General Practitioner Diagnosis Research Articles

Implementation of lung ultrasonography by general practitioners for lower respiratory tract infections: a feasibility study

Objective To evaluate the feasibility of lung ultrasonography (LUS) performed by novice users’ general practitioners (GPs) in diagnosing lower respiratory tract infections (LRTIs) in primary health care settings. Design A prospective interventional multicenter study (December 2019–March 2020). Settings and subjects Patients aged >3 months, suspected of having LRTI consulting in three different general practices (GPs) (rural, semirural and urban) in France. Main outcome measures Feasibility of LUS by GPs was assessed by (1) the proportion of patients where LUS was not performed, (2) technical breakdowns, (3) interpretability of images by GPs, (4) examination duration and (5) patient perception and acceptability. Results A total of 151 patients were recruited, and GPs performed LUS for 111 (73.5%) patients (LUS group). In 99.1% (n = 110) of cases, GPs indicated that they were able to interpret images. The median [IQR] exam duration was 4 [3–5] minutes. LRTI was diagnosed in 70.3% and 60% of patients in the LUS and no-LUS groups, respectively (p = .43). After LUS, GPs changed their diagnosis from ‘other’ to ‘LRTI’ in six cases (+5.4%, p < .001), prescribed antibiotics for five patients (+4.5%, p = .164) and complementary chest imaging for 10 patients (+9%, p < .001). Patient stress was reported in 1.8% of cases, 81.7% of patients declared that they better understood the diagnosis, and 82% of patients thought that the GP diagnosis was more reliable after LUS. Conclusions LUS by GPs using handheld devices is a feasible diagnostic tool in primary health care for LRTI symptoms, demonstrating both effectiveness and positive patient reception. Trial Registration Number Clinicaltrial.gov: NCT04602234, 20/10/2020.

Collective Intelligence Increases Diagnostic Accuracy in a General Practice Setting.

General practitioners (GPs) work in an ill-defined environment where diagnostic errors are prevalent. Previous research indicates that aggregating independent diagnoses can improve diagnostic accuracy in a range of settings. We examined whether aggregating independent diagnoses can also improve diagnostic accuracy for GP decision making. In addition, we investigated the potential benefit of such an approach in combination with a decision support system (DSS). We simulated virtual groups using data sets from 2 previously published studies. In study 1, 260 GPs independently diagnosed 9 patient cases in a vignette-based study. In study 2, 30 GPs independently diagnosed 12 patient actors in a patient-facing study. In both data sets, GPs provided diagnoses in a control condition and/or DSS condition(s). Each GP's diagnosis, confidence rating, and years of experience were entered into a computer simulation. Virtual groups of varying sizes (range: 3-9) were created, and different collective intelligence rules (plurality, confidence, and seniority) were applied to determine each group's final diagnosis. Diagnostic accuracy was used as the performance measure. Aggregating independent diagnoses by weighing them equally (i.e., the plurality rule) substantially outperformed average individual accuracy, and this effect increased with increasing group size. Selecting diagnoses based on confidence only led to marginal improvements, while selecting based on seniority reduced accuracy. Combining the plurality rule with a DSS further boosted performance. Combining independent diagnoses may substantially improve a GP's diagnostic accuracy and subsequent patient outcomes. This approach did, however, not improve accuracy in all patient cases. Therefore, future work should focus on uncovering the conditions under which collective intelligence is most beneficial in general practice. We examined whether aggregating independent diagnoses of GPs can improve diagnostic accuracy.Using data sets of 2 previously published studies, we composed virtual groups of GPs and combined their independent diagnoses using 3 collective intelligence rules (plurality, confidence, and seniority).Aggregating independent diagnoses by weighing them equally substantially outperformed average individual GP accuracy, and this effect increased with increasing group size.Combining independent diagnoses may substantially improve GP's diagnostic accuracy and subsequent patient outcomes.

Graph Neural Network Based Multi-Label Hierarchical Classification for Disease Predictions in General Practice.

General practitioners are supposed to be better diagnostics to detect patients with serious diseases earlier, and conduct early interventions and appropriate referrals of patients. However, in the current general practice, primary general practitioners lack sufficient clinical experiences, and the correct rate of general disease diagnosis is low. To assist general practitioners in diagnosis, this paper proposes a multi-label hierarchical classification method based on graph neural network, which integrates medical knowledge and electronic health record (EHR) data to build a disease prediction model. The experimental results based on data consist of 231,783 visits from EHR show that the proposed model outperforms all baseline models in the general disease prediction task with a top-3 recall of 0.865. The interpretable results of the model can effectively help clinicians understand the basis of the model's decision-making.

Assessing the importance of primary care diagnoses in the UK Biobank

The UK Biobank has made general practitioner (GP) data (censoring date 2016–2017) available for approximately 45% of the cohort, whilst hospital inpatient and death registry (referred to as “HES/Death”) data are available cohort-wide through 2018–2022 depending on whether the data comes from England, Wales or Scotland. We assessed the importance of case ascertainment via different data sources in UKB for three diseases that are usually first diagnosed in primary care: Parkinson’s disease (PD), type 2 diabetes (T2D), and all-cause dementia. Including GP data at least doubled the number of incident cases in the subset of the cohort with primary care data (e.g. from 619 to 1390 for dementia). Among the 786 dementia cases that were only captured in the GP data before the GP censoring date, only 421 (54%) were subsequently recorded in HES. Therefore, estimates of the absolute incidence or risk-stratified incidence are misleadingly low when based only on the HES/Death data. For incident cases present in both HES/Death and GP data during the full follow-up period (i.e. until the HES censoring date), the median time difference between an incident diagnosis of dementia being recorded in GP and HES/Death was 2.25 years (i.e. recorded 2.25 years earlier in the GP records). Similar lag periods were also observed for PD (median 2.31 years earlier) and T2D (median 2.82 years earlier). For participants with an incident GP diagnosis, only 65.6% of dementia cases, 69.0% of PD cases, and 58.5% of T2D cases had their diagnosis recorded in HES/Death within 7 years since GP diagnosis. The effect estimates (hazard ratios, HR) of established risk factors for the three health outcomes mostly remain in the same direction and with a similar strength of association when cases are ascertained either using HES only or further adding GP data. The confidence intervals of the HR became narrower when adding GP data, due to the increased statistical power from the additional cases. In conclusion, it is desirable to extend both the coverage and follow-up period of GP data to allow researchers to maximise case ascertainment of chronic health conditions in the UK.

BT02 Use of teledermatology in the 2-week wait pathway: 5 years of positive data

Abstract Our teledermatology service was started in 2018 to facilitate the dermatology department’s move offsite and allow for the continuation of inpatient care digitally without the need for increasing staff levels or cross-site working. Its implementation also allowed for efficient adaptation and use throughout the COVID-19 pandemic and recovery following this period. Alongside these aspects, teledermatology has become integral in the department’s rapid-access 2-week wait (2WW) capacity. The pathway involves patients attending the department and undergoing a nurse-led review with upload of the history alongside macroscopic and dermoscopic photographs to the platform for consultant review and management. The platform conforms to all aspects of the British Association of Dermatologists’ teledermatology guidelines and is hosted on the National Health Service’s Health and Social Care Network. Of almost 4000 2WW referrals assessed by digital teledermatology between 2018 and the present, no referrals have been rejected for poor quality, only 1% required urgent face-to-face review with a consultant, 32% were discharged, 53% were booked in for a minor surgical procedure of which half were suspected to be cancer or precancer and performed urgently, and 14% were booked for routine review due to risk factors such a multiple moles and need for a full skin check. Our consultants can assess 40 patients on the digital platform in a 4-h period, and the overall cost is approximately half of a face-to-face clinic, including admin. Of all 2WW referrals, general practitioner diagnosis has remained quite constant, with 30% of lesions thought to be malignant melanoma (MM), 36% squamous cell carcinoma (SCC) and 3% basal cell carcinoma (BCC); in the remaining 31%, no diagnosis was made. Of those referred for surgery by the dermatology consultants, the clinical diagnosis was BCC in 20%, SCC in 22% and MM or MM in situ in 6%; the remainder was thought to be precancerous, atypical naevi or other. We have continued to demonstrate through audit that the clinicopathological correlation in the digital teledermatology clinic is as good as face-to-face clinics for lesion diagnosis because of the quality of the history and images. Our most recent satisfaction survey in 2022 showed 96% patient satisfaction with the initial teledermatology assessment; of those booked for a procedure, 88% strongly agreed that all concerns were addressed prior to surgery. All patients would recommend the service to friends and family. With nearly 5 years of data and continuous evaluation, we have shown the effectiveness of a well-designed, connected digital system in supporting our service without duplication of effort by clinicians or prolongation of the patient’s journey.

Classification of Long COVID from General Practitioner Diagnosis Text.

Major challenges exist in identifying Long COVID patients from diagnosis texts recorded by general practitioners. A classification framework is proposed that can be used to identify Long COVID patients given these unstructured diagnostic texts. This framework can be leveraged to provide a general understanding of the risk factors, management strategies, and outcomes associated with Long COVID in Australia.

The Design of Convolutional Neural Networks Model for Classification of Ear Diseases on Android Mobile Devices

An otorhinolaryngologist (ORL) or general practitioner diagnoses ear disease based on ear image information. However, general practitioners refer patients to ORL for chronic ear disease because the image of ear disease has high complexity, variety, and little difference between diseases. An artificial intelligence-based approach is needed to make it easier for doctors to diagnose ear diseases based on ear image information, such as the Convolutional Neural Network (CNN). This paper describes how CNN was designed to generate CNN models used to classify ear diseases. The model was developed using an ear image dataset from the practice of an ORL at the University of Mataram Teaching Hospital. This work aims to find the best CNN model for classifying ear diseases applicable to android mobile devices. Furthermore, the best CNN model is deployed for an Android-based application integrated with the Endoscope Ear Cleaning Tool Kit for registering patient ear images. The experimental results show 83% accuracy, 86% precision, 86% recall, and 4ms inference time. The application produces a System Usability Scale of 76.88% for testing, which shows it is easy to use. This achievement shows that the model can be developed and integrated into an ENT expert system. In the future, the ENT expert system can be operated by workers in community health centres/clinics to assist leading health them in diagnosing ENT diseases early.

Effectiveness of analgesic ear drops as add-on treatment to oral analgesics in children with acute otitis media: study protocol of the OPTIMA pragmatic randomised controlled trial

IntroductionEar pain is the most prominent symptom of childhood acute otitis media (AOM). To control the pain and reduce reliance on antibiotics, evidence of effectiveness for alternative interventions is urgently...

A machine learning approach reveals features related to clinicians' diagnosis of clinically relevant knee osteoarthritis.

To identify highly ranked features related to clinicians' diagnosis of clinically relevant knee OA. General practitioners (GPs) and secondary care physicians (SPs) were recruited to evaluate 5-10 years follow-up clinical and radiographic data of knees from the CHECK cohort for the presence of clinically relevant OA. GPs and SPs were gathered in pairs; each pair consisted of one GP and one SP, and the paired clinicians independently evaluated the same subset of knees. A diagnosis was made for each knee by the GP and SP before and after viewing radiographic data. Nested 5-fold cross-validation enhanced random forest models were built to identify the top 10 features related to the diagnosis. Seventeen clinician pairs evaluated 1106 knees with 139 clinical and 36 radiographic features. GPs diagnosed clinically relevant OA in 42% and 43% knees, before and after viewing radiographic data, respectively. SPs diagnosed in 43% and 51% knees, respectively. Models containing top 10 features had good performance for explaining clinicians' diagnosis with area under the curve ranging from 0.76-0.83. Before viewing radiographic data, quantitative symptomatic features (i.e. WOMAC scores) were the most important ones related to the diagnosis of both GPs and SPs; after viewing radiographic data, radiographic features appeared in the top lists for both, but seemed to be more important for SPs than GPs. Random forest models presented good performance in explaining clinicians' diagnosis, which helped to reveal typical features of patients recognized as clinically relevant knee OA by clinicians from two different care settings.

Comparing cardiovascular disease incidence and prevalence between depressed and non-depressed older persons over time: Cohort differences in the Longitudinal Aging Study Amsterdam

ObjectivePrevious studies suggest that prevention of cardiovascular disease (CVD) is more difficult in depressed persons. Although the prevalence and incidence of CVD decreased over the past decades, it is uncertain whether this is also true for depressed persons. This study examined whether changes in the prevalence and incidence of CVD differ between depressed and non-depressed older persons. MethodsLongitudinal data from three community-based representative birth cohorts aged 55–65 years of the Longitudinal Aging Study Amsterdam were used; N = 1070 born in 1926–1937, N = 995 born in 1938–1947, N = 1019 born in 1948–1957. The outcome included fatal and non-fatal CVD. Depression was defined as ≥16 points on the Center for Epidemiological Studies Depression Scale or a general practitioner's diagnosis of depression. Data were analysed with for age and sex adjusted logistic regression and cox regression models. ResultsComparing cohort 2 and 3 with cohort 1 in the complete sample, the prevalence of CVD decreased by 2% (OR = 0.98, 95%CI = 0.76–1.26) and 32% (OR = 0.68, 95%CI = 0.52–0.89), respectively, and the three-year incidence of CVD decreased by 6% (OR = 0.94, 95%CI = 0.63–1.41) and 26% (OR = 0.74, 95%CI = 0.48–1.15), respectively. The 13-year incidence of CVD decreased by 19% (HR = 0.81, 95%CI = 0.67–0.99) in cohort 2 compared to cohort 1. These decreases did not differ statistically significantly between depressed and non-depressed respondents (p-values of interaction terms: 0.35–0.98). ConclusionSubstantial decreases in the prevalence and incidence of CVD were observed in depressed and non-depressed older persons. Although this is encouraging, cardiovascular risk remained higher in depressed persons over time, warranting tailored prevention programs for depressed older persons.

An Assessment Framework for the Training of General Practitioners and Specialists Based on EPAs.

PurposeThe purpose of this study is to explore the practicality and feasibility of EPA (level 4 theory) for establishing medical training and service model in China.MethodWe opted for a mixed qualitative and quantitative research method, considering both explanatory and exploratory sequential designs. The qualitative research comprehended focus groups and interviews conducted with two panels of experts. The quantitative research was conducted to collect data about the applicability of international entrustable professional activities (EPAs) pediatrics standards in the Chinese context by organizing a seminar with a sample of 60 pediatricians. A questionnaire was designed with EPAs and distributed within professional networks. Structural equation modeling and statistical analysis were used to process the data.ResultsIn this study, Medical Service-Groups Model (MSGM) with four levels was successfully established to measure the correlation between specialized and general EPAs. As expected, results showed that specialized EPAs were built on top of general EPAs. There may be a mediating mechanism that general EPAs contribute to the lower level of specialization EPAs. In addition, levels 1 and 2 were primarily needed to lay the groundwork for levels 3 and 4, and these higher levels of EPAs were still the most informative for specialized Gastroenterology EPAs.ConclusionsThe diagnosis and treatment level of primary general practitioners, as the basis of the pediatric medical service chain, affected the clinical disposal ability of specialists. The establishment of MSGM provided a theoretical basis for the linkage training of general practitioners and specialist physicians. In future studies, scholars must explore China's EPAs based on unique national conditions.

EASI™ 12‑lead ECG with a handheld computer refines cardiovascular diagnosis in general practice

BackgroundElectrocardiogram (ECG) is used to a small extent in general medicine, because of general practitioner (GP) apprehension about interpretation and time consumption. AimThis study tested the hypothesis that user-friendly EASI ECG improves GP diagnosis of cardiovascular symptoms. MethodPatients over 18 years with recent cardiovascular symptoms or auscultation rhythm abnormalities were included in this prospective, multicentric study (10 practices, 17 GPs). ECG recordings were made with Cardiosecur® (4‑lead ECG connected to a handheld computer for EASI™ processing). Besides clinical data, diagnosis/patient referral were noted before and after ECG and interpretation. GP diagnosis and ECG interpretation were compared with a reference diagnosis made by ECG specialist. ResultsThere were 338 patients; 66% had cardiovascular risk factors. ECGs were performed for chest pain (41%), auscultation rhythm abnormalities (33%) or palpitations (19%). Average time to perform ECG was 4.7 ± 2.1 min, with possible home recordings. Compared with standard ECG, improvement provided by Cardiosecur® was scored 9/10 (range 7–10) by GPs. GPs correctly interpreted ECG normality/abnormality in 77% of patients. Diagnosis was correctly changed for 14% of patients thanks to the ECG, and wrongly changed for 2%. One new appropriate final diagnosis was achieved for 9 ECG recordings (p < 0.001). Diagnostic certainty increased 1.9 ± 2.1/10 (p < 0.001). ECG brought about changes in GP decision making: referral or treatment changed for 82 patients (24%) and complementary test for 69 patients (20%). ConclusionThe EASI™ algorithm coupled with a handheld computer facilitates ECG recordings in the primary care setting, providing improved diagnosis.

Impact of the third national 'Be Clear on Cancer' Breast Cancer in Women over 70 Campaign on general practitioner attendance and referral, diagnosis rates and prevalence awareness.

ObjectiveMore than a third of women diagnosed with breast cancer in England, and over half of those who die from it, are over 70. The Be Clear on Cancer Breast Cancer in Women over 70 Campaign, running three times, 2014–2018, aimed to promote early diagnosis of breast cancer in England by raising symptom awareness and encouraging women to see their general practitioner (GP) without delay. We sought to establish whether the third campaign had successfully met its aims.MethodsMetrics covering the patient pathway, including symptom awareness, attending a GP practice with symptoms, urgent GP referral, diagnosis and stage of cancer, were assessed using national cancer databases and two household surveys.ResultsThe third campaign was associated with an increase in urgent cancer referrals, and therefore mammograms and ultrasounds performed. This was associated with an increase in breast cancers diagnosed. There was a delayed effect on GP attendances. Awareness of breast cancer prevalence for the 70‐and‐over age group improved. Impact on these metrics diminished across successive campaigns.ConclusionsFuture campaigns should focus on harder‐to‐reach women and include GPs as targets as this campaign showed a potential to affect referral behaviour.

The impact of clinical risk conditions on influenza and pneumonia diagnoses in England: a nationally representative retrospective cohort study, 2010-2019.

The impact of influenza and pneumonia on individuals in clinical risk groups in England has not previously been well characterized. Using nationally representative linked databases (Clinical Practice Research Database (CPRD), Hospital Episode Statistics (HES) and Office for National Statistics (ONS)), we conducted a retrospective cohort study among adults (≥ 18 years) during the 2010/2011-2019/2020 influenza seasons to estimate the incidence of influenza- and pneumonia-diagnosed medical events (general practitioner (GP) diagnoses, hospitalisations and deaths), stratified by age and risk conditions. The study population included a seasonal average of 7.2 million individuals; approximately 32% had ≥1 risk condition, 42% of whom received seasonal influenza vaccines. Medical event incidence rates increased with age, with ~1% of adults aged ≥75 years hospitalized for influenza/pneumonia annually. Among individuals with vs. without risk conditions, GP diagnoses occurred 2-5-fold more frequently and hospitalisations were 7-10-fold more common. Among those with obesity, respiratory, kidney or cardiovascular disorders, hospitalisation were 5-40-fold more common than in individuals with no risk conditions. Though these findings likely underestimate the full burden of influenza, they emphasize the concentration of disease burden in specific age and risk groups and support existing recommendations for influenza vaccination.

Understanding physician behaviour in the 6–8 weeks hip check in primary care: a qualitative study using the COM-B

ObjectivesA compulsory hip check is performed on an infant at 6–8 weeks in primary care for the detection of developmental dysplasia of the hip (DDH). Missed diagnoses and infants incorrectly...

Deep Neural Frameworks Improve the Accuracy of General Practitioners in the Classification of Pigmented Skin Lesions.

This study evaluated whether deep learning frameworks trained in large datasets can help non-dermatologist physicians improve their accuracy in categorizing the seven most common pigmented skin lesions. Open-source skin images were downloaded from the International Skin Imaging Collaboration (ISIC) archive. Different deep neural networks (DNNs) (n = 8) were trained based on a random dataset constituted of 8015 images. A test set of 2003 images was used to assess the classifiers’ performance at low (300 × 224 RGB) and high (600 × 450 RGB) image resolution and aggregated data (age, sex and lesion localization). We also organized two different contests to compare the DNN performance to that of general practitioners by means of unassisted image observation. Both at low and high image resolution, the DNN framework differentiated dermatological images with appreciable performance. In all cases, the accuracy was improved when adding clinical data to the framework. Finally, the least accurate DNN outperformed general practitioners. The physician’s accuracy was statistically improved when allowed to use the output of this algorithmic framework as guidance. DNNs are proven to be high performers as skin lesion classifiers and can improve general practitioner diagnosis accuracy in a routine clinical scenario.

Study of potentially preventable hospitalisations (PPH) for chronic conditions: what proportion are preventable and what factors are associated with preventable PPH?

IntroductionThe proportion of potentially preventable hospitalisations (PPH) which are actually preventable is unknown, and little is understood about the factors associated with individual preventable PPH. The Diagnosing Potentially Preventable Hospitalisations...

The Added Value of Radiographs in Diagnosing Knee Osteoarthritis Is Similar for General Practitioners and Secondary Care Physicians; Data from the CHECK Early Osteoarthritis Cohort.

Objective: The purpose of this study was to evaluate the added value of radiographs for diagnosing knee osteoarthritis (KOA) by general practitioners (GPs) and secondary care physicians (SPs). Methods: Seventeen GPs and nineteen SPs were recruited to evaluate 1185 knees from the CHECK cohort (presenters with knee pain in primary care) for the presence of clinically relevant osteoarthritis (OA) during follow-up. Experts were required to make diagnoses independently, first based on clinical data only and then on clinical plus radiographic data, and to provide certainty scores (ranging from 1 to 100, where 1 was “certainly no OA” and 100 was “certainly OA”). Next, experts held consensus meetings to agree on the final diagnosis. With the final diagnosis as gold standard, diagnostic indicators were calculated (sensitivity, specificity, positive/negative predictive value, accuracy and positive/negative likelihood ratio) for all knees, as well as for clinically “certain” and “uncertain” knees, respectively. Student paired t-tests compared certainty scores. Results: Most diagnoses of GPs (86%) and SPs (82%) were “consistent” after assessment of radiographic data. Diagnostic indicators improved similarly for GPs and SPs after evaluating the radiographic data, but only improved relevantly in clinically “uncertain” knees. Radiographs added some certainty to “consistent” OA knees (GP 69 vs. 72, p < 0.001; SP 70 vs. 77, p < 0.001), but not to the consistent no OA knees (GP 21 vs. 22, p = 0.16; SP 20 vs. 21, p = 0.04). Conclusions: The added value of radiographs is similar for GP and SP, in terms of diagnostic accuracy and certainty. Radiographs appear to be redundant when clinicians are certain of their clinical diagnosis.

Assessment of an asyncronic tele dermatology system for primary care physicians via social networking app WhatsApp

In Chile, there is little access to Dermatology outpatient consultations in the public health care system, which has favored the development of tele dermatology (TD). To assess satisfaction levels of primary care providers with a TD channel via the social networking app WhatsApp® and the concordance between the diagnosis of general practitioners and dermatologists at Pontificia Universidad Católica de Chile. An electronic survey was answered by the general practitioners who participate in the TD channel. In 417 cases, the diagnostic concordance between general practitioners and specialists was assessed. The survey was answered by 84 practitioners. General satisfaction levels with the platform were over 95%. Satisfaction levels with the response speed and the management suggested by the specialist were over 90%. Over 80% of the practitioners read the consultation sent by dermatologists and considered that their dermatological knowledge improved. Diagnostic concordance between practitioners and specialists was 41%. TD via a WhatsApp® group linked to a university is a low cost and easy to implement intervention, generating high levels of satisfaction among general practitioners.

Identifying the Prescribing Physician in US Healthcare Claims Data.

Identifying the prescribing physician may be essential in studies that use health care claims databases but often there is no single consistent physician identifier (ID) available that spans an entire database and no standard methods have emerged to address this issue. Unique Drug Enforcement Administration (DEA) numbers (N=36,721) were identified from the pharmacy claims of a cohort of postmenopausal (55+) initiators of osteoporosis medications (2008-2011) in the United Healthcare database. The proportion of times a Provider ID appeared at least once on the outpatient medical service claims (OMSCs) in the 14 days before a new prescription (Rx) out of the total number of new Rxs for each DEA was calculated and the Provider ID with the highest proportion was considered the most likely match to the DEA. We used regression models to evaluate how characteristics of OMSCs (N=36,721) related to the probability that the provider would be the prescribing physician on a proximal pharmacy claim for each patient (N=20,058). A total of 37,448,056 new Rxs and 132,135,673 OMSCs were associated with the DEAs. Family and general practitioner, hospitalist, and osteoporosis diagnoses were strong predictors of a match. Our models had sensitivities ranging from 81.9% to 82.8% and specificities ranging from 52.4% to 53.5%. Our algorithms were good at identifying matches for the prescribing physician according to our reference standard but may have included some false positives. Predicted probabilities from our models could be used in other databases where physician IDs are unavailable to help identify the prescriber.