Automating the assessment of door-to-imaging time in stroke management using a clinical data warehouse

Automating the assessment of door-to-imaging time in stroke management using a clinical data warehouse | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Automating the assessment of door-to-imaging time in stroke management using a clinical data warehouse Olivier Hassanaly, Matthieu Doutreligne, Pénélope Troude, Agnès Solomiac, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7538212/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 04 Mar, 2026 Read the published version in Scientific Reports → Version 1 posted 10 You are reading this latest preprint version Abstract Assessment of quality and safety indicators (QSI) remains often based on time-consuming manual Electronic Health Record (EHR) review. Through a pilot study in stroke management, we investigated the feasibility of automating the calculation of one French national QSI, the measure of door-to-imaging time (DTI time) i.e. the time interval between the arrival at hospital and the first stroke diagnostic imaging, in a Clinical Data Warehouse (CDW) through a retrospective observational study of patients hospitalized in the Greater Paris University Hospitals (AP-HP) for an acute stroke in 2022. We automatically computed DTI time for more than 6,000 medical records in the CDW using a systematic approach, and validated this method by matching results against the manual AP-HP EHR review from the 2022 French national QSI audit. On this Study population, CDW and manual EHR review methods agreeed on estimating overall indicators, but showed discrepancies in the case-by-case analysis mainly because of human variability both in EHR completion and manual reviewing. Automation looks promising in a context of limited professional resources but requires structured and validated data, interoperability in the case of inter-institutional stays and validly measurable QSI. Health sciences/Health care Health sciences/Medical research clinical data warehouse electronic health records data reuse health care quality indicators stroke Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction National quality measurement policies have been set in many countries since the 2000s. The calculation of these quality-of-care indicators varies from one country to another, depending on the ability to directly access hospital clinical data electronically and structure it to allow their reuse for healthcare management. In most countries these indicators are measured at the hospital level on a yearly basis. The calculations are either based on diagnostic codes from claims data or from manual collection. For example, since 1999, the American Agency for Healthcare Quality (AHRQ) has been deploying a program of quality indicators based on administrative data from the Healthcare Cost and Utilization Project (Lu, 2023 ). The 60 developed indicators inform authorities to plan needs, monitor and improve quality. Some indicators are used for public information or performance-based payments. AHRQ publishes open-source software enabling hospitals to calculate their own indicators. In Europe, such diagnostic-based indicators have been developed in Italy [Colais et al. 2022 ], the Netherlands (Pross, Geissler, and Busse 2017 ) and Germany (Pross, Geissler, and Busse 2017 ; Busse, Nimptsch, and Mansky 2009). In some countries more detailed clinical indicators were developed at the provider or even medical unit level, also at the heart of national quality programs. For example, in the United Kingdom, the NHS has been developing CQUIN (Commissioning for Quality and Innovation) indicators since 2009, based on manual data collection from patient records, or directly using an information system where available (NHS, 2023). By 2023, 17 indicators targeting priority improvement themes were reported on a quarterly basis, in conjunction with a pay-for-performance scheme. Denmark [Johnsen, Mainz, and Bartels. 2017] and Sweden [Levay 2016 ; Fredriksson et al. 2017 ] have also backed their quality policies with publicly funded registries to calculate clinical indicators. These registries are increasingly retrieving information automatically from Electronic Health Records (EHRs), but the majority of variables are still collected manually. Finally, to our knowledge, Israel is the only country which established its quality program on the basis of fully automated clinical indicators, enabling it to measure indicators at the level of the healthcare provider [Dreiher et al. 2020 ]. This automation allows professional teams to access the indicators during meetings and increases the uptake of clinical guidelines in health care facilities. In France, the national quality and safety indicator (QSI) program piloted by the French National Authority for Health (HAS) is mandatory for all public and private health care facilities. These QSIs aim to improve care quality, to inform patients about the quality of hospital care via the Qualiscope website, and are used in the accreditation of hospitals, and steer decisions at the health facility level, as well as for public authority at the regional and national level. The patient's medical record is the gold standard in France for medical information; it is used for the QSI evaluation campaign aimed to assess and analyze practices quality. To measure a QSI, the HAS asks each health care facility to carry out a retrospective manual EHR review, based on a random sample of records drawn from the French national hospital discharge database, and provides guidelines to professionals to ensure reproducibility of the review. Thus, for a given QSI, each facility manually audits up to 70 records to collect all the necessary data, whatever the size and category of the healthcare facility concerned by the indicator. This relatively low number of records is a compromise between an acceptable workload in a context of constrained care professional resources, and having sufficient statistical precision for the indicator. This manual review yields annual results at the hospital level while mobilizing a substantial number of hospital staff. Moreover, it allows only limited feedback at the specialized clinical units due to limited granularity in both time and space. On the other hand, automation using prospectively implemented clinical data offers an opportunity to have comprehensive measures with more frequent data points. Coupled with restitution tools, it could empower local use of the indicators and directly help professionals taking actions for improving quality of care. In recent years, EHRs stored in large Clinical Data Warehouses (CDWs) have become increasingly available in France [Doutreligne, 2023], facilitating the secondary use of clinical data for research and inhospital quality management. These CDWs appear to be a promising source for automating QSIs [Scholte et al. 2016 ] [Ficheur et al. 2016 ], especially those focused on structured data. By exploiting automatically captured timestamps, it seems possible to overcome the shortcomings of manual reviewing: time-consuming, non-exhaustive and prone to data entry errors [Rau et al. 2025 ]. We take advantage of the widespread development of these CDWs to assess the feasibility of automating the calculation of QSIs in the Greater Paris University Hospitals (AP-HP) CDW. As a proof of concept, we aimed to evaluate the possibility to automatically estimate diagnostic delay in stroke management in AP-HP. The time to diagnosis is indeed one of the major determinants of stroke outcomes and it is particularly meaningful to measure it continuously to improve care with regards to the accurate guidelines. The time between arrival at the hospital and first diagnostic imaging is a critical indicator measured by manual QSIs. Beyond the clinical and quality interest of this indicator, the care delay for stroke is of public health concern in France, since a high heterogeneity is suspected [HAS, 2024] and a deterioration of these delays has been shown [Domecq et al. 2020 , Thevenet et al. 2023 .]. We thus decided to focus this pilot study on assessing the automation of the door-to-imaging time (DTI time) [Camporesi et al. 2023 ] i.e the measure of the delay between the arrival in the hospital and the first diagnostic imaging for in-patients with stroke. Methods Study design We performed a retrospective multicenter cohort study on the CDW of AP-HP, which contains routinely collected clinical and administrative data for 11.4 million patients. Our aim was to assess the automation of the stroke DTI time calculation based on the AP-HP CDW. The validity of automated calculation was assessed by comparing the results to the existing manual EHR review of AP-HP records from the French national QSI audit. We followed the REporting of studies Conducted using Observational Routinely collected Data (RECORD) [Benchimol et al. 2015 ] extension of the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) [von Elm et al. 2008] guidelines. Databases/ sources of data The AP-HP CDW contains data for patients, collected during their care in all the 38 AP-HP university hospitals, based on informed consent statement. This study was approved by the Ethical Committee (IRB00011591, CSE-24-06_IQSS_EDSH_AVC), and was conducted in compliance with the Declaration of Helsinki. The database was frozen on 2024-07-04. In this study, we used data from the billing codes information system, EHR data uploaded in the CDW such as notes and administrative entries and imaging metadata from the Picture Archiving and Communication System (PACS) in which AP-HP centralizes all the Digital Imaging and Communications in Medicine (DICOM) objects. In the DICOM model, a study corresponds to an acquisition session and includes several series. A series generally corresponds to a specific modality or the position of a patient on the acquisition device. The DICOM metadata associated with images stored in the PACS contain information on the type of examination, timestamp, protocol, body part and person identifier. Such metadata are available as structured tables in AP-HP CDW. The French national QSI audit database contains all information manually collected in each health care facility in France during the yearly audit. This audit is based on a manual review of the EHR. This database contains the necessary information for computing the DTI time as part of stroke management, notably date and hour of arrival and 1st imaging for patients diagnosed with stroke. For this study we used a de-identified subset of the stroke French national QSI audit concerning patients from AP-HP. Since this audit assesses the entire stroke care pathway and not only the DTI time, we filtered only relevant manual EHR reviews. See detailed inclusion and flowchart in Annex B. CDW population We included adult patients hospitalized in AP-HP for an acute stroke in 2022. Our inclusion was based on ICD-10 codes related to stroke (I60, I61, I62, I63, I64) for inpatient care. We filtered out intra-hospital transfer in order to exclude intra-hospital strokes, thus focusing on stroke direct admissions. We selected stroke diagnostic imaging based on DICOM metadata. We considered diagnostic imaging when the description of the study and/or its associated series mentioned one of the following areas: “cranium, head, circle of willis, cerebral, cephalic, cervical or supra-aortic trunk”. We ensured that a Carestream PACS series was associated with each study, since it ensures the existence of a radiologist clinical report associated with the study. Thus, we were able to identify stroke diagnostic imaging studies based on the DICOM metadata. We retained only inpatient stays with at least one stroke diagnostic imaging between the day of entry to hospital and the day of discharge. See detailed inclusion and flowchart in Annex A. Automated DTI time calculation method The timestamp of arrival in the care site was obtained from the hospital stay start datetime registered in the EHR software. This timestamp is automatically generated when the patient is registered in the EHR at his arrival to hospital. The earliest study timestamp among stroke diagnostic imaging was considered as the timestamp of first diagnostic imaging. This timestamp is automatically generated when the radiology technologist starts the acquisition. Finally, DTI time were obtained by calculating the difference between the timestamp of the first diagnostic imaging and the timestamp of the arrival at the facility. In the national audit, when a diagnostic imaging is identified with a timestamp earlier than the arrival at the hospital timestamp, manual reviewers have to set the DTI time to 0. In order to avoid biases, we followed the same methodology in the CDW method. The first diagnostic imaging timestamp was set equal to the hospital arrival timestamp if there was a diagnostic imaging in a 24-hours window before the administrative hospital arrival timestamp. Validation of automated calculation of DTI time To assess the validity of the CDW computed DTI time, we assessed agreement with the existing manual AP-HP EHR review of the 2022 French national QSI audit. A probabilistic matching between CDW population and manual EHR reviews was performed based on 5 criteria: age, gender, care site, start date of the stay and length of stay. The matching flowchart in Fig. 1 details the construction of the Study Population. CDW and manual EHR review DTI time estimations on the Study Population were compared both on the overall indicators assessed by the QSI evaluation campaign: median DTI time, interquartile range (IQR) and percentage of records with a DTI time lower than 3 hours, and on individual matched records. To understand DTI time discrepancies, we separately assessed agreements between CDW method and manual EHR reviews for the arrival at hospital and the first diagnostic imaging timestamps in addition to the DTI time. We also performed a manual CDW review of patient records for those with the highest discrepancies in DTI time. Additionally, we audited the quality of the DICOM timestamps estimation of the first diagnostic imaging timestamp, thanks to a dedicated manual annotation campaign of the imaging clinical notes linked to DICOM metadata of 300 records randomly selected among the CDW population. The DICOM timestamp corresponded to the first identified diagnostic imaging study for each record and annotators looked for the clinical image timestamp in the corresponding imaging clinical note. Statistical Analysis For the descriptive statistics of study patients, continuous variables were presented as median values and their IQR and categorical covariates as frequencies and percentages (%). Agreement between automatic and manual calculation was assessed for DTI time, arrival at hospital timestamp and first diagnostic imaging timestamp using Bland Altman diagrams with minute precision and absolute Intraclass Correlation Coefficient (ICC) with their 95% confidence interval (CI) [Shahab Haghayegh et al. 2020 ]. In the case of non-gaussian distribution, we considered quartiles rather than the mean and Limit of Agreement interval [Bland, Altman, 1999 ]. To assess the level of agreement between automatic and manual calculation in determining whether the DTI time was under 3 hours, we used Cohen’s Kappa statistic. Results In the CDW population, median DTI time was estimated at 2.7 hours (0.4–11.1) and 3517 records (52%) were found with a DTI time lower than 3 hours. Among the 392 usable manual EHR reviews, 361 were matched with the CDW population, yielding 92% coverage, and correspond to the Study population. There were no duplicates when matching and the 8% unmatched reviews were most likely absent from the CDW population. Table 1 presents characteristics of in-patients, care pathway and DTI time for both the CDW population (n = 6671 records) and the Study population (n = 361 records). Figure 2 displays the statistical distribution of the DTI time computed on the Study population with both methods. CDW method and manual EHR review provided a 12-minute difference for both median DTI time estimation and IQR with 2.5 (1.1-5.0) hours and 2.7 (1.0-5.1) hours respectively. The two methods agreed on estimating the percentage of records with a DTI time lower than 3 hours with respectively 199 (55%) and 192 (53%) records. A 0.77 Cohen’s Kappa indicated substantial agreement between the two methods. DTI time was set to 0 for 16 records in the manual EHR review and 9 records in the CDW method in the case of a first diagnostic imaging timestamp earlier than the arrival timestamp. Table 1 Description of the CDW population and the Study population CDW population N = 6.671 records Study population N = 361 records Age 70 (57–81) 75 (62–85) Female Sex 3.001 (44%) 177 (49%) Stroke Diagnosis Cerebral infarction (I63) 3.508 (52%) 202 (56%) Stroke, unspecified as hemorrhagic or infarcted (I64) 1.035 (15%) 93 (26%) Intracerebral hemorrhage (I61) 1.102 (16%) 36 (10%) Subarachnoid hemorrhage (I60) 670 (10%) 25 (7%) Other non-traumatic intracranial hemorrhages (I62) 456 (7%) 5 (1%) Length of Stay (days) 9 (3–19) 6 (2–13) Type of Imaging TDM angioTDM IRM angioIRM Unknown 2.259 (33%) 1.279 (19%) 1.820 (27%) 688 (10%) 725 (11%) 106 (30%) 81 (22%) 108 (30%) 35 (10%) 31 (9%) Figure 3 displays the Bland Altman diagram evaluating agreement between the two methods. Among the 361 records from the Study population, 266 (74%) were found with a DTI time difference under 1 hour and 80 records (22%) presented an identical result. The absolute ICC was 0.22 [0.12, 0.32]. Figures 4 and 5 respectively display the Bland Altman diagram evaluating agreement between the two methods for estimating arrival at the hospital and the first diagnostic imaging timestamp. Respectively 302 (84%) and 290 (80%) Study population records were found with a difference under 1 hour on arrival and first diagnostic imaging. Identical results were found in 195 (54%) and 148 (41%) records respectively. Absolute ICC was 0.82 [0.78, 0.85] for arrival and 0.14 [0.04, 0.24] for diagnostic imaging. The review of clinical documentation in CDW for records with an absolute difference in DTI time estimations higher than 4 hours (n = 38) allowed us to further investigate discrepancies and classify them in two types: CDW errors and manual EHR reviews errors. During the annotation campaign of the imaging clinical notes linked to DICOM metadata, we found that 23% of the 300 reports mentioned an earlier diagnostic imaging than the one described in the imaging clinical note, indicating diagnostic imaging missingness in the EHR. We let aside these reports, resulting in 230 reports without a mention of an earlier diagnostic imaging among which 120 did not contain any image timestamp. Among the 110 exploitable reports for measuring concordance between DICOM metadata and clinical notes, 41 (73%) had an exact correspondence and 93 (85%) had a difference below 10 minutes. Absolute ICC was 0.80 [0.71, 0.85]. Figure 6 shows the Bland Altman diagram evaluating the concordance between clinical notes and DICOM metadata. One can notice we mainly evaluated imaging performed during daily hours. Discussion Using the CDW of AP-HP, we were able to automate the calculation of the DTI time in stroke management, based on administrative data and DICOM metadata. On the Study population, DTI time statistical distributions from CDW and manual EHR review methods were close and the agreement between the two methods to determine whether the DTI times were lower than 3 hours was substantial. However, individual agreement was poor, mainly due to discrepancies in imaging timestamps. Explanations for the findings The CDW was able to reliably reproduce manual EHR review for arrival time with more than 1 record out of 2 having an exact agreement on the arrival at the facility estimation. The administrative timestamp may not precisely reflect the patient's actual arrival time due to pre-admission or care before administrative administration. Our results show either that the main source for the manual EHR review was the administrative data, which is uploaded in the CDW with overall good quality, or that administrative arrival data is a great proxy of effective arrival comprehension of reviewers. The impossibility to reproduce manual EHR review DTI time computation arised from the disagreement for the first diagnostic imaging timestamp. Although 4 records out of 10 had an exact correspondence, the CDW method tended to provide an earlier diagnostic imaging timestamp than the manual EHR review (Fig. 5 ). These discrepancies can be explained by the multiplicity of sources in the EHR where manual reviewers extract this piece of information. The first diagnostic imaging timestamp may be found in text in medical history, in DICOM metadata, in imaging summary and in a specific box in the initial clinical examination form. Hence systematically using DICOM timestamp for computing the first diagnostic imaging implies differences when comparing with the non-deterministic manual EHR review. Because discrepancy was mainly due to the diagnostic imaging, we carried out an extra analysis of the imaging timestamp concordance between clinical notes and DICOM metadata. It showed that DICOM metadata were reliable in case of data completeness. One quarter of the random samples mentioned an earlier image than the first available imaging. This could typically occur when the diagnostic imaging was performed out of AP-HP or eventually be due to data missingness. Thus, in cases where the CDW method overestimated the first diagnostic imaging timestamp, it is likely that it did not use the proper image. Improving DICOM metadata collection and/or registering in the PACS when the diagnostic imaging was performed out of AP-HP in a structured field would widely improve the reliability of the first diagnostic imaging timestamp and the CDW DTI time computation. A manual review of the records where the disagreement on DTI time was the highest allowed us to classify outliers into two categories. On the one hand, as explained above, in some cases the first diagnostic imaging was not registered in the PACS, thus the CDW was overestimating the DTI time. For other cases, the difference came from inter-observer variability during manual EHR reviewing since one single information can be stored in several data sources with different values. For stroke management, a specific structured medical record exists in the AP-HP EHR with various forms following the patient’s pathway. While medical history is entered in text-boxes, all dates and hours are registered in structured fields but there is no coherence check or automatic completion from one form to another to ensure unicity of the data. This inter-observer variability is even stronger considering that the professionals who care for patients and the ones carrying out the audit do not necessarily have the same knowledge of the EHR structure, and interprofessional variability also exists in the EHR completion. Overall, sources of discrepancy for our specific DTI time indicator were mainly due to complex stays (transfers, imaging done elsewhere), multiple sources for a single information with different values or manual interpretation errors. While being reproducible and systematizable across all patient records, automation makes it possible to avoid manual errors, but requires greater interoperability in the case of inter-institutional stays. Limitations Diagnostic imaging identification from DICOM metadata was based on the non-standardized “description” field. We also excluded imaging studies without PACS report, i.e without clinical report, associated. Because of these two steps, we might have missed some diagnostic imaging during our screening. We based our initial CDW population selection on ICD10 stroke claims, retrieving similar demographic results than a previous epidemiological study [Gabet et al. 2024]. The higher IQR of DTI time in the CDW population compared to the estimations obtained in the Study population suggests that our selection method for the CDW population was not restrictive enough, including less severe stroke stays. For instance, limiting our inclusion to stroke main diagnosis could avoid coding errors but we didn’t explore this option since our main goal was to match as many possible manual EHR reviews from the 2022 French national QSI audit. We must notice that the structured data available in the CDW at time, which are only a subpart of the EHR, do not allow us to fully target the study population. Work is underway to facilitate and accelerate the integration of specific forms in the CDW. Implications for future researches On the one hand, we have been able to reproduce similar results with the manual EHR review for estimating the median DTI time, IQR and the number of records with a delay under 3 hours. Our CDW approach is functional to estimate the overall indicators assessed by the QSI evaluation campaign at the AP-HP scale. On the other hand, the low absolute ICC showed that the CDW method was statistically poorly reliable to reproduce manual DTI time computation. Even though it corrects manual EHR review errors, this approach is not functional yet to replace the latter for a local use to improve practices. Indeed, on a case-by-case basis, major errors can occur that prevent individual investigations. According to the HAS, results of DTI time must be used with caution: this supplementary information released to hospitals for local improvement purposes is to be interpreted within the context of the patient clinical situation and the organisational and healthcare professional practices. At the very least, a complementary use of the CDW method to double-check manual EHR review must be considered. This pilot study highlighted some challenges one should take into account when automating QSIs. Confronting a systematic DTI time computation methodology to a non-deterministic human manual review exhibited that more work needs to be done on measurement validity and reproducibility, for instance working on improving the CDW interoperability. To be useful to clinicians, QSIs must be validly and reliably measurable. Conclusion DTI time automatic computation was systematizable across all patient records and required few resources. It prevented manual reporting errors and was already functional to produce aggregated results. However, it poorly performed in case of complex stays and was not functional yet to replace manual EHR review for a local use to improve practices. Since neither manual review nor automatic approach were 100% error-free, a combination of these two methods should be considered while waiting to improve CDW data collection and interoperability. Overall, room for improvement has been reported in this study. However, more precise and consensual definitions should be made available by the professional organisations with support of health authorities to allow a reliable calculation of recommended QSIs delays in stroke care pathways. Abbreviations EHR Electronic health record CDW Clinical data warehouse DICOM Digital Imaging and Communications in Medicine PACS Picture archiving and communication system DTI Door-to-imaging QSI Quality and Safety Indicator ICC Intraclass Correlation Coefficient HAS French National Authority for Health AP-HP Greater Paris University Hospitals IQR Inter Quartile Range CI Confidence Interval. Declarations Acknowledgements We thank Malamati Voulgaridou, student medicine internship at the HAS for the time she invested in clinical reports annotation for the DICOM metadata quality audit. We also thank the clinical data warehouse of Greater Paris University Hospitals for providing the infrastructure and support for this project. Author contributions OH and MD conducted data analysis. AS, MD, AS and SM conducted the quality audit for the DICOM metadata. PT, LB, AS and SM defined selection criteria to identify usable records from the French national QSI audit database. OH and PT conducted the manual review of manual CDW review of patient records with highest discrepancies. OH, MD, PT drafted the manuscript. AS, SM and LB conceived and designed the study at the HAS level. EJ and JL designed the study at the AP-HP level. All authors (OH, MD, PT, AS, SM, LB, JL, EJ) were involved in interpreting the results and revising the manuscript. All authors read and approved the final manuscript. Data availability statement The implementation of the AP-HP Clinical Data Warehouse has been authorized by the French Data Protection Agency for conducting research on routine data on premises. The individual data shall not be transferred. Agregate data are available from the corresponding author upon reasonable request. Additional Information Ethics approval and consent to participat The research was performed in accordance with relevant guidelines and regulations and approved by the institutional review board (IRB) of the Greater Paris University Hospitals (IRB00011591), administrative decision CSE-24-06_IQSS_EDSH_AVC. The Clinical Data warehouse of the Greater Paris University Hospitals and its IRB have been authorised by the CNIL (Commission nationale de l'informatique et des libertés – French data protection authority) since 19 January 2017 (authorisation no. 1980120). This research was performed in accordance with the Declaration of Helsinki. French regulation does not require the patient’s written consent for this kind of research but in accordance with the European General Data Protection Regulation the patients were informed and those who opposed the secondary use of their data for research were excluded from the study. Competing interests The authors declare no competing interest. Funding The HAS contributed to the financing of the work carried out by the APHP with a sum covering the costs incurred in providing the data and computing resources necessary for the project. This sum of €15,000 is not intended to be compensation for a commercial activity. References Colais, Paola, Luigi Pinnarelli, Francesca Mataloni, Barbara Giordani, Giorgia Duranti, Paola D’Errigo, Stefano Rosato, Fulvia Seccareccia, Giovanni Baglio, et Marina Davoli. 2022. « The National Outcomes Evaluation Programme in Italy: The Impact of Publication of Health Indicators ». International Journal of Environmental Research and Public Health 19 (18): 11685. https://doi.org/10.3390/ijerph191811685. Lu, Huihua. 2023. « Quality Indicator Empirical Methods », AHRQ Quality Indicators, Pross, Christoph, Alexander Geissler, et Reinhard Busse. 2017. « Measuring, Reporting, and Rewarding Quality of Care in 5 Nations: 5 Policy Levers to Enhance Hospital Quality Accountability ». The Milbank Quarterly 95 (1): 136-83. https://doi.org/10.1111/1468-0009.12248. Busse, Reinhard, Ulrike Nimptsch, et Thomas Mansky. 2009. « Measuring, Monitoring, And Managing Quality In Germany’s Hospitals ». Health Affairs 28 (Supplement 2): w294-304. https://doi.org/10.1377/hlthaff.28.2.w294. NHS England. (2022). Commissioning for Quality and Innovation (CQUIN) 2023/24 guidance: Version 1.2 .https://www.england.nhs.uk/wp-content/uploads/2022/12/Commissioning-for-Quality-and-Innovation-CQUIN-2023-24-guidance-version-1.2-1.pdf Johnsen, Søren Paaske, Jan Mainz, et Paul Daniel Bartels. 2017. « National Clinical Registries: Ten Years of Experience with Improving Quality of Care in Denmark ». In Researching Patient Safety and Quality in Healthcare, édité par Karina Aase et Lene Schibevaag, 103-13. Boca Raton, Florida: CRC Press. https://perpus.univpancasila.ac.id/repository/EBUPT180555.pdf#page=127. Levay, C. 2016. « Policies to Foster Quality Improvement Registries: Lessons from the Swedish Case ». Journal of Internal Medicine 279 (2): 160-72. https://doi.org/10.1111/joim.12438. Fredriksson, Mio, Christina Halford, Ann Catrine Eldh, Tobias Dahlström, Sofie Vengberg, Lars Wallin, et Ulrika Winblad. 2017. « Are data from national quality registries used in quality improvement at Swedish hospital clinics? » International Journal for Quality in Health Care 29 (7): 909-15. https://doi.org/10.1093/intqhc/mzx132. Dreiher, Dalia, Olga Blagorazumnaya, Ran Balicer, et Jacob Dreiher. 2020. « National Initiatives to Promote Quality of Care and Patient Safety: Achievements to Date and Challenges Ahead ». Israel Journal of Health Policy Research 9 (1): 62. https://doi.org/10.1186/s13584-020-00417- x. M. Doutreligne, A. Degremont, P-A. Jachiet, A. Lamer, X. Tannier. “Panorama des entrepôts de données hospitaliers dans les CHU/CHR de France”. Revue d'Épidémiologie et de Santé Publique, Volume 71, Supplement 1,2023,101463,ISSN 0398-7620, https://doi.org/10.1016/j.respe.2023.101463. Scholte, M., van Dulmen, S.A., Neeleman-Van der Steen, C.W.M. et al. Data extraction from electronic health records (EHRs) for quality measurement of the physical therapy process: comparison between EHR data and survey data. BMC Med Inform Decis Mak 16, 141 (2016). https://doi.org/10.1186/s12911-016-0382-4 G. Ficheur, et al., Elderly surgical patients: automated computation of healthcare quality indicators by data reuse of EHR, Stud. Health Technol. Inf. 221 (2016) 92–96. doi:10.3233/978-1-61499-633-0-92 Rau, A., Reisert, M., Frank, B. et al. Analyzing temporal imaging patterns in acute ischemic stroke via DICOM-timestamps. Sci Rep 15, 1239 (2025). https://doi.org/10.1038/s41598-025-85315-5 https://www.has-sante.fr/upload/docs/application/pdf/2024-06/iqss_2023_avc_mco_rapport_analyse_2023.pdf?page=24 S. Domecq, F. Saillour-Glénisson, E. Lesaine, F. Gilbert, M. Maugeais, F. Rouanet. Description des délais de prise en charge des patients victimes d’un accident vasculaire cérébral entre l’apparition des symptômes et le traitement dans 11 établissements de santé entre 2012 et 2019 en ex-Aquitaine. Revue d'Épidémiologie et de Santé Publique, Volume 68, Supplement 3 (2020) Pages S135-S136, ISSN 0398-7620, https://doi.org/10.1016/j.respe.2020.03.072. V Thevenet, E Lesaine, S Domecq, S Miganeh-Hadi, M Maugeais, F Rouanet, I Sibon, F Saillour-Glenisson; ObA2 team. Alert on elongated in-hospital acute stroke management delays. An Aquitain cohort study. Rev Neurol (Paris). 2023 Apr;179(4):368-372. doi: 10.1016/j.neurol.2022.07.008. Epub 2022 Nov 3. Camporesi, J., Strumia, S., Di Pilla, A. et al. Stroke pathway performance assessment: a retrospective observational study. BMC Health Serv Res 23, 1391 (2023). https://doi.org/10.1186/s12913-023-10343-8 von Elm E, Altman DG, Egger M, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: Guidelines for reporting observational studies. PLoS Med . 2007;4(10):e296. doi:10.1371/journal.pmed.0040296 Benchimol EI, Smeeth L, Guttmann A, Harron K, Moher D, Petersen I, et al. (2015). The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) Statement. PLoS Medicine , 12(10), e1001885. https://doi.org/10.1371/journal.pmed.1001885 Bland JM, Altman DG. Measuring agreement in method comparison studies. Stat Methods Med Res. 1999 Jun;8(2):135-60. doi: 10.1177/096228029900800204. PMID: 10501650. Haghayegh S, Kang HA, Khoshnevis S, Smolensky MH, Diller KR. A comprehensive guideline for Bland-Altman and intra class correlation calculations to properly compare two methods of measurement and interpret findings. Physiol Meas. 2020 Jun 15;41(5):055012. doi: 10.1088/1361-6579/ab86d6. PMID: 32252039. Amélie Gabet, Yannick Béjot, Emmanuel Touzé, France Woimant, Laurent Suissa, Clémence Grave, Grégory Lailler, Philippe Tuppin, Valérie Olié. “Epidemiology of stroke in France”. Archives of Cardiovascular Diseases, Volume 117, Issue 12, 2024,Pages 682-692,ISSN 1875-2136. https://doi.org/10.1016/j.acvd.2024.10.327. Additional Declarations No competing interests reported. Supplementary Files supplementaryDTItime.pdf Cite Share Download PDF Status: Published Journal Publication published 04 Mar, 2026 Read the published version in Scientific Reports → Version 1 posted Editorial decision: Revision requested 10 Oct, 2025 Reviews received at journal 09 Oct, 2025 Reviews received at journal 01 Oct, 2025 Reviewers agreed at journal 28 Sep, 2025 Reviewers agreed at journal 26 Sep, 2025 Reviewers invited by journal 25 Sep, 2025 Editor assigned by journal 25 Sep, 2025 Editor invited by journal 09 Sep, 2025 Submission checks completed at journal 09 Sep, 2025 First submitted to journal 08 Sep, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7538212","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":525033963,"identity":"64fd9bf0-1a43-4d94-ba50-ded912ffb9d9","order_by":0,"name":"Olivier Hassanaly","email":"data:image/png;base64,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","orcid":"","institution":"AP-HP, Hôpital Saint Louis","correspondingAuthor":true,"prefix":"","firstName":"Olivier","middleName":"","lastName":"Hassanaly","suffix":""},{"id":525033964,"identity":"50a2b57a-231d-4e8e-bff0-6806d64eaf72","order_by":1,"name":"Matthieu Doutreligne","email":"","orcid":"","institution":"Haute Autorité de Santé","correspondingAuthor":false,"prefix":"","firstName":"Matthieu","middleName":"","lastName":"Doutreligne","suffix":""},{"id":525033965,"identity":"0310fb24-ee1c-4486-be21-de1e07d4715e","order_by":2,"name":"Pénélope Troude","email":"","orcid":"","institution":"University Hospital Lariboisière-Fernand-Widal","correspondingAuthor":false,"prefix":"","firstName":"Pénélope","middleName":"","lastName":"Troude","suffix":""},{"id":525033966,"identity":"e934f410-78ee-4557-a784-d4fa56f89da5","order_by":3,"name":"Agnès Solomiac","email":"","orcid":"","institution":"Haute Autorité de Santé","correspondingAuthor":false,"prefix":"","firstName":"Agnès","middleName":"","lastName":"Solomiac","suffix":""},{"id":525033968,"identity":"0210d420-e5f3-4088-a560-e33b2a1242cf","order_by":4,"name":"Sandrine Morin","email":"","orcid":"","institution":"Haute Autorité de Santé","correspondingAuthor":false,"prefix":"","firstName":"Sandrine","middleName":"","lastName":"Morin","suffix":""},{"id":525033970,"identity":"f3a91914-cae8-4f6c-aecc-8b2f8802730e","order_by":5,"name":"Linda Banaei-Bouchareb","email":"","orcid":"","institution":"Haute Autorité de Santé","correspondingAuthor":false,"prefix":"","firstName":"Linda","middleName":"","lastName":"Banaei-Bouchareb","suffix":""},{"id":525033972,"identity":"770bae28-40fc-4443-8d00-13866f491de9","order_by":6,"name":"Jérôme Lambert","email":"","orcid":"","institution":"AP-HP, Hôpital Saint Louis","correspondingAuthor":false,"prefix":"","firstName":"Jérôme","middleName":"","lastName":"Lambert","suffix":""},{"id":525033976,"identity":"332d7233-4faa-4d3a-8005-956744428a04","order_by":7,"name":"Eric Jouvent","email":"","orcid":"","institution":"University Hospital Lariboisière-Fernand-Widal","correspondingAuthor":false,"prefix":"","firstName":"Eric","middleName":"","lastName":"Jouvent","suffix":""}],"badges":[],"createdAt":"2025-09-04 17:08:19","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7538212/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7538212/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-026-41833-4","type":"published","date":"2026-03-04T15:59:03+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":93241335,"identity":"ab0e494c-e8a0-45be-ab17-c1bce259cb7c","added_by":"auto","created_at":"2025-10-10 14:51:30","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":357140,"visible":true,"origin":"","legend":"","description":"","filename":"paperDTItime.docx","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/1f96b59a15d112d5469cec41.docx"},{"id":93238681,"identity":"e7424086-f268-4166-9ed5-8a93b0e32306","added_by":"auto","created_at":"2025-10-10 14:35:30","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9435,"visible":true,"origin":"","legend":"","description":"","filename":"284541e61c914f1182a1631ec2e27222.json","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/51b3111e35a2bd0aa6c57569.json"},{"id":93238678,"identity":"21ade367-f778-4d6a-b90a-789e302c3b48","added_by":"auto","created_at":"2025-10-10 14:35:30","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":127596,"visible":true,"origin":"","legend":"","description":"","filename":"supplementaryDTItime.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/33c60cce99f7d8cd78cd8718.pdf"},{"id":93238685,"identity":"bd8251fb-7481-4eea-bec2-a3d1a2e73d09","added_by":"auto","created_at":"2025-10-10 14:35:30","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":81675,"visible":true,"origin":"","legend":"","description":"","filename":"284541e61c914f1182a1631ec2e272221enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/c0d5565fcf1d480b942408ca.xml"},{"id":93238693,"identity":"c9689929-4fa1-46ba-b52d-a345a9076ad5","added_by":"auto","created_at":"2025-10-10 14:35:30","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":51559,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/d7652c238693ff8b700d80cf.png"},{"id":93238696,"identity":"da3d7f15-1ea3-4a7f-85fb-8feae87969ef","added_by":"auto","created_at":"2025-10-10 14:35:30","extension":"jpeg","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":206548,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/2659b520d16a28c766792f59.jpeg"},{"id":93238684,"identity":"6d72aefd-0897-4c51-b7fd-7d36b6e39d21","added_by":"auto","created_at":"2025-10-10 14:35:30","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":61652,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/59cbacb696d657fc589d5f8c.png"},{"id":93241339,"identity":"9b92762b-83d5-4626-b569-7bdf6161d062","added_by":"auto","created_at":"2025-10-10 14:51:30","extension":"jpeg","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":293455,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/e0d12ac45e666008e3f2a177.jpeg"},{"id":93238697,"identity":"9b3e596f-df85-49d9-a813-63a0c53d2693","added_by":"auto","created_at":"2025-10-10 14:35:30","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":60678,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/7f22461ce6839b078a8836ec.png"},{"id":93240125,"identity":"4693ad1f-fb56-4992-973f-514bf44876a1","added_by":"auto","created_at":"2025-10-10 14:43:30","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":47462,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/1ffb83701e1cf7cce39244dd.png"},{"id":93238688,"identity":"5a672985-b0cb-4147-a643-4770dbd697ee","added_by":"auto","created_at":"2025-10-10 14:35:30","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":15307,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/febb52c99f60f0ee12aa2586.png"},{"id":93238689,"identity":"e74710a2-d6e0-4303-90e4-ca014ebeea90","added_by":"auto","created_at":"2025-10-10 14:35:30","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":43326,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/379670bbde151e3fbcaea455.png"},{"id":93243629,"identity":"3fda0c1f-978c-456d-b920-7a6c1441a08b","added_by":"auto","created_at":"2025-10-10 14:59:30","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":19392,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/40a311f2ec04f7fad7205206.png"},{"id":93241337,"identity":"54653cc9-81ab-4d99-92ab-c747a6d58667","added_by":"auto","created_at":"2025-10-10 14:51:30","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":57137,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/62e502f773e4cb960864e242.png"},{"id":93240129,"identity":"2c87df1a-e2d8-4a8d-8eed-f5c06bcc15c5","added_by":"auto","created_at":"2025-10-10 14:43:30","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":18545,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/d8509cbda07619c2d9193b54.png"},{"id":93238698,"identity":"57905f5e-3817-4069-8c7d-d086b8221dfd","added_by":"auto","created_at":"2025-10-10 14:35:30","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13447,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/51089909c1d46fbd40ec0ac4.png"},{"id":93240131,"identity":"a05d7870-0c00-452a-a477-b9888a7b9766","added_by":"auto","created_at":"2025-10-10 14:43:30","extension":"xml","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":77341,"visible":true,"origin":"","legend":"","description":"","filename":"284541e61c914f1182a1631ec2e272221structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/f74488b60c2e35f9f009690a.xml"},{"id":93238700,"identity":"0f4d7e1b-187b-4103-a133-19f8e74f0854","added_by":"auto","created_at":"2025-10-10 14:35:30","extension":"html","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":89887,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/8b0476ed573c69e9ed0f28a9.html"},{"id":93238677,"identity":"d0f5d0ca-3056-40bd-b025-774d4b45a68d","added_by":"auto","created_at":"2025-10-10 14:35:30","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":51559,"visible":true,"origin":"","legend":"\u003cp\u003eFlowchart\u003c/p\u003e","description":"","filename":"image1.png","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/81fc773d71f8404c97c2889a.png"},{"id":93240122,"identity":"64d8c72b-4598-415e-8592-d59c67e27416","added_by":"auto","created_at":"2025-10-10 14:43:30","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":24137,"visible":true,"origin":"","legend":"\u003cp\u003eStatistical distribution (hours) of DTI time according to method of measure\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/bf75eb4267f163f3965a08f6.png"},{"id":93240123,"identity":"55a31edd-697c-497c-95fe-4ea98407f504","added_by":"auto","created_at":"2025-10-10 14:43:30","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":61652,"visible":true,"origin":"","legend":"\u003cp\u003eDTI time differences between manual and CDW methods\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/04d58544eaced5caa64eed47.png"},{"id":93241336,"identity":"fb5856d4-8eb3-46c8-85fa-29c336e70949","added_by":"auto","created_at":"2025-10-10 14:51:30","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":58143,"visible":true,"origin":"","legend":"\u003cp\u003eArrival differences between manual and CDW methods\u003c/p\u003e","description":"","filename":"image4.png","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/040dfcc7d4d649aece977a6f.png"},{"id":93238683,"identity":"b637b941-e5e3-4170-825e-f469073f2880","added_by":"auto","created_at":"2025-10-10 14:35:30","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":60678,"visible":true,"origin":"","legend":"\u003cp\u003eFirst diagnostic Imaging differences between manual and CDW methods\u003c/p\u003e","description":"","filename":"image5.png","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/6328115674f423ce5295375a.png"},{"id":93238686,"identity":"f06d3675-4b04-4b29-bd4a-e930cf009bcb","added_by":"auto","created_at":"2025-10-10 14:35:30","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":47462,"visible":true,"origin":"","legend":"\u003cp\u003eImaging hour concordance between DICOM metadata and annotations from clinical notes\u003c/p\u003e","description":"","filename":"image6.png","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/e65f169a25879aa5491c7983.png"},{"id":104252047,"identity":"ef38735a-a88c-4eb3-ba0c-ca8d454d8370","added_by":"auto","created_at":"2026-03-09 16:16:58","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":865979,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/cde5be77-ae68-4a9e-8106-46878cf94fb4.pdf"},{"id":93238680,"identity":"0cafb07c-f902-470b-9271-8af2be6cdab3","added_by":"auto","created_at":"2025-10-10 14:35:30","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":127596,"visible":true,"origin":"","legend":"","description":"","filename":"supplementaryDTItime.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7538212/v1/ebc97e0ac92a518f93b9ecd0.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Automating the assessment of door-to-imaging time in stroke management using a clinical data warehouse","fulltext":[{"header":"Introduction","content":"\u003cp\u003eNational quality measurement policies have been set in many countries since the 2000s. The calculation of these quality-of-care indicators varies from one country to another, depending on the ability to directly access hospital clinical data electronically and structure it to allow their reuse for healthcare management. In most countries these indicators are measured at the hospital level on a yearly basis. The calculations are either based on diagnostic codes from claims data or from manual collection. For example, since 1999, the American Agency for Healthcare Quality (AHRQ) has been deploying a program of quality indicators based on administrative data from the Healthcare Cost and Utilization Project (Lu, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). The 60 developed indicators inform authorities to plan needs, monitor and improve quality. Some indicators are used for public information or performance-based payments. AHRQ publishes open-source software enabling hospitals to calculate their own indicators. In Europe, such diagnostic-based indicators have been developed in Italy [Colais et al. \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2022\u003c/span\u003e], the Netherlands (Pross, Geissler, and Busse \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2017\u003c/span\u003e) and Germany (Pross, Geissler, and Busse \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Busse, Nimptsch, and Mansky 2009). In some countries more detailed clinical indicators were developed at the provider or even medical unit level, also at the heart of national quality programs. For example, in the United Kingdom, the NHS has been developing CQUIN (Commissioning for Quality and Innovation) indicators since 2009, based on manual data collection from patient records, or directly using an information system where available (NHS, 2023). By 2023, 17 indicators targeting priority improvement themes were reported on a quarterly basis, in conjunction with a pay-for-performance scheme. Denmark [Johnsen, Mainz, and Bartels. 2017] and Sweden [Levay \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Fredriksson et al. \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2017\u003c/span\u003e] have also backed their quality policies with publicly funded registries to calculate clinical indicators. These registries are increasingly retrieving information automatically from Electronic Health Records (EHRs), but the majority of variables are still collected manually. Finally, to our knowledge, Israel is the only country which established its quality program on the basis of fully automated clinical indicators, enabling it to measure indicators at the level of the healthcare provider [Dreiher et al. \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2020\u003c/span\u003e]. This automation allows professional teams to access the indicators during meetings and increases the uptake of clinical guidelines in health care facilities.\u003c/p\u003e\u003cp\u003eIn France, the national quality and safety indicator (QSI) program piloted by the French National Authority for Health (HAS) is mandatory for all public and private health care facilities. These QSIs aim to improve care quality, to inform patients about the quality of hospital care via the Qualiscope website, and are used in the accreditation of hospitals, and steer decisions at the health facility level, as well as for public authority at the regional and national level. The patient's medical record is the gold standard in France for medical information; it is used for the QSI evaluation campaign aimed to assess and analyze practices quality. To measure a QSI, the HAS asks each health care facility to carry out a retrospective manual EHR review, based on a random sample of records drawn from the French national hospital discharge database, and provides guidelines to professionals to ensure reproducibility of the review. Thus, for a given QSI, each facility manually audits up to 70 records to collect all the necessary data, whatever the size and category of the healthcare facility concerned by the indicator. This relatively low number of records is a compromise between an acceptable workload in a context of constrained care professional resources, and having sufficient statistical precision for the indicator. This manual review yields annual results at the hospital level while mobilizing a substantial number of hospital staff. Moreover, it allows only limited feedback at the specialized clinical units due to limited granularity in both time and space. On the other hand, automation using prospectively implemented clinical data offers an opportunity to have comprehensive measures with more frequent data points. Coupled with restitution tools, it could empower local use of the indicators and directly help professionals taking actions for improving quality of care. In recent years, EHRs stored in large Clinical Data Warehouses (CDWs) have become increasingly available in France [Doutreligne, 2023], facilitating the secondary use of clinical data for research and inhospital quality management. These CDWs appear to be a promising source for automating QSIs [Scholte et al. \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2016\u003c/span\u003e] [Ficheur et al. \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2016\u003c/span\u003e], especially those focused on structured data. By exploiting automatically captured timestamps, it seems possible to overcome the shortcomings of manual reviewing: time-consuming, non-exhaustive and prone to data entry errors [Rau et al. \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2025\u003c/span\u003e]. We take advantage of the widespread development of these CDWs to assess the feasibility of automating the calculation of QSIs in the Greater Paris University Hospitals (AP-HP) CDW.\u003c/p\u003e\u003cp\u003eAs a proof of concept, we aimed to evaluate the possibility to automatically estimate diagnostic delay in stroke management in AP-HP. The time to diagnosis is indeed one of the major determinants of stroke outcomes and it is particularly meaningful to measure it continuously to improve care with regards to the accurate guidelines. The time between arrival at the hospital and first diagnostic imaging is a critical indicator measured by manual QSIs. Beyond the clinical and quality interest of this indicator, the care delay for stroke is of public health concern in France, since a high heterogeneity is suspected [HAS, 2024] and a deterioration of these delays has been shown [Domecq et al. \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2020\u003c/span\u003e, Thevenet et al. \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2023\u003c/span\u003e.]. We thus decided to focus this pilot study on assessing the automation of the door-to-imaging time (DTI time) [Camporesi et al. \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2023\u003c/span\u003e] i.e the measure of the delay between the arrival in the hospital and the first diagnostic imaging for in-patients with stroke.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eStudy design\u003c/h2\u003e\u003cp\u003eWe performed a retrospective multicenter cohort study on the CDW of AP-HP, which contains routinely collected clinical and administrative data for 11.4\u0026nbsp;million patients. Our aim was to assess the automation of the stroke DTI time calculation based on the AP-HP CDW. The validity of automated calculation was assessed by comparing the results to the existing manual EHR review of AP-HP records from the French national QSI audit. We followed the REporting of studies Conducted using Observational Routinely collected Data (RECORD) [Benchimol et al. \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2015\u003c/span\u003e] extension of the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) [von Elm et al. 2008] guidelines.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eDatabases/ sources of data\u003c/h3\u003e\n\u003cp\u003eThe AP-HP CDW contains data for patients, collected during their care in all the 38 AP-HP university hospitals, based on informed consent statement. This study was approved by the Ethical Committee (IRB00011591, CSE-24-06_IQSS_EDSH_AVC), and was conducted in compliance with the Declaration of Helsinki. The database was frozen on 2024-07-04.\u003c/p\u003e\u003cp\u003eIn this study, we used data from the billing codes information system, EHR data uploaded in the CDW such as notes and administrative entries and imaging metadata from the Picture Archiving and Communication System (PACS) in which AP-HP centralizes all the Digital Imaging and Communications in Medicine (DICOM) objects. In the DICOM model, a study corresponds to an acquisition session and includes several series. A series generally corresponds to a specific modality or the position of a patient on the acquisition device. The DICOM metadata associated with images stored in the PACS contain information on the type of examination, timestamp, protocol, body part and person identifier. Such metadata are available as structured tables in AP-HP CDW.\u003c/p\u003e\u003cp\u003eThe French national QSI audit database contains all information manually collected in each health care facility in France during the yearly audit. This audit is based on a manual review of the EHR. This database contains the necessary information for computing the DTI time as part of stroke management, notably date and hour of arrival and 1st imaging for patients diagnosed with stroke. For this study we used a de-identified subset of the stroke French national QSI audit concerning patients from AP-HP. Since this audit assesses the entire stroke care pathway and not only the DTI time, we filtered only relevant manual EHR reviews. See detailed inclusion and flowchart in Annex B.\u003c/p\u003e\n\u003ch3\u003eCDW population\u003c/h3\u003e\n\u003cp\u003eWe included adult patients hospitalized in AP-HP for an acute stroke in 2022. Our inclusion was based on ICD-10 codes related to stroke (I60, I61, I62, I63, I64) for inpatient care. We filtered out intra-hospital transfer in order to exclude intra-hospital strokes, thus focusing on stroke direct admissions. We selected stroke diagnostic imaging based on DICOM metadata. We considered diagnostic imaging when the description of the study and/or its associated series mentioned one of the following areas: \u0026ldquo;cranium, head, circle of willis, cerebral, cephalic, cervical or supra-aortic trunk\u0026rdquo;. We ensured that a Carestream PACS series was associated with each study, since it ensures the existence of a radiologist clinical report associated with the study. Thus, we were able to identify stroke diagnostic imaging studies based on the DICOM metadata. We retained only inpatient stays with at least one stroke diagnostic imaging between the day of entry to hospital and the day of discharge. See detailed inclusion and flowchart in Annex A.\u003c/p\u003e\n\u003ch3\u003eAutomated DTI time calculation method\u003c/h3\u003e\n\u003cp\u003eThe timestamp of arrival in the care site was obtained from the hospital stay start datetime registered in the EHR software. This timestamp is automatically generated when the patient is registered in the EHR at his arrival to hospital. The earliest study timestamp among stroke diagnostic imaging was considered as the timestamp of first diagnostic imaging. This timestamp is automatically generated when the radiology technologist starts the acquisition. Finally, DTI time were obtained by calculating the difference between the timestamp of the first diagnostic imaging and the timestamp of the arrival at the facility.\u003c/p\u003e\u003cp\u003eIn the national audit, when a diagnostic imaging is identified with a timestamp earlier than the arrival at the hospital timestamp, manual reviewers have to set the DTI time to 0. In order to avoid biases, we followed the same methodology in the CDW method. The first diagnostic imaging timestamp was set equal to the hospital arrival timestamp if there was a diagnostic imaging in a 24-hours window before the administrative hospital arrival timestamp.\u003c/p\u003e\n\u003ch3\u003eValidation of automated calculation of DTI time\u003c/h3\u003e\n\u003cp\u003eTo assess the validity of the CDW computed DTI time, we assessed agreement with the existing manual AP-HP EHR review of the 2022 French national QSI audit. A probabilistic matching between CDW population and manual EHR reviews was performed based on 5 criteria: age, gender, care site, start date of the stay and length of stay. The matching flowchart in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e details the construction of the Study Population. CDW and manual EHR review DTI time estimations on the Study Population were compared both on the overall indicators assessed by the QSI evaluation campaign: median DTI time, interquartile range (IQR) and percentage of records with a DTI time lower than 3 hours, and on individual matched records. To understand DTI time discrepancies, we separately assessed agreements between CDW method and manual EHR reviews for the arrival at hospital and the first diagnostic imaging timestamps in addition to the DTI time. We also performed a manual CDW review of patient records for those with the highest discrepancies in DTI time.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eAdditionally, we audited the quality of the DICOM timestamps estimation of the first diagnostic imaging timestamp, thanks to a dedicated manual annotation campaign of the imaging clinical notes linked to DICOM metadata of 300 records randomly selected among the CDW population. The DICOM timestamp corresponded to the first identified diagnostic imaging study for each record and annotators looked for the clinical image timestamp in the corresponding imaging clinical note.\u003c/p\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003eStatistical Analysis\u003c/h2\u003e\u003cp\u003eFor the descriptive statistics of study patients, continuous variables were presented as median values and their IQR and categorical covariates as frequencies and percentages (%).\u003c/p\u003e\u003cp\u003eAgreement between automatic and manual calculation was assessed for DTI time, arrival at hospital timestamp and first diagnostic imaging timestamp using Bland Altman diagrams with minute precision and absolute Intraclass Correlation Coefficient (ICC) with their 95% confidence interval (CI) [Shahab Haghayegh et al. \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2020\u003c/span\u003e]. In the case of non-gaussian distribution, we considered quartiles rather than the mean and Limit of Agreement interval [Bland, Altman, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e1999\u003c/span\u003e]. To assess the level of agreement between automatic and manual calculation in determining whether the DTI time was under 3 hours, we used Cohen\u0026rsquo;s Kappa statistic.\u003c/p\u003e\u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003eIn the CDW population, median DTI time was estimated at 2.7 hours (0.4\u0026ndash;11.1) and 3517 records (52%) were found with a DTI time lower than 3 hours. Among the 392 usable manual EHR reviews, 361 were matched with the CDW population, yielding 92% coverage, and correspond to the Study population. There were no duplicates when matching and the 8% unmatched reviews were most likely absent from the CDW population. Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e presents characteristics of in-patients, care pathway and DTI time for both the CDW population (n\u0026thinsp;=\u0026thinsp;6671 records) and the Study population (n\u0026thinsp;=\u0026thinsp;361 records). Figure\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e displays the statistical distribution of the DTI time computed on the Study population with both methods. CDW method and manual EHR review provided a 12-minute difference for both median DTI time estimation and IQR with 2.5 (1.1-5.0) hours and 2.7 (1.0-5.1) hours respectively. The two methods agreed on estimating the percentage of records with a DTI time lower than 3 hours with respectively 199 (55%) and 192 (53%) records. A 0.77 Cohen\u0026rsquo;s Kappa indicated substantial agreement between the two methods. DTI time was set to 0 for 16 records in the manual EHR review and 9 records in the CDW method in the case of a first diagnostic imaging timestamp earlier than the arrival timestamp.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eDescription of the CDW population and the Study population\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"3\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eCDW population\u003c/p\u003e\u003cp\u003eN\u0026thinsp;=\u0026thinsp;6.671 records\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eStudy population\u003c/p\u003e\u003cp\u003eN\u0026thinsp;=\u0026thinsp;361 records\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eAge\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e70 (57\u0026ndash;81)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e75 (62\u0026ndash;85)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eFemale Sex\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e3.001 (44%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e177 (49%)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eStroke Diagnosis\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cem\u003eCerebral infarction (I63)\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e3.508 (52%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e202 (56%)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cem\u003eStroke, unspecified as hemorrhagic or infarcted (I64)\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.035 (15%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e93 (26%)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cem\u003eIntracerebral hemorrhage (I61)\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.102 (16%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e36 (10%)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cem\u003eSubarachnoid hemorrhage (I60)\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e670 (10%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e25 (7%)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cem\u003eOther non-traumatic intracranial hemorrhages (I62)\u003c/em\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e456 (7%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e5 (1%)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eLength of Stay (days)\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e9 (3\u0026ndash;19)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e6 (2\u0026ndash;13)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eType of Imaging\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTDM\u003c/p\u003e\u003cp\u003eangioTDM\u003c/p\u003e\u003cp\u003eIRM\u003c/p\u003e\u003cp\u003eangioIRM\u003c/p\u003e\u003cp\u003eUnknown\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2.259 (33%)\u003c/p\u003e\u003cp\u003e1.279 (19%)\u003c/p\u003e\u003cp\u003e1.820 (27%)\u003c/p\u003e\u003cp\u003e688 (10%)\u003c/p\u003e\u003cp\u003e725 (11%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e106 (30%)\u003c/p\u003e\u003cp\u003e81 (22%)\u003c/p\u003e\u003cp\u003e108 (30%)\u003c/p\u003e\u003cp\u003e35 (10%)\u003c/p\u003e\u003cp\u003e31 (9%)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFigure \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e displays the Bland Altman diagram evaluating agreement between the two methods. Among the 361 records from the Study population, 266 (74%) were found with a DTI time difference under 1 hour and 80 records (22%) presented an identical result. The absolute ICC was 0.22 [0.12, 0.32]. Figures\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e and \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e respectively display the Bland Altman diagram evaluating agreement between the two methods for estimating arrival at the hospital and the first diagnostic imaging timestamp. Respectively 302 (84%) and 290 (80%) Study population records were found with a difference under 1 hour on arrival and first diagnostic imaging. Identical results were found in 195 (54%) and 148 (41%) records respectively. Absolute ICC was 0.82 [0.78, 0.85] for arrival and 0.14 [0.04, 0.24] for diagnostic imaging. The review of clinical documentation in CDW for records with an absolute difference in DTI time estimations higher than 4 hours (n\u0026thinsp;=\u0026thinsp;38) allowed us to further investigate discrepancies and classify them in two types: CDW errors and manual EHR reviews errors.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eDuring the annotation campaign of the imaging clinical notes linked to DICOM metadata, we found that 23% of the 300 reports mentioned an earlier diagnostic imaging than the one described in the imaging clinical note, indicating diagnostic imaging missingness in the EHR. We let aside these reports, resulting in 230 reports without a mention of an earlier diagnostic imaging among which 120 did not contain any image timestamp. Among the 110 exploitable reports for measuring concordance between DICOM metadata and clinical notes, 41 (73%) had an exact correspondence and 93 (85%) had a difference below 10 minutes. Absolute ICC was 0.80 [0.71, 0.85]. Figure\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e shows the Bland Altman diagram evaluating the concordance between clinical notes and DICOM metadata. One can notice we mainly evaluated imaging performed during daily hours.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eUsing the CDW of AP-HP, we were able to automate the calculation of the DTI time in stroke management, based on administrative data and DICOM metadata. On the Study population, DTI time statistical distributions from CDW and manual EHR review methods were close and the agreement between the two methods to determine whether the DTI times were lower than 3 hours was substantial. However, individual agreement was poor, mainly due to discrepancies in imaging timestamps.\u003c/p\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003eExplanations for the findings\u003c/h2\u003e\u003cp\u003e The CDW was able to reliably reproduce manual EHR review for arrival time with more than 1 record out of 2 having an exact agreement on the arrival at the facility estimation. The administrative timestamp may not precisely reflect the patient's actual arrival time due to pre-admission or care before administrative administration. Our results show either that the main source for the manual EHR review was the administrative data, which is uploaded in the CDW with overall good quality, or that administrative arrival data is a great proxy of effective arrival comprehension of reviewers. The impossibility to reproduce manual EHR review DTI time computation arised from the disagreement for the first diagnostic imaging timestamp. Although 4 records out of 10 had an exact correspondence, the CDW method tended to provide an earlier diagnostic imaging timestamp than the manual EHR review (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). These discrepancies can be explained by the multiplicity of sources in the EHR where manual reviewers extract this piece of information. The first diagnostic imaging timestamp may be found in text in medical history, in DICOM metadata, in imaging summary and in a specific box in the initial clinical examination form. Hence systematically using DICOM timestamp for computing the first diagnostic imaging implies differences when comparing with the non-deterministic manual EHR review.\u003c/p\u003e\u003cp\u003eBecause discrepancy was mainly due to the diagnostic imaging, we carried out an extra analysis of the imaging timestamp concordance between clinical notes and DICOM metadata. It showed that DICOM metadata were reliable in case of data completeness. One quarter of the random samples mentioned an earlier image than the first available imaging. This could typically occur when the diagnostic imaging was performed out of AP-HP or eventually be due to data missingness. Thus, in cases where the CDW method overestimated the first diagnostic imaging timestamp, it is likely that it did not use the proper image. Improving DICOM metadata collection and/or registering in the PACS when the diagnostic imaging was performed out of AP-HP in a structured field would widely improve the reliability of the first diagnostic imaging timestamp and the CDW DTI time computation.\u003c/p\u003e\u003cp\u003eA manual review of the records where the disagreement on DTI time was the highest allowed us to classify outliers into two categories. On the one hand, as explained above, in some cases the first diagnostic imaging was not registered in the PACS, thus the CDW was overestimating the DTI time. For other cases, the difference came from inter-observer variability during manual EHR reviewing since one single information can be stored in several data sources with different values. For stroke management, a specific structured medical record exists in the AP-HP EHR with various forms following the patient\u0026rsquo;s pathway. While medical history is entered in text-boxes, all dates and hours are registered in structured fields but there is no coherence check or automatic completion from one form to another to ensure unicity of the data. This inter-observer variability is even stronger considering that the professionals who care for patients and the ones carrying out the audit do not necessarily have the same knowledge of the EHR structure, and interprofessional variability also exists in the EHR completion.\u003c/p\u003e\u003cp\u003eOverall, sources of discrepancy for our specific DTI time indicator were mainly due to complex stays (transfers, imaging done elsewhere), multiple sources for a single information with different values or manual interpretation errors. While being reproducible and systematizable across all patient records, automation makes it possible to avoid manual errors, but requires greater interoperability in the case of inter-institutional stays.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003eLimitations\u003c/h2\u003e\u003cp\u003eDiagnostic imaging identification from DICOM metadata was based on the non-standardized \u0026ldquo;description\u0026rdquo; field. We also excluded imaging studies without PACS report, i.e without clinical report, associated. Because of these two steps, we might have missed some diagnostic imaging during our screening.\u003c/p\u003e\u003cp\u003eWe based our initial CDW population selection on ICD10 stroke claims, retrieving similar demographic results than a previous epidemiological study [Gabet et al. 2024]. The higher IQR of DTI time in the CDW population compared to the estimations obtained in the Study population suggests that our selection method for the CDW population was not restrictive enough, including less severe stroke stays. For instance, limiting our inclusion to stroke main diagnosis could avoid coding errors but we didn\u0026rsquo;t explore this option since our main goal was to match as many possible manual EHR reviews from the 2022 French national QSI audit. We must notice that the structured data available in the CDW at time, which are only a subpart of the EHR, do not allow us to fully target the study population. Work is underway to facilitate and accelerate the integration of specific forms in the CDW.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003eImplications for future researches\u003c/h2\u003e\u003cp\u003eOn the one hand, we have been able to reproduce similar results with the manual EHR review for estimating the median DTI time, IQR and the number of records with a delay under 3 hours. Our CDW approach is functional to estimate the overall indicators assessed by the QSI evaluation campaign at the AP-HP scale. On the other hand, the low absolute ICC showed that the CDW method was statistically poorly reliable to reproduce manual DTI time computation. Even though it corrects manual EHR review errors, this approach is not functional yet to replace the latter for a local use to improve practices. Indeed, on a case-by-case basis, major errors can occur that prevent individual investigations. According to the HAS, results of DTI time must be used with caution: this supplementary information released to hospitals for local improvement purposes is to be interpreted within the context of the patient clinical situation and the organisational and healthcare professional practices. At the very least, a complementary use of the CDW method to double-check manual EHR review must be considered. This pilot study highlighted some challenges one should take into account when automating QSIs. Confronting a systematic DTI time computation methodology to a non-deterministic human manual review exhibited that more work needs to be done on measurement validity and reproducibility, for instance working on improving the CDW interoperability. To be useful to clinicians, QSIs must be validly and reliably measurable.\u003c/p\u003e\u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eDTI time automatic computation was systematizable across all patient records and required few resources. It prevented manual reporting errors and was already functional to produce aggregated results. However, it poorly performed in case of complex stays and was not functional yet to replace manual EHR review for a local use to improve practices. Since neither manual review nor automatic approach were 100% error-free, a combination of these two methods should be considered while waiting to improve CDW data collection and interoperability. Overall, room for improvement has been reported in this study. However, more precise and consensual definitions should be made available by the professional organisations with support of health authorities to allow a reliable calculation of recommended QSIs delays in stroke care pathways.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eEHR\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eElectronic health record\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eCDW\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eClinical data warehouse\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eDICOM\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eDigital Imaging and Communications in Medicine\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003ePACS\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003ePicture archiving and communication system\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eDTI\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eDoor-to-imaging\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eQSI\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eQuality and Safety Indicator\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eICC\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eIntraclass Correlation Coefficient\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eHAS\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eFrench National Authority for Health\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eAP-HP\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eGreater Paris University Hospitals\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eIQR\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eInter Quartile Range\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eCI\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eConfidence Interval.\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe thank Malamati Voulgaridou, student medicine internship at the HAS for the time she invested in clinical reports annotation for the DICOM metadata quality audit. We also thank the clinical data warehouse of Greater Paris University Hospitals for providing the infrastructure \u0026nbsp;and support for this project.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOH and MD conducted data analysis. AS, MD, AS and SM\u0026nbsp;conducted the quality audit for the DICOM metadata. PT, LB, AS and SM defined selection criteria to identify usable records from the French national QSI audit database.\u0026nbsp;OH and PT conducted the manual review of\u0026nbsp;manual CDW review of patient records with highest discrepancies.\u0026nbsp;OH, MD, PT drafted the manuscript. AS, SM and LB conceived and designed the study at the HAS level. EJ and JL designed the study at the AP-HP level. All authors (OH, MD, PT, AS, SM, LB, JL, EJ) were involved in interpreting the results and revising the manuscript. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe implementation of the AP-HP Clinical Data Warehouse has been authorized by the French Data Protection Agency for conducting research on routine data on premises. The individual data shall not be transferred. Agregate data are available from the corresponding author upon reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAdditional Information\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participat\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe research was performed in accordance with relevant guidelines and regulations and approved by the institutional review board (IRB) of the Greater Paris University Hospitals (IRB00011591), administrative decision CSE-24-06_IQSS_EDSH_AVC. The Clinical Data warehouse of the Greater Paris University Hospitals and its IRB have been authorised by the CNIL (Commission nationale de l\u0026apos;informatique et des libert\u0026eacute;s \u0026ndash; French data protection authority) since 19 January 2017 (authorisation no. 1980120). This research was performed in accordance with the Declaration of Helsinki. French regulation does not require the patient\u0026rsquo;s written consent for this kind of research but in accordance with the European General Data Protection Regulation the patients were informed and those who opposed the secondary use of their data for research were excluded from the study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe HAS contributed to the financing of the work carried out by the APHP with a sum covering the costs incurred in providing the data and computing resources necessary for the project. This sum of \u0026euro;15,000 is not intended to be compensation for a commercial activity.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eColais, Paola, Luigi Pinnarelli, Francesca Mataloni, Barbara Giordani, Giorgia Duranti, Paola D\u0026rsquo;Errigo, Stefano Rosato, Fulvia Seccareccia, Giovanni Baglio, et Marina Davoli. 2022. \u0026laquo; The National Outcomes Evaluation Programme in Italy: The Impact of Publication of Health Indicators \u0026raquo;. International Journal of Environmental Research and Public Health 19 (18): 11685. https://doi.org/10.3390/ijerph191811685.\u003c/li\u003e\n \u003cli\u003eLu, Huihua. 2023. \u0026laquo; Quality Indicator Empirical Methods \u0026raquo;, AHRQ Quality Indicators,\u003c/li\u003e\n \u003cli\u003ePross, Christoph, Alexander Geissler, et Reinhard Busse. 2017. \u0026laquo; Measuring, Reporting, and Rewarding Quality of Care in 5 Nations: 5 Policy Levers to Enhance Hospital Quality Accountability \u0026raquo;. The Milbank Quarterly 95 (1): 136-83. https://doi.org/10.1111/1468-0009.12248.\u003c/li\u003e\n \u003cli\u003eBusse, Reinhard, Ulrike Nimptsch, et Thomas Mansky. 2009. \u0026laquo; Measuring, Monitoring, And Managing Quality In Germany\u0026rsquo;s Hospitals \u0026raquo;. Health Affairs 28 (Supplement 2): w294-304. https://doi.org/10.1377/hlthaff.28.2.w294.\u003c/li\u003e\n \u003cli\u003eNHS England. (2022). \u003cem\u003eCommissioning for Quality and Innovation (CQUIN) 2023/24 guidance: Version 1.2\u003c/em\u003e.https://www.england.nhs.uk/wp-content/uploads/2022/12/Commissioning-for-Quality-and-Innovation-CQUIN-2023-24-guidance-version-1.2-1.pdf\u003c/li\u003e\n \u003cli\u003eJohnsen, S\u0026oslash;ren Paaske, Jan Mainz, et Paul Daniel Bartels. 2017. \u0026laquo; National Clinical Registries: Ten Years of Experience with Improving Quality of Care in Denmark \u0026raquo;. In Researching Patient Safety and Quality in Healthcare, \u0026eacute;dit\u0026eacute; par Karina Aase et Lene Schibevaag, 103-13. Boca Raton, Florida: CRC Press. https://perpus.univpancasila.ac.id/repository/EBUPT180555.pdf#page=127.\u003c/li\u003e\n \u003cli\u003eLevay, C. 2016. \u0026laquo; Policies to Foster Quality Improvement Registries: Lessons from the Swedish Case \u0026raquo;. Journal of Internal Medicine 279 (2): 160-72. https://doi.org/10.1111/joim.12438.\u003c/li\u003e\n \u003cli\u003eFredriksson, Mio, Christina Halford, Ann Catrine Eldh, Tobias Dahlstr\u0026ouml;m, Sofie Vengberg, Lars Wallin, et Ulrika Winblad. 2017. \u0026laquo; Are data from national quality registries used in quality improvement at Swedish hospital clinics? \u0026raquo; International Journal for Quality in Health Care 29 (7): 909-15. https://doi.org/10.1093/intqhc/mzx132.\u003c/li\u003e\n \u003cli\u003eDreiher, Dalia, Olga Blagorazumnaya, Ran Balicer, et Jacob Dreiher. 2020. \u0026laquo; National Initiatives to Promote Quality of Care and Patient Safety: Achievements to Date and Challenges Ahead \u0026raquo;. Israel Journal of Health Policy Research 9 (1): 62. https://doi.org/10.1186/s13584-020-00417- x.\u003c/li\u003e\n \u003cli\u003eM. Doutreligne, A. Degremont, P-A. Jachiet, A. Lamer, X. Tannier. \u0026ldquo;Panorama des entrep\u0026ocirc;ts de donn\u0026eacute;es hospitaliers dans les CHU/CHR de France\u0026rdquo;. Revue d\u0026apos;\u0026Eacute;pid\u0026eacute;miologie et de Sant\u0026eacute; Publique, Volume 71, Supplement 1,2023,101463,ISSN 0398-7620, https://doi.org/10.1016/j.respe.2023.101463.\u003c/li\u003e\n \u003cli\u003eScholte, M., van Dulmen, S.A., Neeleman-Van der Steen, C.W.M. et al. Data extraction from electronic health records (EHRs) for quality measurement of the physical therapy process: comparison between EHR data and survey data. BMC Med Inform Decis Mak 16, 141 (2016). https://doi.org/10.1186/s12911-016-0382-4\u003c/li\u003e\n \u003cli\u003eG. Ficheur, et al., Elderly surgical patients: automated computation of healthcare quality indicators by data reuse of EHR, Stud. Health Technol. Inf. 221 (2016) 92\u0026ndash;96. doi:10.3233/978-1-61499-633-0-92\u003c/li\u003e\n \u003cli\u003eRau, A., Reisert, M., Frank, B. et al. Analyzing temporal imaging patterns in acute ischemic stroke via DICOM-timestamps. Sci Rep 15, 1239 (2025). https://doi.org/10.1038/s41598-025-85315-5\u003c/li\u003e\n \u003cli\u003ehttps://www.has-sante.fr/upload/docs/application/pdf/2024-06/iqss_2023_avc_mco_rapport_analyse_2023.pdf?page=24\u003c/li\u003e\n \u003cli\u003eS. Domecq, F. Saillour-Gl\u0026eacute;nisson, E. Lesaine, F. Gilbert, M. Maugeais, F. Rouanet. Description des d\u0026eacute;lais de prise en charge des patients victimes d\u0026rsquo;un accident vasculaire c\u0026eacute;r\u0026eacute;bral entre l\u0026rsquo;apparition des sympt\u0026ocirc;mes et le traitement dans 11 \u0026eacute;tablissements de sant\u0026eacute; entre 2012 et 2019 en ex-Aquitaine. Revue d\u0026apos;\u0026Eacute;pid\u0026eacute;miologie et de Sant\u0026eacute; Publique, Volume 68, Supplement 3 (2020) Pages S135-S136, ISSN 0398-7620, https://doi.org/10.1016/j.respe.2020.03.072.\u003c/li\u003e\n \u003cli\u003eV Thevenet, E Lesaine, S Domecq, S Miganeh-Hadi, M Maugeais, F Rouanet, I Sibon, F Saillour-Glenisson; ObA2 team. Alert on elongated in-hospital acute stroke management delays. An Aquitain cohort study. Rev Neurol (Paris). 2023 Apr;179(4):368-372. doi: 10.1016/j.neurol.2022.07.008. Epub 2022 Nov 3.\u003c/li\u003e\n \u003cli\u003eCamporesi, J., Strumia, S., Di Pilla, A. et al. Stroke pathway performance assessment: a retrospective observational study. BMC Health Serv Res 23, 1391 (2023). https://doi.org/10.1186/s12913-023-10343-8\u003c/li\u003e\n \u003cli\u003evon Elm E, Altman DG, Egger M, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: Guidelines for reporting observational studies. \u003cem\u003ePLoS Med\u003c/em\u003e. 2007;4(10):e296. doi:10.1371/journal.pmed.0040296\u003c/li\u003e\n \u003cli\u003eBenchimol EI, Smeeth L, Guttmann A, Harron K, Moher D, Petersen I, et al. (2015). The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) Statement. \u003cem\u003ePLoS Medicine\u003c/em\u003e, 12(10), e1001885. https://doi.org/10.1371/journal.pmed.1001885\u003c/li\u003e\n \u003cli\u003eBland JM, Altman DG. Measuring agreement in method comparison studies. Stat Methods Med Res. 1999 Jun;8(2):135-60. doi: 10.1177/096228029900800204. PMID: 10501650.\u003c/li\u003e\n \u003cli\u003eHaghayegh S, Kang HA, Khoshnevis S, Smolensky MH, Diller KR. A comprehensive guideline for Bland-Altman and intra class correlation calculations to properly compare two methods of measurement and interpret findings. Physiol Meas. 2020 Jun 15;41(5):055012. doi: 10.1088/1361-6579/ab86d6. PMID: 32252039.\u003c/li\u003e\n \u003cli\u003eAm\u0026eacute;lie Gabet, Yannick B\u0026eacute;jot, Emmanuel Touz\u0026eacute;, France Woimant, Laurent Suissa, Cl\u0026eacute;mence Grave, Gr\u0026eacute;gory Lailler, Philippe Tuppin, Val\u0026eacute;rie Oli\u0026eacute;. \u0026ldquo;Epidemiology of stroke in France\u0026rdquo;. Archives of Cardiovascular Diseases, Volume 117, Issue 12, 2024,Pages 682-692,ISSN 1875-2136. https://doi.org/10.1016/j.acvd.2024.10.327.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"clinical data warehouse, electronic health records, data reuse, health care quality indicators, stroke","lastPublishedDoi":"10.21203/rs.3.rs-7538212/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7538212/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eAssessment of quality and safety indicators (QSI) remains often based on time-consuming manual Electronic Health Record (EHR) review. Through a pilot study in stroke management, we investigated the feasibility of automating the calculation of one French national QSI, the measure of door-to-imaging time (DTI time) i.e. the time interval between the arrival at hospital and the first stroke diagnostic imaging, in a Clinical Data Warehouse (CDW) through a retrospective observational study of patients hospitalized in the Greater Paris University Hospitals (AP-HP) for an acute stroke in 2022. We automatically computed DTI time for more than 6,000 medical records in the CDW using a systematic approach, and validated this method by matching results against the manual AP-HP EHR review from the 2022 French national QSI audit. On this Study population, CDW and manual EHR review methods agreeed on estimating overall indicators, but showed discrepancies in the case-by-case analysis mainly because of human variability both in EHR completion and manual reviewing. Automation looks promising in a context of limited professional resources but requires structured and validated data, interoperability in the case of inter-institutional stays and validly measurable QSI.\u003c/p\u003e","manuscriptTitle":"Automating the assessment of door-to-imaging time in stroke management using a clinical data warehouse","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-10 14:35:25","doi":"10.21203/rs.3.rs-7538212/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-10-10T07:00:11+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-10-09T20:17:05+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-10-01T11:31:22+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"25275371455130021069337980394149417199","date":"2025-09-28T11:28:38+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"244593793271737392674131119888197906107","date":"2025-09-26T04:51:08+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-09-25T23:48:02+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-09-25T21:58:49+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-09-09T15:47:29+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-09-09T04:06:05+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-09-08T16:33:57+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"0cf68112-6277-4339-b9e5-566edef15568","owner":[],"postedDate":"October 10th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":55797464,"name":"Health sciences/Health care"},{"id":55797465,"name":"Health sciences/Medical research"}],"tags":[],"updatedAt":"2026-03-09T16:15:39+00:00","versionOfRecord":{"articleIdentity":"rs-7538212","link":"https://doi.org/10.1038/s41598-026-41833-4","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2026-03-04 15:59:03","publishedOnDateReadable":"March 4th, 2026"},"versionCreatedAt":"2025-10-10 14:35:25","video":"","vorDoi":"10.1038/s41598-026-41833-4","vorDoiUrl":"https://doi.org/10.1038/s41598-026-41833-4","workflowStages":[]},"version":"v1","identity":"rs-7538212","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7538212","identity":"rs-7538212","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}