In this first international comparison of cognitive function in nationally representative samples of older adults in the United States and England, US adults performed better than their English counterparts. Using the same cognitive tests administered in the same year, US adults showed significantly better performance on standard tests of memory. The US cognitive advantage was greatest for the "oldest-old," those aged 85 and older. On a population level, the overall difference in cognitive performance between the two countries was quite large, approaching the magnitude associated with about 10 years of aging.
The better cognitive performance of US adults in our study was surprising since, as was found for younger adults aged 55–64 in a recent report, [2
] we found that older US adults had a significantly higher prevalence of cardiovascular risk factors (hypertension and diabetes), heart disease, and stroke. Since increasing evidence suggests that cardiovascular risk factors and cardiovascular disease are associated with cognitive decline and poorer cognitive function, [4
] the higher burden of these conditions among US adults would seem to predict poorer cognitive function compared to English adults with a lower burden of these conditions.
What might account for the apparently better "brain health" among older US adults compared to English adults in the face of a significantly higher burden of cardiovascular risk factors and disease? Our study suggests a number of possibilities. First, US adults had significantly higher levels of education and wealth than English adults in 2002, and these factors accounted for some of the better US cognitive performance in our multivariable analyses. More years of formal education is associated with a reduced risk of cognitive decline and dementia, [18
] likely through multiple causal pathways, including a direct effect on brain development and function, [20
] better health behaviors, [21
] more cognitively stimulating occupations, [19
] and a safer and more enriched social environment.[24
] Greater wealth is also associated with better cognitive function, likely sharing many of the same causal pathways with education.[6
Another interesting and important difference between US and English adults that may have accounted for some of the US cognitive advantage was the significantly lower level of depressive symptoms reported by US adults. Depression is associated with worse cognitive function, although this relationship is likely complex and bidirectional, with depression possibly being a risk factor for cognitive decline, and early cognitive decline possibly leading to depressive symptoms.[26
] We found a significant dose-response relationship between the number of reported depressive symptoms and cognitive performance, with those reporting 4 to 8 depressive symptoms scoring nearly 1 point lower on the cognitive scale in our fully adjusted model (Model 8 in Table ). In our multivariable models, the greater prevalence of depressive symptoms in England explained a portion of the poorer cognitive performance among older English adults. Making valid comparisons of the prevalence of diagnosed clinical depression in different countries is difficult due to differences in study populations and diagnostic criteria. A study of depression prevalence among adults in European countries found significant international variation, from about 9% in Iceland to 24% in Germany.[27
] In England, depression prevalence was 10% in a sample from Liverpool and 17% in London. In the United States a large community-based study in the early 1990s found a 10% prevalence of depression among adults.[28
] Interestingly, and perhaps important to the interpretation of our findings, fewer than 15% of depressed adults in these English samples were receiving medication to treat their depression, while a study of US adults during the same time period (mid-1990s), found that nearly 75% of depressed individuals were receiving medication therapy.[29
] Future research should explore whether more widespread use of anti-depressant medication in the United States may be one reason for the lower level of depressive symptoms, and in turn, the better cognitive performance of older adults in the US compared to England that we found in our study.
One significant difference in health behaviours between the US and England – the consumption of alcohol – likely favoured the cognitive performance of older adults in England. We have previously reported that moderate alcohol consumption, compared to abstinence, was associated with better cognition among those aged 50 and older in the 2002 wave of the ELSA.[30
] We found a similar positive relationship between alcohol consumption and cognition in those aged 65 and older, with those reporting some alcohol intake showing significantly better cognitive function than those who reported abstaining from drinking alcohol. More than 50% of US adults reported no alcohol intake compared to only 15.5% of English adults.
Finally, while US adults reported a higher prevalence of hypertension, they also were more likely to be taking medications to treat hypertension (91% vs. 85% of those with hypertension, p < .001). This finding is in line with a prior cross-national study of hypertension treatment in the 1990's showing a greater likelihood of any hypertension treatment among hypertensive US adults (age 35 to 64) compared to English adults, as well as more aggressive blood pressure lowering among those being treated (84% of treated US adults had a blood pressure of < 160/95 compared to 73% of English adults).[31
] A number of observational studies have shown a relationship between hypertension and an increased risk for cognitive impairment, [3
] as well as a protective effect of antihypertensive therapy for preventing cognitive decline.[33
] Similarly, in our study self-reported use of antihypertensive medications was associated with significantly better cognitive function, controlling for all other covariates. However, while some randomized controlled trials of antihypertensive therapy have shown a benefit for cognitive function, [35
] other RCTs have not shown a clear benefit.[36
To test the hypothesis that more aggressive and effective hypertension treatment in the US vs. England (among those receiving treatment) might be contributing to the US cognitive advantage, we performed an additional regression analysis limited to those with hypertension. After controlling for all of the variables in our analysis (Model 8), an interaction term for hypertension treatment × country (England as reference) was positive (coefficient = 0.64 points) and statistically significant (P = .014). In this sub-analysis, hypertension treatment in the US was associated with a 0.5 point higher cognitive score (among those with hypertension, after controlling for all other covariates), while hypertension treatment in the UK was associated with a 0.1 point lower cognitive score. Future research with more detailed data on hypertension treatment (e.g., number of medications, type of medications, and dose of medications) and measured blood pressure is required to better assess whether more aggressive hypertension treatment in the US is, in fact, helping to protect cognitive function more effectively than in England.
The fact that the greatest cognitive advantage for US adults in our study was among the oldest-old may also support the hypothesis that more aggressive diagnosis and treatment of hypertension, and possibly other cardiovascular risks, in the US in middle-age and older adults leads to less significant cognitive decline among the oldest-old. Given the significant public health and cost implications of cognitive decline and the incidence of dementia in aging populations around the world, future research to identify whether more aggressive treatment of cardiovascular risks such as hypertension, hypercholesterolemia, and obesity leads to improved brain health among older adults could pay significant public health dividends.
The strengths of our study include the large nationally representative samples of adults in the US and England, and the direct assessment of cognition using the same cognitive tests administered in the same year. There are also a number of potential limitations of our study that are important to consider when interpreting our results. First, while both the HRS and ELSA are nationally representative samples, differences between the two studies in overall response rates, and in methods for the recruitment of proxy respondents to answer for sample members, could have important implications for the comparison of cognitive function of those included in our analysis. The overall response rate among all eligible respondents was 87% for the HRS in 2002 and 67% for ELSA. The HRS also included more proxy respondents compared to ELSA in 2002. Among white respondents aged 65 and older in 2002, 1,171 (12%) were represented by a proxy in the HRS, compared to only 96 (2%) in ELSA. If the difference in proxy representation between the two studies is due to the HRS being more likely to use a proxy for a respondent with impaired cognitive function (whereas the ELSA would still use a self-report interview), this could lead to the pattern of better apparent cognitive performance among the HRS self-respondents compared to the ELSA self-respondents included in our study. To assess this possibility, we compared the characteristics of respondents represented by a proxy in the HRS and ELSA in 2002, and did not find evidence that HRS respondents represented by a proxy were "sicker" or more likely to have impaired cognition than those in the ELSA. For instance, compared to ELSA proxies, HRS proxies represented individuals who were younger (81% were aged < 85 in HRS compared to 61% in ELSA, P < .01) and had higher net worth (31% in the top tertile in the HRS, compared to 24% in ELSA, P < .05). In addition, proxies in the HRS rated the overall cognitive function of those whom they represented as somewhat better than ELSA proxies using the same informant scale (the Informant Questionnaire for Cognitive Decline in the Elderly.[37
] Taken together, these comparisons of respondents represented by a proxy in the two studies suggest that the poorer cognitive performance of English adults in our study is not an artifact of differences in the utilization of proxy respondents.
While the same cognitive tests were administered in both the HRS and ELSA, one important difference in the administration of the test should be considered when interpreting the results. About 70% of the HRS sample was interviewed by telephone and 30% in-person, while all ELSA interviews were in-person. If telephone administration of the cognitive tests is associated with systematically better performance compared to in-person administration, this could explain some of the HRS cognitive advantage that we found. However, two prior studies have examined the impact of telephone vs. face-to-face administration of the HRS cognitive tests, and found no significant differences in test scores for the different survey modes.[38
] Finally, it should be noted that our OLS regression analysis may not have accurately identified non-linear relationships between the predictor variables and the cognitive score outcome.