Participants were residents who volunteered to be screened for hearing loss at a community based mobile hearing clinic (MHC) from December 2015 to June 2017. They are likely to be those who suspect that they may have some hearing issue or want to assure themselves that they are fine. Participants who were found to have (1) at least moderate hearing loss (≥40 dB HL) in at least one ear, (2) no prior usage of HA in the last ten years and (3) free from ear related medical complications and (4) had a Mini-Mental State Examination score ≥18 were eligible for this study. Eligible participants were then offered the HA services in the MHC.
The MHC is a community outreach program staffed by a team of audiologists, research assistants and trained audio technicians that offers hearing diagnostic tests, referrals general practitioners (GP) or ENT specialist clinic, HA fitting and short-term post-fitting audiological rehabilitation services. Sited in a purpose-built truck chassis, the MHC has been specially retrofitted with two sound-treated rooms and audiological equipment for the delivery of hearing tests, HA fitting and aural rehabilitation. During the study, the MHC was parked in 38 community centres or near-by neighbourhoods to provide hearing test and HA fitting services to residents.
Hearing Screening at MHC
Research assistants and audiologists screened participants for hearing loss through a series of steps. Otoscopy examination was performed by research assistants to inspect for outer ear conditions such as impacted ear cerumen, ear infection, perforated tympanic membrane and presence of any foreign body in the ear canal . A trained audio technician or audiologist examined middle ear function using tympanometry to detect ear related medical complications such as ear infection, otitis media and middle ear effusion . Pure tone audiometry (PTA) test was then performed to determine hearing levels at 0.25kHz to 8kHz. To facilitate the categorization of hearing loss severity, the pure tone average was computed by averaging hearing levels at 0.5, 1, 2 and 4 kHz . Participants with moderate hearing loss in at least one ear, i.e. PTA ≥40 dB HL were recruited into the study. Participants were also screened for cognitive impairment using the MMSE questionnaire.
Participants with impacted ear cerumen or ear related medical complication were referred to general practitioners or public hospitals and were excluded from the study. Participants with MMSE scores <18 or those with prior usage of HA in the last 10 years were also excluded from the study.
HA Selection, Fitting & Aural Rehabilitation at MHC
After hearing screening, eligible participants who were keen to adopt HA were scheduled for HA selection based on treatment groups they were randomly assigned to (please see study design). During the HA selection session, audiologist discussed with study participant his/her hearing test result, importance, benefits, limitations, and the different types of HA. HA use in both ears were encouraged for participants with binaural hearing loss. Upon decision to adopt HA use and participate in the study, enrolled study participants were then scheduled for HA fitting within the following two weeks.
HA fitting was conducted by an audiologist in accordance to the American Speech Language Hearing Association (ASHA) HA fitting guidelines . Physical fit of HA was checked by the audiologist and verified using a probe microphone measurement. To ensure competency in HA use, participants were briefed on usage, care, and maintenance of the device. Participants practiced HA insertion and removal, turning device on and off, adjusting volume, battery changing and cleaning of HA. Simplified user manual and leaflet that detailed device use, care and maintenance was also provided.
Aural rehabilitation was scheduled two weeks post HA fitting, in a small group setting at the community centre. The aim of the rehabilitation was to re-emphasize use, care, and maintenance of HA, re-learn hearing with the use of the device, practise communication skills and repair strategies and clarify/discuss any issues participants were having with regards to HA use. Two additional rehabilitation sessions were given to participants at one- and three-months post HA fitting. Participants that required more sessions could make an appointment at the MHC.
To establish effectiveness of the HA intervention, suitable controls were required. Residents with hearing loss and refused HA may not be viable controls as there may be systematic differences (e.g. socioeconomic differences) between those who took up the HA and those who did not. An ideal control group would be eligible participants for whom we can delay the HA intervention and rehabilitation.
A prospective delayed-start randomized design was used for this study . Participants who had met the inclusion criteria of this study were randomized into one of the following two groups – (1) immediate-start (Fitted) group where HA was fitted two weeks after HA selection or (2) delayed-start (Not Fitted) group where HA selection and fitting was delayed for three months (Figure 1). After three months, the Not Fitted group were scheduled for HA selection, fitting and followed-up with aural rehabilitation services at the MHC like the Fitted group.
Participant’s demographic, employment status and occupation information were collected during the participant’s first visit to the MHC. Outcomes were collected for both Fitted and Not Fitted groups at baseline and three months (Figure 1). For this economic evaluation, the study outcomes of the Fitted group were compared with the Not Fitted group three months after the index visit to the MHC (Figure 1).
Using the RANDBETWEEN(0,1) function in Microsoft Excel Version 2004, community centres with random number “0” were assigned to the immediate treatment group, while the community centres with random number “1” were allocated to the delayed treatment group. The MHC was assigned to each community centre on a weekly basis.
Sample Size Calculation
To detect a difference in health-related quality of life (HRQoL) as low as 0.06  with an expected standard deviation of 0.2, statistical power of 80%, significance level of 5% and an attrition rate of 20%, a total sample size of 480 participants was required for Fitted (n=240) and Not Fitted (n=240) groups.
Cost Utility Analysis
Cost utility analysis (CUA) was used to compare the cost-effectiveness of being fitted with a HA combined with short-term post audiological rehabilitation (Fitted) with the control group who received no treatment (Not Fitted). Two outcomes were measured for this study – (1) patient-reported utility, using the quality adjusted life years (QALYs) metric and (2) total costs. QALY is a measure of health as a combination of duration of life and health related quality of life (HRQoL). Health utility index (HUI-3) is a psychometric questionnaire that this study’s participants completed in order to measure change in HRQoL. The HUI-3 comprises of eight attributes – vision, hearing, speech, ambulation, dexterity, emotion, cognition, and pain. Each attribute has about 5 to 6 levels of ability or disability. A combination of levels across the eight attributes constitutes as a health state. The health state is converted into a utility score of 0 (dead) to 1 (perfect health). Due to the inclusion criteria of no HA usage prior to study enrolment and upon advice from the tool developers, the HUI-3 was modified to exclude questions regarding HA use (see supplementary Table S1 for modified scoring methodology). The modified HUI-3 was administered to the Not Fitted group at three months. HUI-3 scores were expressed as mean ±SD.
Estimation of Costs
Costs were estimated from a societal perspective and included direct healthcare, direct non-healthcare, and indirect costs (Figure 2). Direct healthcare costs were defined as the expenditures incurred in the healthcare management system for the diagnosis, treatment, and management of hearing loss. In this study, the direct healthcare costs included:
1. All expenditures incurred at the MHC due to hearing loss and
2. All healthcare expenditures incurred due to visits to any other healthcare institutions (such as ENT specialist clinics or primary care clinics) due to hearing loss, three months after the index visit to the MHC.
Expenditure incurred at the MHC included consultation-related (hearing loss evaluation, HA fitting, rehabilitation, and hearing tests) and device-related (HA device, maintenance and repair and consumables) costs. Using surveys, participants were also asked through face to face interviews to recall the number of visits made to any healthcare institutions due to hearing loss in the last three months. The number of units consumed was then multiplied by the norm cost of each visit (Table S2). To capture direct non-healthcare related costs, participants were asked to estimate transportation costs incurred during visits to MHC and/or visits to any other healthcare institutions in the last three months due to hearing loss.
Indirect costs were defined as productivity loss due to hearing impairment Besides study participants, we accounted for productivity losses incurred by working family caregivers who may have taken time off from work to accompany study participants to the MHC and/or other healthcare institutions. Productivity losses for participants and family caregiver were measured in units in time (days) and monetised using the human capital approach . The method of estimating indirect costs were different for working and non-working adult participants. Working participants were asked to estimate the number of days absent from work due to hearing impairment in the three months after the first visit to MHC. Indirect costs were calculated by multiplying number of days absent from work due to hearing loss multiplied by the median earnings (based on individual’s occupation) per capita per day. For retirees or home makers, the estimated number of days taken to visit MHC and/or any healthcare institution due to hearing loss was multiplied by the median market wage for housekeeping (i.e. SGD 550). During the face-to-face interviews at the three-month timepoint, study participants were asked if a working family caregiver accompanied them to the MHC and/or other healthcare institutions during the last three months. Information on occupation of the working family caregiver was recorded, and similar methods was used to calculate productivity losses.
Descriptive analysis was used to compare the baseline characteristics of Fitted and Not Fitted Groups. Costs for Fitted group (MHC costs, healthcare costs, transport costs and productivity losses) were compared with Not Fitted group (healthcare costs, transport costs and productivity losses) (Figure 2). Costs were expressed as mean costs (standard deviation) per patient in 2017 Singapore dollar.
The incremental cost effectiveness ratio (ICER), otherwise known as cost per QALY, was the primary outcome measure of this CUA. ICER was computed by taking the difference between costs of the groups divided by the difference in QALY produced by the two groups. Health interventions with a cost per QALY of <50,000 USD/QALY is cost effective . As cost per QALY decreases, the intervention becomes more cost effective.
ICER was computed based on the primary data and then extrapolated to five years with the following assumptions:
1. 25%, 50% and 70% of the participants continued using the HA for five years
2. Utility gained at three months was constant for five years
3. Healthcare utilization and productivity losses incurred at three months for both groups was projected for five years with a 5% discount rate.
As part of the services of the MHC, participants were followed up for one year. Through telephone interviews, research assistants asked the Fitted group participants if they were still using the HA (Table S3). With a response rate of 66%, 71.4% were still using the device after one year. Currently, there is no evidence of prolonged utility gain over time after HA intervention. Based on the National Institute for Health and Care Excellence (NICE) guidelines, a constant rate of benefit to quality of life for everyone using HA, regardless of age, duration of HA use or level of hearing loss can be assumed . A local study identified that the average usage period of the device to be five years . Hence, we assumed constant utility over the five-year usage period of the device.
To establish a 95% Confidence Interval (CI) for point estimates of ICER, bootstrapping was performed that generated 1000 predicted values for the cost-utility ratio. Similar scenarios were projected with 50% and 25% of participants using HA for 5 years. The computation of costs and QALY were done using Microsoft Excel Version 2004. All other analysis was conducted using R statistical software version 3.4.