Discussion

This consensus study, conducted according to international best practice guidelines,^{Reference Williamson, Altman, Bagley, Barnes, Blazeby and Brookes13} has achieved its goals, leading to the recommendation of the COSNOE diagnostic criteria for necrotising otitis externa (Table 1) and the COSNOE core outcome set for necrotising otitis externa (Table 2). These are the first of their kind worldwide.

The stakeholder group were consulted regarding whether graded, secondary or optional diagnostic items may be useful. There was very clear consensus on this, with all collaborators agreeing that core, minimum diagnostic criteria are preferred, with no role for graded, secondary or optional items. This is to facilitate the clarity, useability and synthesis of future research.

The third goal of the study, regarding disease terminology, reached consensus in Delphi round 1, with ‘necrotising otitis externa’ clearly preferred over ‘malignant otitis externa’ in keeping with published consensus definition recommendations.^{Reference Hodgson, Khan, Patrick-Smith, Martinez-Devesa, Stapleton and Williams12} The term ‘malignant’, first introduced by Chandler in 1968,^{Reference Chandler1} was thought to incorrectly imply a neoplastic process, risking unnecessary patient anxiety. It is important to note that neither term describes osteomyelitis, which is a key feature of advanced necrotising otitis externa. At our consensus meeting, ‘invasive otitis externa’ was proposed as a more accurate term, but the consensus preference for ‘necrotising otitis externa’ prevailed.

On the theme of terminology, a clear consensus was also achieved in Delphi round 1 regarding necrotising otitis externa ‘relapse’, which respondents defined as recurrence of ipsilateral necrotising otitis externa in any timescale following cessation of treatment (core outcome set item 34, Table 2). Previous work has limited the definition of ‘relapse’ to the three-month period following treatment cessation, but our respondents were unanimous in their consensus that an indefinite timescale better reflects the disease process in necrotising otitis externa.

Our patient-derived items were generated via a qualitative study, which was conducted with the specific goal of generating patient-suggested items for the core outcome set. This yielded rich information regarding patient experience of necrotising otitis externa.^{Reference Owen, Abrar and Stapleton7} Patients with necrotising otitis externa tend to be elderly, frail patients with medical co-morbidities^{Reference Arslan, Pekcevik and Cukurova17} and a poor five-year survival rate from both disease-specific and co-morbid causes.^{Reference Stern Shavit, Soudry, Hamzany and Nageris6} For this reason it was agreed not to include patients in the Delphi process, but to optimise data collection via a parallel qualitative study.^{Reference Owen, Abrar and Stapleton7} This is a minor departure from Core Outcome Measures in Effectiveness Trials guidelines to accommodate our patient population and may be a useful precedent for core outcome set methodology involving other frail and co-morbid patient populations.

The only previously published diagnostic criteria for necrotising otitis externa date back to 1987,^{Reference Cohen and Friedman11} when Cohen and Friedman proposed minimum diagnostic criteria. Perhaps unsurprisingly, our consensus diagnostic criteria do not depart greatly from these 36-year-old criteria, with both incorporating pain, otorrhoea, oedema, discharge, granulation tissue, positive imaging and immunocompromise.

One of the main discussion points at our consensus meeting was immunosuppression. The wording ‘at least one risk factor for impaired immune response’ (necrotising otitis externa diagnostic criterion 11, Table 1) was agreed to reflect the inclusion of this complex theme. Consensus for the inclusion of risk factors for immunosuppression was observed in all stages of the stakeholder consensus process, despite the publication of rare cases of necrotising otitis externa in apparently immunocompetent individuals.^{Reference Lui, Peng, Mo and Liang18} It was agreed that frailty, immunosenescence and malnutrition should be included as potential risk factors for impaired immune response, alongside diabetes mellitus and other recognised causes of immunosuppression or immunodeficiency. Local risk factors for necrotising otitis externa did not reach consensus for inclusion. Despite the appearance of local risk factors for necrotising otitis externa in the literature,^{Reference Chandler1,Reference Bisbinas and Stapleton2,Reference Guevara, Mahdyoun, Pulcini, Raffaelli, Gahide and Castillo19} they were felt not to be ubiquitous and therefore not a core diagnostic criterion.

Items associated with the immune response to infection were excluded as diagnostic items (fever and adenopathy) or did not reach consensus for inclusion or exclusion (inflammatory markers including CRP, ESR and WBC count). At our consensus meeting, stakeholders agreed on the exclusion of inflammatory marker items as core diagnostic criteria for necrotising otitis externa. It is recognised that individuals with impaired immunity may not present a classic inflammatory response,^{Reference Pizzo20} hence inflammatory markers cannot be relied on to diagnose necrotising otitis externa because they are non-specific, may be indicative of a non-infectious process and may not be observed during a serious infectious process.^{Reference Jeffcoate and Lipsky21,Reference Moallemi, Niroomand, Tadayon, Forouzanfar and Fatemi22} Our stakeholder consensus was aligned with this, also adding that as a pseudomonal biofilm infection, necrotising otitis externa can be locally invasive without raising a systemic inflammatory response,^{Reference Mulcahy, Isabella and Lewis23} and that absence of systemic inflammatory response may account for delayed necrotising otitis externa diagnoses.^{Reference Bisbinas and Stapleton2} It was agreed that in patients where inflammatory markers are raised in the context of necrotising otitis externa, monitoring them may be a useful marker for response to treatment,^{Reference Chen, Yeh, Shiao and Tu24} a theme beyond the scope of this study that warrants future research trials.

Pain was another diagnostic theme that elicited discussion. Both otalgia and nocturnal pain reached consensus for inclusion as core diagnostic criteria in round 1. At our consensus meeting, it was agreed that pain as a diagnostic criterion can be self-reported or based on its impact on activities of daily living (e.g. eating, sleep) or on analgesia requirement. Although these measures are subjective, severe pain in necrotising otitis externa is ubiquitous and affects patients’ quality of life.^{Reference Bisbinas and Stapleton2} Recognition of pain and optimisation of analgesia are essential.

Whilst respondents agreed that it is essential to send both solid (granulation) and liquid (pus) samples for microbiological analysis to identify a causative organism, positive microbiology reached consensus for exclusion as a core diagnostic criterion for necrotising otitis externa (necrotising otitis externa diagnostic criterion 6, Table 1). If an organism can be cultured, this guides treatment choices, but negative culture will not preclude the empiric treatment of clinically diagnosed necrotising otitis externa. Culture-negative cases of necrotising otitis externa are well recognised^{Reference Gruber, Roitman, Doweck, Uri, Shaked-Mishan and Kolop-Feldman25} and may be attributed to prior antibiotic use, commensal microorganisms producing misleading results or fungal disease. Our respondents discussed the role of deep tissue samples and agreed that whilst these can be deployed for refractory cases and to confirm fungal necrotising otitis externa, they can also be unreliable in patients who have had extensive antibiotic therapy.^{Reference Gruber, Roitman, Doweck, Uri, Shaked-Mishan and Kolop-Feldman25}

The delivery of therapies other than antimicrobials was discussed, including surgery and hyperbaric oxygen therapy^{Reference Amaro, Espiney, Radu and Guerreiro26} for necrotising otitis externa. It was agreed that all interventions for necrotising otitis externa should be reported and ideally published to add to the evidence base.^{Reference Phillips and Jones27} The reporting of surgical intervention, the decision to progress to surgery and the delivery of any additional or alternative therapies reached consensus for inclusion (core outcome set items 23 and 34, Table 2).

Indicators of advanced disease were identified as an important theme for future research, as they can guide treatment decisions and predict prognosis.^{Reference Soudry, Hamzany, Preis, Joshua, Hadar and Nageris28,Reference Durojaiye, Slucka and Kritsotakis29} Initially, cranial nerve palsy did not reach consensus for inclusion or exclusion as a diagnostic marker of advanced necrotising otitis externa. Based on free-text comments in rounds 1 and 2, it was discussed in our consensus meeting, where it reached consensus for inclusion, not as a core diagnostic criterion per se, but as an essential theme for data reporting within the scope of clinicians making a clinical diagnosis of necrotising otitis externa (necrotising otitis externa diagnostic criterion 12, Table 1).

Respondents agreed that whilst cranial nerve palsies, temporo-mandibular joint involvement and extensive radiological changes can all be markers of advanced disease, there is insufficient evidence to include these as core diagnostic criteria. Heterogeneous clinical and radiological stratification and scoring protocols have been proposed,^{Reference Stevens, Lambert, Baker and Meyer30–Reference Cooper, Hildrew, McAfee, McCall, Branstetter and Hirsch33} and none has yet been universally adopted in disease reporting, indicating a lack of robust consensus on this theme.

Notably, positive computed tomography (CT) and magnetic resonance imaging (MRI) scans both reached consensus for inclusion as core diagnostic criteria in round 1 of our consensus process. Imaging recommendations for necrotising otitis externa diagnosis vary across the literature and tend to be based on imaging accessibility and local experience.^{Reference van Kroonenburgh, van der Meer, Bothof, van Tilburg, van Tingeren and Postma34,Reference Mehrotra, Elbadawey and Zammit-Maempel35} Whilst widely accessible, bony erosion on CT scanning reflects late disease, with MRI being superior for detection of early disease and for monitoring response to treatment.^{Reference Loh and Loh32} More work is required to clarify optimal imaging protocols in necrotising otitis externa, with positron emission tomography-CT becoming an increasingly popular choice^{Reference Stern Shavit, Bernstine, Sopov, Nageris and Hilly36} and nuclear imaging declining in popularity, with no single modality of imaging that can provide a complete picture of diagnosis, disease progression and resolution in necrotising otitis externa.^{Reference Khan37}

A validated patient co-morbidity index score reached consensus for inclusion (core outcome set item 22, Table 2). It was noted that the Charlson Comorbidity Index has been frequently used in the literature.^{Reference Charlson, Pompei, Ales and MacKenzie38} Use of a validated score may indicate prognosis and guide clinicians regarding optimal therapy decisions,^{Reference Treviño González, Reyes Suárez and Hernández de León39} and it is also a useful metric for future research synthesis in terms of defining patient populations and examining mortality risk.