Impaired kidney function is associated with lower cognitive function in the elder general population. Results from the Good Aging in Skåne (GÅS) cohort study

Impaired kidney function and cognitive dysfunction are very common conditions in the elderly population [1‐3]. The prevalence of impaired kidney function – defined as glomerular filtration rate (GFR) < 60 ml/min/1,73m² [4, 5] – has been estimated to 23–36% in individuals ≥65 years of age [1], and increases immensely by age to about 55 to 75% in individuals > 80 years of age [6]. In Europe, the prevalence of mild cognitive impairment (MCI) has been estimated to > 20% in individuals ≥85 years of age, and the prevalence of dementia has been estimated to increase from 1% in the age group 65–70 years to 45% in individuals ≥95 years of age [3].

Since previous studies indicate a possible connection between impaired kidney function and cognitive dysfunction [7, 8], impaired kidney function is a known risk factor for vascular disease [9], and cerebrovascular disease is common when cognitive dysfunction is present [10], a potential connection on vascular basis between impaired kidney function and cognitive dysfunction has been suggested [11]. Previous studies investigating the connection between impaired kidney function and declined function in different cognitive domains have shown decline in multiple and various cognitive domains [12], although executive function stands out in some studies [11, 13]. Impairment in multiple cognitive domains have been linked to cerebrovascular disease, although executive function and the speed of processing tend to be more associated [14, 15]. This study aims to investigate if there is an association between GFR and cognitive function in the general elder population, and if so, which specific cognitive domains are associated.

This is a cross-sectional study of data from the general population based cohort study” Good Aging in Skåne” (GÅS), being performed at the Department of Geriatric Medicine, Skåne University Hospital, Sweden [16]. In total 2931 subjects 60 to 93 years of age living in five municipalities in urban and rural areas in the south of Sweden were included from February 2001 to July 2004. They were randomly selected from the national Swedish population register. Home visits were offered to subjects unable to visit the research center. The participants underwent a comprehensive health examination by a physician, registered nurse and psychological test administrator, including medical examination, medical history, physical examination, neuropsychological testing, anthropometrics, interview and self-reported questionnaires and biobanking. The participants’ mean age at baseline was 71.5 years (SD 10.3 years). The participation rate at baseline was 60%. Information on diseases was based on medical records, medical history and examination by a physician based on the ICD-10 and DSM-IV criteria for dementia.

Individuals with impaired kidney function (eGFR < 60 mL/min/1.73 m²) had lower cognitive test results in all the cognitive tests except confidence judgement, compared to individuals with normal kidney function (eGFR ≥60 mL/min/1.73 m²), see Table 2. After adjustments for age, sex, education and country of origin, lower cognitive results remained for MMSE, recognition, word fluency, digit cancellation, pattern comparison and TMT B-A, as presented in Table 3.

The participants were divided into four groups depending on the severity of kidney function (eGFR levels/min/1.73 m²; < 30 mL, 30- < 45 mL, 45- < 60 mL and ≥ 60 mL, with normal kidney function, eGFR ≥60 mL/min/1.73 m², as reference). Adjustments were made for age, sex, education and country of origin. eGFR < 30 mL/ min/1.73 m² was associated with worse results in MMSE and recognition. eGFR 30- < 45 mL/min/1.73 m² was associated with worse results in MMSE, free recall, recognition, word fluency, digit cancellation, TMT B-A and digit span backwards. eGFR 45- < 60 mL/min/1.73 m² was associated with worse results in recognition, word fluency and TMT B-A, see Table 3.

The association of cognitive test results and impaired kidney function (eGFR < 60 mL/min/1.73 m²) was tested with stratification made for age into four age groups, with adjustments made for sex, education and country of origin. In the youngest age group, 60–69 years, impaired kidney function was associated with lower test score results for MMSE, recognition, digit cancellation, pattern comparison, TMT B-A and mental rotations. In the age group 70–79 years, an association was found for MMSE, free recall, word fluency and pattern comparison. An association between impaired kidney function and digit span forward was found in the oldest age group ≥90 years. No other association was found in the two oldest age groups, see Additional file 2.

The association of cognitive test results and the severity of impaired kidney function with eGFR divided into four groups with eGFR ≥60 mL/min/1.73 m² as reference as above, was also tested with stratification made for age into four age groups. Adjustments were again made for sex, education and country of origin. For MMSE an association was found in the age group 60–69 years with eGFR 45- < 60 mL/min/1.73 m², and in the age group 70–79 years with eGFR 30- < 45 mL/min/1.73 m². For digit span forward an association was found in the age group 70–79 years with eGFR 45- < 60 mL/min/1.73 m² and in the age group ≥90 years with eGFR 45- < 60. For free recall an association was found in the age group 70–79 years with eGFR 30- < 45 mL/min/1.73 m². For recognition an association was found in the age group 60–69 years with eGFR 30- < 45 and 45- < 60 mL/min/1.73 m². For word fluency an association was found in the age group 60–69 with eGFR 30- < 45 mL/min/1.73 m². For digit cancellation an association was found in the age group 60–69 years with eGFR 30- < 45 and 45- < 60 mL/min/1.73 m². For pattern comparison an association was found in the age group 60–69 years with eGFR 30- < 45 and 45- < 60 mL/min/1.73 m², in the age group 70–79 years with eGFR 30- < 45 mL/min/1.73 m² and in the age group 80–89 years with eGFR 30- < 45 mL/min/1.73 m². For TMT B-A an association was found in the age group 60–69 years with eGFR 45- < 60 mL/min/1.73 m² and in the age group 70–79 years with eGFR 30- < 45 mL/min/1.73 m². For mental rotations an association was found in the age group 60–69 years with eGFR 45- < 60. No association was found for digit span backwards and confidence judgement, see Additional file 3.

Regarding the cognitive tests MMSE, word fluency, digit span backwards, mental rotations and confidence judgement, interactions were found between impaired/normal kidney function and various demographic variables, see Additional file 4. New analyses for these cognitive tests were performed, including the interactions found, as presented in Additional file 5. With consideration taken for interactions, an association between impaired kidney function and test result on the digit span backwards test, that had not been seen without consideration taken for interactions, was found.

This cross-sectional study showed an association between impaired kidney function as well as the severity of impaired kidney function and the cognitive domains learning and memory, language, complex attention, executive function, as well as global cognitive function, after adjustment for demographic covariates. This association was noted also for mildly impaired kidney function defined as eGFR 45–60 for learning and memory, language and executive function. The lack of association in the most severely impaired eGFR group, was most likely explained by low power, since only 70 individuals had eGFR < 30 mL/min/1.73 m², compared to 241 with eGFR 30- < 45 mL/min/1.73 m² and 483 with eGFR 45- < 60 mL/min/1.73 m².

An association between kidney function and visual perception was seen only in the youngest age group. No association was found between kidney function and meta-memory.

The finding of an association between kidney function and multiple cognitive domains seen in this study is in line with previous studies on the subject, supporting a relationship between kidney function and multiple and various cognitive domains, as described by Elias et al. in a large review analyzing 5 cross-sectional and 10 longitudinal studies [12].

Previous studies show that impairment in multiple cognitive domains are associated with cerebrovascular disease, although executive function and the speed of processing could be more heavily associated than other domains [13, 14]. Our findings on a connection between kidney function and multiple cognitive domains, including executive function and the speed of processing, do not contradict, but rather support, a possible connection on vascular basis between impaired kidney function and cognitive dysfunction. A plausible assumption is that cerebral small vessel disease (CSVD) could constitute a large proportion of this assumed vascular denominator, since CSVD is heavily linked to cognitive impairment and has been estimated to cause up to 45% of dementia and about 20% of all stroke worldwide [36]. In regards to the special connection between executive function or speed of processing and kidney function, we did find an association. However, it is problematic to conclude if these associations are larger than the associations for other cognitive domains due to differences in the psychometric properties of the tests.

Adjustments were made for relevant demographic covariates potentially influencing cognition. Interaction analyses between eGFR and each of the demographic covariates were made. The interactions found are presented in Additional file 5. These interactions are generally hard to interpret, since the effect of normal kidney function seems to be positive for some level(s) of the interacting variable, but negative for other(s). For example, the effect of normal kidney function for those born in Sweden is 0,5 points on the MMSE test, while for foreign-born participants the opposite effect is observed. The same pattern is seen for word fluency, digit span backwards and mental rotations, raising the question if the interactions seen in this study are mainly due to chance (multiple testing).

An interesting finding was the lack of association between reduced kidney function and meta-memory, as described above, suggesting that the awareness of reduced function in memory is not affected in the presence of reduced kidney function. This is new knowledge since no previous study has investigated the association between reduced kidney function and meta-memory.

Another interesting finding was a stronger association found between kidney function and cognitive function in the younger age groups. This can probably at least partly be explained due to lack of power, though only 92–131 participants performed the different cognitive tests in the oldest age group (≥ 90 years), compared to 1229–1255 participants in the youngest age group (60–69 years). Lack of power can however probably not explain the difference between the second youngest age group (70–79 years) where 445–464 participants performed the cognitive tests compared to 489–552 participants in the second oldest age group (80–89 years).

The main strength of this study is the large population sample including both urban and rural areas, thereby increasing generalization of data. Although home visits were offered in order to reduce selection bias of healthier individuals, under-representation of elder and more frail subjects cannot be excluded. This might introduce an underestimation of noted associations.

Another strength was the large number of cognitive tests used. The cognitive domains were assessed with established standardized tests and standardized administration to ensure consistency in test scoring. Moreover, an additional strength was that multiple tests were used to assess the same specific cognitive domain, which showed similar results. The total variance of test scores for five of the cognitive tests determined by the test administrator was between 1.4 to 3.5% [37]. This further supports that the cognitive tests were carried out in a standardized way.

Earlier studies investigating the relationship between kidney function and executive function using TMT to assess executive function, commonly have used TMT B [38, 39]. A strength in our study, was using the difference between TMT B and A (TMT B-A), to measure mental flexibility (executive function), but in the same time avoid measuring the speed of processing [26].

A further strength in this study was estimating kidney function by use of both cystatine C and creatinine derived GFR analyses from one and the same batch providing a more reliable estimate than separate analyses of one of the two [19].

A weakness in this study was that no test assessed the cognitive domain social cognition [20].

The cardiovascular risk factors hypertension, smoking and diabetes type 2 were not proportionally distributed between the group with impaired kidney function and the group with normal kidney function, see Table 1. Adjustments for cardiovascular risk factors were not done in the analyses, adding a possible weakness to the study.

Another weakness is that the data was collected in 2001–2004. We cannot exclude the plausible association between kidney function and cognitive function since then has decreased somewhat, due to advancements in cardiovascular primary and secondary prophylaxis.

Impaired kidney function as well as the severity of impaired kidney function is associated with impairment in learning and memory, language, complex attention, executive function and global cognitive function, but not meta-memory. The multiple cognitive domains affected in association with impaired kidney function might be related to a common denominator, such as cerebral small vessel disease. Further longitudinal studies and MRI studies are needed to explore possible predictors and causality for cognitive decline related to kidney function.