Objectives: To investigate whether it is clinically effective and cost-effective to provide (i) a unilateral cochlear implant for severely to profoundly deaf people (using or not using hearing aids), and (ii) a bilateral cochlear implant for severely to profoundly deaf people with a single cochlear implant (unilateral or unilateral plus hearing aid).
Data sources: Main electronic databases [MEDLINE; EMBASE; Cochrane Database of Systematic Reviews; CENTRAL; NHS EED; DARE; HTA (NHS-CRD); EconLit; National Research Register; and ClinicalTrials.gov] searched in October 2006, updated July 2007.
Review methods: A systematic review of the literature was undertaken according to standard methods. A state-transition (Markov) model of the main care pathways deaf people might follow and the main complications and device failures was developed.
Results: The clinical effectiveness review included 33 papers, of which only two were RCTs. They used 62 different outcome measures and overall were of moderate to poor quality. All studies in children comparing one cochlear implant with non-technological support or an acoustic hearing aid reported gains on all outcome measures, some demonstrating greater gain from earlier implantation. The strongest evidence for an advantage from bilateral over unilateral implantation was for understanding speech in noisy conditions (mean improvement 13.2%, p < 0.0001); those receiving their second implant earlier made greater gains. Comparison of bilateral with unilateral cochlear implants plus an acoustic hearing aid was compromised by small sample sizes and poor reporting, but benefits were seen with bilateral implants. Cochlear implants improved children's quality of life, and those who were implanted before attending school were more likely to do well academically and attend mainstream education than those implanted later. In adults, there was a greater benefit from cochlear implants than from non-technological support in terms of speech perception. Increased age at implantation may reduce effectiveness and there is a negative correlation between duration of deafness and effectiveness. Speech perception measures all showed benefits for cochlear implants over acoustic hearing aids [e.g. mean increase in score of 37 points in noisy conditions (p < 0.001) with BKB sentences]; however, prelingually deafened adults benefited less than those postlingually deafened (mean change scores 20% versus 62%). For unilateral versus bilateral implantation, benefits in speech perception were significant in noisy conditions on all measures [e.g. 76% for HINT sentences (p < 0.0001)]. Quality of life measured with generic and disease-specific instruments or by interview mostly showed significant gains or positive trends from using cochlear implants. The Markov model base-case analysis estimated that, for prelingually profoundly deaf children, the incremental cost-effectiveness ratio (ICER) for unilateral implantation compared with no implantation was 13,413 pounds per quality-adjusted life-year (QALY). Assuming the utility gain for bilateral implantation is the same for adults and children, the ICERs for simultaneous and sequential bilateral implantation versus unilateral implantation were 40,410 pounds and 54,098 pounds per QALY respectively. For postlingually sensorineurally profoundly deaf adults, the corresponding ICERs were 14,163 pounds, 49,559 pounds and 60,301 pounds per QALY respectively. Probabilistic threshold analyses suggest that unilateral implants are highly likely to be cost-effective for adults and children at willingness to pay thresholds of 20,000 pounds or 30,000 pounds per QALY. There are likely to be overall additional benefits from bilateral implantation, enabling children and adults to hold conversations more easily in social situations.
Conclusions: Unilateral cochlear implantation is safe and effective for adults and children and likely to be cost-effective in profoundly deaf adults and profoundly and prelingually deaf children. However, decisions on the cost-effectiveness of bilateral cochlear implants should take into account the high degree of uncertainty within the model regarding the probable utility gain.