Summary
Background
Despite favorable results of randomized studies and several guidelines, screening for abdominal aortic aneurysm is poorly implemented in most countries. In order to implement an effective abdominal aortic aneurysm screening programme, training of physicians other than cardiovascular imaging specialists is necessary. Also, the use of pocket-sized ultrasound systems seems an appealing alternative to conventional echography machines for large-scale screening.
Aims
To test the hypothesis that, after a short period of specific training with a pocket-sized ultrasound system, novice operators could reliably measure the abdominal aortic diameter. We assessed the agreement between abdominal aortic diameter measurements from novice operators using a pocket-sized ultrasound system and experts using conventional machines.
Methods
After focused training of novice operators, the abdominal aortic diameter was independently measured at least four times: by two experts using conventional ultrasound, by one expert using a pocket-sized ultrasound system and by at least one novice operator using the pocket-sized system; each operator was blinded to the others.
Results
The aortic diameters of 56 patients were measured. The intraclass correlation coefficients between the four sets of measurement were all > 0.91 and the mean difference between the measurements was negligible (<1 mm). The interoperator variability for experts using conventional machines versus novices using pocket-sized machines was ≤ 4 mm in 92.0% of cases. No learning curve over time was noted.
Conclusion
In order to screen for abdominal aortic aneurysm, the abdominal aortic diameter can be accurately measured by non-specialist physicians with pocket-sized ultrasound devices after a short period of training.
Résumé
Contexte
Malgré la publication de résultats favorables et les recommandations émises, le dépistage des anévrysmes de l’aorte abdominale est peu réalisé dans la plupart des pays. Il apparaît donc nécessaire de former d’autres médecins que les spécialistes. La diffusion des échographes de poche semble également bien adaptée à un dépistage à large échelle.
Objectif
Pour tester l’hypothèse que des médecins novices, après une courte formation ciblée, et avec un échographe de poche, pourraient mesurer le diamètre de l’aorte abdominale de manière fiable. Évaluer l’accord entre les mesures du diamètre de l’aorte abdominale effectuées par des médecins novices utilisant un échographe de poche, et celles réalisées par des experts utilisant un échographe classique.
Méthodes
Après une courte formation ciblée des novices, le diamètre de l’aorte abdominale de chaque patient a été mesuré de manière indépendante à au moins quatre reprises, en insu des autres mesures réalisées : deux fois par des experts avec l’échographe classique, une fois par un expert utilisant l’échographe de poche et au moins une fois par un novice avec le même échographe de poche.
Résultats
Les diamètres aortiques de 56 patients ont été mesurés. Le coefficient de corrélation intra-classe entre les quatre mesures était > 0,91 et la moyenne des différences des mesures était négligeable (< 1 mm) ; la variabilité inter-observateur était ≤ 4 mm dans 92 % des cas ; et il n’y a pas de courbe d’apprentissage de la mesure avec le temps.
Conclusions
Le diamètre de l’aorte abdominale peut être mesuré de manière valide par des médecins inexpérimentés, après une courte formation, à l’aide d’un échographe de poche. Ceux-ci peuvent acquérir rapidement des performances très acceptables, et être aptes à la réalisation du dépistage d’anévrysme de l’aorte abdominale.
Introduction
Abdominal aortic aneurysm (AAA) is conventionally defined as an aortic diameter enlarged by at least 50%. The main risk of AAA is rupture, with associated high mortality . This risk of rupture increases with AAA diameter. AAA-related mortality can be reduced by ultrasound screening of individuals at risk, with prompt intervention for the larger lesions (>50–55 mm), as is recommended in several countries . The simplest recommendations have been published in the UK, where all men aged > 65 years should be screened . In France, a recent national guideline document recommends AAA screening in men aged 65–75 years with a history of smoking or aged 55–75 years if they have a family history of AAA . The application of such recommendations implies great availability of cardiovascular ultrasound specialists. Some countries (e.g. UK and USA) have opted for large-scale screening, with some difficulties in implementation , while in others, including France, population screening has not yet been implemented. To improve screening implementation in our country, it is necessary to train other physicians to carry out the measurement of abdominal aortic diameter, so that they can screen their patients at risk of AAA. Moreover, such a strategy would require the use of handheld ultrasound devices which, due to their small size, high mobility and low cost, are more suitable for large-scale screening.
We hypothesized that, after a short period of focused training, novice operators using a pocket-sized ultrasound device would be able to reliably measure the abdominal aortic diameter. The aim of this study was to assess the agreement between abdominal aortic diameter measurements performed by novice operators using pocket-sized ultrasound systems and those obtained by experts using conventional ultrasound.
Methods
This prospective study assessed the agreement of measurements taken in a teaching hospital between May and July 2012 in two successive phases, according to level of expertise (novice versus expert) and type of ultrasound machine (pocket-sized versus conventional).
In the first phase, novice operators (medical students) were instructed during three 3-hour training sessions in the use of a pocket-sized ultrasound system (Vscan ® ; GE Healthcare, Wauwatosa, WI, USA) to measure the abdominal aortic diameter, with a phased-array probe 1.7–3.8 MHz. Initially, we set a theoretical and practical training programme for ultrasound imaging of the abdominal aorta. Two hands-on sessions were then organized to learn the settings and manipulation of the pocket-sized ultrasound system, to identify the aorta and its adjacent structures in different tomographical planes and to measure the abdominal aortic diameter.
In the second phase, we compared the measurements performed by novice operators with those obtained by experts, using either the pocket-sized system or a conventional ultrasound machine (iE33; Philips Healthcare, Boston, MA, USA). For this machine, we used a 5 MHz phased-array probe. Each operator used only one of the two ultrasound devices on each patient and averaged the results of three measurements of the external anteroposterior diameter of the infrarenal aorta in the transverse view, immediately above its bifurcation. In case of AAA (defined as a diameter > 30 mm), the maximal external anteroposterior diameter was required. Each operator was blinded to the results of the other operators. For each patient, at least four sets of measurements were performed: by two experts using the conventional system, by one expert using the pocket-sized machine and by at least one novice operator using the pocket-sized machine. For each patient, the experts involved were randomly selected from the team of nine physicians in our laboratory. Measurements by two experts using a conventional machine, one expert using the pocket-sized machine and at least one novice using the pocket-sized machine were each compared with each other ( Fig. 1 ).
All patients included in this study were initially hospitalized for cardiovascular diseases other than aortic disease and were invited to take part into this study. Patients who had previously undergone operations on the abdominal aorta and those with an unstable haemodynamic state or any other condition jeopardizing their immediate prognosis were excluded from the study. Patients who declined our invitation to participate in the study were excluded and refusals were reported. Informed consent was obtained from all participants. The study was approved by the ethical committee of our institution (Committee for Persons Protection, Southwestern France-IV) on 12th April 2012.
Based on the literature , we took a difference of ≤ 4 mm between two measurements of the abdominal aortic diameter to represent good interoperator agreement. We also assessed this reproducibility with more stringent thresholds of ≤ 3 mm and ≤ 2 mm. For an expected intraclass correlation coefficient (ICC) of 0.80 (i.e. good agreement) with a precision of 0.10 (95% confidence interval [CI] 0.70–0.90), the number of evaluable patients needed for this study was 51. To account for non-evaluable cases, estimated at 15%, a total of 60 patients was necessary.
Qualitative variables are presented as frequencies and percentages; quantitative variables as means ± standard deviations. To assess the agreements, ICCs were calculated using the Shrout-Fleiss method (using the first-case ICC where operators were selected at random for each subject). This coefficient varies between 0 and 1. The ICC reflects a good agreement between the measurements when it is 0.71–0.90 and a very good one when it is ≥ 0.91. It is presented with its 95% CIs according to Smith’s method .
Bland-Altman plots were also plotted for each pair of measurements. These plots represent the differences between two measurements as a function of the mean of the two measurements. The limits of agreement for each plot are also presented to illustrate the fact that we would expect most of the differences between the two measurements to lie between this interval. The statistical analyses were performed with SAS 9.3 software (SAS Institute, Cary, NC, USA).
Results
Overall, 62 patients were recruited. Six patients were excluded because they did not undergo four sets of measurements during their hospitalization. Therefore, 56 patients (42 men and 14 women) were included in the analysis. The patients were hospitalized for peripheral artery disease ( n = 30), coronary bypass surgery ( n = 24) and cerebrovascular disease ( n = 2).
Table 1 displays the results of the abdominal aorta measurements. The estimation of the ICC showed good or very good agreement between the pairs of measurements without any statistical difference according to the level of expertise and type of ultrasound machine used ( Table 2 ). Accordingly, the Bland-Altman plots show good concordance between pairs of measurements ( Fig. 2 ), although some measurement differences are outside the ± 4 mm limits of agreement; these correspond to patients who were overweight and therefore had poorer imaging quality. Overall, the mean differences were small: 0.1 mm for expert/conventional versus expert/conventional ( Fig. 2 A); 0 mm for experts/conventional versus expert/pocket-sized ( Fig. 2 B); 0.1 mm for expert/pocket-sized versus novice/pocket-sized ( Fig. 2 C); and 0.7 mm for experts/conventional versus novice/pocket-sized ( Fig. 2 D), which is not clinically relevant.
