ORIGINAL PAPER

A model-based cost-effectiveness analysis of a grommets-led carepathway for children with cleft palate affected by otitis mediawith effusion

Syed Mohiuddin • Katherine Payne •

Elisabeth Fenwick • Kevin O’Brien •

Iain Bruce

Received: 7 August 2013 / Accepted: 8 May 2014

� Springer-Verlag Berlin Heidelberg 2014

Abstract There is a paucity of evidence to guide the

management of otitis media with effusion (OME), which is

a common problem causing significant hearing impairment

in children with cleft palate. The insertion of grommets is

currently being used to correct hearing impairment and

prevent complications of unmanaged OME, but there is

ongoing discussion about whether the benefits of grommets

outweigh the costs and risks. A decision-tree model was

developed to assess the surgical insertion of grommets with

two non-surgical alternatives (hearing-aids and do-nothing

strategies) in cleft palate children with persistent bilateral

OME. The model assumed a 2-year time horizon and a UK

National Health Service perspective. Outcomes were val-

ued using quality-adjusted life-years (QALYs) estimated

by linking utility values with potential hearing gains

measured in decibels. Multiple data sources were used,

including reviews of the clinical effectiveness, resource use

and utility literature, and supplemented with expert opin-

ion. Uncertainty in the model parameters was assessed

using probabilistic sensitivity analysis. Expected value of

perfect information analysis was used to calculate the

potential value of future research. The results from the

probabilistic sensitivity analysis indicated that the grom-

mets strategy was associated with an incremental cost-

effectiveness ratio of £9,065 per QALY gained compared

with the do-nothing strategy, and the hearing-aids strategy

was extended dominated by the grommets strategy. The

population expected value of perfect information was

£5,194,030 at a willingness to pay threshold of £20,000 per

QALY, implying that future research could be potentially

worthwhile. This study found some evidence that the

insertion of grommets to manage cleft palate children with

bilateral OME is likely to be cost-effective, but further

research is required to inform this treatment choice.

Keywords Otitis media with effusion � Cleft palate �Grommets � Hearing aids � Cost-effectiveness � Expected

value of perfect information

JEL Classification I10

Introduction

Otitis media with effusion (OME), also called glue ear, is a

common condition in children across the world. The con-

dition is particularly prevalent in children who are born

with cleft palate (CP). Among children with CP, most

OME cases tend to occur in under the age of 6 years [1–3].

It is universally accepted that approximately 90 % of

children with CP experience non-trivial OME [2, 4–6],

which is more likely to be a persistent problem in children

S. Mohiuddin � K. O’Brien � I. Bruce

The Healing Foundation Cleft and Craniofacial Clinical

Research Centre, University of Manchester, Manchester, UK

S. Mohiuddin (&) � K. Payne

Manchester Centre for Health Economics, Institute of Population

Health, University of Manchester, Oxford Road,

Manchester M13 9PL, UK

e-mail: syed.mohiuddin@manchester.ac.uk

E. Fenwick

Institute of Health and Wellbeing, University of Glasgow,

Glasgow, UK

K. O’Brien

School of Dentistry, University of Manchester, Manchester, UK

I. Bruce

Paediatric ENT Department, Royal Manchester Children’s

Hospital, Manchester, UK

123

Eur J Health Econ

DOI 10.1007/s10198-014-0610-8

with CP as opposed to non-CP children [2], owing to the

effects of palatal dysfunction on eustachian tube function.

OME is said to be persistent when effusion or fluid within

the middle ear space lasts for at least 3 months [7]. OME

can have an impact on a child in a number of ways. The

most obvious negative outcome is conductive hearing loss,

which can impair the child’s health-related quality of life

and have other knock on effects in terms of speech pro-

duction, language development and ability to learn [2, 3,

8]. The high incidence of OME in children with CP has led

to the suggestion that surgical treatment would ameliorate

the inevitable hearing loss and other symptoms, but the

benefits of the operation must be set against the costs of the

procedure and the risks of postoperative complications.

Besides, the decision to operate is not always clear-cut as

there is a possibility of spontaneous resolution of OME,

and hence, the use of health care resources may not be

necessary.

There are three potential actions available to a clinician

when a child with CP is diagnosed with bilateral OME and

persistence of the condition has been confirmed over a

period of 3 months. The first option is to consider an

‘extended period of watchful waiting’ (hereafter termed

‘do-nothing strategy’) in which no intervention is made [8–

10], but the child is treated as needed with an appropriate

course of antibiotics to manage emergent infection such as

in acute otitis media (AOM) [11–13]. This approach

assumes that OME will resolve spontaneously without

surgery. An alternative approach, effectively comparable to

the do-nothing strategy, involves giving the child hearing

aids (hereafter termed ‘hearing-aids strategy’) to help with

day-to-day activities and reduce the negative impact of

hearing loss on health-related quality of life. Another

option available is surgical intervention that involves the

insertion of ‘ventilation tubes’ or ‘tympanostomy tubes’

(hereafter termed ‘grommets strategy’). A grommet is a

very small plastic tube that is inserted into the tympanic

membrane (hereafter termed ‘eardrum’) under general

anaesthetic to equalise the pressure either side of the ear-

drum. Thereby, a grommet prevents the development of the

negative middle ear pressure resulting from eustachian tube

dysfunction, which is thought to be a central causal factor

in childhood OME. A qualitative study recently explored

the views of parents with a child who has OME and CP.

The findings from this study suggested that parents showed

a preference for grommets as a treatment strategy, and

their views were formed by varying perceptions about

anticipated risks and benefits of the available management

options [14]. However, with alternative options for man-

aging OME available, the implications for health care

resource use and health outcomes of the grommets strategy

should be weighed against these other management

options. There is some evidence to suggest that parents do

not consider grommets and hearing aids as equal treatments

[14], but currently, there is no clear evidence to show

whether grommets or hearing aids are the most effective

option for children with CP [3, 7].

Armstrong [15] first described the use of grommets for

the management of chronic OME some 60 years ago.

Despite the growing use of grommets over an extended

period of time, there remain different opinions regarding the

benefits, costs and risks of this surgical procedure [16, 17].

Likewise, there are substantial differences of opinion

amongst specialists who treat children with CP about

whether the potential benefits of grommets outweigh the

risks and costs of repeated surgical intervention. This

reflects concerns regarding the chronicity of OME in chil-

dren with CP and apprehension about the consequences of

repeated surgery on the long-term health of the eardrum.

This variation is reflected across the specialist centres

treating children with CP in the UK. However, there is

limited evidence from clinical trials to inform whether the

grommets strategy is effective [3]. This is due to the chal-

lenges in conducting such trials in children and the uncer-

tainty about what are the best outcomes to use in such trials.

There is some previous economic evidence that explored

whether the use of grommets was a cost-effective use of

health care resources, but these evaluations were in children

without CP [18–20]. In 2008, a guideline was published on

the surgical management of OME in children, commis-

sioned by the National Institute for Health and Clinical

Excellence [now called the National Institute for Health and

Care Excellence (NICE)] and prepared by the National

Collaborating Centre for Women’s and Children’s Health

(NCCWCH). This guideline recommended a general care

pathway of watchful waiting for 3 months followed by the

surgical insertion of grommets or the provision of hearing

aids [7]. As part of preparing the guideline, a de novo

decision-analytic model was structured to identify the

incremental costs and benefits of four strategies (hearing

aids, grommets, grommets plus adjuvant adenoidectomy,

and do nothing) for the management of OME persisting

after 3 months of watchful waiting. However, neither the

recommendations from the guideline nor the results of the

previously published economic evaluations can be gener-

alised to a population of children with CP.

The main aim of this study was to perform a model-

based economic analysis to assess the potential impact of

the surgical insertion of grommets as compared with non-

surgical alternatives, such as the hearing-aids and do-

nothing strategies, in the management of bilateral OME

persisting after watchful waiting for 3 months in children

with CP. This analysis is an early economic evaluation, as

defined in Sculpher et al. [21], because the model was

populated with the current available evidence base. This

model-based economic analysis is part of a larger study

S. Mohiuddin et al.

123

funded by the UK’s National Institute for Health Research

Health Technology Assessment Programme called the

Management of Otitis Media with Effusion in Children

with Cleft Palate (MOMENT) study. MOMENT is a fea-

sibility study to understand whether there is a need for

further research regarding the use of grommets in children

with CP.

Methods

A decision-analytic model was developed to determine the

incremental cost-effectiveness associated with the grom-

mets strategy compared with the hearing-aids and do-

nothing strategies for the management of persistent bilat-

eral OME in children with CP. The model followed a

hypothetical cohort of 10,000 children under the age of

12 years with CP and persistent OME. This age cut-off was

used since the ear problems appear to settle in children with

CP after the age of 12 years [2], and as children grow

older, the incidence [1] and consequences [22] of the

condition decline. The model assumed hearing gain as the

primary clinical outcome and assumed a UK National

Health Service perspective for costs. The model structure

was informed by a previously published model that had

been used to inform national guidance on the management

of children with OME [7]. The model represented care

pathways for a 24-month time horizon, compared with the

12-month time horizon used in the previously published

model [7]. The timeframe for the current model was based

on the advice from clinical experts (n = 3) in the

MOMENT Study Management Group and the literature

which indicated two reasons: (1) 24 months is a reasonable

follow-up period to detect key outcomes following an

intervention for OME [7, 23, 24]; and (2) the total number

of potential multiple insertions and retractions of grommets

could be performed within a 24-month period. All costs

and benefits incurred beyond 12 months were discounted at

an annual rate of 3.5 %, as recommended by the NICE

‘Guide to the methods of technology appraisal’ [25].

The decision-analytic model used a decision-tree struc-

ture, and a schematic of the care pathways represented is

shown in Fig. 1. Following the criteria for selecting an

appropriate modelling approach set out by Barton et al.

[26], a decision-tree structure was considered to be the

most appropriate modelling approach for this decision

problem since the children within the model are assumed to

be independent of each other, the care pathways outlined

following consultation with clinical experts can be ade-

quately represented by a probability tree and a short fol-

low-up time horizon was considered. The model was built

in Microsoft Excel� and programmed using Visual Basic

for Applications� to estimate the expected costs and

benefits for each strategy. The three strategies compared

were the grommets strategy, the hearing-aids strategy and

the do-nothing strategy. These strategies and the assump-

tions made are now described.

Grommets strategy

The grommets strategy assumes that children with CP are

recommended for the surgical insertion of grommets once

the persistence of OME has been confirmed over a 3-month

period. The model starts at the point when the first surgical

insertion of grommets occurs. Some children are assumed

to have a second insertion of grommets, and some of these

will undergo a third insertion because of persistent or

relapsing OME [2, 7], or early extrusion of the grommets

from the eardrum [3, 7]. To avoid the high frequency of

insertion in a single year, the second surgical insertion of

grommets is assumed to occur within the first year of

management follow-up, and the third surgical insertion is

assumed to occur within the second year of management

follow-up. From the literature [2, 7, 27], the maximum

number of insertions of grommets per child was limited to

three because in children who undergo several insertions of

grommets the risk of conductive hearing loss is increased

in the long run [28], and also to prevent scarring of the

eardrum from repeated operations.

The surgical insertion of grommets is commonly

accepted as a safe operation, but there remain possibilities

of some potential postoperative complications. Hence,

following each surgical insertion, there is some chance of

complications such as otorrhoea, granulation tissue for-

mation and eardrum perforation. These have been found to

occur proportionately on the basis of estimates extracted

from the literature (see Table 1). Fortunately, none of the

potential complications related to the grommet-insertion

operation are life-threatening. The risk of eardrum perfo-

ration is usually higher after repeated insertion of grom-

mets [23, 29]; thus, a higher risk is expected for eardrum

perforations in subsequent operations. Cholesteatoma (the

abnormal collection of skin cells) formation was not

included as a complication of OME because NCCWCH

[7], Kwan et al. [27], Fior and Veljak [30] and Vlastarakos

et al. [31] among many others suggested that insertion of

grommets can avert sequelae of OME such as cholestea-

toma formation. Although there may be a chance of cal-

cium deposition within the eardrum with subsequent

increased eardrum rigidity (tympanosclerosis) that could be

due to AOM [32], OME itself [33], or insertion of grom-

mets [16], no incidence of tympanosclerosis formation was

predicted as part of any of the strategies included in this

analysis given that the actual cause of tympanosclerosis is

not fully understood [34].

Cost-effectiveness analysis of grommets-led care pathway

123

No serious injury or surgical death was assumed to

occur since it is very unusual and extremely unlikely for

children experience serious injury or death from an

insertion of grommets under modern anaesthesia [35, 36].

Occasionally, there may be a need for the grommets to be

removed [7, 23], and the impact of this has been explored

Fig. 1 The decision tree. OME otitis media with effusion

S. Mohiuddin et al.

123

in the model as part of the complications. It is estimated

that children will have their first ear, nose and throat (ENT)

review within 6 weeks of an operation, and subsequent

ENT reviews every 26 weeks thereafter until the mean

‘extrusion time’ (i.e. the time by which the grommets

should naturally fall out) of 39 weeks [7]. It is also esti-

mated that children will require one or two audiological

reviews after each operation on the basis of the advice by

NCCWCH [7] that hearing levels of children who have

undergone the insertion of grommets for management of

OME should be reassessed postoperatively. The model

assumed that some children who have otorrhoea and/or

granulation tissue formation will need to visit a general

practitioner (GP) for a course of antibiotics or ear drops.

Hearing-aids strategy

This strategy assumes that children with CP are offered

hearing aids once the persistence of OME has been con-

firmed over a 3-month period. It is estimated that in some

children OME will resolve spontaneously by the end of

12 months [9, 10, 12]. However, the model assumed that

children in whom OME had not naturally resolved by

12 months will continue using their hearing aids in the

hope of spontaneous resolution without surgery. The initial

costs of this strategy include the hearing aids, batteries for

the hearing aids, ear moulds (to help fit the hearing aids

into a child’s ear and allow the amplified sound to enter the

ear canal), a hearing-aid care kit, and the fitting of hearing

aids in an audiology department. Batteries for hearing aids

are estimated to need replacing every 4 weeks [7]. In

addition, ear moulds are estimated to need replacing every

13 weeks [7] because ear moulds repeatedly turn yellow

and become inflexible with time, and hence, require

replacement at regular intervals. For some children, some

of these costs are expected to be incurred again owing to

breakage or loss of hearing aids [7].

AOM is a common sequela in children with OME, and if

the OME is left untreated, then it usually leads to episodes

of AOM that require treatment with an appropriate course

of antibiotics [11–13]. AOM is the commonest reason for

children to take antibiotics [11], and to be cured of the

AOM infection, the use of an antibiotic such as amoxicillin

has been highly cited in the literature [13, 37]. This strategy

included the use of antibiotics, and on the basis of the

meta-analysis conducted by Rosenfeld and Kay [12], an

average of 2.8 episodes of AOM [95 % confidence interval

(CI) 2.2–3.4 episodes] have been predicted to occur each

year. AOM is one of the foremost causes of physicians’

consultations [11], and prescribing antibiotics for AOM is

known to encourage GP visits for subsequent episodes [13].

Thus, the children with untreated OME in this strategy are

assumed to make 2.8 GP visits (on average) every year

because of AOM episodes [11–13]. It is estimated that

children will make one or two ENT visits every year [10,

37]. Furthermore, it is estimated that children will have

their first audiological review after 13 weeks and sub-

sequent audiological reviews every 26 weeks thereafter

[7]. Previous work has suggested that adherence to wearing

hearing aids is a problem [9, 10], whereby a child would

frequently remove the hearing aid. To reflect the impact of

the acceptability of wearing the hearing aids, the model

Table 1 Probability data for complications related to the insertion of grommets

Complication Probability Distribution for

PSA

Source Notes

Otorrhoea 0.25 Beta*(3, 9) Maheshwar et al. [9] In view of the incidence of otorrhoea, a more conservative

value has been used. The value of 0.25 reported (for the

CP population) by Maheshwar et al. [9] is in line with the

meta-analysis value of 0.26 reported (for the non-CP

population) by Kay et al. [29]. Maheshwar et al. [9]

conducted their retrospective study in the UK. Russell

et al. [51] claimed a slightly lower risk of otorrhoea for

the CP population than for the non-CP population

Granulation tissue

formation

0.042 Beta*(37, 850) Kay et al. [29] Granulation tissue formation is a postoperative

complication suggested by Kay et al. [29] and NCCWCH

[7] among others. The value of 0.042 is reported by Kay

et al. [29] from their meta-analysis

Eardrum perforation

(1st procedure)

0.024 Beta*(2, 81) Phua et al. [23] This was a retrospective study that was conducted recently

in New Zealand. This was the only identified study of

children with CP from the systematic review that reported

subsequent higher risk of perforations of the eardrum

owing to repeated insertion of grommets

Eardrum perforation

(C2 procedures)

0.098 Beta*(4, 37) Phua et al. [23]

CP cleft palate, NCCWCH National Collaborating Centre for Women’s and Children’s Health, PSA probabilistic sensitivity analysis

Cost-effectiveness analysis of grommets-led care pathway

123

also included an estimate of the level of adherence to

wearing the hearing aids and the associated impact on

quality-adjusted life-years (QALYs). Neither AOM infec-

tion nor non-adherence to hearing aids is shown on the

decision-tree diagram because neither of these impact on

the probability of spontaneous resolution.

Do-nothing strategy

The do-nothing strategy reflects extending the initial

watchful waiting period of 3 months by a further 24 months.

This strategy assumes no intervention is planned for children

with CP [8–10], but they are offered an appropriate course of

antibiotics to treat any emerging instances of AOM [11–13].

Similarly to the hearing-aids strategy, children in this strat-

egy are expected to experience 2.8 episodes of AOM (95 %

CI 2.2–3.4 episodes) every year [12]. As such, every year,

the children with untreated OME are assumed to make 2.8

visits to the GP (on average) because of AOM episodes [11–

13]. Furthermore, the children are assumed to require

ongoing contact with health care services, including one or

two audiological reviews [7, 9, 10, 37] and one or two ENT

visits [10, 37] every year. The model assumed that, apart

from the direct costs related to hearing aids and the need for

any subsequent audiological reviews, the resource conse-

quences of this strategy will effectively be similar to those of

the hearing-aids strategy.

Model inputs

The data used to populate the model were derived from a

variety of sources, including systematic reviews of clinical

effectiveness and existing economic evaluation literature,

and rapid reviews of resource use and utility literature. The

model was also supplemented with data from published

expert opinion where other evidence to populate the model

was not available. Ponduri et al. [3] conducted a systematic

review to identify all studies that reported on the associa-

tion between early insertion of grommets and subsequent

outcome in children with CP. Most of the studies identified

from their systematic review of the relevant clinical liter-

ature were judged to be of low quality. The main chal-

lenges were that the identified studies were small, without

sample size calculations and generally had poor reporting

of data. It was not possible to perform a meta-analysis to

estimate an overall measure of clinical effect because of

study heterogeneity; thus, it was necessary to select the

articles deemed to have most direct relevance to the study

population of interest. The model inputs in terms of

probabilities, clinical effectiveness, utility values, resource

use and unit costs are now described.

The probabilities identified for each aspect of the care

pathway associated with the grommets strategy (and

sources of data) are shown in Table 1 (for complications

related to grommets) and in Table 2 (for insertion of

grommets).

For the hearing-aids strategy, a probability for breakage

or loss over a 12-month period of 16.44 % was used within

the model. This was calculated from estimates that 25 % of

children break or lose their hearing aids over a period of

21 months, presented by NCCWCH [7], under the

assumption of a constant hazard. Rosenfeld and Kay [12]

reported a meta-analysis that generated a value for spon-

taneous resolution of chronic OME documented for

3 months or longer, and in line with this, the model pre-

dicted a spontaneous resolution rate of 30.8 % by the end

of the first 12 months for children receiving the non-sur-

gical strategies. The resolution of OME was assumed to

have a constant rate over 12 months, following advice from

clinical experts and the evidence base. In calculating the

QALY gain associated with the hearing-aids strategy, the

model assumed that 90.9 % of the cohort of children will

adhere to wearing their hearing aids, on the basis of a

published non-adherence rate of 9.1 % [10]. The 9.1 % of

children who do not adhere to wearing their hearing aids

are then assumed to have gain in QALY equivalent to that

in the do-nothing strategy.

Hearing level in decibels (dBHL) over two predefined

time periods of 12 and 24 months was used to value the

impact of each strategy on children with CP and OME.

The study conducted by Maw and Bawden [38] was

identified as the primary source to provide estimates for

hearing gain associated with each strategy, and in the

baseline analysis, the assumed hearing gains were (1)

13.06 dBHL after 12 months and 12.24 dBHL after

24 months for the grommets strategy, and (2) 4.88 dBHL

after 12 months and 7.57 dBHL after 24 months for the

non-surgical strategies (hearing-aids and do-nothing

strategies). QALYs were used to value health gain and

were assumed to be a linear function of potential

improvement in hearing. To calculate the QALYs, a

utility gain per unit increase in hearing gain was attached

to the identified dBHLs for each strategy. Published

expert opinion was used to apply a value for the assumed

utility gain associated with the per unit increase in dBHL.

This estimate of a utility gain per unit increase in dBHL

of 0.00874 (95 % CI 0.005–0.012) [7] was based on

interpretation of an unpublished study by Kubba [39] that

collected individual-patient data on the Health Utilities

Index Mark III for children with a median age of 5 years.

The use of grommets can improve a child’s quality of

hearing by approximately 50.5 % (95 % CI 47.0–54.5 %)

when compared with no intervention (i.e. do-nothing

strategy) [40]. Therefore, the utility gain per unit increase

in dBHL of 0.00874 has been reduced by 50.5 % for the

children in the do-nothing strategy.

S. Mohiuddin et al.

123

Table 3 summarises the point estimates of resource use

and unit costs used for each strategy and pathway for the

model together with the assumed ranges and distributions

used in the probabilistic sensitivity analysis (PSA). All

prices are presented in British pounds for the year

2010–2011.

One-way sensitivity analysis

Two one-way sensitivity analyses were performed to assess

how sensitive the baseline results were to predefined

parameters used in the model. The selected value for each

parameter of interest was varied within a range of lower

and upper quartile values, while keeping all the other

parameters constant. The following two key parameters

were selected for inclusion, on the basis of input from

clinical experts, because a priori they were expected to

potentially have an impact on the relative costs and benefits

of the grommets strategy: (1) the unit cost of insertion of

grommets (£891) was varied between £646 (lower quartile)

and £1,043 (upper quartile) (NHS Reference Costs

2010–11; Healthcare Resource Groups code CZ08T; day

case); and (2) the potential hearing gain after 12 months

(13.06 dBHL) was varied between 6.70 dBHL (lower

quartile) and 19.50 dBHL (upper quartile), and after

24 months (12.24 dBHL) it was varied between

6.10 dBHL (lower quartile) and 18.40 dBHL (upper quar-

tile). The lower and upper quartiles for the potential change

in hearing gain were calculated, assuming a normal dis-

tribution, from the reported mean values and standard

deviations in Maw and Bawden [38].

Probabilistic sensitivity analysis

PSA was used to quantify the joint uncertainty in the model

parameters by assigning a range and specific distribution to

each of the input parameters. Gamma distributions were

used to represent the uncertainty in the cost parameters (see

Table 3) because these values are constrained to be zero or

positive. The gamma distribution is parameterised by two

parameters (i.e. shape and scale), which are expressed as

functions of the expectation and variance of the distribu-

tion. Beta distributions were used to represent the uncer-

tainty in the probability parameters (see Tables 1, 2)

because these values are defined on the interval with a

minimum (0) and maximum (1) value. The beta distribu-

tion is parameterised by two parameters (i.e. alpha and

beta); alpha corresponds to the number of events observed

and beta corresponds to the number of non-events

observed. Normal distributions were used to represent the

uncertainty in the hearing gain parameters to reflect the

likelihood of a unit increase or decrease in dBHL during

the recovery period: Normal*(13.06, 9.49) after

12 months and Normal*(12.24, 9.1) after 24 months for

the grommets strategy and Normal*(4.88, 11.11) after

12 months and Normal*(7.57, 12.76) after 24 months for

the non-surgical strategies [38]. The utility gain per unit

increase in dBHL was parameterised by a gamma distri-

bution [Gamma*(24.38, 0.0004)] with the shape (24.38)

and scale (0.0004) determined from the mean and variance

reported earlier [7, 39].

The probability of spontaneous resolution was sampled

from the distribution labelled as Beta*(61, 137) [12], and

the probability of breakage or loss of hearing aids was

reestimated within the PSA on the basis of uncertainty

surrounding the original 21-month data that was repre-

sented by Beta*(6, 18) [10]. Adherence to the hearing-

aids strategy was sampled from Beta*(2, 20) [10], and

expected episodes of AOM were sampled from Nor-

mal*(2.8, 0.3) [12]. On the basis of the statement given

earlier that grommets can improve a child’s quality of

hearing by approximately 50.5 % (95 % CI 47.0–54.5 %)

[40] when compared with the do-nothing strategy, the

QALY gain associated with the do-nothing strategy was

Table 2 Probability data for the grommets strategy

Parameter Probability Distribution for PSA Source Notes

Removal of grommets

(1st procedure)

0.072 Beta*(6, 77) Phua et al. [23] This was the only identified study of children

with CP from the systematic review that

reported different risk of retraction following

subsequent grommet-insertion proceduresRemoval of grommets

(C2 procedures)

0.171 Beta*(7, 34) Phua et al. [23]

Reinsertion of 2nd

grommets

0.38 Beta*(68, 110) Sheahan et al. [2] This was a questionnaire-based study of children

with CP. This is the only identified study that

reported a reinsertion rate for two or more

grommets. A previous model used a reinsertion

rate of 0.25 for general children [7]

Reinsertion of 3rd

grommets

0.38 Beta*(68, 110) Sheahan et al. [2]

Time to extrusion Grommets fall

out by 39 weeks

Grommets fall out

by Normal*(39,

2.93)a

NCCWCH [7] The time to extrusion is defined as the time by

when the grommets should fall out naturally

a This is estimated on the basis of information that grommets fall out between 26 and 52 weeks [7]

Cost-effectiveness analysis of grommets-led care pathway

123

Ta

ble

3R

eso

urc

eu

sean

du

nit

cost

dat

a

Par

amet

erR

eso

urc

eu

seC

ost

So

urc

eN

ote

s

Mea

nd

escr

ipti

on

(per

chil

d)

Dis

trib

uti

on

for

PS

A

Mea

nu

nit

cost

(£)

Dis

trib

uti

on

for

PS

A

Inse

rtio

no

f

gro

mm

ets

1p

roce

du

reF

ixed

89

1G

amm

a*(1

30

.45

,6

.83

)

NS

RC

1a

20

10

–2

01

1(H

RG

cod

e

CZ

08

T;

day

case

)

Min

or

ear

pro

ced

ure

sfo

rch

ild

ren

B1

8y

ears

old

thro

ug

hth

e

tym

pan

icm

emb

ran

e.1

pro

ced

ure

iseq

uiv

alen

tto

inse

rtin

g2

gro

mm

ets

for

each

chil

do

win

gto

bil

ater

alO

ME

Ty

mp

ano

pla

sty

1p

roce

du

reF

ixed

1,8

31

Gam

ma*

(10

0.1

1,

18

.29

)

NS

RC

1a

20

10

–2

01

1(H

RG

cod

e

CZ

10

U;

day

case

)

Maj

or

ear

pro

ced

ure

sfo

rch

ild

ren

B1

8y

ears

old

ow

ing

to

per

fora

tio

no

fth

eea

rdru

m

Rem

ov

alo

f

gro

mm

ets

1p

roce

du

reF

ixed

89

1G

amm

a*(1

30

.45

,6

.83

)

NS

RC

1a

20

10

–2

01

1(H

RG

cod

e

CZ

08

T;

day

case

)

Min

or

ear

pro

ced

ure

sfo

rch

ild

ren

B1

8y

ears

old

thro

ug

hth

e

tym

pan

icm

emb

ran

e.1

pro

ced

ure

iseq

uiv

alen

tto

rem

ov

ing

1o

r2

gro

mm

ets

for

each

chil

do

win

gto

bil

ater

al

OM

E

Hea

rin

gai

d2

un

its

Fix

ed£

80

Fix

edG

DG

esti

mat

e[7

]h

asb

een

infl

ated

on

the

bas

iso

fth

eH

CH

Sin

dex

20

10

–2

01

1(P

SS

RU

)

2h

eari

ng

aid

sar

ere

qu

ired

for

each

chil

do

win

gto

bil

ater

alO

ME

Ear

mo

uld

8u

nit

sp

ery

ear

Fix

ed1

7F

ixed

GD

Ges

tim

ate

[7]

has

bee

nin

flat

ed

on

the

bas

iso

fth

eH

CH

Sin

dex

20

10

–2

01

1(P

SS

RU

)

2ea

rm

ou

lds

are

req

uir

edfo

rea

ch

chil

dev

ery

13

wee

ks

[7]

Hea

rin

g-a

id

care

kit

1u

nit

Fix

ed2

0.8

8F

ixed

htt

p:/

/ww

w.c

on

nev

ans.

co.u

k

(acc

esse

d3

Dec

20

12

)

Th

isco

stis

incu

rred

on

ceo

win

gto

the

mai

nte

nan

ceo

fth

eh

eari

ng

aid

(s)

Hea

rin

g-a

id

bat

tery

26

un

its

per

yea

rF

ixed

£0

.49

Fix

edh

ttp

://w

ww

.hea

rin

g-a

id-b

atte

ries

.

org

.uk

(acc

esse

d3

Dec

20

12

)

Bat

teri

esar

ere

qu

ired

tob

e

rep

lace

dev

ery

4w

eek

s[7

]

Hea

rin

g-a

id

fitt

ing

1p

roce

du

reF

ixed

76

Gam

ma*

(71

.03

,1

.07

)

NH

SR

efer

ence

Co

sts

(20

05

-06

)

(ser

vic

eco

de

AS

1F

A)

has

bee

n

infl

ated

on

the

bas

iso

fth

eH

CH

S

ind

ex2

01

0–

20

11

(PS

SR

U)

Fit

tin

go

fh

eari

ng

aid

sin

an

aud

iolo

gy

dep

artm

ent.

1

pro

ced

ure

iseq

uiv

alen

tto

fitt

ing

2h

eari

ng

aid

sfo

rea

chch

ild

ow

ing

tob

ilat

eral

OM

E

GP

vis

it/

anti

bio

tic

med

icat

ion

Gro

mm

ets

stra

teg

y:

0.2

5(o

torr

ho

ea)

?

0.0

42

(gra

nu

lati

on

)ti

mes

See

Tab

le1

GP

vis

it:

41

An

tib

ioti

c

med

icat

ion

:1

1

Fix

edG

Pv

isit

:P

SS

RU

(un

itco

sts

of

hea

lth

and

soci

alca

re

20

10

–2

01

1)

An

tib

ioti

cm

edic

atio

n:

GD

G

esti

mat

e[7

]h

asb

een

infl

ated

on

the

bas

iso

fth

eH

CH

Sin

dex

20

10

–2

01

1(P

SS

RU

)

GP

vis

itan

dco

stfo

ra

cou

rse

of

anti

bio

tics

or

ear

dro

ps

for

man

agem

ent

of

oto

rrh

oea

,

gra

nu

lati

on

,an

d/o

rA

OM

Hea

rin

g-a

ids

stra

teg

y:

2.8

tim

es

No

rmal*

(2.8

,0

.3)

Do

-no

thin

gst

rate

gy

:

2.8

tim

es

No

rmal*

(2.8

,0

.3)

S. Mohiuddin et al.

123

Ta

ble

3co

nti

nu

ed

Par

amet

erR

eso

urc

eu

seC

ost

So

urc

eN

ote

s

Mea

nd

escr

ipti

on

(per

chil

d)

Dis

trib

uti

on

for

PS

A

Mea

nu

nit

cost

(£)

Dis

trib

uti

on

for

PS

A

Au

dio

log

ical

rev

iew

Gro

mm

ets

stra

teg

y:

1.5

vis

its

foll

ow

ing

each

op

erat

ion

Un

ifo

rm*

(1,

2)

48

Gam

ma*

(64

,0

.75

)

NH

SR

efer

ence

Co

sts

(20

05

-06

)

(ser

vic

eco

de

AS

1F

U)

has

bee

n

infl

ated

on

the

bas

iso

fth

eH

CH

S

ind

ex2

01

0–

20

11

(PS

SR

U)

1o

r2

vis

its

foll

ow

ing

each

op

erat

ion

for

the

gro

mm

ets

stra

teg

y[7

].1

vis

itw

ith

in

13

wee

ks

and

sub

seq

uen

tv

isit

s

ever

y2

6w

eek

sth

erea

fter

for

the

hea

rin

g-a

ids

stra

teg

y[7

].1

or

2

vis

its

ever

yy

ear

for

the

do

-

no

thin

gst

rate

gy

[7,

9,

10,

37]

Hea

rin

g-a

ids

stra

teg

y:

1v

isit

init

iall

y?

1.5

vis

its

per

yea

r

ther

eaft

er

Fix

ed

Do

-no

thin

gst

rate

gy

:

1.5

vis

its

per

yea

r

Un

ifo

rm*

(1,

2)

EN

Tsp

ecia

list

vis

it

Gro

mm

ets

stra

teg

y:

2.5

vis

its

foll

ow

ing

each

op

erat

ion

Fix

ed9

1.7

2G

amm

a*(8

4.1

4,

1.0

9)

NS

RC

1a

20

10

–2

01

1

(ser

vic

eco

de

12

0)

1v

isit

wit

hin

6w

eek

so

fea

ch

op

erat

ion

and

sub

seq

uen

tv

isit

s

ever

y2

6w

eek

sth

erea

fter

un

til

the

tim

eo

fex

tru

sio

n[7

].1

or

2

vis

its

ever

yy

ear

[10,3

7]

for

bo

th

the

hea

rin

g-a

ids

stra

teg

yan

dth

e

do

-no

thin

gst

rate

gy

Hea

rin

g-a

ids

stra

teg

y:

1.5

vis

its

per

yea

r

Un

ifo

rm*

(1,

2)

Do

-no

thin

gst

rate

gy

:

1.5

vis

its

per

yea

r

Un

ifo

rm*

(1,

2)

AO

Mac

ute

oti

tis

med

ia,

EN

Tea

r,n

ose

and

thro

at,

GD

GG

uid

elin

eD

evel

op

men

tG

rou

p,

GP

gen

eral

pra

ctit

ion

er,

HC

HS

Ho

spit

alan

dC

om

mu

nit

yH

ealt

hS

erv

ices

,H

RG

Hea

lth

care

Res

ou

rce

Gro

up

s,N

HS

Nat

ion

alH

ealt

hS

erv

ice,

OM

Eo

titi

sm

edia

wit

hef

fusi

on

,P

SS

RU

Per

son

alS

oci

alS

erv

ices

Res

earc

hU

nit

aN

HS

Ref

eren

ceC

ost

s2

01

0-1

1A

pp

end

ixN

SR

C1

Cost-effectiveness analysis of grommets-led care pathway

123

adjusted according to a normal distribution [Nor-

mal*(0.505, 0.02)]. The PSA involved 10,000 iterations.

Expected value of perfect information

Using the decision-tree structure and subsequent PSA, an

expected value of perfect information analysis was con-

ducted to estimate the potential value of future research.

Equation 1 shows that EVPI estimates the difference

between the expected value of a decision made with perfect

information and the expected value of a decision made on

the basis of the current evidence base:

EVPI ¼ Eh maxj

NBðj; hÞ �maxj

EhNBðj; hÞ; ð1Þ

where Eh maxj NBðj; hÞ represents the expected value of

the decision with perfect information and maxj EhNBðj; hÞrepresents the expected value without perfect information.

In its simplest form, EVPI represents the maximum amount

that a decision-maker would be willing to pay to gain

access to perfect information. However, the societal value

of research should ideally be estimated across the popula-

tion of future patients for whom the decision is pertinent

since the information provided by research is a public

good. Equation 2 shows the calculation of the population

level of EVPI (pEVPI):

pEVPI ¼ EVPI �XT

t¼1

It

ð1þ rÞt; ð2Þ

where T is the effective lifetime of a technology, It is the

incidence of the condition relevant to the health technology

over the effective lifetime of the technology and r is dis-

count rate. Thus, pEVPI represents an upper bound of the

expected benefit of conducting further research. If pEVPI is

greater than the expected cost of conducting further

research, then it should potentially be considered worth-

while to conduct the further research. Here the estimate of

the population was based on an assumption that every year

720 children1 (I) might be eligible for grommets in the UK,

that the decision would be relevant for 10 years2 (T), with a

discount rate of 3.5 %3 (r).

Results

The results of the baseline (deterministic) and probabilistic

analyses are shown in Table 4. All results were generated

for a hypothetical cohort of 10,000 children, with results

presented in terms of costs and outcomes per child. The

results based on the expected values from the PSA showed

that insertion of grommets was the most costly strategy,

with a cost of £2,086 per child, compared with £1,237 per

child for the hearing-aids strategy and £593 per child for

the do-nothing strategy. The associated QALYs were 0.218

for the grommets strategy, 0.102 for the hearing-aids

strategy and 0.053 for the do-nothing strategy. The grom-

mets strategy was associated with the highest cost, but it

was the most effective option in terms of additional QA-

LYs gained.

The incremental cost-effectiveness ratios (ICERs) per

QALY based on the expected values from the PSA were

(1) £13,143 per QALY gained for the hearing-aids strategy

compared with the do-nothing strategy, and (2) £7,338 per

QALY gained for the grommets strategy compared with

the hearing-aids strategy. Applying the principle of

extended dominance, the hearing-aids strategy was found

to be extended dominated by the grommets strategy

because the hearing-aids strategy compared with the do-

nothing strategy has an ICER (£13,143 per QALY gained)

greater than that of the grommets strategy compared with

the hearing-aids strategy (£7,338 per QALY gained). As

such, the ICER for the grommets strategy compared with

the do-nothing strategy was £9,065 per QALY gained.

Although the hearing-aids strategy was extended domi-

nated by the grommets strategy on the basis of the mean

point estimates, it was not excluded from the subsequent

PSA and EVPI analysis because there could be some

possible realisations of the uncertainty where the hearing-

aids strategy would not be extended dominated by the

grommets strategy (Fig. 2).

The results of the one-way sensitivity analyses demon-

strated the relationship between two specific parameters

and the expected costs and QALYs. When the baseline unit

cost of insertion of grommets was varied between £646

(lower quartile) and £1,043 (upper quartile), the resulting

ICER for the grommets strategy compared with the do-

nothing strategy ranged between £6,791 and £10,453.

When the baseline hearing gain of 13.06 dBHL (after

12 months) was varied between 6.70 dBHL (lower quar-

tile) and 19.50 dBHL (upper quartile) and that of

12.24 dBHL (after 24 months) was varied between

6.10 dBHL (lower quartile) and 18.40 dBHL (upper quar-

tile), the resulting ICER for the grommets strategy com-

pared with the do-nothing strategy ranged between £26,017

and £5,461. The one-way sensitivity analysis indicated that

the relative cost-effectiveness of the grommets strategy,

1 Data from the CRANE Database showed there were 800 children

born with CP in England, Wales and Northern Ireland in 2012 [http://

www.craniofacialsociety.org.uk/downloads/CRANE%20Annual%20

Report%202013%20(Final).pdf]. Of which, 720 (90 % [2, 4–6] of

800) were assumed to have OME.2 Armstrong [15] first described the use of grommets in 1954, and

since then, use of grommets to restore hearing to normal has been

increased. Given the historical longevity of the technology revealed in

the literature, it seemed reasonable to assume that grommets will be

used for at least another 10 years before a new technology comes

along and replaces grommet technology.3 As recommended by the NICE reference case [25].

S. Mohiuddin et al.

123

when compared with the do-nothing strategy, is affected by

the assumed unit cost of surgical insertion of grommets. As

expected, as the unit cost of surgical insertion increases, the

grommets strategy will become relatively less cost-effec-

tive compared with the alternative strategies.

Figure 3 shows a scatter plot of the expected incre-

mental costs and effects (gain in QALYs) for each of the

10,000 simulations. The do-nothing strategy was assumed

to be the status quo, and hence, was anchored at the

origin. Comparing the values for positive expected

incremental costs and gain in QALYs, some 77 % of the

simulated values for the hearing-aids strategy versus the

do-nothing strategy and 89 % for the grommets strategy

versus the do-nothing strategy fell in the northeast

quadrant of the cost-effectiveness plane. The northeast

quadrant represents values for which an intervention

would be more costly and more effective versus its

comparators, and it then becomes necessary to make a

decision about the threshold value for the willingness to

pay for an additional QALY. Approximately 82 % of the

simulated values of the northeast quadrant indicated that

the grommets strategy would be cost-effective compared

with the do-nothing strategy below the willingness to pay

threshold of £20,000 per QALY. Furthermore, the PSA

revealed that the hearing-aids strategy was extended

dominated by the grommets strategy for some 62 % of

the simulated realisations.

Figure 4 presents cost-effectiveness acceptability curves

for each of the strategies based on the results of the PSA.

The cost-effectiveness acceptability curves illustrate that

the grommets strategy is most likely to be cost-effective for

values of the ceiling ratio threshold beyond £7,800 per

QALY gained. The probability that the grommets strategy

is cost-effective is 0.49 at a ceiling ratio of £10,000 per

QALY and 0.64 at a ceiling ratio of £20,000 per QALY.

The EVPI at both the individual and the population level

for a range of different values of the threshold are pre-

sented in Table 5.

Figure 5 shows the relationship between pEVPI over a

decision horizon of 10 years and different values of the

ceiling ratio per QALY. This figure also shows that the

value of further research exceeds £4 million for all values

of the threshold beyond £10,000 per QALY gained. This is

likely to exceed the total cost of future research. At a

threshold ceiling ratio of £20,000 per QALY, the value of

further research exceeds £5 million. However, should the

threshold ceiling ratio be less than £1,500, pEVPI is zero,

indicating that there is no value in additional information

from future research.

Table 4 Expected cost and quality-adjusted life-year (QALY) gain for each strategy from the deterministic and probabilistic analyses

Strategy Deterministic Probabilistic ICER per QALY gained

Cost (£) QALY Cost (£) QALY Deterministic Probabilistic

Do-nothing strategy 592 0.0528 593 0.0529 – –

Hearing-aids strategy 1,235 0.1017 1,237 0.1019 Extended dominated

(by the grommets strategy)

Extended dominated

(by the grommets strategy)

Grommets strategy 2,083 0.2175 2,086 0.2176 £9,053 (grommets strategy

vs do-nothing strategy)

£9,065 (grommets strategy vs

do-nothing strategy)

ICER incremental cost-effectiveness ratio

£0

£500

£1,000

£1,500

£2,000

£2,500

0.00 0.05 0.10 0.15 0.20 0.25

Co

st

QALY gain

do-nothing hearing-aids grommets

Fig. 2 Expected cost and quality-adjusted life-year (QALY) gain for

each strategy0

500

1000

1500

2000

2500

-0.5 -0.2 0.1 0.4 0.7 1

Incr

emen

tal c

ost

Incremental QALY gain

hearing-aids grommets ICER threshold at £20,000

Fig. 3 Scatter plot of incremental expected costs and QALYs

obtained from probabilistic sensitivity analysis (PSA). The circles

represent the expected value from the PSA. The do-nothing strategy

was anchored at the origin. ICER incremental cost-effectiveness ratio

Cost-effectiveness analysis of grommets-led care pathway

123

Discussion

The association between children with CP and hearing loss

that results from OME is well documented [41, 42]. The

surgical insertion of grommets is one of the commonest

surgical procedures in childhood today [16, 35]. Although

the disagreement regarding the relative benefits and risks

for the insertion of grommets in children with CP is

unresolved within the surgical community, affected chil-

dren still desire to function better and parents want their

children to be in a position to participate fully in education.

Despite a large body of evidence on incidence and preva-

lence rates of OME, there is still a paucity of research on

the potential impact of the surgical insertion of grommets

in children with CP. Therefore, it was essential to assess

whether the surgical insertion of grommets can have a

positive impact on expected health benefits. A systematic

search of the published literature was conducted to identify

decision-analytic-model-based economic evaluations of

surgical insertion of grommets in the management of per-

sistent bilateral OME in children with CP. No economic

evaluations were identified that were relevant to children

who are born with CP. Therefore, a model-based cost-

effectiveness analysis was developed to assess the impact

of surgical insertion of grommets compared with non-sur-

gical alternatives in the management of bilateral OME

persisting after watchful waiting for 3 months in children

with CP.

In this model-based cost-effectiveness analysis, the

grommets strategy was found to be the optimal strategy,

with the expected value of the ICER from the PSA of

£9,065 per QALY gained compared with the do-nothing

strategy, and the hearing-aids strategy was extended

dominated by the grommets strategy. The ICER for

grommets is well below the willingness to pay threshold of

£20,000 per QALY, which is commonly taken by NICE as

a reasonable level of the willingness to pay for an addi-

tional QALY [43]. The gain in QALY resulting from

improvement in dBHL associated with the grommets

strategy to manage OME came at a higher cost, which is

mainly driven by the resource use attributable to the sur-

gical process. The results of the PSA indicated the exis-

tence of considerable uncertainty surrounding the existence

and extent of the incremental QALY gain associated with

the grommets strategy compared with the do-nothing

strategy. In addition, there was considerable uncertainty

surrounding the extent of the incremental cost associated

with the grommets strategy compared with the do-nothing

strategy. This observed uncertainty perhaps explains why

there remains disparity in the medical community regard-

ing the use of grommets in individual children with CP for

management of OME. The results from this early economic

analysis should not be used to inform any current changes

in clinical practice; it was conducted to understand whether

there is need for further research regarding the use of

grommets in children with CP and persistent OME.

A key strength of this study was that it extended the

analysis beyond PSA and calculated the EVPI to provide a

measurable insight into whether further research in this

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

£0 £20,000 £40,000 £60,000 £80,000 £1,00,000

Pro

bab

ility

co

st e

ffec

tive

Value of ceiling ratio

grommetshearing-aidsdo-nothing

Fig. 4 Cost-effectiveness acceptability curve for the three strategies

Table 5 Expected values of perfect information (EVPI) at the indi-

vidual level and the population level at different ceiling ratios

Ceiling

ratio (£)

Individual

EVPI (£)

Population EVPI

over a 10-year

decision horizon (£)

Population EVPI

over a 5-year

decision horizon (£)

5,000 103 639,462 347,161

10,000 642 3,977,639 2,159,445

15,000 720 4,465,138 2,424,106

20,000 838 5,194,030 2,819,818

25,000 967 5,993,883 3,254,056

30,000 1,102 6,827,204 3,706,462

£0

£20,00,000

£40,00,000

£60,00,000

£80,00,000

£1,00,00,000

£0 £10,000 £20,000 £30,000 £40,000 £50,000

Po

pu

lati

on

EV

PI

Value of ceiling ratio

Fig. 5 Expected value of perfect information (EVPI) at the popula-

tion level at various ceiling ratios

S. Mohiuddin et al.

123

area is potentially worthwhile. The feasibility and impli-

cations of using the EVPI method for informing the future

research prioritisation process has previously been well

described, and its use is recommended in the context of

commissioned health technology assessments [44]. The

calculated EVPI of £838 for every child and £5,194,030 for

a population of children in England, Wales and Northern

Ireland, assuming a willingness to pay threshold of £20,000

per QALY and a decision horizon of 10 years, suggest that

further research in this area is potentially worthwhile.

This model was an early economic evaluation that used

all the available data from multiple sources. However,

there were some limitations in terms of the availability and

relevance of the data used for parameter inputs, which

should be revised and reassessed once more relevant clin-

ical effectiveness, resource use and utility data become

available. This analysis used a utility value taken from a

published source, and the same value was used previously

in an appraisal completed by NCCWCH [7]. The calcula-

tion of QALYs was driven by a combination of utility

values and effectiveness data, specified in terms of dBHL.

The one-way sensitivity analysis illustrated that changes in

the dBHL data had a major impact on the estimates of cost-

effectiveness associated with the grommets strategy. In

addition, the second one-way sensitivity analysis indicated

that changes in the unit cost of grommet insertion impacted

on the cost-effectiveness of the grommets strategy. The

results of these one-way sensitivity analyses suggest that

these might be important areas for future research, and

although outside the scope of this article, this could be

confirmed by a parameter EVPI analysis. A further con-

sideration for future research is to explore whether gains in

QALYs, which focus on measuring improvements in health

status, are the only relevant outcome to assess an inter-

vention aimed at young children. Other non-health gains,

such as improvements in educational attainment and ability

of children to play with their peers, may also be important

outcomes for children and their parents.

The eligible patient population was also an area of

uncertainty suggested to be important by clinical experts.

This economic evaluation focussed on a CP population of

children younger than 12 years. This focus was necessary

because of a paucity of epidemiological data for other age

groups. This age group was selected for this analysis

because clinical experts considered this group may repre-

sent children in which the condition is most prevalent.

However, further work is needed, informed by robust

epidemiological data, to understand the relative cost-

effectiveness of the insertion of grommets in different age

groups, and also the most appropriate age for the surgical

procedure in a child with CP. This analysis used the only

available source for estimates of health care resource use.

These data were not directly relevant to a population of CP

children, but provided the best evidence in the absence of

directly relevant data. Using these data was likely to be a

conservative approach since OME is relatively commoner

in CP children than in non-CP children and has an extended

recovery period [2]. This analysis posed two technical

challenges. Interpretation of this analysis should be

undertaken with caution because there are no definitive

guidelines to follow for the treatment of OME in children

with CP. Therefore, the clinical pathways used to structure

the economic model were developed using assumptions

based on an existing economic model, used in a previous

appraisal conducted by NCCWCH [7], and adapted for a

population of children with CP using advice from clinical

experts. Furthermore, the limited number of studies meant

that it was difficult to generate ranges based on empirical

data around some parameters included in the PSA.

This is the first model-based economic evaluation to

identify and quantify the costs and benefits of different

management options for persistent bilateral OME in chil-

dren with CP. The model has demonstrated the potential

for resources to be released from other health care inter-

ventions when insertion of grommets is applied for man-

aging OME. The total cost of the grommets strategy is

relatively high, but this intervention appears to provide

good value for money, on the basis of the current evidence

base, if it is used after the initial 3-month period of

watchful waiting as a means to correct significant hearing

impairment and prevent complications of untreated OME.

The early management of OME-related complications

should generate expected net benefits that might compen-

sate the additional expenditure incurred because of repe-

ated clinic visits [13] and prompt rapid hearing gain, which

is also important for childhood speech development and

associated educational attainment [45, 46]. Schonweiler

et al. [47] showed that language development depends

more on hearing ability than on the severity or surgical

repair of CP. Paradise and Bluestone [48] some 40 years

ago advocated a policy of early insertion of grommets and

replacement when necessary in order to decrease the long-

term otological complications and minimise the effects on

speech and language development.

Furthermore, the use of grommets could release clini-

cians from the pressure to prescribe antibiotics to manage

multiple instances of infections, which could impact on

antibiotic resistance [49, 50]. However, this analysis has

shown limitations in the current evidence base and identi-

fied that it is potentially worthwhile undertaking further

research in this area. In particular, it is likely to be useful to

understand the relevance of using hearing gain alone as an

outcome, the link between hearing gain and utility gain, the

actual use of health care resources and clinical effective-

ness data to inform the appropriate age for the insertion of

grommets in children with OME.

Cost-effectiveness analysis of grommets-led care pathway

123

Acknowledgments This study was supported by the National

Institute for Health Research Health Technology Assessment Pro-

gramme at the National Health Service and the Healing Foundation,

UK. The Management of Otitis Media with Effusion in Children with

Cleft Palate (MOMENT) Study Management Group (Iain Bruce,

Peter Callery, Nicola Harman, Syed Mohiuddin, Kevin O’Brien,

Katherine Payne, Bill Shaw, Tri Tat, Stephanie Tierney and Paula

Williamson). The Healing Foundation Cleft and Craniofacial Clinical

Research Centre, University of Manchester, and Paul Jacklin (senior

health economist, National Collaborating Centre for Women’s and

Children’s Health, UK) are thanked for providing advice and

guidance.

Conflict of interest None.

References

1. Møller, P.: Hearing, middle ear pressure and otopathology in a

cleft palate population. Acta Otolaryngol. 92, 521–528 (1981)

2. Sheahan, P., Miller, I., Sheahan, J.N., Earley, M.J., Blayney,

A.W.: Incidence and outcome of middle ear disease in cleft lip

and/or cleft palate. Int. J. Pediatr. Otorhinolaryngol. 67, 785–793

(2003)

3. Ponduri, S., Bradley, R., Ellis, P.E., Brookes, S.T., Sandy, J.R.,

Ness, A.R.: The management of otitis media with early routine

insertion of grommets in children with cleft palate—a systematic

review. Cleft Palate Craniofac. J. 46, 30–38 (2009)

4. Stool, S.E., Randall, P.: Unexpected ear disease in infants with

cleft palate. Cleft Palate J. 4, 99–103 (1967)

5. Doyle, W.J., Cantekin, E.I., Bluestone, C.D.: Eustachian tube

function in cleft palate children. Ann. Otol. Rhinol. Laryngol.

Suppl. 89, 34–40 (1980)

6. Grant, H.R., Quiney, R.E., Mercer, D.M., Lodge, S.: Cleft palate

and glue ear. Arch. Dis. Child. 63, 176–179 (1988)

7. National Collaborating Centre for Women’s and Children’s

Health: Surgical management of otitis media with effusion in

children. NICE clinical guideline 60. National Institute for Health

and Clinical Excellence, London (2008)

8. Maw, R., Wilks, J., Harvey, I., Peters, T.J., Golding, J.: Early

surgery compared with watchful waiting for glue ear and effect

on language development in preschool children: a randomised

trial. Lancet 353, 960–963 (1999)

9. Maheshwar, A.A., Milling, M.A.P., Kumar, M., Clayton, M.I.,

Thomas, A.: Use of hearing aids in the management of children

with cleft palate. Int. J. Pediatr. Otorhinolaryngol. 66, 55–62

(2002)

10. Gani, B., Kinshuck, A.J., Sharma, R.: A review of hearing loss in

cleft palate patients. Int. J. Otolaryngol. (2012)

11. Rovers, M.M., Schilder, A.G.M., Zielhuis, G.A., Rosenfeld,

R.M.: Otitis media. Lancet 363, 465–473 (2004)

12. Rosenfeld, R.M., Kay, D.: Natural history of untreated otitis

media. Laryngoscope 113, 1645–1657 (2003)

13. Rovers, M.M., Glasziou, P., Appelman, C.L., Burke, P.,

McCormick, D.P., Damoiseaux, R.A., Gaboury, I., Little, P.,

Hoes, A.W.: Antibiotics for acute otitis media: a meta-analysis

with individual patient data. Lancet 368, 1429–1435 (2006)

14. Tierney, S., O’Brien, K., Harman, N.L., Sharma, R.K., Madden,

C., Callery, P.: Otitis media with effusion: experiences of chil-

dren with cleft palate and their parents. Cleft Palate Craniofac. J.

(2013)

15. Armstrong, B.W.: A new treatment for chronic secretory otitis

media. AMA Arch. Otolaryngol. 59, 653–654 (1954)

16. Lous, J., Burton, M.J., Felding, J.U., Ovesen, T., Rovers, M.M.,

Williamson, I.: Grommets (ventilation tubes) for hearing loss

associated with otitis media with effusion in children. Cochrane

Database Syst. Rev. CD001801 (2005)

17. Schilder, A.G.M., Rovers, M.M.: International perspective on

management. In: Rosenfeld, R.M., Bluestone, C.D. (eds.) Evi-

dence-Based Otitis Media, pp. 325–332. Hamilton, BC Decker

(2003)

18. Berman, S., Roark, R., Luckey, D.: Theoretical cost effectiveness

of management options for children with persisting middle ear

effusions. Pediatrics 93, 353–363 (1994)

19. Gates, G.A.: Cost-effectiveness considerations in otitis media

treatment. Otolaryngol. Head Neck Surg. 114, 525–530 (1996)

20. Hartman, M., Rovers, M.M., Ingels, K., Zielhuis, G.A., Severens,

J.L., van der Wilt, G.J.: Economic evaluation of ventilation tubes

in otitis media with effusion. Arch. Otolaryngol. Head Neck Surg.

127, 1471–1476 (2001)

21. Sculpher, M., Drummond, M., Buxton, M.: The iterative use of

economic evaluation as part of the process of health technology

assessment. J. Health Serv. Res. Policy 2, 26–30 (1997)

22. Gould, H.J.: Hearing loss and cleft palate: the perspective of time.

Cleft Palate J. 27, 36–39 (1990)

23. Phua, Y.S., Salkeld, L.J., de Chalain, T.M.B.: Middle ear disease

in children with cleft palate: protocols for management. Int.

J. Pediatr. Otorhinolaryngol. 73, 307–313 (2009)

24. Liu, L., Sun, Y., Ma, L., Zhao, W., Wu, R.: Effect of ventilation

tube insertion on otitis media with effusion in cleft palate chil-

dren. Zhonghua Er Bi Yan Hou Ke Za Zhi. 39, 216–218 (2004)

25. National Institute for Clinical Excellence: Guide to the methods

of technology appraisal. http://www.nice.org.uk/niceMedia/pdf/TAP_Methods.pdf (2004). Accessed 10 Apr 2013

26. Barton, P., Bryan, S., Robinson, S.: Modelling in the economic

evaluation of health care: selecting the appropriate approach.

J. Health Serv. Res. Policy 9, 110–118 (2004)

27. Kwan, W.M.Y., Abdullah, V.J., Liu, K., van Hasselt, C.A., Tong,

M.C.F.: Otitis media with effusion and hearing loss in Chinese

children with cleft lip and palate. Cleft Palate Craniofac. J. 48,

684–689 (2011)

28. Sheahan, P., Blayney, A.W., Sheahan, J.N., Earley, M.J.:

Sequelae of otitis media with effusion among children with cleft

lip and/or cleft palate. Clin. Otolaryngol. Allied Sci. 27, 494–500

(2002)

29. Kay, D.J., Nelson, M., Rosenfeld, R.M.: Meta-analysis of tym-

panostomy tube sequelae. Otolaryngol. Head Neck Surg. 124,

374–380 (2001)

30. Fior, R., Veljak, C.: Late results and complications of tympan-

ostomy tube insertion for prophylaxis of recurrent purulent otitis

media in pediatric age. Int. J. Pediatr. Otorhinolaryngol. 8,

139–146 (1984)

31. Vlastarakos, P.V., Nikolopoulos, T.P., Korres, S., Tavoulari, E.,

Tzagaroulakis, A., Ferekidis, E.: Grommets in otitis media with

effusion: the most frequent operation in children. But is it asso-

ciated with significant complications? Eur. J. Pediatr. 166,

385–391 (2007)

32. Gibb, A.G.: President’s address. Tympanosclerosis. Proc. R. Soc.

Med. 69, 155–162 (1976)

33. Williamson, I.: Otitis media with effusion in children. Clin. Evid.

(2011)

34. NHS Choices: Complications of glue ear. http://www.nhs.uk/

Conditions/Glueear/Pages/Complications.aspx. Accessed 10 Jul

2013

35. Hoffmann, K.K., Thompson, G.K., Burke, B.L., Derkay, C.S.:

Anesthetic complications of tympanostomy tube placement in

children. Arch. Otolaryngol. Head Neck Surg. 128, 1040–1043

(2002)

S. Mohiuddin et al.

123

36. Children’s Hospital: Grommets surgery for glue ear—informa-

tion for patients. Oxford Radcliffe Hospitals NHS Trust. http://

www.ouh.nhs.uk/patient-guide/leaflets/files%5C101018grom

mets.pdf (2012). Accessed 10 Apr 2013

37. Gunasekera, H., O’Connor, T.E., Vijayasekaran, S., Mar, C.B.D.:

Primary care management of otitis media among Australian

children. Med. J. Aust. 191 (2009)

38. Maw, R., Bawden, R.: Spontaneous resolution of severe chronic

glue ear in children and the effect of adenoidectomy, tonsillec-

tomy, and insertion of ventilation tubes (grommets). BMJ 306,

756–760 (1993)

39. Kubba, H.: Quality of life assessment in paediatric otolaryngol-

ogy. MD thesis, University of Glasgow (2004)

40. Rosenfeld, R.M., Bhaya, M.H., Bower, C.M., Brookhouser, P.E.,

Casselbrant, M.L., Chan, K.H., Cunningham, M.J., Derkay, C.S.,

Gray, S.D., Manning, S.C., Messner, A.H., Smith, R.J.: Impact of

tympanostomy tubes on child quality of life. Arch. Otolaryngol.

Head Neck Surg. 126, 585–592 (2000)

41. Paradise, J.L.: Middle ear problems associated with cleft palate.

An internationally-oriented review. Cleft Palate J. 12, 17–22

(1975)

42. Bluestone, C.D.: Eustachian tube obstruction in the infant with

cleft palate. Ann. Otol. Rhinol. Laryngol. 80(Suppl 2), 1–30

(1971)

43. Rawlins, M.D., Culyer, A.J.: National Institute for Clinical

Excellence and its value judgments. BMJ 329, 224–227 (2004)

44. Claxton, K., Ginnelly, L., Sculpher, M., Philips, Z., Palmer, S.: A

pilot study on the use of decision theory and value of information

analysis as part of the NHS Health Technology Assessment

programme. Health Technol. Assess. Winch. Engl. 8, 1–103

(2004)

45. Schonweiler, R., Lisson, J.A., Schonweiler, B., Eckardt, A., Ptok,

M., Trankmann, J., Hausamen, J.E.: A retrospective study of

hearing, speech and language function in children with clefts fol-

lowing palatoplasty and veloplasty procedures at 18–24 months of

age. Int. J. Pediatr. Otorhinolaryngol. 50, 205–217 (1999)

46. Kalcioglu, M.T., Cokkeser, Y., Kizilay, A., Ozturan, O.: Follow-

up of 366 ears after tympanostomy tube insertion: why is it

draining? Otolaryngol. Head Neck Surg. 128, 560–564 (2003)

47. Schonweiler, R., Schonweiler, B., Schmelzeisen, R.: Hearing

capacity and speech production in 417 children with facial cleft

abnormalities. HNO 42, 691–696 (1994)

48. Paradise, J.L., Bluestone, C.D.: Early treatment of the universal

otitis media of infants with cleft palate. Pediatrics 53, 48–54

(1974)

49. Britten, N., Ukoumunne, O.: The influence of patients’ hopes of

receiving a prescription on doctors’ perceptions and the decision

to prescribe: a questionnaire survey. BMJ 315, 1506–1510 (1997)

50. Macfarlane, J., Holmes, W., Macfarlane, R., Britten, N.: Influ-

ence of patients’ expectations on antibiotic management of acute

lower respiratory tract illness in general practice: questionnaire

study. BMJ 315, 1211–1214 (1997)

51. Russell, C., Black, O., Dutt, D., Ray, A., Devlin, M., Wynne, D.:

Are ventilation tubes (grommets) in cleft children truly associated

with increased complication rates? Results of a nested case

control study of cleft and non-cleft children. Br. J. Oral Max-

illofac. Surg. 50, S2–S3 (2012)

Cost-effectiveness analysis of grommets-led care pathway

123