Teleconsultation is a diverse collection of technologies and clinical applications used by electronically transferring information from one site to another.18 It has been proven effective for many purposes and has been used in numerous clinical settings including orthopaedics.5-13,21,23,26,27 However, commercial teleconsultation systems are expensive and not available in many hospitals. Advancements in information technology have made it easy for conventional analog radiographs to be digitized by taking high-resolution pictures with a digital camera, which then can be sent via E-mail to an expert for assessment at a low cost.3,24,27 The few studies regarding using digital images for teleconsultation in orthopaedics7,19-22 suggest no difference in diagnostic accuracy from conventional radiographs and ready acceptance.7,19-22 However, the power to detect a difference in these studies is questionable because of the small number of patients included. Also, the patients in these studies had displaced fractures, which might have been too obvious to reveal any substantial difference between digital images and conventional radiographs.
Patients with nondisplaced or minimally displaced fractures are at greater risk for misdiagnosis and improper management, decreased confidence in the quality of hospital service, or malpractice suits.1,6,15-17 Therefore, there is a need to test the validity and reliability of digital images for patients at risk for misdiagnosis before adopting them for wide use. Experience in radiographic interpretation may determine the usefulness of digital images.
We hypothesized that the reliability and validity of digital camera images would be similar to assessing the original analog radiographs as performed by groups of assessors with differing levels of experience in diagnosing fractures.
MATERIALS AND METHODS
Our department has used digital cameras for quality assurance of orthopaedic trauma care since 1999. All patients presenting to our emergency unit have digital clinical photographs taken of the injured site and digital images taken from the conventional analog radiographs using a high-resolution digital camera by the on-duty senior resident. By 2005, we had more than 5000 patient files. From these files, 100 patients with nondisplaced or minimally displaced fractures (displaced less than 3 mm) were selected by a systematic sampling technique by a panel of two orthopaedic staff members (BK, NK) and one radiologist (WT) who were not involved in the interpretation. We used clinical and radiographic followup data as the gold standard. When there were no such existing data, the opinion from a panel (two orthopaedists and one radiologist) was used as a gold standard. We selected only patients with isolated fractures and no dislocation or other injury. Fifty patients among those originally selected proved to have no fracture so the same files were used to select additional patients using simple random sampling according to injury location. Each patient had three sets of images: digital images taken from conventional analog radiographs (digitized radiograph), digital clinical photographs of the injured site, and conventional analog radiographs (Fig 1). All radiographs had at least two projections (anteroposterior and lateral). The patient order was randomly allocated. This study was performed with the approval of the Institutional Review Board of our institution.
We used a SONY MVC-CD 300 3.3-megapixels digital camera (Sony, Nagoya, Japan) with a 2046 × 1536 pixel resolution.
The file format was Joint Photographic Experts Group (JPEG), and files were recorded on recordable compact discs. The digital images were viewed on a personal computer monitor using a 15 × 19-inch liquid crystal display (LCD) screen. Magnifying or adjusting the image was allowed. Interpretation of conventional radiographs was done using a standard viewing box.
There were three groups of assessors with varying degrees of experience in radiographic interpretation. The staff group had 5 or more years experience, the senior resident group had greater than 3 years but less than 5 years experience, and the junior resident group had less than 3 years experience. All evaluations were performed independently and blind to clinical results. The assessors determined whether each patient had a fracture and specified the fracture location. The results were recorded on the data collection form. There were two stages of interpretations for each assessor. In the first stage, the images were evaluated before transferring via E-mail. In the second stage, the same images were reevaluated after transmitting via E-mail. The first stage was divided into three phases to investigate the usefulness of digitized radiographs and digital clinical photographs compared with conventional radiographs. In the first phase, all assessors evaluated only the digitized radiographs. The next week, the same files were reevaluated with a supplementary clinical photograph of each patient. The conventional radiographs were examined 1 week later. One month after the first stage, all digitized information was transferred via E-mail to every assessor for reevaluation. There were 5400 total observations. The fracture locations were divided into four regions: upper extremity, lower extremity, pelvis, and spine. The mean age of the patients was 45.5 years (range, 1-80 years).
A kappa statistic was used to test for agreement among each source of images and group of assessors. The level of agreement was categorized into five levels;14 slight (0-0.2), fair (0.21-0.4), moderate (0.41-0.6), good (0.61-0.8), and excellent (0.81-1). Sensitivity, specificity, and accuracy were calculated from cross- tabulation among the results of interpretation of each group of assessors and the gold standard. A chi square test was used to determine the gross association between variables and misdiagnosis. Multiple logistic regression analysis was used to identify independent variables affecting misdiagnosis.
The overall reliability (kappa), sensitivity, and specificity of reading the digitized radiographs were 0.57%, 83.2% and 80.7%, respectively. The reliability of reading the digitized radiographs was at a good level in the staff group, at a moderate to good level in the senior resident group, and at a fair to moderate level in the junior resident group. Adding clinical photographs improved the overall reliability (0.67) and sensitivity (87.2%), but decreased the specificity (79.2%). The reliability of reading conventional radiographs was better than reading digitized radiographs in the staff group and worse in both resident groups. Improved sensitivity was seen when reading digitized radio- graphs supplemented with clinical photographs only in the staff and senior resident groups. Assessors having more experience in radiographic interpretation had greater sensitivity, specificity, and accuracy compared with the less experienced groups (Table 1). There was no deterioration in reliability, sensitivity, or specificity of reading digitized radiographs after transmitting via E-mail (Table 2). Reliability of reading the images was not affected by the patient age groups or fracture region. A moderate to good level of agreement was shown in each region and each age group (Table 3).
The overall misdiagnosis was 15.5%. The only factor associated (p < 0.05) with misdiagnosis of fracture was the assessment group (Table 4). After adjusting for confounders, the assessment group was still the only independent variable affecting misdiagnosis. Senior resident and junior resident groups were 1.6 and 2.5 times, respectively, more likely to misdiagnose compared with the staff group (Table 5).
Teleconsultation via E-mail using digital camera images was valid and reliable compared with conventional radio- graphs. Digitized radiographs should be sent with clinical photographs of the injured site to improve reliability and accuracy of diagnosis. The only factor associated with misdiagnosis was level of experience of the assessor doing the radiographic interpretation. The consultant should have at least 5 years experience in radiographic interpretation. Digital camera images are a low-cost teleconsultation option for consultation of fracture care. Consultation with an expert via E-mail is advantageous for patients and physicians for proper and prompt management at a lower cost than commercial teleconsultations. Transferring the digital images for consultation from remote areas via E-mail is a new strategy, especially areas where specialists are unavailable or transportation is not available. However, Andres et al4 reported that resolution of the digital camera should be at least 1.3 megapixels, and macromode must be used for imaging small objects with an ability to focus at less than 5 inches.
Our study has some limitations. First, the time lag between each assessment may not have been long enough to abolish the memory effect in the interpretation. Second, the result of assessment of teleconsultation via E-mail might be improved because of previous experience in reading the image. Third, the large number of patients in each assessment session may have decreased the attention span of the assessors. The use of Landis and Koch criteria for assessing agreement are arbitrary. The level of acceptability depends on the implications of errors. Serious or life-threatening conditions might need a greater level of agreement than low, virulent injuries. However, our study included only patients with nondisplaced or minimally displaced fractures, therefore, using this criteria might not seriously jeopardize our findings.
Our study has three strengths. First, we included only patents with nondisplaced or minimally displaced fractures. Patients with these fractures are at risk for misdiagnosis, therefore our results showed the worse-case scenario. Second, the number of patients included is larger than in previous studies.19-21 Third, we included assessors with different levels of experience in radiographic interpretation, and all assessors were blinded to the clinical evaluation to decrease bias.
Comparing the results of our study with results of other studies is difficult because of differences in design, assessors, and patients selected. However, our results seem to be better than those of Scott et al,25 who reported an overall accuracy in detecting subtle fracture of only 60% compared with 80% using conventional radiographs. This is also comparable to the results of Palombo et al,17 who revealed a sensitivity of 82.6% in detecting displaced fractures using conventional radiographs as a gold standard.
The greater sensitivity and reliability in reading images from a digital camera compared with conventional radio- graphs might be explained by image manipulation capacity, such as contrast adjustment in the viewing station when making the interpretation. The improvement in sensitivity after adding a clinical photograph of the injured site could be the result of additional information. Junior residents showed no improvements in reliability or sensitivity as a result of the photographs. This might be because lack of experience prevents them from ascertaining the appropriate additional information.
We found images of musculoskeletal trauma taken by a digital camera can be used for teleconsultation from a remote area. Sending clinical and radiographic digital images will increase the level of sensitivity for diagnosis. However, the consultant should have enough experience in radiographic interpretations. Despite the potential usefulness of this technology, the risk of litigation for incorrect diagnosis and confidentiality is a concern.2,18 The decision regarding large-scale implementation should be based on the government healthcare policy and regulation. In the United States, hospitals must comply with the Health Insurance Portability and Accountability Act (HIPAA) requirements to avoid potential threats to patients' privacy and security.
Images of patients with musculoskeletal injuries taken with a digital camera can be sent via E-mail as a low-cost teleconsultation option, but good image quality and interpretation by a consultant with adequate experience are prerequisites. We recommend sending images of the radiographs and the clinical features of the injured site.
We thank Apiradee Lim for suggestions regarding data analysis.
1. Aaland MO, Smith K. Delayed diagnosis in a rural trauma center. Surgery
2. Abboud JA, Bozentka DJ, Beredjiklian PK. Telemedicine consultation for patients with upper extremity disorders is reliable. Clin Orthop Relat Res
3. Agbamu DA, Sim E. The data security aspects of telepathology via the internet have been ignored. Hum Pathol
4. Andres BM, Khanna AJ, Wenz JF Sr, Faust AF, Frassica FJ. A comparison of digital cameras: features essential for the orthopaedic surgeon. Clin Orthop Relat Res
5. Baldwin AJ, Langton SG. Postoperative monitoring of flaps by digital camera and internet link. Br J Maxillofacial Surg
. 2001;39: 120-121.
6. Ballas MT, Tytko J, Mannarino F. Commonly missed orthopedic problems. Am Fam Physician
7. Baruffaldi F, Gualdrini G, Toni A. Comparison of asynchronous and realtime teleconsultating for orthopaedic second opinions. J Telemed Telecare
8. Benger J. A review of telemedicine in accident and emergency: the story so far. J Accid Emerg Med
9. Buntic RF, Siko PP, Buncke GM, Ruebeck D, Kind GM, Buncke HJ. Using the internet for rapid exchange of photographs and x-ray images to evaluate potential extremity replantation candidates. J Trauma
10. Couturier P, Tyrrell J, Tonetti J, Rhul C, Woodward C, Franco A. Feasibility of orthopaedic teleconsultation in a geriatric rehabilitation service. J Telemed Telecare
11. Egol KA, Helfet DL, Koval KJ. Efficacy of telemedicine in the initial management of orthopaedic trauma. Am J Orthop
. 2003;32: 356-360.
12. Glode LM. Challenges and opportunities of the internet for medical oncology. J Clin Oncol
13. Hailey D, Roine R, Ohinmaa A. Systemic review of evidence for the benefits of telemedicine. J Telemed Telecare
14. Landis JR, Koch GC. The measurement of observer agreement for categorical data. Biometrics
15. Larson A, Lynch DA, Zeligman B, Harlow C, Vanoni C, Thieme G, Kilcoyne R. Accuracy of diagnosis of subtle chest disease and subtle fractures with a teleradiology system. AJR Am J Roentgenol
16. Murphey MD, Bramble JM, Cook LT, Martin NL, Dwyer SJ3rd. Nondisplaced fractures: spatial resolution requirements for detection with digital skeletal imaging. Radiology
17. Palombo A, Haigh T, Ferguson J, Pedley D. Can pediatric radio- graphs be accurately interpreted using an inter-hospital telemedicine system? J Temed Telecare
18. Perednia DA, Allen A. Telemedicine technology and clinical applications. JAMA
19. Pysher L, Harlow C. Teleradiology using low-cost consumer- oriented computer hardware and software. AJR Am J Roentgenol
20. Raikin SM, Bley LA, Leb RB. Emerging technology: remote analysis of traumatic musculoskeletal radiographs transmitted by electronic mail. J Orthop Trauma
21. Ricci WM, Borrelli J. Teleradiology in orthopaedic surgery: impact on clinical decision making for acute fracture management. J Orthop Trauma
22. Ricci WM, Borrelli J. Teleradiology in orthopaedics. Clin Orthop Relat Res
23. Rind DM, Kohane IS, Szolovits P, Safran C, Chueh HC, Barnett GO. Maintaining the confidentiality records shared over the Internet and World Wide Web. Ann Intern Med
24. Roine R, Ohinmaa A, Hailey D. Assessing telemedicine: a systematic review of the literature. CMAJ
25. Scott WW Jr, Rosenbaum JE, Ackerman SJ, Reichle RL, Magid D, Weller JC, Gitlin JN. Subtle orthopaedic fracture: teleradiology workstation versus film interpretation. Radiology
26. Szot A, Jacobson FL, Munn S, Jazayeri D, Nardell E, Harrison D, Drosten R, Ohno-Machado L, Smeaton LM, Fraser HS. Diagnostic accuracy of chest x-rays acquired using a digital camera for low- cost teleradiology. Int J Med Inf
© 2006 Lippincott Williams & Wilkins, Inc.
27. Wade FA, Oliver CW, McBride K. Digital imaging in trauma and orthopaedic surgery: is it worth it? J Bone Joint Surg Br. 2000;82: 791-794