Introduction
Venous thromboembolism (VTE) comprises deep vein thrombosis (DVT) and pulmonary embolism (PE), with an estimated average incidence rate ranging from 104 to 183 per 100,000 person-years [
1]. The signs and symptoms of VTE at the time of presentation can be notably non-specific [
2], with VTE diagnosed in only 10% of those in whom it is suspected [
3]. Concern has been expressed about the delay in diagnosis, particularly where clinical presentations are often less explicit [
4,
5]. Initiation of treatment is advised urgently to prevent additional morbidity, disability and the risk of death [
6]. Other concerns include extra patient burden when unnecessarily referred to the hospital for additional examinations, and increased healthcare costs incurred through overuse of healthcare resources [
7].
The most commonly accepted diagnostic strategy for VTE is based on the use of a clinical decision rule to determine the pre-test probability, complemented with a measurement of D-dimer, (possibly) followed by a confirmatory imaging assessment (compression ultrasonography [CUS] for DVT, and computed tomography pulmonary angiography [CTPA] for PE) [
8]. The practical utility of the clinical decision rule is the safe identification of patients with suspected VTE who are at low risk, and therefore unlikely to require urgent hospital referral for further investigation. Over the years, several clinical decision rules have been developed for use in primary care, and have been shown to give a comparable performance [
8]. Oudega et al. developed a set of clinical decision rules in 2005 using a study of 1295 patients with suspected DVT in whom patients with suspected PE were excluded [
9]. The performance of this simplified decision rule was compared with the more established Wells score [
10] and subsequently validated in 525 patients by Toll et al. in 2006 [
11]. These derivation and validation studies [
9‐
11] used quantitative laboratory-based D-dimer tests, and revealed that the decision rule could not only rule out a diagnosis of DVT safely, but also reduce the number of unnecessary patient referrals to secondary care for further investigations. Recently, the use of an age-related cut-off value for D-dimer has been proposed, offering an improved diagnostic performance of the decision rule, although this has not been widely adopted to date [
3].
Most practice guidelines advocate the use of a clinical decision rule in combination with the measurement of D-dimer. Recent guidelines have supported the introduction of D-dimer measurement as part of a clinical decision rule when the method of detection has a D-dimer sensitivity of ≥ 95% [
12,
13]. The use of age-related D-dimer cut-off values (where diagnostic yield is increased in low- and medium-risk populations) [
14], and the use of D-dimer where the risk score is below a fixed threshold [
15], have also been recommended. Applying point-of-care testing (POCT) is suggested as an option if operated in collaboration with an accredited laboratory for reasons of quality control [
16] and patient safety [
15].
The National Institute for Health and Care Excellence (NICE) VTE guidelines, updated in March 2020, recommend the use of the Wells score for DVT and PE [
17]. Review of the evidence indicated that if a patient is suspected with DVT, a two-level DVT Wells score should be offered to estimate the clinical probability of DVT [
17,
18]. Patients with a ‘likely’ DVT Wells score (≥ 2 points) should be offered a proximal leg vein ultrasound scan within 4 h and confirmed with a D-dimer test if the scan results are negative. In cases where proximal leg vein ultrasound scan results cannot be obtained within 4 h, a D-dimer test is recommended followed by a proximal leg vein ultrasound within 24 h [
17]. In patients where DVT is suspected and when the two-level DVT Wells score is ‘unlikely’ (≤ 1 point), a D-dimer test is recommended, with a proximal leg vein ultrasound performed if the D-dimer test result is positive [
17].
Similarly, patients with suspected PE should be assessed using a two-level PE Wells score [
17,
19]. Those with a likely PE Wells score (> 4 points) should be offered prompt CTPA or a suitable alternative [
17]. D-dimer testing is recommended for patients with an ‘unlikely’ PE Wells score (≤ 4 points), with imaging diagnosis methods undertaken if the test results are positive [
17,
19].
The revised NICE guidelines support the use of a quantitative D-dimer test, with consideration given to the use of POCT if laboratory facilities are not immediately available [
17,
20]. Furthermore, when using point-of-care (POC) or laboratory D-dimer test, an age-adjusted D-dimer test threshold for people aged over 50 years should be considered [
17,
20].
Analysis of five case studies by the All-Party Parliamentary Thrombosis Group demonstrated that the availability of diagnostic tools, such as the POC D-dimer tests, in primary care led to the redesigning of local community-based DVT pathways [
21]. This not only improved primary care services and patient experience by allowing patients to be treated closer to their homes, but also facilitated substantial cost savings for the local health economy by reducing unnecessary hospital admissions [
21]. This review aimed to summarise the diagnostic performance of POCT for D-dimer when used as an integral part of a clinical decision rule in adult patients presenting with symptoms of VTE in primary care, and to explore some of the practical implications of applying widespread POCT for D-dimer in primary care, including its cost-effectiveness.
Methods
A search on PubMed, Cochrane Library, CINAHL and EMBASE databases was undertaken on November 25, 2019. The search was limited to English language publications over the period 2000–2019; this was considered the main period in which POCT technologies for D-dimer have evolved. Search terms included terms for D-dimer (fibrin fragment D) test, POCT, primary care (general practice), VTE, and their synonyms, including relevant medical subject headings; the same terms were used for each of the databases interrogated, with syntax adjusted according to the database.
A reviewer screened the citations using the title and abstract, and selected the studies for further investigation. Studies were included if they were conducted in a primary care setting; reported a randomised controlled, observational or validation study; and investigated the performance characteristics of POCT for D-dimer, used in combination with a clear decision rule with a reference standard for diagnosis, or the practical implications of applying POCT for D-dimer (e.g., cost-effectiveness). Case studies, study protocols and editorials were excluded, together with studies where other comorbidities were reported in the patient cohort. The references cited by identified systematic reviews were investigated for additional potentially relevant citations. This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.
Discussion
To our knowledge, a total of four systematic reviews have to date investigated the use of D-dimer as part of clinical decision rules in the diagnosis of VTE [
17,
38‐
40]; however, none of these reviews have focussed on the use of D-dimer as part of a clinical decision rule in the primary care setting. Two reviews have explored the use of POCT for D-dimer in the primary care setting, but were limited to patients presenting with suspected PE [
2,
41]. DVT is notably more common than PE in the primary care setting, but reviews on the value of POCT for D-dimer in all VTE are lacking. An understanding of the evidence base specifically for primary care is required as patients tend to present with a larger variety of less specific and less severe clinical presentations. Moreover, rapid decisions in the absence of easily accessible hospital diagnostics are required to prevent ongoing diagnostic uncertainty and poor patient outcomes. There is a lower prevalence of VTE and a lesser severity of the condition in the primary care setting, as the patient is likely to be seen at an earlier stage of the condition. It is, therefore, important to explore the diagnostic performance of POCT for D-dimer in a primary care setting to guide diagnostic decision-making.
The key findings from the two primary studies addressing the research question were that in 49% and 45% of the study populations with suspected DVT [
22] and PE [
23], respectively, the condition could be ruled out, with a false negative rate of 1.4% and 1.5%, respectively (Table
1). Geersing et al. considered the fact that the false negative rate was lower than 2%, and the CI crossed the pre-defined 2% limit as clinically significant. However, it should be remembered that both of the studies employed a qualitative D-dimer assay based on a cut-off of 80 ng/mL. As no formal methods for power calculations of model validation studies exist, further discussion among clinicians is required regarding the proportion that can be considered acceptable as the upper limit [
23]. Using POCT for D-dimer as part of a strategy to exclude DVT in primary care has also been found to be cost-effective compared to secondary care-based strategies [
36].
Strengths and Limitations of This Study
The strength of this review is that the data reported are derived solely from studies in which patients have presented with suspected VTE in the primary care setting. While there are only two studies addressing the primary research questions, they represent contributions from a large number of GP practices. The review also focuses on the use of POCT for D-dimer, as this technology is distinct from laboratory D-dimer assessment methods and may have additional benefits within the primary care setting. There are certain limitations in the conduct of the studies that constitute the evidence base for this review. They include: (i) a heterogeneous population recruited in some of the studies, with potential contamination due to an age profile that may influence the baseline D-dimer value, such as elderly patients, or a previous history of VTE; (ii) variation in the choice of reference method (imaging) for confirming the diagnosis of VTE; (iii) those performing the reference method not necessarily blinded to the results of the index test, which may have resulted in bias; (iv) the reference test (imaging) not being undertaken on all patients; (v) use of a range of clinical decision rules; (vi) recruitment of patients with a previous history of either DVT or PE in some studies; (vii) a reliance on clinical follow-up to detect missed thrombotic disease; and (viii) the majority of the studies were performed in the Dutch healthcare system—this may impact on the generalisability to other healthcare systems. Furthermore, none of the studies included in this review adopted a randomised controlled trial approach with the majority employing a ‘prospective management study’ approach equating to a validation study. However, it may be argued that this approach was able to generate ‘real-world evidence’.
Barriers to the Adoption of POCT in Primary Care
The introduction of POCT into primary care practice has been recognised as challenging in several surveys of GPs [
42‐
45].
There have been numerous studies of analytical performance of POCT for D-dimer in addition to clinical studies. Oude Elferink et al. reported on a clinical evaluation of seven quantitative, laboratory D-dimer tests and one qualitative POC D-dimer test, which highlighted the poor harmonisation in calibration between assays, resulting in the need to use individualised decision threshold values [
46]. This finding has also been observed in results from external quality assurance schemes for laboratory-based, quantitative tests [
47], and there have been calls for suitable reference material [
48]. These observations need to be considered when comparing data and harmonising local clinical guidelines for the use of POCT and laboratory-based services.
Lucassen et al. compared the diagnostic performance of qualitative (POCT in primary care) and quantitative (laboratory-based) assays, which indicated that the false negative rates were 1.5% and 0.4%, respectively. The quantitative test appeared to be safer than the qualitative test; however, the difference was not statistically significant (Table
1) [
27]. Geersing et al., in a diagnostic meta-analysis of four D-dimer assays evaluated in varying outpatient settings, also found that the sensitivity was better using a quantitative assay (0.96 [95% CI 0.91–0.98] and 0.93 [95% CI 0.88–0.97]) compared with qualitative POCT assays (0.87 [95% CI 0.81–0.91] and 0.85 [95% CI 0.78–0.90]) [
49]. Geersing et al. analysed data of 577 consecutive primary care patients with suspected DVT in a primary care diagnostic centre; clinical decision rule was omitted while forming the diagnosis. The group compared the diagnostic performance of four quantitative and one qualitative POCT D-dimer assay against a reference comparator (CUS) [
50]. The quantitative assays employed different sample types (two with citrated plasma, one with citrated blood, and the other with lithium-heparinised blood). The D-dimer cut-off values varied from 196 to 570 µg fibrinogen equivalent units (FEU); laboratory workers performed all of the tests, albeit with no previous experience of POCT. The sensitivities reported with the quantitative tests varied between 94 and 99%, with the specificities between 39 and 62%. The qualitative POCT device demonstrated a sensitivity of 91% and a specificity of 64%. This study also included a questionnaire on the user-friendliness of the devices, conducted with 20 nurses from a thrombosis service. The main issue raised related to the interpretation of the results—judged to be problematic in 25% of the responses regarding the qualitative POCT device, and 0–5% in the case of the quantitative assays.
Our findings are similar to those reported by the evidence review supporting the revised NICE guideline in patients with suspected DVT, with comparable high sensitivities between POCT and laboratory-based D-dimer tests, with the evidence considered to be of low quality [
20]. Evidence supporting the revised NICE guidelines in patients with suspected PE suggests that POCT D-dimer devices offer lower sensitivity (88% [0.84–0.91] vs 93% [0.91–0.94]), but higher specificity (63% [0.57–0.69] vs 48% [0.43–0.53]) compared with laboratory-based tests [
20]. Crucially for the study aim of this NICE review, these results were based on patient presentations in a range of clinical settings, as well as a range of POCT technologies, with varying performance characteristics [
20]; thus, no conclusions relating to the use of POCT in primary care could specifically be drawn.
The majority of the studies of POCT for D-dimer in primary care utilised qualitative POCT systems [
22,
23]. Recently, a comparison of the analytical performance of five quantitative POCT assays with a hospital reference test showed that four performed analytically well with a set of 238 plasma samples from patients clinically suspected of VTE in general practice. Most devices were considered easy to use in a primary care setting [
51]. In addition to the potentially superior diagnostic performance offered by the quantitative POCT assays enabling closer harmonisation with the local laboratory D-dimer assay, such tests would allow the use of age-related cut-off values for the older population. The application of an age-adjusted D-dimer threshold is regarded as improving the management of patients with suspected VTE, especially with access to direct oral anticoagulants [
52]. We would expect clinical performance of decision rules to be improved with the use of quantitative D-dimer tests, as well as age-adjusted cut-off points; this can only be proved with further studies.
Implementation Plan
Implementation of POCT is recognised as a challenge and has been evaluated in a POCT-facilitated diagnostic pathway for DVT in primary care involving 450 GPs, with the aid of educational outreach visits, financial reimbursements and periodical newsletters [
53]. The researchers addressed ‘acceptability’, ‘feasibility’, ‘fidelity’ and ‘sustainability’. The study showed an increase in the use of the pathway from 42% to an expected continuation of use of 91%. Regarding clinical outcomes, 54% of patients were not referred to the hospital, missing six cases of DVT (1.8% [95% CI 0.7–3.9%]), reflecting similar observations of other studies. However, it was also noted that during the implementation study, the pathway guidelines were found to have not been used correctly for 32% of patients. The researchers concluded that the study had demonstrated evidence of high acceptability, feasibility and expected sustainability. Interestingly, the implementation strategy described by Kingma et al. [
53] included reimbursement, addressing one of the concerns expressed in the adoption of POCT in primary care [
42,
44].
Investment and Disinvestment
A total of five case studies reported the redesigning of the local DVT pathway by transferring the diagnosis and treatment of non-complex DVT patients into primary care to avoid unnecessary hospital admissions; four of these five case studies included measurements using POCT D-dimer devices [
21]. A consistent finding in the case studies was the reduction in the number of patients referred to the hospital for diagnosis or anticoagulation management, and greater patient satisfaction. There was also a significant reduction in cost to local healthcare services, despite the need to purchase diagnostic equipment. Of note, very limited quantitative data were provided in these studies [
21].
Evidence from a cost-consequence model developed in support of NICE guidelines indicated that POCT for D-dimer results in a small, statistically significant increase (4 per 1000 people) in the number of false negative results and a large, statistically significant decrease (138 per 1000) in the number of false positive results. Excluding primary care costs, the overall POCT strategy was found to be less costly than laboratory testing (−£1331 [95% credible interval, −£10,777 to £8721]). When primary care costs are included, the overall POCT strategy becomes significantly less costly (−£20,166 [95% credible interval, −£30,296 to −£9527]) [
20]. Clearly this data will need to be updated when studies employing a quantitative POCT tests have been completed.
Therefore, evidence of the potential cost-effectiveness of adopting POCT for D-dimer rests on demonstrating: (i) a reduction in the number of patients with suspected VTE requiring referral to hospital for further imaging investigations; and (ii) a reduction in the number of patients admitted to hospital with complications of VTE due to a failure to make a timely diagnosis.
Conclusions
The evidence, albeit limited in terms of the number of studies, indicates that POCT for D-dimer can be employed in the primary care setting to reliably guide diagnostic and management strategies for patients presenting with suspected VTE, reducing the time to diagnosis and treatment. A good D-dimer POCT device used to rule out a diagnosis could safely reduce DVT and PE referrals to hospital. Some of these cost savings can be used to invest in POCT technology and infrastructure. Evidence suggests that a quantitative POCT assay is the preferred technology choice; additionally, this will enable the use of age-related cut-off values. The choice of device, and its associated cut-off values, should be harmonised with the service provided by the laboratory serving the local hospital. Furthermore, the operators of the D-dimer testing system in primary care should participate in regular quality control and quality assurance programmes, and collaborate with POCT experts from an accredited laboratory.
Acknowledgements
The literature database search was undertaken by Sushmita Roy Nawathe and Maria Haughton (both of imc integrated medical communications, UK), and sponsored by LumiraDx.