This is a 18-months prospective, multi-center, historical control study among 30 stroke centers in Zhejiang Province, China. Stroke centers with in-hospital stroke patients voluntarily participate in this trial. All medical documents of consecutive AIS patients admitted in the stroke center according to ICD-10 (I63 and G45) are collected by investigators from Zhejiang Stroke Quality Control Center (ZSQCC). Only the de-identified documents are preserved in a safe information database. The clinical data involved in this study belong to the scope of information collection of routine clinical practice. The local infrastructure and characteristics of each recruited center are also recorded. The study protocol is approved by the institutional review board of the Second Affiliated Hospital of Zhejiang University (SAHZU). Because patient information in the CASE is de-identified and anonymized before being released to the researchers, the informed consent requirement is waived by SAHZU Institutional Review Board. All the protocol item is standardized (see the additional file 1). It was registered at ClinicalTrial. gov (NCT03684629).
There are two stages in this trial: pre-intervention stage (the initial 6 months) and post-intervention stage (the next 12 months). We collect the consecutive data of each center during the pre-intervention stage without feedback of KPIs, and then monthly send its own KPIs to each center during the post-intervention stage. We finally compare the changes of KPIs from pre-intervention stage to the last 6 months of post-intervention stage. (Figure 1)
The inclusion criteria for the analysis of medical documents:
- Patients with 18 years or older;
- AIS confirmed by computed tomography (CT) or Diffusion weighted imaging (DWI);
- Admitted within 7 days from symptom onset.
Data monitoring and Confidentiality
After discharge, the original medical document of each patient during hospital stay is saved as images or portable document format (PDF). Specific software recognizes and pre-processes the above material, and send it to multiple Optical Character Recognition (OCR)(15) engines to build documents with recognized text, which are subsequently re-segmented and synthesized in the post-processing. Required data is extracted from the processed text. 300 medical documents were randomly selected to evaluate the error rate of the extraction, ie. the rate of inconsistent data extracted by computer-based medical data collecting system and manual input of neurologists, which was finally 3.4%. Previous report has also demonstrated generalizability and applicability of OCR with a positive predictive value of 94.6%(16). This software also allows the trained study investigators to quickly check the quality of extracted data by pointing out the original extraction site, enabling the authenticity and traceability of original data. The checked data are then imported into the analysing system, and KPIs will be calculated according to the designed formula. (Figure 2)
Intervention for medical care improvement
Individualized quality improvement process is divided into 3 steps :
- Each stroke center receives their own KPIs monthly, as well as the average KPIs level of all 30 stroke centers as contrast, which are automatically calculated by computer analyzing system.
- Based on the performance of KPI data, stroke experts from ZSQCC identify the items which need to be improved, provide each center with relevant education of guidelines, and help the clinicians to examine the causes of non-compliance of guideline-based care and develop programs to decrease their frequency via regular tele-conference per month and face-to-face lectures quarterly. Meantime, stroke centers with the best KPI performance also share their experiences on medical care improvement.
- Each stroke center is then required to provide a detailed improvement plan for each KPI and check whether it is implemented one by one during the next tele-conference.
Primary Outcome: 7 KPIs including the accomplish of National Institutes of Health Stroke Scale (NIHSS), intravenous thrombosis within 4.5 hours of onset, antiplatelet agents therapy within the first 48 hours, statin use after hospitalization and at discharge, antithrombotic therapy at discharge, anticoagulant treatment for patients with atrial fibrillation at discharge(8, 9, 11-13). (see the additional file 2)
Secondary Outcome: 9 KPIs including prevention of deep vein thrombosis, blood vessel evaluation within one week of hospitalization, swallowing function evaluation, blood lipid level assessment, rehabilitation evaluation and implementation, recommendation of life style change, anti-hypertensive therapy, hypoglycemic treatment at discharge, and recurrence rate(10, 11, 17). (see the additional file 2)
Among all the KPI, the adherence to intravenous thrombosis (IVT) was poor in china(18). Result of investigation in Zhejiang before the trial indicated that only approximately 8% of ischemic stroke patients received IVT 4.5h after stroke onset. To detect 10 % improvement in the adherence to IVT, with α=0.05 and power=90%, at least 16570 patients will be needed in pre-intervention stage and post-intervention stage.
Planned data statistical analysis
Statistical analysis will be performed with SPSS 22.0 (SPSS, Inc, Chicago, USA). Baseline characteristics of each stroke center (see the additional file 3) and the patients enrolled will be analyzed. The achieved rates of all KPIs of 30 hospitals during pre-intervention stage and the last 6 months of post-intervention stage are compared by the paired t test. KPIs are also collected at multiple time points before and after intervention. Interrupted time series (ITS) analysis are performed to assess the efficiency on outcome one year after the intervention is carried out(19, 20).The multiple time points before the intervention allow the underlying tend to be estimated, the multiple time points after the intervention allow the intervention effect to be estimated accounting for the underlying trend. No additional adjusted analyses will be performed. All the information to calculat the KPIs were included in the medical records. The missing data not extracted by computer will be extracted manually by clinician. There are no anticipated problems that are detrimental to the participant so that there are no formal stopping rules for the trial. The schedule of enrolment, interventions, and assessments was included in the Figure 3.
Trial management and monitoring
the Trial Management Group (TMG) will be responsible for all aspects of local organization and the day-to-day running of the trial, including monitoring the operation and maintenance of the Computer Analysing system, outcome data collection, and ensuring the feedback of KPIs to the relevant hospitals every month. The trial will be overseen by an independent Trial Steering Committee (TSC), which will meet at least monthly to consider and address strategic issues. A Data Monitoring Committee (DMC), members of which will act independently of the TSC, TMG and funder, will monitor data, check the quality of extracted data, and make recommendations to the TSC on whether there are any ethical or safety reasons why the trial should not continue. The Serious Adverse Events (SAEs) which will be recorded and reported in this trial are deaths by any cause in the 1 year after hospitalization. The DMC will monitor the occurrence of SAEs. Patients information in the CASE is de-identified and anonymized extrated from medical document after discharge.