Introduction
Necrotising otitis externa is a serious infective condition originating from the external auditory canal, largely but not exclusively in patients with recognised risk factors.Reference Bisbinas and Stapleton1 The first report of necrotising otitis externa appears in the literature in 1830,Reference Toulmouche2 with the first case series described by Chandler in 1968.Reference Chandler3 Necrotising otitis externa has a profound effect on patients’ quality of lifeReference Owen, Abrar and Stapleton4 and carries potentially fatal complications.Reference Stern Shavit, Soudry, Hamzany and Nageris5
Necrotising otitis externa has attracted considerable interest from researchers due to its apparent rising incidence between 2002 and 2018Reference Linton and Stapleton6 with theories for this phenomenon including increasing prevalence of diabetes mellitus,Reference Chawdhary, Liow, Democratis and Whiteside7 antibiotic resistance,Reference Byun, Patel, Nguyen and Lambert8,Reference Le Clerc, Verillaud, Duet, Guichard, Herman and Kania9 ageing population, and increased clinician awareness of the condition.Reference Linton and Stapleton6
In patients with diabetes mellitus, microangiopathy and hypoperfusion are thought to contribute to the pathogenesis of necrotising otitis externa.Reference Unadkat, Kanzara and Watters10,Reference Singh, Al Khabori and Hyder11 However, non-diabetic and non-immuno-compromised cases are appearing in the literature.Reference van Tol and van Rijswijk12,Reference Bhatt, Pahade and Nair13 Pseudomonas aeruginosa remains the most common pathogen isolated in necrotising otitis externa.Reference Stapleton and Watson14 Pseudomonal resistance has been attributed to the increased use of fluoroquinolones, and the formation of biofilms.Reference Loh and Loh15,Reference Berenholz, Katzenell and Harell16
There is a paucity of evidence regarding the diagnostic criteria, management and monitoring of necrotising otitis externa.Reference Stapleton and Watson14 Although consensus definitions have been recommended,Reference Hodgson, Khan, Patrick-Smith, Martinez-Devesa, Stapleton and Williams17 standardised diagnostic criteria, treatment recommendations and outcomes measures remain elusive. The resulting heterogeneity in data prevents best practice from being defined.Reference Takata, Hopkins, Alexander, Bannister, Dalton and Harrison18 The management of necrotising otitis externa therefore remains a challenge.
Several reviewsReference Stapleton and Watson14,Reference Takata, Hopkins, Alexander, Bannister, Dalton and Harrison18 summarise antibiotic regimens used to manage necrotising otitis externa in published case series and highlight variation within these, but detailed analysis of these regimens and their outcomes is yet to be performed. This systematic review explores antimicrobial therapies and outcomes reported in published case series, aiming to identify the most effective approach to antimicrobial therapy in terms of medication choice, delivery method, dual versus single therapy, and empiric versus culture-guided therapy. This review also addresses reported outcomes regarding treatment efficacy and safety profiles of antimicrobial therapies.
Methods
The review was conducted according to the Preferred Reporting Items for Systematic Review and Meta-Analyses (‘PRISMA’) guidelines.Reference Liberati, Altman, Tetzlaff, Mulrow, Gøtzsche and Ioannidis19 The review was pre-registered on the PROSPERO database (ID CRD42022353244).
The search terms used are outlined in Table 1. These include all variations of ‘necrotising otitis externa’ identified through a pilot search. Electronic literature searches were performed in PubMed (9 October 2022; 1945–2022), Embase via OVID (7 October 2022; 1974–2022), CINAHL Plus (9 October 2022, no date limitations) and the Cochrane Library databases (9 October 2022; no date limitations; English-only publications).
Table 1. Search terms used to complete literature searches
Eligible studies included the use of diagnostic criteria for necrotising otitis externa and included at least one outcome measure in the context of patients receiving antimicrobial therapy for necrotising otitis externa. Studies were limited to adult patients. Randomised, controlled trials, case control studies, and multi- and single-centre case series with 10 or more patients were included. Case reports and case series with fewer than 10 patients were excluded, as were review articles, opinion pieces and paediatric and animal studies.
Two researchers independently screened all abstracts and full-text papers, with review by a third author in the event of non-agreement. Duplicates were removed. Titles and abstracts were screened using the inclusion and exclusion criteria. If further information was needed to determine whether abstracts were eligible, full texts were also screened. Short-listed studies underwent full text assessment to exclude ineligible studies. Of the studies selected for full text assessment, reference lists were screened to identify relevant studies not detected by the search. Additional papers were also identified using the Google search engine free text search. Data were extracted from full manuscripts for analysis (summarised in Table 2). Statistical analysis included calculating the total number of patients and/or the percentage of patients pertaining to each data item, where appropriate. If this was not possible, the total number of studies addressing each variable of interest was calculated. The average total duration of treatment across studies was calculated.
Table 2. Data items for extraction from selected studies
Risk-of-bias assessments were completed using the Risk of Bias Assessment of Non-randomised Studies (RoBANS) criteria. Certainty of evidence was assessed using the Grading of Recommendations, Assessments, Development and Evaluation (GRADE) criteria.
Results
A four-phase Preferred Reporting Items for Systematic Review and Meta-Analyses (‘PRISMA’) flow diagram displays the literature search (Figure 1). Of the 548 records that were identified, 393 abstracts were screened and 72 articles were assessed for eligibility. Twenty-eight articles were included in the final synthesis.
Figure 1. Preferred Reporting Items for Systematic Review and Meta-Analyses (‘PRISMA’) flow diagram displaying the systematic search methodology. CINAHL = Cumulative Index to Nursing and Allied Health Literature
Study characteristics
The characteristics of each study, including the first author, year of publication, study type and number of patients, are outlined in Table 3. Four cohort studies, 19 case series, one descriptive study, two quasi-experimental studies and two retrospective case reviews were identified.
Table 3. Characteristics of studies included in the final synthesis
Risk-of-bias assessment
The risk-of-bias assessment for each study is outlined in Table 4. All studies rated ‘high’ for selection bias in participant selection due to the lack of control groups. All studies also rated ‘high’ for intervention measurements due to the nature of necrotising otitis externa diagnosis (lack of best-evidence guidelines) and retrospective diagnosis. With the exception of a study by Franco-Vidal et al.,Reference Franco-Vidal, Blanchet, Bebear, Dutronc and Darrouzet42 studies also rated ‘high’ for selection bias in confounding variable control. The majority of studies rate ‘low’ for detection and attrition bias because, although studies did not have blinding, the outcome measures were deemed unlikely to be influenced by this, and reasons for missing data were provided. Reporting bias ratings were variable. In studies with a ‘high’ rating for reporting bias, outcomes were reported incompletely. As a result, meta-analysis could not be performed.
Table 4. Risk of bias assessment results for studies included in the final analysis
Certainty of assessment
Sixty-one per cent (17/28) of studies were deemed to have ‘low’ overall certainty, meaning that the true effect is likely to be markedly different from the presented effect (Table 5). Thirty-six per cent (10/28) of studies were deemed to have ‘moderate’ overall certainty, meaning that the true effect is likely to be close to the presented effect. One study was deemed to have ‘very low’ overall certainty, meaning that the true effect is likely to be markedly different from the presented effect.
Table 5. Certainty assessment results for studies included in the final analysis
Antibiotic classifications
Seventeen different antibiotic classes were identified in the included studies (Table 6). Fluoroquinolones were the most prescribed antibiotic class (68 per cent), with ciprofloxacin the most used antibiotic (64 per cent).
Table 6. Antibiotic usage by class
Antifungal therapy
Antifungal therapies are mentioned in several studies, delivered empirically in cases of non-response to antibiotic therapyReference Hasibi, Ashtiani, Motassadi Zarandi, Yazdani, Borghei and Kuhi36,Reference Soudry, Hamzany, Preis, Joshua, Hadar and Nageris41 or guided by positive fungal culture.Reference Stern Shavit, Soudry, Hamzany and Nageris5,Reference Le Clerc, Verillaud, Duet, Guichard, Herman and Kania9 However, outcome reporting was inconsistent, therefore outcomes could not be compared, particularly because patients often received a variety of therapies during their treatment.
Antimicrobial combinations
Due to incomplete and heterogeneous reporting of data, exact antimicrobial combinations could not be identified in all studies (Table 7). Most patients received dual therapy first line (57 per cent), and monotherapy second line (23 per cent).
Table 7. Antibiotic combinations
Route of antimicrobial delivery
Most patients received intravenous therapy first line (82 per cent), and oral therapy second line (31 per cent), as summarised in Table 8.
Table 8. Route of antibiotic delivery
Setting of antimicrobial delivery
Antimicrobials were delivered most often in the in-patient setting (50 per cent). In 45 per cent of participants, this was followed by the out-patient setting (Table 9).
Table 9. Setting of antibiotic delivery
Basis for antimicrobial delivery
Half of the studies (50 per cent) involved empirical antimicrobial prescription, with regimens amended based on culture results, as outlined in Table 10.
Table 10. Basis for antibiotic delivery
Treatment duration
The mean duration of total antimicrobial treatment was 65 days, based on treatment duration detail provided by 10 studies (Table 11). There was wide variation, with mean treatment durations ranging from 21 to 122 days across the included studies.
Table 11. Antibiotic treatment duration
Criteria for cessation of antimicrobial treatment
Ten studies reported on criteria used for cessation of antibiotic treatment (Table 12). These varied across the included studies, but most included clinical and radiological markers.
Table 12. Criteria for the cessation of antibiotic therapy
PET = positron emission tomography; EAC = external auditory canal; VAS = visual analogue scale; CRP = C-reactive protein; ESR = erythrocyte sedimentation rate
Side effects
Seven studies reported side effects of antimicrobial therapy for necrotising otitis externa (Table 13). Of 196 patients in studies where side effects were reported, the most common were allergy (9 per cent), renal impairment (9 per cent), hepatotoxicity (7 per cent) and leukopenia (6 per cent), with only one published study focusing on the complications of antimicrobial therapy for necrotising otitis externa.Reference Ijaz, Williams, Cole and Watson23
Table 13. Side effects of antibiotic therapy for necrotising otitis externa
Deaths
Of the 20 studies reporting mortality (n = 708 patients), 19 per cent of patients died (n = 131/708) within the variable stated follow-up periods. Five per cent of deaths were attributed to necrotising otitis externa (34/708). No deaths were attributed to complications of antimicrobial therapy.
Disease progression, relapse and readmission
Seven studies reported disease progression (n = 280 patients). Fourteen per cent of patients (40/280) experienced disease progression whilst receiving antibiotic therapy. In the 16 studies reporting relapse (n = 702 patients), 7 per cent (51/702) of patients experienced at least one relapse following cessation of antibiotic therapy. In the five studies reporting on readmission (n = 153 patients), 34 per cent (52/153 patients) of patients required at least one readmission to hospital after discharge.
Correlation between antimicrobial therapy and outcomes
Due to the heterogeneity of reported data regarding antimicrobial therapy and outcomes, in addition to the results of the bias and certainty assessments, it was not possible to form any conclusion regarding optimal treatment regimen or strategy in the treatment of necrotising otitis externa.
Discussion
The aim of this systematic review was to summarise antimicrobial regimens used to manage necrotising otitis externa reported in published literature, and to identify most effective and safe antimicrobial regimens and strategies. Due to insufficient and heterogeneous variable reporting, the most efficacious and safe regimens could not be determined, and an accepted process for treatment cessation could not be identified.
Complete antimicrobial therapy details could not be identified due to underreporting of drug combinations, doses used and the reasons for these. Nevertheless, this review identified that fluoroquinolones were the most prescribed antimicrobial (68 per cent), with ciprofloxacin being most used. Most patients received dual therapy first line (57 per cent), and monotherapy second line (23 per cent).
Prior to the introduction of ciprofloxacin, patients with necrotising otitis externa traditionally were treated with in-patient intravenous dual therapy, consisting of an aminoglycoside and semisynthetic penicillin.Reference Berenholz, Katzenell and Harell16 Since the introduction of ciprofloxacin in the 1980s, patients have been treated largely with ciprofloxacin as a first-line choice.Reference Lang, Goshen, Kitzes-Cohen and Sadé43,Reference Sadé, Lang, Goshen and Kitzes-Cohen44 Increasingly, dual therapy with both ciprofloxacin and an additional antibiotic to cover ciprofloxacin-resistant strains of pseudomonas is chosen as a first-line treatment.Reference Berenholz, Katzenell and Harell16 Notably, Pulcini et al. reported better prognosis with dual therapy, in comparison to monotherapy.Reference Pulcini, Mahdyoun, Cua, Gahide, Castillo and Guevara48
The incidence of ciprofloxacin-resistant strains of pseudomonas in aural swabs has remained stable at 7–20 per cent over the past two decades.Reference Linton and Stapleton6
Regarding delivery method, most patients received intravenous (82 per cent) therapy first line, and oral (31 per cent) therapy second line. Treatment tended to be initiated in the in-patient setting (95 per cent) and continued in the out-patient setting (50 per cent). Since the introduction of ciprofloxacin, which has good oral bioavailability, increasing numbers of patients are managed in the out-patient setting.Reference Lang, Goshen, Kitzes-Cohen and Sadé43,Reference Sadé, Lang, Goshen and Kitzes-Cohen44 However, evidence to support decisions regarding intravenous versus oral treatment, or switch between the two, is lacking. There are no studies that directly compare the efficacy of oral and intravenous antibiotic treatment. Nevertheless, early studies report that in comparison to intravenous ciprofloxacin, oral ciprofloxacin is associated with fewer side effects.Reference Sadé, Lang, Goshen and Kitzes-Cohen44 Moreover, studies suggest that intravenous monotherapy produces comparable outcomes to co-administration of antipseudomonal penicillin and aminoglycoside antibiotics.Reference Sadé, Lang, Goshen and Kitzes-Cohen44,Reference Meyers, Mendelson, Parisier and Hirschman49 Future research could usefully focus on comparing monotherapy, dual therapy, and oral and intravenous antibiotic treatment through double-blind prospective clinical trials.
Necrotising otitis externa patients tend to be frail, immunosuppressed, and have medical co-morbidities.Reference Bisbinas and Stapleton1 Necrotising otitis externa patients also tend to receive prolonged parenteral antimicrobial treatment. It is therefore unsurprising that 95 per cent of studies in this review involved delivery of at least initial courses of treatment in an in-patient setting. Non-in-patient settings identified in this review were variable. In 45 per cent of studies, in-patient treatment was followed by treatment in the out-patient setting, and in one study (5 per cent), treatment was solely delivered in the out-patient setting.Reference Durojaiye, Slucka and Kritsotakis22 The latter assessed home-administered intravenous treatment, which was proposed as an alternative to in-patient treatment in resource-poor settings.
Out-patient antibiotic therapy studies for necrotising otitis externa are surprisingly limited, especially since this is a commonplace practice. Out-patient antibiotic therapy has been shown to be highly effective in the management of other infections.Reference Durojaiye, Bell, Andrews, Ntziora and Cartwright50 Failure of out-patient antibiotic therapy in necrotising otitis externa has been associated with extensive disease on imaging, facial-nerve involvement, and medical co-morbidities.Reference Durojaiye, Slucka and Kritsotakis22 Although the delivery of out-patient antibiotic therapy and home-based treatment could be limited by the need for training and education, they remain cost-effective options for the management of necrotising otitis externa.Reference Durojaiye, Bell, Andrews, Ntziora and Cartwright50
Most included studies (50 per cent) initiated treatment on an empirical basis, and amended treatment based on microbiological culture results. Although this is the most popular approach, reporting of outcomes was not sufficiently robust to identify whether this approach was more efficacious or safe than other strategies. Notably, microbiological sampling techniques varied between studies and are discussed in several studies. Culture-negative cases of clinically and radiologically diagnosed necrotising otitis externa are well recognised,Reference Chaabouni, Achour, Yousfi, Thabet, Trigui and Kallel20 and may be explained by prior antibiotic use, fungal necrotising otitis externa or inadequate sampling technique.Reference Gruber, Roitman, Doweck, Uri, Shaked-Mishan and Kolop-Feldman51 In culture-negative and non-resolving cases, deep tissue sampling is increasingly being usedReference Rosenfeld, Schwartz, Cannon, Roland, Simon and Kumar52 to optimise antimicrobial therapy choices and strategies.
Mean antimicrobial treatment duration identified in this review was approximately nine weeks. A median could not be calculated due to limited and variable reporting of treatment duration between studies, reflecting the lack of consensus on optimal duration. Although some studies documented treatment duration, they did not provide explanations for treatment durations, or variations between these within their own patient cohort.
It would be ideal to identify optimal treatment duration, or criteria for treatment cessation, because extended courses of antimicrobials such as fluoroquinolones can be associated with complications. For example, prolonged use is associated with side effects such as tendon rupture and clostridium difficile infection,Reference Durojaiye, Slucka and Kritsotakis22 and may contribute to increased pseudomonal resistance.Reference Berenholz, Katzenell and Harell16,Reference Durojaiye, Slucka and Kritsotakis22 This is concerning because pseudomonas has intracellular efflux mechanisms that reduce antibiotic penetration, rendering it a challenge to eradicate.Reference Berenholz, Katzenell and Harell16 Although challenging to attribute specific antimicrobial regimens due to heterogeneous reporting, it is clear from this review that the side effects of antimicrobial therapy are common and can be serious.Reference Ijaz, Williams, Cole and Watson23
Criteria for cessation of treatment were provided by 10 studies. These included variable combinations of disease resolution based on clinical symptoms and examination, imaging and inflammatory markers. Inflammatory markers must be used with caution in immunocompromised populations because they can be unreliable.Reference Pizzo53–Reference Moallemi, Niroomand, Tadayon, Forouzanfar and Fatemi55 Due to insufficient and heterogeneous variable reporting, an accepted process for treatment cessation could not be identified.
All-cause mortality was reported by 71 per cent of included studies, with a mean of 19 per cent, although follow-up periods varied. Disease-specific mortality was estimated as being five per cent across studies. These figures may be unreliable for several reasons, including inadequate reporting across studies, variability in patient follow-up duration and disease-specific mortality identification in a frail patient population with multiple co-morbidities.
Outcomes such as disease progression, relapse and readmission were reported variably between studies, rendering it impossible to compare outcomes or to pair outcomes with treatment strategies, either within or between studies. Furthermore, these terms were variably defined between studies, if indeed they were defined at all. One study reported that relapse is not associated with antibiotic treatment duration, but rather, complex disease.Reference Hodgson, Sinclair, Arwyn-Jones, Oh, Nucken and Perenyei24 Some studies reported that fungal infections and/or antibiotic-resistant organisms can contribute to advanced disease.Reference Le Clerc, Verillaud, Duet, Guichard, Herman and Kania9,Reference Hobson, Moy, Byers, Raz, Hirsch and McCall56 Other studies report no association between pathogens and outcomes.Reference Loh and Loh15,Reference Hobson, Moy, Byers, Raz, Hirsch and McCall56
Strengths and limitations
This systematic review is the first to assess antimicrobial regimens and associated clinical outcomes in necrotising otitis externa. Date restrictions were not applied, to allow a comprehensive review to be completed. The review identified numerous areas requiring attention and research. The low quality of published data on necrotising otitis externa therapies and outcomes is a major limitation. Due to the wide variability and inconsistent data reporting, it was not possible to attribute outcomes to treatment regimens, preventing the identification of optimal therapy choices in terms of efficacy and safety. It is likely that duplicated patient data exists in the literature, with separately published case series arising from the same centre and authors, with overlapping time periods. The role of antifungal therapies could not be analysed due to lack of robust reported data.
Published reviewsReference Stapleton and Watson14–Reference Takata, Hopkins, Alexander, Bannister, Dalton and Harrison18 have highlighted a lack of quality research in necrotising otitis externa. Existing literature consists largely of uncontrolled, retrospective case series with high bias and low certainty of assessment results. The limitations of this review highlight a further valuable perspective – that robust and consistent outcome reporting is essential for analysis of the efficacy and safety of treatment regimens. Our team recently have completed a collaborative national Delphi study to reach consensus on the diagnostic criteria and core outcome set for necrotising otitis externa. It is hoped that this will improve outcome reporting in future studies and facilitate optimal management strategies to be identified through consistency in reporting, as well as facilitate future prospective clinical trials focused on comparing and identifying the safest and most effective treatment strategies.
Conclusion
Heterogeneous, inadequate and variable reporting of treatment strategies and clinical outcomes in necrotising otitis externa research mean that the most effective and safe regimens cannot be determined, and an appropriate process for treatment cessation cannot be identified. The lack of primary research on this theme, together with inconsistencies in reporting results from single-centre case series, are barriers to identifying best practice.
Robust, consistent reporting of clinical outcomes is important, and will be facilitated by the publication of a core outcome set for necrotising otitis externa. Whilst consensus processes can be useful in the identification of best practice regarding the initiation, cessation, and modification of treatment, and disease monitoring in the absence of high-level evidence, robust clinical trials are indicated to definitively answer the question that repeatedly arises in necrotising otitis externa publications: What is the optimal treatment strategy for necrotising otitis externa?
Acknowledgements
This research was supported by the National Institute for Health Research Manchester Biomedical Research Centre.
Financial support
No funding was received for this work.
Competing interests
The authors declare none.
Data availability statement
All systematic review data is included within the manuscript.
Authorship statement
ES designed the work; SL, ST and ES acquired and analysed data, drafted, revised and approved the manuscript; all authors agree to be accountable for all aspects of the work.
