acknow

ledgm

ents

Qual

ity

of

care

feedbac

kpro

cedure

Annual

report

s

Liv

e‘‘o

n-d

em

and’’

report

s

Indiv

idual

hosp

ital

report

s

Par

tof

anat

ional

pro

gram

of

active

qual

ity

impro

vem

ent

Com

mon

tohav

em

ultip

lem

eth

ods

of

com

munic

atin

gperf

orm

ance

topar

-

tici

pat

ing

hosp

ital

s

Not

alw

ays

clear

or

report

ed


Cadilhac et al. 35

collection at regional stroke centres in Ontario, as wellas a biennial audit on a random sample of all consecu-tive patients seen at every acute care hospital in theprovince, with this sampling strategy designed to min-imize the biases that can be associated with selectiveenrolment of hospitals or patients.24 Follow-up inter-views were no longer performed, but the registry waslinked to population-based administrative databases toprovide information on mortality, readmissions, phys-ician services and medication prescriptions. The regis-try data have been used for national, provincial andregional public stroke performance reports.11

In the UK (excluding Scotland), three separate pro-grams have be used to monitor the quality of care pro-vided to patients hospitalized with stroke. The NationalSentinel Stroke Audit (NSSA) conducted seven bian-nual audit snapshots of stroke care (�50 consecutivecases contributed per hospital) from 1998 to 2010 inthe majority of acute hospitals. This was supersededby a continuous prospective Stroke ImprovementNational Audit Programme (SINAP) between 2010and 2012, followed by the Sentinel Stroke NationalAcute Programme (SSNAP) now collecting informa-tion for patients with stroke in England, Wales andNorthern Ireland since December 2012.31

In Scotland, the Scottish Stroke Care Audit (SSCA)was established in 2002, and collected data from allhospitals since 2005. It registers over 8000 inpatientsand about 3000 outpatients per year and includes anestimated >90% of patients with stroke.24 Performanceagainst nationally agreed standards is fed back to hos-pitals, health boards monthly to drive improvements,and data are made publicly available each year. Thisregistry is mandatory and embedded in a nationalstroke improvement programme.

The Get With the Guidelines (GWTG)-Stroke regis-try, run by the American Heart Association with leader-ship from stroke expert volunteers, is the larger of the twocurrently active national stroke registries in theUSA.37 Itwas launched in 2001 in a selected number of sites, andthenbecameavailable nationally in 2003.GWTG-Strokeis a voluntary registry with participation from a total of2523 hospitals over the past 15 years contributing over3.5 million patient encounters. In calendar year 2014,over 460,000 new cases were added. The registry hasbeen the data source for 70 peer-reviewed publications,some of which included linkage to administrative claimsdata for long-term follow-up information.

Since GWTG is web-based, clinicians can downloadover 60 individual or summary measure ‘‘live’’ reportsof their data benchmarked against local, regional ornational benchmarks or directly download their datafor analysis. Periodic data quality reports are also pro-duced and hospitals receive a personalized version ofthe report that highlights areas of data quality, out of

range values and missing data. A nearly identical data-set is used in the other national stroke registry in theUSA, the Paul Coverdell National Acute StrokeRegistry, which is coordinated through the Centersfor Disease Control and Prevention and administeredvia grants awarded competitively to State Departmentsof Public Health.38 In some states, participation in thestroke registry is mandatory for stroke center accredit-ation. The GWTG-Stroke registry has inspired theestablishment of a similar registry in Taiwan.29

Similar to the system in the UK, stroke care moni-toring in Australia began with biennial clinical audits(established in 2007 and funded by the National StrokeFoundation) of up to 40 consecutive cases per hos-pital.39 The Australian Stroke Clinical Registry(AuSCR), established in 2009 receives funding from arange of government, non-government and industryorganizations for limited periods.8 Lack of consistentfunding has hampered expansion into more hospitals.As of 2015, more than 50 public and private hospitalsacross Australia were contributing registry data, with8226 patients entered in 2014 of approximately 35,000annual Australian stroke hospitalizations. SinceAuSCR is web-based, clinicians can download up to11 ‘‘live’’ reports of their data benchmarked againstthe national average or directly download their datafor analysis. An annual report is also produced andhospitals receive a personalized version of the reportthat compares their performance to other hospitals.Unlike in the UK, the Australian biennial audits andthe national registry continue to co-exist.

Variables collected for reporting on thequality of care

Table 3 provides a summary of the variables collectedin the national stroke registries that were listed as indi-cators of the quality of care. The majority of registriescollected information on use of intravenous thromboly-sis, in-hospital antithrombotic therapy, discharge onantithrombotic medication and stroke unit care. Themost commonly collected outcome measures includedsurvival status, functional status, recurrent stroke sincehospital discharge, and place of residence. For furtherdetails of process and outcome metrics collected inOECD countries, see the supplementary material avail-able online with this article (Tables S1 and S2).

Discussion

This article provides an overview of national strokeregistries designed to systematically monitor the qualityof stroke care. In some countries, the registry evolvedfrom a cross-sectional audit program (e.g. England).However, it was beyond the scope of this article to



detail fully the latter method of quality monitoring.The two major approaches to monitoring the qualityof care with a registry were: 1) traditional registries,which provide the most comprehensive data, but areresource-intensive to run and have optimal coveragewhen mandatory; and 2) administrative database link-age, which is cheaper with greater coverage, but oftenlacks detail and is feasible in only some jurisdictions.Registries that have mandatory participation of all hos-pitals with inclusion of all eligible patients and follow-up processes provide the most accurate country-wideassessment of quality of care.

Having adequate budget for data collection andactive feedback to hospitals is also important toencourage greater behaviour change. Unfortunately,we were unable to confidently summarize the methodsused by the identified registries to feedback on the qual-ity of care to hospitals or patients, since this informa-tion was often lacking. We found that some nationalstroke registries were complementary to active qualityimprovement programs in their respective countries.For example, in Scotland, inspections of quality ofcare are made and staff members in hospitals partici-pate in professional forums to share best practice meth-ods. Investments to enhance care in hospitals are madeas part of this broader program. This is also similar to

the approach in the GWTG-Stroke registry where par-ticipating hospitals are provided with clinical tools andpatient education resources, and staff can attend edu-cation workshops. An alternative to this may be to haveconcurrent audits that provide snapshots of care in alarge sample of hospitals, or collection of data in arepresentative region as has been the evolution inCanada. In the case of Australia, where the AuSCRdoes not have full national coverage, an audit of 40consecutive cases at all hospitals admitting patientswith stroke is undertaken biennially by the NationalStroke Foundation, and complements the data col-lected in the AuSCR.

The breadth of variables used was not explicitlyreported for all registries identified, and sometimes hadto be deduced from several publications. There were alsodifficulties compiling information on the active dates ofthe registries, patient eligibility; data acquisition meth-ods; hospital participation; governance arrangements;and funding sources. We encourage authors to reportthese features in their registry publications. Further,data custodians for registries should endeavour toaudit and report on the quality of their registry dataincluding coverage as is recommended by Bufalinoet al.40 and provide more detail on how the data areused to close the ‘‘quality-of-care’’ gaps identified.

Table 3. Most commonly reported areas of focus collected to measure quality of care.

Reported to be collected in the majority of registries (>14 registries)a

Intravenous thrombolysis (recombinant tissue plasminogen activator)

Antithrombotic therapy during hospitalization

Discharge on antithrombotic medication

Managed on a stroke unit

Collected in multiple registries (2–14 registries)a

Adequate fluid and nutrition

Assessment of nutritional risk

Assessment for rehabilitation

Brain or vascular imaging

Carotid endarterectomy/stenting

Continence plan

Craniectomy

Discharge on antihypertensive medication

Discharge on statin/lipid-lowering medication

Discussion with relative/carer

Dysphagia screening

Educational (materials) provided

Length of stay

Rehabilitation services in hospital, including early mobilization

Smoking cessation counselling

Stroke unit team management

Time between onset, door, scan and needle (e.g. tPA bolus or groin puncture)

Transport to hospital

Note: Some metrics were re-worded or combined to simplify presentation.aProvided as a qualitative description of the most common aspects of stroke care monitored in national stroke registries.


Cadilhac et al. 37

We found that a large number of variables were col-lected but that definitions, numerators and denomin-ators for similar variables were inconsistent acrossregistries, making international comparisons difficult.Only a small number of clinical process variableswere identified that were commonly collected in themajority of registries. In some of the registries,follow-up of patients was only possible with patients’written informed consent. Therefore, selection bias maybe an issue if inability to obtain consent is common,especially among the most disabled or vulnerable popu-lations.3 For several registries, linkage with nationaladministrative datasets allowed ascertainment of mor-tality and tracking of secondary prevention therapiesprovided over time in all participants.

We recognize that there may be some national strokeregistries that were missed in our search. Defining thecriteria for inclusion of national stroke registries in thisreview was not straightforward. Many population-based studies of stroke incidence and mortality werereferred to as registry projects, but were not used formeasuring the quality of care in hospitals. We foundthat the majority of the included registries did not havefull national coverage. This may reflect their level ofmaturity, e.g. phase of scaling-up; limitations withfunding; voluntary versus mandatory participationstatus; or decisions made to have selective samplingfrom a certain number of hospitals or region to providea national picture of the quality of stroke care in hos-pitals. It is also acknowledged that other types of regis-tries exist (e.g. regional or multi-disease) and these mayhave larger number of patients than some of those high-lighted here. Further, some countries have a number ofregistries or datasets that clinicians contribute to thatmay be focussed on different aspects of clinical care.For example, in Japan, there is a national stroke regis-try, a regional stroke registry and a thrombolysisregistry.41

National stroke registries may also serve otherimportant purposes. Information about stroke inci-dence and mortality is limited in many countries,42

and registries may offer a suitable proxy for theseimportant epidemiological measures in situationswhere the majority of new cases are treated in a hos-pital. We encourage countries without a standardizedcollection of data for stroke care to consider our find-ings. Understanding the common features of the regis-tries may assist other countries to establish asustainable registry program for stroke care and toselect the most appropriate variables to collect. Therehave been several programs established now to encour-age consistent data collection across countries that offerimportant progress in this field, but currently thesefocus on outcomes rather than processes of hospitalcare (see Online supplement).

Conclusions

The findings from this review provide an overview ofthe current use of national stroke registries, a descrip-tion of their common features and other considerationsin relation to monitoring the quality of acute strokecare in hospitals. Future directions in this field includehaving better reporting on governance, data collection,including variables and their definitions, sources offunding and quality of care feedback mechanisms andtheir effectiveness. This information would be useful forother countries aiming to establish national strokeregistries.

Acknowledgments

We acknowledge the contribution of a peer reviewer whoprovided additional information about the Japanese

Standard Stroke Registry.

Declaration of Conflicting Interests

The author(s) declared the following potential conflicts ofinterest with respect to the research, authorship, and/or pub-lication of this article: Dominique Cadilhac is the Data

Custodian for the AuSCR; and Joosup Kim is a researchfellow working on the AuSCR. Natasha Lannin is on theManagement Committee for the AuSCR. Moira Kapral pro-vides oversight for the Ontario Stroke Registry. Lee

Schwamm is the chair of the Stroke Clinical Workgroup ofGWTG-Stroke, a stroke systems consultant on the PaulCoverdell National Acute Stroke Registry in Massachusetts,

and a technical expert panelist for the Joint CommissionPrimary and Comprehensive Stroke Center programs.Martin Dennis provides oversight of the Scottish Stroke

Care Audit; and Bo Norrving oversight of Riksstroke. AtteMeretoja has provided oversight and direction for the Finnish(PERFECT) Stroke Registry.

Funding

The author(s) disclosed receipt of the following financial sup-

port for the research, authorship, and/or publication of thisarticle: Dominique Cadilhac was supported by a fellowshipfrom the National Health and Medical Research Council(NHMRC; 1063761 co-funded by National Heart

Foundation) and Joosup Kim receives salary support froma partnership project grant funded by the NHMRC(1034415). Moira Kapral was supported by a Career

Investigator Award from the Heart and Stroke Foundation,Ontario Provincial Office.

References

1. Asplund K, Hulter Asberg K, Appelros P, et al. The Riks-

Stroke story: Building a sustainable national register for

quality assessment of stroke care. Int J Stroke 2011; 6:

99–108.2. Silver FL, Kapral MK, Lindsay MP, Tu JV and Richards

JA. International experience in stroke registries: Lessons



learned in establishing the Registry of the CanadianStroke Network. Am J Prev Med 2006; 31: S235–S237.

3. Australian Commission on Safety and Quality in Health

Care (ACS&QHC). (2008). Operating Principles andTechnical Standards for Australian Clinical QualityRegistries. Sydney: ACS&QHC.

4. Rankin J. Cerebral vascular accidents in patients over the

age of 60. III. Diagnosis and treatment. Scot Med J 1957;2: 254–268.

5. Caplan LR. The last 50 years of cerebrovascular disease:

Part 1. Int J Stroke 2006; 1: 104–108.6. Tanne D, Koton S and Bornstein NM. National stroke

registries: What can we learn from them? Neurology 2013;

81: 1257–1259.7. Sposato LA, Esnaola MM, Zamora R, et al. Quality of

ischemic stroke care in emerging countries: The

Argentinian National Stroke Registry (ReNACer).Stroke 2008; 39: 3036–3041.

8. Cadilhac DA, Lannin NA, Anderson CS, et al. Protocoland pilot data for establishing the Australian Stroke

Clinical Registry. Int J Stroke 2010; 5: 217–226.9. Greisenegger S, Seyfang L, Kiechl S, Lang W, Ferrari J

and Austrian Stroke Unit Registry C. Thrombolysis in

patients with mild stroke: Results from the AustrianStroke Unit Registry.[Erratum appears in Stroke 2014;45: e70]. Stroke 2014; 45: 765–769.

10. Al Banna M, Baldawi H, Kadhim A, Humaidan H andWhitford DL. Stroke in Bahrain: Rising incidence, mul-tiple risk factors, and suboptimal care. Int J Stroke 2015;10: 615–618.

11. Fang J, Kapral MK, Richards J, Robertson A,Stamplecoski M and Silver FL. The Registry ofCanadian Stroke Network: An evolving methodology.

Acta Neurol Taiwan 2011; 20: 77–84.12. Wang Y, Cui L, Ji X, et al. The China National Stroke

Registry for patients with acute cerebrovascular events:

design, rationale, and baseline patient characteristics. IntJ Stroke 2011; 6: 355–361.

13. Wildenschild C, Mehnert F, Thomsen RW, et al.

Registration of acute stroke: validity in the DanishStroke Registry and the Danish National Registry ofPatients. Clin Epidemiol 2014; 6: 27–36.

14. Meretoja A, Roine RO, Kaste M, et al. Stroke monitor-

ing on a national level: PERFECT Stroke, a comprehen-sive, registry-linkage stroke database in Finland. Stroke2010; 41: 2239–2246.

15. Seidel G, Hermanek P, Kolominsky-Rabas P, et al. Thequality of acute stroke care – An analysis of evidence-based indicators in 260 000 patients. Dtsch Arztebl Int

2014; 111: 759–765.16. Ovary C, Szegedi N, May Z, Gubucz I and Nagy Z.

Comparison of stroke ward care versus mobile stroketeams in the Hungarian stroke database project. Eur J

Neurol 2007; 14: 757–761.17. Briggs R, McDonagh R, Mahon O and Harbison J. In-

hospital stroke: characteristics and outcomes. Irish Med J

2015; 108: 24–25.18. Kobayashi S and Japan Stroke Scale Registry Study G.

International experience in stroke registry: Japanese

Stroke Databank. Am J Prev Med 2006; 31: S240–S242.

19. Nazifah SN, Azmi IK, Hamidon BB, Looi I, Zariah AAand Hanip MR. National Stroke Registry (NSR):Terengganu and Seberang Jaya experience. Med J

Malaysia 2012; 67: 302–304.20. Cantu-Brito C, Ruiz-Sandoval JL, Murillo-Bonilla LM,

et al. The first Mexican multicenter register on ischaemicstroke (the PREMIER study): Demographics, risk fac-

tors and outcome. Int J Stroke 2011; 6: 93–94.21. Ruiz-Sandoval JL, Chiquete E, Banuelos-Becerra LJ,

et al. Cerebral venous thrombosis in a Mexican multi-

center registry of acute cerebrovascular disease: theRENAMEVASC study. J Stroke Cerebrovasc Dis 2012;21: 395–400.

22. Niewada M, Skowronska M, Ryglewicz D, et al. Acuteischemic stroke care and outcome in centers participatingin the Polish National Stroke Prevention and Treatment

Registry. Stroke 2006; 37: 1837–1843.23. Hachinski V, Donnan GA, Gorelick PB, et al. Stroke:

working toward a prioritized world agenda. Stroke2010; 41: 1084–1099.

24. Turner M, Barber M, Dodds H, et al. Implementing asimple care bundle is associated with improved outcomesin a national cohort of patients with ischemic stroke.

Stroke 2015; 46: 1065–1070.25. Venketasubramanian N, Chang HM, Chan BP, et al.

Countrywide stroke incidence, subtypes, management

and outcome in a multiethnic Asian population: theSingapore Stroke Registry – methodology. Int J Stroke2015; 10: 767–769.

26. Lee BC, Roh JK and Korean Stroke R. International

experience in stroke registries: Korean Stroke Registry.Am J Prev Med 2006; 31: S243–S245.

27. Kim BJ, Han MK, Park TH, et al. Current status of acute

stroke management in Korea: A report on a multicenter,comprehensive acute stroke registry. Int J Stroke 2014; 9:514–518.

28. Asplund K, Hulter Asberg K, Norrving B, et al. Riks-stroke – a Swedish national quality register for strokecare. Cerebrovasc Dis 2003; 15: 5–7.

29. Hsieh FI, Lien LM, Chen ST, et al. Get With theGuidelines-Stroke performance indicators: Surveillanceof stroke care in the Taiwan Stroke Registry: Get Withthe Guidelines-Stroke in Taiwan. Circulation 2010; 122:

1116–1123.30. Nilanont Y, Nidhinandana S, Suwanwela NC, et al.

Quality of acute ischemic stroke care in Thailand: A pro-

spective multicenter countrywide cohort study. J StrokeCerebrovasc Dis 2014; 23: 213–219.

31. Smith CJ, Bray BD, Hoffman A, et al. Can a novel clin-

ical risk score improve pneumonia prediction in acutestroke care? A UK multicenter cohort study. J AmHeart Assoc 2015; 4: e001307.

32. Hills NK and Johnston SC. Duration of hospital partici-

pation in a nationwide stroke registry is associated withimproved quality of care. BMC Neurol 2006; 6: 20.

33. LaBresh KA, Reeves MJ, Frankel MR, Albright D and

Schwamm LH. Hospital treatment of patients with ische-mic stroke or transient ischemic attack using the ‘‘GetWith The Guidelines’’ program. Arch Intern Med 2008;

168: 411–417.


Cadilhac et al. 39

34. Reeves MJ, Broderick JP, Frankel M, et al. The PaulCoverdell National Acute Stroke Registry: initial resultsfrom four prototypes. Am J Prev Med 2006; 31:

S202–S209.35. Malmivaara A, Meretoja A, Peltola M, et al. Comparing

ischaemic stroke in six European countries. TheEuroHOPE register study. Eur J Neurol 2015; 22:

284–e26.36. Tu JV, Willison DJ, Silver FL, et al. Impracticability of

informed consent in the Registry of the Canadian Stroke

Network. New Engl J Med 2004; 350: 1414–1421.37. Fonarow GC, Reeves MJ, Smith EE, et al.

Characteristics, performance measures, and in-hospital

outcomes of the first one million stroke and transientischemic attack admissions in get with the guidelines-stroke. Circ Cardiovasc Qual Outcome 2010; 3: 291–302.

38. Wattigney WA, Croft JB, Mensah GA, et al. Establishingdata elements for the Paul Coverdell National AcuteStroke Registry: Part 1: proceedings of an expert panel.

Stroke 2003; 34: 151–156.39. Harris D, Cadilhac D, Hankey GJ, Hillier S, Kilkenny M

and Lalor E. National Stroke Audit: The Australianexperience. Clin Audit 2010; 2: 25–31.

40. Bufalino VJ, Masoudi FA, Stranne SK, et al. TheAmerican Heart Association’s recommendations forexpanding the applications of existing and future clinical

registries: A policy statement from the American HeartAssociation. Circulation 2011; 123: 2167–2179.

41. Toyoda K. Epidemiology and registry studies of stroke in

Japan. J Stroke 2013; 15: 21–26.42. Thrift AG, Cadilhac DA, Thayabaranathan T, et al.

Global stroke statistics. Int J Stroke 2014; 9: 6–18.



national stroke registries for monitoring and improving ... · systematic review national stroke...

Documents