Search strategy

We selected studies through electronic searches on five health databases: LILACS; PsychINFO; PubMed; SCOPUS and Web of Science, using the following combination of descriptors and filters: ("P300" OR "electrophysiological" OR "evoked cognitive potentials" OR "evoked potential" OR "auditory evoked potentials") AND "elderly"; Activated filters: Research in humans, Aged: 65+ years, 80 and over: 80+ years.

Study selection

Two independent "blind" reviewers conducted the search for titles and abstracts in July 2016. The studies that addressed the guiding questions were selected for full reading using a selection protocol previously created in Microsoft Excel^®2010. In addition, a strategy of cross-reference searching was performed, by which all the references from the selected studies that addressed the guiding questions and met the inclusion criteria were selected for full reading. Reading and discussion of the studies were carried out intensively in a consensus meeting held on February 05, 2017 with the presence of authors.

Figure 1 depicts a summary of the steps for the study identification and selection procedure with numerical data.

Figure 1 Summary of the study search and selection procedure. São Carlos, 2017. 

Data collection, analysis and synthesis of results

For the analysis and extraction of the data from the articles retrieved, a form devised by the authors collecting the following information was used: authors; publication year; country; study design; aims; sociodemographic information on the sample; auditory or visual ERP; complementary evaluation; statistical tests used; and main findings. The information was gathered in spreadsheets. The quantitative data were presented as simple frequency, percentage and mean. The qualitative analyses of the studies were carried out together with all the authors and presented in text and tables.

As a secondary and complementary result, the studies that reported amplitude and latency information for the healthy elderly were described and summarized in the review.

General characteristics of the studies

Of the 26 articles selected, 22 were cross-sectional studies and four randomized clinical trials. Regarding the publication date, most were published in 2014 (30.8%) and 2013 (30.8%). The studies were conducted in 12 different countries, 26.9% in the United States, and 11.6% in Japan, Italy, Brazil, Canada, and China.

Concerning the sample characteristics, the studies included information from 1,463 participants, of which 940 were healthy elderly. Regarding participant age, 23 studies cited the mean age of the healthy elderly groups, where the lowest average found was 60.0 years¹³ and the highest was 75.8 years.¹⁴ The majority (n = 10) had a mean age of between 70 and 74 years.^2,15-20 Two studies did not cite the mean age, only the age range of the participants, which was 15-85 years²¹ for one study and 42-78 years for the other.²²

With regard to the type of stimulus used for obtaining the P300, 50% used auditory stimulation and 50% visual stimulation, and each study used different settings for recording the ERP. Most studies (92.3%) used complementary evaluations and 23% also included electrophysiological evaluation of waves other than the P300.

P300 evaluation

The evaluation of P300 was used in 53.8% of the articles to identify the influence of age on cognitive processing performance.^{5,15-17,19,21,23-29} In 19.2% of the studies, the P300 was used to compare the electrophysiological evaluation of healthy and cognitively impaired participants,^{6,13,14,30,31} 15.4% used electrophysiological evaluation to monitor the effects of therapies in clinical trials,^32-35 one study evaluated the P300 according to age, depression, and cognition,² one evaluated depression in post-stroke patients²² and one paper studied the difference between older monolingual and bilingual adults.¹⁸ A summary of the results is given in Table 1.

P300 values

Of the 26 studies, eight reported means and standard deviations of the P300 values (amplitude and latency), as summarized in Table 2. These values are only for the healthy elderly groups, serving as a reference for discussion of the subsequent studies.

The statistical procedures most used for the treatment of P300 data were those recommended when the data adhere to normality. The variance analysis by ANOVA was the most used (61.5%), the t-test was the most used procedure for comparison of groups and therapy effects (28%), and the non-parametric option predominantly used to compare the ranking between groups was the Mann-Whitney test. The Spearman (16%) and Pearson (8%) correlation coefficients were used for correlation analyses.

Regarding the most evident findings, half of the articles found an influence of age on P300 values. In 19.2% of the studies, associations between neurocognitive impairment and worse electrophysiological evaluation^{6,13,14,30,31} were found. Education was an item commonly not associated with P300 evaluation.^2,16,27 Two articles presented contrasting results for the association between sex and P300 values - the first study found that latency in the male group was greater with advancing years²¹ whereas the second found that response to the stimulus in the frontal region was faster in men.⁵ Four studies performed an association analysis between neuropsychological and cognitive tests with P300 values, although two of these failed to identify a relationship between the two variables. In both studies that demonstrated this association, the authors confirmed a relationship between P300 and the working memory domain in healthy elderly²⁹ and for general scores on word list recognition, constructional praxis, and word fluency neuropsychological tests of the MMSE and CERAD-K in elderly with Alzheimer's disease.¹⁴ Finally, one study examined the difference between monolingual and bilingual elderly in terms of brain response during the performance of multiple tasks.¹⁸

The amplitude and latency values of P300 in healthy elderly are given in Table 2, showing no pattern of response to the stimuli among the studies described. The amplitude values ranged from 2.2 µV (±0.2)³¹ to 18.5 µV (±9.2)³⁴ and latency values from 320 ms (±20.2)²² to 484 ms (±21),³¹ independently of the channel (electrode) used to capture the response to the stimulus.

P300: sociodemographics and values in healthy elderly

Of the cross-sectional studies that included a control group, 14 evaluated the effects of healthy aging on endogenous evoked potentials.^{5,15-17,19-21,23-29} In all studies, age influenced cognitive processing. These data corroborate a review study that indicated a decrease in P300 amplitude at advanced ages.⁹ Another integrative review reported that latency tends to increase, and amplitude to decrease, with advancing age.³⁶

On average, the P300 latency increases by around 2 ms per year, as wave amplitude is linearly reduced in older people.²¹ The data found corroborates the findings of a study that estimated the latency effect of P300 in 62 healthy elderly subjects and subdivided them into three groups according to age: 60-64 years, 65-69 years and 70-74 years. The researchers found a linear significant increase of 2.85 ms per year in the age groups 65-69 years and 70-74 years.⁸

The articles also showed that the P300 amplitude values ranged from 2.2 µV to 18.5 µV and the latency values from 320 ms to 484 ms. These variations found may be related to the characteristics of the samples studied, the variables analyzed and the methods adopted in each study.

The increase in wave amplitude may result in increased use of frontal executive functions as a compensatory mechanism to adequately perform the tasks.¹⁶ The reaction time to the stimuli differed between elderly and adults groups. For the elderly, it was a mean of 555 ms, in middle-age adults 515 ms while the average among the young was 480 ms.²⁵ In one of the studies, poor performance on tasks involving working memory was associated with a higher latency value in elderly, and in relation to the amplitude, compared to the young, the elderly had higher values in frontal regions and smaller values in posterior regions.²⁹ In the elderly, the activity is more distributed in the temporal and superior temporal lobes.⁵ Besides the changes related to age, differences were identified in the parietal and frontal regions.^24,28

One of the articles analyzed showed that the changes in P300 are more pronounced in men, where latency increased by 2.3 ms/year in men and 1.6 ms/year in women.²¹ However, on topographic imaging exams, elderly men had greater distribution in the frontal region compared to women, responding to the stimulus faster.⁷

Education and cognitive tests showed no relationship with the results of the electrophysiological evaluations. However, the young had better performance on the MMSE¹⁷ and had over 10 years more education than the elderly,²⁷ a factor which may have influenced age-related P300 results. These findings corroborate the results of the study by Miranda et al. (2012)² in which no correlation was found between increase in P300 latency and education in elderly with hearing loss. A study carried out in Colombia of healthy elderly revealed an inverse relationship between latency and amplitude values; and, regarding gender and education, no differences were observed for the latency and amplitude of the P300 wave.³⁷ One study found that bilingual elderly performed better than monolingual elderly on the electrophysiological measures during cognitive tasks.¹⁸

P300: comparison between healthy elderly and those with mental disorders

Regarding the use of P300 in mental disorders, most of the studies had used event-related potentials to evaluate whether differences existed between cognitively healthy elderly and subjects with cognitive loss and/or Alzheimer's disease (AD).

Three studies found that patients with AD had lower amplitude of the P300 wave when compared to healthy elderly.^6,14,30 Although latency did not differ between the two groups in the study by Lee et al.,¹⁴ it was higher for elderly with AD compared to healthy elderly with high and low risk of dementia in the study of Asaumi et al. (2014).⁶ In addition to these investigations, another study observed better latency and amplitude values for the group without subjective memory complaints than the group with complaints.³¹ In contrast, another investigation,² despite identifying greater latency with increasing age, found no correlation of the P300 results with cognitive evaluation tests and depressive symptoms in elderly with hearing loss.

The relationship of P300 with cognitive loss and AD has been previously described in other studies. Since the P300 response is related to fundamental elements of cognitive function, it may be useful in the early diagnosis of dementia, serving as a complementary tool to existing instruments.¹¹

A literature review evaluated whether the P300 was able to estimate the risk of progression from mild cognitive impairment (MCI) to AD. Eight studies were evaluated and found that the electrode positioned in the parietal region is the most effective to evaluate this progression, evidencing increased latency and decreased amplitude. However, despite being a promising method for evaluating this progression, such investigations remain scarce, with small samples and heterogeneous results. This demonstrates a need for further studies so that this tool can be used in clinical practice.³⁸ Another meta-analysis of 13 papers investigated whether the P300 can serve as a useful neurophysiological marker to discriminate MCI and predict its progression. Differences were identified in P300 latency of patients with MCI when compared to controls and to patients with AD, indicating that the this may be useful for identifying early cognitive impairment and its progression, including for AD.⁴ A meta-analysis of twenty studies sought to characterize the P300 in probable AD compared with healthy controls and found that the amplitude was lower in individuals with AD.³⁹ Another systematic review of eight investigations found a consensus that the P300 latency is increased in elderly with AD, but no consensus on amplitude (which may also be due to the methodological variations of the studies assessed).³⁶

One of the evaluated articles in this review analyzed the association of P300 latency with the clinical stage of Parkinson's disease (PD). Latency values of the elderly with PD that exceeded two units from the standard deviation of the latency values of the healthy subjects were considered altered. In the group at initial stages of PD, 10% were considered altered and 31% at the advanced stages, showing an association between PD severity and increased P300 latency.¹³ The fact that the P300 serves as a possible predictor of PD evolution based on higher latency has been described previously.⁴⁰

Another mental disorder found in the present review was depression. Zhang et al. (2013)²² evaluated stroke patients with and without depression versus a control group. The P300 latency values were higher, and amplitude values lower, in post-stroke patients with depression compared to the no depression and healthy groups. These data corroborate a study conducted in Australia that found greater amplitude and lower latency in patients with depression compared to those without depression,⁴¹ and also another in India, that found a delay in latency among patients with depression compared to healthy controls, which was proportional to disease severity.⁴²

P300 in clinical trials with healthy elderly

Four trials with a total of 232 participants^32-35 used the P300 measurement to identify the effects of the use of Bacopa monnieri³² and arachidonic acid,³³ Tai Chi practice³⁴ and several sports modalities³⁵ on cognitive processing in healthy elderly compared to controls. In all studies, the evaluation of P300 was sensitive for detecting the effects of therapies. Protocols ranged from 12 weeks to 18 months. These trials yielded important information, such as: when the treatment was interrupted, the P300 values returned to similar values to those before treatment began.³² During the therapy period, the control group of elderly continued to show increased latency and decreased amplitude compared to the elderly undergoing intervention,³⁵ the electrophysiological evaluation may be improved with the use of other waves, such as Mismatch Negativity (MMN).³⁴

One of the trials indicated that Pz would be a more sensitive electrode for detecting the effects of cognitive processing interventions.³⁴ One of the articles reported information only for the electrode Pz³³ while another only for the electrode Cz.³⁵ Perhaps the sensitivity of the electrodes for detecting the effects depends on the characteristics of the sample of participants and type of intervention used. Therefore, it is up to the authors to decide what information from which electrode should be analyzed. Future studies could focus on the sensitivity of each electrode.

In the literature, other studies involving participants with different characteristics used the P300 to detect effects on cognitive processing of drug therapies, such as levetiracetam in the treatment of epilepsy,⁴³ use of perospirone in people with schizophrenia⁴⁴ and donepezil in the elderly with vascular dementia type.⁴⁵ The P300 has also proved sensitive in clinical trials involving additional and complementary therapies, such as vitamin B12 supplementation⁴⁶ and acupuncture.⁴⁷

In clinical trials for the diagnosis of disorders, the P300 appears to be useful for identifying cognitive loss, age-related or otherwise, cognitive changes after complex surgical procedures and diagnoses for Alzheimer's disease.³⁶ In other types of controlled clinical trials, the P300 was effective in identifying the influence of unhealthy habits, such as smoking.⁴⁸ In summary, the electrophysiological evaluation by P300 seems to be a useful and sensitive tool to detect effects on cognitive processing in therapies with clinical trials and in studies of accuracy for diagnosing disorders.

Main findings of the study

This review aimed to systematize the latest information related to the use of P300 as a tool for cognitive processing assessment in healthy elderly. We found that:

Based on these results, further investigations should be conducted in order to:

In conclusion, the reviewed scientific evidence suggests that there is an increase in latency and a decrease in amplitude with advancing age. Latency impairment is more pronounced in individuals with mental disorders, but no consensus exists with respect to amplitude. Normative values of amplitude and latency in healthy elderly cannot be established due to the variability in methodological characteristics, such as number of channels, electrode locations, impedance values, and differences in stimulation procedures. Thus, there is no evidence on the association between gender, education and the P300 in healthy elderly. In clinical trials, the P300 has proven a useful tool to evaluate the effects of interventions in the elderly, but most studies used healthy elderly only as a control group.

Based on these results, it is suggested that further investigations should be conducted to evaluate the true relationship between age, education, gender and the P300 in elderly. Finally, as this review identified a lack of studies designed to specifically investigate cognitive processing in healthy aging using the P300, it is recommended that future studies include healthy elderly as the main investigation group.