Abstract
Objective: Ruptured abdominal aortic aneurysms (AAA) carry a mortality rate as high as 80% . Early detection through a screening ultrasound can lead to a large mortality reduction. Point-of-care ultrasound (POCUS) has preliminary data suggesting it is as accurate as hospital-based ultrasounds performed by a sonographer. This validation study investigated the relative concordance of family physicians using POCUS to determine aortic diameter compared with hospital-based ultrasound studies.
Methods: The study was a cross-sectional, multi-observation study conducted at 3 office practices. Five family physicians with varying degrees of training and experience utilized various ultrasound machines to measure maximal aortic diameter at the proximal, mid, and distal aorta. Hospital-based ultrasound or Computed Tomography (CT) served as the validation scan. Pairwise comparisons were made, with statistical testing for difference using the T-TEST command with the PAIRS subcommand.
Results: Forty-four independent observations were completed by the 5 physicians on the 18 patients (n = 18). The mean difference between the POCUS and validation scans was 0.2 cm (95% CI −1.10 to 0.40). The family physicians generally underestimated the aortic diameter. The proximal aorta had the largest mean difference in aortic size (0.23 cm; P = .003). Type of ultrasound device used, the width of the largest aortic segment, and low patient body mass index had significant relations.
Conclusion: This small study found data suggesting that family physicians with variable POCUS experience can accurately perform AAA screening in the ambulatory setting with either handheld or cart-based POCUS machines.
- Abdominal Aortic Aneurysm
- Cross-Sectional Studies
- Family Medicine
- Family Physicians
- POCUS
- Point-of-Care Systems
- Screening
- Technology
- Ultrasound
Introduction
Abdominal aortic aneurysm (AAA) refers to an enlargement of the abdominal aorta, which occurs slowly over many years, sometimes culminating in sudden rupture – a catastrophic event with mortality as high as 80%.1 Reported AAA prevalence varies from 1.9% to 18.5% among men and from 0.1% to 4.2% among women, with risk factors other than sex including a smoking history, cardiovascular disease, and a family history of AAA.2 One-time ultrasound screening of high-risk individuals has been demonstrated to reduce AAA-related mortality by between 50% and 65%.1,3,4 Consequently the US Preventive Services Task Force (USPSTF) recommends that all men with any smoking history be screened once between ages 65 and 75, and that persons with aortic diameters ≥3.0 cm be followed periodically, with the frequency of rescreening depending on the size of the aneurysm.5–7 This recommendation has been incorporated into Medicare guidelines, which recommend one-time screening of men who have smoked 100 or more cigarettes in their life as part of the “Welcome to Medicare” visit.8 Elective repair of a AAA is recommended when the size has reached 5.5 cm.5
Unfortunately, implementation of the USPSTF recommendations has been limited. In one US academic medical center, for example, only 13% of eligible patients were screened during the 2 years after reaching age 65, with screening more common among patients who were White and had a primary care physician.9 Just having a primary care clinician seems to not be a strong factor for achieving the needed screening, however, as a study from a network of primary care practices documented that only 9.2% of eligible patients obtained the study.10 When national Medicare data were examined, reported rates were even lower – less than 1% per year after introduction of AAA screening as a Medicare benefit11; and low income was identified as an additional risk factor for nonscreening.11 These rates are considerably lower than for other recommended screening procedures, such as for colon cancer. Additional factors implicated as causes of low screening include lack of physician awareness, out-of-pocket insurance copays, the limited window of screening eligibility, and physician reimbursement issues.12
An important contributor to low screening uptake is the fact that obtaining the test requires a trip to a hospital radiology department(particularly burdensome for people living in rural areas) creating a disconnect between the office ordering the test and its implementation. To address these issues, 2 initiatives from the British National Health Service sent mobile ultrasound units to primary care sites. Both studies achieved high screening rates (86%–91%),13,14 but the effort was costly, and no such outreach exists in the United States.
Growth of point-of-care ultrasound (POCUS), and its increasing adaptation by family medicine practitioners and residency training programs,15 offers an exciting potential new avenue for increasing AAA screening,16 particularly now that low-cost, higher-sensitivity portable devices are available.17–19 Already, use of POCUS has become a standard in emergency medicine, where a meta-analysis of 9 studies demonstrated a pooled sensitivity of 98.3% and specificity of 99.8% for identifying AAA.20 It is not surprising, then, that calls have been made to incorporate POCUS screening for AAA into office practice, and preliminary studies in Canada and Europe have been encouraging in terms of validity of family physician-conducted studies.21–23
Many steps must be taken, however, for primary care POCUS screening for AAA to become a routine standard of care. Family physician testing must be demonstrated to be noninferior to formal ultrasonography, and the requirements for accurate testing must be identified in terms of the amount of training needed and the type of machine used. In addition, issues of coding, billing, and insurance reimbursement will need to be addressed. To shed light on some of these issues, we conducted a preliminary validation study involving 5 family physicians with different levels of experience, using several different POCUS machines, and studied the relative concordance of their findings with hospital-based ultrasound studies.
Methods
Study Design
The study was a cross-sectional, multi-observation study of abdominal aortic diameter measurement by family physicians on patient volunteers meeting USPSTF criteria for AAA screening. Study settings were 3 office practices affiliated with an academic department of family medicine, and 5 family physicians with varying degrees of training and experience. Data for these analyses were collected on 18 study participants, with up to 4 independent ultrasound measurements obtained on each participant, resulting in 46 separate ultrasound studies. Participant demographic and historic data were collected by medical student research assistants; additional data and confirmatory hospital-based ultrasound findings were abstracted from participant medical records. Study procedures were approved by the Institutional Review Board of the University of North Carolina at Chapel Hill (IRB No. 2321007).
Participants
Patient participants were identified by screening the electronic health records of men aged 65 to 74 who had ever smoked, had no record of ever having had their aortic size determined (either through screening or incidentally as part of scanning done for other reasons), and were regular patients at 1 of 3 clinics – a university-based clinic and 2 rural federally qualified health centers. To assure that the sample included individuals with abnormal aortic diameters, a subsample of 7 patients were recruited with a known abdominal aortic diameter greater than 3.0 centimeters, and aortic diameter (known/unknown) status of all participants was blinded to the family physician ultrasonographers. Individuals were excluded who did not speak English, had been recently hospitalized, resided in a long-term care facility, had a severe chronic mental disorder, or were actively being treated for cancer. Each potential subject’s primary care clinician was contacted for permission to approach potential participants; those not excluded for any of the above reasons were approached by telephone. A small grant provided $100 to reimburse each of up to 25 participants, a number which was reduced as copays for certain insurance plans required additional payment to maintain a $100 participation incentive. In total, 19 individuals were studied and 18 yielded complete data for analysis.
Five family physicians conducted the ultrasound procedures, measuring the largest diameter (outer wall to outer wall [OTO]) at each of 3 sites (proximal, middle, and distal) in each participant’s abdominal aorta. Two of the physicians were highly experienced in POCUS, each of whom reported between 300 to 450 hours of formal or self-study training and approximately 500 abdominal ultrasounds. The other 3 were less experienced, reporting 40 to 70 hours of training and between 5 and 40 abdominal scans each.
Data Collection
Data collection took place in 3 half-day clinics at the family medicine practices where the patients obtained primary care, in a dedicated space when the family physician ultrasonagraphers were not on clinical duty. Data collected on or associated with each observation included: a) participant demographic information (age, height, weight, smoking history, cardiovascular disease status, hours since last meal, and clinic location); b) FP ultrasonagrapher data (years since residency, amount of ultrasound training, POCUS experience); c) ultrasound measurements (date, machine used, and largest OTO diameter in millimeters of each abdominal aortic segment (proximal, middle, distal), and FP ultrasonagrapher impression of ease or difficulty of scan (and reasons for difficulty); and d) validation data (type – ultrasound or CT scan, date, largest aortic diameter measurement reported). To blind the FP ultrasonagraphers to data on the study participants, including which were patients for screening and which had known enlarged aortas, all patient identification, recruitment, orientation, interviews, and data recording were done by medical student research assistants, and patient participants were asked to not share any personal data with the FP ultrasonagraphers until after the procedure. Laptop-based devices used included GE Venue Go, GE Logiq E, and Sonosite Edge. The handheld ultrasound device used was a first-generation Butterfly probe.
Statistical Analyses
Analyses were conducted using SPSS Statistics, Version 29.0.2.0. There were 44 observations with FP sonography results, representing 19 unique patients. Of these observations, 1 patient did not have confirmatory measurements and thus the 5 FP ultrasonographer exams on this patient were removed, leaving 39 records for which comparisons between POCUS and confirmatory measurements could be made. For the overall comparisons, we used the maximum diameter reported across proximal, middle, and distal aorta segments for both the POCUS measurements and the confirmatory measurements. Pairwise comparisons were made, with statistical testing for difference using the T-TEST command with the PAIRS subcommand.
Results
Characteristics of the patient participants, physician participants, and scanning methods are summarized in Table 1. Eighteen patients were recruited, studied, and provided validation data for these analyses. All were men over 65 years of age with significant smoking histories (mean of 17.7 pack-years). The majority had hypertension (56%); and Body Mass Index (BMI) status was evenly divided into obese (“≥30” and “nonobese”). The 5 FP ultrasonagraphers were all faculty with recent training (mean, 3.4 years since completing residency). For analytic purposes they were divided into those with more experience (2 individuals, each with 376 average hours of training and approximately 500 prior abdominal ultrasound scans) and those with less experience (3 individuals, averaging 107 hours of training and 20 prior abdominal scans). Of the 44 independent observations by these 5 physicians on the 18 patients, 39 were done with a laptop device and 5 with a handheld device; of the hospital-based validation scans, 33 were ultrasound scans and 6 were computerized tomographic scans.
Description of the Study Sample
Table 2 displays the observed differences between the POCUS results obtained by the family physicians and the validation scans. Across all observations, the mean difference between the POCUS and validation scans was 0.2 centimeters, with the family physicians generally underestimating the aortic diameter. Taking 95% of observations by dropping the lowest and highest difference values, the difference range was between −1.10 and 0.40 centimeters.
Comparison of Family Physician and Hospital-Based Estimates of Abdominal Aorta Size
Pairwise comparisons (Table 2) identify a number of findings that were statistically significant at P ≤ .05. The mean difference in estimating proximal aortic size (0.23 cm; P = .003) was larger than that of the other aortic segments. Other factors showing significant relationships were the type of ultrasound device used, the width of the largest aortic segment, and low patient body mass index.
Other comparisons generated less clear differences. The mean difference between screening and validation studies varied little between the relatively experienced physicians (0.21 cm) and the relatively inexperienced ones (0.17 cm); the former was statistically significant, but the larger sample size may have been responsible. Similar findings are present regarding the type of validation scan and the time between the primary care scan and the validation scan (Table 2).
Discussion
This preliminary study of differences between family physician and hospital-based evaluations of abdominal aortic size has resulted in several intriguing if preliminary findings. Overall, 95% of family physician estimates were within 0.2 centimeters of the validation study, with the median difference less than 0.2 centimeters, and the largest error and variation noted in observation of the proximal abdominal aorta. Interestingly, modest training and experience was not associated with poorer ability to estimate aortic size, and neither obesity nor reported difficulty performing the scan was associated with poorer concordance between family physician observations and those of hospital-based scans. Neither did the type of hospital-based scan (ultrasound or CT), although not surprisingly a longer lag time between the family medicine scan and the validation scan was associated with a trend for greater discordance. Finally, though data are sparse, available evidence suggests that handheld devices may perform as well as laptop-based models.
Perception of difficulty learning POCUS has been a barrier to its integration into primary care. However, our study is consistent with other literature in suggesting that scanning an aorta to screen for AAA does not require extensive training and experience. Bonnafy et al, for example, found that a 9-hour training and practical hands-on program could teach medical students with no previous ultrasound experience to reliably measure aortic diameter.24 Another medical student-based study found that 8 hours of training resulted in no statistical difference between the students and professional ultrasound technologists.25 Similar results have been found using internal medicine residents.26 Thus, training time should not be a barrier to wide dissemination of primary-care based screening.
The mean difference in our study between the family physician scan and the validation scan was 0.2 cm (95% CI −1.10 to 0.40). This finding was similar to the report by Mai et al, who had a mean difference of 0.26 cm among second-year medical students,27 and Blois, who found an absolute difference of 0.2 cm (95% CI, 0.15 to 0.25 cm) among rural family medicine physicians.21 Similar interobserver differences have been reported among radiologists reading ultrasound studies −0.2 cm or less in 2/3 of cases in one study28; a mean difference of 0.16 cm in another.29 Several human factors come into play that can impact the reproducibility and accuracy of a scan. For example, 2 experienced sonographers may scan the same aorta and measure at a slightly different location, have a different probe angle that is not equally perpendicular to the vessel, or have the measuring calibers slightly off, all which contribute to error.24 The proximal aorta view can also have significant variability as seen in this study, as the inferior costal margin and bowel gas from the stomach and duodenum can limit how proximal the measurement is obtained. While it is important to obtain accurate views in each segment, it is helpful to consider that approximately 85% of aneurysms occur distal to the renal arteries.30 However, given these data on variability and the small number and size of trials in the literature, to prevent missing some smaller AAAs, a lower screening cutoff, such as 2.7 or 2.8 cm, instead of the 3.0 cm used in hospital-based screening, could be considered for primary care.
Findings from abdominal CT scans were used as confirmatory data for many of the participants in the “known AAA” group – testimony to the fact that so few screening scans are done in our study settings that many known AAA’s are detected serendipitously, usually during an emergency department evaluation for another problem. For purposes of our study, we accepted a lag time between primary care screening and the confirmatory scans of up to 1 year, which resulted in significant, though small, increases in scan discordance with increased time between scans (Table 2). The modest size of this discordance is consistent with the fact that aortic enlargement generally proceeds slowly, with AAA mean growth averaging 0.26 cm per year.31 Use of CT data led to one other potential source of variation in our results, as ultrasound has been demonstrated to underestimate AAA size compared with CT in a range of 0.01–0.5 centimeters.32–36 While these sources of variation only modestly affected our findings, future studies should avoid these limitations by further minimizing the lag time between POCUS exams and the hospital-based ultrasound scans, and either exclusively using ultrasound for confirmation or having adequate sample sizes to independently evaluate both ultrasound and CT confirmation.
Limitations of the study include the small study sample and the limited patient diversity, both of which reduce the generalizability of the findings. To address these issues, effort was made to diversify the sample by adding participants with known aortic enlargement, by recruiting patients from both a university-based clinic and 2 rural community health centers, and by including family physician ultrasonagraphers with varied POCUS experience. Other limitations involve the time intervals between screening and validation, which was evaluated analytically (Table 2), and the fact that some subjects with known AAA had validation by CT scan, whereas the remainder were by ultrasound.
Within the above limitations, this study suggests that family physicians with variable ultrasound experience can adequately perform AAA screening in the ambulatory setting. The level of training was not associated with reduced accuracy, and very preliminary results suggest that handheld devices may perform as well as laptop-based models and should be targeted in future research.37,38 If AAA screening in primary care could become standard practice, this more patient-centered approach could increase AAA detection and decrease mortality.3 It also should be feasible even in a busy practice, as scanning can be performed in under 5 minutes.21,39–42 One barrier to primary care-based screening is the cost and limited portability of cart-based machines; therefore, the innovation of low-cost handheld devices with similar diagnostic accuracy represents a potential game changer, if our preliminary findings and those of others are replicated in larger studies.24,27,39,40 Therefore, primary care-based studies in the future should especially focus on handheld devices.
Conclusion
In our small study, we have found data suggesting that family physicians with variable ultrasound experience can accurately perform AAA screening in the ambulatory setting. The level of training was not associated with reduced accuracy, nor did reported difficulty performing the scan impact the results. Handheld devices may perform as well as laptop based models and should be targeted in future research.
Notes
This article was externally peer reviewed.
Funding: Partial support was provided by a small grant from the Department of Family Medicine of the University of North Carolina at Chapel Hill.
Conflict of interest: None.
- Received for publication June 2, 2025.
- Revision received July 7, 2025.
- Accepted for publication August 11, 2025.
References
Jump to section
Related Articles
Cited By...
- No citing articles found.