Abstract

Introduction

Cognitive impairment is an intermediate stage between normal cognitive aging and dementia. Although diagnosing subclinical cognitive impairment is essential for identifying individuals at high risk of dementia, this task is complex and challenging due to a lack of standardized diagnostic criteria.¹1. Richardson C, Stephan BC, Robinson L, Brayne C, Matthews FE; Cognitive Function and Ageing Study Collaboration. Two-decade change in prevalence of cognitive impairment in the UK. Eur J Epidemiol. 2019;34:1085-92. Cognitive impairment varies from strict clinical criteria for mild cognitive impairment (MCI)²2. Winblad B, Palmer K, Kivipelto M, Jelic V, Fratiglioni L, Washlund LO, et al. Mild Cognitive Impairment – beyond controversies, towards a consensus: report of the International Working Group on Mild Cognitive Impairment. J Intern Med. 2004;256:240-46.

3. Petersen RC, Roberts RO, Knopman DS, Boeve BF, Gefa YE Ivnik Rj, et al. Mild cognitive impairment: ten years later. Arch Neurol. 2009;66:1447-55.

4. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5). Arlington: American Psychiatric Publishing; 2013.-⁵5. Albert MS, DeKosky ST, Dickson D, Dubois B, Feldman HH, Fox NC, et al. The diagnosis of mild cognitive impairment due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement. 2011;7:270-9. to definitions of cognitive decline used in population-based studies, which cover a broader range of dysfunctions.⁶6. Bassuk SS, Wypij D, Berkman LF. Cognitive impairment and mortality in the community-dwelling elderly. Am J Epidemiol. 2000;151:676-88.

7. Sachs GA, Carter R, Holtz LR, Smith F, Stump TE, Tu W, et al. Cognitive impairment: an independent predictor of excess mortality: a cohort study. Ann Intern Med. 2011;155:300-8.-⁸8. Park MH, Kwon DY, Jung JM, Han C, Jo I, Jo SA. Mini-mental status examination as predictors of mortality in the elderly. Acta Psychiatr Scand. 2013;127:298-304.

Six previous longitudinal studies from high-income countries have shown that individuals with cognitive impairment have a higher mortality risk than healthy controls.⁷7. Sachs GA, Carter R, Holtz LR, Smith F, Stump TE, Tu W, et al. Cognitive impairment: an independent predictor of excess mortality: a cohort study. Ann Intern Med. 2011;155:300-8.

8. Park MH, Kwon DY, Jung JM, Han C, Jo I, Jo SA. Mini-mental status examination as predictors of mortality in the elderly. Acta Psychiatr Scand. 2013;127:298-304.

9. Santabárbara J, Gracia-García P, Pírez G, López-Antón R, De La Cámara C, Ventura T, et al. Mortality in mild cognitive impairment diagnosed with DSM-5 criteria and with Petersen's criteria: A 17-year follow-up in a community study. Am J Geriatr Psychiatry. 2016;24:977-86.

10. de Bruijn RF, Akoudad S, Cremers LG, Hofman A, Niessen WJ, van der Lugt A, et al. Determinants, MRI correlates, and prognosis of mild cognitive impairment: the Rotterdam study. J Alzheimers Dis. 2014;42 Suppl 3:S239-49.

11. Perna L, Wahl HW, Mons U, Saum KU, Holleczek B, Brenner H. Cognitive impairment, all-cause and cause-specific mortality among non-demented older adults. Age Ageing. 2015;44:445-51.-¹²12. Vassilaki M, Cha RH, Aakre JA, Therneau TM, Geda YE, Mielke MM, et al. Mortality in mild cognitive impairment varies by subtype, sex, and lifestyle factors: the Mayo Clinic Study of Aging. J Alzheimers Dis. 2015;45:1237-45. However, these studies measured cognitive impairment differently. In four of these studies, cognitive impairment was identified by one-time administration of a cognitive functioning scale,⁶6. Bassuk SS, Wypij D, Berkman LF. Cognitive impairment and mortality in the community-dwelling elderly. Am J Epidemiol. 2000;151:676-88.

7. Sachs GA, Carter R, Holtz LR, Smith F, Stump TE, Tu W, et al. Cognitive impairment: an independent predictor of excess mortality: a cohort study. Ann Intern Med. 2011;155:300-8.-⁸8. Park MH, Kwon DY, Jung JM, Han C, Jo I, Jo SA. Mini-mental status examination as predictors of mortality in the elderly. Acta Psychiatr Scand. 2013;127:298-304.,¹¹11. Perna L, Wahl HW, Mons U, Saum KU, Holleczek B, Brenner H. Cognitive impairment, all-cause and cause-specific mortality among non-demented older adults. Age Ageing. 2015;44:445-51. while in the other two it was diagnosed according to MCI criteria.⁹9. Santabárbara J, Gracia-García P, Pírez G, López-Antón R, De La Cámara C, Ventura T, et al. Mortality in mild cognitive impairment diagnosed with DSM-5 criteria and with Petersen's criteria: A 17-year follow-up in a community study. Am J Geriatr Psychiatry. 2016;24:977-86.,¹⁰10. de Bruijn RF, Akoudad S, Cremers LG, Hofman A, Niessen WJ, van der Lugt A, et al. Determinants, MRI correlates, and prognosis of mild cognitive impairment: the Rotterdam study. J Alzheimers Dis. 2014;42 Suppl 3:S239-49.,¹²12. Vassilaki M, Cha RH, Aakre JA, Therneau TM, Geda YE, Mielke MM, et al. Mortality in mild cognitive impairment varies by subtype, sex, and lifestyle factors: the Mayo Clinic Study of Aging. J Alzheimers Dis. 2015;45:1237-45. Data on the association between cognitive impairment and mortality is scarce in low- and-middle-income countries: a single Tanzanian study¹³13. Paddick SM, Kisoli A, Dotchin CL, Gray WK, Chaote P, Longdon A, et al. Mortality rates in community-dwelling Tanzanians with dementia and mild cognitive impairment: a 4-year follow-up study. Age Ageing. 2015;44:636-41. applied MCI criteria and found that, after adjusting for the effects of age, sex, and education, a higher mortality risk was associated with cognitive impairment.

The cognitive decline model used in population-based studies is different from and more prevalent than strict MCI criteria, including a higher rate of progression to dementia.¹⁴14. Stephan BC, Matthews FE, McKeith IG, Bond J, Brayne C; Medical Research Council Cognitive Function and Aging Study. Early cognitive change in the general population: how do different definitions work?J Am Geriatr Soc. 2007;55:1534-40.,¹⁵15. Hogan DB, Ebly EM. Primitive reflexes and dementia: results from the Canadian study of health and aging. Age Ageing. 1995;24:375-81. Consistently lower rates of progression from MCI to dementia are found in community services than in specialty and clinical research services.¹⁶16. Bruscoli M, Lovestone S. Is MCI really just early dementia? A systematic review of conversion studies. Int Psychogeriatr. 2004;16:129-40.,¹⁷17. Mitchell AJ, Shiri-Feshki M. Rate of progression of mild cognitive impairment to dementia -- meta-analysis of 41 robust inception cohort studies. Acta Psychiatr Scand. 2009;119:252-65. Furthermore, the findings of population-based studies indicate that many individuals with MCI remain stable or even revert to normal during follow-up.¹⁶16. Bruscoli M, Lovestone S. Is MCI really just early dementia? A systematic review of conversion studies. Int Psychogeriatr. 2004;16:129-40.

17. Mitchell AJ, Shiri-Feshki M. Rate of progression of mild cognitive impairment to dementia -- meta-analysis of 41 robust inception cohort studies. Acta Psychiatr Scand. 2009;119:252-65.-¹⁸18. Matthews FE, Stephan BC, McKeith IG, Bond J, Brayne C; Medical Research Council Cognitive Function and Ageing Study. Two-year progression from mild cognitive impairment to dementia: to what extent do different definitions agree? J Am Geriatr Soc. 2008;56:1424-33. Thus, the objective of the current study was to compare mortality in individuals with and without cognitive impairment according to population-based criteria over 15 years of follow-up in Brazil, a middle-income country.

Methods

Study area and population

This analysis was carried out using data from a 1997 population-based prospective cohort study on aging in Bambui, a city of approximately 15,000 inhabitants in Minas Gerais, Brazil. The city’s United Nations’ Human Development Index was 0.70, implying medium development, and its life expectancy was 70.2 years in 1991.¹⁹19. Lima-Costa MF, Firmo JO, Uchoa E. Cohort profile: the Bambuí (Brazil) cohort study of ageing. Int J Epidemiol, 2011;40:862-7. The city has one general hospital and approximately one physician per 1,000 inhabitants. Its main economic activities are agriculture (particularly farming) and commerce.¹⁹19. Lima-Costa MF, Firmo JO, Uchoa E. Cohort profile: the Bambuí (Brazil) cohort study of ageing. Int J Epidemiol, 2011;40:862-7.

The procedures of the Bambui Cohort Study of Aging have been described elsewhere.¹⁹19. Lima-Costa MF, Firmo JO, Uchoa E. Cohort profile: the Bambuí (Brazil) cohort study of ageing. Int J Epidemiol, 2011;40:862-7. Briefly, the baseline cohort population consisted of all residents aged 60 years and over on January 1, 1997, who were identified through a private census of the city. Of a total of 1,742 older residents identified in this census, 92.2% (1,606) were interviewed and 85.9% (1,496) underwent clinical examination during the baseline survey, which consisted of hematological assays, biochemical assays, and serological tests for Chagas disease, electrocardiogram, and anthropometric and blood pressure measurement.

The interviews lasted approximately 90 min and were carried out in the participants’ homes. Each interview was completed in one or two visits (within 1 week), depending upon the respondent’s need for rest. Neither the interviewers nor the respondents were aware of the clinical results at the time of the interview. A sample of 10% of participants was selected for reliability assessment. Cohort members underwent annual follow-ups, consisting of standardized interviews and death certificate verification.

Mortality data source

Deaths occurring up to December 31, 2011 were included in this analysis. Deaths were reported by next of kin during the annual follow-up interview and were verified through the Brazilian Ministry of Health’s mortality information system. Similar procedures were used to verify the death of individuals who moved away from the study area (5.2%). The endpoint in this analysis was death from any cause.

Cognitive assessment

The Bambui Cohort Study of Aging questionnaire included a standard Brazilian version of the Mini-Mental State Examination (MMSE).²⁰20. Seabra MLV, Concilio GV, Villares JB, Carlini EA. Avaliação do teste “Mini-mental state” em volunta?rios e pacientes brasileiros. Rev ABP-APL. 1990;12:1-7. In the Brazilian validation study, some questions were modified according to their relevance to the target population. Questions about the season, the building, and the floor in the original orientation section²¹21. Folstein MF, Folstein SE, McHugh PR. Mini-mental state. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12:189-98. were replaced with the period of the day, the room, and the address in the Brazilian version.²⁰20. Seabra MLV, Concilio GV, Villares JB, Carlini EA. Avaliação do teste “Mini-mental state” em volunta?rios e pacientes brasileiros. Rev ABP-APL. 1990;12:1-7. In the registration and recall section, the words used were gato (cat), árvore (tree), and violão (guitar). In the attention and calculation section, serial fives replaced serial sevens, and for the reverse spelling task, “World” was replaced by “Maria,” due to is common usage in Brazilian culture. Previous analysis showed that the MMSE version used in this study had cross-cultural equivalence and invariance or stability over time in its factor structure.²²22. Castro-Costa E, Fuzikawa C, Ferri C, Uchoa E, Firmo J, Lima-Costa MF, et al. Dimensions underlying the mini-mental state examination in a sample with low-education levels: the Bambui health and aging study. Am J Geriatr Psychiatry. 2009;17:863-72.,²³23. Castro-Costa E, Dewey ME, Uchôa E, Firmo JO, Lima-Costa M F, Stewart R. Construct validity of mini-mental state examination across time in a sample with low-education levels: 10-year follow-up of the Bambuí cohort study of ageing. Int J Geriatr Psychiatry. 2014;29:1294-303. According to total MMSE scores, cognitive impairment was classified as none (1.0 standard deviation [SD] above cutoff means), mild (1.0 SD below cutoff means) or moderate (2.0 SD below cutoff means). These groupings were applied to allow comparison with previous research.⁸8. Park MH, Kwon DY, Jung JM, Han C, Jo I, Jo SA. Mini-mental status examination as predictors of mortality in the elderly. Acta Psychiatr Scand. 2013;127:298-304.

Sociodemographic characteristics

Data were collected on the following variables: age, sex, number of complete years of education (< 4 years vs. ≥ 4 years: 4 years was the statutory primary education for this population), marital status (married or living together vs. single, separated, divorced, or widowed), and self-reported ethnicity (Caucasian vs. non-Caucasian, e.g. African or mixed ethnicity).

Lifestyle variables

Alcohol consumption was determined through the question: “During the previous 12 months, on average, how often have you drunk alcohol?”; caffeine consumption was determined through the question: “How many cups of coffee, tea, or other caffeinated drinks have you had daily in the last 12 months?” Current smokers were defined as participants who had smoked at least 100 cigarettes in their lifetime and were still smokers. Physical activity level was determined through the question: “During your leisure time in the previous 90 days, how many times did you go for a walk, work out, or participate in sports for at least 20-30 min?”

Functional status and health conditions

Functional impairment was defined as the inability to independently perform at least one of the following five activities of daily living: dressing, walking across a room, bathing or showering, eating, getting in or out of bed, and using the toilet. Body mass index (BMI) was assessed in a visit to the Bambui field clinic using the standard protocol. Diabetes mellitus was defined as a fasting blood glucose level ≥ 126 mg/dL and/or current use of insulin or oral antidiabetic drug treatment.²⁴24. American Diabetes Association. Classification and diagnosis of diabetes: Standards of Medical Care in Diabetes. Diabetes Care. 2018;41(Suppl. 1):S13-27. Infection with T. cruzi was assessed by two parallel methods. Infection was defined as seropositivity in both indirect hemagglutination assay (Biolab-Mérieux, Rio de Janeiro, Brazil) and enzyme-linked immunosorbent assay (Abbott, São Paulo, Brazil). Information on lifetime history of stroke was obtained by interview according to methods described elsewhere.²⁵25. Plan and operation of the Third National Health and Nutrition Examination Survey, 1988-94. Series 1: programs and collection procedures. Vital Health Stat 1. 1994;(32):1-407. Hypertension was defined as systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥ 90 mmHg and/or the use of antihypertensive drugs.²⁶26. Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA, Izzo JL Jr, et al. The seventh report of the joint national committee on prevention, detection, evaluation, and treatment of high blood pressure: the JNC 7 report. JAMA. 2003;289:2560-72. Genomic DNA for APOE genotyping was extracted from blood samples using a Wizard^® Genomic DNA Purification System (Promega, Madison, USA). In the present analysis, we considered three groups of gene composition for the ɛ4 allele: absent, homozygous (ɛ4ɛ4), and heterozygous (42.8% is ɛ2ɛ4 and 57.2% is ɛ3ɛ4).

Depressive symptoms were assessed by the General Health Questionnaire-12, using a cutoff of 4/5 to define case-level symptomatology.²⁷27. Costa E, Barreto SM, Uchoa E, Firmo JO, Lima-Costa MF, Prince M. Is the GDS-30 better than the GHQ-12 for screening depression in the elderly people in the community? The Bambui health aging study (BHAS). Int Psychogeriatr. 2006;18:493-503.

Statistical analysis

The follow-up period was from the date of the first examination until the date of death or the study endpoint (December 31, 2011). Mortality rates were estimated using the person-time method the (number of deaths divided by person-years of follow-up). Hazard ratios (HR) for death in each cognitive impairment category were estimated by Cox’s proportional hazards regression modeling. Gender, years of education, being unmarried (including separated, divorced, or widowed status), self-reported ethnicity, alcohol consumption, smoking, exercise, physical function, diabetes mellitus, infection with T. cruzi, stroke, hypertension, APOE ɛ4 allele, and depressive symptoms were used as categorical measures, while age, caffeine consumption, and BMI were used as continuous measures.

The analysis was based on four models, which were performed for the total sample and then stratified according to gender. First, we estimated the crude association between cognitive impairment and mortality, and then adjusted incrementally for 1) demographic variables, 2) lifestyle, and 3) health measures. A formal test of the proportional-hazards assumption based on Schoenfeld residuals was also performed. All analyses were performed in Stata version 14.1.

Ethics

The project was approved by the ethics committee of the Fundação Oswaldo Cruz, Brazil.

Results

The mean follow-up duration was 11.0 years (median: 13.3), corresponding to 14,120.4 person-years at risk. A total of 639 subjects died during follow-up (crude mortality rates for mild and moderate cognitive impairment were 69.1 per 1,000 person-years and 93.9 per 1,000 person-years, respectively). There were no significant differences between participants included and excluded from the analysis regarding gender (degrees of freedom [df] = 1; χ² = 0.09; p = 0.752) and education (df = 1; χ² = 0.89; p = 0.343), although the included participants were younger (df = 1; χ² = 4.33; p = 0.03). Table 1 summarizes the baseline sociodemographic and other characteristics of the 1,281 participants with complete baseline information for all variables, divided according to cognitive impairment level.

Association between cognitive impairment and all-cause mortality

Table 2 shows the HR for all-cause mortality according to cognitive impairment for all participants and stratified according to gender. In the unadjusted HR model, the mortality risk was increased in participants with mild (HR: 1.71; 95% confidence interval [95%CI] 1.36-2.15) and moderate cognitive impairment (HR: 2.50; 95%CI: 1.54-4.05). However, the associations did not remain significant in the final model (HR MCI: 1.15; 95%CI 0.89-1.48 and HR moderate cognitive impairment: 1.35; 95%CI 0.81-2.24).

After gender stratification, only MCI increased the mortality risk in the unadjusted analysis (HR: 1.92; 95%CI 1.40-2.62), although it did not remain significant in the final model (HR: 1.35; 95%CI 0.96-1.88) for female participants. In male participants, mortality risk was increased by both MCI (HR; 1.42; 95%CI 1.01-2.00) and moderate cognitive impairment (HR: 3.30; 95%CI 1.69-6.45) in the unadjusted analysis, although only moderate cognitive impairment remained significant in the final model (HR: 2.30; 95%CI 1.12-2.4.74).

Figure 1 shows the estimated survivor function for the six curves (each cognitive functioning group for each sex). Compared to women with no cognitive impairment (the reference group), men with moderate cognitive impairment had the highest mortality risk (HR: 3.37; 95%CI 1.71-6.62) followed by men with MCI (HR: 1.55; 95%CI 1.05-2.25).

Figure 1
Cox’s proportional hazards regression according to cognitive impairment group. 95%CI = 95% confidence interval; HR = hazard ratios.

Discussion

Our results indicated that mild and moderate cognitive impairment, as determined by a single screening with the MMSE, were not associated with mortality risk in the entire sample after controlling for covariables. However, we found that gender did moderate mortality risk in this population, since men with moderate cognitive impairment had a significantly higher independent risk of mortality.

To our knowledge, only two other investigations, both in developed countries, have examined the association of cognitive decline and excess mortality using cognitive assessment similar to that of the present study. In the first, Park et al.⁸8. Park MH, Kwon DY, Jung JM, Han C, Jo I, Jo SA. Mini-mental status examination as predictors of mortality in the elderly. Acta Psychiatr Scand. 2013;127:298-304. investigated whether cognitive impairment is independently associated with increased long-term mortality in community-dwelling Koreans aged 60 years or older. At baseline, their participants were screened for cognitive impairment with the MMSE and were categorized into groups with no, mild, or moderate cognitive impairment. Their regression analysis included several variables, representing either confounders or mediators in the relationship between cognitive impairment (including different domain measurements) and mortality. Their results indicated that both mild and moderate cognitive impairment are associated with an increased risk of mortality. Although their sample was larger than ours (2,712 vs. 1,276), the adjusted analysis was not rigorous enough to rule out confounders, since it did not include age, gender, education, exercise, current smoking, depression, osteoporosis, hyperlipidemia, and cerebrovascular diseases as factors. However, we used two approaches to investigate the extent to which a relationship between cognitive impairment and mortality was independent of potential confounding variables. First, we included numerous additional confounding factors, including APOE ɛ4 allele, impaired physical functioning, BMI, alcohol and caffeine consumption, and self-reported ethnicity. Second, we carried out a proportional hazards test, which was consistent with a constant HR over the follow-up period.

The second study, conducted by Bassuk et al. in New Haven, CT, USA,⁶6. Bassuk SS, Wypij D, Berkman LF. Cognitive impairment and mortality in the community-dwelling elderly. Am J Epidemiol. 2000;151:676-88. investigated the effects of cognitive impairment on excess mortality among community-dwelling adults aged 68 years or older, using MMSE scores to assess cognition. Their results indicated a strong relationship between both mild and severe cognitive impairment and mortality in participants under 80 years of age over a 3-year follow-up period. However, this study differed from ours in that MMSE scores were not assessed according to normative cognitive modeling,²⁸28. Steinerman JR, Hall CB, Sliwinski MJ, Lipton RB. Modeling cognitive trajectories within longitudinal studies: a focus on older adults. J Am Geriatr Soc. 2010;58 Suppl 2:S313-8. which reduces ceiling and floor effects and would be desirable in a study of the increasing burden of multiple forms of pathological damage to the brain.²⁹29. Jack CR Jr, Knopman DS, Jagust WJ, Shaw LM, Aisen PS, Weiner MW, et al. Hypothetical model of dynamic biomarkers of the Alzheimer's pathological cascade. Lancet Neurol. 2010;9:119-28. Thus, their results are not strictly comparable with ours.

Other investigations have assessed the association between cognition and mortality in different ways. A study by Paddick et al. in Tanzania,¹³13. Paddick SM, Kisoli A, Dotchin CL, Gray WK, Chaote P, Longdon A, et al. Mortality rates in community-dwelling Tanzanians with dementia and mild cognitive impairment: a 4-year follow-up study. Age Ageing. 2015;44:636-41. a low-income country, demonstrated that cognitive decline (i.e., MCI) was associated with higher mortality rates than no cognitive impairment. In Indianapolis, IN, USA, Sachs et al.⁷7. Sachs GA, Carter R, Holtz LR, Smith F, Stump TE, Tu W, et al. Cognitive impairment: an independent predictor of excess mortality: a cohort study. Ann Intern Med. 2011;155:300-8. investigated whether cognitive impairment is independently associated with increased long-term mortality in primary care patients aged 60 years or older. At baseline, they screened patients for cognitive impairment with the Short Portable Mental Status Questionnaire, categorizing them into groups with no, mild, or moderate to severe cognitive impairment. They found that both mild and moderate to severe cognitive impairment are associated with an increased risk of mortality.

Our results showed an association between cognitive decline and excess mortality, particularly among men. Comparison with other studies could be compromised by methodological differences, such as study design, assessment instruments, sociocultural characteristics, healthcare access, treatment options, etc. However, such discrepancies were found even in the most directly comparable studies, in which cognitive impairment was measured and classified with the MMSE.⁶6. Bassuk SS, Wypij D, Berkman LF. Cognitive impairment and mortality in the community-dwelling elderly. Am J Epidemiol. 2000;151:676-88.,⁸8. Park MH, Kwon DY, Jung JM, Han C, Jo I, Jo SA. Mini-mental status examination as predictors of mortality in the elderly. Acta Psychiatr Scand. 2013;127:298-304. Despite the methodological differences, however, our findings on gender differences agree with one German and two American studies in that men with cognitive impairment had shorter survival times.¹¹11. Perna L, Wahl HW, Mons U, Saum KU, Holleczek B, Brenner H. Cognitive impairment, all-cause and cause-specific mortality among non-demented older adults. Age Ageing. 2015;44:445-51.,³⁰30. Moritz DJ, Fox PJ, Luscombe FA, Kraemer HC. Neurological and psychiatric predictors of mortality in patients with Alzheimer disease in California. Arch Neurol. 1997;54:878-85.,³¹31. Perls TT, Morris JN, Ooi WL, Lipsitz LA. The relationship between age, gender and cognitive performance in the very old: the effect of selective survival. J Am Geriatr Soc. 1993;41:1193-201.

The mechanism for gender differences in the association between cognitive impairment and mortality risk has seldom been investigated. Some researchers have proposed that dementia severity is a less significant predictor of survival in women than in men.³⁰30. Moritz DJ, Fox PJ, Luscombe FA, Kraemer HC. Neurological and psychiatric predictors of mortality in patients with Alzheimer disease in California. Arch Neurol. 1997;54:878-85.

Alternatively, the same group proposed that the MMSE may be a more valid indicator of dementia severity in men than women, perhaps because social or cultural factors influence MMSE scores differently by sex.³⁰30. Moritz DJ, Fox PJ, Luscombe FA, Kraemer HC. Neurological and psychiatric predictors of mortality in patients with Alzheimer disease in California. Arch Neurol. 1997;54:878-85. Furthermore, other mechanisms for dementia in men and women have been suggested, such as different profiles of comorbid illness and different causes of cognitive impairment.³¹31. Perls TT, Morris JN, Ooi WL, Lipsitz LA. The relationship between age, gender and cognitive performance in the very old: the effect of selective survival. J Am Geriatr Soc. 1993;41:1193-201.

The association between high mortality risk and moderate cognitive impairment found in men could also be related to lifestyle differences with women. Univariate analysis revealed robust gender differences in alcohol consumption (40.1in men vs. 5.5% in women) and smoking (29.3 in men vs. 10.6% in women). Although several large epidemiological studies have reported a lower risk of dementia in light to moderate drinkers than abstainers,³²32. Topiwala A, Ebmeier KP. Effects of drinking on late-life brain and cognition. Evid Based Ment Health. 2018;21:12-5.

33. Ruitenberg A, van Swieten JC, Witteman JC, Mehta KM, van Duijn CM, Hofman A, et al. Alcohol consumption and risk of dementia: the Rotterdam Study. Lancet. 2002;359:281-6.

34. Mukamal KJ, Kuller LH, Fitzpatrick AL, Longstreth WT Jr, Mittleman MA, Siscovick DS. Prospective study of alcohol consumption and risk of dementia in older adults. JAMA. 2003;289:1405-13.-³⁵35. Heffernan M, Mather KA, Xu J, Assareh AA, Kochan NA, Reppermund S, et al. Alcohol consumption and incident dementia: evidence from the Sydney memory and ageing study. J Alzheimers Dis. 2016;52:529-38. a large longitudinal study in Norway recently reported a higher rate of dementia-related deaths in moderate drinkers than abstainers.³⁶36. Ormstad H, Rosness TA, Bergem AL, Bjertness E, Strand BH; GENIDEM-Group. Alcohol consumption in the elderly and risk dementia-related death--a Norwegian prospective study with a 17-year follow-up. Int J Neurosci. 2016;126:135-44. This inconsistency has been ascribed to residual confounding, and Mendelian randomization has been recommended.³⁷37. Au Yeung Sl, Jiang C, Cheng KK, Cowling BJ, Liu B, Zhang W, et al. Moderate alcohol use and cardiovascular disease from Mendelian randomization. PLoS One. 2013;8:e68054. The two studies that applied this technique found that moderate drinking has neither a protective effect on cognitive decline in older men (defined as MMSE < 23/30; 6-year follow-up)³⁸38. Almeida OP, Hankey GJ, Yeap BB, Golledge J, Flicker L. Alcohol consumption and cognitive impairment in older men: a mendelian randomization study. Neurology. 2014;82:1038-44. nor a protective association with MMSE performance or word recall (cross-sectional study).³⁷37. Au Yeung Sl, Jiang C, Cheng KK, Cowling BJ, Liu B, Zhang W, et al. Moderate alcohol use and cardiovascular disease from Mendelian randomization. PLoS One. 2013;8:e68054. Regarding smoking a risk factor for dementia and cognitive decline, a meta-analysis of 19 prospective studies³⁹39. Anstey KJ, von Sanden C, Salim A, O'Kearney R. Smoking as a risk factor for dementia and cognitive decline: a meta-analysis of prospective studies. Am J Epidemiol. 2007;166:367-78. found that current older smokers at baseline had an increased risk of dementia and cognitive decline than older adults who never or formerly smoked.

Our study is the first to address the association between cognitive impairment and mortality in South America, adding to the limited evidence from low- and middle-income countries. Its strengths include: 1) cognitive modeling based on MMSE scores, which reduces the influence of short-term changes in mood, pain, sleep, stress or physiological measures (such as serum glucose, blood pressure, and oxygen saturation); 2) the use of an MMSE version that has been validated for the Brazilian population; 3) mortality was determined over 15 years of follow-up in a large population-based cohort with a high response rate and minimal attrition; and 4) the analysis included adjustment for a range of potential confounders, including APOE ɛ4 allele, impaired physical functioning, BMI, etc. The fact that the proportional hazards were consistent throughout the follow-up period further reduces the likelihood of confounding.

However, our study has some limitations that could affect interpretation of the results. Generalizability may be restricted because participants were obtained from a single city in southeastern Brazil, and most participants had low educational and socioeconomic status. However, the demographic characteristics of our study participants are very similar to the overall Brazilian population. Since cognition was only assessed at baseline, we could not evaluate the trajectory of cognitive impairment and its effects on mortality risk over the follow-up period, which was a significant weakness. The low accuracy of cause-of-death coding in the mortality information system did not allow a deeper exploration of the data on this issue. There were only nine males in the moderate cognitive impairment group, which could have led to lower precision and power. Thus, the HR (CI) was not as high as expected. Finally, the HR (CI) for moderate cognitive impairment in men and women overlapped. Statisticians have traditionally considered CI overlap between two independent groups to demonstrate no statistically significant difference between them.⁴⁰40. Cumminng G, Finnch S. Inference by eye: confidence intervals and how to read pictures of data. Am Psychol. 2005;60:170-80.

41. Cumming G. Inference by eye: Reading the overlap of independent confidence intervals. Stat Med. 2009;28:205-20.-⁴²42. Nakagawa S, Cuthill IC. Effect size, confidence interval and statistical significance: a practical guide for biologists. Biol Rev Camb Philos Soc. 2007;82:591-60. However, this discussion has referred to normal populations and symmetrical CI for sample means. When comparing proportions bounded by categories (as we did), such CIs are generally asymmetrical, i.e., that the lower and upper arms are of unequal length.³³33. Ruitenberg A, van Swieten JC, Witteman JC, Mehta KM, van Duijn CM, Hofman A, et al. Alcohol consumption and risk of dementia: the Rotterdam Study. Lancet. 2002;359:281-6. In such cases, overlapping CI could indicate a statistically significant difference when it is about half the average of arm length or p ≤ 0.01, as we found in our analysis.⁴¹41. Cumming G. Inference by eye: Reading the overlap of independent confidence intervals. Stat Med. 2009;28:205-20.

In conclusion, our findings suggest an association between moderate cognitive impairment and all-cause mortality in men in a large Brazilian cohort of older adults. However, future studies are needed to both replicate these results, determine which mechanisms underly this phenomenon, and determine whether there are effective measures to reduce mortality in such cases.

Acknowledgements

The Bambui Cohort Study of Aging was carried out by a team of researchers based at the Fundação Oswaldo Cruz in the state of Minas Gerais, Brazil. Funding was provided by the Financiadora de Estudos e Projetos (FINEP), the Conselho Nacional de Desenvolvimento Científico e Tecnológicos (CNPq), and the Fundação de Amparo à Pesquisa do Estado de Minas Gerais (FAPEMIG). JOAF, MFL-C, BSD, and SLB received grants from CNPq (305274/2014-7). BSD currently receives research support from the National Institute of Mental Health (MH114970-01, MH115953-01, and R03MH115212-01).

References

Publication Dates

History