Digestive Endoscopy
Detection of abnormal lesions recorded by capsule endoscopy: A prospective study comparing endoscopist's and nurse's accuracy

https://doi.org/10.1016/j.dld.2006.03.019Get rights and content

Abstract

Background and study aim

Capsule endoscopy is a non-invasive technique for small bowel examination but its evaluation is time consuming. The aim of this study was to assess whether, following adequate training, an endoscopy nurse is capable of picking up all significant images without reducing the diagnostic accuracy of the procedure.

Patients and methods

Between April 2003 and December 2004, a total of 41 consecutive capsule endoscopy studies were blindly reviewed by both an endoscopy nurse and an endoscopist. The two operators had to select all significant images independently and to complete a structured questionnaire. Thirty-nine capsule endoscopy examinations (two studies discharged for premature battery failure) were evaluated. The agreement between the two operators was calculated by kappa statistics (coefficient of agreement).

Results

Agreement was excellent for all kind of selected lesions (mean κ > 0.85); the agreement was complete (κ = 1) for site identification, active bleeding, stenosis and negative studies. The greater disagreement (κ = 0.77) was found in cases of subtle mucosal abnormalities (i.e. reduction of villi), which were over-estimated by the nurse.

Conclusions

The preview recordings made by the nurse may increase the cost/effectiveness of the study, by considerably reducing the time needed for the endoscopist to make the final report (about 5–10 min), without compromising final diagnosis.

Introduction

Capsule endoscopy (CE) is a non-invasive endoscopic technique introduced in clinical practice since 2001. It is a revolutionary new diagnostic tool for the detection of small bowel diseases consisting of a swallowable capsule, able to record video images of the intestinal tract and transmit them to a data recorder, worn on a belt outside the body. After about 8 h of recording, the patient returns the data recorder, and images are downloaded to a computer workstation (RAPID® workstation, Given Imaging), where they can be displayed at a rate of 1–40 images/s. During the 8 h of recording the capsule acquires about 54,000 images.

The primary clinical indication for CE is obscure gastrointestinal bleeding (OGIB) [1], but this technique is also indicated for other small bowel diseases, such as Crohn's disease (CD), hereditary polyposis syndrome, small bowel damage secondary to non-steroidal anti-inflammatory drugs (NSAIDs), chronic diarrhoea [2], [3], [4], and for the follow-up of patients who underwent intestinal transplantation [5]. In a number of previous studies CE has showed its superiority over push-enteroscopy and radiographic examinations, such as small bowel enteroclysis, multi-slice computed tomography enteroclysis, magnetic resonance enteroclysis and scintigraphic methods in detecting small bowel lesions [6], [7], [8].

Major drawback of the technique is the time needed to download and, in particular, to review stored images. In this context, the traditional role of the nurse includes preparation of the patient with application of sensor arrays, administration of the capsule and, subsequently, download of images, while the physician is asked to interpret the videos. However, the mean time required for images review is considerably high, ranging from 60 to 120 min [7], [8]. On the other hand, CE is an operator-independent procedure, since the operator cannot modify in any view the characteristics of the images (i.e. by changing the camera's orientation); moreover, several different operators may review the same video without affecting both the quality and quantity of stored images. We hypothesise that a well-trained nurse might accurately review video images and identify all relevant abnormalities. If the nurse carefully selects critical images, the accuracy of the subsequent diagnosis made by the physician would be preserved with a considerable sparing of time.

In a small study by Levinthal et al. [9], the diagnostic accuracy by a nurse was assessed in 20 consecutive CE studies, all of which had been read in a blinded fashion by the gastroenterologist. The efficacy in abnormal images detection was estimated by comparing the number of significant lesions detected by the endoscopy nurse and the gastroenterologist. The nurse reached a 93% sensitivity (CI 74–99%) in identifying significant lesion when compared to the endoscopist. A recent study aimed at evaluating the ability of an experienced endoscopy nurse to pinpoint abnormal findings and to prepare CE records for physician interpretation, has demonstrated that a preliminary review by the nurse appear to be safe, reliable and cost-effective [10]. In both these studies [9], [10], the GI nurse and the physician were trained with 10 or 15 videos, respectively, before the study. However, the nurse's abnormal findings were compared with those of the physician's in a face-to-face session, without any formal statistical analysis.

The aim of our study was to assess whether an experienced endoscopy nurse might adequately select all significant images without affecting the diagnostic accuracy of the procedure. Moreover, the average cost of the procedure with the two different ways of analysis has been calculated.

Section snippets

Materials and methods

The study was carried out by an endoscopy nurse (G.C.) with 13 years of experience in following over 40,000 diagnostic or operative video endoscopies (gastroscopy, colonoscopy, push-enteroscopy and ERCP), and a by an endoscopist (V.A.) with 20-years experience with over 60,000 diagnostic or operative endoscopies (gastroscopy, colonoscopy and push-enteroscopy) performed. They were both trained to analyse CE videos by attending a specific 5-day course organised by the capsule distributor

Statistical analysis

For the analysis of agreement among the operators, the Cohen kappa coefficient was used [11] and calculated by the SPSS 13.0 for windows package software. Kappa coefficient may verify that agreement exceeds chance levels, according to the Fleiss’ scale [12]; more specifically, the agreement is scored as marginal (κ = 0–0.4), good (κ  0.4–0.75) and excellent (κ  0.75).

The cost of CE examination was calculated taking into account the following parameters:

  • 1.

    cost of 1 h of physician's time;

  • 2.

    cost of 1 h of

Results

The caecum was reached in 37 out of 41 cases (90%); all patients showed an adequate small bowel cleansing. In four subjects the capsule did not reach the caecum (two cases due to premature battery failure, and two cases because of small bowel stenosis, respectively). The mean small bowel transit time was 4 h and 33 min (range 65–490 min).

The nurse and the endoscopist selected a total of 65 and 53 thumbnails of abnormal lesions, respectively. The characteristics of detected lesions and relative

Discussion

CE is a new tool for the examination of small bowel disease with a clear-cut advantage over other endoscopic and radiologic techniques [6], [7], [8]. Its main drawback is that it is time consuming, especially in the evaluation of stored images. We hypothesised that a well-trained nurse may accurately select all significant images with a good agreement with the endoscopist; if this was the case, the time to make the final diagnosis would be considerably shortened (from 60–120 to 5–10 min),

References (14)

There are more references available in the full text version of this article.

Cited by (0)

View full text