Joint associations of smartphone use and gender on multidimensional cognitive health among community-dwelling older adults: a cross-sectional study

A face-to-face survey was conducted among registered residents aged over 60 years from July 1st to October 20th, 2016, in Xiamen, China. A multistage random sampling method was used for the sampling. The primary sampling unit was the district in Xiamen (involving two urban and four rural districts in total), the secondary sampling unit was the sub-district, and the community in each sub-district was then regarded as the tertiary sampling unit. According to the proportion of eligible older adults in each community, individuals were randomly chosen after controlling for gender and age. Ultimately, a total of 3230 respondents from 44 communities in six districts were interviewed, 3061 (94.77%) of which provided valid responses. A more detailed study design and methods have been reported previously [24]. This study was approved by the ethical review committee of the School of Public Health, Xiamen University. Written informed consent was obtained from each respondent prior to the participation in the questionnaire survey.

The primary outcomes of the present study were general and multi-domain cognitive functions, which were evaluated by the Beijing version of the Montreal Cognitive Assessment (MoCA), a Chinese adaptation of the MoCA tailored to the local context. This adaptation has been reported to have good sensitivity as well as specificity for cognitive screening among Chinese elderly [25]. Six cognitive subdomains were extracted from the MoCA [26], including delayed memory (five points), visuospatial ability (four points), executive ability (four points), attention (six points), language (six points), and orientation (six points). Of note, two points were added for older adults with ≤6 years of education, while one point was added for those with 7–12 years of education (if the adjusted score exceeded 30, 30 was assumed for analyses) to overcome the educational bias of the MoCA [27].

The exposure of interest was smartphone usage, assessed by the question of “Do you use a smartphone in your daily life?”, and the two options of yes and no were provided. If the answer was yes, the respondents would be further asked “Which of the following smartphone functions did you use (check all that apply)?” offering a list of 10 items (phone call/message, Internet searching/news reading, social networking services, listening to music or radio, online shopping, video viewing, camera use, playing games, learning, and others). The number of smartphone functions used (NSFU) for each individual was further classified into three categories (0, 1, and 2+) by taking account of its distribution and explainable meaning. Participants with a score of 0 were regarded as non-smartphone users, those with a score of 1 were regarded as traditional smartphone users because the majority of these were mainly used calling and messaging, and those with a score of 2+ were regarded as advanced users due to the more advanced functions they used. The NSFU indicates the acceptance and adoption of new technology.

Additionally, several demographic and socioeconomic characteristics (age, gender, region, marital status, education level, income, and occupation) as well as medical and health factors (smoking, drinking, hypertension, diabetes, and depression) were considered as covariates. These covariates were included to account for possible confounding effects since the literature suggested that they are closely related to the exposure (smartphone use) [28] or outcome (multi-domain cognition) [29]. Among these, smoking (or alcohol consumption) was measured by asking “Do you smoke cigarettes (drink alcohol)?” and three options of “never”, “smoking (drinking) now”, and “have quit now” were provided. Chronic diseases were assessed by the question “Do you suffer from the following physician-diagnosed chronic diseases (check all that apply)?” Hypertension and diabetes were included in this list of chronic diseases. In addition, depression was assessed by the 15-item Geriatric Depression Scale (GDS-15) and a GDS < 5 would be regarded as no depression [30].

For the current study, 444 participants were excluded whose MoCA scores were very low (lower than the education-adjusted cut-offs for dementia, which were 11, 14, and 16 for ≤5, 6–8, and ≥ 9 years of education, respectively [31]) to avoid potentially occurring reverse causality since severe cognitive impairment likely impacts smartphone use. Additionally, 17 individuals with missing information with regard to smartphone use or smartphone function use were excluded. Finally, 2600 individuals were included in the study.

Firstly, demographic characteristics of the remaining 2600 participants were described via means and standard deviations or percentages. In particular, smartphone usage was summarized according to different functions stratified by gender, which were then assessed with the Chi-square test. Secondly, the characteristics of the adjusted MoCA total score and six subdomain scores were summarized according to gender specified NSFU (0, 1, and 2+) by using descriptive statistics (mean and standard deviations). The mean differences were compared via analysis of variance (ANOVA). Thirdly, general and subdomain cognitive functions were modelled on NSFU only, gender only, and NSFU and gender combined by using the seven proportional-odds cumulative logit model [32]. A four-category quantile classification was used for the total MoCA score (P₂₅, P₅₀, and P₇₅) and this quantile and six subdomains’ score were used as the ordinal dependent variables for each model, respectively. All models were adjusted for the afore-mentioned covariates. For each model, a Chi-square score test indicated the appropriateness of the ordinal logistic regression; the parallel slopes assumption was not rejected at a p-value > 0.10. Since seven parallel analyses were performed, Bonferroni corrections were applied and thus, was set to 0.007. Fourthly, similar models were used to evaluate the joint associations of gender and NSFU on general and six subdomains of cognitive function by considering them based on the following possible mixed conditions: male & NSFU = 0, male & NSFU = 1, male & NSFU = 2, female & NSFU = 0, female & NSFU = 1, and female & NSFU = 2. Odds ratios (ORs) and corresponding 95% confidence intervals (CIs) were estimated, using men who did not report to have used smartphones as the reference category. All analyses were performed using SAS version 9.4 (SAS Institute, Cary NC, 2017).

Descriptive statistics are presented in Table 1. Among the 2600 remaining participants, the mean age was 69.06 ± 7.06 years (ranging from 60 to 96 years). Male participants, who were currently married and lived in urban areas accounted for 53.88, 74.55, 54.62%, respectively. More than a quarter (25.33%) and almost two-thirds (64.51%) of respondents were illiterate and engaged in manual labor, respectively, while ~ 16% of older adults had insufficient income. Most respondents never smoked (62.38%) or drank alcohol (79.49%). More than one-third (35.04%) of the older adults suffered from hypertension while the prevalence of diabetes and depression were both ~ 10%. Overall, the education-adjusted mean MoCA score was 21.98 ± 4.61, and people who were younger, male, married, lived in urban areas, had higher education levels, had a balanced income, and had non-manual jobs were more likely to achieve higher MoCA scores.

Figure 1 shows the gender differences with regard to the characteristics of smartphone use. Nearly 30% of the sampled older adults were smartphone users, 473 (60.72%) of which were men. The top three most frequently used smartphone functions were phone call/message (men: 26.84%, women: 20.27%), social networking services (men: 13.92%, women: 12.18%), and Internet searching/news reading (men: 13.20%, women: 8.17%). More men than women used each smartphone function with significant differences in phone call/message (p < 0.001), Internet searching/news scanning (p < 0.001), and listening to music or radio (p < 0.05).

ANOVA (Table 2) indicated that, for both men and women, the mean scores of each cognitive domain differed significantly among groups with different NSFU, and higher scores in all subdomains were more likely to be attained by smartphone users with a NSFU of 2 + .

NSFU was positively associated with general cognition and the cognition of all other subdomains except for memory and orientation (model 1 in Table 3), and increasing associations were observed between them with higher NSFU level except for memory and orientation. With or without constant NSFU, significant differences were not found in general cognition between men and women; however, women significantly outperformed men in memory while being surpassed by men in visuospatial ability (models 2 and 3).

Figure 2 shows ORs with corresponding 95% CIs for joint associations of NSFU and gender on general and sub-domain cognitive function. Male with NSFU = 0 was used as the general reference group, and an increasing pattern in the odds of general and four domains of cognitive ability was observed with increasing NSFU level (except for memory and orientation). In general, men with a NSFU of 2+ had the highest odds of better cognition (OR (95% CI): 2.20 (1.63–2.96)). For non-smartphone users, women significantly performed better in memory than men (OR (95% CI): 1.60 (1.24–2.06)); however, this advantage might be tempered with higher NSFU. In contrast, with regard to their visuospatial ability, women performed significantly worse than men (OR (95% CI): 0.45 (0.35–0.58)) for non-smartphone users, and this disadvantage became insignificant for females with a NSFU of 2+. With regard to executive function, attention, language, and orientation, women had an inferior position at first compared to men, while they were better with a NSFU of 2+, with each OR = 1.71, 1.42, 1.83, and 1.31, respectively (although not all of these were significant).

This study has a number of strengths. One of the main strengths is that it focused on the association of using smartphones with cognitive function, which has rarely been studied before. Moreover, six domain-specific cognitive functions were simultaneously considered by gender, and furthermore, a joint association between gender and smartphone use on multi-domain cognition was detected, which offers a new and more comprehensive perspective for specific individual interventions. In addition to, the conducted survey was based on a large randomly selected sample, which allows for greater generalization of the obtained results. Nonetheless, the current study has several limitations. First, this study is a cross-sectional study, and no causal effect could be concluded between smartphone use and multi-domain cognition. However, to decrease the reverse causality to a certain extent, participants with dementia or with severe cognitive impairment were excluded since demented subjects will have greater limitations with respect to smartphone use. The remaining participants, including cognitively normal participants and those with mild cognitive impairment, should have normal ability with regard to daily activities [50]. Second, the associations of the domain-specific cognition with the number of smartphone function use were explored; however, associations with specific smartphone functions are still missing. However, we will focus on this next and will strive to identify the most beneficial smartphone functions corresponding to each domain-specific cognition. Third, no detailed information on the habit of smartphone use was collected, such as frequency and length of usage time, which obstructs a deeper understanding of its relationship with cognition. Fourth, domain-specific cognitive functions were assessed by the MoCA only, which might impede a stable result. Moreover, although only the MoCA was used as a cognitive screener at first, a host of studies demonstrated its stability for cognitive assessment [51, 52]. Nevertheless, more fine-grained neuropsychological assessment methods should be used to gain a more stable result and to achieve better comparability to similar studies.