Development and implementation of acute stroke care pathway in a tertiary care hospital in India: A cluster-randomized study
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.178038
Source of Support: None, Conflict of Interest: None
Background: In-hospital care of stroke patients can reduce the risk of death and disability. There is an emerging evidence for the routine use of care pathways (CPs) for acute stroke management. Our aim was to develop evidence-based CPs and test the hypothesis that acute stroke management by CPs is superior to that provided by conventional care.
Keywords: Acute stroke; care pathway development; stroke management
Stroke is a major health problem in both the developed and the developing world. In Asian countries, it is the second most common cause of death. Half of all stroke survivors are left with a permanent disability. Care in a hospital stroke unit can reduce the risks of death and disability after stroke. Care pathways (CPs) aim to promote organized and efficient patient care based on the best evidence and guidelines. There is an emerging evidence for the routine use of CPs for acute stroke management. Most of the evidence has been derived from nonrandomized trials conducted in Western countries. None of these studies were done in India. Acute stroke management in India is affected by the lack of knowledge, stroke units, facilities for performance of an urgent computed tomographic (CT) scan and availability of tissue plasminogen activator, as well as poor affordability of the patients. Less than 10% of patients are covered by health insurance in this region and, therefore, have to rely on their own resources or on the subsidized treatment available at government hospitals, which are limited in number and capacity. Most of the available stroke pathways are not based upon the available health care resources. There is an emerging evidence for the need to properly follow a CP in this unique cultural and resource-poor setting. Ours is the first study of its kind that is attempting to develop and evaluate a regional, evidence-based CP that is relevant in a developing nation like India. The aim of the present study was, therefore, to develop a regionally acceptable, evidence-based CP and test the hypothesis that acute stroke management by the CP is superior to the conventional care (CC) in reducing the incidence of aspiration pneumonia, for better monitoring of clinical parameters, and for improving death and dependency outcomes at 90 days.
The project received ethical clearance from the institutional ethics committee. All individual patients gave consent to participate in the study and had the opportunity to withdraw from the study at any point in time. An acute stroke CP was designed to meet the definition of European Pathway Association and was in accordance with the recent ASA/AHA stroke guidelines  and with 2010 IAN/Indian Stroke Association (ISA) guidelines on stroke management in India.  The development of CP was done in several steps: (1) Preparation of the draft of the local CP using various pathways available on public domain (Calgary Regional Health Pathway, Southern Heath and Social Care (HSC) Trust-Acute Stroke Integrated CP, Leeds Acute Stroke CP, Model of Stroke Care-State of Western Australia); (2) peer review; (3) review by nursing staff assigned to implement the CP; (4) each modified draft version was tested; and, (5) the 14 th version of CP was pretested and again modified based on the clinical assessment for the establishment of the final CP. The CP was composed of the following three components: (1) An integrated, multidisciplinary, care protocol document in the form of a checklist to guide the process of care in which important treatment guidelines and various assessment scores were included at the end of CP; (2) swallowing assessment flow chart; and, (3) patient and caregiver education on acute stroke care, rehabilitation, and the need for secondary prevention and lifestyle modifications.
This was a prospective, open-label, nonrandomized study, in which patients with stroke were enrolled either to the CP group or the CC group depending on the ward in which the patient got admission. Initially, the study was planned as a randomized control trial, but while testing the CP, it became clear that contamination was impossible to prevent if the same nurse was caring for patients of both study arms. The study, therefore, followed a cluster-randomized interventional design, with one ward receiving the CP and the other, the CC.
Patients aged 18 years or more, with acute stroke of all types (ischemic/hemorrhagic), admitted in the neurology wards within the first 72 hours of the symptom-onset were enrolled in the study. We excluded patients with subarachnoid hemorrhage and those who refused to give consent for the study.
The diagnosis of stroke and its type were established by a CT/magnetic resonance imaging (MRI) scan done at the time of admission. Protocols were implemented depending on the place/ward of admission. One ward on the fourth floor (NS4) was the interventional arm (CP) and another ward on the fifth floor (NS5), the control arm (CC). The recruiting physician had no control on the choice of the patients getting admission to these wards (NS4/NS5). Work was reviewed on a daily basis by both the treating and the recruiting neurologists for ensuring the proper conduct of the study.
Integrated CP arm-special nursing staff (who received training for CP by the treating physician), dedicated to stroke care, were involved in the implementation of CPs. The staff members were given educational sessions on acute stroke management and rehabilitation at the start of the study and periodically during the study period. This CP was intended towards guiding, sequencing and evaluating the timing of care and treatment decisions specific to stroke, but an integration with the clinical judgment was the primary objective. A swallowing screening protocol was culturally adapted to the regional food habits (pureed food in the Western stroke protocols were replaced by 'Khichri/Kheer' (based on the North Indian diet), or Payasam (based on the South Indian diet). Swallowing assessment was done by the resident doctor-in-charge, beginning from the first day of hospitalization of the patient.
Conventional care protocol that had already been existing in the ward was followed. As the patient reached the ward, vital parameters like blood pressure, heart rate, respiratory rate, and temperature were recorded and managed accordingly with the help of the treating physician. Monitoring of clinical parameters, changing the patients' position, oral suctioning, and chest physiotherapy were done periodically or when needed. All neurology patients with or without stroke were managed together based on similar protocols by the same nursing staff. No special dysphagia assessment protocol was followed, and feeding was started by the resident doctor depending on his clinical judgment.
The data on age, sex, demography, stroke subtype, neurological deficits, and premorbid conditions were collected at baseline for all the patients enrolled. Assessment was done with Barthel Activities of Daily Living Index, modified Rankin Scale (mRS) and National Institutes of Health (NIH) Stroke Scale (NIHSS; at admission and at the time of discharge) by an NIHSS-certified physician (NIHSS certification, given by the National Stroke Association, was obtained during the study period).
The diagnosis of aspiration pneumonia (AP) was made if the patients fulfilled three of the following five criteria: (1) Fever greater than 38ΊC, (2) purulent secretions/sputum, (3) pulmonary consolidation on physical examination, (4) presence of a new or progressive radiographic infiltrate, and, (5) leukocytosis (>12000/mm 3 ) or leukopenia (<4000//mm 3 ).
Other outcome measures being assessed were: (1) Control of clinical parameters like blood pressure, blood sugar, and body temperature; (2) quantitative measures of stroke-related deficits by the NIH Stroke Scale; (3) complications occurring during the hospital stay such as the development of pressure sores or seizures; (4) death during the hospital stay or during the first 3 months of enrolment; (5) Barthel Index and mRS score at discharge and at a follow-up of three months; and, (6) patient and caregiver satisfaction. In view of difficulty in mobilizing the stroke patients, a follow-up assessment at 3 months was done on telephone for all patients, and their Barthel Index and mRS were assessed.
Statistical analyses were carried out using SPSS for Windows (version 21, IBM, USA). For categorical variables, chi-square test and Fisher's exact test were used to calculate the exact P values. For outcomes with continuous data, we first checked the data for normality of distribution. For normally distributed data, comparisons of means were performed using an independent t-test; otherwise, the nonparametric Mann-Whitney U test was used. Spearman's rank correlation coefficient was used to determine correlation. Binary logistic regression was done using model-building steps according to Hosmer and Lemesho.  Briefly, an initial bivariate analysis between the baseline and output predictors was carried out. To investigate if the reduction in death in the CP group was determined by the reduction in the incidence of aspiration pneumonia, we then entered the variable "aspiration pneumonia" into the regression analysis.
GCS score was categorized as a good outcome (9-15) and a bad outcome (3-8). Regression analysis was again done with the new GCS categories. To study the interaction between the CP and the GCS categories, we used the Cochran's Mantel-Haenszel analysis interaction between the CP and GCS categories in predicting death at 90 days of admission. The Cochran's Mantel-Haenszel test stratified for the GCS subgroup scores (3-8 and 9-15) was used to evaluate the data and to further determine the risk ratio. Predictor variables associated with a P < 0.25 were entered into the logistic regression model.
We prospectively enrolled 162 patients, 77 patients in the intervention group and 85 patients in the control group. The mean age was 55.7 years (standard deviation: 15.7 years), 73.5% were male, and ischemic stroke was the most common stroke type (76.5%). The mean age was 55.3 years in the intervention group and 50.3 years in the control group. The two groups were similar for most of the baseline characteristics [Table 1] except that the intervention group had significantly (P = 0.019) more ischemic strokes (84% vs. 69%).
Of the total 162 acute stroke patients enrolled, AP occurred in 18 patients: 5 of the 77 patients (6.49%) in the CP group and 13 of the 85 patients (15.29%) in the CC group. The CP arm had a lower incidence of AP in comparison with the CC arm (6.5% vs. 15.3%, risk ratio [RR] = 0.42, 95% confidence interval [CI] = 0.16-1.14, P = 0.062). On the multivariable regression analysis, CP showed trends toward significance compared with the conventional arm (odds ratio [OR] = 0.33, 95% CI = 0.09-1.22, P = 0.097).
The CP group had a decreased risk of requirement of mechanical ventilation (7.8% vs. 17.6%, OR = 0.39, 95% CI = 0.14-1.07, P = 0.05). The institution of CPs led to a 9.8% absolute risk reduction in the incidence of the need for mechanical ventilation with a number needed to treat (NNT) of 10. Four of six patients in the CP group and 12 of 15 in the CC arm were intubated and put on mechanical ventilator owing to AP. One patient (1.3%) in the CP group and four patients (4.7%) in the CC group died during their hospital stay [Table 2], of which two (50%) deaths in the CC group but none in the CP group were attributable to infection.
After discharge, a telephonic follow-up was done at 90 days, and the mRS and Barthel Index were assessed in all the patients. In total, 50 patients (64.9%) in the CP arm and 53 (65.4%) in the care arm had the follow-up Barthel Index >60; 44 (57.1%) in the CP arm and 49 (57.6%) in the CC arm had a 90-day mRS ≤2. Both the follow-up Barthel Index and mRS showed nonsignificant differences between the groups. Death at 90 days showed a significant difference between the two groups (7.8% vs. 20%, P = 0.022). After adjustment for confounding variables (age, admission GCS, aspiration pneumonia), the CP arm showed a significant trend toward a reduction in the number of deaths during the follow up assessment. There were no significant differences in the other secondary outcome measures such as an adequate control of clinical parameters during the hospital stay (blood glucose, body temperature, blood pressure) and the incidence of seizures and pressure sores. The bivariable correlation analysis is shown in [Table 3].
On higher level interaction analysis between aspiration pneumonia, CP and death at 90 days, the CP lost statistical significance, indicating that reduction in the risk of death with the CP was mitigated through a reduction in aspiration pneumonia. With regard to the inter-relationship among clinically important variables, we found a significant association of the CP group with death within 90 days of admission (P = 0.026), a trend toward significance for AP (P = 0.076), but not for age and admission GCS score. Bivariate logistic regression values of all important prognostic factors are given in [Table 4]. In regression analysis model 1 [Table 5], the AP (P = 0.001) and GCS score at admission (P = 0.005) were significantly associated with death at 90 days of admission, but not with age, CP (P = 0.139), and stroke type (ischemic/hemorrhagic). [Table 6] shows that the two significant determinants of death at 90 days were the admission GCS and CP.
A significant association was found between the categorized GCS score at admission (P < 0.0005) and the CP [Table 6]. We verified the selection of variables in model building based on our clinical judgment through a stepwise method, and the final model had the same two variables, CP and admission GCS. Binary logistic regression analysis showed a significant interaction (P = 0.005) between the CP and the GCS category in predicting death at 90 days of admission. In the subgroup of patients with GCS 9-15, the total number and percentage of deaths at 90 days of admission were 2 (2.8%) in the integrated CP and 13 (20.3%) in the conventional arm [Table 7]. CP was better than CC in reducing the risk of death at 90 days of admission in the GCS subgroup scoring 9-15, reducing the risk of death by 84% when compared with the CC arm.
The first purpose of this study was to develop a regionally acceptable and evidence-based acute stroke CP. The second purpose was to gauge the effects of the implementation of this pathway on standard medical care. We found that CP may benefit patients by reducing the risk of AP (P = 0.06), the need for mechanical ventilation (P = 0.05), and the risk of death at a follow-up of 90 days (P = 0.02). Our study showed a trend towards clinically significant reduction in the incidence of AP by 57.5% (P = 0.062). Similar results were reported by a previous study of 121 patients showing a decrease in AP from 6.7% to 4.1% (a reduction of 38.7%) but the results could not attain statistical significance, probably because of the small numbers.  This study by Odderson et al., clearly mentioned that a screening for the function of swallowing was done on the first day of hospitalization, and the patients did not get any oral feeds until their swallowing function was assessed by a speech therapist or a certified registered nurse.  On the contrary, four nonrandomized studies with 797 patients in the Cochrane review  and another study  with 1480 patients showed no significant difference in the occurrence of AP. However, swallowing assessment was reported to be done only in two of the aforementioned five studies. ,
In a systematic review of dysphagia after stroke by Martino et al., dysphagia occurred in 51%-55% of cases using the clinical testing. There was an increased risk for pneumonia in patients with dysphagia (RR = 3.17, 95% CI = 2.07-4.87) and an even greater risk in patients with aspiration (RR = 11.56, 95% CI, 3.36-39.77). 
A cluster randomized control trial (RCT) by Middleton et al., which focused on an evidence-based management of swallowing dysfunction, reported that the prevalence of AP did not differ between the interventional and noninterventional groups (13 [2%] of 603 patients managed for dysphagia vs. 13 [3%] of 483 controls, P = 0.82).  In another study with 2532 patients, a formal dysphagia screening reduced the rates of pneumonia from 5.4% to 2.4% (P = 0.0016).  Overall, the severity of stroke, dysphagia and aspiration, as well as an advanced age of the patient are the other commonly associated risk factors for stroke-associated pneumonia. We showed that the reduction in AP was partly due to the continuous usage of swallowing assessment protocols.
Although our study was underpowered, we found a significant reduction in the number of patients who needed mechanical ventilation in the CP arm (7.8% vs. 17.6%, OR = 0.39, 95% CI = 0.14-1.07, P = 0.05, NNT = 10). This appears as a rather unrealistically large benefit and may still be due to chance, as the present study was not a randomized controlled trial.
Death, when assessed at a 90-day follow-up, showed a significant reduction in the CP arm (P = 0.026) when compared with the CC arm. A recent clinical study also demonstrated that recorded patient care was the only variable inversely associated with both in-hospital mortality and long hospital stay in acute stroke patients.  On the contrary, in a Cochrane review by Kwan et al., (1 RCT and 2 non-RCT, 783 patients), the aggregate results showed no difference in mortality at 90 days (OR = 0.88, 95% CI = 0.49-1.57, P = 0.7).  Two other recent cluster RCTs with 476 patients  and 1039 patients,  respectively, showed a similar nonsignificant effect on follow-up mortality. In a cluster RCT by Panella et al., of 476 subjects, patients in the CP arm had a significantly lower risk of mortality at 7 days (OR = 0.10, 95% CI = 0.01-0.95) but not at a follow up of 30 days. Our study demonstrates a significant benefit of utilizing CP in reducing the risk of death at 90 days. Studies from specialist stroke units, in particular, have shown pathways to have a limited effect on outcomes, perhaps because the established care practices may already be close to an optimum standard.  Pathways have shown to have a significant effect on dependency rates of patients or their clinical parameters when applied at regional hospitals and in a rural setting. , A previous study has demonstrated better functional prognosis in the lacunar infarct subtype of stroke.  However, the sample size of the present study was inadequate to study this effect.
We found that the risk of death in the good GCS subgroup (9-15) was reduced by 84% in the CP arm when compared with the CC arm. Most of the previous studies done on the CP concept were not analyzed based on the stratification of stroke severity. 
Similar to many other acute stroke studies, our study has several limitations. We had a small sample size and a nonrandomized, nonblinded protocol. The quality of life and cost of hospital stay were not considered. Patients with severe strokes were under-represented. This under-representation of severe stroke was due to the exclusion of patients who presented to us 3 days after the onset of stroke for diverse reasons (the admission of the patient to an intensive care unit or the performance of decompressive hemicraniectomy, or hematoma evacuation). However, patients with severe stroke admitted directly to the department for palliative care were included in the study. Exclusion of these patients with severe stroke might have also accounted for the nonsignificant differences between the groups with respect to functional dependency and mortality. Patients with an intracerebral hematoma were more in the CC group. Deep-seated bleed (basal ganglionic and thalamic) was more common in the CC group, which could be responsible for a worse outcome in the CC group.
This study shows that implementation of CPs improves care and outcome in a group of acute stroke patients. CPs provide benefit by reducing the risk of aspiration pneumonia, the need for mechanical ventilation, and the risk of death when assessed at a follow up of 90 days. We have also shown that a formal dysphagia screening lowers the risk of aspiration pneumonia. This, in turn, probably reduces the 90-day mortality. To the best of our knowledge, this is the first integrated stroke management study in a setting of a developing country like India. The finding in this study needs confirmation in a larger cluster-randomized trial using the final version of the pathway. If confirmed and disseminated, this strategy of using CPs has a significant potential for improving the outcome of stroke patients.
We would like to thank the ward nurses, whose dedicated help was necessary for the successful completion of this study.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5], [Table 6], [Table 7]