Lung Function and Cognitive Ability in a Longitudinal Birth Cohort Study
MARCUS RICHARDS, PHD, DAVID STRACHAN, MD, REBECCA HARDY, PHD, DIANA KUH, PHD,
AND MICHAEL WADSWORTH, PHD
Objective: The objective of this study was to examine the association between forced expiratory volume in 1 second (FEV1) and
cognitive ability in midlife in the normal population.Methods:Multiple regression was used to test associations between FEV1 and
cognitive function in 1778 men and women in the MRC National Survey of Health and Development, also known as the British
1946 birth cohort. Analyses were adjusted for sex, body size (birth weight, adult height, weight, and chest circumference),
socioeconomic status, lifetime smoking, and a range of health indicators, including early respiratory vulnerability (infant lower
respiratory infection, childhood asthma, and exposure to atmospheric pollution). Results: FEV1 at 43 years was associated with
slower psychomotor speed (peg placement) at the same age and with slower decline in psychomotor speed (letter search speed) from
43 to 53 years, independently of the previously mentioned potential confounders. These independent associations were not
observed, however, for adult verbal ability, verbal memory, or rate of decline in memory, which were significantly explained by
socioeconomic status and adolescent cognitive ability. In a subsequent analysis, adolescent cognition was positively associated with
FEV1, although not with rate of decline in FEV1 from 43 to 53 years, again independently of the previously mentioned confounders.
Conclusions: Cognitive function and FEV1 are positively associated across the life course. One possible explanation lies in the
parallel action of endocrine, autonomic, and motor control systems on respiration and higher mental function. Because respiration
and mental function are both associated with functional capacity and survival, this is a matter of potential clinical significance.
Key words: lung function, cognitive function, birth cohort.
AH4 � Alice Heim group abilities test; COPD � chronic obstruc-
tive pulmonary disease; FEV1 � forced expiratory volume within 1
second; FVC � forced vital capacity; NSHD � National Survey of
Health and Development; NART � National Adult Reading Test;
SES � socioeconomic status.
INTRODUCTION
Several lines of evidence indicate that lung function isassociated with cognitive performance. Patients with
chronic obstructive pulmonary disease (COPD) show slowed
electrocortical activity (1,2) and impaired cognitive function
(3,4). Conversely, cognitive function is improved after stabi-
lization of circadian variation in ventilatory flow in asthmatics
(5). The association between respiratory and cognitive func-
tion is also observed in the normal population in cross-sec-
tional (6–8) and longitudinal (9–11) studies, in which, to
varying degrees, relevant demographic, socioeconomic, and
health-related confounders are controlled. This link between
respiratory and cognitive function is of clinical importance,
because decreased pulmonary function is associated with in-
creased risk of dementia in the population, even after control-
ling for COPD (12), and pulmonary and cognitive function are
both associated with survival (13,14).
In terms of underlying mechanisms, the most widely held
view is that impaired lung function causes changes in the
central nervous system (CNS) through possible processes such
as vascular disease resulting from inflammation, impaired
fibrinolytic activity, oxidative stress or cardiovascular risk
factors (see 15), or hypoxia-induced changes in neurotrans-
mitter metabolism (16). Such CNS changes then lead to lower
cognitive function. However, given that respiratory control
requires the operation of brain structures across many levels of
the neuraxis (17), it may be that cognitive function is a marker
of the general integrity of neurorespiratory regulation. If so,
the association between respiration and cognition may be
evident at an earlier stage of the life course.
The British 1946 birth cohort study has measured pulmo-
nary function in midlife before lung disease has had a major
impact on survival. We investigated the association between
lung function and cognition at 43 years, and between lung
function at this age and change in cognitive function from 43
to 53 years, controlling for a range of sociodemographic,
socioeconomic, and health-related indicators. We then inves-
tigated the association between childhood cognition and adult
lung function and change.
METHODS AND MATERIALS
Study Participants
Participants were drawn from the MRC National Survey of Health and
Development (NSHD), a birth cohort study stratified by social class and
initially consisting of 5362 people selected from all births that occurred in
England, Scotland, and Wales during 1 week in March 1946 (18). Information
about sociodemographic factors and medical, cognitive, and psychological
function has been repeatedly obtained by interview and examination, most
recently in 1999 at age 53 years, when sample size was 3035. At this time, the
cohort was shown still to be a representative sample, in most respects, of the
U.K. population legitimately and singly born in the immediate postwar era
(19). Exceptions were an overrepresentation among nonresponders of those
never married and those least advantaged in terms of cognitive ability,
educational attainment, and social class. Ethical approval for this study came
from the North Thames Multicenter Research Ethics Committee.
Lung Function
Lung function was represented by forced expiratory volume in 1 second
(FEV1) measured at 43 years and 53 years using a Micro Medical Micro Plus
spirometer, administered by a trained nurse. On both occasions, participants
were asked to stand, fill their lungs to capacity, make an airtight seal around
the mouthpiece with their lips, then blow as hard and fast as possible until
their lungs were empty. Participants did not wear a noseclip. The nurse first
demonstrated the technique and then provided verbal encouragement during
From the MRC National Survey of Health and Development, University
College London, U.K. (M.R., R.H., D.K., M.W.); and the Department of Public
Health Sciences, St. George’s Hospital Medical School, London, U.K. (D.S.).
Address correspondence and reprint requests to Marcus Richards, PhD,
University College London, Department of Epidemiology & Public Health,
1–19 Torrington Place, London WC1E 6BT, U.K. E-mail: m.richards@
ucl.ac.uk
Received for publication April 22, 2004; revision received January 25,
2005.
The National Survey of Health and Development is funded by the Medical
Research Council. Data collection at 53 years was carried out by the National
Centre for Social Research.
DOI: 10.1097/01.psy.0000170337.51848.68
602 Psychosomatic Medicine 67:602–608 (2005)
0033-3174/05/6704-0602
Copyright © 2005 by the American Psychosomatic Society
the trials. Three trials were given at 43 years, and 2 trials were given at 53
years. The higher value of the first 2 trials, and of either trial, respectively,
was used for analysis. When the survey member’s spirometer technique at 53
years was rated as unsatisfactory by the nurse, values for that trial were
classified as missing data. Nine survey members were excluded from the
analysis because their technique was unsatisfactory on both trials. These nurse
ratings were not undertaken at 43 years. Variation in FEV1 across both trials
was within 5% for 77.5% of the sample at 43 years and 79.6% of the sample
at 53 years.
Adult Cognitive Measures
At 43 years, they were given a timed peg placement test (overall mean of
3 trials for each hand subjected to a log transformation to improve distribu-
tion), and at 43 and 53 years, they were given repeat tests of memory (a 3-trial
15-item word list), and speed and concentration (timed letter search). A
different word list was given to each half of the cohort at 43 years, and then
these lists were reversed at 53 years. Target letters were in different positions
on the page at 43 and 53 years. At age 53, they also took the National Adult
Reading Test (NART) (20), a test of verbal ability requiring the pronunciation
of 50 irregular words of increasing difficulty. All cognitive measures were
administered after appropriate training to the nurses.
Potential Confounding Variables
Body size is a potential confounder because birth weight is associated
with ventilatory function (21), because FEV1 increases with height, and
because pre- and postnatal growth is associated with cognitive function
(22,23). The following parameters were used: birth weight, adult leg length,
trunk length, expanded chest circumference, and weight current to FEV1 at 43
years. Adult standing height was measured at 53 years to the nearest 0.5 cm
using a portable stadiometer, as was sitting height (to represent trunk length).
Leg length was calculated as the difference between sitting and standing
height. Expanded chest circumference was also measured at 53 years in men
at nipple level and in women immediately below the breasts.
Socioeconomic status (SES) was represented by father’s social class,
mother’s education, material home conditions, own educational attainment,
and own current or last occupational social class at 43 years. An aggregate
variable representing material home conditions at age 4 years comprised
ratings by a health visitor of age, cleanliness, and state of repair of the
dwelling; number of people per room; and cleanliness and condition of
clothing and shoes of the survey member. This total score was then catego-
rized into very good, good, modest, or poor (24). For own educational
attainment, the highest educational qualifications and their training equiva-
lents attained by 26 years were classified as none, vocational only, ordinary
secondary (O levels), advanced secondary (A levels), or degree level or
equivalent. Social class was classified as professional, managerial, interme-
diate, skilled manual, semiskilled manual, or unskilled, according to the
Registrar General.
Smoking impairs FEV1 and is associated with cognitive decline (25).
Information on daily cigarette consumption was obtained at ages 20, 25, 31,
36, 43, and 53 years. A measure of lifetime smoking to 43 years was
calculated by multiplying pack-years at each age by the number of years to the
next age (e.g., smoking exposure from 25–31 years was calculated as pack-
years at 25 multiplied by 6), and then totalling and averaging these products.
To include adolescent smoking, pack-years at 20 years was multiplied back to
16 years.
Other health-related measures at 43 years such as physical exercise, which
is protective of cognitive decline (26), was classified as 0, 1 to 4, or 5�
sporting or recreational activities per month based on an equivalent scheme
for activities at 36 years (27). Resting pulse and systolic and diastolic blood
pressure were measured by a research nurse. Significant respiratory problems
were classified by self-reports of any of the following: a wheezy or whistling
chest most days or nights, usually bringing up phlegm or coughing in the
morning or during the day or night in winter for at least 3 months each year,
or any chest illness (e.g., bronchitis, pneumonia) that required sickness
absence of a week or more. Affective state was measured by the Psychiatric
Symptom Frequency (PSF) scale (28).
Information on lower respiratory infection in infancy, which is a risk
factor for adult respiratory problems (29,30), was provided by the respon-
dents’ mothers. A small number of survey members (n � 58) had also been
noted as having asthma by a school doctor at 6, 11, or 15 years. Because this
cohort was born 10 years before the Clean Air Act of 1956, exposure to air
pollution from incomplete coal combustion from birth to 9 years, estimated
from validated measures (31), was also used as a potential confounder.
To investigate a possible neurodevelopmental pathway to lung function,
the effect of adjusting for cognitive ability at 15 years was examined. At this
age, children took the Heim AH4 test (32), the Watts-Vernon Reading Test
(33), and a mathematics test. The AH4 is a 130-item ability test, with verbal
items (analogies, comprehension, and numerical reasoning) and nonverbal
items (matching, spatial analysis, and nonverbal reasoning) summed to yield
a general ability score. The Watts-Vernon is a test of reading comprehension,
requiring selection of appropriate words to complete 35 sentences. The
mathematics test consisted of 47 items, requiring the use of arithmetic,
geometry, trigonometry, and algebra. Concerning reliability of these tests,
Pidgeon (33) quoted a figure of 0.92 for test–retest consistency of the total
AH4 score with an interval of 1 month, a figure of 0.89 for reliability of the
Watts-Vernon reading test as calculated from the Kuder-Richardson Formula
20 test, and noted a similar figure for test–retest analysis.
Scores for these three tests were summed to obtain an overall score
representing general cognitive ability. This score was normally distributed
within the present study sample.
Statistical Analysis
The association between FEV1 at 43 and midlife cognitive function was
tested using memory, search speed, and peg placement speed at 43 years,
and the NART at 53 years as outcomes. Conditional models of change were
used to investigate the association between FEV1 at 43 and change in memory
and search speed from 43 to 53 years, adjusting the memory and search speed
scores at 53 years for their corresponding scores at 43 years. These analyses
were performed in stages, controlling for sex and body size (model 1), then
SES (model 2), then lifetime smoking (model 3), and then cognition at 15
years (model 4). We then tested the association between cognitive ability at 15
years and FEV1 at 43 years, and, also using conditional analysis of change,
rate of FEV1 decline from 43 to 53 years, similarly adjusting for covariates in
stages, according to models 1 to 4.
For all analyses, FEV1, the age 15 cognitive ability score, lifetime smok-
ing, and all body size parameters were entered as continuous variables,
whereas education and social class were entered as categorical variables.
Finally, all stage 4 models were adjusted for each remaining health
indicator in turn, i.e., physical exercise, pulse, blood pressure, adult lung
disease, affective state (PSF), infant lower respiratory infection, childhood
asthma, and atmospheric pollution.
RESULTS
Sample Size
Of the 3035 cohort members who provided information at
53 years, 1778 had nonmissing data for FEV1 at 43 and 53
years, all cognitive variables, parental and adult SES, all body
size parameters, and smoking. Those with missing data for
any of these variables had lower cognitive ability in childhood
(p � .001).
Forced Expiratory Volume in 1 Second and Midlife
Cognition
Tables 1 to 3 show regression coefficients and 95% confi-
dence intervals representing mean increase in peg placement
time at 43 years and verbal ability (NART) at 53 years, and
mean difference (change) in memory and search speed for
given memory and search speed scores at 43 years, per unit
increase in FEV1 at 43 years, in which positive coefficients
LUNG FUNCTION AND COGNITION
603Psychosomatic Medicine 67:602–608 (2005)
TABLE 1. Regression Coefficients and 95% Confidence Intervals Representing Mean Difference in Peg Placement Speed and NART Score Per
Unit Increase in Forced Expiratory Volume in 1 Second at 43 Years,a Progressively Adjusting for Sex and Body Size,b Socioeconomic Status,c
Smoking, and Adolescent Cognitive Ability (n � 1778)
Peg placement speed (43 Yr) NART (53 Yr)
Regression Coefficient
(95% CI)
p
Regression Coefficient
(95% CI)
p
Model 1
(FEV1 adjusted for sex and body size
b) �0.05 (�0.07, �0.03) �.001 2.84 (1.99, 3.70) �.001
Model 2
(FEV1 further adjusted for SES
c) �0.05 (�0.07, �0.03) �.001 0.73 (0.01, 1.44) .05
Model 3
(FEV1 further adjusted for smoking) �0.05 (�0.07, �0.03) �.001 0.76 (0.02, 1.50) .04
Model 4
(FEV1further adjusted for cognition at age 15 yr) �0.05 (�0.07, �0.03) �.001 0.53 (�0.12, 1.19) .11
a 53 years for the NART.
b Birth weight, weight at 43 years, and adult leg length, trunk length and expanded chest circumference.
c Paternal social class, maternal education, early material home conditions, own educational attainment and own social class at 43 years.
TABLE 2. Regression Coefficients and 95% Confidence Intervals Representing Mean Difference in Memory at 43 Years, and Rate of Change in
Memory From 43 to 53 Years, per Unit Increase in FEV1 at 43 Years Progressively Adjusting for Sex and Body Size,
a SES,b Smoking, and
Adolescent Cognitive Ability (n � 1778)
Verbal Memory at 43 Yr Decline in Memory From 43 to 53 Yr
Regression Coefficient (95% CI) p Regression Coefficient (95% CI) p
Model 1
(FEV1 adjusted for sex and body size
a) 1.20 (0.67, 1.72) �.001 0.39 (�0.02, 0.80) .06
Model 2
(FEV1 further adjusted for SES
b) 0.55 (0.08, 1.02) .02 0.20 (�0.21, 0.60) .34
Model 3
(FEV1 further adjusted for smoking) 0.54 (0.07, 1.02) .02 0.16 (�0.25, 0.57) .45
Model 4
(FEV1 further adjusted for cognition at 15 yr) 0.30 (�0.15, 0.76) .19 0.06 (�0.35, 0.46) .79
Note. Positive coefficients represent higher memory scores at 43 years, and slower memory decline from 43 to 53 years with better lung function.
a Birth weight, weight at 43 yr, and adult leg length, trunk length and expanded chest circumference.
b Paternal social class, maternal education, early material home conditions, own educational attainment and own social class at 43 yr.
TABLE 3. Regression Coefficients and 95% Confidence Intervals Representing Mean Difference in Search Speed at 43 Years, and Rate of
Change in Search Speed From 43 to 53 Years, per Unit Increase in FEV1 at 43 Years, Progressively Adjusting for Sex and Body Size,
a SES,b
Smoking, and Adolescent Cognitive Ability (n � 1778)
Search Speed at 43 Yr Decline in Search Speed From 43 to 53 Yr
Regression coefficient (95% CI) p Regression coefficient (95% CI) p
Model 1
(FEV1 adjusted for sex body size
a) 3.42 (�3.02, 9.85) .30 12.00 (6.52, 17.50) �.001
Model 2
(FEV1 further adjusted for SES
b) 0.59 (�5.88, 7.06) .86 9.84 (4.31, 15.37) �.001
Model 3
(FEV1 further adjusted for smoking) �0.49 (�7.05, 6.06) .88 8.92 (3.31, 14.53) .002
Model 4
(FEV1 further adjusted for cognition at 15 yr) �0.87 (�7.44, 5.70) .80 7.94 (2.35, 13.54) .005
Note. Positive coefficients represent faster search time at 43 years, and slower decline in search time from 43 to 53 years, with increasing lung function.
a Birth weight, weight at 43 years, and adult leg length, trunk length and expanded chest circumference.
b Paternal social class, maternal education, early material home conditions, own educational attainment and own social class at 43 years.
M. RICHARDS et al.
604 Psychosomatic Medicine 67:602–608 (2005)
represent slower decline. These associations were adjusted for
sex and body size (model 1), then SES (model 2), then lifetime
smoking (model 3), and then cognitive ability at 15 years
(model 4).
FEV1 was positively associated with peg placement speed
(model 1) in that a negative coefficient represented faster task
completion time with better lung function. The strength of this
association was unaffected by adjusting for SES, smoking, or
adolescent cognitive ability (models 2–4).
Although FEV1 was strongly associated with the NART at
53 years with sex- and body size-adjusted (model 1), the
strength of this association was greatly reduced after control-
ling for SES. The coefficient was essentially unchanged by
subsequent control for lifetime smoking. Indeed, when smok-
ing was added to model 1 without adjustment for SES, FEV1
was still strongly associated with the NART (p � .001).
However, the association was further reduced by control for
adolescent cognitive ability (model 4), after which it was no