Can otolaryngologists accurately diagnose and triage patients with suspected peritonsillar abscess using oropharyngeal examinations recorded using standard smartphone cameras?
In this comparative effectiveness research study, 31 patients with suspected peritonsillar abscess were evaluated telemedically, with a mean diagnostic accuracy of 81%. Comparing patients who were deemed to require prompt otolaryngology evaluation and those with peritonsillar abscess by the criterion standard, mean sensitivity was 90%.
Telemedical consultation is a viable, cost-conscious, efficient, and safe approach to peritonsillar abscess management.
Telemedicine is rapidly gaining traction as a way to reduce costs and connect patients with medical experts outside their local communities. Peritonsillar abscess (PTA) is a logical pathologic condition to evaluate for effectiveness of remote diagnosis given its prevalence and the paucity of on-site otolaryngologists at many institutions.
To explore the potential of otolaryngology telemedical consultation in triaging and diagnosing patients with suspected PTA.
Design, Setting, and Participants
A comparative effectiveness research study was conducted from January 1 to June 30, 2018, at 3 tertiary care hospitals among 31 consecutive patients aged 18 to 85 years for whom the otolaryngology department was consulted to assess for PTA. Statistical analysis was conducted from July 1 to September 30, 2018.
Telemedical evaluation of suspected PTA by 5 attending otolaryngologists blinded to patients’ history aside from the chief report of odynophagia. Otolaryngologists rated each patient video on whether they believed the patient had a PTA and whether the case warranted prompt evaluation by an otolaryngologist. Predictions were compared with the criterion standard of drainage or negative needle aspiration. Otolaryngologists additionally assessed video quality.
Main Outcomes and Measures
Rates of accurate diagnosis and triage of PTA based on otolaryngologists’ review of oropharyngeal examinations recorded using standard smartphone cameras, as well as percentage of videos of oropharyngeal examinations using standard smartphone cameras deemed of sufficiently high quality for clinical decision-making.
A total of 31 patients (16 women [51.6%]; mean age, 31.9 years [range, 18-62 years]) were recruited, and 16 patients (51.6%) had a PTA. Comparing otolaryngologists’ predictions with PTA status by the criterion standard, the prediction was consistent with that of the criterion standard 81% of the time averaged across otolaryngologists (mean diagnostic accuracy, 0.81). Similarly, the mean diagnostic accuracy was 0.83 when comparing the otolaryngologist’s suggestion for a prompt in-person evaluation with actual PTA status by the criterion standard. Comparing patients who were deemed to require prompt otolaryngology evaluation and those with PTA by the criterion standard, mean sensitivity was 90%. Videos were rated as of sufficiently high quality to make a diagnosis in 154 of 155 videos (99.4%).
Conclusions and Relevance
This study suggests that telemedical consultation is a viable, cost-conscious, efficient, and safe approach to PTA management. Despite having some difficulty diagnosing PTAs based on “history concerning for PTA” and oropharyngeal video alone, otolaryngologists are able to determine, with high sensitivity, which patients require prompt otolaryngology evaluation. The recording of consistently high-quality video using a standard smartphone camera is achievable without formal training.
As our health care system moves toward alternative payment models in an effort to provide more cost-effective, high-quality care, telemedicine has emerged as a potential strategy. Within the field of otolaryngology, studies have suggested the potential of telemedicine, showing that it can be reliable and accurate while also leading to cost savings1,2 in both real time and delayed settings.3,4 Telemedicine is appealing for treating patients with otolaryngologic pathologic conditions, as very few emergency departments or urgent care centers have access to an on-site otolaryngologist, particularly on nights and weekends.
Within otolaryngology, one of the most common symptoms among patients presenting to an emergency department is odynophagia with concern for a peritonsillar abscess (PTA). Peritonsillar abscesses are the most common head and neck deep space infections, with an annual incidence of 30 to 37 per 100 000 in the general population and 125 per 100 000 in patients between 14 and 21 years of age.5,6 Hundreds of thousands of new cases of PTA are treated annually, representing an annual cost of more than $150 million to treat the condition.6 The incidence of PTA has trended upward over time, increasing by 18% during the past 10 years, in part owing to indications for tonsillectomy becoming more narrowed over time.7 Along with this trend, the demand for otolaryngology consultations and transfers for otolaryngologic evaluations has also increased. Health systems, payers, and patients might benefit if telemedicine could be deployed as part of an easily implemented PTA screening protocol.
The application of telemedicine within otolaryngology often depends on expensive commercial telemedical systems outfitted with specific equipment, such as nasopharyngoscopes or thermal imaging hardware.8 The initial financial outlay for these instruments undercuts the utility of the approach, as the costs are untenable for lower-volume centers that could otherwise benefit the most from seeking outside expertise. Owing to the ubiquity of smartphones with high-definition video cameras and Health Insurance Portability and Accountability Act (HIPAA)–compliant communication services, physicians can now easily take and send high-quality video of patients’ physical examinations to other physicians, allowing real-time feedback while avoiding HIPAA violations. Although the use of smartphone video applications allows for efficient and cost-effective remote consultation, robust investigations into the use of this technology for the diagnosis of PTA are lacking, to our knowledge.8 This study explores the feasibility and practicality of using telemedical consultation of otolaryngologists in the management of patients with suspected PTA.
We created and subsequently evaluated video recordings of oropharyngeal examinations performed by 2 consulting otolaryngology resident physicians (J.R.M. and R.D.) on consecutive patients with suspected PTA at 3 tertiary care hospitals in both the emergency department and inpatient settings from January 1 to June 30, 2018. Patients were eligible for the study if they were between age 18 and 85 years and if the otolaryngology department was consulted to assess for PTA. All videos were recorded via the hospital-sanctioned, HIPAA-compliant software TigerText (TigerConnect). The consulting otolaryngologists used 2 tongue blades, one on top of the other. Videos were obtained using the standard camera application on a Samsung Galaxy S7 smartphone (Samsung) with the camera flash enabled and flashlight turned on. Videos did not include audio and ranged from 14 to 27 seconds (mean, 19 seconds). The Hartford Healthcare Institutional Review Board approved this study. Patients provided verbal consent.
Five otolaryngologists (T.F., N.K., S.S., C.G., and S.W.) evaluated the aggregated videos without any additional history or examination findings. Each video was evaluated on several metrics. First, the video quality was evaluated, and each video was assigned 1 of 3 designations: video quality and content were sufficient for diagnosis, video quality was sufficient for diagnosis but the content was insufficient for diagnosis, or video quality was insufficient for diagnosis. Next, the attending physicians were asked to comment on whether they believed the patient in question had a PTA or not. Finally, the attending physicians were asked to report on whether they thought the patient warranted a timely in-person consultation and/or aspiration or drainage. These predictions were independent of one another. For example, if the evaluating otolaryngologist thought that the patient likely did not have a PTA but the results of the examination were concerning, they could still recommend that the patient have prompt in-person otolaryngology evaluation despite a prediction of “no PTA.” The attending physicians in the study represented both private practice and academia; all were affiliated with the University of Connecticut otolaryngology residency program.
The diagnosis that resulted from the initial examination by the consulting otolaryngologist served as the criterion standard for the presence or absence of a PTA. A patient was deemed to have a PTA if purulence was drained or an abscess was identified on computed tomography (CT) scan. A patient was deemed to not have a PTA if no purulence was expressed with formal incision and drainage or with needle aspiration performed in 3 distinct locations along the tonsillar capsule provided that any CT or magnetic resonance imaging results, if previously obtained, were also negative for a PTA. Our institutions do not regularly use ultrasonography for suspected PTA. In the event of failed drainage in a patient with high suspicion for PTA or in any patient with concerning findings, a senior-level resident and/or an attending otolaryngologist were involved.
Statistical analysis was conducted from July 1 to September 30, 2018. Agreement between the diagnoses obtained via video and the criterion standard diagnosis was measured by calculating the percentage of correct predictions, sensitivity, specificity, positive predictive value, negative predictive value, and Cohen κ. These 6 values were calculated for both of the 2 separate predictions made by the 5 otolaryngologists reviewing the videos. Specifically, these measures were calculated comparing the criterion standard diagnosis with the reviewers’ predictions of (1) whether the patients had a PTA and (2) whether they thought that the patient warranted prompt in-person evaluation by an otolaryngologist. Percentage of correct predictions was calculated by the number of correct predictions (consistent with the criterion standard prediction) divided by the total number of predictions. Cohen κ was adjusted for chance agreement in all applicable scenarios. Overall, we calculated the mean diagnostic accuracy, which was defined as the mean percentage of correct predictions across otolaryngologists. We also used the majority predictions to calculate the 6 diagnostic values (eg, the majority prediction would be “yes” if 3 of 5 predictions were “yes” compared with 2 “no” predictions). The majority prediction is expected to produce robustly high diagnostic values assuming the raters are similarly skilled.
In addition, the agreement among the 5 attending otolaryngologists was assessed using Fleiss κ. Based on work by Landis and Koch,9 interrater reliabity was defined as poor for values less than 0, slight for values between 0 and 0.20, fair for values between 0.21 and 0.40, moderate for values between 0.41 and 0.60, substantial for values between 0.61 and 0.80, and almost perfect for values between 0.81 and 1.00.
A total of 31 patients (16 women [51.6%]; mean age, 31.9 years [range, 18-62 years]) were recruited for the study; 16 patients (51.6%) received a diagnosis of PTA by the criterion standard during in-person examination.
Comparing otolaryngologists’ PTA predictions with actual PTA status by the criterion standard (Table 1), the prediction was consistent with that of the criterion standard 81% of time averaged across otolaryngologists (mean diagnostic accuracy, 0.81). The agreement among the otolaryngologists was moderate (Fleiss κ = 0.57), in which the agreement between each otolaryngologist and the criterion standard ranged from 0.49 to 0.81. However, the prediction based on the majority prediction produced high diagnostic values: sensitivity of 0.81 (95% CI, 0.54-0.96), specificity of 1.00 (95% CI, 0.78-1.00), positive predictive value of 1.00 (95% CI, 0.75-1.00), and negative predictive value of 0.83 (95% CI, 0.59-0.96).
Comparing otolaryngologist’s suggestions for a prompt in-person evaluation with actual PTA status by the criterion standard (Table 2), the prediction was consistent with that of the criterion standard 83% of time averaged across otolaryngologists (mean diagnostic accuracy, 0.83). The agreement among the otolaryngologists was moderate (Fleiss κ = 0.60), in which the agreement between each otolaryngologist and the criterion standard ranged from 0.55 to 0.74. However, the prediction based on the majority prediction produced high diagnostic values: sensitivity of 1.00 (95% CI, 0.79-1.00), specificity of 0.80 (95% CI, 0.52-0.96), positive predictive value of 0.84 (95% CI, 0.60-0.97), and negative predictive value of 1.00 (95% CI, 0.74-1.00). Comparing patients who were deemed to require prompt otolaryngology evaluation and those with PTA by the criterion standard, mean sensitivity was 90%.
When the attending physicians rated the quality of the videos, 154 of 155 (99.4%) were rated to be of sufficiently high quality to make a diagnosis; 128 of 155 videos (82.6%) were deemed to be of sufficiently high quality and to have sufficient content to support a diagnosis.
The data illustrate that smartphone video of oropharyngeal examination findings in patients with suspected PTA reviewed by an otolaryngology attending physician has good reliability of diagnosis when compared with evaluation in person by an otolaryngology resident. The sensitivity is even higher when considering which patients were deemed to benefit from in-person evaluation by an otolaryngologist. As a screening test, this high sensitivity is particularly encouraging. The interrater reliability as measured by the Fleiss κ was near the cutoff of moderate and substantial agreement based on parameters defined by Landis and Koch.9
Based on an independent, unanimous vote, all but 1 of the videos taken with a standard smartphone camera were deemed to be of sufficient quality to enable a diagnosis to be made remotely. This finding suggests that emergency and urgent care clinicians can use easy-to-take videos on readily available equipment to engage with otolaryngologists on clinical assessment and the need for intervention, including drainage, as well as potential patient transfer. The quality and capabilities of the video applications available on smartphones have consistently improved over time, supporting the expanded use of these applications for telemedicine in the future.
By limiting the attending physicians’ knowledge of the patients’ presenting history and other examination findings, we artificially lowered our diagnostic accuracy by removing important context clues. This protocol was done to isolate the utility of the video in making a diagnosis so that the otolaryngologists rating the videos could not infer a diagnosis from the content of the history or the manner in which an otolaryngology resident provided it. Because not all emergency and urgent care clinicians will know which aspects to focus on when taking a history for a patient suspected of a PTA, we think this process allows for broad generalizability of the results. In practice, otolaryngologists could use easily attainable information (eg, a history of trismus that improved with corticosteroids, a muffled voice, or high fevers with leukocytosis) to further improve their accuracy.
Owing to changing trends in patients with PTA, making a diagnosis based on history alone is becoming increasingly challenging. Over time, the average patient with a PTA has become older, is less likely to have had an antecedent tonsillitis of significant length, and is more likely to have already failed adequate oral antibiotic treatment.10 These trends accentuate the importance of having an oropharyngeal examination evaluated by a specialist. Although other methods of diagnosis may have higher sensitivity or specificity, they have other associated costs to consider. Computed tomography is moderately expensive and involves risks of radiation. Magnetic resonance imaging is even more expensive and is time consuming. Ultrasonography is very operator dependent, requires purchasing of equipment and training personnel, and presents potential technical challenges in a patient with severe trismus. Smartphone videos are fast and have a marginal cost approaching $0.
Telemedical consultation also serves to ameliorate the disparities in access to an otolaryngologist that are commonplace in rural settings.11 Otherwise, the lack of immediate access often leads to treatment decisions being made exclusively by a clinician who does not often treat such patients or prompts transfer to another institution. These transfers are time intensive, labor intensive, and cost intensive.
Given the high sensitivity with which raters identified patients requiring in-person consultation, it is unlikely that a patient with a PTA would be missed. Furthermore, not all PTAs require drainage. Souza et al12 found that 95% of patients with small PTAs (mean size, 1.3 cm based on largest dimension detected on the CT scan) can be safely and effectively managed with antibiotics alone. Thus, even if relatively small PTAs are missed by tele-examination, the chance of untoward effects is low.
Based on these results, we propose an easily implemented, cost-effective telemedical screening protocol for patients with a history or physical examination results suggestive of PTA, as outlined in the Figure. Despite uncertainties regarding reimbursement, the cost savings potential is immense if unnecessary imaging, in-person consultation, or patient transfer can be avoided. The Centers for Medicare & Medicaid services have recently proposed13 separate reimbursement for “Interprofessional Internet Consultation,” further supporting the financial viability of such telemedical consultations. These procedures are also supported by Current Procedural Terminology codes 994X6, 994X0, and 99446 to 99449. Relative value unit valuations have yet to be fully elucidated. Patients should be given the same opportunity to consent to remote consultations as they do for in-person consultations.
Owing to the associated training burden, our protocol does not include ultrasonographic evaluation, which some institutions use with good results. Intraoral ultrasonography has a sensitivity of 89% to 95% and a specificity of 78% to 100% but is highly dependent on the training of the person performing and interpreting the results. Transcutaneous ultrasonography, while less invasive and uncomfortable, has a slightly lower sensitivity and specificity.14
This study had some limitations. As the primary goal was to assess the feasibility of using the smartphone video application to arrive at a diagnosis, enrollment was limited. Given that each of the hospitals included are tertiary care centers, selection bias may have skewed the sample toward more acutely ill patients. Despite these limitations, we think that telemedicine offers a potentially useful strategy for diagnosing PTA. Further research, including randomization to telemedical vs standard in-person consultation, is needed to more fully understand the role that telemedicine may play in otolaryngology.
Telemedical consultation is a viable, cost-conscious, efficient, and safe approach to PTA management. Despite having some difficulty diagnosing PTAs based on nothing more than a patient history concerning for PTA and an oropharyngeal video, otolaryngologists are able to determine, with high sensitivity, which patients require prompt otolaryngology evaluation. The recording of consistently high-quality video using a standard smartphone camera is achievable even without formal training.
Accepted for Publication: June 4, 2020.
Corresponding Author: Jonathan R. Mallen, MD, Division of Otolaryngology, Department of Surgery, University of Connecticut Health, 263 Farmington Ave, Farmington, CT 06030 (firstname.lastname@example.org).
Published Online: August 20, 2020. doi:10.1001/jamaoto.2020.1972
Author Contributions: Dr Mallen had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Mallen, Shah, Lotterman, Bonaiuto.
Acquisition, analysis, or interpretation of data: Mallen, Drake, Kreicher, Falcone, Karter, Schoem, Grindle, Wolfe, Kuo, Mu, Bonaiuto.
Drafting of the manuscript: Mallen, Shah, Wolfe, Kuo, Lotterman.
Critical revision of the manuscript for important intellectual content: Mallen, Drake, Kreicher, Falcone, Karter, Schoem, Grindle, Mu, Bonaiuto.
Statistical analysis: Mallen, Shah, Drake, Wolfe, Kuo, Mu.
Administrative, technical, or material support: Shah, Kreicher.
Supervision: Falcone, Karter, Schoem, Grindle, Wolfe, Lotterman, Bonaiuto.
Conflict of Interest Disclosures: Dr Falcone reported receiving personal fees from GI Reviewers LLC outside the submitted work. No other disclosures were reported.
Additional Information: All work was done at hospitals affiliated with the University of Connecticut Health System. All authors are affiliated with the University of Connecticut Health System.
Additional Contributions: Tara McLaughlin, PhD, Hartford Hospital Department of Surgery, and Roslyn Cooper assisted with authorship. Neither party has any conflicts of interest or received direct compensation for their work.
D. Otolaryngology consultations by real-time telemedicine. Ulster Med J
. 2002;71(1):26-29.PubMedGoogle Scholar
P. Comparison of medical therapy alone to medical therapy with surgical treatment of peritonsillar abscess. Otolaryngol Head Neck Surg
. 2018;158(2):280-286. doi:10.1177/0194599817739277
et al. Comparison of medical versus surgical management of peritonsillar abscess: a retrospective observational study. Laryngoscope
. 2016;126(7):1529-1534. doi:10.1002/lary.25960