OUP user menu

Do centres with well-developed protocols, training and infrastructure have higher rates of thrombolysis for acute ischaemic stroke?

J.D.H. Van Wijngaarden, M. Dirks, L.W. Niessen, R. Huijsman, D.W.J. Dippel
DOI: http://dx.doi.org/10.1093/qjmed/hcr075 785-791 First published online: 24 May 2011


Background: The introduction of intravenous thrombolysis with recombinant tissue Plasminogen Activator (rt-PA) has greatly improved the effectiveness of acute ischaemic stroke care. However, in most hospitals only 2–10% of all admitted stroke patients are treated with thrombolysis.

Aim: The purpose of this study is to identify if available protocols, training and infrastructure influence the thrombolysis rate.

Design: Cohort study of 12 hospitals in the Netherlands.

Methods: In a cohort of patients admitted with acute stroke within 24 h from onset of symptoms, data were obtained. Stroke service characteristics of 12 hospitals were acquired through structured interviews with intra- and extramural representatives, in order to asses (i) protocols, (ii) training and (iii) complexity of infrastructure. Data were analysed with multi-level logistic regression to relate the likelihood of treatment with thrombolysis to availability and completeness of protocols, training and infrastructure both outside (extramural) and inside (intramural) each centre.

Results: Overall 5515 patients were included in the study. Thrombolysis rates varied from 5.7% to 21.7%. An association was observed between thrombolysis rates and extramural training [odds ratio (OR): 1.11; 95% confidence interval (CI): 0.99–1.25] and availability of intramural protocols (OR: 1.46; 95% CI: 1.12–1.91).

After adjustment for hospital size and teaching vs. nonteaching hospital, these associations became stronger; extramural training [adjusted OR (aOR): 1.14; 95% CI: 1.01–1.30] and availability of intramural protocols (aOR: 1.77; 95% CI: 1.30–2.39).

Conclusions: Extramural training and intramural protocols are important tools to increase thrombolysis rates for acute ischaemic stroke in hospitals. Intramural protocols and extramural training should be aimed at all relevant professionals.


Treatment of acute stroke patients with recombinant tissue plasminogen activator (rtPA), within 3 h after onset of the symptoms, may save 1 in 10 patients from death or dependency.1 Even in the 3–4.5 h window a considerable effect has been demonstrated.2 Still, only a limited proportion of stroke patients receive thrombolytic treatment in most hospitals.3,4 Research suggests that both inside and outside the hospital, there are barriers to quick referral and management of these patients.5,6

Considering the available scientific evidence, the European Stroke Initiative (EUSI) Executive Committee has suggested different measures to lift these barriers. The EUSI emphasizes that ‘teaching the public about symptoms and signs of stroke is one of the highest priorities of public medical education’.7 Health professionals need to learn that they are ‘important and competent partners in the team providing acute stroke care’.7 Second, they point out that ‘written protocols are a prerequisite for standardised patient care’.7 Furthermore, surveys suggest that infrastructure and availability of resources may also be of influence.6

Although the recommendations of the EUSI have a high face value, a beneficial effect has not always been firmly established.5 Research did show however that the introduction of training programmes and protocols as part of a larger focused quality improvement effort can improve thrombolysis rates.8 The purpose of this study is to identify which (combination of) structural characteristics influence the thrombolysis rate, in a cohort study of 12 hospitals in the Netherlands. We particularly addressed differences in training, in the availability of protocols and infrastructure.


Study design and sample

This study was designed as a cohort study in 12 centres covering 11% of all hospitals in the Netherlands; it ran from October 2005 to October 2007. Centres were recruited by setting out a call among 21 centres participating in a programme for improving organized stroke care. At the start of the study the participating hospitals had a mean thrombolysis rate of 5% (range: 0–10%), similar to the mean rate in the Netherlands at that time.

Because structural characteristics are often multidimensional and involve complex social interactions, we used a mixed methods approach9 with both qualitative and quantitative methods based on a Delphi approach.10

Patient population

During a period of 2 years all patients >18 years of age were included who were admitted with acute stroke, i.e. patients with an acute focal neurological deficit. Onset of symptoms was not >24 h before admission. Information for all patients on age, sex, time since onset of symptoms and treatment with thrombolysis was registered. For patients with an ischaemic stroke admitted within 4 h from onset, more detailed clinical data were gathered: stroke severity (NIH stroke scale), contra-indications for thrombolysis and cardiovascular risk factors. For patients who were treated with thrombolysis door-to-needle time was also registered.


The primary outcome in all registered patients was treatment with thrombolysis or not. On hospital level, we calculated a thrombolysis rate by dividing the number of stroke patients treated with thrombolysis, by the total number of stroke patients admitted within 24 h from symptom onset during the inclusion period.

Structural characteristics

Structural characteristics that are expected to influence thrombolysis rates were identified by international publications5,7 and the results of a survey we did among 15 Dutch hospitals.6 We distinguished between factors focused on lifting barriers outside the hospital (extramural) and inside the hospital (intramural). We focused on training given in the preceding 2-year period, because from previous research it is known that it is important to use periodic reinforcement messages in public campaigns11 and changes in personnel should be taken into account. Furthermore, The Netherlands is densely populated and there is always a hospital with thrombolysis facilities in reach within 15 min driving. In urban areas there are often more hospitals, not all of which perform thrombolysis treatment or have beds available at a certain moment. The number of hospitals in a region is therefore a relevant infrastructural characteristic. For these and each other factor an indicator was developed (Table 1).

View this table:
Table 1

Structural characteristics in the continuum of care for thrombolysis in stroke

    Agreements and protocolsGeneral practitionersProtocol present (0,1)
Triage nurses in GP-serviceProtocol present (0,1)
Ambulance personnelProtocol present (0,1)
Emergency incident roomProtocol present (0,1)
    TrainingGeneral publicCoverage in percentage (0–1)
General practitionersCoverage in percentage (0–1)
Ambulance personnelCoverage in percentage (0–1)
    InfrastructureHospitalsNo. of hospitals in region (1/n)
    Agreements and protocolsStaff emergency serviceProtocol present (0,1)
Staff priority ECGProtocol present (0,1)
Staff priority CTProtocol present (0,1)
Who interprets CT: (resident-)Protocol present (0,1)
radiologist/neurologist)Protocol present (0,1)
Staff priority lab-resultsProtocol present (0,1)
Allocated bedsProtocol present (0,1)
Stroke nursesCompleteness of protocols (0–1)
NeurologistsCompleteness of protocols (0–1)
    TrainingEmergency nursesCoverage of general training/information packages on stroke care (0–1)
Nursing staff of general and neurology departments‘dummy runs’ (0,1)
Coverage training/information packages to perform thrombolysis (0–1)
Staff radiology department‘dummy runs’ (0,1)
Coverage of information supply concerning thrombolysis-related procedures (0–1)
    InfrastructureMedical staff for thrombolysis treatmentNo. of neurologists and assistants
Hours a week that all resources are available for thrombolysis (n/168)

Acquisition of centre-related data

Data on structural characteristics were gathered during the 2-year inclusion period. The leading neurologists in all 12 centres were asked to keep a diary of their activities to improve the thrombolysis rate and onset-to-needle time. These diaries were used as input for face-to-face interviews with them. Also, representatives of each unit involved in the care-process (from onset of stroke, to admission on the stroke unit) of all participating hospitals were interviewed by telephone. These units were general practitioners (GP's) in the region, ambulance services, casualty, laboratory and departments of radiology and neurology. An open-ended questionnaire was developed for each unit, based on the indicators (Table 1). All interviews were audio taped and transcribed by an independent transcriptionist.

Scoring procedure

The data from the interviews and the protocols were used to attribute scores to each of the structural characteristics (Table 1). Each indicator was scored 0, 0.5 or 1, depending on presence. Protocols for both neurologists and nurses on thrombolysis were also judged on completeness based on a scoring list developed by two neurologists (Table 2). Finally, sum scores were made for all structural characteristics. The sum score for (intramural) protocols is a combination of the presence and completeness scores.

View this table:
Table 2

Checklist for completeness of protocols for stroke nurses and neurologists

Date, author, subjectScore for availability of information in protocol
Patient level
    IndicationsYes (1), Partly (0.5), No (0)
    Contra-indicationsYes (1), Partly (0.5), No (0)
    Patient informationYes (1), Partly (0.5), No (0)
    Preparation of infusionYes (1), Partly (0.5), No (0)
    Administration routeYes (1), Partly (0.5), No (0)
    How to administerYes (1), Partly (0.5), No (0)
    Frequency vital signs monitoringYes (1), Partly (0.5), No (0)
    Limits for checksYes (1), Partly (0.5), No (0)
    What to do in case of complicationsYes (1), Partly (0.5), No (0)
    Whom to call in case of complicationsYes (1), Partly (.5), No (0)
Total scoreTotal score/11
View this table:
Table 3

Baseline characteristics of the 12 hospitals and the admitted stroke patients in the Netherlands over a 2-year period by quantiles of the sum score of structural characteristics

Low sum scoreHigh sum score
Hospital characteristicsa
    Sumscore of structural characteristics (median, range)16 (12.6–18.7)21 (20.5–27.4)
    Number of hospitals66
    Teaching hospital14
    University hospital11
    Mean thrombolysis rate (%)12.114.4
Patient characteristics
    Stroke, admitted within 24 h
        Number of patients26302885
        Mean age (SD)71 (13)70 (14)
        Male sex, n (%)1303 (50)1434 (50)
    Ischaemic stroke, admitted within 4 h
        Number of patients, n (%)714 (27)943 (33)
        Mean age (SD)71 (13)70 (14)
        Male sex, n (%)390 (54)512 (54)
        Median NIHSS (IQR)6 (3–12)5 (3–10)
        Onset to door time, median (IQR)75 (50–120)80 (53–123)
        Door to needle time, median (IQR)60 (46–80)64 (45–85)
        Absence of contraindications in tPA-treated patients70 (18)57 (10)b
        Absence of contraindications in not-tPA-treated patients308 (97)377 (99)c
        History of stroke, n (%)137 (24)124 (21)
        History of myocardial infarction, n (%)90 (15)76 (14)
        History of heart failure, n (%)33 (6)69 (12)
        Diabetes mellitus, n (%)91 (16)100 (17)
        Atrial fibrillation, n (%)84 (14)103 (18)
  • aData are averages over centres as units of observation.

  • bP = 0.001, X2.

  • cP = 0.03, X2.

View this table:
Table 4

Association of structural characteristics with the likelihood of being treated with intravenous thrombolysis in twelve centres in the Netherlands

ScoreMean (range)OR (95% CI) (unadjusted)
Structural characteristics19.5 (12.6–27.4)1.03 (0.98–1.08)
Training6.7 (1.5–12.5)1.04 (0.98–1.10)
Protocols5.8 (2.6–7.8)1.02 (0.87–1.19)
Infrastructure7 (4.7–10.5)1.03 (0.91–1.16)
Extramural training4.3 (0.7–7)1.11 (0.99–1.25)
Extramural protocols2.6 (0.42–4.2)0.92 (0.77–1.11)
Extramural infrastructure4.2 (2.5–5.7)2.41 (0.19–30.48)
Intramural training2.4 (0–6.9)1.03 (0.93–1.14)
Intramural protocols3.2 (2.2–3.8)1.46 (1.12–1.91)
  • The OR represents the relative increase in likelihood of being treated per point item score.

The full transcripts of the interviews were made anonymous and were analysed independently by two rates (M.D. and D.D) in a modified Delphi approach. If after three rounds no consensus could be reached a third rater (L.N.) was involved to tip the balance.

Statistical analysis

For the analysis of the association between treatment with thrombolysis and structural characteristics we used a multilevel logistic regression model, in order to adjust for the potential clustering effect.12 Statistical significance was set at 5%. We performed both an unadjusted analysis and an analysis with adjustments for hospital size and teaching facilities. We considered confounding by differences in distribution of patient characteristics between the centres. This was explored by correlating patient characteristics to quantiles of the sum score of both training and protocols. Data were analysed using STATA version 10 (Stata Corp, College Station, Texas, USA).


Overall 5515 stroke patients were registered, 701 (12.7%) were treated with intravenous rtPA. Thrombolyis rates varied between 5.7 and 21.7% for the 12 centres, for a mean of 12.7%, SD 3.9% (Table 3). Increases in thrombolysis rate were not associated with an increase in non-adherence to protocols or occurrence of adverse events. The proportion of patients who were not treated, but had no contra-indication was higher in the group of patients from hospitals with low structural scores (18% vs. 10%, P = 0.001). Also, the proportion of patients who were treated but actually had a contra-indication was higher in the group of patients from hospitals with low structural scores (3% vs. 1%, P = 0.003). We did not have to adjust for patient characteristics, because no relevant statistical associations were found with quantiles of the sum score of structural characteristics. Also, no statistical associations were found between door-to-needle time and quantiles of the sum score of structural characteristics.

Association between structural characteristics and thrombolysis rates

The unadjusted multilevel logistic regression shows a significant association between thrombolysis rates and intramural protocols (Table 4). A higher score on intramural protocols (sum score: presence of all intramural protocols and completeness of protocols for neurologists and nurses) increases the likelihood for patients of being treated with thrombolysis [odds ratio (OR) 1.46, 95% confidence interval (CI) 1.12–1.91]. After adjusting for hospital size and teaching vs. non-teaching hospital, the strength of the association increased [adjusted OR (aOR) 1.77, CI 130–2.39]. The analysis also shows a significant association between extramural training and thrombolysis rate after adjustment (aOR 1.14, CI 1.01–1.30). An analysis of all the individual factors (Table 1) that make up the sum scores of extramural training and intramural protocols, showed no significant association with thrombolysis rate. When extramural training and intramural protocols are combined in one multilevel logistic regression model, only intramural protocols showed a significant association with thrombolysis rate (OR 1.68, CI 1.15–2.46).


This study shows that differences in thrombolysis rates between centres can partly be explained by differences in extramural training and intramural protocols. For extramural training we looked at the availability of education and training in the last 2 years, for the general public in the region, general practitioners and ambulance personnel. Separately these items have no significant association with thrombolysis rates, but combined and after adjustment for hospital size and teaching vs. non-teaching hospital, a higher score increased the likelihood of receiving thrombolysis per patient. Intramural protocols have the strongest association with thrombolysis rates. This characteristic is a sum score made up of the evaluation of thrombolysis protocols for different departments. For the neurologists and stroke nurses we also scored, using a checklist, how complete their protocols are. Again, the individual item-scores (of protocols for different departments and completeness of protocols for neurologists and nurses) have no significant association with thrombolysis rates, but combined they have a strong association. The data suggest that protocol adherence is better in centres with well-developed protocols, as treatment or non-treatment is stronger related to the presence of contra-indicators in these centres. When extramural training and intramural protocols are evaluated together in one multilevel regression model, only intramural protocols show a significant association with thrombolysis rates. It seems that centres with well-developed protocols are also often active in extramural training.

Our findings confirm the EUSI recommendations for public education, professional training and written standards for in-hospital delays.7 Based on our calculations a reasonable improvement in intramural protocol score of ½ SD (0.3 points) might lead to increases of 20% (1% absolute) in thrombolysis rates.

As a single-item public education showed no significant association with thrombolysis rates, probably because the differences between centres were small. Moreover, all centres may have benefitted from a national campaign in 2006, informing the Dutch public about stroke. Intramural training and extramural protocols also showed no significant association with thrombolysis rates. Intramural training was defined as: at least one formal training or presentation given in the last 2 years. In most hospitals formal trainings were given >2 years ago. This may still be sufficient if knowledge is passed on to new employees during day-to-day practice. Although there were differences in extramural protocols for GP's, triage nurses and incident emergency rooms, all ambulance services used the same standardized protocols. Extramural training may also be more effective then extramural protocols. Furthermore, infrastructure of both extramural and intramural showed no significant association with thrombolysis rates. The Netherlands is very densely populated and for most inhabitants a hospital is in reach by ambulance within 15 min. Also if ambulance services have good protocols, it does not matter if there are more hospitals in the region. This also applies for the number of neurologists that are involved. Finally, although there is a large difference in how many hours a week a neurologist is actually present who can decide to perform thrombolysis (58 vs. 168 h), there is no significant association with thrombolysis rate. Possibly because in all centres a neurologist can be present within half an hour and there is always personnel that can start the routing for thrombolysis (EKG, CT, Laboratory).

A limitation of this study is the number of centres involved: only 12 hospitals participated of the ∼100 hospitals in the Netherlands (11%). However, these hospitals cover the whole spectrum of hospital types, from small rural hospitals to the larger academic hospitals and the very large community hospitals. The limited number of hospitals also made it possible to collect not only patient-related numerical data, but also rich contextual information from the different centres. Moreover, we were able to interview a substantial number of health professionals (in total ∼100). We have tried to compensate for subjective bias by using data from different sources: interviews, protocols and diaries to confirm our findings (triangulation).

Although, it has been estimated that 25% of patients with acute stroke could benefit from treatment with thrombolysis,4,13,14 only 2–10% receive this treatment in most hospitals.3,4 This study shows that to improve the thrombolysis rate, extramural training and the introduction of complete intramural protocols involving all relevant professionals are important factors to address.


Dutch Organization for Health Research and Development ZON-MW (grant number 945-14-217). ZON-MW is the national health council appointed by the Ministry of Health (VWS) and the Dutch Organization for Scientific Research (NWO) to promote quality and innovation in the field of health research and care.

Conflict of interest: None declared.


We extend our gratitude to all stroke care teams who participated in this study, and to the PRACTISE investigators. The PRACTISE investigators and their affiliations are listed in the published trial protocol.15


View Abstract