Background Electronic health records (EHRs) offer a wealth of observational data. Machine-learning (ML) methods are efficient at data extraction, capable of processing the information-rich free... Show moreBackground Electronic health records (EHRs) offer a wealth of observational data. Machine-learning (ML) methods are efficient at data extraction, capable of processing the information-rich free-text physician notes in EHRs. The clinical diagnosis contained therein represents physician expert opinion and is more consistently recorded than classification criteria components. Objectives To investigate the overlap and differences between rheumatoid arthritis patients as identified either from EHR free-text through the extraction of the rheumatologist diagnosis using machine-learning (ML) or through manual chart-review applying the 1987 and 2010 RA classification criteria. Methods Since EHR initiation, 17,662 patients have visited the Leiden rheumatology outpatient clinic. For ML, we used a support vector machine (SVM) model to identify those who were diagnosed with RA by their rheumatologist. We trained and validated the model on a random selection of 2000 patients, balancing PPV and sensitivity to define a cutoff, and assessed performance on a separate 1000 patients. We then deployed the model on our entire patient selection (including the 3000). Of those, 1127 patients had both a 1987 and 2010 EULAR/ACR criteria status at 1 year after inclusion into the local prospective arthritis cohort. In these 1127 patients, we compared the patient characteristics of RA cases identified with ML and those fulfilling the classification criteria. Results The ML model performed very well in the independent test set (sensitivity=0.85, specificity=0.99, PPV=0.86, NPV=0.99). In our selection of patients with both EHR and classification information, 373 were recognized as RA by ML and 357 and 426 fulfilled the 1987 or 2010 criteria, respectively. Eighty percent of the ML-identified cases fulfilled at least one of the criteria sets. Both demographic and clinical parameters did not differ between the ML extracted cases and those identified with EULAR/ACR classification criteria. Conclusions With ML methods, we enable fast patient extraction from the huge EHR resource. Our ML algorithm accurately identifies patients diagnosed with RA by their rheumatologist. This resulting group of RA patients had a strong overlap with patients identified using the 1987 or 2010 classification criteria and the baseline (disease) characteristics were comparable. ML-assisted case labeling enables high-throughput creation of inclusive patient selections for research purposes. Show less
Verstappen, M.; Niemantsverdriet, E.; Matthijssen, X.M.E.; Cessie, S. le; Helm-van Mil, A.H.M. van der 2020
BackgroundSustained DMARD-free remission (SDFR) is increasingly achievable. The pathogenesis underlying SDFR development is unknown and patient characteristics at diagnosis poorly explain whether... Show moreBackgroundSustained DMARD-free remission (SDFR) is increasingly achievable. The pathogenesis underlying SDFR development is unknown and patient characteristics at diagnosis poorly explain whether SDFR will be achieved. To increase the understanding, we studied the course of disease activity scores (DAS) over time in relation to SDFR development. Subsequently, we explored whether DAS course could be helpful identifying RA patients likely to achieve SDFR.Methods772 consecutive RA patients, promptly treated with csDMARDs (mostly methotrexate and treat-to-target treatment adjustments), were studied for SDFR development (absence of synovitis, persisting minimally 12 months after DMARD stop). The course of disease activity scores (DAS) was compared between RA patients with and without SDFR development within 7 years, using linear mixed models, stratified for ACPA. The relation between 4-month DAS and the probability of SDFR development was studied with logistic regression. Cumulative incidence of SDFR within DAS categories (<1.6, 1.6-2.4, 2.4-3.6, 3.6) at 4 months was visualized using Kaplan-Meier curves.ResultsIn ACPA-negative RA patients, those achieving SDFR showed a remarkably stronger DAS decline within the first 4 months, compared to RA patients without SDFR; -1.73units (95%CI, 1.28-2.18) versus -1.07units (95%CI, 0.90-1.23) (p <0.001). In APCA-positive RA patients, such an effect was not observed, yet SDFR prevalence in this group was low. In ACPA-negative RA, DAS decline in the first 4 months and absolute DAS levels at 4 months (DAS(4 months)) were equally predictive for SDFR development. Incidence of SDFR in ACPA-negative RA patients was high (70.2%) when DAS(4 months) was <1.6, whilst SDFR was rare (7.1%) when DAS(4 months) was >= 3.6.ConclusionsIn ACPA-negative RA, an early response to treatment, i.e., a strong DAS decline within the first 4 months, is associated with a higher probability of SDFR development. DAS values at 4 months could be useful for later decisions to stop DMARDs. Show less
Niemantsverdriet, E.; Dakkak, Y.J.; Burgers, L.E.; Bonte-Mineur, F.; Steup-Beekman, G.M.; Kooij, S.M. van der; ... ; Helm-van Mil, A.H.M. van der 2020
Background: We present a study protocol for a randomized, double-blind, placebo-controlled trial that investigates the hypothesis if intervention in the symptomatic phase preceding clinical... Show moreBackground: We present a study protocol for a randomized, double-blind, placebo-controlled trial that investigates the hypothesis if intervention in the symptomatic phase preceding clinical arthritis (clinically suspect arthralgia (CSA)) is effective in preventing progression from subclinical inflammation to clinically apparent persistent arthritis. Currently, rheumatoid arthritis (RA) can be recognized and diagnosed when arthritis (joint swelling) has become detectable at physical examination. Importantly, at this time, the immune processes have already matured, chronicity is established, and patients require long-standing treatment with disease-modifying anti-rheumatic drugs. The TREAT EARLIER trial studies the hypothesis that intervention in the symptomatic phase preceding clinical arthritis is more often successful in permanent disease modification because of less matured underlying disease processes.Methods: A two-level definition to identify patients that are prone to develop RA is used. First, patients should have CSA and recent-onset arthralgia (< 1 year) that is suspect to progress to RA according to the expertise of the treating rheumatologist. Second, patients need to have subclinical inflammation of the hand or foot joints at 1.5 T MRI. The trial aims to recruit 230 participants from secondary care hospital settings across the south-west region of The Netherlands. Intervention will be randomly assigned and includes a single-dose of intramuscular 120 mg methylprednisolon followed by methotrexate (increasing dose to 25 mg/week orally) or placebo (both; injection and tablets) over the course of 1 year. Thereafter, participants are followed for another year. The primary endpoint is the development of clinically detectable arthritis, either fulfilling the 2010 criteria for RA or unclassified clinical arthritis of >= 2 joints, which persists for at least 2 weeks. DMARD-free status is a co-primary endpoint. The patient-reported outcomes functioning, along with workability and symptoms, are key secondary endpoints. Participants, caregivers (including those assessing the endpoints), and scientific staff are all blinded to the group assignment.Discussion: This proof-of-concept study is the logical consequence of pre-work on the identification of patients with CSA with MRI-detected subclinical joint inflammation. It will test the hypothesis whether intervention in patients in this early phase with the cornerstone treatment of classified RA (methotrexate) hampers the development of persistent RA and reduce the disease burden of RA. Show less
Matthijssen, X.M.E.; Wouters, F.; Boeters, D.M.; Boer, A.C.; Dakkak, Y.J.; Niemantsverdriet, E.; Helm-van Mil, A.H.M. van der 2019