1 Chief Scientist Office Form 4 Final report form CSO reference number: Please complete this form in Verdana 10 point font size Project title: Investigation of observed increases in breastfeeding rates in deprived areas Start date: 1 July 2010 Finish date: 30th June 2011 Investigators: Mr Bruce Whyte Glasgow Centre for Population Health Dr Pauline M Craig NHS Health Scotland (formerly Glasgow Centre for Population Health) Dr James Chalmers Information Services, NHS National Services Scotland Structure of final report: 1. Summary This study had two aims. It sought to explain increases in breastfeeding in selected neighbourhoods in Glasgow and, it aimed to provide new analyses of breastfeeding trends across Scotland through the creation of a new child and maternal linked dataset. A contextual analysis of the Glasgow neighbourhoods identified important compositional changes in the population of each area that are likely to have affected trends in breastfeeding. However, in two of the areas, Greater Gorbals and Govanhill, there were rises in breastfeeding among British born mothers and mid-aged and older mothers that, additionally, may reflect the potential positive influence of a ten year ‘peer support’ breastfeeding initiative. Through this study a new child and maternal linked dataset has been created. The analysis of this resource has highlighted the influence of a range of demographic, socio- economic, lifestyle-related, birth and neo-natal related factors on types and duration of infant feeding. Older mothers, mothers living in less deprived areas, those in higher status jobs, those married or living with a partner, mothers who did not smoke, mothers who were not obese or underweight and those born outside the British Isles or with a non-British ethnic origin had higher levels of exclusive breastfeeding. Many of the factors that predict likelihood of breastfeeding are known from the existing literature, but have not been described before for a Scottish population. This new linked dataset is a rich resource that could be developed and analysed further to improve understanding of infant feeding in Scotland and, for example, its likely impact on infant health. 2. Original aims Our aims were to: explain increases in breastfeeding rates observed between 1997 and 2009 i in five ii selected neighbourhoods within NHS Greater Glasgow and Clyde. i The period of analysis has been extended from 1997-2006 proposed in the original SCPHP application to 1997-2009 due to access to more recent linked and administrative data. ii Originally there were 4 selected neighbourhoods. However to bring greater clarity to the contextual analysis Greater Gorbals and Govanhill, which were previously combined, were split and analysed separately.
21
Embed
Investigation of observed increases in breastfeeding rates in deprived areas
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
Chief Scientist Office Form 4
Final report
form
CSO reference number:
Please complete this form in Verdana 10 point font size
Project title: Investigation of observed increases in breastfeeding rates in
deprived areas
Start date: 1 July 2010 Finish date: 30th June 2011
Investigators:
Mr Bruce Whyte Glasgow Centre for Population Health
Dr Pauline M Craig NHS Health Scotland (formerly Glasgow
Centre for Population Health)
Dr James Chalmers Information Services, NHS National
Services Scotland
Structure of final report:
1. Summary
This study had two aims. It sought to explain increases in breastfeeding in selected
neighbourhoods in Glasgow and, it aimed to provide new analyses of breastfeeding
trends across Scotland through the creation of a new child and maternal linked dataset.
A contextual analysis of the Glasgow neighbourhoods identified important compositional
changes in the population of each area that are likely to have affected trends in
breastfeeding. However, in two of the areas, Greater Gorbals and Govanhill, there were
rises in breastfeeding among British born mothers and mid-aged and older mothers that,
additionally, may reflect the potential positive influence of a ten year ‘peer support’
breastfeeding initiative.
Through this study a new child and maternal linked dataset has been created. The
analysis of this resource has highlighted the influence of a range of demographic, socio-
economic, lifestyle-related, birth and neo-natal related factors on types and duration of
infant feeding. Older mothers, mothers living in less deprived areas, those in higher
status jobs, those married or living with a partner, mothers who did not smoke, mothers
who were not obese or underweight and those born outside the British Isles or with a
non-British ethnic origin had higher levels of exclusive breastfeeding.
Many of the factors that predict likelihood of breastfeeding are known from the existing
literature, but have not been described before for a Scottish population. This new linked
dataset is a rich resource that could be developed and analysed further to improve
understanding of infant feeding in Scotland and, for example, its likely impact on infant
health.
2. Original aims
Our aims were to:
explain increases in breastfeeding rates observed between 1997 and 2009i in fiveii
selected neighbourhoods within NHS Greater Glasgow and Clyde.
i The period of analysis has been extended from 1997-2006 proposed in the original SCPHP application to
1997-2009 due to access to more recent linked and administrative data. ii Originally there were 4 selected neighbourhoods. However to bring greater clarity to the contextual analysis
Greater Gorbals and Govanhill, which were previously combined, were split and analysed separately.
2
investigate new ways of analysing breastfeeding trends across Scotland making
use of linked data.
We have attempted to answer four specific research questions.
Q1 Has the population composition of the neighbourhoods changed between 1997 and
2009 in ways that might impact on breastfeeding rates, such as changes in
deprivation, household structure, housing tenure and population movement?
Q2 What are the best data sources and analytical methods to use to understand
breastfeeding trends in Greater Glasgow and Clyde?
Q3 What is the impact of maternal age, previous breastfeeding history, deprivation and
ethnic background on likelihood to breastfeed and length of breastfeeding in these
neighbourhoods and in Scotland as a whole?
Q4 Can we improve on the current Scottish measures of breastfeeding trends by
employing alternative sources such as those used in this project?
3. Methodology
In order to answer the first research question we undertook a contextual analysis of
selected neighbourhoods in Greater Glasgowiii by gathering a range of demographic,
housing and socioeconomic data from national and local sources. Trends and patterns in
these data were analysed for the period 1997-2009 and related to changes in patterns of
infant feeding over the same period. A draft report of this work has been completed and
a summary of the findings is included in the results section of this report.
To address the second research question we initially undertook a detailed quality
assurance of a child health surveillance dataset iv. This identified strengths, weaknesses
and gaps in this data source prior to linkage and subsequent analysis. A draft report of
this work has been written and will be published on the GCPH website.
Our next step was to plan and undertake a linkage to create a new child and maternal
linked dataset for Scotland. An application was made to NHS National Services
Scotland’s Privacy Advisory Committee requesting a linkage of child and maternal data
schemes in February 2010. This was approved in May 2010.
Working in collaboration with ISD Scotland, a record linkage applying probability
matching techniques (Gill and Baldwin (1987), Newcombe et al. (1989), Kendrick and
Clarke (1993), Kendrick (1997), Blakely and Salmond (2002), Mason and Tu (2008))
was undertaken to create a new child/maternal linked dataset . This new dataset
combines General Registrar of Scotland (GROSv) birth records, maternal obstetric
discharge records (SMR02), Neonatal discharge records (SMR11), Scottish Birth Records
(SBR), Scottish Still Birth and Infant Death Survey (SSBID) records and Child Health
Surveillance Programme – Pre-school System (CHSP-PS) records. The dataset covers all
Additionally, initially only South Lenzie was to be included, but during data collection it was decided to widen
the study area to the whole of Lenzie, by including North Lenzie – which is similar socio-economically and had
high rates of breastfeeding, although no notable upward trend - in order to provide more robust figures for a
larger population. iii The neighbourhoods were Govanhill, Greater Gorbals (both in South East Glasgow), Sighthill, Roystonhill &
Germiston (in North Glasgow), Lenzie (in East Dunbartonshire) and, Temple & Anniesland (in West Glasgow)
These neighbourhoods were selected because health profiles (Whyte B, 2008, GCPH) had identified increases in
rates of any breastfeeding in these areas between 1997-99 and 2004-06. iv CHSP-PS - Child Health Surveillance Programme Pre-School System v Now National Records of Scotland (NRS)
3
births in Scotland in the period 1997-2009. Figure 1 provides a summary of the
linkagevi.
Figure 1 Linkage Summary
Additional variables were appended to the new child and maternal linked dataset after
linkage:
a name classification algorithm, Onomap (Mateos, Webber and Longley (2007)) was
applied to the dataset to create a marker of maternal and paternal ethnic and
cultural background
geographical and deprivation variables were added
the dataset provided to the research team was anonymised and a random ID number was created for every child
The quality assurance and linkage, alongside the analyses described in the next section,
help to answer the question we have posed regarding the best data sources and
methods for understanding breastfeeding trends in Greater Glasgow and Clyde.
The third research question was aimed at discovering the impact of a range of factors on
likelihood and duration of breastfeeding. There were several parts to our approach.
vi
Further details of the linkage methodology are available in a separate report which will be published on the
GCPH website.
613,900
Child Health Records
SMR11 records
GRO birth records
SBR
SSBID
GRO SBID
CHI database
Each child health record is linked against the maternal and neonatal database and can find a best match against any record type. The vertical arrow shows that once a best match is found to any record type, the record can indirectly link to all other record types within that mother/baby patient group
All record types within a mother/baby group are permanently linked together in the maternal and neonatal linked file
Initial linkage of GRO birth records to CHI database allows CHI number to be added to each birth record in the linked maternity file.
Maternal & Neonatal
Linked Database
SMR02 records
4
A detailed literature review was undertaken. The review looked at the policy context
relating to infant feeding internationally and in Scotland, investigated issues of
measurement and sought evidence from the literature on factors that could influence
breastfeeding.
The main analytic phase of the project involved using the newly created child and
maternal linked dataset to undertake a series of analyses of infant feeding trends in
relation to a range of socio-economic, child-related and maternal factors (or predictive
variables). Analyses were initially carried out on a univariate basis to quantify the
relationship between predictive variables and different types of infant feeding - exclusive
breastfeeding, mixed feeding and bottle feeding. The main source of information on
infant feeding has been CHSP-PS data, which has allowed analysis of infant feeding at
different stages from birth through to the 6 to 8 week review.
Subsequently, these demographic, maternal and infant health variables were input into a
multivariate model and used to predict breastfeeding using manual and step-wise
(forwards) logistic regression (STATA v11).
An answer to research question 4 – ‘Can we improve on the current Scottish measures of
breastfeeding trends by employing alternative sources such as those used in this
project?’ – is given in the discussion and conclusions based on the overall findings from
the project.
4. Results
Contextual analysis
In this analysis we have examined both contextual factors (e.g. house building,
deprivation) and compositional factors (e.g. ethnicity, age of mothers) in the study
neighbourhoods and how these have changed over the period 1997-2009. Breast-
feeding rates are likely to be affected by a range of factors, including ethnicity, culture,
age, education, health, economic status, social networks, etc. Our methods do not
establish causation but do allow explanations of trends to be postulated that are firmly
rooted in relevant data.
Four areas included in this study were included in a Glasgow Breastfeeding Initiativevii
which was initiated via GP services between 1997 and 2000 and continued until 2007viii.
The project aimed to increase the duration and level of breastfeeding at 6 to 8 weeks in
participating localities. All of the selected neighbourhoods had been identified from
GCPH’s health profiles (Whyte B, 2008) as showing increases in breastfeeding at 6 to 8
weeks over the last decade; with the largest and most sustained increases occurring in
Greater Gorbals and Govanhill.
From the evidence gathered within the contextual analysis, summarised in Appendix 1, it
seems likely that in Govanhill changes in population composition – more mothers born
outside Britain and more births to older mothers – possibly coupled with the impact of
the peer support initiative on some maternal groups (e.g. British born mothers), have
contributed to increases in breastfeeding.
In Greater Gorbals, it is possible that breastfeeding trends have been influenced by a
combination of compositional changes and other factors – such as the Greater Glasgow
‘peer support’ breastfeeding initiative. For example, breastfeeding at birth among British
born mothers rose from 38% in 2001/2003 to 42% in 2007/2009, and at first visit
increased from 27% in 1997/1999 to 43% in 2007/2009ix. There was also evidence of a
vii The Glasgow Breastfeeding initiative was set up in parts of Greater Glasgow – mainly Gorbals, Govanhill and the Riverside LHCC. The initiative was set up in Greater Gorbals/Govanhill in 1997, in Sighthill/Roystonhill (but only for mothers registered with a Townhead GP) in 1997 and in Temple and Anniesland in 2000 as part of the Riverside LHCC. Linda Wolfson (Infant Feeding Coordinator, NHS Greater Glasgow and Clyde) personal correspondence viii Information provided by Marion McPhillips, Gorbals Health Centre ix 3 year rolling rates used to adjust for small numbers
5
sustained increase in levels of breastfeeding among mid-aged mothers (25-34 year old)
and older mothers (35 years and over) at birth – see Figure 2 - and at first visit.
Figure 2 Breastfeeding trends at birth by maternal age, Greater Gorbals
Greater Gorbals: Any breastfeeding (exclusive and mixed) at birth by maternal age
In the three other neighbourhoods, a variety of socio-economic and breastfeeding trends
were identified. In Sighthill, Roystonhill and Germiston, it is likely that increasing
population diversity, older mothers and a less deprived population have contributed to
the rise in breastfeeding in the last decade. There has also been a notable increase in
mixed feeding, particularly among non-British born mothers.
In Temple and Anniesland, population composition has remained relatively stable,
albeit the population has become slightly more ethnically diverse. Exclusive
breastfeeding and mixed feeding have increased, but only marginally. There have been
increases in any breastfeeding among older mothers.
In Lenzie, the most affluent of the selected neighbourhoods, breastfeeding levels have
remained high and have not changed greatly in the last ten years. The population,
which is largely of British origin, has remained relatively stable.
Linkage
The child and maternal linked dataset created for this study was the key resource for
analysing infant feeding both nationally and locally.
The final child and maternal linked dataset contained 731,611 records relating to
children born in Scotland between 1997 and 2009, including 613,900 child health
records with links to other records. This difference is because the child health data set
does not have national coverage (currently 85%)x.
Figure 3 provides a diagrammatic summary of the main outcome and predictor variables
that have been analysed from the final linked dataset.
x There are currently 12 participating Boards in the Child Health Surveillance Programme Pre -School System covering 85% of Scotland’s pre-school population.
6
Figure 3 Overview of data schemes
Literature Reviewxi
There is considerable evidence and policy support for breastfeeding as the main source
of nutrition in the first 6 months of life. The practice of infant feeding is greatly affected
by the socio-cultural and demographic characteristics of the mother. Influential factors
include age, education, deprivation, maternal physical and psychological health,
maternal and societal beliefs/ attitudes, social networks and experience of health care.
Evidence specific to the Scottish context is relatively sparse, although some research has
been carried out using the Infant Feeding Survey (Bolling K et al (2007)), Growing Up in
Scotland (Bradshaw P at al (2008)) and various ISD held datasets.
It was noted that there are still information gaps (particularly in Scotland) on the feeding
practices and effective interventions in deprived areas; the role of supportive
environments; media campaigns and non – health sector interventions in promoting
breastfeeding.
Univariate analyses
Our analysis has highlighted the influence of a range of demographic, socio-economic,
lifestyle-related, birth and neo-natal related factors on types and duration of infant
feeding. Across Scotland, older mothers, mothers living in less deprived areas, those in
higher status jobs, those married or living with a partner, mothers who did not smoke,
mothers who were not obese or underweight and those born outside the British Isles or
with a non-British ethnic origin had higher levels of exclusive breastfeeding.
In contrast, bottle feeding correlates with almost the polar opposite of these
characteristics. Additionally, mode of delivery, parity, length of maternal stay in hospital,
gestation and infant health also clearly influenced type and duration of infant feeding.
xi The literature review is being reviewed and will be published on the GCPH website as a working paper
7
The following graphs illustrate a sample of these findings in relation to ethnic origin,
marital status and maternal weight.
Figure 4, based on Scottish data, illustrates an expected decline in exclusive
breastfeeding with time from birth but also that levels of exclusive breastfeeding were
higher among non-British born and non-British originxii mothers and in contrast were
lowest among mothers who were British born and of British origin.
Figure 4 Breastfeeding trends by ethnicity
Scotland: Breastfeeding trends (exclusive) and maternal ethnicity and place of
xvii First visit limited to infants reviewed within 6 to 30 days from birth between 2003 and 2009 xviii 6 to 8 week review limited to infants reviewed within 35 – 70 days from birth between 2003 and 2009
10
Demographic/maternal/infant factors % n Odds ratio (95% CI)
Summary findings from contextual analysis of selected neighbourhoods in
Greater Glasgow
In the contextual analysis demographic, socio-economic, housing and infant feeding
trends were investigated in five neighbourhoodsxxiii in Greater Glasgow - Govanhill,
Greater Gorbals (both in South East Glasgow), Sighthill, Roystonhill & Germiston (in
North Glasgow), Lenzie (in East Dunbartonshire) and, Temple & Anniesland (in West
Glasgow).
Data from a variety of administrative sources were accessed for this analysis: small area
population estimates (Scottish Government); births, including mothers’ country of birth
(GROS Birth Record); housing completions/demolitions, tenure and type, household
structure and ethnicity of children in education (Glasgow City Council); worklessness
(DWP); deprivation (SIMD); house prices (SNS); and asylum seekers (CoSLA).
Trends and patterns in these data were analysed over the period 1997-2009 and related
to changes in patterns of infant feeding over the same period. Trends were analysed
using annualised data. Not all of the data were available across the full period and this
is noted in the text and footnotes. In this regard it is to worth noting that the CHSP_PS
system only began to record infant feeding at birth from 2000 and data are only valid
from 2001. However, infant feeding data reported for the first review and 6 to 8 week
review are available back to 1997.
It is also relevant to note that four of the study areas - Govanhill, Greater Gorbals,
Sighthill, Roystonhill & Germiston and Temple & Anniesland - were included within a
Greater Glasgow Breastfeeding Initiativexxiv implemented between 1997 and 2000 and
continued until 2007xxv. The aim of the initiative was to increase the duration and level
of breastfeeding at 6 to 8 weeks in the participat ing localities. The strategy focused on
providing appropriate training and resources and monitoring progress for staff and
recruited volunteers supporting women ante-natally and postpartum. A key aim of the
contextual analysis has been to assess the potential impact of this scheme on
breastfeeding.
This summary is drawn from a longer report which will be published on the GCPH
website in the summer of 2011.
Govanhill
Overview of the area
Situated in south east Glasgow, Govanhill has a population of 15,839 (2009) and is
made up of 18 datazones. It has a rich multicultural identity with a history of migration
from Ireland, Pakistan and more recently, Poland and Slovakiaxxvi. In 2008, the
xxiii
These neighbourhoods were selected because health profiles (Whyte B, 2008, GCPH) had identified
increases in rates of any breastfeeding in these areas between 1997-99 and 2004-06. NB Initially only South
Lenzie was included, but during data collection it was decided to widen the study area to the whole of Lenzie,
by including North Lenzie – which is similar socio-economically and had high rates of breastfeeding, although
no notable upward trend - in order to provide more robust figures for a larger population.
xxiv The Glasgow Breastfeeding initiative was set up in parts of Greater Glasgow – mainly Gorbals, Govanhill and
the Riverside LHCC. The initiative was set up in Greater Gorbals/Govanhill in 1997, in Sighthill/Roystonhill (but
only for mothers registered with a Townhead GP) in 1997 and in Temple and Anniesland in 2000 as part of the
Riverside LHCC. Linda Wolfson (Infant Feeding Coordinator, NHS Greater Glasgow and Clyde) personal
correspondence
xxv Information provided by Marion McPhillips, Gorbals Health Centre xxvihttp://library.nhsggc.org.uk/mediaAssets/CHP%20South%20East%20Glasgow/SE%20Glasgow %20Health%20Improvement
%20Plan%202009-2012%20Public%20Summary.pdf
17
minority ethnic population was estimated to comprise about a third of the Govahill
neighbourhood (29%). In 2010, an estimated 272 asylum seekersxxvii lived in the area.
Residents of Govanhill have a lower life expectancy (males: 71 years; females 77 years)
compared to the overall Glasgow/Scottish average (72 years/74.5years and 78
years/79.5 years respectively)xxviii. It is relatively deprived with 46% of the population
living within the most deprived quintile – SIMD 1 (SIMD,2009).
Demographic, socio-economic and housing trends
Over the period from 1997-2009, Govanhill’s population rose by 12% to 15,839, due to
inward migration and natural increase. Govanhill had a birth rate of 16 births per 1000
population in 2009 (Scotland – 11 births per 1000 population). Births to teenage
mothers decreased from 56 births to 28 births per 1000 between 1997 and 2009 and
births to older mothers increased (e.g. the birth rates in mothers aged between 30 and
34 years rose from 66 births to 93 births per 1000 between 1997 and 2009. About half
of the births were to mothers living within the most deprived quintile.
Overall, 13% of the mothers in Govanhill smoked at first visit between 2001 and 2009,
compared to 21% across Scotland and the proportion mothers smoking declined over the
study period, similar to the trends observed in Greater Glasgow and Clyde and Scotland.
The proportion of the population from an ethnic minority population was estimated to
have risen to 29% by 2008 (from 19% in 2001). The proportion of mothers of non-
British birth increased from 26% in 1997 to 57% in 2009. While Govanhill has remained
relatively deprived, there was a drop in employment deprivation but a slight increase in
income deprivation. There was an increase in the proportion of the population privately
renting and a decrease in owner-occupation. Average house prices more than doubled
between 1997 and 2008 and house sales increased by 50%.
Breastfeeding trends
The pattern of infant feeding in Govanhill changed subtly between 2001 and 2009 at
different points after birth. While there was no overall change in any breastfeeding at
birth (~73%), mixed feeding increased to over 10% by the end of the period. Exclusive
and mixed feeding rose over the period both at the first review (7-10 days after birth)
and the 6 to 8 week review. Overall, exclusive and mixed feeding represented 16% and
32% of infant feeding, respectively, at the 6 to 8 week review.
As in all the other areas and in line with the evidence from national data, levels of any
breastfeeding were higher among non-British mothers and older mothers. Analysis of
breastfeeding by country of birth within the neighbourhood showed no change in breast
feeding among non-British born mothers. There were fluctuations in breastfeeding levels
among British born mothers at birth and at 6 to 8 weeks but a slight increase in any
breastfeeding at first visit, rising from 43% in 1997 to 65% in 2009.
There was an overall increase in breastfeeding among the maternal age groups over the
study period. Between 1997/1999 and 2007/2009, any breastfeeding increased from
54% to 64% in mothers aged less than 25 years; 59% to 72% in mothers aged between
25 and 34 years and 58% to 68% in mothers aged 35 years or older.
Conclusions in relation to breastfeeding and other social trends
The influence on breastfeeding trends of compositional change in the neighbourhood’s
population is clearly important. The overall population and the proportion of people from
a non-British origin has increased to nearly 30% of the total population. Perhaps most
significantly the proportion of mothers of non-British birth has increased from a quarter
to nearly 60%. The age profile of mothers has also risen. The area remains deprived but
the income and employment deprivation indicators have contrasting trends.
While the Greater Glasgow ‘peer support’ breastfeeding initiative may have had some
influence on infant feeding levels - there is evidence at first visit of an increase in
xxvii Source: recent f igures as at April 2010 from COSLA xxviii
Estimates 2001-2005, GCPH community health profiles http://w w w .gcph.co.uk/communityprofiles