You are here

  • Home
  • Archive
  • Volume 69, Issue 6
  • Pulmonary function as a risk factor for dementia death: an individual participant meta-analysis of six UK general population cohort studies

Article Text

Article menu

Download PDFPDF

Research report
Pulmonary function as a risk factor for dementia death: an individual participant meta-analysis of six UK general population cohort studies
  1. Tom C Russ1,2,3,4,
  2. John M Starr1,2,3,
  3. Emmanuel Stamatakis5,6,7,
  4. Mika Kivimäki5,
  5. G David Batty1,3,5
  1. 1Alzheimer Scotland Dementia Research Centre, University of Edinburgh, Edinburgh, UK
  2. 2Scottish Dementia Clinical Research Network, NHS Scotland, UK
  3. 3Centre for Cognitive Ageing & Cognitive Epidemiology, University of Edinburgh, Edinburgh, UK
  4. 4Division of Psychiatry, University of Edinburgh, Edinburgh, UK
  5. 5Department of Epidemiology and Public Health, University College, London, UK
  6. 6Charles Perkins Centre, University of Sydney, Australia
  7. 7Exercise and Sport Sciences, Faculty of Health Sciences, University of Sydney, Australia
  1. Correspondence to Dr Tom C Russ, Division of Psychiatry, University of Edinburgh, Kennedy Tower, Royal Edinburgh Hospital, Morningside Terrace, Edinburgh EH10 5HF, UK; tc.russ{at}ed.ac.uk

Abstract

Background In addition to being associated with all-cause mortality and cardiovascular disease mortality, lung function has been linked with dementia. However, existing studies typically provide imprecise estimates due to small numbers of outcome events and are based on unrepresentative samples of the general population.

Methods Individual participant meta-analysis of six cohort studies from the Health Survey for England and the Scottish Health Survey (total N=54 671). Dementia-related mortality was identified by mention of dementia on any part of the death certificate (mean follow-up 11.7 years). Study-specific Cox proportional hazard models of the association between lung function and dementia-related death were pooled using random effect meta-analysis to produce overall results.

Results There was a dose–response association between poorer lung function and a higher risk of dementia-related death (age- and sex-adjusted HR compared to highest quartile of forced expiratory volume in 1 s (FEV1), 95% CI: second quartile 1.32, 0.99 to 1.76; third quartile 1.78, 1.30 to 2.43; fourth (lowest) quartile 2.74, 1.73 to 4.32). There was no significant heterogeneity in study-specific estimates (I2=0%). Controlling for height, socioeconomic status, smoking and general health attenuated but did not remove the association (second quartile 1.15, 0.82 to 1.62; third quartile 1.37, 0.96 to 1.94; fourth quartile 2.09, 1.17 to 3.71). Results for forced vital capacity and peak flow were similar.

Conclusions In these general population samples, the relation between three measures of lung function and dementia death followed a dose–response gradient. Being in the bottom quartile of lung function was associated with a doubling of the risk.

  • AGEING
  • MENTAL HEALTH
  • META ANALYSIS
  • EPIDEMIOLOGY

https://doi.org/10.1136/jech-2014-204959

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Dementia is a major, growing public health priority affecting more than 800 000 people in the UK and 44 million worldwide.1 Since there are still no disease-modifying treatments,2 the focus of research has moved to prevention. However, modification of cardiovascular risk factors in late life does not reduce dementia risk.3 This may relate to the slow development of the neuropathological processes of dementia which can begin years, or even decades, before the onset of symptoms4–6 and, as a result, many now advocate the application of the life course paradigm to dementia epidemiology.7 This approach suggests that exposures throughout life might have a detrimental or protective effect on disease risk.8 In the absence of studies with follow-up encompassing entire life-spans, identifying exposures which capture experiences from different stages of life are important in the investigation of life course effects on disease. For example, height—regarded as a marker of early life illness, adversity, nutrition or psychosocial stress9—has been shown to be associated with dementia.10 Pulmonary function is also affected by multiple factors throughout the life course, not merely in early life, notably smoking, and it has been suggested that decreased lung function is linked to both cardiovascular disease11 ,12 and cognitive decline,13–19 as well as to overall mortality.12 ,20 ,21 However, studies linking lung function and dementia are scarce22–27 and the largest include around 10 000 participants. Few of these studies targeted the general population. Therefore, we add to the limited evidence base the first meta-analysis of individual-participant data using six large general population cohort studies.

    Methods

    Data source

    The study sample comprised four Health Surveys for England (HSE28 1995–1997 and 2001) and two Scottish Health Surveys (SHS29 1998 and 2003). These are annual (HSE) or intermittent (SHS), representative, general population-based cross-sectional studies sampling community-dwelling individuals in the UK. The majority of study members (82.2%) consented to mortality surveillance by linkage to the UK National Health Service death register, thus converting the original cross-sectional studies into prospective cohort studies. Study participants gave full informed consent and ethical approval was granted for all aspects of these studies by the London Research Ethics Council or the Local Research Ethics Councils prior to each survey year data collection.

    Measurement of lung function and covariables

    Pulmonary function was measured using a Vitalograph spirometer. The spirometer was calibrated at the start of each day and the temperature in the survey participant's home entered prior to its use. Participants blew into the spirometer five times while standing up and the researcher recorded whether these were technically satisfactory blows. The highest value for forced vital capacity (FVC), forced expiratory volume in 1 s (FEV1) and peak flow (PF) was recorded.

    Informants were visited by an interviewer, who measured height and weight—allowing computation of body mass index (BMI) using the usual formula (weight (kg)/(height(m))2)—and subsequently by a nurse. Ethnicity was coded as white, black, Asian and ‘Chinese or other’. Age on leaving full-time education was recorded as <15, 15, 16, 17, 18, >18 years, ‘never went to school’ and ‘still in full-time education’. Information on occupational social class was collected during the interview and coded according to the Registrar General Classification (professional, intermediate, skilled non-manual, skilled manual, part-skilled and unskilled), a standard approach in the UK. Smoking status was classified as never a regular smoker, ex-smoker and smoker with daily consumption recorded. Alcohol consumption was recorded as frequency of drinking. During the interview, participants were asked to rate their health on a five-point scale (very good/good/fair/bad/very bad) and asked whether or not they suffered from a long-standing illness.

    Mortality data

    Study participants were linked to mortality data until the first quarter 2011 and all causes recorded on death certificates (up to 10) were obtained in addition to date of death. Dementia was identified from death certification according to International Classification of Diseases Ninth Edition (ICD-9) codes 290.0-290.4, 294.9, 331.0-331.2 and 331.9 and ICD-10 codes F00, F01, F03, F09, G30 and G31. Preliminary analyses suggested that using dementia as underlying cause of death gave the same results as using any mention of dementia but, due to fewer deaths, resulted in lower statistical power (see online supplementary table S1). Therefore, any mention of dementia in any part of the death certificate was used as the outcome of interest, a standard approach in dementia epidemiology since people with dementia often die of other conditions.30–33

    Statistical analysis

    After ascertaining that the proportional hazards assumption had not been violated, we used Cox proportional hazards models34 to compute study-specific effect estimates with accompanying SEs for the association between pulmonary function and dementia-related death. We then pooled these study-specific estimates in random effects meta-analyses. We report HR with accompanying 95% CIs for four categories of each marker of lung function in relation to dementia-related death. Lung function categories were defined by choosing pulmonary function test cut-offs which provided an approximately equal number of dementia deaths in each category. Since preliminary results suggested a linear relationship between lung function and dementia death, we also report HRs per SD disadvantage (decrease) in pulmonary function. Calendar time (months) was the underlying time scale; for participants who did not die from dementia, data were censored at the linkage date.

    There was no effect modification by gender, so data for men and women were pooled and (age- and) sex-adjusted. Models were then additionally adjusted for a series of covariates: height, ethnicity, occupational social class, educational attainment, smoking, alcohol consumption, BMI, self-rated general health and longstanding illness. We then combined all these variables in a multivariable model. We compared the effects of controlling for different confounding and mediating variables on the magnitude of the association by examining the change in size of HR rather than a change in significance level.

    We computed models examining the association between lung function and all-cause mortality for comparison with the dementia-related death models. We also carried out a number of sensitivity and subgroup analyses. We repeated the multivariable model restricting the sample to never-smokers. We also examined the effect on the multivariable model of excluding deaths occurring in the first 5 years of follow-up to investigate the possibility that individuals with undiagnosed dementia might have found it more difficult to comply with pulmonary function testing. Individuals with data missing for one or more variable and those with no missing data were compared using Student's t test for continuous variables and χ2 tests for categorical variables. We then produced five multiply imputed data sets in SPSS V.21 and repeated the multivariable model using these data. All other analyses were conducted using R V.2.15.2. The reporting of our analyses conforms to the STROBE statement.35

    Results

    Figure 1 shows the derivation of the sample used in these analyses. From an initial sample of 73 859 participants, 5774 (7.8%) did not consent to record linkage and 13 414 (18.1%) were missing essential data. This resulted in an analytic sample of 54 671 (mean (SD) age 46.8 (17.6) years). Online supplementary table S2 shows details of the individual cohort studies and comparisons between individuals who did and did not consent to mortality follow-up are shown in online supplementary table S3.

    Figure 1
    Figure 1

    Derivation of sample: individual participant meta-analysis (N=54 671) of cohort studies from four Health Surveys for England and two Scottish Health Surveys.

    Table 1 shows the baseline characteristics of the sample by pulmonary function. Poorer lung function was generally associated with a less favourable risk factor profile: shorter stature, fewer remaining in education after the compulsory school leaving age, a higher proportion from a manual occupational social class, a larger proportion of smokers, slightly higher BMI, a smaller proportion rating their health as good or very good, and a larger proportion with a longstanding illness. Individuals with poorer lung function were less likely to drink alcohol at least once weekly.

    View this table:
    Table 1

    Baseline characteristics of study members according to pulmonary function: individual participant meta-analysis (N=54 671) of cohort studies from four Health Surveys for England and two Scottish Health Surveys

    Of 7327 deaths during a mean (SD) follow-up of 11.7 (3.7) years, 459 were dementia related (139 where dementia was recorded as the underlying cause of death). The relationship between the different lung function tests was high (correlation coefficient range 0.77–0.92, all p<0.001; online supplementary table S4); we therefore report only results for FEV1 in relation to dementia death (see online supplementary table S5 for the meta-analysis results for FVC and PF). Poorer FEV1 was associated with increased risk of dementia death in a dose–response pattern (figure 2 and table 2): age- and sex-adjusted HR (95% CI) per SD decrease in FEV1 1.65 (1.42 to 1.93; p<0.001; figure 3). Adjusting individually for height, educational attainment, occupational social class, smoking status and self-reported general health decreased the magnitude of this association somewhat; other covariates had negligible effects. Adjusting for groups of covariates (socioeconomic status, health behaviours and illness) and multiple variables simultaneously further reduced the strength association but the association remained statistically significant (multivariable-adjusted HR per SD decrease in FEV1, 95% CI 1.43; 1.18 to 1.72; p<0.001).

    View this table:
    Table 2

    HRs (95% CIs) for the association between FEV1* and dementia-related death: individual participant meta-analysis (N=54 671) of cohort studies from four Health Surveys for England and two Scottish Health Surveys

    Figure 2
    Figure 2

    Kaplan-Meier cumulative hazard plot by forced expiratory volume in 1 s (FEV1) quartile: individual participant meta-analysis (N=54 671) of cohort studies from four Health Surveys for England and the two Scottish Health Surveys. FEV1 quartiles were defined using the following cut-points to give an approximately equal number of dementia deaths in each category: 1.36, 1.81 and 2.35L.

    Figure 3
    Figure 3

    Forest plot of the association between decreasing lung function (age- and sex-adjusted HR per SD decrease in forced expiratory volume in 1 s) and dementia death: individual participant meta-analysis (N=54 671) of cohort studies from four Health Surveys for England and two Scottish Health Surveys.

    Sensitivity and subgroup analyses

    We performed a series of subgroup analyses to ascertain the robustness of these associations (see online supplementary table S6). First, restricting the sample to never-smokers did not alter our conclusions, although the reduction in sample size reduced the statistical power of the analyses, making the dose–response association seen in the main results slightly less clear. Second, excluding deaths occurring within the first 5 years of follow-up to explore reverse causality (the possibility that people in the early stages of dementia were less able to perform pulmonary function tests effectively) did not affect the observed association. Third, accounting for missing data by multiple imputation also did not change our findings. Individuals with missing data for one or more variables were slightly younger, more likely to be female, less likely to be white British, more likely to have more than compulsory education, more likely to have a manual occupation, more likely to smoke, less likely to drink alcohol at least once weekly, had a lower BMI, were less likely to rate their health as good or very good, and were more likely to have a longstanding illness than participants with complete data for all variables (see online supplementary table S7). Thus, individuals with missing data did not invariably have an unfavourable risk factor profile and the statistical significance of the differences is likely to relate, in part, to the large sample size.

    Discussion

    We have shown a strong dose–response association between lung function and dementia-related death over an 11-year follow-up of six UK general population samples. Part of this association was accounted for by height, smoking, socioeconomic status and self-reported general health but the association remained even after adjustment for these factors. Being in the bottom quartile of lung function was associated with a doubling of dementia-related death.

    Comparison with other studies

    Our findings are in agreement with previous results from smaller-scale studies. The association between decreasing lung function and mortality has been well-described.12 ,20 ,21 However, there are fewer prospective studies linking pulmonary function with dementia. The largest study followed 10 211 men from 13 cohorts (including geographically-defined and occupational samples) over 40 years and identified 160 cases of dementia from death certification.23 Participants in the highest quartile of FVC had better survival compared to the lowest quartile although the linear trend was not statistically significant (multivariable-adjusted HR; 95% CI 0.54; 0.30 to 0.98; ptrend=0.28). Another study followed 9837 people aged 47–70 over a median 14.1 years.24 However, there was substantial attrition from the initial representative sample (N=15 792). They found worse survival in the lowest quartile of FEV1 (multivariable-adjusted HR, 95% CI 1.59; 0.91 to 2.78; ptrend=0.03) and FVC (2.08; 1.16 to 3.72; ptrend=0.006), compared to the highest quartile. Furthermore, the 205 dementia cases were identified from hospitalisation; while people with dementia are at an increased risk of hospitalisation,30 the majority of diagnoses are made in the community, not everyone is admitted to hospital and a dementia diagnosis is frequently unrecorded on discharge.36 ,37 Another study (N=9480) identified 2767 incident dementia cases over four decades follow-up but found no association between pulmonary function and dementia.25 The Age, Gene/Environment Susceptibility-Reykjavik Study (N=3665), over 23 years, found an association between mid-life pulmonary function (FEV1/height2) and later dementia, diagnosed by a three-stage screening process (OR=0.68, 95% CI 0.55 to 0.84, p<0.001).27 A study following 1291 women born in 1908, 1914, 1918, 1922 and 1930 systematically sampled from the census for 29 years, identified 147 cases based on clinical assessment.22 The authors reported an association between all three measures of lung function used and dementia (multivariable-adjusted HR per SD increase in pulmonary function; 95% CI PF 0.77; 0.65 to 0.91; FVC 0.71; 0.57 to 0.92; FEV1 0.75; 0.59 to 0.95). Only one of these studies included men and women and only one was truly representative of the general population,25 the other having been subject to substantial attrition.24

    Our findings are also in agreement with studies examining the relationship between pulmonary function and cognition, which is impaired in dementia. The largest followed 3036 Japanese-American men in Hawaii for at least 23 years and measured their cognition at the end of that time using the Cognitive Abilities Screening Instrument: baseline FEV1 was a predictor of cognitive score in a linear regression.18 Higher FEV1 aged 43 in 1778 men and women from the British 1946 birth cohort was associated with better psychomotor speed at that age as well as slower decline over 10 years in psychomotor speed but not verbal ability or memory.13 Another study of 1192 men and women living in the community using structural equation modelling found that, in addition to education, strenuous activity and self-efficacy predicted cognitive change over 2 years.17 A study of 857 men from Finland, the Netherlands and Italy found that baseline lung function was associated with cognitive function at least 25 years later only in non-carriers of APOE ε4 (pinteraction<0.05).14 A study assessing 832 male and female twins seven times over 19 years found changes in lung function were associated with subsequent cognitive change, particularly fluid cognitive abilities.16 However, interpretation of studies looking at cognition at a single point in time are complicated by the fact that higher mental ability in childhood is associated with better lung function in later life.15

    Strengths and limitations

    The present study is by far the largest study to date of the association between lung function and dementia and is based on general population samples. The use of individual participant meta-analysis adequately accounts for clustering within cohort studies. There has been minimal attrition due to the record linkage methodology.

    However, there are a number of limitations in these data which must be acknowledged. While we were able to include a large number of important variables in our models, the possibility of residual confounding remains. There was no information available on baseline cognitive function, respiratory illness or genetic factors in the HSE or SHS data sets. A further potential criticism is the use of dementia mortality as the outcome, though this has been used previously in another study of lung function and dementia.23 The problems associated with the use of death certification in identifying dementia cases in observational studies include the unavoidable under-diagnosis of dementia in the community,38 under-recording of dementia on death certificates39 and inaccurate coding of diagnoses. However, given the non-differential loss of data, it is likely that death certification is an adequate outcome for observational studies determining relative risk. A recent study found that almost three-quarters of a memory clinic sample diagnosed with Alzheimer disease had dementia correctly recorded on their death certificates.30

    Implications

    Neither the sensitivity analysis in the present study nor a detailed previous longitudinal study16 found evidence to suggest that the identified association between poorer lung function and dementia is attributable to reverse causality, but further research is needed to confirm this given the association between early life cognition and later life lung function.15 The precise mechanisms underlying the association between lung function and dementia remain unclear. Dementia may share some aetiology with cardiovascular disease40 and this overlap in the conditions may explain the association, since lung function is associated with cardiovascular disease.11 ,12 Alternatively, it may be that lung function contributes to cognitive reserve, protecting against cognitive decline.15 A further possibility is that, similarly to height, lung function may reflect life course exposures which also modify an individual's risk of dementia.8 ,12 We have been able to include a number of possible candidates in our models as covariates, including smoking and early life socioeconomic status (educational attainment, albeit based on recall), but these factors did not fully explain the observed association. Other hypothesised mechanisms for the association between pulmonary function and dementia include hypoperfusion and hypoxia affecting cerebral energy metabolism and leading to oxidative stress.22 The finding that lung function is only associated with cognitive change in non-carriers of APOE ε4 may suggest a more complex relationship between pulmonary function and dementia.14

    Conclusions

    We have shown a dose–response association between poorer lung function and dementia-related death in a large, general population-based sample of men and women. Further research is required to characterise this association in more detail and to identify whether interventions to improve lung function can reduce dementia risk.

    What is already known on this subject

    • Dementia is increasingly being acknowledged as a condition of major public health importance. Furthermore, there is a growing recognition that exposures across the life course contribute to the risk of developing manifest dementia in later life. Lung function is a putative marker for potentially relevant life course exposures but studies linking it with dementia are scarce.

    What this study adds

    • This large, general population-based study identifies a dose–response association between poorer lung function and the risk of dementia death with individuals at the bottom quartile of lung function having a twofold increased risk compared to those at the top quartile. Further work examining the mechanism underlying this association may shed light on the aetiology of dementia and, ultimately, guide preventative strategies.

    Acknowledgments

    The Health Survey for England is part of a programme of surveys commissioned by The UK National Health Service Information Centre for health and social care, and carried out since 1994 by the Joint Health Surveys Unit of the National Centre for Social Research (NatCen) and the Department of Epidemiology and Public Health at the University College London Medical School. Supported by Alzheimer Scotland from 2009 to 2013, TCR is now employed by the University of Edinburgh and NHS Lothian. TCR, JMS and GDB are members of both the Alzheimer Scotland Dementia Research Centre funded by Alzheimer Scotland and of the University of Edinburgh Centre for Cognitive Ageing and Cognitive Epidemiology, part of the cross council Lifelong Health and Wellbeing Initiative (G0700704/84698). Funding from the BBSRC, EPSRC, ESRC and MRC for the latter is gratefully acknowledged. ES is funded by a NIHR Career Development Fellowship and MK is supported by the Medical Research Council, UK (K013351), the US National Institutes of Health (R01HL036310; R01AG034454) and a professorial fellowship from the Economic and Social Research Council. All researchers are independent of funders who played no role in this study.

    References

    1. ↵
      1. Prince M,
      2. Guerchet M,
      3. Prina M
      ; Alzheimer's Disease International. Policy brief for heads of government: the global impact of dementia 2013–2050. London: Alzheimer Disease International, 2013.
    2. ↵
      1. Luengo-Fernandez R,
      2. Leal J,
      3. Gray A
      . Dementia 2010: the economic burden of dementia and associated research funding in the United Kingdom. A report produced by the Health Economics Research Centre, University of Oxford for the Alzheimer's Research Trust. Cambridge: Alzheimer's Research Trust, 2010.
    3. ↵
      1. McGuinness B,
      2. Todd S,
      3. Passmore P, et al
      . Blood pressure lowering in patients without prior cerebrovascular disease for prevention of cognitive impairment and dementia. Cochrane Database Syst Rev 2009;4:CD004034. doi:10.1002/14651858.CD004034.pub3
      OpenUrlPubMed
    4. ↵
      1. Braak H,
      2. Braak E
      . Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol 1991;82:239–59. doi:10.1007/BF00308809
      OpenUrlCrossRefPubMedWeb of Science
    5. ↵
      1. Braak H,
      2. Braak E
      . Frequency of stages of Alzheimer-related lesions in different age categories. Neurobiol Aging 1997;18:351–7. doi:10.1016/S0197-4580(97)00056-0
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Sperling RA,
      2. Aisen PS,
      3. Beckett LA, et al
      . Toward defining the preclinical stages of Alzheimer's disease: recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimers Dement 2011;7:280–92. doi:10.1016/j.jalz.2011.03.003
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. Whalley LJ,
      2. Dick FD,
      3. McNeill G
      . A life-course approach to the aetiology of late-onset dementias. Lancet Neurol 2006;5:87–96. doi:10.1016/S1474-4422(05)70286-6
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Ben-Shlomo Y,
      2. Kuh D
      . A life course approach to chronic disease epidemiology: conceptual models, empirical challenges and interdisciplinary perspectives. Int J Epidemiol 2002;31:285–93. doi:10.1093/ije/31.2.285
      OpenUrlFREE Full Text
    9. ↵
      1. Batty GD,
      2. Shipley MJ,
      3. Langenberg C, et al
      . Adult height in relation to mortality from 14 cancer sites in men in London (UK): evidence from the original Whitehall study. Ann Oncol 2006;17:157–66. doi:10.1093/annonc/mdj018
      OpenUrlAbstract/FREE Full Text
    10. ↵
      1. Russ TC,
      2. Kivimaki M,
      3. Starr JM, et al
      . Height in relation to dementia death: individual-participant meta-analysis of eighteen UK prospective cohort studies. Br J Psychiatry 2014;205:348–54. doi:10.1192/bjp.bp.113.142984
      OpenUrlAbstract/FREE Full Text
    11. ↵
      1. Sin DD,
      2. Wu L,
      3. Man SF
      . The relationship between reduced lung function and cardiovascular mortality: a population-based study and a systematic review of the literature. Chest 2005;127:1952–9. doi:10.1378/chest.127.6.1952
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. Batty GD,
      2. Gunnell D,
      3. Langenberg C, et al
      . Adult height and lung function as markers of life course exposures: associations with risk factors and cause-specific mortality. Eur J Epidemiol 2006;21:795–801. doi:10.1007/s10654-006-9057-2
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Richards M,
      2. Strachan D,
      3. Hardy R, et al
      . Lung function and cognitive ability in a longitudinal birth cohort study. Psychosom Med 2005;67:602–8. doi:10.1097/01.psy.0000170337.51848.68
      OpenUrlAbstract/FREE Full Text
    14. ↵
      1. Giltay EJ,
      2. Nissinen A,
      3. Giampaoli S, et al
      . Apolipoprotein E genotype modifies the association between midlife lung function and cognitive function in old age. Dement Geriatr Cogn Disord 2009;28:433–41. doi:10.1159/000255600
      OpenUrlCrossRefPubMed
    15. ↵
      1. Deary IJ,
      2. Whalley LJ,
      3. Batty GD, et al
      . Physical fitness and lifetime cognitive change. Neurology 2006;67:1195–200. doi:10.1212/01.wnl.0000238520.06958.6a
      OpenUrlCrossRef
    16. ↵
      1. Emery CF,
      2. Finkel D,
      3. Pedersen NL
      . Pulmonary function as a cause of cognitive aging. Psychol Sci 2012;23:1024–32. doi:10.1177/0956797612439422
      OpenUrlAbstract/FREE Full Text
    17. ↵
      1. Albert MS,
      2. Jones K,
      3. Savage CR, et al
      . Predictors of cognitive change in older persons: MacArthur studies of successful aging. Psychol Aging 1995;10:578. doi:10.1037/0882-7974.10.4.578
      OpenUrlCrossRefPubMedWeb of Science
    18. ↵
      1. Chyou PH,
      2. White LR,
      3. Yano K, et al
      . Pulmonary function measures as predictors and correlates of cognitive functioning in later life. Am J Epidemiol 1996;143:750–6. doi:10.1093/oxfordjournals.aje.a008812
      OpenUrlAbstract/FREE Full Text
    19. ↵
      1. Emery CF,
      2. Pedersen NL,
      3. Svartengren M, et al
      . Longitudinal and genetic effects in the relationship between pulmonary function and cognitive performance. J Gerontol B Psychol Sci Soc Sci 1998;53:P311–17. doi:10.1093/geronb/53B.5.P311
      OpenUrlPubMedWeb of Science
    20. ↵
      1. Sabia S,
      2. Shipley M,
      3. Elbaz A, et al
      . Why does lung function predict mortality? Results from the Whitehall II cohort study. Am J Epidemiol 2010;172:1415–23. doi:10.1093/aje/kwq294
      OpenUrlAbstract/FREE Full Text
    21. ↵
      1. Buchman A,
      2. Boyle P,
      3. Wilson R, et al
      . Pulmonary function, muscle strength and mortality in old age. Mech Ageing Dev 2008;129:625–31. doi:10.1016/j.mad.2008.07.003
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      1. Guo X,
      2. Waern M,
      3. Sjögren K, et al
      . Midlife respiratory function and Incidence of Alzheimer's disease: a 29-year longitudinal study in women. Neurobiol Aging 2007;28:343–50. doi:10.1016/j.neurobiolaging.2006.01.008
      OpenUrlCrossRefPubMedWeb of Science
    23. ↵
      1. Alonso A,
      2. Jacobs DR Jr.,
      3. Menotti A, et al
      . Cardiovascular risk factors and dementia mortality: 40 years of follow-up in the Seven Countries Study. J Neurol Sci 2009;280:79–83. doi:10.1016/j.jns.2009.02.004
      OpenUrlCrossRefPubMed
    24. ↵
      1. Pathan SS,
      2. Gottesman RF,
      3. Mosley TH, et al
      . Association of lung function with cognitive decline and dementia: the Atherosclerosis Risk in Communities (ARIC) Study. Eur J Neurol 2011;18:888–98. doi:10.1111/j.1468-1331.2010.03340.x
      OpenUrlCrossRefPubMed
    25. ↵
      1. Exalto LG,
      2. Quesenberry CP,
      3. Barnes D, et al
      . Midlife risk score for the prediction of dementia four decades later. Alzheimers Dement 2014;10:562–70. doi:10.1016/j.jalz.2013.05.1772
      OpenUrlCrossRefPubMedWeb of Science
    26. ↵
      1. Schaub RT,
      2. Munzberg H,
      3. Borchelt M, et al
      . Ventilatory capacity and risk for dementia. J Gerontol A Biol Sci Med Sci 2000;55:M677–83. doi:10.1093/gerona/55.11.M677
      OpenUrlAbstract/FREE Full Text
    27. ↵
      1. Vidal JS,
      2. Aspelund T,
      3. Jonsdottir MK, et al
      . Pulmonary function impairment may be an early risk factor for late-life cognitive impairment. J Am Geriatr Soc 2013;61:79–83. doi:10.1111/jgs.12069
      OpenUrlCrossRefPubMed
    28. ↵
      1. Mindell J,
      2. Biddulph JP,
      3. Hirani V, et al
      . Cohort profile: the Health Survey for England. Int J Epidemiol 2012;41:1585–93. doi:10.1093/ije/dyr199
      OpenUrlAbstract/FREE Full Text
    29. ↵
      1. Gray L,
      2. Batty GD,
      3. Craig P, et al
      . Cohort profile: the Scottish Health Surveys cohort. Int J Epidemiol 2009;39:345–50. doi:10.1093/ije/dyp155
      OpenUrlPubMedWeb of Science
    30. ↵
      1. Russ TC,
      2. Batty GD,
      3. Starr JM
      . Cognitive and behavioural predictors of survival in Alzheimer disease: results from a sample of treated patients in a tertiary-referral memory clinic. Int J Geriatr Psychiatry 2012;27:844–53. doi:10.1002/gps.2795
      OpenUrlCrossRefPubMed
    31. ↵
      1. Russ TC,
      2. Hamer M,
      3. Stamatakis E, et al
      . Psychological distress as a risk factor for dementia death. Arch Intern Med 2011;171:1858–9. doi:10.1001/archinternmed.2011.501
      OpenUrlCrossRefPubMedWeb of Science
    32. ↵
      1. Russ TC,
      2. Hamer M,
      3. Stamatakis E, et al
      . Does the Framingham cardiovascular disease risk score also have predictive utility for dementia death? An individual participant meta-analysis of 11,887 men and women. Atherosclerosis 2013;228:256–8. doi:10.1016/j.atherosclerosis.2013.02.020
      OpenUrlCrossRefPubMedWeb of Science
    33. ↵
      1. Russ TC,
      2. Stamatakis E,
      3. Hamer M, et al
      . Socioeconomic status as a risk factor for dementia death: an individual participant meta-analysis of 86 508 men and women from the UK. Br J Psychiatry 2013;203:10–17. doi:10.1192/bjp.bp.112.119479
      OpenUrlAbstract/FREE Full Text
    34. ↵
      1. Cox DR
      . Regression models and life-tables. J Roy Stat Soc B 1972;34:187–220.
      OpenUrl
    35. ↵
      1. von Elm E,
      2. Altman D,
      3. Egger M, et al
      . Strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ 2007;335:806–8. doi:10.1136/bmj.39335.541782.AD
      OpenUrlFREE Full Text
    36. ↵
      1. Scott IA,
      2. Ward M
      . Public reporting of hospital outcomes based on administrative data: risks and opportunities. Med J Aust 2006;184:571.
      OpenUrlPubMedWeb of Science
    37. ↵
      1. Russ TC,
      2. Parra MA,
      3. Lim AE, et al
      . Predictors of general hospital admission in people with dementia: cohort study. Br J Psychiatry 2015;206:153–9.
    38. ↵
      1. Sampson EL,
      2. Blanchard MR,
      3. Jones L, et al
      . Dementia in the acute hospital: prospective cohort study of prevalence and mortality. Br J Psychiatry 2009;195:61–6. doi:10.1192/bjp.bp.108.055335
      OpenUrlAbstract/FREE Full Text
    39. ↵
      1. Martyn CN,
      2. Pippard EC
      . Usefulness of mortality data in determining the geography and time trends of dementia. J Epidemiol Community Health 1988;42:134. doi:10.1136/jech.42.2.134
      OpenUrlAbstract/FREE Full Text
    40. ↵
      1. Stephan BC,
      2. Brayne C
      . Vascular factors and prevention of dementia. Int Rev Psychiatry 2008;20:344–56. doi:10.1080/09540260802094456
      OpenUrlCrossRefPubMedWeb of Science

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

      Files in this Data Supplement:

    Footnotes

    • Contributors TCR conceived and designed the study. ES and GDB were responsible for acquisition of data. TCR and GDB were responsible for analysis and interpretation of data. TCR and GDB drafted the manuscript. All authors critically revised the manuscript for important intellectual content. TCR and GDB did the statistical analysis. GDB obtained funding. MK and GDB were responsible for study supervision. TCR and GDB are the study guarantors.

    • Funding Alzheimer Scotland; Economic and Social Research Council (author MK only); US NIH (author MK only) R01AG034454 and R01HL036310; NIHR Career Development Fellowship (author ES only); Medical Research Council (author MK only) K013351; Lifelong health and wellbeing G0700704/84698.

    • Competing interests None.

    • Ethics approval London Research Ethics Council or the Local Research Ethics Councils.

    • Provenance and peer review Not commissioned; externally peer reviewed.