Introduction Diabetic foot ulcers are feared complications of diabetes mellitus (DM), requiring extensive treatment and hospital admissions, ultimately leading to amputation and increased mortality... Show moreIntroduction Diabetic foot ulcers are feared complications of diabetes mellitus (DM), requiring extensive treatment and hospital admissions, ultimately leading to amputation and increased mortality. Different factors contribute to the development of foot ulcers and related complications. Onychomycosis, being more prevalent in patients with diabetes, could be an important risk factor for developing ulcers and related infections. However, the association between onychomycosis and diabetic complications has not been well studied in primary care.Research design and methods To determine the impact of onychomycosis on ulcer development and related complications in patients with diabetes in primary care, a longitudinal cohort study was carried out using routine care data from the Extramural Leiden University Medical Center Academic Network. Survival analyses were performed through Cox proportional hazards models with time-dependent covariates.Results Data from 48 212 patients with a mean age of 58 at diagnosis of DM, predominantly type 2 (87.8%), were analysed over a median follow-up of 10.3 years. 5.7% of patients developed an ulcer. Onychomycosis significantly increased the risk of ulcer development (HR 1.37, 95% CI 1.13 to 1.66), not affected by antimycotic treatment, nor after adjusting for confounders (HR 1.23, 95% CI 1.01 to 1.49). The same was found for surgical interventions (HR 1.54, 95% CI 1.35 to 1.75) and skin infections (HR 1.48, CI 95% 1.28 to 1.72), again not affected by treatment and significant after adjusting for confounders (HR 1.32, 95% CI 1.16 to 1.51 and HR 1.27, 95% CI 1.10 to 1.48, respectively).Conclusions Onychomycosis significantly increased the risk of ulcer development in patients with DM in primary care, independently of other risk factors. In addition, onychomycosis increased the risk of surgeries and infectious complications. These results underscore the importance of giving sufficient attention to onychomycosis in primary care and corresponding guidelines. Early identification of onychomycosis during screening and routine care provides a good opportunity for timely recognition of increased ulcer risk. Show less
Onychomycosis is the most prevalent nail disease and is frequently encountered in clinical practice. Despite having multiple therapeutic options, of which systemic antifungals are the most... Show moreOnychomycosis is the most prevalent nail disease and is frequently encountered in clinical practice. Despite having multiple therapeutic options, of which systemic antifungals are the most effective, treatment is not always mandatory in all patients. Especially when considering systemic treatment, the risk of adverse reactions may outweigh the potential benefits of treatment. In this case report, we present a clinical case of a 49-year-old male patient with a blank past medical history who experienced a severe drug eruption from terbinafine prescribed for mild onychomycosis that required discontinuation of terbinafine, additional evaluation, and treatment of this adverse reaction. Show less
Background Lifestyle intervention programmes target behavioural risk factors that contribute to cardiovascular diseases (CVDs). Unfortunately, sustainable implementation of these programmes can be... Show moreBackground Lifestyle intervention programmes target behavioural risk factors that contribute to cardiovascular diseases (CVDs). Unfortunately, sustainable implementation of these programmes can be challenging. Gaining insights into the barriers and facilitators for successful implementation is important for maximising public health impact of these interventions. The Healthy Heart (HH) programme is an example of a combined lifestyle intervention programme.Aim To analyse the reach, adoption, and implementation of the HH programme.Design & setting A mixed-methods study conducted in a general practice setting in The Netherlands.Method Quantitative data were collected from the Healthy Heart study (HH study), a non-randomised cluster stepped-wedge trial to assess the effect of the HH programme on patients at high risk of developing CVDs at practice level. Qualitative data were obtained through focus groups.Results Out of 73 approached general practices, 55 implemented the HH programme. A total of 1082 patients agreed to participate in the HH study, of whom 64 patients were referred to the HH programme and 41 patients participated. Several barriers for participation were identified such as time investment, lack of risk perception, and being confident in changing lifestyle on their own. Important barriers for healthcare providers (HCPs) to refer a patient were time investment, lack of information to sufficiently inform patients, and preconceived notions regarding which patients the programme was suitable for.Conclusion This study has offered insights from a patient and HCP perspective regarding barriers and facilitators for implementation of the group-based lifestyle intervention programme. The identified barriers and facilitators, and the suggested improvements, can be used by others who wish to implement a similar programme. Show less
AbstractBackgroundMolluscum contagiosum (MC) can cause significant burden in children. So far, pharmacological treatment has not been proven beneficial. More rigorous interventions have not been... Show moreAbstractBackgroundMolluscum contagiosum (MC) can cause significant burden in children. So far, pharmacological treatment has not been proven beneficial. More rigorous interventions have not been well studied. Current guidelines advise a “wait and see” policy. However, children and their parents frequently visit their GP requesting intervention. Therefore, the aim of this study was to gain insight into the approach to MC by GPs and parents’ expectations and to investigate willingness to participate in an interventional study.MethodsA survey study was carried out among GPs and parents using a questionnaire for each group inquiring about MC and potential study participation. Descriptive statistics were used to analyze results and logistical regression to investigate factors influencing participation.ResultsThe majority of GPs (88%) preferred an expectative approach; only 21% were willing to participate in a trial as proposed. GPs estimating ≥ 50% of parents would request treatment, were more likely to participate. Most responding parents did or would visit their GP requesting treatment. In contrast to GPs, 58% were willing to participate. Parents preferring cryotherapy or curettage were more likely to participate.ConclusionOur study demonstrated that the majority of GPs preferred a conservative approach, adhering to current guidelines. However, most parents preferred treatment to resolve MC and symptoms. Parents’ willingness to participate was much higher than GP’s, reflecting parents’ desire for treatment. These findings underscore the need for continued therapeutic research. Careful preparation and selection of GPs and patients will be essential to ensure the feasibility of such an endeavor. Show less
Boeijen, J.A.; Pol, A.C. van de; Uum, R.T. van; Smit, K.; Ahmad, A.; Rijswijk, E. van; ... ; Zwart, D.L.M. 2023
ObjectiveDuring the COVID-19 pandemic new collaborative-care initiatives were developed for treating and monitoring COVID-19 patients with oxygen at home. Aim was to provide a structured overview... Show moreObjectiveDuring the COVID-19 pandemic new collaborative-care initiatives were developed for treating and monitoring COVID-19 patients with oxygen at home. Aim was to provide a structured overview focused on differences and similarities of initiatives of acute home-based management in the Netherlands.MethodsInitiatives were eligible for evaluation if (i) COVID-19 patients received oxygen treatment at home; (ii) patients received structured remote monitoring; (iii) it was not an ‘early hospital discharge’ program; (iv) at least one patient was included. Protocols were screened, and additional information was obtained from involved physicians. Design choices were categorised into: eligible patient group, organization medical care, remote monitoring, nursing care, and devices used.ResultsNine initiatives were screened for eligibility; five were included. Three initiatives included low-risk patients and two were designed specifically for frail patients. Emergency department (ED) visit for an initial diagnostic work-up and evaluation was mandatory in three initiatives before starting home management. Medical responsibility was either assigned to the general practitioner or hospital specialist, most often pulmonologist or internist. Pulse-oximetry was used in all initiatives, with additional monitoring of heart rate and respiratory rate in three initiatives. Remote monitoring staff’s qualification and authority varied, and organization and logistics were covered by persons with various backgrounds. All initiatives offered remote monitoring via an application, two also offered a paper diary option.ConclusionsWe observed differences in the organization of interprofessional collaboration for acute home management of hypoxemic COVID-19 patients. All initiatives used pulse-oximetry and an app for remote monitoring. Our overview may be of help to healthcare providers and organizations to set up and implement similar acute home management initiatives for critical episodes of COVID-19 (or other acute disorders) that would otherwise require hospital care. Show less
Background: Onychomycosis, the most common cause of nail dystrophy, is generally diagnosed by clinical examination. Current guidelines for Dutch general practice advise confirmatory testing only in... Show moreBackground: Onychomycosis, the most common cause of nail dystrophy, is generally diagnosed by clinical examination. Current guidelines for Dutch general practice advise confirmatory testing only in cases of doubt or insufficient response to treatment. However, making a correct diagnosis can be challenging given the wide variety of clinical features and differential diagnosis. Aim: To establish accuracy of clinical diagnosis of onychomycosis by GPs. Design & setting: A diagnostic accuracy study based on GPs' clinical diagnosis of primary care patients suspected of onychomycosis. Method: Using 137 complete datasets from the Onycho Trial, diagnostic accuracy of clinical diagnosis as the index test was compared with confirmatory testing as the reference test. A sensitivity analysis was performed to determine diagnostic values for different combinations of index and reference test. Logistical regression was used to assess which clinical characteristics were associated with the positive predictive value (PPV) of the index test. Results: Clinical accuracy, that is the PPV of the index test, was 74.5%. Sensitivity analysis showed no significant difference in diagnostic values. Male sex and a history of any previous treatment significantly increased clinical accuracy with an odds ratio (OR) of 3.873 (95% confidence interval [CI] = 1.230 to 12.195, P = 0.021) and OR 4.022 (95% CI = 1.075 to 15.040, P = 0.039), respectively. Conclusion: The study demonstrated that the GPs' clinical diagnosis of onychomycosis was insufficiently accurate to initiate treatment without confirmatory testing. Further research is needed to investigate how to increase clinical accuracy and reduce potentially unnecessary exposure to treatment. Show less
Background and ObjectivesFemale-specific factors and psychosocial factors may be important in the prediction of strokebut are not included in prediction models that are currently used. We... Show moreBackground and ObjectivesFemale-specific factors and psychosocial factors may be important in the prediction of strokebut are not included in prediction models that are currently used. We investigated whetheraddition of these factors would improve the performance of prediction models for the risk ofstroke in women younger than 50 years.MethodsWe used data from the Stichting Informatievoorziening voor Zorg en Onderzoek, population-based, primary care database of women aged 20–49 years without a history of cardiovasculardisease. Analyses were stratified by 10-year age intervals at cohort entry. Cox proportionalhazards models to predict stroke risk were developed, including traditional cardiovascularfactors, and compared with models that additionally included female-specific and psychosocialfactors. We compared the risk models using the c-statistic and slope of the calibration curve at afollow-up of 10 years. We developed an age-specific stroke risk prediction tool that may helpcommunicating the risk of stroke in clinical practice.ResultsWe included 409,026 women with a total of 3,990,185 person-years of follow-up. Strokeoccurred in 2,751 women (incidence rate 6.9 [95% CI 6.6–7.2] per 10,000 person-years).Models with only traditional cardiovascular factors performed poorly to moderately in all agegroups: 20–29 years: c-statistic: 0.617 (95% CI 0.592–0.639); 30–39 years: c-statistic: 0.615(95% CI 0.596–0.634); and 40–49 years: c-statistic: 0.585 (95% CI 0.573–0.597). After addingthe female-specific and psychosocial risk factors to the reference models, the model discrimi-nation increased moderately, especially in the age groups 30–39 (Dc-statistic: 0.019) and 40–49years (Dc-statistic: 0.029) compared with the reference models, respectively.DiscussionThe addition of female-specific factors and psychosocial risk factors improves the discrimina-tory performance of prediction models for stroke in women younger than 50 years. Show less
AbstractBackground: Onychomycosis, the most common cause of nail dystrophy, is generally diagnosed by clinical examination. Current guidelines for Dutch general practice advise confirmatory testing... Show moreAbstractBackground: Onychomycosis, the most common cause of nail dystrophy, is generally diagnosed by clinical examination. Current guidelines for Dutch general practice advise confirmatory testing only in cases of doubt or insufficient response to treatment. However, making a correct diagnosis can be challenging given the wide variety of clinical features and differential diagnosis.Aim: To establish accuracy of clinical diagnosis of onychomycosis by GPs.Design & setting: A diagnostic accuracy study based on GPs' clinical diagnosis of primary care patients suspected of onychomycosis.Method: Using 137 complete datasets from the Onycho Trial, diagnostic accuracy of clinical diagnosis as the index test was compared with confirmatory testing as the reference test. A sensitivity analysis was performed to determine diagnostic values for different combinations of index and reference test. Logistical regression was used to assess which clinical characteristics were associated with the positive predictive value (PPV) of the index test.Results: Clinical accuracy, that is the PPV of the index test, was 74.5%. Sensitivity analysis showed no significant difference in diagnostic values. Male sex and a history of any previous treatment significantly increased clinical accuracy with an odds ratio (OR) of 3.873 (95% confidence interval [CI] = 1.230 to 12.195, P = 0.021) and OR 4.022 (95% CI = 1.075 to 15.040, P = 0.039), respectively.Conclusion: The study demonstrated that the GPs' clinical diagnosis of onychomycosis was insufficiently accurate to initiate treatment without confirmatory testing. Further research is needed to investigate how to increase clinical accuracy and reduce potentially unnecessary exposure to treatment. Show less
Patients with severe infection have an increased risk of cardiovascular events. A possible underlying mechanism is inflammation-induced platelet aggregation. We investigated whether... Show morePatients with severe infection have an increased risk of cardiovascular events. A possible underlying mechanism is inflammation-induced platelet aggregation. We investigated whether hyperaggregation occurs during infection, and whether aspirin inhibits this. In this multicentre, open-label, randomised controlled trial, patients hospitalised due to acute infection were randomised to receive 10 days of aspirin treatment (80 mg 1dd or 40 mg 2dd) or no intervention (1:1:1 allocation). Measurements were performed during infection (T1; days 1-3), after intervention (T2; day 14) and without infection (T3; day > 90). The primary endpoint was platelet aggregation measured by the Platelet Function Analyzer (R) closure time (CT), and the secondary outcomes were serum and plasma thromboxane B2 (sTxB2 and pTxB2). Fifty-four patients (28 females) were included between January 2018 and December 2020. CT was 18% (95%CI 6;32) higher at T3 compared with T1 in the control group (n = 16), whereas sTxB2 and pTxB2 did not differ. Aspirin prolonged CT with 100% (95%CI 77; 127) from T1 to T2 in the intervention group (n = 38), while it increased with only 12% (95%CI 1;25) in controls. sTxB2 decreased with 95% (95%CI - 97; - 92) from T1 to T2, while it increased in the control group. pTxB2 was not affected compared with controls. Platelet aggregation is increased during severe infection, and this can be inhibited by aspirin. Optimisation of the treatment regimen may further diminish the persisting pTxB2 levels that point towards remaining platelet activity. This trial was registered on 13 April 2017 at EudraCT (2016-004303-32). Show less
Bulk, S. van den; Petrus, A.H.J.; Willemsen, R.T.A.; Boogers, M.J.; Meeder, J.G.; Rahel, B.M.; ... ; Bonten, T.N. 2023
Introduction Chest pain is a common reason for consultation in primary care. To rule out acute coronary syndrome (ACS), general practitioners (GP) refer 40%-70% of patients with chest pain to the... Show moreIntroduction Chest pain is a common reason for consultation in primary care. To rule out acute coronary syndrome (ACS), general practitioners (GP) refer 40%-70% of patients with chest pain to the emergency department (ED). Only 10%-20% of those referred, are diagnosed with ACS. A clinical decision rule, including a high-sensitive cardiac troponin-I point-of-care test (hs-cTnI-POCT), may safely rule out ACS in primary care. Being able to safely rule out ACS at the GP level reduces referrals and thereby alleviates the burden on the ED. Moreover, prompt feedback to the patients may reduce anxiety and stress.Methods and analysis The POB HELP study is a clustered randomised controlled diagnostic trial investigating the (cost-)effectiveness and diagnostic accuracy of a primary care decision rule for acute chest pain, consisting of the Marburg Heart Score combined with a hs-cTnI-POCT (limit of detection 1.6ng/L, 99th percentile 23ng/L, cut-off value between negative and positive used in this study 3.8ng/L). General practices are 2:1 randomised to the intervention group (clinical decision rule) or control group (regular care). In total 1500 patients with acute chest pain are planned to be included by GPs in three regions in The Netherlands. Primary endpoints are the number of hospital referrals and the diagnostic accuracy of the decision rule 24 hours, 6 weeks and 6 months after inclusion.Ethics and dissemination The medical ethics committee Leiden-Den Haag-Delft (the Netherlands) has approved this trial. Written informed consent will be obtained from all participating patients. The results of this trial will be disseminated in one main paper and additional papers on secondary endpoints and subgroup analyses. Show less
Background Prediction models for risk of cardiovascular events generally do not include young adults, and cardiovascular risk factors differ between women and men. Therefore, this study aimed to... Show moreBackground Prediction models for risk of cardiovascular events generally do not include young adults, and cardiovascular risk factors differ between women and men. Therefore, this study aimed to develop prediction models for first-ever cardiovascular event risk in men and women aged 30 to 49 years.Methods and Results We included patients aged 30 to 49 years without cardiovascular disease from a Dutch routine care database. Outcome was defined as first-ever cardiovascular event. Our reference models were sex-specific Cox proportional hazards models based on traditional cardiovascular predictors, which we compared with models using 2 predictor subsets with the 20 or 50 most important predictors based on the Cox elastic net model regularization coefficients. We assessed the C-index and calibration curve slopes at 10 years of follow-up. We stratified our analyses based on 30- to 39-year and 40- to 49-year age groups at baseline. We included 542 141 patients (mean age 39.7, 51% women). During follow-up, 10 767 cardiovascular events occurred. Discrimination of reference models including traditional cardiovascular predictors was moderate (women: C-index, 0.648 [95% CI, 0.645-0.652]; men: C-index, 0.661 [95%CI, 0.658-0.664]). In women and men, the Cox proportional hazard models including 50 most important predictors resulted in an increase in C-index (0.030 and 0.012, respectively), and a net correct reclassification of 3.7% of the events in women and 1.2% in men compared with the reference model.Conclusions Sex-specific electronic health record-derived prediction models for first-ever cardiovascular events in the general population aged <50 years have moderate discriminatory performance. Data-driven predictor selection leads to identification of nontraditional cardiovascular predictors, which modestly increase performance of models. Show less
BackgroundPrediction models for risk of cardiovascular events generally do not include young adults, and cardiovascular risk factors differ between women and men. Therefore, this study aimed to... Show moreBackgroundPrediction models for risk of cardiovascular events generally do not include young adults, and cardiovascular risk factors differ between women and men. Therefore, this study aimed to develop prediction models for first‐ever cardiovascular event risk in men and women aged 30 to 49 years.Methods and ResultsWe included patients aged 30 to 49 years without cardiovascular disease from a Dutch routine care database. Outcome was defined as first‐ever cardiovascular event. Our reference models were sex‐specific Cox proportional hazards models based on traditional cardiovascular predictors, which we compared with models using 2 predictor subsets with the 20 or 50 most important predictors based on the Cox elastic net model regularization coefficients. We assessed the C‐index and calibration curve slopes at 10 years of follow‐up. We stratified our analyses based on 30‐ to 39‐year and 40‐ to 49‐year age groups at baseline. We included 542 141 patients (mean age 39.7, 51% women). During follow‐up, 10 767 cardiovascular events occurred. Discrimination of reference models including traditional cardiovascular predictors was moderate (women: C‐index, 0.648 [95% CI, 0.645–0.652]; men: C‐index, 0.661 [95%CI, 0.658–0.664]). In women and men, the Cox proportional hazard models including 50 most important predictors resulted in an increase in C‐index (0.030 and 0.012, respectively), and a net correct reclassification of 3.7% of the events in women and 1.2% in men compared with the reference model.ConclusionsSex‐specific electronic health record‐derived prediction models for first‐ever cardiovascular events in the general population aged <50 years have moderate discriminatory performance. Data‐driven predictor selection leads to identification of nontraditional cardiovascular predictors, which modestly increase performance of models. Show less
Bulk, S. van den; Spoelman, W.A.; Dijkman, P.R.M. van; Numans, M.E.; Bonten, T.N.; Leiden Univ Med Ctr LUMC 2022
Background: The prevalence of coronary artery disease is increasing due to the aging population and increasing prevalence of cardiovascular risk factors. Non-acute chest pain often is the first... Show moreBackground: The prevalence of coronary artery disease is increasing due to the aging population and increasing prevalence of cardiovascular risk factors. Non-acute chest pain often is the first symptom of stable coronary artery disease. To optimise care for patients with non-acute chest pain and make efficient use of available resources, we need to know more about the current incidence, referral rate and management of these patients. Methods: We used routinely collected health data from the STIZON data warehouse in the Netherlands between 2010 and 2016. Patients > 18 years, with no history of cardiovascular disease, seen by the general practitioner (GP) for non-acute chest pain with a suspected cardiac origin were included. Outcomes were (i) incidence of new non-acute chest pain in primary care, (ii) referral rates to the cardiologist, (iii) correspondence from the cardiologist to the GP, (iv) registration by GPs of received correspondence and; (v) pharmacological guideline adherence after newly diagnosed stable angina pectoris. Results: In total 9029 patients were included during the study period, resulting in an incidence of new non-acute chest pain of 1.01/1000 patient-years. 2166 (24%) patients were referred to the cardiologist. In 857/2114 (41%) referred patients, correspondence from the cardiologist was not available in the GP's electronic medical record. In 753/1257 (60%) patients with available correspondence, the GP did not code the conclusion in the electronic medical record. Despite guideline recommendations, 37/255 (15%) patients with angina pectoris were not prescribed antiplatelet therapy nor anticoagulation, 69/255 (27%) no statin and 67/255 (26%) no beta-blocker. Conclusion: After referral, both communication from cardiologists and registration of the final diagnosis by GPs were suboptimal. Both cardiologists and GPs should make adequate communication and registration a priority, as it improves health outcomes. Secondary pharmacological prevention in patients with angina pectoris was below guideline standards. So, proactive attention needs to be given to optimise secondary prevention in this high-risk group in primary care. Show less
Background: There is increasing attention on machine learning (ML)-based clinical decision support systems (CDSS), but their added value and pitfalls are very rarely evaluated in clinical practice.... Show moreBackground: There is increasing attention on machine learning (ML)-based clinical decision support systems (CDSS), but their added value and pitfalls are very rarely evaluated in clinical practice. We implemented a CDSS to aid general practitioners (GPs) in treating patients with urinary tract infections (UTIs), which are a significant health burden worldwide. Objective: This study aims to prospectively assess the impact of this CDSS on treatment success and change in antibiotic prescription behavior of the physician. In doing so, we hope to identify drivers and obstacles that positively impact the quality of health care practice with ML. Methods: The CDSS was developed by Pacmed, Nivel, and Leiden University Medical Center (LUMC). The CDSS presents the expected outcomes of treatments, using interpretable decision trees as ML classifiers. Treatment success was defined as a subsequent period of 28 days during which no new antibiotic treatment for UTI was needed. In this prospective observational study, 36 primary care practices used the software for 4 months. Furthermore, 29 control practices were identified using propensity score-matching. All analyses were performed using electronic health records from the Nivel Primary Care Database. Patients for whom the software was used were identified in the Nivel database by sequential matching using CDSS use data. We compared the proportion of successful treatments before and during the study within the treatment arm. The same analysis was performed for the control practices and the patient subgroup the software was definitely used for. All analyses, including that of physicians' prescription behavior, were statistically tested using 2-sided z tests with an alpha level of .05. Results: In the treatment practices, 4998 observations were included before and 3422 observations (of 2423 unique patients) were included during the implementation period. In the control practices, 5044 observations were included before and 3360 observations were included during the implementation period. The proportion of successful treatments increased significantly from 75% to 80% in treatment practices (z=5.47, P<.001). No significant difference was detected in control practices (76% before and 76% during the pilot, z=0.02; P=.98). Of the 2423 patients, we identified 734 (30.29%) in the CDSS use database in the Nivel database. For these patients, the proportion of successful treatments during the study was 83%-a statistically significant difference, with 75% of successful treatments before the study in the treatment practices (z=4.95; P<.001). Conclusions: The introduction of the CDSS as an intervention in the 36 treatment practices was associated with a statistically significant improvement in treatment success. We excluded temporal effects and validated the results with the subgroup analysis in patients for whom we were certain that the software was used. This study shows important strengths and points of attention for the development and implementation of an ML-based CDSS in clinical practice. Trial Registration: ClinicalTrials.gov NCT04408976; https://clinicaltrials.gov/ct2/show/NCT04408976 Show less
Background: There is increasing attention on machine learning (ML)-based clinical decision support systems (CDSS), but their added value and pitfalls are very rarely evaluated in clinical practice.... Show moreBackground: There is increasing attention on machine learning (ML)-based clinical decision support systems (CDSS), but their added value and pitfalls are very rarely evaluated in clinical practice. We implemented a CDSS to aid general practitioners (GPs) in treating patients with urinary tract infections (UTIs), which are a significant health burden worldwide.Objective: This study aims to prospectively assess the impact of this CDSS on treatment success and change in antibiotic prescription behavior of the physician. In doing so, we hope to identify drivers and obstacles that positively impact the quality of health care practice with ML.Methods: The CDSS was developed by Pacmed, Nivel, and Leiden University Medical Center (LUMC). The CDSS presents the expected outcomes of treatments, using interpretable decision trees as ML classifiers. Treatment success was defined as a subsequent period of 28 days during which no new antibiotic treatment for UTI was needed. In this prospective observational study, 36 primary care practices used the software for 4 months. Furthermore, 29 control practices were identified using propensity score-matching. All analyses were performed using electronic health records from the Nivel Primary Care Database. Patients for whom the software was used were identified in the Nivel database by sequential matching using CDSS use data. We compared the proportion of successful treatments before and during the study within the treatment arm. The same analysis was performed for the control practices and the patient subgroup the software was definitely used for. All analyses, including that of physicians’ prescription behavior, were statistically tested using 2-sided z tests with an α level of .05.Results: In the treatment practices, 4998 observations were included before and 3422 observations (of 2423 unique patients) were included during the implementation period. In the control practices, 5044 observations were included before and 3360 observations were included during the implementation period. The proportion of successful treatments increased significantly from 75% to 80% in treatment practices (z=5.47, P<.001). No significant difference was detected in control practices (76% before and 76% during the pilot, z=0.02; P=.98). Of the 2423 patients, we identified 734 (30.29%) in the CDSS use database in the Nivel database. For these patients, the proportion of successful treatments during the study was 83%—a statistically significant difference, with 75% of successful treatments before the study in the treatment practices (z=4.95; P<.001).Conclusions: The introduction of the CDSS as an intervention in the 36 treatment practices was associated with a statistically significant improvement in treatment success. We excluded temporal effects and validated the results with the subgroup analysis in patients for whom we were certain that the software was used. This study shows important strengths and points of attention for the development and implementation of an ML-based CDSS in clinical practice. Show less
Background: There is increasing attention on machine learning (ML)-based clinical decision support systems (CDSS), but their added value and pitfalls are very rarely evaluated in clinical practice.... Show moreBackground: There is increasing attention on machine learning (ML)-based clinical decision support systems (CDSS), but their added value and pitfalls are very rarely evaluated in clinical practice. We implemented a CDSS to aid general practitioners (GPs) in treating patients with urinary tract infections (UTIs), which are a significant health burden worldwide.Objective: This study aims to prospectively assess the impact of this CDSS on treatment success and change in antibiotic prescription behavior of the physician. In doing so, we hope to identify drivers and obstacles that positively impact the quality of health care practice with ML.Methods: The CDSS was developed by Pacmed, Nivel, and Leiden University Medical Center (LUMC). The CDSS presents the expected outcomes of treatments, using interpretable decision trees as ML classifiers. Treatment success was defined as a subsequent period of 28 days during which no new antibiotic treatment for UTI was needed. In this prospective observational study, 36 primary care practices used the software for 4 months. Furthermore, 29 control practices were identified using propensity score-matching. All analyses were performed using electronic health records from the Nivel Primary Care Database. Patients for whom the software was used were identified in the Nivel database by sequential matching using CDSS use data. We compared the proportion of successful treatments before and during the study within the treatment arm. The same analysis was performed for the control practices and the patient subgroup the software was definitely used for. All analyses, including that of physicians’ prescription behavior, were statistically tested using 2-sided z tests with an α level of .05.Results: In the treatment practices, 4998 observations were included before and 3422 observations (of 2423 unique patients) were included during the implementation period. In the control practices, 5044 observations were included before and 3360 observations were included during the implementation period. The proportion of successful treatments increased significantly from 75% to 80% in treatment practices (z=5.47, P<.001). No significant difference was detected in control practices (76% before and 76% during the pilot, z=0.02; P=.98). Of the 2423 patients, we identified 734 (30.29%) in the CDSS use database in the Nivel database. For these patients, the proportion of successful treatments during the study was 83%—a statistically significant difference, with 75% of successful treatments before the study in the treatment practices (z=4.95; P<.001).Conclusions: The introduction of the CDSS as an intervention in the 36 treatment practices was associated with a statistically significant improvement in treatment success. We excluded temporal effects and validated the results with the subgroup analysis in patients for whom we were certain that the software was used. This study shows important strengths and points of attention for the development and implementation of an ML-based CDSS in clinical practice. Show less
Background Implementation of digital health (eHealth) generally involves adapting pre-established and carefully considered processes or routines, and still raises multiple ethical and legal... Show moreBackground Implementation of digital health (eHealth) generally involves adapting pre-established and carefully considered processes or routines, and still raises multiple ethical and legal dilemmas. This study aimed to identify challenges regarding responsibility and liability when prescribing digital health in clinical practice. This was part of an overarching project aiming to explore the most pressing ethical and legal obstacles regarding the implementation and adoption of digital health in the Netherlands, and to propose actionable solutions. Methods A series of multidisciplinary focus groups with stakeholders who have relevant digital health expertise were analysed through thematic analysis. Results The emerging general theme was 'uncertainty regarding responsibilities' when adopting digital health. Key dilemmas take place in clinical settings and within the doctor-patient relationship ('professional digital health'). This context is particularly challenging because different stakeholders interact. In the absence of appropriate legal frameworks and codes of conduct tailored to digital health, physicians' responsibility is to be found in their general duty of care. In other words: to do what is best for patients (not causing harm and doing good). Professional organisations could take a leading role to provide more clarity with respect to physicians' responsibility, by developing guidance describing physicians' duty of care in the context of digital health, and to address the resulting responsibilities. Conclusions Although legal frameworks governing medical practice describe core ethical principles, rights and obligations of physicians, they do not suffice to clarify their responsibilities in the setting of professional digital health. Here we present a series of recommendations to provide more clarity in this respect, offering the opportunity to improve quality of care and patients' health. The recommendations can be used as a starting point to develop professional guidance and have the potential to be adapted to other healthcare professionals and systems. Show less