« Previous | Contents |
Listen
The Independent Evaluation of 'Starting
Well' Final Report
Footnotes
1 Greater Easterhouse is used here as a shorthand for
the following areas since the project did not cover Greater
Easterhouse in its entirety: Cranhill, Ruchazie, Craigend,
Garthamlock, Easterhouse and Gartloch.
2 Health visitors stated that postnatal tiredness and a
lack of time were very common reasons for not opting into
the study.
3 See also the website
http://www.ualr.edu/~crtldept/home4.htm,
4 At the time of writing, 73 participants had either
voluntarily withdrawn from the study (N=26) or moved
without leaving a forwarding address (N=47). Opt-ins who
could not be contacted for at least one assessment were not
included in analyses in order to maximise the number of
predictor variables available for modelling.
5 The contested nature of community development (Popple,
2001) is discussed more fully within the body of the
report.
6 Greater Easterhouse is used here as a shorthand for
the following areas since the project did not cover Greater
Easterhouse in its entirety: Cranhill, Ruchazie, Craigend,
Garthamlock, Easterhouse and Gartloch.
7 Because of the wide range of the evaluation and the
methods that it utilises, the description of methods that
is provided within this section is relatively brief. Where
more detailed technical descriptions of particular aspects
of the evaluation exist elsewhere they are referenced
rather than replicated in full.
8 predominantly the North Glasgow LHCC, including
Springburn and Possil. Lower than expected recruitment
rates here necessitated extension of the comparison area to
encompass two further LHCCs in the west of Glasgow
(Drumchapel and Clydebank)
9 40% of which were comparison area births, 60%
intervention births.
10 Health visitors stated that postnatal tiredness and a
lack of time were very common reasons for not opting into
the study.
11 At the time of writing, 73 participants had either
voluntarily withdrawn from the study (N=26) or moved
without leaving a forwarding address (N=47). Opt-ins who
could not be contacted for at least one assessment were not
included in the analysis in order to maximise the number of
predictor variables available for modelling.
12 as assessed by the Edinburgh Postnatal Depression
Scale, Cox, Holden and Sagovsky, 1987 (Appendix III)
13 Analysis for all qualitative data was analysed as
follows. All interviews and focus groups were
tape-recorded, transcribed in full and analysed using
either ATLAS -ti or N5 (Computer Assisted Qualitative Data
Analysis Software packages). Transcripts were coded for
both a priori themes outlined in the interview schedules
and themes emerging from the data. Data were analysed
thematically using constant comparative techniques.
Quotations have been selected to illustrate the range of
views on a given theme. The participant codes have been
removed from quotations since these may render respondents
identifiable. Where the names of persons or place appear in
the data these have been replaced with the word "Person" or
"Place" to preserve anonymity. Where views are believed by
the researcher to be extreme or unique, this is signalled
in the text. Some quotations are shortened (this is
indicated by the presence of the following punctuation
within the quoted text: …); this shortening indicates that
aspects of the quotation are either irrelevant to the issue
that is being discussed, repetitive or unintelligible.
14 Of the seven women not interviewed a second time,
three were contacted but failed to attend successive
appointments and four could not be contacted.
15 14 health-visitors took part in at least one
interview, some providing more than one case.
16 For an exception, see Elkan et al (2000) which
contains a selective review of the British literature
17 as defined by whole or part postcode sectors. See
appendix IV for a comparative contextual description.
18 As part of the UK's immigration and asylum policy,
several hundred families of asylum-seekers were 'dispersed'
(temporarily settled) to the area covered by the Northern
LHCC during the study recruitment period. Births to these
families could not be included due to variations in the
developing health visiting service and to lack of
interpreting support.
19 Appendix IV describes an approximation of the main
comparison area comprised of whole postcode sectors.
20 Here, and in the rest of the section, this refers to
Greater Glasgow NHS Board Child Health Information Team
21 Note these figures include births to families who
transferred out of the area before they could be approached
for consent and therefore differ slightly to those
published in previous outputs that did not consider these
to be 'eligible' births.
22 In practice, missing values for some predictor
variables further restricted the number of cases available
for analysis. This explains the reduction in N in some
regression tables below.
23 In fact, all minority participants live in the
southern project area, in a well-established
Scottish Asian community largely composed of first, second
and third generation Pakistani Muslims. Disclosure concerns
prevent release of 2001 Census ethnicity data at
small-areas levels (e.g. postcode sector), however, just
over two (2.2) percent of all households in the city are
headed by people of this ethnicity (General Register Office
(Scotland), 2003).
24 This is confirmed by contextual comparisons of
area-level material resources (see Appendix IV).
25 Note that additional outcomes (e.g. maternal
self-esteem, functional support, parenting confidence) are
not included here both for reasons of parsimony and the
fact that they tap similar dimensions to the chosen
measures (e.g. EPDS and self-esteem; HOME score and
parenting confidence).
26 Due to differing skills, emphases, training, etc, it
is not easy to confidently state that one health visitor
home contact has the same therapeutic significance as one
health support worker or one nursery nurse contact. For
these reasons, visits by these health professionals were
not included in analyses.
27 This is only true for comparisons
across models (e.g. 6- and 18-month assessments),
to the extent that the same predictors are entered with
similar sample sizes.
28 'NS-SEC class 8' classifies respondents as never
having worked or being long-term unemployed
29 Actual c
2 statistics and degrees of freedom available on
request.
30 Significant positive intervention effects are also
observed with the other two measures described in table
2.11 but are not reported for reasons of space.
31 Interestingly, given the prominence of Olds' work in
the initial design of the project, only three studies (Olds
et al 1986; 1994; Kitzman et al, 1997) use the HOME as an
outcome. Only one (Kitzman et al) shows significant
intervention effects but in relation to a highly selective
sample of disadvantaged African-American primiparous
mothers who had given birth to very premature infants
(<29 weeks gestation).
32 A number of studies in the HTA review (e.g. Larson,
1980; Black et al, 1995) report findings that are only just
significant at the p=0.05 level, whilst others (e.g. Black
et al, 1994) just fail to reach this level.
33 As both HOME and EPDS were collected at the same
assessment, it is difficult to make causal statements
regarding their inter-relationship.
34 A similar conclusion was reached in a recent British
study by Davies, Howells and Jenkins (2003) which involved
repeated EPDS screenings over the first postnatal year.
Here 'new' cases were often women who had moved into the
practice area and had not been screened previously. This
could not be a factor in our study but the finding that the
number of significant life-events predicted high scores at
eighteen months could point to the aetiology of depressed
mood amongst this 'late-onset' group.
35 As a corollary, a recent qualitative study of parents
of pre-school children in three Scottish NHS Boards
(including Greater Glasgow) described the many perceived
barriers to registration, including confusion as to the
appropriate age to do so, parental fear, hostile attitudes
to staff, accessibility issues and lack of information and
incentives to attend (Morrison, Macpherson and Binnie,
2000). Future qualitative work might explore the extent to
which the observed gains in intervention areas are
attributable to challenging these attitudes.
36 The practice of articulating a project's Theory of
Change, however, throws up a serious of further practical
and conceptual challenges for the evaluator (Barnes et al,
2003; Mackenzie and Blamey, in press; Mackenzie and
Benzeval, in preparation).
37 In practice more home visits are often delivered. In
the quasi-experimental study comparison group, the median
number of home visits to six months (i.e. an intermediate
point between the two interviews) was 3 with range 0-16.
The capacity to deliver these visits is however, severely
limited by caseload size and tends to be focussed on the
very vulnerable.
38 Several health visitors noted that this often meant
that input
increased over time, meaning that the scheduled
decrease in visit frequency did not always occur. This
exacerbated already-heavy workloads (see next section).
39 Copies of the Family Health Plan available on
request.
40 This evaluation was not commissioned to look in
detail at the effectiveness of these individual tools in
changing practice; rather it aimed to focus on
general lessons concerning changing practice.
Various pieces of evaluation commissioned internally by the
Starting Well team focus on specific tools of practice.
41 The joint working between health visitors and health
support workers is discussed in section 5.3.
42 The issues that arise where these boundaries are not
maintained are discussed further in 3.3.3.3
43 This is a small number of examples of good practice
undertaken at a local level and is not an exhaustive
list.
44 Digitised boundary data obtained from the UKBORDERS
service:
http://www.ukborders.ac.uk
45 Further details can be obtained from the project
team.
46 see definitions in section 2.2.1, page 14.
47 in analyses not shown, this sub-sample of comparison
families were found to be non-significantly less affluent
than the 'full' cohort
48 for example, those composing the 'additional'
recruitment areas of Drumchapel and Clydebank
49 All 2001 Census data was downloaded from the General
Register of Scotland's 'Scottish Census Results On-Line'
(SCROL), accessed at http://www.scrol.gov.uk.
50This latter statistic seems at odds with the impact
study finding that the intervention cohort are more
disadvantaged than the comparison cohort. This may be due,
however, to the fact that the census data refers to parents
with dependent children of
all ages; parents of the youngest children may
compose a more disadvantaged sub-group.
51 compared to a Glasgow City figure of 31.5% and a
Scottish figure of 14.7% (GRO(Scotland), 2004)
52 for example a higher proportion of NS-SEC Class 8
residents in the intervention area but better women's
self-reported health in the comparison area
« Previous | Contents |