In this study, we found that adherence to evidence-based performance measures in the acute care of ischemic stroke in a sample of public hospitals in the Metropolitan Region of Santiago was very low. Several programs to improve the quality and delivery of care have prospectively investigated performance measures before and after intervention programs around the world and have described similar results before organizational interventions in acute stroke care [20
In our study, a CT scan was performed within 24
hours of symptom onset in less than 50% of patients and within 4.5
hours in less than 10%, limiting the possibility of acute care, especially intravenous thrombolysis or hemicraniectomy. The literature reveals considerable variability in arrival times to hospital. In the Paul Coverdell Registry, the median time from hospital arrival to brain imaging among all patients was 1
minutes (mean: 2
], while in Korea the median time interval from symptom onset to hospital arrival was 7
]. Consistent with the finding of delays from symptom onset to CT confirmation of diagnosis, intravenous thrombolysis was administered in only 1.7% of the patients, a lower percentage than that in high-income countries [16
], and similar to that reported in Argentina and China, where 1.05% and 2.4% of patients with acute ischemic stroke were treated with intravenous thrombolysis [30
]. This low use was to be expected, since systematic organizational changes have not been implemented in hospitals and thrombolysis is usually administered in hospitals with stroke units [16
We detected substantial deficiencies in the thoroughness and accuracy of registration of the times of symptom onset and CT confirmation of diagnosis. Monitoring the time from symptom onset to the distinct interventions is essential to improve patient care and adherence to evidence-based performance measures in the acute management of stroke and to reduce pre-admission and post-admission barriers to treatment [15
]. Of 56,969 patients in the Paul Coverdell Registry, the time from symptom onset to hospital arrival was unknown or not recorded in 57.8% [27
This study shows that clinical neurological evaluation measures such as the National Institute of Health Stroke Scale (NIHSS) have not been incorporated in routine clinical practice in the emergency departments of the hospitals in the sample. The NIHSS has established validity and reliability for use in the clinical evaluation of stroke patients and is strongly predictive of early functional recovery and long-term outcome. This scale can be administered by physicians, research workers, and nurses alike and has proven intra- and inter-rater reliability [33
]. In addition to the low use of the NIHSS, we also found that early evaluation of neurological deficit was not systematically performed in all the hospitals studied.
Dysphagia screening was particularly low in our study and could strongly influence the incidence of pneumonia. In the Canadian Stroke Network Registry, dysphagia screening was performed in 56% of patients [28
]. A study using the Paul Coverdell Registry showed that unscreened patients were at a higher risk of pneumonia than screened patients [34
In the present study, more than one-fifth of the patients developed in-hospital pneumonia. Post-stroke pneumonia is a potentially preventable complication that is associated with poor outcome [35
]. In the multivariable analysis, the factors associated with pneumonia were age, female sex, a reduced level of consciousness on admission, aphasia, hemiplegia and a diagnosis of diabetes. Neither dysphagia screening nor clinical evaluation care contributed significantly to post-stroke pneumonia. Dysphagia screening was carried out in a very small proportion of patients and could not fully be evaluated in the multivariable model. Clinical evaluation on admission, as defined in current clinical practice, was insufficiently specific to identify other variables with proven prognostic value in the risk of post-stroke pneumonia, such as pre-existing dependency, non-lacunar vs lacunar stroke, chronic obstructive pulmonary disease and other factors [36
]. Finlayson reported that health care determinants, such as stroke unit admission did not predict the occurrence of pneumonia but were, however, associated with decreased mortality from this infection, indicating the need to intensify acute stroke care [36
]. Like Aslayen et al. [37
], we found an increased risk of pneumonia in women and patients with diabetes, which could be explained by the higher risk of infections in diabetic patients.
Among secondary prevention measures, compliance with recommendations on the prescription of antithrombotic agents at discharge was lower than expected and these drugs were not prescribed in a third of the eligible patients; variations among hospitals were also found. This finding is particularly worrisome as prescription of these drugs at discharge is one of the most widely used and cheapest secondary preventive measures. In most registries, compliance with this performance measure was usually above 90% in both high- and low-medium income countries [20
]. Likewise, no record was found in the medical charts of disability evaluation at discharge or the need for community-based rehabilitation. Although not usually considered performance measures of acute stroke care, both measures are associated with improved outcomes [38
The case fatality rates in our study are similar to the mean case fatality rates reported in countries belonging to the Organization for Economic Cooperation and Developmet (OECD) [39
]. Like other studies [35
], we found that the factors that contributed to explaining the likelihood of death at 30
days were pneumonia, age [36
], speech disturbance, reduced level of consciousness and aphasia. Nevertheless, the two interventions selected did not have a significant effect on this outcome.
Our study has several limitations that could bias the results. We were unable to retrieve 17.5% of the sample because of the poor quality of medical registration in the patients’ charts and the impossibility of retrieving the charts in one hospital after the 2010 earthquake. However, no difference was found by demographic characteristics and 30-day fatality between patients included in the sample and those not included. Therefore, we believe that the probability of selection bias is small and that our results reflect the reality of stroke care in these hospitals. Nevertheless, they may not reflect the care in hospitals in more isolated and less populated regions of the country, where standards could be lower and outcomes poorer, as shown in rural Australia by Cadilhac et al. [40
Another limitation of this study is that we could not measure the effect of organized patient care (stroke units) as these facilities are only recently being implemented in some public hospitals in the Metropolitan Region of Santiago. In addition, we had no standardised summary measure of stroke severity such as the NIHSS and relied on other objective clinical measures at time of admission to account for stroke severity in our statistical models.
A strength of this study is that we were able to obtain data from hospitals admitting more that 50% of all patients with ischemic stroke in the country. We believe the findings of this study have good external validity for Chile as the age and sex distribution of the patients included was similar to those of the patients in the only community-based stroke study published to date the PISCIS study [3