Article Text

Download PDFPDF

Overutilisation of imaging studies for diagnosis of pulmonary embolism: are we following the guidelines?
  1. Prajwal Dhakal1,2,
  2. Mian Harris Iftikhar3,
  3. Ling Wang3,
  4. Varunsiri Atti3,
  5. Sagar Panthi3,
  6. Xiao Ling3,
  7. Mark T P Mujer3,
  8. Om Dawani3,
  9. Manoj P Rai3,
  10. Shilpa Tatineni3,
  11. Shiva Shrotriya3,
  12. Supratik Rayamajhi3
  1. 1 Department of Internal Medicine, Division of Oncology and Hematology, University of Nebraska Medical Center, Omaha, Nebraska, USA
  2. 2 Fred and Pamela Buffett Cancer Center, University of Nebraska Medical Center, Omaha, Nebraska, USA
  3. 3 Department of Medicine, Michigan State University, East Lansing, Michigan, USA
  1. Correspondence to Dr Manoj P Rai, Department of Medicine, Michigan State University, East Lansing, MI 48824, USA; manojrai029{at}


Objective To evaluate if imaging studies such as CT pulmonary angiography (CTPA) or ventilation–perfusion (V/Q) scan are ordered according to the current guidelines for the diagnosis of pulmonary embolism (PE).

Methods We performed a retrospective observational cohort study in all adult patients who presented to the Sparrow Hospital Emergency Department from January 2014 to December 2016 and underwent CTPA or V/Q scan. We calculated the Wells’ score retrospectively, and d-dimer values were used to determine if the imaging study was justified.

Results A total of 8449 patients underwent CTPA (93%) or V/Q scan (7%), among which 142 (1.7%) patients were diagnosed with PE. The Wells’ criteria showed low probabilities for PE in 96 % and intermediate or high probabilities in 4 % of total patients. Modified Wells’ criteria demonstrated PE unlikely in 99.6 % and PE likely in 0.4 % of total patients. D-dimer was obtained in only 37 % of patients who were unlikely to have a PE or had a low score on Wells’ criteria. Despite a low or unlikely Wells’ criteria score and normal d-dimer levels, 260 patients underwent imaging studies, and none were diagnosed with PE.

Conclusion More than 99 % of CTPA or V/Q scans were negative in our study. This suggests extraordinary overutilisation of the imaging methods. D-dimer, recommended in patients with low to moderate risk, was ordered in only one-third of patients. Much greater emphasis of current guidelines is needed to avoid inappropriate utilisation of resources without missing diagnosis of PE.

  • Pulmonary embolism
  • Computed tomography pulmonary angiography
  • Ventilation–perfusion scan
  • Wells’ criteria
  • D-dimer

Statistics from


Pulmonary embolism (PE), a common and potentially fatal condition, causes approximately 300 000 annual deaths in the USA.1 The diagnosis of PE is often challenging due to its varied presentations. Current guidelines recommend calculation of the pretest probability using a scoring system such as Wells’ criteria (table 1) before initiating diagnostic workup for PE.2 3 Either the conventional three-tier model or the modified two-tier scoring may be used.4 If the scoring is low or unlikely, high sensitivity d-dimer test is done. If d-dimer is elevated, an imaging study should be ordered. Intermediate scores in the three-tier system may undergo either d-dimer or imaging studies based on clinical suspicion. If the pretest probability is high or PE likely, the next step is imaging studies without checking d-dimer levels. Low probability on Wells’ criteria and normal d-dimer assays has been shown to safely exclude PE in most cases.5 6 However, the imaging studies such as CT pulmonary angiography (CTPA) or ventilation–perfusion (V/Q) scan are often ordered in clinical practice to avoid missing a diagnosis of PE.3 7 In this context, we evaluated if CTPA or V/Q scan were ordered appropriately while investigating patients with suspected acute PE.

Table 1

Wells’ criteria for PE


Study sample and characteristics

We performed a retrospective observational cohort study of all patients who presented to the Sparrow Hospital Emergency Department (ED) from January 2014 to December 2016. All adult patients who underwent an imaging study for diagnosis of PE (CTPA or V/Q scan) were included. We excluded patients who underwent CTPA to rule out other medical conditions including aortic dissection.

Calculation of Well’s score and use of d-dimer values

A retrospective chart review was done on all patients to look for documented Wells’ score in ED provider notes. If no specific score was mentioned or calculated, we applied the Wells’ criteria, and the score was calculated retrospectively with the help of the notes and data at the time of presentation. D-dimer data were collected on all assays done. D-dimer values, reported in μg/mL, were considered positive if greater than 0.5 µg/mL. We used Wells’ criteria scores and d-dimer values to determine if an imaging study was justified. Depending on clinical suspicion, we accepted as adherence to guidelines that an imaging study may be ordered for the patients on an intermediate score on the traditional three-tier model.3 We rated non-adherence to guidelines when an imaging study was ordered without checking d-dimer or following negative d-dimer in patients with a low score on the three-tier model or PE unlikely in the two-tier model.

Statistical analysis

Continuous data are presented as medians and range of scores; categorical data are presented in numbers and proportions in each group. Multivariate logistic regression was performed to predict the probability of PE outcome using Wells’ score. Statistical significance was set at p<0.05. Statistical tests were performed using SAS V.9.4 software (SAS).


We included a total of 8449 patients in the study. The median age was 57 years (range 18–107 years); 5244 (62.1%) were females (table 2). CTPA was done in 7837 (92.8%), and V/Q scan was done in 612 (7.2%) patients (table 3). Wells’ criteria showed low probabilities for PE in 8057 (95.4%) and intermediate or high probabilities in 392 (4.6%) patients. An unlikely modified Wells’ score for PE was present in 8414 (99.6%), while a likely Wells’ score was present in only 35 (0.4%) patients. Among those with low Wells’ score, d-dimer was done in 3108 (36.8%) patients, out of which 2849 (92%) had an elevated d-dimer. PE was not diagnosed in any of the patients with low Wells’ score regardless of d-dimer values (figure 1). Among those with intermediate or high Wells’ score, 142 (36.2%) were diagnosed with PE. D-dimer was done in 3192 (37.8%) patients who were unlikely to have a PE on modified Wells’ scores and on 2932 (91.8%) patients with likely PE (figure 2). Among those who were unlikely to have a PE but had high d-dimer, PE was diagnosed in 34 (1%) patients. No patient who was unlikely to have a PE and had a normal d-dimer was diagnosed with PE. Among those with likely PE scores, 27 (77.1%) were diagnosed with PE.

Figure 1

D-dimer tests based on Wells’ criteria. PE, pulmonary embolism.

Figure 2

D-dimer tests based on modified Wells’ criteria. PE, pulmonary embolism.

Table 2

Patients’ demographics and Wells’ criteria comparison between patients with PE and those without PE

Table 3

Imaging studies done and Well’s score in patients with and without PE

A total of 142 (1.7% of total) patients were diagnosed with PE: 128 (90% of PE positive) with CTPA and 14 (10%) with V/Q scan. Wells’ criteria showed intermediate to high pretest probability for PE in 100% of patients diagnosed with PE. Modified Wells’ scores likely for PE were present in 19% (n=27) of patients and unlikely Wells’ scores were found in 81% (n=115) of patients. D-dimer assay was done in 37 (26%) of total PE-positive patients and was elevated in all instances. Logistic regression was conducted to determine if there was a significant relationship between PE status and Wells’ score for the entire sample (n=8504). Results revealed significant association (p<0.0001) and showed that if Wells’ scores increased by one point, the odds of having PE is predicted to be 18-fold higher (95% CI, 12.4 to 26.6). A receiver-operating characteristic (ROC) curve was also computed to assess the accuracy of the ability of Well’s score to predict PE. The area under the ROC curve was 0.99 (figure 3). Out of 31 PE-positive patients who underwent lower extremity duplex ultrasound, 68% (n=21) were positive for deep vein thrombosis. Upper extremity deep vein thrombosis was diagnosed in one PE-positive patient.

Figure 3

ROC curve for wells’ criteria. ROC, receiver-operating characteristic.


In our study, less than 1% of people who underwent imaging studies were diagnosed with PE. This is considerably lower compared with other studies which reported a diagnostic yield of 7%–33%.7–15 Failure to adhere to guidelines for using Wells’ criteria and d-dimer studies were found to be the major reason for overutilisation of imaging studies.1 Since PE can be difficult to diagnose and may cause fatal complications, the relative ease of obtaining the imaging studies and the additional information they offer to rule in or rule out various cardiopulmonary conditions may lead to their overuse in clinical practice.7 8 16 In addition, time constraints due to an increased focus of ED's on turnaround time and an increased patient load may not allow the ED physicians to follow the clinical guidelines to risk stratify patients.

More than 95% of our study population had a low pretest probability on Wells’ criteria, and none were diagnosed with PE. Similarly, no patients with normal levels of d-dimer had PE. Thus, as validated in previous studies, use of Wells’ scores followed by a negative d-dimer is a safe way to rule out PE.4 17 18 This applies to both, individuals with a history of PE and those who are hospitalised.19–21 Since d-dimer has limited specificity and may be elevated in a number of conditions, it is important to order d-dimer in the appropriate population as recommended by guidelines.21 22 However, more than half of our patients with a low probability of PE were not tested for d-dimer before an imaging study was ordered. A number of these imaging studies could have been avoided if clinical guidelines were followed, and d-dimer levels were checked. Additionally, imaging studies were ordered in 260 patients despite normal d-dimer levels. Since d-dimer is a highly sensitive test, imaging could have been avoided. D-dimer was also ordered in one-third of high probability cases which is another deviation from current recommended practice.

The reports of non-adherence to guidelines has been a common theme in previous studies, with guidelines followed accurately in only 46%–73% of cases.11–13 15 23 Inappropriate use of imaging studies is associated with many complications along with increased healthcare costs.24 25 Different allergic and non-allergic adverse effects related to intravenous contrast with CTPA may occur.24 25 Contrast-induced nephropathy, one of the serious complications, has been reported in more than 10% of cases. In addition, there is exposure to ionising radiation and increased risk of cancer secondary to radiation exposure.26 V/Q scan, although associated with fewer adverse effects than CTPA, is more expensive and may not be available in all health centres. Moreover, it is not as specific as CTPA in PE diagnosis.27 In our study, no calculations were done, but the overuse of these imaging modalities likely increased the healthcare costs. Additionally, although we did not look at any of the complications associated with the imaging, even one occurrence would be an undesired event which could have been avoided with the appropriate use of the guidelines. Overutilisation of imaging studies also leads to incidental findings, which may or may not be clinically relevant and may lead to further workup.

Different ways have been studied in an attempt to decrease unnecessary imaging to diagnose PE. Use of pulmonary embolism rule-out criteria alone or in conjunction with low probability on Well’s score has been used safely to exclude PE.2 8 28 Since d-dimer levels increase with age, adjusting the d-dimer cut-off levels for age more than 50 years have been studied which can increase the sensitivity further.29 Other measures such as increasing awareness regarding current guidelines for PE diagnosis and mandatory documentation of pretest probability score and d-dimer scores may also be used.

Our study should be viewed in the context of the following limitations. The retrospective observational nature of the study carries an inherent risk of selection bias. Wells’ scores were calculated retrospectively in most of the cases, which might not replicate the clinical suspicion of the physician at the exact time of presentation. Age-adjusted cut-offs for d-dimer, which is shown to be more sensitive, were not used. Despite such limitations, our large study population represents a real-world clinical scenario.

To conclude, our study shows overuse of imaging studies for the diagnosis of PE. There is a significant deviation from guidelines producing an extremely low diagnostic yield. This translates into a waste of resources and also potential harm to patients due to unnecessary radiation and contrast exposure. Strict implementation of validated guidelines should help in the appropriate utilisation of imaging studies without missing any diagnosis of PE.

Main messages

  • CT pulmonary angiogram (CTPA) and ventilation–perfusion (V/Q) scan are overused to rule out or for the diagnosis of pulmonary embolism (PE).

  • There is a need for improvement in the adherence to guidelines, use of pretest probability scoring and d-dimer for risk stratification of patients requiring imaging studies (V/Q scan or CTPA) to rule out PE.

  • Improvement in adherence to guidelines will reduce the reliance and overutilisation of imaging studies.

Current research questions

  • What are the barriers leading to inadequate adherence to PE diagnosis guidelines?

  • Are the clinicians overusing imaging studies to rule out PE due to the fear of missing PE and possible litigation?

  • What are the chances of missing PE with strict adherence to current guidelines on PE?


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.


  • Presented at This study was presented as an abstract, after preliminary data analysis, at American College of Physicians, Michigan Chapter Residents day, 2018 and Mid-Michigan Research Day, 2018.

  • Correction notice This article has been corrected since it was published Online First. There was an error to the percentage of patients who were diagnosed with PE which is 142 out of 8449. This was corrected in the Abstract and the main text.

  • Contributors All the authors contributed to the paper. PD: planned the study and contributed to majority of the manuscript. MHI: wrote the manuscript including the introduction and the discussion. LW: assisted us in this study with statistics which included creating graphs. VA, XL, SS: assisted with data collection and with writing discussion. SP, MM, OD, ST: assisted with data collection. MPR: assisted with editing the manuscript, submission and revision of the manuscript. SR: is the senior author in the paper and he assisted with writing discussion and editing the manuscript.

  • Funding This research was funded by Resident-led research mini-grant from Graduate Medical Education, Inc, Sparrow Hospital.

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Ethics approval Our study was approved by the institutional review boards at Michigan State University and Sparrow Hospital (IRB # x17-460e).

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.