Hearing assessment

Most of the studies used pure tone audiometry as the main auditory assessment method, and this measures hearing sensitivity. Self-reported or assessor-reported hearing loss was used by 10 studies,^{Reference Yu and Woo41–Reference Gallagher, Kiss, Lanctot and Herrmann44,Reference Yang and Gu46,Reference Gurgel, Ward, Schwartz, Norton, Foster and Tschanz48,Reference Heward, Stone, Paddick, Mkenda, Gray and Dotchin51,Reference Benito-León, Mitchell, Vega and Bermejo-Pareja64,Reference DeVore71,Reference Heywood, Gao, Nyunt, Feng, Chong and Lim72} and in 1 paper the assessment method was not reported.^{Reference Han, Tang and Ma50} Some studies used central auditory function tests, such as dichotic digits test, speech audiometry, synthetic sentence identification with ipsilateral competing message and so on.^{Reference Uhlmann, Larson, Rees, Koepsell and Duckert7,Reference Gates, Cobb, Linn, Rees, Wolf and D'Agostino49,Reference Lister, Harrison Bush, Andel, Matthews, Morgan and Edwards56,Reference Quaranta, Coppola, Casulli, Barulli, Lanza and Tortelli59,Reference Rahman, Mohamed, Albanouby and Bekhet61–Reference Gates, Gibbons, McCurry, Crane, Feeney and Larson63,Reference Gates, Anderson, Feeney, McCurry and Larson65,Reference Frisina and Frisina68,Reference Strouse, Hall and Burger70} The definition of hearing loss from the World Health Organization⁷³ was used by most studies, but the frequencies tested ranged from: 500 to 4000 Hz,^{Reference Deal, Betz, Yaffe, Harris, Purchase-Helzner and Satterfield45,Reference Fischer, Cruickshanks, Schubert, Pinto, Carlsson and Klein47,Reference Bruckmann and Pinheiro54,Reference Bucks, Dunlop, Taljaard, Brennan-Jones, Hunter and Wesnes55,Reference Deal, Sharrett, Albert, Coresh, Mosley and Knopman57,Reference Lin, Ferrucci, Metter, An, Zonderman and Resnick60,Reference Tay, Kifley, Lindley, Landau, Ingham and Mitchell66} 1000 to 4000 Hz,^{Reference Van Boxtel, Beijsterveldt van, Houx, Anteunis, Metsemakers and Jolles67} 250 to 2000 Hz^{Reference Zhang, Yang, Liu, Huang, Chen and Zhang58} and 500 to 3000 Hz.^{Reference Uhlmann, Larson, Rees, Koepsell and Duckert7} One study used the hearing threshold at 4 kHz only to separate the different groups of hearing loss.^{Reference Frisina and Frisina68}

Cognitive assessment

The cognitive assessment tool that was used by most researchers was the Mini Mental State Examination^{Reference Uhlmann, Larson, Rees, Koepsell and Duckert7,Reference Schubert, Cruickshanks, Fischer, Pinto, Chen and Huang40,Reference Vaccaro, Zaccaria, Colombo, Abbondanza and Guaita42,Reference Yang and Gu46,Reference Fischer, Cruickshanks, Schubert, Pinto, Carlsson and Klein47,Reference Gates, Cobb, Linn, Rees, Wolf and D'Agostino49,Reference Han, Tang and Ma50,Reference Iliadou, Bamiou, Sidiras, Moschopoulos, Tsolaki and Nimatoudis53,Reference Bruckmann and Pinheiro54,Reference Zhang, Yang, Liu, Huang, Chen and Zhang58,Reference Quaranta, Coppola, Casulli, Barulli, Lanza and Tortelli59,Reference Lin, Ferrucci, Metter, An, Zonderman and Resnick60,Reference Idrizbegovic, Hederstierna, Dahlquist, Kämpfe Nordström, Jelic and Rosenhall62,Reference Benito-León, Mitchell, Vega and Bermejo-Pareja64,Reference Tay, Kifley, Lindley, Landau, Ingham and Mitchell66,Reference DeVore71,Reference Heywood, Gao, Nyunt, Feng, Chong and Lim72} or its modified version, the Modified Mini Mental State tool.^{Reference Deal, Betz, Yaffe, Harris, Purchase-Helzner and Satterfield45,Reference Gurgel, Ward, Schwartz, Norton, Foster and Tschanz48,Reference Strouse, Hall and Burger70} A score below 24 on the Mini Mental State Examination^{Reference Folstein, Folstein and McHugh74} was considered abnormal by most studies.^{Reference Uhlmann, Larson, Rees, Koepsell and Duckert7,Reference Schubert, Cruickshanks, Fischer, Pinto, Chen and Huang40,Reference Yang and Gu46,Reference Fischer, Cruickshanks, Schubert, Pinto, Carlsson and Klein47,Reference Iliadou, Bamiou, Sidiras, Moschopoulos, Tsolaki and Nimatoudis53,Reference Zhang, Yang, Liu, Huang, Chen and Zhang58,Reference Tay, Kifley, Lindley, Landau, Ingham and Mitchell66} The cut-off score was 27 in two studies.^{Reference DeVore71,Reference Heywood, Gao, Nyunt, Feng, Chong and Lim72} In another study, Mini Mental State examination score thresholds were adjusted for education level.^{Reference Han, Tang and Ma50} Other cognitive assessment tests used included the Montreal Cognitive Assessment,^{Reference Iliadou, Bamiou, Sidiras, Moschopoulos, Tsolaki and Nimatoudis53,Reference Lister, Harrison Bush, Andel, Matthews, Morgan and Edwards56} Cognitive Drug Research Computerised Assessment System,^{Reference Bucks, Dunlop, Taljaard, Brennan-Jones, Hunter and Wesnes55} verbal fluency,^{Reference Vaccaro, Zaccaria, Colombo, Abbondanza and Guaita42,Reference Bucks, Dunlop, Taljaard, Brennan-Jones, Hunter and Wesnes55,Reference Deal, Sharrett, Albert, Coresh, Mosley and Knopman57,Reference Benito-León, Mitchell, Vega and Bermejo-Pareja64} National Adult Reading Test,^{Reference Bucks, Dunlop, Taljaard, Brennan-Jones, Hunter and Wesnes55} Delayed Word Recall Test,^{Reference Deal, Sharrett, Albert, Coresh, Mosley and Knopman57} Digit Symbol Substitution Test,^{Reference Deal, Sharrett, Albert, Coresh, Mosley and Knopman57} Wechsler Adult Intelligence Test,^{Reference Zhang, Yang, Liu, Huang, Chen and Zhang58} Clinical Dementia Rating Scale,^{Reference Gallagher, Kiss, Lanctot and Herrmann44,Reference Quaranta, Coppola, Casulli, Barulli, Lanza and Tortelli59,Reference Heywood, Gao, Nyunt, Feng, Chong and Lim72} Frontal Assessment Battery,^{Reference Quaranta, Coppola, Casulli, Barulli, Lanza and Tortelli59} Free and Cued Selective Reminding Test,^{Reference Lin, Ferrucci, Metter, An, Zonderman and Resnick60} Trail Making Test (parts A and B),^{Reference Schubert, Cruickshanks, Fischer, Pinto, Chen and Huang40,Reference Vaccaro, Zaccaria, Colombo, Abbondanza and Guaita42,Reference Lin, Ferrucci, Metter, An, Zonderman and Resnick60,Reference Gates, Gibbons, McCurry, Crane, Feeney and Larson63,Reference Benito-León, Mitchell, Vega and Bermejo-Pareja64} Rey–Osterrieth Complex Figure Recall Test,^{Reference Vaccaro, Zaccaria, Colombo, Abbondanza and Guaita42} Stroop Letter and Category Fluency,^{Reference Lin, Ferrucci, Metter, An, Zonderman and Resnick60,Reference Gates, Gibbons, McCurry, Crane, Feeney and Larson63} the American version of the Nelson Adult Reading Test,^{Reference Lin, Ferrucci, Metter, An, Zonderman and Resnick60} Abbreviated Memory Inventory for the Chinese,^{Reference Yu and Woo41} Cambridge Cognitive Examination,^{Reference Rahman, Mohamed, Albanouby and Bekhet61} Clock Drawing Test,^{Reference Vaccaro, Zaccaria, Colombo, Abbondanza and Guaita42,Reference Gates, Gibbons, McCurry, Crane, Feeney and Larson63} Cognitive Abilities Screening Instrument,^{Reference Gates, Gibbons, McCurry, Crane, Feeney and Larson63,Reference Gates, Anderson, Feeney, McCurry and Larson65} Letter-Digit Symbol Test,^{Reference Van Boxtel, Beijsterveldt van, Houx, Anteunis, Metsemakers and Jolles67} Auditory Verbal Learning Test,^{Reference Van Boxtel, Beijsterveldt van, Houx, Anteunis, Metsemakers and Jolles67} Clinical Dementia Rating Scale^{Reference Gates, Karzon, Garcia, Peterein, Storandt and Morris69} and Storandt Battery.^{Reference Strouse, Hall and Burger70}

Hearing loss in people with mild cognitive impairment

Peripheral hearing loss

Across six cohort studies comparing mild cognitive impairment between people with peripheral hearing loss (n = 8235) and without peripheral hearing loss (n = 17 891), the risk ratio was 2.06 (random-effects, 95 per cent CI = 1.35 to 3.15; p = 0.0008; I² = 97 per cent); significantly more people with peripheral hearing loss had mild cognitive impairment compared with those without, although a high level of statistical heterogeneity was evident (Figure 5).

Fig. 5.

Incidence of mild cognitive impairment with and without peripheral hearing loss (risk ratio). M-H = Mantel-Haenszel; CI = confidence interval

Across two cohort studies reporting the between-group difference as a hazard ratio (i.e. no raw data were available) comparing mild cognitive impairment between people with and without peripheral hearing loss, the hazard ratio was 1.40 (random-effects, 95 per cent CI = 1.64 to 1.95; p < 0.00001; I² = 0 per cent); significantly more people with peripheral hearing loss had mild cognitive impairment compared with those without peripheral hearing loss (Figure 6).

Fig. 6.

Incidence of mild cognitive impairment with and without peripheral hearing loss (hazard ratio). SE = standard error; IV = Instrumental variable; CI = confidence interval

The between-group difference for one cohort study reporting the outcome as an odds ratio (i.e. no raw data were available) comparing mild cognitive impairment between people with and without peripheral hearing loss was 1.70 (random-effects, 95 per cent CI = 1.30 to 2.22; p = 0.0001); significantly more people with peripheral hearing loss had mild cognitive impairment compared with those without peripheral hearing loss (Figure 7).

Fig. 7.

Odds ratio outcomes in between-group difference (with and without peripheral hearing loss). SE = standard error; IV = Instrumental variable; CI = confidence interval

The between-group differences for one cohort study reporting the outcome as risk ratio (i.e. no raw data were available) comparing mild cognitive impairment between people with and without mild, moderate or severe peripheral hearing loss were: mild 1.26 (random-effects, 95 per cent CI = 1.15 to 1.38; p < 0.0001), moderate 1.29 (random-effects, 95 per cent CI = 1.13 to 1.47; p = 0.0002) and severe 1.37 (random-effects, 95 per cent CI = 1.06 to 1.7; p = 0.02); significantly more people with peripheral hearing loss had mild cognitive impairment compared with those without peripheral hearing loss in all categories (Figure 8).

Fig. 8.

Risk ratio outcomes in between-group difference (with and without peripheral hearing loss). SE = standard error; IV = Instrumental variable; CI = confidence interval

Hearing impairment and early dementia

The outcomes of the studies on hearing impairment and mild cognitive impairment are presented in Table 6.

Table 6.

Outcomes from the studies of prevalence of hearing loss amongst patients with mild cognitive impairment

Similar to the studies described above, the results in this category vary considerably. Four studies out of 15 failed to find any strong association between mild cognitive impairment and hearing loss.^{Reference Heward, Stone, Paddick, Mkenda, Gray and Dotchin51,Reference MacDonald, Keller, Brewster and Dixon52,Reference Benito-León, Mitchell, Vega and Bermejo-Pareja64,Reference Heywood, Gao, Nyunt, Feng, Chong and Lim72} Some studies found a strong association between mild cognitive impairment and central hearing loss.^{Reference Han, Tang and Ma50,Reference Lister, Harrison Bush, Andel, Matthews, Morgan and Edwards56,Reference Rahman, Mohamed, Albanouby and Bekhet61–Reference Gates, Gibbons, McCurry, Crane, Feeney and Larson63,Reference Gates, Anderson, Feeney, McCurry and Larson65,Reference Gates, Karzon, Garcia, Peterein, Storandt and Morris69,Reference Strouse, Hall and Burger70}

Peripheral hearing loss was found to be significantly associated with mild cognitive impairment in two studies,^{Reference Uhlmann, Larson and Koepsell32,Reference DeVore71} but no such association was observed in two other studies.^{Reference Idrizbegovic, Hederstierna, Dahlquist, Kämpfe Nordström, Jelic and Rosenhall62,Reference Gates, Karzon, Garcia, Peterein, Storandt and Morris69} There was considerable variance in the number of recruited participants per study from 20^{Reference Strouse, Hall and Burger70} to 2146.^{Reference Benito-León, Mitchell, Vega and Bermejo-Pareja64} There was no consistency in the outcome measures used (e.g. in DeVore's study (1992), the participants did not undergo any formal audiometry^{Reference DeVore71} whereas in the study by Gates et al. (1995) they underwent several tests including pure tone audiometry, speech audiometry, synthetic sentence identification with ipsilateral competing message, distortion product otoacoustic emissions, and auditory brainstem response.^{Reference Gates, Karzon, Garcia, Peterein, Storandt and Morris69} Lister et al. (2016) sought to identify an association between cortical auditory evoked potentials and mild cognitive impairment by means of changes to the P1-N1-P2 complex.^{Reference Lister, Harrison Bush, Andel, Matthews, Morgan and Edwards56} Their findings might be consistent with further changes of inhibition, in the presence of mild cognitive impairment, with fewer overall resources being available to devote to the task. The cognitively normal older adults had significantly larger P2 amplitudes than those with probable mild cognitive impairment (3.7 (SD, 1.9) vs 1.8 (SD, 1.0)), but P1-N1-P2 latencies were similar in both groups. Gates et al. (2010) recruited 313 patients from the longitudinal Adult Changes in Thought study and grouped them according to their cognitive function as normal (n = 232), memory-impaired (n = 60) or demented (n = 21).^{Reference Gates, Gibbons, McCurry, Crane, Feeney and Larson63} One SD poorer executive function was associated with a −9.2 per cent point difference in synthetic sentence identification with ipsilateral competing message, −15 per cent point difference in Dichotic Sentence Identification Test and −8.4 per cent point difference in Dichotic Digits Test. Finally, Uhlmann et al. (1989) postulated that the risk of dementia was increased for mild and moderate hearing loss and reached statistical significance for hearing loss of more than 40 dB (p < 0.05).^{Reference Uhlmann, Larson, Rees, Koepsell and Duckert7}

Early dementia or cognitive decline and hearing impairment

The outcomes of studies of early dementia or cognitive decline amongst patients with hearing impairment are presented in Table 7. A significant correlation between hearing loss and incidence of early dementia or cognitive decline was reported in 7 out of 11 studies included in this category.^{Reference Yu and Woo41–Reference Fischer, Cruickshanks, Schubert, Pinto, Carlsson and Klein47} The number of recruited participants ranged from 1662^{Reference Gates, Cobb, Linn, Rees, Wolf and D'Agostino49} to 13 731.^{Reference Yang and Gu46} The means of assessing hearing varied between the studies (pure tone audiometry^{Reference Schubert, Cruickshanks, Fischer, Pinto, Chen and Huang40,Reference Deal, Betz, Yaffe, Harris, Purchase-Helzner and Satterfield45} versus self-reported.^{Reference Yu and Woo41–Reference Gallagher, Kiss, Lanctot and Herrmann44,Reference Yang and Gu46} The majority of studies used the Mini Mental State Examination or Modified Mini Mental State to assess cognition, but in a few studies the Clinical Dementia Rating Scale,^{Reference Gallagher, Kiss, Lanctot and Herrmann44} abbreviated Memory Inventory for the Chinese^{Reference Yu and Woo41} and subjective cognitive decline tests^{Reference Curhan, Willett, Grodstein and Curhan43} were used.

Table 7.

Studies of incidence of early dementia or cognitive decline amongst patients with hearing impairment

CI = confidence interval; HR = hazard ratio

In a study by Gurgel et al. (2014), all-cause dementia was observed in 16.3 per cent of patients with hearing loss (at baseline) but only in 12.1 per cent of those without hearing loss (p < 0.001).^{Reference Gurgel, Ward, Schwartz, Norton, Foster and Tschanz48} Following multivariate analysis, hearing loss was found to be an independent factor for dementia. When evaluating the subgroup of patients who were cognitively intact at baseline and taking into account all covariates, hearing loss was not found to be a strong independent risk factor for developing dementia (p = 0.09). Gates et al. (1996) postulated that central hearing dysfunction precedes the emergence of cognitive decline and dementia and recommended that both peripheral and central hearing tests be obtained as part of the general health evaluation of the elderly.^{Reference Gates, Cobb, Linn, Rees, Wolf and D'Agostino49}