Suzanne Desirée van der Werff

BACKGROUND: Conventional manual surveillance of healthcare-associated infections is labour-intensive and therefore often restricted to areas with high-risk patients. Fully automated surveillance of hospital-onset bacteraemia and fungaemia (HOB) may facilitate hospital-wide surveillance. AIM: To develop an algorithm and minimal dataset (MDS) required for automated surveillance of HOB and apply it to real-life routine data in four European hospitals METHODS: Through consensus discussion, a HOB definition with MDS suitable for automated surveillance was developed and applied in a retrospective multicentre observational study including all adult patients admitted to hospitals in the Netherlands, Germany, Sweden and Switzerland (2018–22). Annual HOB rates were calculated per 1,000 patient days for hospital, intensive care unit (ICU) and non-ICU settings. RESULTS: HOB was defined as a positive blood culture with a recognised pathogen 2 or more days after hospital admission. For common commensals, two blood cultures with the same commensal within 2 days were required. HOB rates were comparable between the four hospitals (1.0–2.2/1,000 patient days). HOB rates were substantially higher in ICU than non-ICU settings, and HOB with common commensals accounted for 14.8–28.2% of all HOB. HOB rates per 1,000 patient days were consistent over time, but higher in 2020–21. HOB caused by Staphylococcus aureus comprised 8.4–16.0% of all HOB. CONCLUSION: Automated HOB surveillance using a common definition was feasible and reproducible across four European hospitals. Future studies should investigate clinical relevance and preventability of HOB, and focus on strategies to make the automated HOB metric an actionable infection control tool.

Healthcare‐associated infections (HAIs) are common adverse events, and surveillance is considered a core component of effective HAI reduction programmes. Recently, efforts have focused on automating the traditional manual surveillance process by utilizing data from electronic health record (EHR) systems. Using EHR data for automated surveillance, algorithms have been developed to identify patients with (ventilator‐associated) pneumonia and (catheter‐related) bloodstream, surgical site, (catheter‐associated) urinary tract and Clostridioides difficile infections (sensitivity 54.2%–100%, specificity 63.5%–100%). Mostly methods based on natural language processing have been applied to extract information from unstructured clinical information. Further developments in artificial intelligence (AI), such as large language models, are expected to support and improve different aspects within the surveillance process; for example, more precise identification of patients with HAI. However, AI‐based methods have been applied less frequently in automated surveillance and more frequently for (early) prediction, particularly for sepsis. Despite heterogeneity in settings, populations, definitions and model designs, AI‐based models have shown promising results, with moderate to very good performance (accuracy 61%–99%) and predicted sepsis within 0–40 h before onset. AI‐based prediction models detecting patients at risk of developing different HAIs should be explored further. The continuous evolution of AI and automation will transform HAI surveillance and prediction, offering more objective and timely infection rates and predictions. The implementation of (AI‐supported) automated surveillance and prediction systems for HAI in daily practice remains scarce. Successful development and implementation of these systems demand meeting requirements related to technical capabilities, governance, practical and regulatory considerations and quality monitoring.

We conducted a population-based cohort study in Stockholm, Sweden, to investigate the effect of COVID-19 vaccination on the risk of developing persistent post–COVID-19 condition (PCC) in individuals surviving the first year after a SARS-CoV-2 infection. In total, 331 042 individuals were included, of which 852 had persistent PCC. The adjusted risk ratio for developing persistent PCC compared with unvaccinated individuals was 0.81 (95% confidence interval [CI], .59–1.10) for 1 dose, 0.42 (95% CI, .35–.52) for 2 doses, and 0.37 (95% CI, .27–.52) for 3 doses. Reduced risks for vaccinated individuals were also observed when restricting the analyses to pre-Omicron and Omicron, as well as all subgroups including sex, age, and previous infection.

Objective: To assess the performance of an algorithm for automated grading of surgery-related adverse events (AEs) according to Clavien-Dindo (C-D) classification. Background: Surgery-related AEs are common, lead to increased patient morbidity, and raise health care costs. Resource-intensive manual chart review is still standard, and, to our knowledge, algorithms using electronic health record (EHR) data to grade AEs according to C-D classification have not been explored. Methods: The algorithm was developed in a research database containing all EHR data of Karolinska University Hospital Stockholm and returns a C-D grade for each AE within 30 days. This raw score was used to grade the postoperative recovery of 1379 elective colorectal procedures according to C-D classification and Comprehensive Complication Index. Agreement with manual annotation of colorectal surgeon (gold standard) and research nurse (current practice) was assessed in a random sample of 399 procedures. Results: For the C-D classification, kappa was 0.77 (95% CI: 0.71 to 0.84) for algorithm versus surgeon and 0.74 (95% CI: 0.67 to 0.82) for algorithm versus nurse. The kappa value increased to 0.89 (95% CI: 0.84 to 0.95) after the correction of misclassified annotations by the surgeon. The intraclass correlation for Comprehensive Complication Index between algorithm and surgeon was 0.89 (95% CI: 0.87 to 0.91) after correction and 0.76 (95% CI: 0.71 to 0.80) for algorithm versus nurse. Conclusions: The performance of the algorithm motivates in our opinion implementation to real-time data under continuous scientific evaluation of the impact on AEs in different types of surgery. In the future, local EHR data could be used to enhance risk prediction with machine learning techniques.

Automation of surveillance of infectious diseases—where algorithms are applied to routine care data to replace manual decisions—likely reduces workload and improves quality of surveillance. However, various barriers limit large-scale implementation of automated surveillance (AS). Current implementation strategies for AS in surveillance networks include central implementation (i.e. collecting all data centrally, and central algorithm application for case ascertainment) or local implementation (i.e. local algorithm application and sharing surveillance results with the network coordinating center). In this perspective, we explore whether current challenges can be solved by federated AS. In federated AS, scripts for analyses are developed centrally and applied locally. We focus on the potential of federated AS in the context of healthcare associated infections (AS-HAI) and of severe acute respiratory illness (AS-SARI). AS-HAI and AS-SARI have common and specific requirements, but both would benefit from decreased local surveillance burden, alignment of AS and increased central and local oversight, and improved access to data while preserving privacy. Federated AS combines some benefits of a centrally implemented system, such as standardization and alignment of an easily scalable methodology, with some of the benefits of a locally implemented system including (near) real-time access to data and flexibility in algorithms, meeting different information needs and improving sustainability, and allowance of a broader range of clinically relevant case-definitions. From a global perspective, it can promote the development of automated surveillance where it is not currently possible and foster international collaboration.The necessary transformation of source data likely will place a significant burden on healthcare facilities. However, this may be outweighed by the potential benefits: improved comparability of surveillance results, flexibility and reuse of data for multiple purposes. Governance and stakeholder agreement to address accuracy, accountability, transparency, digital literacy, and data protection, warrants clear attention to create acceptance of the methodology. In conclusion, federated automated surveillance seems a potential solution for current barriers of large-scale implementation of AS-HAI and AS-SARI. Prerequisites for successful implementation include validation of results and evaluation requirements of network participants to govern understanding and acceptance of the methodology.

Background: Conventional manual surveillance of healthcare-associated infections is labour-intensive and therefore often restricted to areas with high-risk patients. Fully automated surveillance of hospital-onset bacteraemia and fungaemia (HOB) may facilitate hospital-wide surveillance. Aim: To develop an algorithm and minimal dataset (MDS) required for automated surveillance of HOB and apply it to real-life routine data in four European hospitals. Methods: Through consensus discussion a HOB definition with MDS suitable for automated surveillance was developed and applied in a retrospective multicentre observational study including all admitted adult patients (2018-2022). HOB was defined as a positive blood culture with a recognised pathogen two or more days after hospital admission. For common commensals, two blood cultures with the same commensal within two days were required. Annual HOB rates were calculated per 1,000 patient days for the hospital and for intensive care units (ICU) and non-ICU. Results: HOB rates were comparable between the four hospitals (1.0 to 2.2 per 1,000 patient days). HOB rates were substantially higher in ICU than non-ICU across the four hospitals, and HOB with common commensals accounted for 14.8-28.2% of all HOB. HOB rates per 1,000 patient days were rather consistent over time, but were higher in 2020 and 2021. HOB caused by Staphylococcus aureus accounted for 8.4-16.0% of all HOB. Conclusion: Automated HOB surveillance using a common definition was feasible and reproducible across four European hospitals. Future studies should investigate clinical relevance and preventability of HOB, and focus on strategies to make the automated HOB metric an actionable infection control tool.

Background Fever is a common symptom among travellers returning from tropical/subtropical areas to Europe, and promptly distinguishing severe illnesses from self-limiting febrile syndromes is important but can be challenging due to non-specific clinical presentation. Methods A cross-sectional study enrolled adults and children who sought care during 2015–2020 at Karolinska University Hospital, Stockholm, Sweden with fever within 2 months after returning from travel to a tropical/subtropical area. Data on symptoms and laboratory parameters were prospectively and retrospectively collected. Two separate scoring systems for malaria and dengue were developed based on backward elimination regressions. Results In total, 2113 adults (18–94 years) and 202 children (1–17 years) were included, with 112 (4.8%) confirmed malaria by blood thick smear and 90 (3.9%) PCR/serology dengue-positive cases. Malaria was more likely in a patient who had visited sub-Saharan Africa and presented with combination of thrombocytopenia, anaemia and fever ≥39.5°C. Leucopenia, muscle pain and rash after travelling to Asia or South/Latin America indicated high probability of dengue. Two scoring systems with points between 0 and 7 for prediction of malaria or dengue were created based on the above predictors. Scores ≥3 indicated >80% sensitivity and specificity for malaria and >90% specificity for dengue in children and adults (area under the curve (AUC) for dengue: 0.92 in adults (95% CI 0.90 to 0.95) and 0.95 in children (95% CI 0.88 to 1.0); AUC for malaria: 0.93 in adults (95% CI 0.91 to 0.96) and 0.88 in children (95% CI 0.78 to 0.99)). Internal validation of optimism and overfitting was managed with bootstrap. Conclusion The presented scoring systems provide novel tools for structured assessment of patients with tropical fever in a non-endemic area and highlight clinical signs associated with a potential severe aetiology to direct the need for microbial investigation.

Background: Continuous surveillance for healthcare-associated infections such as central venous catheter-related bloodstream infections (CVC-BSI) is crucial for prevention. However, traditional surveillance methods are resource-intensive and prone to bias. This study aimed to develop and validate fully-automated surveillance algorithms for CVC-BSI. Methods: Two algorithms were developed using electronic health record data from 1000 admissions with a positive blood culture (BCx) at Karolinska University Hospital from 2017: (1) Combining microbiological findings in BCx and CVC cultures with BSI symptoms; (2) Only using microbiological findings. These algorithms were validated in 5170 potential CVC-BSI-episodes from all admissions in 2018–2019, and results extrapolated to all potential CVC-BSI-episodes within this period (n = 181,354). The reference standard was manual record review according to ECDC’s definition of microbiologically confirmed CVC-BSI (CRI3-CVC). Results: In the potential CVC-BSI-episodes, 51 fulfilled ECDC’s definition and the algorithms identified 47 and 49 episodes as CVC-BSI, respectively. Both algorithms performed well in assessing CVC-BSI. Overall, algorithm 2 performed slightly better with in the total period a sensitivity of 0.880 (95%-CI 0.783–0.959), specificity of 1.000 (95%-CI 0.999–1.000), PPV of 0.918 (95%-CI 0.833–0.981) and NPV of 1.000 (95%-CI 0.999–1.000). Incidence according to the reference and algorithm 2 was 0.33 and 0.31 per 1000 in-patient hospital-days, respectively. Conclusions: Both fully-automated surveillance algorithms for CVC-BSI performed well and could effectively replace manual surveillance. The simpler algorithm, using only microbiology data, is suitable when BCx testing adheres to recommendations, otherwise the algorithm using symptom data might be required. Further validation in other settings is necessary to assess the algorithms’ generalisability.

Background: In patients who underwent colorectal surgery, an existing semi-automated surveillance algorithm based on structured data achieves high sensitivity in detecting deep surgical site infections (SSI), however, generates a significant number of false positives. The inclusion of unstructured, clinical narratives to the algorithm may decrease the number of patients requiring manual chart review. The aim of this study was to investigate the performance of this semi-automated surveillance algorithm augmented with a natural language processing (NLP) component to improve positive predictive value (PPV) and thus workload reduction (WR). Methods: Retrospective, observational cohort study in patients who underwent colorectal surgery from January 1, 2015, through September 30, 2020. NLP was used to detect keyword counts in clinical notes. Several NLP-algorithms were developed with different count input types and classifiers, and added as component to the original semi-automated algorithm. Traditional manual surveillance was compared with the NLP-augmented surveillance algorithms and sensitivity, specificity, PPV and WR were calculated. Results: From the NLP-augmented models, the decision tree models with discretized counts or binary counts had the best performance (sensitivity 95.1% (95%CI 83.5–99.4%), WR 60.9%) and improved PPV and WR by only 2.6% and 3.6%, respectively, compared to the original algorithm. Conclusions: The addition of an NLP component to the existing algorithm had modest effect on WR (decrease of 1.4–12.5%), at the cost of sensitivity. For future implementation it will be a trade-off between optimal case-finding techniques versus practical considerations such as acceptability and availability of resources.

Background: Automated surveillance methods that re-use electronic health record data are considered an attractive alternative to traditional manual surveillance. However, surveillance algorithms need to be thoroughly validated before being implemented in a clinical setting. With semi-automated surveillance patients are classified as low or high probability of having developed infection, and only high probability patients subsequently undergo manual record review. The aim of this study was to externally validate two existing semi-automated surveillance algorithms for deep SSI after colorectal surgery, developed on Spanish and Dutch data, in a Swedish setting. Methods: The algorithms were validated in 225 randomly selected surgeries from Karolinska University Hospital from the period January 1, 2015 until August 31, 2020. Both algorithms were based on (re)admission and discharge data, mortality, reoperations, radiology orders, and antibiotic prescriptions, while one additionally used microbiology cultures. SSI was based on ECDC definitions. Sensitivity, specificity, positive predictive value, negative predictive value, and workload reduction were assessed compared to manual surveillance. Results: Both algorithms performed well, yet the algorithm not relying on microbiological culture data had highest sensitivity (97.6, 95%CI: 87.4–99.6), which was comparable to previously published results. The latter algorithm aligned best with clinical practice and would lead to 57% records less to review. Conclusions: The results highlight the importance of thorough validation before implementation in other clinical settings than in which algorithms were originally developed: the algorithm excluding microbiology cultures had highest sensitivity in this new setting and has the potential to support large-scale semi-automated surveillance of SSI after colorectal surgery.