Diagnostic classification in patients with suspected deep venous thrombosis: physicians' judgement or a decision rule?

Clinical decision rules in primary care: necessary investments for sustainable healthcare

Clinical judgement in primary care is more often decisive than in the hospital. Clinical decision rules (CDRs) can help general practitioners facilitating the work-through of differentials that follows an initial suspicion, resulting in a concrete 'course of action': a 'rule-out' without further testing, a need for further testing, or a specific treatment. However, in daily primary care, the use of CDRs is limited to only a few isolated rules. In this paper, we aimed to provide insight into the laborious path required to implement a viable CDR. At the same time, we noted that the limited use of CDRs in primary care cannot be explained by implementation barriers alone. Through the case study of the Oudega rule for the exclusion of deep vein thrombosis, we concluded that primary care CDRs come out best if they are tailor-made, taking into consideration the specific context of primary health care. Current CDRs should be evaluated frequently, and future decision rules should anticipate the latest developments such as the use of point-of-care (POC) tests. Hence, such new powerful diagnostic CDRs could improve and expand the possibilities for patient-oriented primary care.

Two point-of-care test-based approaches for the exclusion of deep vein thrombosis in general practice: a cost-effectiveness analysis

BACKGROUND

In the diagnostic work-up of deep vein thrombosis (DVT), the use of point-of-care-test (POCT) D-dimer assays is emerging as a promising patient-friendly alternative to regular D-dimer assays, but their cost-effectiveness is unknown. We compared the cost-effectiveness of two POCT-based approaches to the most common, laboratory-based, situation.

METHODS

A patient-level simulation model was developed to simulate the diagnostic trajectory of patients presenting with symptoms of DVT at the general practitioner (GP). Three strategies were defined for further diagnostic work-up: one based on current guidelines ('regular strategy') and two alternative approaches where a POCT for D-dimer is implemented at the 1) phlebotomy service ('DVT care pathway') and 2) GP practice ('fast-POCT strategy'). Probabilities, costs and health outcomes were obtained from the literature. Costs and effects were determined from a societal perspective over a time horizon of 6 months. Uncertainty in model outcomes was assessed with a one-way sensitivity analysis.

RESULTS

The Quality-Adjusted Life Years (QALYs) scores for the three DVT diagnostic work-up strategies were all around 0.43 across a 6 month-time horizon. Cost-savings of the two POCT-based strategies compared to the regular strategy were €103/patient for the DVT care pathway (95% CI: -€117-89), and €87/patient for the fast-POCT strategy (95% CI: -€113-67).

CONCLUSIONS

Point-of-care-based approaches result in similar health outcomes compared with regular strategy. Given their expected cost-savings and patient-friendly nature, we recommend implementing a D-dimer POCT device in the diagnostic DVT work-up.

A Machine Learning Approach to Predict Deep Venous Thrombosis Among Hospitalized Patients

Deep venous thrombosis (DVT) is associated with significant morbidity, mortality, and increased healthcare costs. Standard scoring systems for DVT risk stratification often provide insufficient stratification of hospitalized patients and are unable to accurately predict which inpatients are most likely to present with DVT. There is a continued need for tools which can predict DVT in hospitalized patients. We performed a retrospective study on a database collected from a large academic hospital, comprised of 99,237 total general ward or ICU patients, 2,378 of whom experienced a DVT during their hospital stay. Gradient boosted machine learning algorithms were developed to predict a patient's risk of developing DVT at 12- and 24-hour windows prior to onset. The primary outcome of interest was diagnosis of in-hospital DVT. The machine learning predictors obtained AUROCs of 0.83 and 0.85 for DVT risk prediction on hospitalized patients at 12- and 24-hour windows, respectively. At both 12 and 24 hours before DVT onset, the most important features for prediction of DVT were cancer history, VTE history, and internal normalized ratio (INR). Improved risk stratification may prevent unnecessary invasive testing in patients for whom DVT cannot be ruled out using existing methods. Improved risk stratification may also allow for more targeted use of prophylactic anticoagulants, as well as earlier diagnosis and treatment, preventing the development of pulmonary emboli and other sequelae of DVT.

Point-of-Care Testing for D-Dimer in the Diagnosis of Venous Thromboembolism in Primary Care: A Narrative Review

Venous thromboembolism (VTE) is regarded as a significant cause of mortality and disability, affecting 1-2 per 1000 people annually, presenting with a relatively wide range of symptoms, which can pose a diagnostic challenge. Historically, people in whom VTE is suspected will have been taken to hospital for diagnosis and treatment; however, a high proportion of patients are found not to have VTE. Concerns have been expressed about potential delays in treatment, with the risk of additional morbidity and disability, and death. Diagnostic strategies are typically based on the use of a clinical prediction rule to determine the pre-test probability, complemented with a measurement of D-dimer, with confirmation by imaging assessment. This narrative review explores the literature on the use of point-of-care testing (POCT) for the measurement of D-dimer, as part of a clinical decision rule, for the diagnosis of deep vein thrombosis (DVT) and pulmonary embolism (PE) in the primary care setting. In the two main prospective management (validation) studies that included D-dimer POCT or similar technologies, with a total cohort of 1600 participants, DVT was ruled out in 49% of patients, with a false negative rate of 1.4%, whereas PE was ruled out in 45% of patients, with a false negative rate of 1.5%. This suggests that uptake of POCT D-dimer in primary care has the potential to reduce the number of referrals to hospitals for imaging confirmatory investigation, with consequent cost savings. Thus, adopting POCT for D-dimer in primary care can offer clinical and cost benefits, particularly when quantitative POCT assays are being used. Furthermore, POCT should be undertaken in collaboration with the local laboratories to ensure the harmonisation of D-dimer methods and quality assurance to improve the diagnosis of VTE.

Identifying frailty in primary care: a qualitative description of family physicians' gestalt impressions of their older adult patients

Background

Many tools exist to guide family physicians’ impressions about frailty status of older adults, but no single tool, instrument, or set of criteria has emerged as most useful. The role of physicians’ subjective impressions in frailty decisions has not been studied. This study explores how family physicians conceptualize frailty, and the factors that they consider when making subjective decisions about patients’ frailty statuses.

Methods

Descriptive qualitative study of family physicians who practice in a large urban academic family medicine center as they participated in one-on-one “think-aloud” interviews about the frailty status of their patients aged 80 years and over. Of 23 eligible family physicians, 18 shared their impressions about the frailty status of their older adult patients and the factors influencing their decisions. Interviews were audio-recorded, transcribed, and thematically analyzed.

Results

Four themes were identified, the first of which described how physicians conceptualized frailty as a spectrum and dynamic in nature, but also struggled to conceptualize it without a formal definition in place. The remaining three themes described factors considered before determining patients’ frailty statuses: physical characteristics (age, weight, medical conditions), functional characteristics (physical, cognitive, social) and living conditions (level of independence, availability of supports, physical environment).

Conclusions

Family physicians viewed frailty as multifactorial, dynamic, and inclusive of functional and environmental factors. This conceptualization can be useful to make comprehensive and flexible evaluations of frailty status in conjunction with more objective frailty tools.

Electronic supplementary material

The online version of this article (10.1186/s12875-018-0743-4) contains supplementary material, which is available to authorized users.

Cost-effectiveness analysis of rivaroxaban for treatment and secondary prevention of venous thromboembolism in the Netherlands

BACKGROUND

Until recently, standard treatment of venous thromboembolism (VTE) concerned a combination of short-term low-molecular-weight heparin (LMWH) and long-term vitamin-K antagonist (VKA). Risk of bleeding and the requirement for regular anticoagulation monitoring are, however, limiting their use. Rivaroxaban is a novel oral anticoagulant associated with a significantly lower risk of major bleeds (hazard ratio = 0.54, 95% confidence interval = 0.37-0.79) compared to LMWH/VKA therapy, and does not require regular anticoagulation monitoring.

AIMS

To evaluate the health economic consequences of treating acute VTE patients with rivaroxaban compared to treatment with LMWH/VKA, viewed from the Dutch societal perspective.

METHODS

A life-time Markov model was populated with the findings of the EINSTEIN phase III clinical trial to analyze cost-effectiveness of rivaroxaban therapy in treatment and prevention of VTE from a Dutch societal perspective. Primary model outcomes were total and incremental quality-adjusted life years (QALYs), as well as life expectancy and costs.

RESULTS

Over a patient's lifetime, rivaroxaban was shown to be dominant, with health gains of 0.047 QALYs and cost savings of €304 compared to LMWH/VKA therapy. Dominance was robustly present in all sensitivity analyses. Major drivers of the differences between the two treatment arms were related to anticoagulation monitoring (medical costs, travel costs, and loss of productivity) and the occurrence of major bleeds.

CONCLUSION

Rivaroxaban treatment of patients with venous thromboembolism results in health gains and cost savings compared to LMWH/VKA therapy. This conclusion holds for the Dutch setting, both for the societal perspective, as well as the healthcare perspective.

A systematic review of clinical prediction scores for deep vein thrombosis

Objective Diagnosis of deep vein thrombosis remains a challenging problem. Various clinical prediction rules have been developed in order to improve diagnosis and decision making in relation to deep vein thrombosis. The purpose of this review is to summarise the available clinical scores and describe their applicability and limitations. Methods A systematic search of PubMed, MEDLINE and EMBASE databases was conducted in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidance using the keywords: clinical score, clinical prediction rule, risk assessment, clinical probability, pretest probability, diagnostic score and medical Subject Heading terms: 'Venous Thromboembolism/diagnosis' OR 'Venous Thrombosis/diagnosis'. Both development and validation studies were eligible for inclusion. Results The search strategy returned a total of 2036 articles, of which 102 articles met a priori criteria for inclusion. Eight different diagnostic scores were identified. The development of these scores differs in respect of the population included (hospital inpatients, hospital outpatients or primary care patients), the exclusion criteria, the inclusion of distal deep vein thrombosis and the use of D-dimer. The reliability and applicability of the scores in the context of specific subgroups (inpatients, cancer patients, elderly patients and those with recurrent deep vein thrombosis) remains controversial. Conclusion Detailed knowledge of the development of the various clinical prediction scores for deep vein thrombosis is essential in understanding the power, generalisability and limitations of these clinical tools.

Vulnerable pregnant women in antenatal practice: Caregiver's perception of workload, associated burden and agreement with objective caseload, and the influence of a structured organisation of antenatal risk management

INTRODUCTION

pregnancy care for vulnerable women is often perceived as a burden by caregivers as vulnerable clients require complex case management, additional time, and more often show adverse perinatal outcomes. Vulnerable clients bring about additional work strain for the caregiver, especially when the workload is high. We define client vulnerability as coexistence of psychopathology, psychosocial problems or substance use, together with features of deprivation. We investigated, as part of a national programme, whether the subjective caregiver's perception of workload and the objective registry-based caseload of vulnerable clients are in agreement, and whether a structured organisation of antenatal risk management reduces the burden associated with the perceived workload, in particular if the objective caseload is high.

METHODS

we combined three data sources: (1) at the unit level (i.e. midwifery practice, obstetric unit) interview data from caregivers, from which we derived a) the (subjective) caregiver's perception of workload, b) the associated burden and c) organisational structure of antenatal risk management, (2) at the unit level perinatal registry data, from which we derived a) unit characteristics and b) (objective) unit specific caseload, and (3) at the individual client level survey data collected during the first antenatal visit, from which the prevalence of vulnerable clients was derived. The study area was the South-West Netherlands (2.5 million inhabitants), containing areas with varying degrees of urbanisation and deprivation.

FINDINGS

sixteen units had complete data on all measures. Generally, subjective workload and objective caseload were only weakly related, the relation being modified by the organisation of antenatal risk management. If the organisational structure of antenatal risk management was low, the experienced burden was high, even if the objective caseload was low. Highly structured antenatal risk management was associated with a medium to low burden.

DISCUSSION

our observational study suggests that even a high caseload can be dealt with by structured antenatal risk management. A change from the current individual case-finding policies towards a more universal screen-like approach may thus benefit both the client and the caregiver.

Ruling Out Pulmonary Embolism in Primary Care: Comparison of the Diagnostic Performance of "Gestalt" and the Wells Rule

PURPOSE

Diagnostic prediction models such as the Wells rule can be used for safely ruling out pulmonary embolism (PE) when it is suspected. A physician's own probability estimate ("gestalt"), however, is commonly used instead. We evaluated the diagnostic performance of both approaches in primary care.

METHODS

Family physicians estimated the probability of PE on a scale of 0% to 100% (gestalt) and calculated the Wells rule score in 598 patients with suspected PE who were thereafter referred to secondary care for definitive testing. We compared the discriminative ability (c statistic) of both approaches. Next, we stratified patients into PE risk categories. For gestalt, a probability of less than 20% plus a negative point-of-care d-dimer test indicated low risk; for the Wells rule, we used a score of 4 or lower plus a negative d-dimer test. We compared sensitivity, specificity, efficiency (percentage of low-risk patients in total cohort), and failure rate (percentage of patients having PE within the low-risk category).

RESULTS

With 3 months of follow-up, 73 patients (12%) were confirmed to have venous thromboembolism (a surrogate for PE at baseline). The c statistic was 0.77 (95% CI, 0.70-0.83) for gestalt and 0.80 (95% CI, 0.75-0.86) for the Wells rule. Gestalt missed 2 out of 152 low-risk patients (failure rate = 1.3%; 95% CI, 0.2%-4.7%) with an efficiency of 25% (95% CI, 22%-29%); the Wells rule missed 4 out of 272 low-risk patients (failure rate = 1.5%; 95% CI, 0.4%-3.7%) with an efficiency of 45% (95% CI, 41%-50%).

CONCLUSIONS

Combined with d-dimer testing, both gestalt using a cutoff of less than 20% and the Wells rule using a score of 4 or lower are safe for ruling out PE in primary care. The Wells rule is more efficient, however, and PE can be ruled out in a larger proportion of suspected cases.

A systematic review of studies comparing diagnostic clinical prediction rules with clinical judgment

Background

Diagnostic clinical prediction rules (CPRs) are developed to improve diagnosis or decrease diagnostic testing. Whether, and in what situations diagnostic CPRs improve upon clinical judgment is unclear.

Methods and Findings

We searched MEDLINE, Embase and CINAHL, with supplementary citation and reference checking for studies comparing CPRs and clinical judgment against a current objective reference standard. We report 1) the proportion of study participants classified as not having disease who hence may avoid further testing and or treatment and 2) the proportion, among those classified as not having disease, who do (missed diagnoses) by both approaches. 31 studies of 13 medical conditions were included, with 46 comparisons between CPRs and clinical judgment. In 2 comparisons (4%), CPRs reduced the proportion of missed diagnoses, but this was offset by classifying a larger proportion of study participants as having disease (more false positives). In 36 comparisons (78%) the proportion of diagnoses missed by CPRs and clinical judgment was similar, and in 9 of these, the CPRs classified a larger proportion of participants as not having disease (fewer false positives). In 8 comparisons (17%) the proportion of diagnoses missed by the CPRs was greater. This was offset by classifying a smaller proportion of participants as having the disease (fewer false positives) in 2 comparisons. There were no comparisons where the CPR missed a smaller proportion of diagnoses than clinical judgment and classified more participants as not having the disease. The design of the included studies allows evaluation of CPRs when their results are applied independently of clinical judgment. The performance of CPRs, when implemented by clinicians as a support to their judgment may be different.

Conclusions

In the limited studies to date, CPRs are rarely superior to clinical judgment and there is generally a trade-off between the proportion classified as not having disease and the proportion of missed diagnoses. Differences between the two methods of judgment are likely the result of different diagnostic thresholds for positivity. Which is the preferred judgment method for a particular clinical condition depends on the relative benefits and harms of true positive and false positive diagnoses.

Accuracy in diagnosing deep and pelvic vein thrombosis in primary care: an analysis of 395 cases seen by 58 primary care physicians

BACKGROUND

Ruling out a deep vein thrombosis (DVT) is difficult in general practice because the clinical manifestations of DVT are nonspecific and more often due to other diseases. The aim of diagnostic screening in primary care must be to rule out a DVT with high accuracy in most patients, so that only those who are likely to have a DVT will undergo further testing. In this study, we tested the accuracy of exclusion of DVT by the combination of a clinical score (the Wells score) with either a bedside D-dimer test or selective compression sonography.

METHOD

This cohort study included all patients who presented to the participating primary care physicians and were suspected of having a DVT on the basis of pre-defined inclusion criteria. To rule out DVT, a Wells score was determined for all patients, and all patients additionally underwent either a D-dimer test or selective compression sonography as required by the clinical algorithm. Patients were seen six weeks later in follow-up to determine whether they had actually had a DVT (gold standard). The negative predictive value (NPV) for the exclusion of DVT in this way was determined, as was the NPV of clinical judgment alone, without knowledge of Wells score or D-dimer results.

RESULTS

395 patients were evaluated by 58 primary care physicians for suspected DVT; 59 were ultimately found to have had a definite DVT, and 9 a probable DVT. Exclusion of DVT with the study protocol had an NPV of 99.0% (95% CI, 96.3 to 99.8)-i.e. only one case of DVT in 100 patients was missed (maximum: 4, minimum: 0)-while clinical judgment alone had an NPV of 95.0% (95% CI, 90.7 to 97.7).

CONCLUSION

We recommend the Wells score combined with either a D-dimer test or selective compression sonography according to the algorithm used in this study for use in primary care to rule out DVT. Clinical judgment alone is less effective.

Aggregating published prediction models with individual participant data: a comparison of different approaches

During the recent decades, interest in prediction models has substantially increased, but approaches to synthesize evidence from previously developed models have failed to keep pace. This causes researchers to ignore potentially useful past evidence when developing a novel prediction model with individual participant data (IPD) from their population of interest. We aimed to evaluate approaches to aggregate previously published prediction models with new data. We consider the situation that models are reported in the literature with predictors similar to those available in an IPD dataset. We adopt a two-stage method and explore three approaches to calculate a synthesis model, hereby relying on the principles of multivariate meta-analysis. The former approach employs a naive pooling strategy, whereas the latter accounts for within-study and between-study covariance. These approaches are applied to a collection of 15 datasets of patients with traumatic brain injury, and to five previously published models for predicting deep venous thrombosis. Here, we illustrated how the generally unrealistic assumption of consistency in the availability of evidence across included studies can be relaxed. Results from the case studies demonstrate that aggregation yields prediction models with an improved discrimination and calibration in a vast majority of scenarios, and result in equivalent performance (compared with the standard approach) in a small minority of situations. The proposed aggregation approaches are particularly useful when few participant data are at hand. Assessing the degree of heterogeneity between IPD and literature findings remains crucial to determine the optimal approach in aggregating previous evidence into new prediction models.

Self-monitoring of listening abilities in normal-hearing children, normal-hearing adults, and children with cochlear implants

BACKGROUND

Self-monitoring has been shown to be an essential skill for various aspects of our lives, including our health, education, and interpersonal relationships. Likewise, the ability to monitor one's speech reception in noisy environments may be a fundamental skill for communication, particularly for those who are often confronted with challenging listening environments, such as students and children with hearing loss.

PURPOSE

The purpose of this project was to determine if normal-hearing children, normal-hearing adults, and children with cochlear implants can monitor their listening ability in noise and recognize when they are not able to perceive spoken messages.

RESEARCH DESIGN

Participants were administered an Objective-Subjective listening task in which their subjective judgments of their ability to understand sentences from the Coordinate Response Measure corpus presented in speech spectrum noise were compared to their objective performance on the same task.

STUDY SAMPLE

Participants included 41 normal-hearing children, 35 normal-hearing adults, and 10 children with cochlear implants.

DATA COLLECTION AND ANALYSIS

On the Objective-Subjective listening task, the level of the masker noise remained constant at 63 dB SPL, while the level of the target sentences varied over a 12 dB range in a block of trials. Psychometric functions, relating proportion correct (Objective condition) and proportion perceived as intelligible (Subjective condition) to target/masker ratio (T/M), were estimated for each participant. Thresholds were defined as the T/M required to produce 51% correct (Objective condition) and 51% perceived as intelligible (Subjective condition). Discrepancy scores between listeners' threshold estimates in the Objective and Subjective conditions served as an index of self-monitoring ability. In addition, the normal-hearing children were administered tests of cognitive skills and academic achievement, and results from these measures were compared to findings on the Objective-Subjective listening task.

RESULTS

Nearly half of the children with normal hearing significantly overestimated their listening in noise ability on the Objective-Subjective listening task, compared to less than 9% of the adults. There was a significant correlation between age and results on the Objective-Subjective task, indicating that the younger children in the sample (age 7-12 yr) tended to overestimate their listening ability more than the adolescents and adults. Among the children with cochlear implants, eight of the 10 participants significantly overestimated their listening ability (as compared to 13 of the 24 normal-hearing children in the same age range). We did not find a significant relationship between results on the Objective-Subjective listening task and performance on the given measures of academic achievement or intelligence.

CONCLUSIONS

Findings from this study suggest that many children with normal hearing and children with cochlear implants often fail to recognize when they encounter conditions in which their listening ability is compromised. These results may have practical implications for classroom learning, particularly for children with hearing loss in mainstream settings.

Search filters for finding prognostic and diagnostic prediction studies in Medline to enhance systematic reviews

Background

The interest in prognostic reviews is increasing, but to properly review existing evidence an accurate search filer for finding prediction research is needed. The aim of this paper was to validate and update two previously introduced search filters for finding prediction research in Medline: the Ingui filter and the Haynes Broad filter.

Methodology/Principal Findings

Based on a hand search of 6 general journals in 2008 we constructed two sets of papers. Set 1 consisted of prediction research papers (n = 71), and set 2 consisted of the remaining papers (n = 1133). Both search filters were validated in two ways, using diagnostic accuracy measures as performance measures. First, we compared studies in set 1 (reference) with studies retrieved by the search strategies as applied in Medline. Second, we compared studies from 4 published systematic reviews (reference) with studies retrieved by the search filter as applied in Medline. Next – using word frequency methods – we constructed an additional search string for finding prediction research. Both search filters were good in identifying clinical prediction models: sensitivity ranged from 0.94 to 1.0 using our hand search as reference, and 0.78 to 0.89 using the systematic reviews as reference. This latter performance measure even increased to around 0.95 (range 0.90 to 0.97) when either search filter was combined with the additional string that we developed. Retrieval rate of explorative prediction research was poor, both using our hand search or our systematic review as reference, and even combined with our additional search string: sensitivity ranged from 0.44 to 0.85.

Conclusions/Significance

Explorative prediction research is difficult to find in Medline, using any of the currently available search filters. Yet, application of either the Ingui filter or the Haynes broad filter results in a very low number missed clinical prediction model studies.