Background
To increase the clinical usefulness of the D‐dimer test in diagnosis of deep vein thrombosis (DVT), two strategies have been proposed: the age‐adjusted, and the clinical pre‐test ...probability (CPTP) adjusted interpretation. However, it is not known which of these strategies is superior.
Objective
To conduct an individual patient data (IPD) meta‐analysis that compares the sensitivity, specificity, negative predictive value (NPV), and utility (the proportion of all patients who have a negative D‐dimer test) when the two strategies are used to interpret D‐dimer results.
Methods
Using an established IPD database, we conducted a meta‐analysis to compare the two strategies. A bivariate random effects regression model was used to estimate and compare the pooled sensitivity and specificity simultaneously. The pooled NPV and utility of the two strategies was compared using a univariate random effects model.
Results
Four studies were eligible for this analysis, with a total of 2554 patients. Overall prevalence of DVT was 12% with substantial heterogeneity between studies (P value < .001). Both strategies have high pooled NPVs (99.8%) with a difference of 0% (95% confidence interval CI: −0.1, 0.1). The difference between the pooled specificity of the CPTP‐adjusted strategy (57.3%) and the age‐adjusted strategy (54.7%) was 2.6% (95% CI: −7.7, 12.8). The CPTP‐adjusted strategy (49.4%) has a marginally greater pooled utility compared with the age‐adjusted approach (47.4%), with a pooled difference of 1.9% (95% CI: −0.1, 3.9).
Conclusions
Both D‐dimer interpretation strategies were associated with a high and similar NPV, and similar utility.
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.
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.
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.
This guideline has been written on behalf of the International Council for Standardisation in Haematology (ICSH) and focuses on two point of care haematology tests used within primary care, namely ...International Normalised Ratio (INR) and D‐dimer. Primary care covers out of hospital settings and can include General Practice (GP), Pharmacy and other non‐hospital settings (although these guidelines would also be applicable to hospital out‐patient settings). The recommendations are based on published data in peer reviewed literature and expert opinion; they should supplement regional requirements, regulations or standards.
Summary
Glucocorticoid treatment increases venous thromboembolism (VTE) risk. Whether this is due to the medication or the underlying disease, or affects the risk of VTE recurrence, has been ...difficult to determine. The aim of our present study was to quantify the risk for first and recurrent VTE associated with oral glucocorticoids use, considering the underlying disease. A total of 2547 patients with VTE from the Multiple Environmental and Genetic Assessment of Risk Factors for Venous Thrombosis (MEGA) study were linked to the Dutch Pharmaceutical Statistics register. The risk of first VTE during periods of exposure with oral glucocorticoids was estimated by the self‐controlled case series method and that of recurrent VTE was examined in a cohort design. The incidence rate ratio (IRR) of first VTE in the period of glucocorticoid treatment was 3·51 95% confidence interval (CI) 2·55–4·80. This IRR was 2·53 (95% CI 1·10–5·72) in the week before treatment started, 5·28 (95% CI 2·89–9·53) in the first 7 days of treatment, remained elevated afterwards and decreased to 1·55 (95% CI 0·85–3·12) after 6 months, as compared to unexposed periods. The hazard ratio for recurrence was 2·72 (95% CI 1·64–4·78) in treatment periods as compared with no treatment. The increased risk of VTE associated with oral glucocorticoid treatment is due to a combined effect of the treatment and the underlying disease, remaining high during the first months of prescription.
This is the 2nd update to the 9th edition of these guidelines. We provide recommendations on 17 PICO (Population, Intervention, Comparator, Outcome) questions, four of which have not been addressed ...previously.
We generate strong and weak recommendations based on high-, moderate-, and low-certainty evidence, using GRADE (Grading of Recommendations, Assessment, Development, and Evaluation) methodology.
The panel generated 29 guidance statements, 13 of which are graded as strong recommendations, covering aspects of antithrombotic management of VTE from initial management through secondary prevention and risk reduction of postthrombotic syndrome. Four new guidance statements have been added that did not appear in the 9th edition (2012) or 1st update (2016). Eight statements have been substantially modified from the 1st update.
New evidence has emerged since 2016 that further informs the standard of care for patients with VTE. Substantial uncertainty remains regarding important management questions, particularly in limited disease and special patient populations.
Background
Current clinical decision rules to exclude deep vein thrombosis (DVT) are underused partly because of their complexity. A simplified rule that can be easily applied would be more appealing ...to use in clinical practice.
Methods
We used individual patient data from prospective diagnostic studies of patients suspected of DVT to develop a new clinical decision rule. The primary outcome was presence of DVT either at initial testing or during follow‐up. DVT was considered safely excluded if the upper 95% confidence interval (CI) of DVT prevalence was <2%.
Results
Four studies and 3368 patients were eligible for this analysis. Overall prevalence of DVT was 17%. In addition to D‐dimer, two variables, calf swelling and DVT as the most likely diagnosis, are included in the new rule. Based on these two variables, two clinical pretest probability (CPTP) groups were defined; low (none of the two items present) and high (at least one of the items present). DVT can be safely excluded in patients with low CPTP with a D‐dimer <500 ng/mL (prevalence = 0.1%; 95% CI, 0.0‐0.8), low CPTP with a D‐dimer between 500 ng/ml and 1000 ng/ml (prevalence = 0.3%; 95% CI, 0.0‐1.7), and D‐dimer <500 ng/ml in patients with high CPTP (prevalence = 0.3%; 95% CI, 0.0‐1.0).
Conclusions
The combination of D‐dimer and Wells items resulted in a simple clinical decision rule with 3 items. The results suggest that the rule can safely exclude DVT. Prospective validation is required.
This is the 2nd update to the 9th edition of these guidelines. We provide recommendations on 17 PICO (Population, Intervention, Comparator, Outcome) questions, four of which have not been addressed ...previously.
We generate strong and weak recommendations based on high-, moderate-, and low-certainty evidence, using GRADE (Grading of Recommendations, Assessment, Development, and Evaluation) methodology.
The panel generated 29 guidance statements, 13 of which are graded as strong recommendations, covering aspects of antithrombotic management of VTE from initial management through secondary prevention and risk reduction of postthrombotic syndrome. Four new guidance statements have been added that did not appear in the 9th edition (2012) or 1st update (2016). Eight statements have been substantially modified from the 1st update.
New evidence has emerged since 2016 that further informs the standard of care for patients with VTE. Substantial uncertainty remains regarding important management questions, particularly in limited disease and special patient populations.
Patients with acute venous thromboembolism (VTE) require anticoagulant therapy to prevent recurrent VTE and death, which exposes them to an inherent increased risk of bleeding. Identification of ...patients at high risk of bleeding, and mitigating this risk, is an essential component of the immediate and long‐term therapeutic management of VTE. The bleeding risk can be estimated by either implicit judgment, weighing individual predictors (clinical variables or biomarkers), or by risk prediction tools developed for this purpose. Management of bleeding risk in clinical practice is, however, far from standardized. International guidelines are contradictory and lack clear and consistent guidance on the optimal management of bleeding risk. This report of the ISTH subcommittee on Predictive and Diagnostic Variables in Thrombotic Disease summarizes the evidence on the prediction of bleeding in VTE patients. We systematically searched the literature and identified 34 original studies evaluating either predictors or risk prediction models for prediction of bleeding risk on anticoagulation in VTE patients. Based on this evidence, we provide recommendations for the standardized management of bleeding risk in VTE patients.
Patients with delayed diagnosis all had PE related symptoms on arrival Normal: diagnosed in ED 87.6% Delayed: 1-48 h after admission 12.4% Smith14 2012 United States (1) 400 Retrospective cohort of ...ED patients presenting with symptoms suggestive of PE Normal: within 12 h of arrival 81.7% Delayed: >12 h after ED arrival 18.3% Torres-Macho11 2013 Spain (3) 436 Retrospective cohort of ED patients presenting with symptoms suggestive of PE Normal: during initial ED visit 66.5% Delayed: during admission to hospital 21.6% Delayed: during subsequent ED visit 11.9% Walen32 2016 Netherlands (1) 261 Retrospective cohort of patients referred from primary care to pulmonology clinic Normal: < 24 h after evaluation 82.0% Delayed: 24-72 h after evaluation 11.9% Delayed: >72 h after evaluation 6.1% Mansella10 2020 Switzerland (1) 226 Retrospective cohort of ED patients with any cardiac or pulmonary testing Normal: within 24 h of arrival 81.5% Delayed: 1-30 days after arrival 19.5% ED=emergency department; PE=pulmonary embolism. A 2007 systematic review identified several elements from patients’ history and clinical examination associated with PE, including dyspnoea (positive likelihood ratio, PLR 1.4), recent surgery (PLR 1.6), prior VTE (PLR 1.5), syncope (PLR 2.4), and active cancer (PLR 1.7).17 While haemoptysis is associated with PE (PLR 3.4) in patients under 50 (but not in older patients18), it is present in less than 10% of cases of PE, limiting its usefulness.1819 Although 40% of patients (range 16-51%) with PE do not have specific risk factors for PE, such as malignancy or recent orthopaedic surgery, 19202122 physicians may perceive the risk of PE to be so low in these patients that risks of further PE evaluation (eg, false positive tests leading to unnecessary anticoagulation) could outweigh any benefit. Since many symptoms of PE are common and can originate from a non-PE cause, they may be attributed to more common conditions. ...a retrospective cohort study of 436 emergency department patients in Spain found that individuals with PE who were initially misdiagnosed often presented with fever, haemoptysis, or a pulmonary infiltrate on chest x ray.1113 Indeed, most missed PEs are misdiagnosed as respiratory infections (43%), heart failure (27%), or exacerbations of chronic obstructive pulmonary disease (COPD) (8%), suggesting that physicians may prematurely close the differential diagnosis in patients with dyspnoea or cough,1123 particularly if they have a history of chronic cardiopulmonary disease such as heart failure and chronic obstructive pulmonary disease.11 For example, in a multicentre cross sectional study of 740 French patients with a history of COPD who were admitted to hospital with worsening respiratory symptoms, combined use of a clinical probability assessment, D-dimer, and CTPA within 48 hours of admission identified PE (that had not been diagnosed at initial presentation) in 6% of patients.24 Diagnostic delay or misdiagnosis also appears to be more common in patients aged 65 or older,2325 in people with altered mental status,9 or in those with cough as the main presenting symptom.12 Additionally, patients with less acute PE presentations, such as gradual onset dyspnoea, may be more likely to experience an initial misdiagnosis because physicians associate PE with a sudden onset of symptoms.23 Although two single centre Italian cohort studies concluded that gradual onset symptoms were present in only 3-6% of PE cases,2026 a larger multicentre cohort from 12 centres in the US found that 45% of patients with PE reported a gradual onset of symptoms.27 In contrast, patients with chest pain,101228 syncope,2329 or recent surgery29 appear less likely to experience delay or misdiagnosis of PE.30 It is unclear whether sex of the patient makes a difference in diagnostic delay. Randomised controlled trials show that anticoagulation therapy after PE is associated with a substantially lower risk of VTE recurrence.3334 Observational data additionally support an association between early treatment and reduced mortality and morbidity related to PE.925 For example, a prospective cohort study of 161 patients in the US who were diagnosed with PE after hospital admission from an emergency department experienced a longer median time to heparin administration (33 hours v 7.7 hours, P<0.001) and had more frequent adverse in-hospital outcomes, including intubation, shock, or death (30% v 8.5%, P=0.01) compared with patients diagnosed with PE in an emergency department before hospital admission.9 Similarly, in a retrospective cohort study of 2250 Australian patients who presented to an emergency department with breathlessness or chest pain, diagnosis of PE in the emergency department was associated with lower in-hospital mortality (odds ratio, OR, 0.31; 95% confidence interval, CI, 0.24
We present an illustrative application of methods that account for covariates in receiver operating characteristic (ROC) curve analysis, using individual patient data on D-dimer testing for excluding ...pulmonary embolism.
Bayesian nonparametric covariate-specific ROC curves were constructed to examine the performance/positivity thresholds in covariate subgroups. Standard ROC curves were constructed. Three scenarios were outlined based on comparison between subgroups and standard ROC curve conclusion: (1) identical distribution/identical performance, (2) different distribution/identical performance, and (3) different distribution/different performance. Scenarios were illustrated using clinical covariates. Covariate-adjusted ROC curves were also constructed.
Age groups had prominent differences in D-dimer concentration, paired with differences in performance (Scenario 3). Different positivity thresholds were required to achieve the same level of sensitivity. D-dimer had identical performance, but different distributions for YEARS algorithm items (Scenario 2), and similar distributions for sex (Scenario 1). For the later covariates, comparable positivity thresholds achieved the same sensitivity. All covariate-adjusted models had AUCs comparable to the standard approach.
Subgroup differences in performance and distribution of results can indicate that the conventional ROC curve is not a fair representation of test performance. Estimating conditional ROC curves can improve the ability to select thresholds with greater applicability.