socio-economic position across the life course and hysterectomy in three british cohorts: a...

Socio-economic position across the life course and hysterectomyin three British cohorts: a cross-cohort comparative study

Rachel Cooper,a Debbie A. Lawlor,b Rebecca Hardy,a Shah Ebrahim,b

David A. Leon,c Michael E.J. Wadsworth,a Diana Kuha

Objective To examine the association between indicators of lifetime socio-economic position and rates ofhysterectomy in three British cohorts.

Design Cross-cohort comparative study.

Setting Two cohorts: England, Scotland and Wales. Third cohort: Aberdeen, Scotland.

Population Three thousand two hundred and eight women born between 1919 and 1940, participating in theBritish Women’s Heart and Health Study (BWHHS); 1394 women from the MRC National Survey ofHealth and Development (NSHD), followed up since birth in 1946; 3208 women born between 1950 and1955, participating in the Aberdeen Children of the 1950s study, all with complete information on lifetimesocio-economic position and hysterectomy status.

Methods Relative indices of inequality were derived for markers of socio-economic position in childhood andadulthood. Cox’s regression models were used to test the association between these markers and hysterectomy.

Main outcome measure Self-reported hysterectomy with or without oophorectomy.

Results Adverse socio-economic position in childhood and as indicated by educational status was associatedwith reduced rates of hysterectomy in the oldest of the three cohorts (BWHHS), whereas conversely in theNSHD and Aberdeen cohorts it was associated with increased rates of hysterectomy. The unadjusted hazardsratios for hysterectomy comparing worst to best socio-economic position for father’s social class were 0.73(0.56, 0.96) for women from the BWHHS, 1.77 (1.19, 2.65) for those from the NSHD and 2.06 (1.46, 2.89)for those from the Aberdeen cohort. Associations between markers of adult socio-economic position andhysterectomy tended to be weaker in all three cohorts and often did not reach conventional levels ofstatistical significance.

Conclusions Our results show that hysterectomy rates are influenced by childhood socio-economic positionand educational attainment, but that the nature of this association varies across these three British cohortsborn in different decades of the 20th century. That there were no consistent or strong associations betweenadult SEP and hysterectomy rates suggest that social factors influencing rates of hysterectomy are likely tobe those experienced or which develop in early life rather than those which develop later.

INTRODUCTION

Approximately 20% of women in the UK undergo hys-

terectomy by age 60 years1,2 and this commonly performed

procedure is usually discretionary.2

Biological and social factors acting at different stages of

the life course3,4 may increase the likelihood of hysterecto-

my. Indicators of adverse socio-economic position (SEP) in

adulthood have often been found to be associated with in-

creased risk of hysterectomy.5 – 14 Education has been found

to be inversely related to hysterectomy with more consisten-

cy between studies than other SEP variables.5 – 9,12 – 15 Edu-

cation may influence risk of hysterectomy through its impact

on health choices, adult circumstances and behaviours,16

but also because it is a marker of social conditions in ear-

lier life.17,18 Other factors influencing risk of hysterecto-

my, including obesity,19 are associated with childhood

socio-economic disadvantage.20 There are conflicting find-

ings about whether SEP in early life affects the risk of

hysterectomy.9,11 The influence of social factors may vary

by geographic location and by birth cohort and be driven

by differences in medical attitudes, culture or disease risk

between places and over time,21 but investigators rarely

have the opportunity to examine this within one study.

We aimed to examine associations of child and adult

SEP with hysterectomy rates in three British cohorts, of

women born in the 1920s and 1930s [British Women’s

Heart and Health Study (BWHHS)], in 1946 [MRC National

BJOG: an International Journal of Obstetrics and GynaecologyAugust 2005, Vol. 112, pp. 1126–1133

D RCOG 2005 BJOG: an International Journal of Obstetrics and Gynaecology www.blackwellpublishing.com/bjog

aMRC National Survey of Health and Development,

Department of Epidemiology and Public Health, University

College London, UKbDepartment of Social Medicine, University of Bristol, UKcDepartment of Epidemiology and Population Health,

London School of Hygiene and Tropical Medicine, UK

Correspondence: Ms R. Cooper, Department of Epidemiology and Public

Health, University College London, 1-19 Torrington Place, London WC1E

6BT, UK.

DOI: 10.1111/j .1471-0528.2005.00654.x

Survey of Health and Development (NSHD)] and between

1950 and 1955 [Aberdeen Children of the 1950s study

(Aberdeen study)].

METHODS

The BWHHS is a cohort of 4286 women, born between

1919 and 1940, randomly selected from general practition-

er lists in 23 British towns. Baseline data were collected

between 1999 and 2001, when study members were 60–

79 years old, when gynaecological history and details of

childhood socio-economic circumstances were obtained ret-

rospectively and details of contemporary socio-economic

circumstances were recorded.22,23

The MRC NSHD is a nationally representative, pro-

spective birth cohort. Information on health and social fac-

tors has been collected regularly since their births in March

1946. Questions on hysterectomy and oophorectomy were

included at two home visits (ages 43 and 53 years) and in

postal questionnaires sent annually between the ages of 47

and 54 and 57 years.24 Of the 2547 women in the original

cohort 750 (29.4%) have not answered a questionnaire since

before 1989.

The Aberdeen study is a cohort of 12,150 men and wom-

en born in Aberdeen between January 1950 and December

1955 who attended Aberdeen primary schools, participated

in surveys in 1962, of whom 99% were successfully traced

(N ¼ 12,115) in 1999.25 As of October 2003, 291 of those

traced had emigrated, 62 were in the armed forces or in-

stitutionalised and 479 had died. The remaining 11,282

were mailed a follow up health questionnaire to which 3751

(67.7%) of the females responded. This postal question-

naire sent in 2000/1, when participants were 45–51 years

old, collected information on a range of information in-

cluding SEP and hysterectomy.

Relevant British ethics committee approval was received

for all three studies.

Hysterectomy was defined as any self-report of hyster-

ectomy with or without oophorectomy. All hysterectomies

with a valid age at procedure as reported at baseline as-

sessment in the BWHHS, up to age 57 years in the NSHD

and in the 2000/1 questionnaire in the Aberdeen cohort were

included.

To ensure comparability of results, socio-economic var-

iables available in at least two of the three cohorts were

selected.

Markers of childhood SEP were as follows: father’s oc-

cupational social class; level of crowding (categorised as

�1; 1.1–1.9; �2 persons per room); family access to a car;

running hot water in the home; indoor bathroom; sharing a

bedroom; and housing tenure. All these measures were

retrospectively recalled in the BWHHS and were collected

prospectively at age 11 years in the NSHD (or at ages 15, 4

or 8 years if missing at age 11, this varied by variable but

applied to up to 8 women except for father’s social class

where it applied to 77). In the Aberdeen cohort, father’s

social class was recorded at birth and all other measures

were collected retrospectively in the 2000/1 questionnaire

by asking about conditions at age 12 years.

Education was included in analyses as highest level of

educational attainment (grouped in five categories as fol-

lows: degree or higher; ‘A’ levels or equivalent; ‘O’ levels

or equivalent; CSE, clerical course or equivalent; none),

and whether cohort members were still in education after

reaching minimum legal school leaving age or not. Markers

of adult SEP were as follows: occupational social class of

self and head of household; housing tenure; and access to a

car. In the BWHHS and Aberdeen cohorts all measures of

adult SEP and education were measured at the time of base-

line data collection and 2000/1 postal questionnaire respec-

tively. In the NSHD all measures of adult SEP were taken

at a home visit at age 43 years (or at age 53, 36 or 26 years

if missing at age 43, this varied by variable but applied to

between 93 and 281 women) and educational measures were

taken at age 26 years.

Table 1. Age-specific hysterectomy rates in three British cohorts.

Age group ( years) Hysterectomy rate per 1000 women years

BWHHS (N ¼ 4286)

birth years: 1919–1940

NSHD (N ¼ 1790)

birth years: 1946

Aberdeen (N ¼ 3685)


No. of cases Rate (95% CI) No. of cases Rate (95% CI) No. of cases Rate (95% CI)

0 – 39 191 1.12 (0.97, 1.29) 109 1.54 (1.27, 1.85) 244 1.69 (1.50, 1.92)

40– 44 215 10.79 (9.44, 12.34) 99 12.31 (10.11, 14.99) 193 10.29 (8.94, 11.85)

45– 49 204 10.81 (9.43, 12.40) 110 15.11 (12.54, 18.22) 98 9.49 (7.79, 11.57)

50– 54 134 7.47 (6.30, 8.84) 63 10.03 (7.84, 12.84) 1 –

55– 59* 62 3.54 (2.76, 4.53) 14 6.16 (3.65, 10.39) – –

�60 93 2.77 (2.26, 3.39) – – – –

Total 899 3.23 (3.02, 3.45) 395 4.17 (3.78, 4.60) 536 3.09 (2.84, 3.37)

BWHHS ¼ British Women’s Heart and Health Study; NSHD ¼ National Survey of Health and Development; CI ¼ confidence interval.* The rate of hysterectomy in women of the NSHD presented is for the age group 55– 57 years.

LIFETIME SOCIO-ECONOMIC POSITION AND HYSTERECTOMY 1127

D RCOG 2005 BJOG: an International Journal of Obstetrics and Gynaecology 112, pp. 1126–1133

Occupational social class was classified according to the

Registrar General’s classification and grouped into six cate-

gories: I; II; III non-manual; III manual; IV; and V. Housing

tenure was categorised as owner occupied versus other. All

other variables unless otherwise stated were coded as bina-

ry variables.

Relative indices of inequality (RII) enable direct com-

parison between cohorts of SEP variables as they take ac-

count of differences between cohorts in the proportions of

women in the different categories of a socio-economic

variable.26 For each indicator of SEP a score between 0

(highest SEP) and 1 (lowest SEP) was assigned to each

Table 2. Comparison of rates of hysterectomy in women excluded due to incomplete data on SEP with those who are included in further analyses.

Status Hysterectomy rate per 1000 women years

BWHHS birth years: 1920– 1940 NSHD birth years: 1946 Aberdeen birth years: 1950– 1955


Included 705 2.73 (2.37, 3.14) 305 4.12 (3.68, 4.61) 448 2.97 (2.71, 3.26)

Excluded 194 3.40 (3.16, 3.67) 90 4.34 (3.53, 5.34) 88 3.90 (3.16, 4.80)

P 0.005 0.55 0.02

BWHHS ¼ British Women’s Heart and Health Study; NSHD ¼ National Survey of Health and Development; CI ¼ confidence interval.

Table 3. Unadjusted hazards ratios for hysterectomy comparing worst socio-economic condition with best for each indicator of socio-economic position from

across the life course in three British cohorts using RII.

BWHHS (1920– 1940) N ¼¼¼ 3208,

hysterectomy ¼¼¼ 705

NSHD (1946) N ¼¼¼ 1394,


Aberdeen (1950– 1955) N ¼¼¼ 3208,


HR (95% CI) HR (95% CI) HR (95% CI)

Childhood SEP

Father’s social class 0.73 (0.56, 0.96) 1.77 (1.19, 2.65) 2.06 (1.46, 2.89)

P 0.02 0.005 <0.001

Level of crowding – 1.22 (0.77, 1.94) 1.63 (1.14, 2.32)

P – 0.40 0.007

Family access to a car 0.63 (0.43, 0.92) – 1.80 (1.24, 2.62)

P 0.02 – 0.002

Running hot water 0.60 (0.46, 0.88) 1.31 (0.77, 2.24) –

P 0.006 0.33 –

Bathroom in house 0.62 (0.45, 0.86) 0.90 (0.51, 1.61) –

P <0.001 0.73 –

Own bedroom 0.73 (0.41, 1.28) 1.34 (0.85, 2.11) –

P 0.27 0.20 –

Family housing – 1.58 (0.97, 2.56) 1.74 (1.06, 2.86)

P – 0.07 0.03

Age at leaving school 0.59 (0.43, 0.81) Up to 44 yearsy

2.12 (1.08, 4.17)

44–57 years*

1.03 (0.56, 1.90)

2.34 (1.62, 3.37)

P 0.001 0.03 0.93 <0.001

Highest educational attainment – 2.45 (1.33, 4.54) 0.95 (0.55, 1.65) 1.90 (1.35, 2.66)

P – 0.004 0.87 <0.001

Adult SEP

Own social class 0.93 (0.70, 1.23) 1.57 (1.05, 2.37) 1.29 (0.90, 1.86)

P 0.60 0.03 0.18

Head of household social class 1.08 (0.82, 1.42) Up to 44 yearsy

2.07 (1.14, 3.76)

44–57 years*

1.10 (0.63, 1.92)

–

P 0.57 0.02 0.74 –

Housing 0.75 (0.51, 1.12) 1.26 (0.70, 2.26) 1.26 (0.78, 2.04)

P 0.16 0.44 0.35

Access to a car 0.43 (0.30, 0.63) – 1.29 (0.71, 2.35)

P <0.001 – 0.40


Numbers in parentheses after the study name indicate the birth year(s) of the cohort.* Age 44–57 years (N ¼ 1219, no. of hysterectomies ¼ 164).y Up to age 44 years (N ¼ 1394, no. of hysterectomies ¼ 141).

1128 R. COOPER ET AL.


category based on the proportion of the population above the

mid-point in that category. For example, if 10% of the

population are in social class I women in this group are

represented by the range 0–0.1 and so are allocated the

score 0.05 (0.1/2). If 20% of the population are in the next

highest group, social class II, then this social class is

allocated a score 0.20 (0.1 þ 0.2/2) and so on. The RII is

then obtained by regressing the outcome on each of these

SEP scores and is directly interpretable for each SEP in-

dicator used as comparing women of the lowest SEP (1)

with the highest SEP (0).

The rate of hysterectomy in five-year age bands was cal-

culated in each cohort. To ensure comparability of results

in all subsequent analyses only women who had complete

data on hysterectomy and all markers of SEP (BWHHS

N ¼ 3208; NSHD N ¼ 1394; Aberdeen N ¼ 3208) were

included. Rates of hysterectomy in those excluded because

of incomplete SEP information were compared with rates

in those with complete data, in each cohort.

Cox’s regression models were used to test the associa-

tion between markers of SEP and hysterectomy. All models

were run on the age-time scale with follow up in years from

birth until hysterectomy. Follow up times of women who had

not had a hysterectomy were censored at the age of com-

pleting the baseline questionnaire in the BWHHS, when

they last completed a questionnaire in the NSHD and when

completing the 2000/1 questionnaire in the Aberdeen study.

Assumptions of proportionality were assessed by examina-

tion of plots and testing for evidence of an interaction

between each SEP variable and the age-time scale. Where

there was evidence of an interaction, for education and

adult head of household social class in NSHD, a cut point

of age 44 years was selected (based on previous analyses)10

and piecewise models were fitted assuming proportional

hazards within each of the two age groups: up to 44 years and

44–57 years. Estimates are presented separately for these

two age ranges where necessary. For all other associations

in all three cohorts proportionality assumptions held.

In the first set of analyses, each marker of SEP was en-

tered into a Cox regression model individually. Multivariable

models were then used to examine whether indicators of

childhood SEP were associated with hysterectomy after ad-

justment for adult SEP variables and education. In all models

the SEP measures were entered as continuous index of

inequality scores.

RESULTS

Age-specific hysterectomy rates were highest in the

NSHD, with women between ages 40 and 50 years expe-

riencing the highest rates in all three cohorts (Table 1). In

the BWHHS and Aberdeen cohorts those with complete

data on hysterectomy who were excluded from analyses

because of incomplete SEP data had higher rates of

hysterectomy than those who were included (Table 2).

In unadjusted analyses (Table 3), women in the BWHHS

who lived in adverse socio-economic conditions in child-

hood had lower hysterectomy rates than women who lived

in better conditions. Conversely, in the NSHD and Aber-

deen cohorts adverse socio-economic conditions in child-

hood were associated with higher hysterectomy rates, with

results weaker in the NSHD. As an example, Table 4 shows

how hysterectomy rates decreased with worsening father’s

social class among women in BWHHS and increased with

worsening father’s social class in the other two cohorts.

Lower educational status was associated with reduced

rates of hysterectomy in the BWHHS. In the NSHD and

Aberdeen cohorts, lower educational status was associated

with increased rates of hysterectomy. These effects atten-

uated with age in NSHD.

Neither own nor head of household social class were

associated with hysterectomy in the BWHHS or Aberdeen

cohorts, but there was a trend in the relationship between

own occupational class and hysterectomy and in the rela-

tionship between head of household social class and hys-

terectomy up to age 44 years in NSHD, with women of

lower SEP having the highest rates of hysterectomy.

The relationships between childhood indicators of SEP and

hysterectomy were independent of adult SEP as demonstrated

Table 4. Hysterectomy rates by father’s social class in three British cohorts.

Father’s social class Hysterectomy rate per 1000 women years

BWHHS (N ¼ 3208)


NSHD (N ¼ 1394)

birth years: 1946

Aberdeen (N ¼ 3208)

birth years: 1950– 1955


I 27 4.01 (2.75, 5.85) 15 2.98 (1.80, 4.95) 5 1.26 (0.52, 3.02)

II 58 3.82 (2.95, 4.95) 45 3.21 (2.40, 4.30) 24 1.77 (1.19, 2.64)

IIINM 91 3.76 (3.06, 4.64) 49 4.13 (3.12, 5.47) 59 2.91 (2.25, 3.76)

IIIM 272 3.12 (2.77, 3.51) 104 4.33 (3.57, 5.25) 202 2.84 (2.47, 3.26)

IV 179 2.98 (2.57, 3.44) 72 4.86 (3.86, 6.12) 77 3.84 (3.09, 4.82)

V 78 3.04 (2.44, 3.81) 20 4.61 (2.98, 7.15) 81 3.69 (2.97, 4.59)

Total 705 2.73 (2.36, 3.14) 305 4.12 (3.68, 4.61) 448 2.97 (2.71, 3.26)




by the fact that the strength of the associations shown in

Table 3 were not attenuated greatly after adjustment for adult

SEP (Table 5). Additional adjustment for education attenu-

ated the size of many of the associations but the majority of

markers of childhood SEP remained associated with hyster-

ectomy in the BWHHS cohort. In the NSHD and Aberdeen

cohorts the rate of hysterectomy remained higher among

women whose father’s social class was low than among

those whose father’s social class was high even after adjust-

ment for indicators of adult SEP and education.

DISCUSSION

Adverse lifetime socio-economic conditions, especially

in childhood and as indicated by educational status, were as-

sociated with reduced rates of hysterectomy in the BWHHS

(the oldest cohort) whereas in the other two cohorts adverse

socio-economic conditions were associated with increased

rates of hysterectomy, with some suggestion of a stronger

association in the Aberdeen than the NSHD cohort. The rela-

tionships of hysterectomy with indicators of childhood SEP

were largely independent of SEP in later life. There were no

consistent or strong associations between adult SEP and hys-

terectomy rates suggesting that social factors influencing

hysterectomy are likely to be those experienced or which

develop in early life rather than those which develop later.

Previous work in the NSHD8,10,11 mostly focussed on

adult SEP and education and found an inverse gradient

between SEP and hysterectomy up to age 52 years. Now,

with information on hysterectomy in the NSHD up to age

57 years, we have examined the relationship with early life

SEP and made comparisons with two other cohorts, born in

different decades of the 20th century. The clearly defined

Table 5. Adjusted hazards ratios for hysterectomy comparing worst socio-economic condition with best for each childhood indicator of socio-economic

position and education in three British cohorts using RII.

BWHHS (1920– 40) N ¼¼¼ 3208,


NSHD (1946) N ¼¼¼ 1394, hysterectomy ¼¼¼ 305 Aberdeen (1950– 55) N ¼¼¼ 3208,


Adjusted for

adult SEP,a

HR (95% CI)

Adjusted for

adult SEP and

education,b

HR (95% CI)

Adjusted for

adult SEP,a

HR (95% CI)

Adjusted for

adult SEP and

education,b

HR (95% CI)

Adjusted for

adult SEP,a

HR (95% CI)

Adjusted for

adult SEP and

education,b

HR (95% CI)

Father’s

social class

0.74 (0.56, 0.97) 0.77 (0.58, 0.99) 1.62 (1.06, 2.47) 1.64 (1.03, 2.59) 2.07 (1.46, 2.90) 1.69 (1.19, 2.44)

P 0.02 0.05 0.02 0.04 <0.001 0.004

Level of

crowding

– – 1.12 (0.70, 1.80) 1.07 (0.66, 1.75) 1.56 (1.09, 2.25) 1.36 (0.96, 1.96)

P – – 0.64 0.78 0.03 0.09

Family access

to a car

0.62 (0.43, 0.91) 0.66 (0.44, 0.94) – – 1.75 (1.20, 2.56) 1.61 (1.10, 2.35)

P 0.02 0.03 – – 0.004 0.01

Running

hot water

0.63 (0.47, 0.90) 0.66 (0.48, 0.93) 1.23 (0.72, 2.10) 1.19 (0.69, 2.06) – –

P 0.007 0.01 0.46 0.53 – –

Bathroom

in house

0.62 (0.45, 0.86) 0.65 (0.47, 0.89) 0.84 (0.47, 1.49) 0.80 (0.44, 1.44) – –

P <0.001 0.002 0.54 0.46 – –

Own bedroom 0.70 (0.40, 1.26) 0.76 (0.43, 1.31) 1.28 (0.81, 2.01) 1.24 (0.78, 1.97) – –

P 0.25 0.31 0.29 0.37 – –

Family housing – – 1.43 (0.87, 2.35) 1.39 (0.83, 2.33) 1.68 (1.02, 2.81) 1.27 (0.76, 2.12)

P – – 0.16 0.21 0.03 0.40

Age at

leaving school

0.70 (0.51, 0.94) – Up to 44 yearsy

1.85 (0.88, 3.87)

44–57 years*

0.82 (0.42, 1.62)

– 2.44 (1.66, 3.49) –

P 0.03 – 0.10 0.57 – <0.001 –

Highest

educational

attainment

– – 2.31 (1.15, 4.63) 0.70 (0.37, 1.35) – 2.12 (1.42, 2.93) –

P – – 0.02 0.29 – <0.001 –


Numbers in parentheses after the study name indicate the birth year(s) of the cohort.a Adjusted for all indicators of adult SEP for each cohort as listed in Table 3 (where cohorts have both own and household adult occupational social class

adjustment has been made just for own social class and other adult measures of SEP).b Adjusted for all indicators of adult SEP (as described above) and in addition education (for BWHHS age at leaving full-time education only; for other two

cohorts age at leaving full-time education and highest educational attainment). In the analyses of NSHD, allowance has been made for the attenuation of the

effect of education with age in adjusting for these variables.* Age 44– 57 years (N ¼ 1219, no. of hysterectomies ¼ 164).y Up to age 44 years (N ¼ 1394, no. of hysterectomies ¼ 141).



times of birth make it possible to identify cohort specific

effects which other studies, grouping together women span-

ning generations and birth decades or focussing on only one

birth cohort, are unable to do.

In the studies that have explicitly examined the relation-

ship between lifetime SEP and hysterectomy9,11 one found

evidence of an inverse association11 between early life SEP

and hysterectomy while the other, a US study, found none.9

The US results9 may not apply to a UK population where

hysterectomy rates are lower and organisation of health ser-

vice provision is different. UK studies2,15,27 other than the

NSHD have found limited evidence in support of the influ-

ence of adult SEP on hysterectomy. However, these studies

rely on self-reported adult social class data only which may

not capture fully the aspects of SEP which are influential.

In all three cohorts hysterectomy was self-reported, but a

previous study found that using self-reported hysterectomy

did not bias analyses of associated factors when compared

with hospital record information.28

In the BWHHS and Aberdeen study most SEP data were

based on retrospective self-reports with recall of up to 70

years in BWHHS. In the Aberdeen cohort data on father’s

social class from obstetric records were available and were

also self-reported in adulthood. For the main analyses

father’s social class from obstetric records was used be-

cause this is likely to be more accurate. Any effects of

misclassification due to self-report were assessed using a

sensitivity analysis, which examined the association of adult

report of father’s social class with hysterectomy. Results

using self-reported father’s social class [unadjusted HR (95%

CI): 2.03 (1.42, 2.92); and HR adjusted for adult SEP and

education: 1.68 (1.18, 2.43)] were very similar to those

using father’s social class from obstetric records shown in

Tables 3 and 5. The distributions of father’s social class

based on adult retrospective report in the BWHHS are

consistent with known distributions of men in the age range

20–55 years in the 1931 census (15% of men were in social

classes I and II and 38% were in social classes IV and V

compared with 12% and 43%, respectively, for adult

retrospective report in BWHHS).29 Despite the potentially

greater level of imprecision in measurements of other

indicators of childhood SEP compared with father’s social

class in Aberdeen, results for all indicators are consistent.

The data for BWHHS and Aberdeen were collected

using a cross sectional design and so only survivors at least

to age 60 and 45 years, respectively, can be included, which

may have introduced survivor bias. This may lead to under-

estimation of effect size.

Women with missing information on hysterectomy sta-

tus and incomplete SEP data were excluded from analyses.

In the BWHHS and Aberdeen cohorts women excluded due

to incomplete SEP data had higher rates of hysterectomy

than others (Table 2). However, in univariable analyses

including all those with data on the variable of interest and

hysterectomy (results not shown), there was little difference

in the size or direction of effects compared with the findings

presented, suggesting that the results are unlikely to be

markedly influenced by selection bias due to exclusions.

This study suggests that early life SEP and education are

predictors of hysterectomy rates independent of adult SEP

and that this relationship appears to have changed direction

over time. These discordant results have been found despite

the fact that the BWHHS and Aberdeen studies collected data

using similar methods and hence face similar limitations.

Differences in directions of associations could be an arte-

fact of differences between the three data sets. Follow up of

women in each cohort is to different ages. In analyses with

follow up restricted to age 45 years in all cohorts (results not

shown), the directions of association and differences between

cohorts were similar to those presented. It remains possible

that the effect of SEP could attenuate with age in Aberdeen,

if follow up continued, as it does in NSHD.

Although SEP was measured slightly differently in each

study, sensitivity analyses in the Aberdeen study and com-

parisons of self-report with 1931 census data in the BWHHS,

reported above, suggest that there is no major bias due to

retrospective self-report and so these differences are unlikely

to fully explain our findings.

Markers of SEP change over time and indicators such as

running hot water and access to an indoor bathroom are

more a marker of severe deprivation in younger compared

with older cohorts. The relative index of inequality

accounts for such cohort differences in the proportion of

women in the most deprived category of an SEP indicator.

Using RII rather than socio-economic variables in their

original categories could potentially result in differences in

findings. However, in this study, when survival analyses in

which socio-economic variables were entered into models

as categories (results not shown) the findings were similar

and the conclusions remained the same as those presented.

If the results are not simply an artefact, what other ex-

planations might account for our findings? The differences

could be partially explained by differences between places.

The NSHD and BWHHS cohorts included women from all

areas of Britain in their sampling frame whereas women in

the Aberdeen cohort were sampled from one town in Scot-

land. There are characteristics of populations in specific

geographical areas that could potentially influence hyster-

ectomy rates and the social gradients in these. These char-

acteristics include religion, potentially associated factors

such as family size and contraceptive use and opinions and

preferences about the management of menstruation and men-

strual disorders especially after reaching desired family size.

Such influences would not be expected to be seen in larger

national populations, which are more heterogeneous with

respect to these characteristics. One characteristic of Aber-

deen, although possibly not the only important factor, is

that from the late 1940s the city has been a centre of ex-

cellence with respect to obstetrics and gynaecology. Dugald

Baird, influential in Aberdeen, often expressed his posi-

tive opinions of postpartum sterilisation as a means of birth

control especially for women of lower SEP.30 – 32 It is thus



possible that a culture of acceptability of gynaecological

procedures existed in the mothers of the Aberdeen cohort,

particularly those in the most adverse SEP and that this

persisted among the participants. While this explains why

the strength of association may be greater in Aberdeen, it

does not explain the difference in direction of effect be-

tween BWHHS and the other two cohorts.

Changes over time are a plausible explanation of our

findings. These include the introduction of the NHS in

1948, changes in access to medical care, attitudes and ac-

countability of doctors, availability of oral contraceptives,

changes in average number of children and timing of birth,

increasing levels of obesity, changes in the acceptability of

some conditions (e.g. menstrual disorders) as appropriate

indications for hysterectomy and the introduction of alter-

natives to hysterectomy and patterns in uptake of these.

Some of these factors such as patterns of childbearing are

likely to have become more socially differentiated, while

others such as access to medical care will have become less

socially graded.

The paucity of information regarding trends in rates and

reasons for hysterectomy in the UK makes it difficult to

establish which of the above factors influence differences

in the relationship between SEP and hysterectomy between

cohorts over time. However, it is likely that the introduction

of the NHS and resultant changes to medical access and

opinions of medical care and the changing patterns of

childbearing are influential.

Whether the mechanism responsible for the differences

in the association between education, childhood SEP and

hysterectomy in these three cohorts is related to age, geog-

raphy or time, our results demonstrate the dynamic nature

of the relationship between SEP and health outcomes.21

When considering policy aimed at social inequalities in

health and when controlling for SEP in epidemiological

studies, it is important to appreciate the possibility of marked

changes over time and between places in the direction and

strength of effect of SEP on health outcomes. Researchers

should not automatically assume that results from one co-

hort are generalisable to cohorts from different periods or

locations.

Acknowledgments

BWHHS: We thank all of the women who participated

in the study, their general practitioners who supported data

collection and Rita Patel, Carol Bedford, Alison Emerton,

Nicola Frecknall, Karen Jones, Mark Taylor, Simone Wat-

son and Katherine Wornell for collecting and entering data.

MRC National Survey of Health and Development: We

thank all the women who have participated in the study

throughout their lives.

Aberdeen Children of the 1950s cohort: We are very

grateful to Raymond Illsley for providing us with the data

from the Aberdeen Child Development Survey and for his

advice about the study. Graeme Ford played a crucial role

in identifying individual cohort members and in helping

us initiate the process of revitalisation. Sally Macintyre,

Doris Campbell, George Davey Smith, Marion Hall, Bianca

de Stavola, Susan Morton, David Batty, David Godden,

Diana Kuh, Debbie Lawlor, Glyn Lewis and Viveca Ost-

berg collaborated with David Leon to revitalise the cohort.

Heather Clark managed the study at the Dugald Baird Cen-

tre, Aberdeen with the assistance of Margaret Beveridge.

Finally, we thank the study participants who responded to a

mailed questionnaire 40 years after the original survey was

completed.

Competing interestsNone declared.

ContributionsDK and DA Lawlor conceived the idea for the study and

along with RC developed the objectives and design of the

study. DA Lawlor and RC ran the analyses and act as

guarantors for the paper. RC drafted the paper and all

authors commented on this and contributed to the final

version.

FundingThe British Women’s Heart and Health Study is funded

by the UK Department of Health. The MRC National

Survey of Health and Development is funded by the

Medical Research Council. The Aberdeen Children of the

1950s Study was funded as a component project of a

Medical Research Council Co-operative Group on Life-

course and trans-generational influences on disease risk.

RC is funded by a Medical Research Council research

studentship. DA Lawlor is funded by a UK Department of

Health Career Scientist Award. RH, MEJW and DK are

funded by the Medical Research Council. The views ex-

pressed in the paper are those of the authors and not nec-

essarily those of any funding body.

References

1. Maresh MJA, Metcalfe MA, McPherson K, et al. The VALUE

national hysterectomy study: description of the patients and their

surgery. Br J Obstet Gynaecol 2002;109:302 –312.

2. Vessey MP, Villard-Mackintosh L, McPherson K, Coulter A, Yeates

D. The epidemiology of hysterectomy: findings in a large cohort

study. Br J Obstet Gynaecol 1992;99:402 –407.

3. Kasper AS. Hysterectomy as social process. Women Health 1985;10:

109– 127.

4. Hardy R, Kuh D. Menopause and gynaecological disorders: a life course

perspective. In: Kuh D, Hardy R, editors. A Life Course Approach to

Women’s Health. Oxford: Oxford University Press, 2002:64– 85.

5. Dharmalingam A, Pool I, Dickson J. Biosocial determinants of hys-

terectomy in New Zealand. Am J Public Health 2000;90:1455– 1458.

6. Harlow BL, Barbieri RL. Influence of education on risk of hys-

terectomy before age 45 years. Am J Epidemiol 1999;150:843 – 847.

7. Kjerulff K, Langenberg P, Guzinski G. The socioeconomic correlates

of hysterectomies in the United States. Am J Public Health 1993;

83:106 –108.



8. Kuh D, Stirling S. Socioeconomic variation in admission for diseases

of female genital system and breast in a national cohort aged 15– 43.

BMJ 1995;311:840– 843.

9. Marks NF, Shinberg DS. Socioeconomic differences in hysterectomy:

the Wisconsin longitudinal study. Am J Public Health 1997;87:1507–

1514.

10. Marshall SF, Hardy RJ, Kuh D. Socioeconomic variation in hys-

terectomy up to age 52: national, population based, prospective cohort

study. BMJ 2000;320:1579.

11. Marshall S, Hardy R, Kuh D. Changes with age in the socioeconomic

gradient in hysterectomy: findings from a national cohort. J Epidemiol

Community Health 2000;54:782.

12. Meilahn EN, Matthews KA, Egeland G, Kelsey SF. Characteristics of

women with hysterectomy. Maturitas 1989;11:319– 329.

13. Santow G, Bracher M. Correlates of hysterectomy in Australia. Soc

Sci Med 1992;34:929– 942.

14. Settnes A, Jorgensen T. Hysterectomy in a Danish cohort. Prevalence,

incidence and socio-demographic characteristics. Acta Obstet Gynecol

Scand 1996;75:274 –280.

15. Coulter A, McPherson K. Socioeconomic variations in the use of

common surgical operations. BMJ 1985;291:183– 187.

16. Ross CE, Wu C. The links between education and health. Am Sociol

Rev 1995;60:719– 745.

17. Kuh D, Power C, Blane D, Bartley M. Socioeconomic pathways

between childhood and adult health. In: Kuh D, Ben-Shlomo Y,

editors. A Life Course Approach to Chronic Disease Epidemiology.

Oxford: Oxford University Press, 2004:371–395.

18. Sacker A, Schoon I, Bartley M. Social inequality in educational

achievement and psychosocial adjustment throughout childhood: mag-

nitude and mechanisms. Soc Sci Med 2002;55:863 –880.

19. Lake J, Power C, Cole T. Significant social class gradient in menstrual

disorders. BMJ 1995;311:1503.

20. Power C, Parsons T. Overweight and obesity from a life course

perspective. In: Kuh D, Hardy R, editors. A Life Course Approach to

Women’s Health. Oxford: Oxford University Press, 2002:304 – 328.

21. Davey Smith G, Lynch J. Life course approaches to socioeconomic

differentials in health. In: Kuh D, Ben-Shlomo Y, editors. A Life

Course Approach to Chronic Disease Epidemiology. Oxford: Oxford

University Press, 2004:77–115.

22. Lawlor DA, Bedford C, Taylor M, Ebrahim S. Geographical variation

in cardiovascular disease, risk factors, and their control in older

women: British Women’s Heart and Health Study. J Epidemiol

Community Health 2003;57:134– 140.

23. Lawlor DA, Ebrahim S, Davey Smith G. The association of socio-

economic position across the life course and age at menopause: the

British Women’s Heart and Health Study. Br J Obstet Gynaecol

2003;110:1078– 1087.

24. Kuh D, Hardy R. Women’s health in midlife: findings from a British

birth cohort study. J Br Menopause Soc 2003;9:55– 60.

25. Batty GD, Morton SMB, Campbell D, et al. The Aberdeen Children of

the 1950s cohort study: background, methods, and follow-up infor-

mation on a new resource for the study of life course and inter-

generational influences on health. Paediatr Perinat Epidemiol 2004;

18:221– 239.

26. Mackenbach JP, Kunst AE. Measuring the magnitude of socio-economic

inequalities in health: an overview of available measures illustrated with

two examples from Europe. Soc Sci Med 1997;44:757–771.

27. Kennedy TM, Jones RH. The epidemiology of hysterectomy and irri-

table bowel syndrome in a UK population. Int J Clin Prac 2000;54:

647– 650.

28. Brett KM, Madans JH. Hysterectomy use: the correspondence between

self-reports and hospital records. Am J Public Health 1994;84:1653–

1655.

29. Registrar General. The Registrar-General’s Decennial Supplement,

England and Wales 1931: Part IIa. Occupational Mortality. London:

HMSO, 1934.

30. Thompson B, Baird D. Follow-up of 186 sterilised women. Lancet

1968;I:1023– 1027.

31. Baird D. Sterilization and therapeutic abortion in Aberdeen. Br J

Psychiatry 1967;113:701– 709.

32. Baird D. The Galton Lecture 1970: the obstetrician and society.

J Biosoc Sci 1971;(Suppl 3):93 – 111.

Accepted 13 February 2005



socio-economic position across the life course and hysterectomy in three british cohorts: a...

Documents