Outcomes of primary myringoplasty in indigenous children from the Northern Territory of Australia

International Journal of Pediatric Otorhinolaryngology 127 (2019) 109634

Contents lists available at ScienceDirect

International Journal of Pediatric Otorhinolaryngology journal homepage: www.elsevier.com/locate/ijporl

Outcomes of primary myringoplasty in indigenous children from the Northern Territory of Australia

T

Stephen O'Learya,b,∗, Amelia Darkea, Kathy Curriec, Katie Ozdowskaa, Hemi Pateld a

Department of Otolaryngology - Surgery, University of Melbourne, Royal Victorian Eye and Ear Hospital, Australia Royal Victorian Eye and Ear Hospital, Melbourne, Australia c Department of Health, Northern Territory Government, Australia d Royal Darwin Hospital, Northern Territory Government, Australia b

ARTICLE INFO

ABSTRACT

Keywords: Tympanoplasty Myringoplasty Indigenous First nations Australian Aboriginal

Aim: To report the surgical and audiological outcomes of myringoplasty (Type I tympanoplasty) in Indigenous Australian children living in remote and regional communities in northern Australia. Method: An observational cohort study, with prospective recording of the details of surgery. Audiological outcomes were collected independently, and these data were integrated in the present study. Children aged 5–18 year underwent myringoplasty in the Northern Territory during a program initiated by the Australian Government. Surgery was performed by surgeons drawn from across Australia. Results: 412 primary myringoplasties were performed. The mean age at surgery was 11 years. The tympanic membrane was closed in 64.2% of cases. Fascial grafting was associated with greater surgical success than cartilage. Dryness of the ear at surgery did not affect drum closure. Post-operative aural discharge was half that reported in historical literature. Surgical success was independent of the patient's age at surgery. Post-operative audiograms were available on 216 cases. At last review, hearing had improved even when the operation was not a surgical success, with hearing aid candidacy falling from 84 to 34%. Hearing was similar irrespective of the size of the perforation at surgery or the graft used and did not change with the time between surgery and review. The best hearing was associated with drum closure and Types A or C tympanograms. A conductive hearing loss persisted after surgery that was greater when there was an immobile drum. Conclusions: Indigenous children benefited from myringoplasty, even when the operation was not a “surgical success” as deemed by drum closure. There lower incidence of post-operative discharge from persistent perforations suggests an improvement in the ear health of the population. A persistent conductive loss persists, likely a consequence of the underlying disease but possibly from the surgery.

1. Introduction

these circumstances patient has benefited is not well documented. It may be that drum-closure is too narrow a definition of the success of the operation, and that a broader view based upon aural discharge and hearing improvement should be considered. Here we report on the surgical and hearing outcomes of a large cohort of Indigenous Australian children, who underwent myringoplasty during a governmental program from 2008 to 2010. All recipients lived in regional or remote communities in the Northern Territory of Australia. Otitis media is exceptionally prevalent in this population. The post-nasal space is colonised with pathological bacteria within the first 3 months of life [8], most children have experienced otitis media by the age of one year, and a quarter have tympanic membrane perforations in early childhood [9]. Many of these

Myringoplasty (Type I tympanoplasty) is the treatment of choice for tympanic membrane perforation, as it prevents further aural discharge and improves the hearing. It is widely practiced to treat tympanic membrane perforation in Aboriginal and Torres Strait Islander peoples, but with mixed success. Surgeons typically judge the success of the operation on the basis of tympanic membrane closure, but the drumclosure-rates reported are typically low (around 30–70%) [1–5] compared with those reported in urban hospital-based practices where the median closure rate upon review is ~90 (ranging from 25 to 93%) [6,7]. A persistent perforation means that the ear could still experience aural discharge, the hearing may not improve and whether the under

⁎ Corresponding author. Chair of Otolaryngology, University of Melbourne, 5th floor, Peter Howson Wing, Royal Victorian Eye and Ear Hospital, 32 Gisborne St, Melbourne East 3002, Australia. E-mail address: [email protected] (S. O'Leary).

https://doi.org/10.1016/j.ijporl.2019.109634 Received 18 June 2019; Received in revised form 6 August 2019; Accepted 7 August 2019 Available online 13 August 2019 0165-5876/ © 2019 Elsevier B.V. All rights reserved.

International Journal of Pediatric Otorhinolaryngology 127 (2019) 109634

S. O'Leary, et al.

perforations persist, causing a conductive hearing loss and frequent aural discharge [10]. Hearing outcomes following myringoplasty are often reported as pure-tone averages. While this may be convenient, it obscures detail available that could provide insights into the success of the surgery and speech perception [11,12]. Specifically, an examination of the air-bone gap across frequency may provide insights into the likely presence of ossicular and/or tympanic membrane scarring [13], outcomes that may indicate whether the surgery could be improved. For this reason, we report frequency-specific audiometric results. There is diverse literature on the importance of graft types [14], perforation size [15–19], chronic active otitis media at the time of surgery [20–23], age at surgery [7,17,22,24–27] upon the success of myringoplasty. Much of this literature is derived from urban tertiary hospitals, so it cannot be assumed that conclusions drawn from this literature will apply in the remote and regional context, where the severity of otitis media is much greater. These surgical factors were explored in this study, to help provide guidance for their application to the treatment of Indigenous ear disease.

range of locations and at various times depending on the patient's community and access to follow-up services. The acoustic environment in which the audiogram was performed was recorded in all cases. Most reviews were scheduled, but some were received opportunistically if there was a nearby clinic running. Children were followed up one or more times. ENT assessments performed by surgeons in the field (including the authors) were collated prospectively by two of the authors (HP and KC) and entered into a database. Indices recorded included (a) The size of the tympanic membrane perforation (pinpoint, < 40% of the size of the tympanic membrane, > 40%, or sub/total perforation); (b) The presence or absence of aural discharge; (c) The dryness of the ear at surgery; (d) the surgical diagnosis at surgery, namely either chronic otitis media (COM)- Inactive, COM-active. At the post-operative assessment, the integrity of the tympanic membrane as was its mobility, the dryness of the ear and the clinical diagnosis (atelectatic drum, atrophic drum, COM healed. COM active, Dry perforation (COM -inactive), granular myringitis, intact drum, minimal perforation, moist perforation, tympanic membrane not viewed or child uncooperative, otitis media with effusion, retraction, or “too early” if the review was made too soon after the surgery for an assessment to be made. The timing and location of pre- and post-operative ENT assessments mirrored those described for audiological assessment. ENT and audiological assessments may, or may not have occurred on the same day, depending upon scheduling of visits of ENT and/or audiological teams to hospitals and communities.

2. Methods 2.1. Patient cohort The patients underwent myringoplasty as part of Australian Government's Northern Territory Emergency Response (NTER), a program that intended to reduce the burden of disease amongst Indigenous Australians living in the Territory. Patients requiring an ENT opinion were identified during the NTER's Child Health Check Initiative (CHCI). The clinical assessments, surgical recommendations, operations and follow-up were undertaken by teams of ENT surgeons and audiologists who visited hospitals in remote and regional Northern Territory, namely Alice Springs Hospital (ASH), Gove District Hospital (GDH), Katherine District Hospital (KDH) or Tennant Creek Hospital (TCH). The ENT surgeons involved in this work were not necessarily sub-specialised in either otology or Indigenous health, but all were given cultural awareness training before working with Indigenous people.

2.4. Ethics and data storage Clinical records of the myringoplasties and their outcomes were maintained by the NT Department of Health. Audiograms, in the form of a scanned copy of the paper record, were transcribed into digital format by author AD, using the Easy Audiometer iPad App written by Dr Luke Campbell (2014). Data release and transfer was overseen by of the Human Research Ethics Committee (HREC) of the Royal Victoria Eye and Ear Hospital (reference #16/1267HS), NT Department of Health/Menzies School of Health Research (reference #2016–2635) and the Central Australian Ethics Committee (reference #16–411). Potentially identifiable information on the audiological reports (the patient's name, carer's name and the patient's community) were removed from both surgical and audiological data prior to being released to the investigators.

2.2. Surgical approach All myringoplasties were performed between April 2008 and October 2010. Myringoplasty was undertaken by either a postaural, endaural or transcanal approach, depending on the preference of the surgeon. Similarly, the selection of graft material was judged on a caseby-case basis, and included temporal fascia, cartilage with or without perichondrium, Gelfoam™ or fat. The latter two grafting materials were grouped together because of the small numbers of each. The identity of the surgeon performing each operation was withheld from the investigators.

2.5. Data analysis All data were imported into MATLAB (R2018a, Mathworks, MA, USA) for the linking of surgical and audiological records, and recoding of variables. All analysis was coded in MATLAB for reproducibility, and this program was used for data association, graphical representation and statistical analysis. The terminology used to describe the surgical diagnosis of the ear at surgery and at review changed sightly after the last surgery but during the follow-up period and was not conducive to global analysis of these data given that several of the diagnostic categories overlapped. In order to account for this, the data were recoded (Table 1) into variables that reflected specific characteristics of the tympanic membrane, namely whether or not the ear was intact, dry, retracted or had a middle ear effusion. In the way, desired tympanic membrane appearances, such as dry (or wet) and intact or perforated (or intact) could be specified. The temporal disassociation between audiological and surgical reviews introduced some complexities into the analysis. For example, surgical reviews were documented for all patients, as were pre-operative audiograms, but post-operative audiograms were available on only a sub-set of those undergoing surgery. The dates of surgical and

2.3. Audiometric and ENT assessment Myringoplasty outcomes were assessed in two distinct ways: through audiological reviews performed by audiologists both before and after surgery, and surgical follow-ups performed by ENT specialists. The last surgical review was in June, 2016. Audiological assessment involving otoscopy, pure tone audiometry and tympanometry was undertaken by registered audiologists. The majority of myringoplasty patients underwent pre-operative audiological assessment in hospital immediately prior to surgery. For some, the assessment occurred at a hospital or in a community-based clinic, usually within a month of surgery, but always within a year. Audiograms were completed in either soundproof or non-soundproof conditions depending on the location of the testing. All hospitals and many larger clinics had sound-proofed booths, but many some smaller communities did not. Post-operative audiometry was undertaken in a 2

International Journal of Pediatric Otorhinolaryngology 127 (2019) 109634

S. O'Leary, et al.

Table 1 Recoding of ENT diagnostic codes into categories describing the status of the tympanic membrane and middle ear. AOM – acute otitis media; COM – Chronic otitis media; perf. – perforation; TM – tympanic membrane. Codes after Feb. 2010

Alternative codes (before Feb 2010)

AOM with Perforation (perf.) COM MUCOSAL ACTIVEa (Discharging perf.) COM MUCOSAL INACTIVEb (Dry perf.) COM SQUAMOUS INACTIVE (Retraction) Eustachian tube dysfunction Healed perforation NORMAL NORMAL Post-operation Otitis media with effusion OTHER Myringitis a b

Tympanic Membrane Descriptors for analysis

minimal perf. atelectatic COM healed atrophic, dry intact No view of TM, too early to assess or uncooperative granular myringitis

numerical code

normal

intact

effusion

wet

retracted

1 2 3 4 5 6 7 8 9 10 11

FALSE FALSE FALSE FALSE TRUE TRUE TRUE TRUE FALSE no value FALSE

FALSE FALSE FALSE TRUE TRUE TRUE TRUE TRUE TRUE no value TRUE

TRUE FALSE FALSE FALSE FALSE FALSE FALSE FALSE TRUE no value FALSE

TRUE TRUE FALSE FALSE FALSE FALSE FALSE FALSE FALSE no value TRUE

FALSE FALSE FALSE TRUE FALSE FALSE FALSE FALSE FALSE no value FALSE

COM Mucosal Active and Chronic active otitis media are equivalent. COM Mucosal Inactive and Chronic inactive otitis media are equivalent.

audiological review were frequently not the same, and the time between surgery and review varied considerably between individuals. In order to make the best use of the data available, two main analyses were performed. The first was a detailed assessment of the surgical procedures, and demographic pre-, peri- and post-operative factors that may have been associated with these. The second analysis was performed in those patients in which both post-operative audiograms and surgical outcomes were available.

Table 3 Details of all myringoplasties performed.

3. Results 3.1. Demographics and outcomes of surgical procedures Surgical results were available from 412 primary myringoplasties. The majority of these cases were undertaken at three rural and remote hospitals; Alice Springs Hospital, Katherine District Hospital and Gove District Hospital, with a smaller proportion performed at Tennant Creek Hospital. These hospitals differ in their local geography. Alice Springs is in the deserts of central Australia. Katherine is approximately 300 km inland from the northern coast of the Territory, and Gove is situated close to the coast in East Arnhem Land. Myringoplasties were performed on a higher number of females (57.3%) than males. The mean age of patients was 11.0 ( ± 2.7) years, ranging from 5 to 18 years (Table 2). Details of the surgical procedures undertaken are detailed in Table 3. Similar numbers of patients underwent myringoplasty by the post-auricular, endaural and trans-canal approaches. Temporal fascia was most often chosen for the graft, followed by cartilage. The perforation size, as estimated by the operating surgeon, ranged from pinhole to subtotal. Approximately a third of the ears operated upon were wet on the day of surgery. The size of the perforation did not influence whether the ear was wet or dry at surgery (χ2 = 0.59, p = 0.89), but it did influence the graft type chosen (χ2 = 10.1,p = 0.006) with fat/Gelfoam, although used infrequently (2.7%) having been selected rarely in a discharging ear. For both wet and dry ears at surgery, cartilage and fascia were used

Surgical approach Postaural Endaural Transcanal Not recorded

113 105 125 69

Graft type Temporalis fascia Cartilage Fat/Gelfoam Not recorded

218 116 11 67

Perforation size Pinhole Under 40% Over 40% Subtotal/total Not recorded

9 95 132 96 80

Condition of perforation at surgery Dry Wet Not recorded

363 49 0

in similar proportions for perforations of < 40% of the drum, while fascia was more frequently used for larger perforations. Surgical success, defined as an intact tympanic membrane at review, was seen in 64.2%. The proportions of ears with either intact drums, dry or wet perforations was not related to the dryness of the ear at surgery (χ2 = 0.1, p = 0.95). The proportion of discharging ears, relative to dry ones, seen at follow-up did not change at different times throughout the year. 3.2. Surgical and audiometric outcomes in primary myringoplasty Audiograms were not available on all patients after surgery, and

Table 2 Demographics of myringoplasties. Age mean ± SD (range)

Male Female Total

ASH

GDH

KDH

TCH

All

11.9 ± 3 (5–18))

10.4 ± 2.3 (5–17)

11.6 ± 2.1 (7–16)

12.2 ± 2.6 (5–16)

11.0 ± 2.7 (5–18)

63 83 156 (35.5%)

60 76 142 (32.3%)

47 67 126 (28.6%)

4 12 16 (3.6%)

179 (43%) 233 (57%) 412

3

International Journal of Pediatric Otorhinolaryngology 127 (2019) 109634

S. O'Leary, et al.

Table 4 Surgical Success by hospital.

ASH KDH GDH TCH

Intact

Dry Perforation

Wet Perforation

missing

44 44 45 1

16 5 15 0

8 4 10 1

7 6 10 0

Hospitals and surgical success for patients with pre- and post-operative audiological data. When the tympanic membrane could not be observed, the outcome was classed as “missing” in the table. Legend: ASH – Alice Springs Hospital, KDH – Katherine District Hospital, GDH – Gove District Hospital, TCH – Tennant Creek Hospital.

The mean patient age of the matched cohort was 10.7 years, which was comparable to that of all surgical cases (p = 0.24, t-test). The percentage of myringoplasties undertaken at ASH, KDH, GDH and TCH were 34.6, 27.3, 37 and 0.9% respectively, and these did not differ significantly from all surgical cases (χ2 = 3.7, p = 0.29). In the matched cohort, 54.5% of patients were female. Of the tympanic membranes observed, the surgical success rate for this group was 69.4%, with 18.6% having a dry perforation, and 11.9% a wet perforation. If it is assumed that those ears not observed had failed, the success rate would be 62.0%. The success at each surgical site is summarised in Table 4. It is apparent that a greater proportion of patients had surgical success at KDH. The proportions of patients having wet and dry ears at surgery (Table 5) did not differ between the entire population and the audiologically matched cohort (χ2 = 1.52, p = 0.21). Neither did the perforation size (Table 5) (χ2 = 0.28, p = 0.96), the graft type selected (Table 5) (χ2 = 1.10, p = 0.58) or the surgical success (χ2 = 0.64, p = 0.72). The relations between graft choice and perforation size were similar between all patients and the matched groups. Audiometric outcomes for this matched cohort are summarised in Fig. 2. Patients were classified according to the surgical outcome of the myringoplasty, where the tympanic membrane was observed clinically at follow-up. It is apparent that hearing improved after surgery, irrespective of whether the operation succeeded in closing the tympanic membrane, or not (Fig. 2). These results are often described as pure tone averages, and the improvement of the 4-tone pure tone average in this study (PTA4) was 10.25 dB. But this analysis obscures the finding apparent in Fig. 2, that hearing gained were even greater at lower frequencies, where there was the largest potential for improvement.

Fig. 1. Pure tone threshold for all available data, and patients in which there was both a pre- and a post-operative audiogram available (the “matched” group).

because surgical follow-ups and hearing tests were done independently fewer patients had both surgical follow-up and post-operative hearing assessments. The air-conduction thresholds, taken from the last-available audiogram for all available data, are shown in Fig. 1a. It is apparent that the mean audiometric thresholds improved through the surgical intervention. These data are shown for patients who had both pre- and post-operative audiometry are available in Fig. 1b. The improvement in hearing is greater, illustrating the importance of comparing pre- and post- intervention audiometric outcomes within-patient’. Further analysis was undertaken on this “matched” cohort. (The data in Fig. 1 are also presented as pure-tone averages in 10 dB bands in Supplementary Fig. 1, to aid comparison with previous literature.). Detailed hearing and surgical analysis are reported in patients with both pre- and post-operative audiometry and a post-operative follow-up occurred. Two hundred at sixteen (216) cases met these criteria. When reporting the hearing outcomes, the pre-operative and the last-available audiograms were considered. Surgical success was derived from the last-available surgical follow-up, when the tympanic membrane was observed. In 23 of the follow-ups, the drum could not be assessed because either the child was uncooperative or the drum could not be examined, which meant that there were 193 cases available in which surgical success could be related to other factors, such as hearing outcomes or surgical technique. The dates of the post-operative surgical and audiological reviews were often not the same, given that these services were provided independently.

Table 5 Relationships between ear dryness, perforation size and choice of graft. Table 5(a)

Pinpoint

< 40%

> 40%

Subtotal

wet at surgery dry at surgery

1 5

8 47

9 64

6 40

Table 5(b) wet at surgery dry at surgery

Fat 0 7

Fascia 16 103

Cartilage 13 43

Table 5(c) Fat Fascia Cartilage

Pinpoint 3 0 3

< 40% 3 23 24

> 40% 0 54 13

Subtotal 0 36 7

Legend: Intraoperative surgical factors for patients with pre- and post-operative audiological data. (a) Perforation size (pinpoint to subtotal) against the dryness of the ear at surgery. Missing data rates increase because of the cross-tabulation. For this table is missing rate is 16.7%. (b) Graft type against dryness of the ear at surgery. Missing data: 15.7%. (c) Graft type against perforation size. Missing data: 23%. 4

International Journal of Pediatric Otorhinolaryngology 127 (2019) 109634

S. O'Leary, et al.

Fig. 2. Pure tone thresholds, bone conduction and the air-bone gap for patients classified according to the surgical outcome at their last clinical review.

Prior to surgery 84% of children would have met Australian Hearing Service (AHS)'s criteria for amplification in the operated ear. Following surgery 34% met the AHS criteria for a hearing aid. Hearing was best when the perforation was closed. The mean bone conduction thresholds were within normal limits, irrespective of the surgical outcomes of the operation (Fig. 2) and almost all of the hearing deficit could be accounted for by the air-bone gap (Fig. 2). The most remarkable observation was the magnitude of this persistent air-bone gap, which was observed even when the tympanic membrane was intact. The air-bone gap was greater in response to low-frequency stimulation, and least at 2 kHz. Although some of the audiological data in the database were collected in communities, and outside sound-proof booths, all of the data presented here were (per chance) made within a booth, so the acoustic environment in which the tests were performed was not a consideration in the interpretation of these data. In intact ears after myringoplasty, tympanometry was available in all but 18 cases. The air-conducted hearing was better in the 57 patients with a Type A tympanogram than those with Type B (Fig. 3). Because there were only 5 Type C tympanograms, a statistical comparison was made between mobile drums (tympanograms Types A and C) and immobile drums (Type B), where it was found that a mobile drum significantly improved air-conducted hearing (F(94,1) = 5.98, p = 0.016). Significance was not reached for the air-bone gap, possibly due to the smaller sample given that not all patients had bone conduction tested. The patient's age at surgery did not impact upon the post-operative observation of a mobile tympanic membrane (F(94,1) = 0.01, p = 0.95). The ages of patients with mobile and immobile drums did not differ significantly (p = 0.94, t-test) and immobile drums were observed across the age-range. The median time from surgery to the last follow-up was 534 days, with the 5–95% range covering 40–2086 days. The correlation between follow-up duration and the four-frequency pure tone average (PTA4) was not significant (R = −0.033, p = 0.61). The child's age at surgery had no bearing upon the pure tone average hearing (R = 0.11, p = 0.11). Pre-operative audiograms were similar for both wet and dry ears at operation (Fig. 4). Perforation size influenced the pre-operative hearing considerably, with poorer hearing being associated with larger

Fig. 3. Tympanograms of intact tympanic membranes at last follow-up.

perforations (Fig. 4). Ears chosen for fat or Gelfoam grafting had the best pre-operative hearing. Audiometric thresholds were similar in ears where fascia or cartilage were chosen (Fig. 4). Hearing at last follow-up was similar irrespective of the dryness of the ear at surgery, the size of the perforation at surgery or the graft type chosen (Fig. 5). This was reflected statically in the PTA4, which was not affected by dryness of the ear at surgery (t-test, p = 0.31), the perforation size (F(3,189) = 1.56, p = 0.20) or the graft type (F (2,195) = 0.28, p = 0.75) (Fig. 5). Similarly, none of these factors 5

International Journal of Pediatric Otorhinolaryngology 127 (2019) 109634

S. O'Leary, et al.

Fig. 4. Pre-operative audiograms for three surgical factors.

Fig. 5. Post-operative audiograms for three surgical factors. 6

International Journal of Pediatric Otorhinolaryngology 127 (2019) 109634

S. O'Leary, et al.

influenced the bone conduction thresholds or the air-bone gaps.

funded hearing-health programs running in the NT [10]. But these data, like those reported here, are prone to a selection bias. Both of these programs were referral-based, so they may not reflect prevalence, and in the case of myringoplasty differing opinions about operating on the discharging ear amongst participating surgeons may have affected patient selection. An observation that argues against a major effect of selection bias is that the proportion of children with wet ears at surgery was 29%, which was similar to that seen after surgery by Foreman [1]. We conclude that the most reasonable interpretation is that myringoplasties performed now are less likely to result in a discharging ear than before, and whether this is due to improved socioeconomic determinants, service delivery, or both, could not be gleaned from these data. The size of the tympanic membrane perforation did not influence the rate of closure of the tympanic membrane. This adds weight to the same finding in other recent studies [12,15–19], and represents a change from previous observations where healing was reported to be less likely with larger perforations [15–19]. Chronic active otitis media at the time of surgery did not influence the success of myringoplasty in this study. Several previous studies found the same result [20,21], but given the exceptionally high rate of perioperative aural discharge in the present population there were legitimate, albeit unsupported, concerns about poorer surgical success as other investigators had observed [22,23]. These results do not support the proposition that operating on wet ears may be harmful, through perpetuation of the infection or increasing the risk of graft failure. Neither do the hearing outcomes, that were improved even in discharging ears that did not achieve surgical success. In ears that were dry at surgery, fascia was used three time more often than cartilage in surgeries that resulted in an intact ear at followup. In wet ears fascia and cartilage were used with similar frequency, and with similar degrees of success. We expect that cartilage was used more often in wet ears because it was thought to be more robust than fascia. But why cartilage tended to fail more often in dry ears is difficult to understand. Previous literature suggests that tympanic membrane closure rates are good with either [30], but superiority of cartilage [14] was not found here. This paper supports recent literature suggesting that patient's age of surgery does not influence myringoplasty outcomes [12,16,31–33], a finding that contrasts with earlier literature [7,17,22,24–27].

4. Discussion Myringoplasty was found to be effective in improving hearing even when it was not a “surgical” success, namely that a tympanic membrane perforation persisted. In intact drums, the best hearing was observed when the tympanic membrane was mobile. Myringoplasty meant that the number of ears eligible for a hearing aid according to Australian Hearing criteria dropped from 84 to 34% with this intervention. A significant outcome of this study is that the performance of myringoplasty had a greater impact upon hearing than its surgical success. This applied even in the “worst case” scenario, when the postoperative surgical approach was a wet perforation. The reason for this is not apparent from the data collected, but we suspect the near-universal improvement in hearing with surgery results from a reduction in the size of the tympanic membrane perforation [13]. Regardless, the implication for clinical practice is that myringoplasty is of benefit to hearing, irrespective of whether the tympanic membrane is closed, or not. 4.1. Surgical success Surgical outcomes in this study compared favourably with previous reports from the same region (the Northern Territory). A drum-close rate was similar to the 70% reported by Foreman from the 1990's [1]. That study, like the present one had good rates of surgical follow-up. Much poorer rates of drum closure (~30%) have recently been reported in the northern communities of Western Australia [5]. The present investigation confirms that closure of the tympanic membrane, while within the range of outcomes previously reported [6], is poorer than that seen in the majority of series conducted in urban contexts. By way of contrast, another recent study reported a much higher rate of surgical success (84%) amongst Indigenous Australians [28], using the endoscopic “push-through” technique. It is not yet known whether this approach is superior in the remote Australian setting, or whether other peri-operative factors such the consistency of the surgical team or the quality of the follow-up were better. The present study found differing rates of surgical success amongst the participating hospitals. Higher rates of drum close or a dry perforation were achieved at Katherine Hospital (approximately 300 km inland in the north) than either Gove (a coastal town) or Alice Springs (a central Australian city). Outcomes amongst the 13 patients who had surgery at Tennant Creek (inland, between Katherine and Alice Springs) were particularly disappointing with only one graft taking. Variation in surgical success rates has been reported to differ considerably by place in remote Australia [1], and it is interesting to note that Katherine hospital had the best results in the previously study too. It seems unlikely that the geographic location can explain this, because both best and worst results were from desert communities. The main surgical difference identified was that fascia was used for most grafts at Katherine, while both fascia and cartilage were used in similar proportions at other sites. The authors are aware that the primary surgeon at this site was unchanged for both studies, which might suggest that individual surgical technique or after-care played a role. Other servicerelated factors (such as the quality of peri-operative care) or community-related factors (such as the nature of the swimming environment [29]) may have been of significance but were beyond the scope of the present study. The prevalence of wet ears at follow-up was lower in the current study (12%) than that reported previously (26%) [1]. One possible explanation is that the incidence of discharging ears (chronic active otitis media) has diminished in these remote Australian communities over the last 20 years. In keeping with this interpretation are reports from the Australian Institute of Health and Welfare that cases of CSOM have been falling amongst children referred into the government-

4.2. Hearing outcomes The post-operative hearing, reported from the matched cohort is expected to be representative of the unmatched cohort because the patient demographics, pre-operative status of the operated ears, choice of grafts and post-operative surgical success were similar across those with and without audiological follow-up. The time from surgery to the last audiogram varied from months to years in this series. The lack of any change in hearing over this duration of time suggests that hearing is stable in Indigenous children after myringoplasty. Similar stability has been reported previously [4]. The shortest observation period of around 40 days was very short, and substantially less than the median duration of 534 days, but to pre-determine a minimum follow-up period would have been an entirely arbitrary decision, given the lack of any correlation between the time to last follow-up and the pure-tone-average (R = −0.033, p = 0.61). In light of these considerations, inclusion of all audiological outcomes, irrespective of the duration of follow-up, seemed justified. As expected from the literature, the size of the tympanic membrane perforation was the main determinant of the severity of the pre-operative hearing loss [13,34]. Consistent with the outcomes of previous studies of myringoplasty in Indigenous Australian children [1,2,4,5], post-operative hearing was poorer than that reported in studies from urban hospitals. Pure tone and air-bone averages were 10 dB or more poorer than those reported in a meta-analysis of paediatric 7

International Journal of Pediatric Otorhinolaryngology 127 (2019) 109634

S. O'Leary, et al.

myringoplasties [7]. Compared with the scant frequency-specific data available [12], pure-tone thresholds were 7–10 dB poorer at all frequencies from 0.5 to 4 kHz. Despite these poorer than anticipated outcomes, the procedure was effective in improving hearing, more-so when the grafting was successful, but even when it was not, with 50% fewer individuals in whom a hearing aid needed to be considered. We conclude that the procedure is worthwhile from a hearing perspective, even when not a surgical success. The best hearing was achieved when the tympanic membrane was both intact and mobile following surgery, as anticipated, as these findings reflect restoration of drum function and coupling of the ossicular chain [11,13]. But only half of the intact drums were mobile. Given the chronic otitis media that precedes myringoplasty, which is often active even at the time of surgery, we speculate that drum immobility results from inadequate aeration of the middle ear, with or without fibrosis surrounding the ossicles [13]. Excessive thickening of the tympanic membrane could also contribute [13] to the persistent airbone gap. It is possible that otitis media with effusion underlies the Type B tympanograms, but this seems unlikely because of the absence of skew towards younger patients. Almost all of the residual hearing loss in these patients was conductive, and the extent of the conductive heating loss after myringoplasty is consistent with the results of previous reports involving the Indigenous Australian population [1,3–5,28], even in the recent series with higher rates of drum closure [28]. Most previous studies have reported the average air-bone gap only, but the data here have been reported across frequency. This provides some insights into the status of the middle ear in these children. That the air-bone gap greatest in response to low-frequency stimulation suggests failure of the tympanoplasty to restore ossicular coupling, which in health contributes 20–25 dB to hearing at frequencies below 1 kHz, and progressively less at higher frequencies [13,35]. As noted above, inadequate middle ear aeration, ossicular scarring or tympanic membrane thickening may underlie this. But which of these predominate, and whether this might be a consequence of the underlying chronic otitis media or the response of these ears to surgery is not yet known. None of the surgical factors studied (the dryness of the ear at surgery, the size of the perforation or the graft type) had any significant impact upon the hearing outcomes. Previously observations that cartilage gives superior hearing outcome [14] were not supported. From these observations we infer that similar hearing can be expected after surgery, irrespective of the perforation size or the choice of graft. These findings support our conclusion above that the residual hearing loss is due to factors, such as middle ear disease, that are beyond the scope of a simple myringoplasty.

4.4. Implications for service provision The reduction in hearing aid requirement following myringoplasty has implications for service design and delivery. The efficacy of myringoplasty in reducing the burden of hearing loss suggests that this is the preferred option to rehabilitate hearing, obviating the need for a life-time of hearing aid and battery provision and professional visits to fit and maintain them. When successful in restoring hearing, the surgical approach also eliminates risks of stigmatisation from wearing hearing aids, and persistent social and economic disadvantage if hearing loss persists because a hearing aid is not worn. Once a perforation is identified, there is a strong case for prioritising and resourcing myringoplasty, even for children under 10 years of age. Most ears after myringoplasty were dry, even when the perforation persisted. This is a departure from previous reports, where approximately 25% of ears were wet after surgery [1]. This, together with the improvement in hearing from the procedure, means that most patients derive benefit, which should positively influence investment in this surgical procedure as a way of sustaining long-term hearing improvement for Indigenous children. Furthermore, the counselling of patients' expectations, and surgeons’ interpretation of the risk of the procedure should be revised in light of these findings. Although current surgical practice is successful in improving hearing, the long-term hearing benefit would be improved further if the nature of the persistent conductive hearing loss were better understood and appropriate surgical mitigation strategies were developed. This is a field worthy of further research investment. Conflicts of interest The authors declare no conflicts of interest. Acknowledgements SOL was funded by a Practitioner Fellowship from the National Health and Medical Research Council of Australia. Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.ijporl.2019.109634. References [1] A. Foreman, Aust. J. Oto Laryngol. 3 (1999) 235–239 1999. [2] D. Mak, A. Mackendrick, M. Bulsara, H. Coates, F. Lannigan, D. Lehmann, L. Leidwinger, S. Weeks, Clin. Otolaryngol. Allied Sci. 29 (2004) 606–611 2004. [3] D. Mak, A. Mackendrick, S. Weeks, A.J. Plant, J. Laryngol. Otol. 114 (2000) 26–32 2000. [4] D.B. Mak, A. MacKendrick, M.K. Bulsara, S. Weeks, L. Leidwinger, H. Coates, F.J. Lannigan, D. Lehmann, Med. J. Aust. 179 (2003) 324–325 2003. [5] S. Soumya, A. Hinton-Bayre, H. Coates, E.H. Ooi, J. Kuthubutheen, Aust. J. Otolaryngol. 1 (2018) 18 2018. [6] S.S. Chandrasekhar, U. Devgan, U. Devgan, Arch. Otolaryngol. Head Neck Surg. 121 (1995) 873–878 1995. [7] J.T. Vrabec, R.W. Deskin, J.J. Grady, Arch. Otolaryngol. Head Neck Surg. 125 (1999) 530–534 1999. [8] A.J. Leach, J.B. Boswell, V. Asche, T.G. Nienhuys, J.D. Mathews, Pediatr. Infect. Dis. J. 13 (1994) 983–989 1994. [9] P.S. Morris, A.J. Leach, P. Silberberg, G. Mellon, C. Wilson, E. Hamilton, J. Beissbarth, BMC Pediatr. 5 (2005) 27 2005. [10] Australian Institute of Health and Welfare, Northern Territory Outreach Hearing Health Program: July 2012 to December 2016. Cat No, IHW 189, Canberra, 2017. [11] H.G. Choi, D.H. Lee, K.H. Chang, S.W. Yeo, S.H. Yoon, B.C. Jun, Clin. Exp. Otorhinolaryngol. 4 (2011) 126–130 2011. [12] D.T. Kent, D.J. Kitsko, T. Wine, D.H. Chi, JAMA Otolaryngol. - Head Neck Surg. 140 (2014) 106–111 2014. [13] S.N. Merchant, M.E. Ravicz, S. Puria, S.E. Voss, K.R. Whittemore Jr., W.T. Peake, J.J. Rosowski, Am. J. Otol. 18 (1997) 139–154 1997. [14] K. Onal, S. Arslanoglu, M. Songu, U. Demiray, I.A. Demirpehlivan, J. Laryngol. Otol. 126 (2012) 22–25 2012. [15] M.M. Carr, C.P. Poje, M.L. Nagy, M.P. Pizzuto, L.S. Brodsky, J. Otolaryngol. 30

4.3. Methodological factors These data were acquired through routine clinical practice, and due to the structure of the health system, surgical and audiological assessments were, in most cases, conducted independently. Incomplete data collection is commonplace in research in the remote Australian context, as it relies upon visiting health practitioners covering vast distances, servicing peoples who themselves need to travel hundreds of kilometres from their communities for medical assessment and treatment. The temporal disassociation of the audiometric and surgical outcomes explains why some patients with surgical follow up did not have a postoperative audiogram, given that these patients may not have been present in their communities when the audiological team visited. The difference in timing means also that the status of the ear could have differed between the last surgical review and the audiogram, although the lack of any trend in hearing or surgical success over post-operative time suggests that this will have had minimal effect upon the interpretation of these data. 8

International Journal of Pediatric Otorhinolaryngology 127 (2019) 109634

S. O'Leary, et al. (2001) 199–202 2001. [16] D. Merenda, K. Koike, M. Shafiei, H. Ramadan, Otolaryngol. Head Neck Surg. 136 (2007) 189–192 2007. [17] C.H. Raine, S.D. Singh, J. Laryngol. Otol. 97 (1983) 217–221 1983. [18] G.B. Singh, T.S. Sidhu, A. Sharma, N. Singh, Int. J. Pediatr. Otorhinolaryngol. 69 (2005) 1071–1076 2005. [19] Y. Uyar, B. Keleş, S. Koç, K. Öztürk, H. Arbaǧ, Int. J. Pediatr. Otorhinolaryngol. 70 (2006) 1805–1809 2006. [20] R. Caylan, A. Titiz, M. Falcioni, G. De Donato, A. Russo, A. Taibah, M. Sanna, Otolaryngol. Head Neck Surg. 118 (1998) 709–713 1998. [21] D. Ophir, M. Porat, G. Marshak, Arch. Otolaryngol. Head Neck Surg. 113 (1987) 1288–1290 1987. [22] R.R. Macdonald III, R.P. Lusk, H.R. Muntz, Arch. Otolaryngol. Head Neck Surg. 120 (1994) 138–143 1994. [23] K. Onal, M.Z. Uguz, K.C. Kazikdas, S.T. Gursoy, H. Gokce, Clin. Otolaryngol. 30 (2005) 115–120 2005. [24] J. Friedberg, T. Gillis, J. Otolaryngol. 9 (1980) 165–168 1980. [25] W.M. Koch, E.M. Friedman, T.J.I. McGill, G.B. Healy, Arch. Otolaryngol. Head Neck Surg. 116 (1990) 35–40 1990.

[26] S. Sarkar, A. Roychoudhury, B.K. Roychaudhuri, Eur. Arch. Oto-Rhino-Laryngol. 266 (2009) 627–633 2009. [27] L. Shih, T. De Tar, J.A. Crabtree, Otolaryngol. Head Neck Surg. 105 (1991) 74–77 1991. [28] T.Z. Loh, S. Rangius, H. Patel, G. Crossland, Aust. J. Oto Laryngol. (2018) 1–22 2018. [29] D. Lehmann, M.T. Tennant, D.T. Silva, D. McAullay, F. Lannigan, H. Coates, F.J. Stanley, Br. Med. J. 327 (2003) 415–419 2003. [30] A. Zwierz, K. Haber, A. Sinkiewicz, P. Kalinczak-Gorna, J. Tyra, J. Mierzwinski, Eur. Arch. Oto-Rhino-Laryngol. 276 (2019) 323–333 2019. [31] R. Albera, V. Ferrero, M. Lacilla, A. Canale, Ann. Otol. Rhinol. Laryngol. 115 (2006) 875–879 2006. [32] N. Umapathy, P.J. Dekker, Arch. Otolaryngol. Head Neck Surg. 129 (2003) 1053–1055 2003. [33] M. Yung, C. Neumann, S.L. Vowler, Otol. Neurotol. 28 (2007) 353–355 2007. [34] R.P. Mehta, J.J. Rosowski, S.E. Voss, E. O'Neil, S.N. Merchant, Otol. Neurotol. 27 (2006) 136–143 2006. [35] H. Kurokawa, R.L. Goode, Otolaryngol. Head Neck Surg. 113 (1995) 349–355 1995.

9