Search

Henrik Albaek Jacobsen,1,2 Tine Jess,1,2 Lone Larsen1,2 1Department of Gastroenterology and Hepatology, Aalborg University Hospital, Aalborg, Denmark; 2Center for Molecular Prediction of Inflammatory Bowel Disease, PREDICT, Department of Clinical Medicine, Aalborg University, Copenhagen, DenmarkCorrespondence: Henrik Albaek Jacobsen, National Center of Excellence PREDICT, Department of Gastroenterology and Hepatology, Aalborg University Hospital, Mølleparkvej 4, Aalborg, 9000, Tel +45 97656267, Email henrik.jacobsen@rn.dkPurpose: The Danish National Patient Registry (DNPR) is recognized for providing high-quality data. However, only a few minor studies have validated inflammatory bowel disease (IBD) diagnoses in the DNPR, reporting various degrees of validity. To pave the way for large-scale studies of IBD in Denmark, we aimed to investigate the validity of IBD among > 8000 patients registered in the DNPR between 2002 and 2020 in the North Denmark Region.Patients and Methods: To evaluate the reliability of the diagnoses in the DNPR, we initially compared all patients registered with one IBD diagnosis during 2002– 2020 to a list of already verified patients in the regional IBD database GASTROBIO. Medical records on all DNPR registered patients not on the list were manually reviewed by a gastroenterologist to verify or dismiss the IBD diagnosis. Positive predictive values (PPV) were calculated.Results: Of 8040 patients with at least one IBD diagnosis in DNPR, 5263 were already confirmed cases, leaving 2777 for medical record evaluation, of whom 849 had IBD. In total, 6112 were correctly registered with IBD based on one diagnosis, and 1343 were incorrectly registered, resulting in a PPV of 0.82 (95% CI, 0.81– 0.83). For patients registered with at least two diagnoses, the PPV was 0.95 (95% CI, 0.95– 0.96), and with at least three diagnoses, the PPV was 0.98 (95% CI, 0.98– 0.99). Results were similar for UC and CD separately. Of note, the completeness of valid cases went from 6112 to 4606 (75%; 95% CI, 74%-76%) when demanding at least two registered diagnoses and to 3320 (54%; 95% CI, 53%-56%) when demanding at least three registered diagnoses.Conclusion: Reassuringly, the validity of IBD diagnoses in DNPR is high, especially for patients registered more than once. However, the reduced completeness when applying a true case definition of at least two registered diagnoses should be considered.Keywords: inflammatory bowel disease, Danish National Patient Registry, validation, diagnosis codes, positive predictive value, completeness

Jacob Bodilsen,1 Lykke Larsen,2 Christian Thomas Brandt,3,4 Lothar Wiese,4 Birgitte Rønde Hansen,5 Christian Østergaard Andersen,6 Hans Rudolf Lüttichau,7 Jannik Helweg-Larsen,8 Merete Storgaard,9 Henrik Nielsen1,10 1Department of Infectious Diseases, Aalborg University Hospital, Aalborg, 9000, Denmark; 2Department of Infectious Diseases, Odense University Hospital, Odense, 5000, Denmark; 3Department of Infectious Diseases, Nordsjællands Hospital, Hillerød, 3400, Denmark; 4Department of Infectious Diseases, Sjælland University Hospital, Roskilde, 4000, Denmark; 5Department of Infectious Diseases, Hvidovre University Hospital, Copenhagen, 2650, Denmark; 6Department of Clinical Microbiology, Hvidovre University Hospital, Copenhagen, 2650, Denmark; 7Department of Infectious Diseases, Herlev Gentofte Hospital, Copenhagen, 2730, Denmark; 8Department of Infectious Diseases, Rigshospitalet, Copenhagen, 2100, Denmark; 9Department of Infectious Diseases, Aarhus University Hospital, Aarhus, 8200, Denmark; 10Department of Clinical Medicine, Aalborg University, Aalborg, 9000, DenmarkCorrespondence: Jacob BodilsenDepartment of Infectious Diseases, Aalborg University Hospital, Mølleparkvej 4, Aalborg, DK9000, DenmarkTel +45 97660566Email Jacob.bodilsen@rn.dkAbstract: Central nervous system (CNS) infections are rare diseases that are associated with considerable morbidity and mortality. Increased knowledge based on contemporary data is a prerequisite for improved management and prevention of these serious conditions. Yet, population-based databases of patients hospitalized with CNS infections remain scarce. The Danish Study Group of Infections of the Brain (DASGIB) has prospectively registered information on all adults ≥ 18 years of age admitted with CNS infections at departments of infectious diseases in Denmark since 2015. The main variables collected are baseline demographics, blood and cerebrospinal fluid tests, imaging results, and outcome using the Glasgow Outcome Scale score. To investigate important aspects for each type of CNS infections, additional variables are included specifically for bacterial meningitis, viral meningitis, encephalitis, brain abscess, neurosyphilis, and Lyme neuroborreliosis. From 2015 to 2020, a total of 3579 cases of CNS infections have been recorded in the DASGIB database. Using the unique civil registration number assigned to all Danish residents, the database can be unambiguously linked with nationwide healthcare registries at the individual level. This enables researchers to conduct detailed population-based and longitudinal observational studies of risk and prognosis of CNS infections and to compare them with matched population cohorts. The database is well-suited for epidemiological research and the DASGIB network forms a solid infrastructure for future national and international collaborations.Keywords: CNS infections, epidemiology, database, cohort studies, bacterial meningitis, viral meningitis, encephalitis, brain abscess, cerebral abscess, neurosyphilis, syphilis, Lyme neuroborreliosis, borreliosis

Lone Larsen,1 Michael Dam Jensen,2 Michael Due Larsen,3 Rasmus Gaardskær Nielsen,4 Niels Thorsgaard,5 Ida Vind,6 Signe Wildt,7 Jens Kjeldsen8 1Department of Gastroenterology and Hepatology, Aalborg University Hospital, Aalborg,2Department of Internal Medicine, Section of Gastroenterology, Lillebaelt Hospital Vejle, Vejle, 3Department of Clinical Research, Faculty of Health Sciences, University of Southern Denmark, 4Department of Paediatrics, Hans Christian Andersen Children's Hospital, Odense University Hospital, Odense C, 5Department of Internal Medicine, Section of Gastroenterology, Hospital Unit West, Herning, 6Department of Gastroenterology, Copenhagen University Hospital, Hvidovre, 7Department of Medicine, Section of Gastroenterology, Køge Hospital, Køge, 8Department of Medical Gastroenterology, Odense University Hospital, Odense C, Denmark Aim: The aims of The Danish National Registry for Biological Therapy in Inflammatory Bowel Disease are to ensure that biological therapy and the clinical management of patients with inflammatory bowel disease (IBD) receiving biological treatment are in accordance with the national clinical guidelines and, second, the database allows register-based clinical epidemiological research. Study population: The study population comprises all Danish patients with IBD (both children and adults) with ulcerative colitis, Crohn's disease, and IBD unclassified who receive biological therapy. Patients will be enrolled consecutively when biological treatment is initiated. Main variables: The variables in the database are: diagnosis, time of diagnosis, disease manifestation, indication for biological therapy, previous biological and nonbiological therapy, date of visit, clinical indices, physician's global assessment, pregnancy and breastfeeding (women), height (children), weight, dosage (current biological agent), adverse events, surgery, endoscopic procedures, and radiology. Descriptive data: Eleven clinical indicators have been selected to monitor the quality of biological treatment. For each indicator, a standard has been defined based on the available evidence. National results will be published in an annual report and local results on a quarterly basis. The indicators will be reported as department-specific proportions with 95% confidence intervals, and the national average will be provided for comparison. An estimated 1,200–1,300 new biological therapies are initiated each year in Danish patients with IBD. Conclusion: The database will be available for research during 2016. Data will be made available by The Danish Clinical Registries (www.rkkp.dk). Keywords: inflammatory bowel disease, biological therapy, anti-TNF-α agents, quality indicators, database

Water availability has historically been one of the most significant threats to African regional social and economic well-being. Over the Sahel region, a megadrought during the 1960s and 1970s induced by an abrupt and substantial rainfall reduction caused widespread famine and death. The postdrought recovery, which is still ongoing, has been characterized by gradual increases in rainfall, but with dramatic fluctuations. The large negative human impacts, slow recovery, and variability raise important questions of why and how rainfall dynamics evolve and interact with other components of the regional climate system. Here we provide an observational assessment of rainfall interactions mechanisms (informed by directional transfer of information entropy) that regulate Sahel rainfall. We quantitatively demonstrate that (1) sea surface temperature over the Gulf of Guinea controls moisture advection and transport to the West Sahel region and (2) strong bidirectional interactions exist between local vegetation dynamics and rainfall patterns. Then we assess the directional interaction patterns from nine state-of-the-art Earth System Models (ESMs). We find that most ESMs are able to represent either the unidirectional control of sea surface temperature on precipitation or the bidirectional interaction between vegetation and precipitation. However, none of the ESMs represents both interactive patterns. The GFDL and IPSL-CM5A-LR models successfully reproduced observed patterns over ~50% of the West Sahel region but were not accurate in reproducing observed precipitation regional trends or interannual variation. We propose that the directional information transfer is a powerful mechanistic benchmark to assess model fidelity at the process level.

We developed a suite of models using deep learning to make hindcast predictions of the 7‐day average backward‐looking nitrate concentration at 46 predominantly agricultural sites across the midwestern and eastern United States. The models used daily observations of discharge and meteorological variables and watershed attributes describing anthropogenic modification to hydrology, nitrogen application, climate, groundwater, land use, watershed physiographic attributes, and soils. Across all sites, discharge and watershed soil and physiographic attributes showed a strong influence on model performance. Analysis of drivers across sites revealed considerable regional differences related to controlling processes such as groundwater contributions. We tested several ways to pool data across sites to develop accurate models and make the most effective use of available data. Single‐site models, in which models are trained and tested at a single location, showed generally strong predictive performance (median Kling‐Gupta Efficiency = 0.66), and accuracy at poorly performing sites could be improved by grouping sites with similar characteristics. Developing a single model for all sites reduced performance at several locations with distinct characteristics, suggesting that there is a threshold of dissimilarity beyond which more data does not improve the model. While many deep learning studies have shown that national or even global models can outperform local models, it is not clear that this is true for water quality constituents. This study demonstrates how data can be combined effectively, using deep learning to develop accurate and interpretable models of instream nitrate at sites where varying processes are responsible for changes in nitrate concentration. Plain Language Summary: Developing accurate models of water contaminants is a critical tool for understanding and predicting their concentrations in our waterways. Models rely on observational data collected at specific locations, but it is unclear if and in what manner data can be pooled across locations to develop more accurate predictions and deeper understanding of the underlying processes driving water quality changes. We developed a series of models for 46 sites across the midwestern and eastern United States using machine learning to investigate how data can be grouped together to create accurate models and better understand the processes that drive water quality. When we grouped data from all sites, we demonstrate that discharge, basin size, stream length, and soil properties are critical to predicting nitrate concentrations. In some areas, other features, such as groundwater nitrate concentration and land use are very important. We show that poorly performing models can be improved by informing the models with data from similar sites. The results show that similar sites can be grouped to create accurate models, but adding more data from dissimilar sites does not produce more accurate models. Key Points: Machine learning models can make accurate predictions of 7‐day average nitrate concentration, particularly in wetter conditionsPoorly performing single‐site models can be improved with data from similar sites, but more data does not always increase accuracyA model using data from 46 sites in 9 U.S. states, reveals that watershed soil and physical properties are key drivers of nitrate [ABSTRACT FROM AUTHOR]

Accurate streamflow predictions are essential for water resources management. Recent studies have examined the use of hybrid models that integrate machine learning models with process‐based (PB) hydrologic models to improve streamflow predictions. Yet, there are many open questions regarding optimal hybrid model construction, especially in Mediterranean‐climate watersheds that experience pronounced wet and dry seasons. In this study, we performed model benchmarking to (a) compare hybrid model performance to PB and machine learning models and (b) examine the sensitivity of hybrid model performance to PB model parameter calibration, structural complexity, and variable selection. Hybrid models were generated by post‐processing process‐based models using Long Short‐Term Memory neural networks. Models were benchmarked within two northern California watersheds that are managed for both municipal water supplies and aquatic habitat. Though model performance varied substantially by watershed and error metric, calibrated hybrid models frequently outperformed both the machine learning model (for 72% of watershed‐model‐metric combinations) and the calibrated process‐based models (for 79% of combinations). Furthermore, hybrid models were relatively insensitive to PB model calibration and structural complexity, but sensitive to PB model variable selection. Our results demonstrate that hybrid models can improve streamflow prediction in Mediterranean‐climate watersheds. Additionally, hybrid model insensitivity to PB model parameter calibration and structural complexity suggests that uncalibrated or less complex PB models could be used in hybrid models without any loss of streamflow prediction accuracy, improving model construction efficiency. Moreover, hybrid model sensitivity to the selection of PB model variables suggests a strategy for diagnosing poorly performing PB model components. Key Points: Hybrid streamflow prediction models frequently outperformed both machine learning and process‐based (PB) modelsHybrid models were relatively insensitive to PB model calibration and structural complexity, but sensitive to PB model variable selectionHybrid models can improve streamflow prediction accuracy, efficiency, and diagnostics in Mediterranean‐climate watersheds [ABSTRACT FROM AUTHOR]

Background: A nasal spray containing the antiallergy agent olopatadine hydrochloride is being developed for the treatment of seasonal allergic rhinitis (SAR). Objective: To evaluate the safety and efficacy of 2 concentrations of olopatadine nasal spray vs placebo in patients with SAR. Methods: This was a multicenter, randomized, double-blind, placebo-controlled study. After a 3- to 21-day placebo run-in, 565 patients aged 12 to 80 years were randomized to receive 0.4% or 0.6% olopatadine or placebo, 2 sprays per nostril twice daily for 2 weeks. Patients evaluated morning and evening reflective and instantaneous nasal symptoms (sneezing, stuffy nose, runny nose, and itchy nose, which compose the total nasal symptom score [TNSS]) and ocular symptoms and completed the Rhinoconjunctivitis Quality of Life Questionnaire (RQLQ). Results: Olopatadine spray (0.4% and 0.6%) was significantly superior to placebo for percentage change from baseline in overall reflective (P = .004 and P < .001, respectively) and instantaneous (P = .02 and P = .003, respectively) TNSSs. Also, 0.6% olopatadine was significantly superior to placebo for reducing the reflective and instantaneous assessments of sneezing runny and itchy nose, and itchy eyes; the instantaneous assessments of watery eyes; and the overall and all 7 domain scores of the RQLQ (P < .05). Olopatadine spray exhibited a safety profile comparable with that of placebo. Conclusions: Olopatadine nasal spray (0.4% and 0.6%) provided statistically significant improvements in allergic rhinitis symptoms compared with placebo regarding TNSSs (reflective and instantaneous) and in quality-of-life variables in patients with SAR. Olopatadine nasal spray administered twice daily was safe and well tolerated in adolescents and adults.

Background/Objectives: Cardiovascular disease (CVD) remains a major cause of death worldwide. Whereas dairy generally is associated with a neutral or a beneficial CVD effect, the consumption of ultra-high temperature (UHT)-treated milk has been reported to increase levels of low-density lipoprotein cholesterol (LDL-C) in an uncontrolled study. Our aim was to examine whether semi-skimmed UHT dairy milk increases the risk of CVD development compared with pasteurized (PAST) dairy milk in overweight healthy adults. Subjects/Methods: Nineteen healthy men and women participated in a randomized, controlled, crossover study. The effect of intake of 1.5 l of UHT dairy milk or PAST milk, similar in nutritional content, was examined as a supplement to the participant's habitual diet for 21 days in each intervention period. Intake of other dairy products was not allowed during the intervention period. Clinical evaluation and blood samples took place preintervention and postintervention. Results: There was no significant effect by type of milk on LDL-C (P=0.29). No effects of type of milk were observed in other blood lipid levels, such as total cholesterol, high-density lipoprotein cholesterol or triglycerides. No effects of type of milk were found for blood pressure, insulin, glucose concentration and insulin resistance (homeostasis model assessment of insulin resistance) or body weight. Conclusions: This study does not support the hypothesis that UHT processing of milk increases the risk of CVD., Author(s): C K Hansen [sup.1] , L Klingenberg [sup.1] , L B Larsen [sup.2] , J K Lorenzen [sup.1] , K V Sørensen [sup.1] , A Astrup [sup.1] Author Affiliations: [...]