Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
There is an acknowledged need for better information to guide resource allocation and
service planning in the health services. Despite the recognition of the important role of socioeconomic factors, difficulties with the appropriate presentation of data have so far proved insuperable. Social area analysis (SAA), which is a generic name for a number of methods employing
census and other data to classify small areas into similar socioeconomic groups, is an approach
which quantifies data in a useful fashion and has important applications in medical, epidemiological,
and health services research. Most previous British exercises in SAA have been in the field of town
planning. The potentialities of the approach for community medicine are evaluated, by the use
of information from two existing studies. This is shown to discriminate more effectively than does
existing health information between hypotheses concerning geographical variations in mortality,
and it provides adequate explanations for urban area differentials in infant mortality, the uptake of
vaccination, and the incidence of infectious disease. Specific applications of SAA in health planning
and research are discussed.
SUMMARY
The recent introduction of national systems of of the population (the ecological fallacy). Social
health care planning and of resource allocation class is better employed to classify data than to
(Department of Health and Social Security, 1976a explain differences (Central Statistical Office, 1975).
and b) has concentrated attention on the effectiveness None the less, information on social class, housing
of existing health information systems and on the conditions, population density, and other relevant
relevance of available data. Both leave much to be variables has in the past been collected by health
desired. National norms of provision for service planners to produce composite socioeconomic
planning are still, in general, based on a crude valuations of areas (Coulter and Guralnick, 1959),
population count; and although the introduction of and such exercises can influence the allocation of
mortality data as indicators of need for resource scarce resources at the local level (Donaldson, 1976).
One approach, as yet untried in health care
allocation is a step in the right direction, it is widely
acknowledged that a considerable degree of refine- planning, which appears to permit the reliable use of
ment will be required. The importance of socio- socioeconomic data at population level, is that of
economic factors as determinants both of health social area analysis (SAA). This has been defined as
(Registrar General, 1971; Adelstein and White, 1976; 'a set of integrated procedures designed to study
Syme and Berkman, 1976) and use of health services characteristics of groups or subpopulations of
(Cartwright and O'Brien, 1976; Forster, 1976) is well people who live in defined geographic areas'
recognised, but it has not as yet proved possible to (Struening, 1974). Early studies in this field used
quantify variations in a way useful to health planners. census data to derive multivariate classifications
Census information on socioeconomic status by which ranked census tracts by social status and
occupation is readily available (Office of Population urbanisation (Shevky and Williams, 1949). Although
Censuses and Surveys, 1970, 1975) but problems the initial emphasis was on describing groups,
arise when this is employed to compute rates of research workers soon realised the value of relating
morbidity or mortality by social class (Buechley the classifications to non-ensus data such as crime
et al., 1956; Office of Population Censuses and and suicide rates (Shevky and Bell, 1955). ApplicaSurveys, 1970; Nagi and Stockwell, 1973). In tions in the field of health care, however, were slow
addition, doubts are cast upon the validity of this to develop; psychiatry is the sole health care topic
one-dimensional information as a proxy for the in which SAA has been employed in Britain (Philip
multidimensional socioeconomic aspects of need in and McCulloch, 1966; Giggs, 1973; Skrimshire,
health care planning because it imposes the necessity 1976a). Possible uses in the analysis of health data
to generalise from the level of the individual to that have been suggested (Craig, 1975), but actual
199
200
Alex Scott-Samuel
LIVERPOOL SOCIAL AREA STUDY
201
Table 1 Social class distribution between the five Families (social class of head of household as percentage of city total)
Family
Social Class
58
6-6
12-4
54
2-1
0 7
8-2
20-6
IandII
IIIN
Subtotal I, II, IIIN
IIIM
IV
V
SubtotalIIIM,IV,V
Total
(28)
(32)
(60)
(26)
(10)
(3)
(39)
(99)
1-2
2-2
3-4
2*7
1*7
1-0
54
8-8
(14)
(25)
(39)
(31)
(19)
(11)
(61)
(100)
3
0 3
1 0
1-3
2-0
2-5
3.1
7-6
(85)
8-9
(99)
(3)
(11)
(14)
(22)
(28)
(35)
2-0
58
7-8
12-6
7-8
4-8
252
33-0
Total
(6)
(18)
(24)
(38)
(24)
(15)
(77)
(101)
1-7
4-4
6*1
10-9
6-7
4*5
22-1
28-2
(6)
(16)
(22)
(39)
(24)
(16)
(79)
(101)
11-0
20-0
31*0
33-6
20-8
14-1
68.5
99 5
Table 2 Infant mortality, triple vaccination, and infectious diseases within the five Families
Family
Populationl
1
2
3
4
5
Total
125 921
52 790
52600
189 776
158 843
579 930
Live births2
289
220
163
433
565
1670
x2(4DF)
Infant
mortality3
25
36
25
35
68
189
13 6
<0-01
Infant
mortality
Triple
vaccine
rate4
14-4
27-3
256
13-5
20-1
uptake5
18-9
215
47
17
8
53
90
31-6
<0-001
Vaccine
uptakco
Infectious
49
23
15
37
48
39
84
73
78
140
174
rate
diseaseO
Infectious
disease
rate7
27
55
59
30
44
549
49.5
<0-001
38
202
Alex Scott-Samuel
uptake of the latter might reflect its lower incidences and the ratio was highly significant (F = 24.94,
of overcrowding and residential mobility. The lower DF = 29 428; P <0-001). A further variance ratio
infectious disease rate in Family 4 is probably due test was then performed for SMR between clusters
to its advantages in housing stock. Such comparisons and between AHAs; the ratio of the between-cluster
to the between-AHA variance was also highly
as these between Families 4 and 5 stress the importance of quantifying the influences of socioeconomic significant (F = 3 55, DF = 29 112; P = <0 001).
variables other than social class; this is emphasised Area types derived by SAA are thus sensitive to
by the inverse relationship between infant mortality mortality in their grouping of districts, and are more
and infant vaccination rates which has been demon- meaningful in explaining mortality variations than
strated at national level (West and Lowe, 1976), and are the geographically defined AHAs.
is evident in Families 1, 2 and 3, but does not appear
Discussion
to be true for Families 4 and 5.
SAA OF BRITISH LOCAL AUTHORITY DISTRICTS
203
204
Publications Association for Research in Nervous and
Mental Disease, 47, 130-138.
Struening, E. L., Rabkin, J. G., Cohen, P., Raabe, G.,
Muhlin, G. L., and Cohen, J. (1973). Family, ethnic,
and economic indicators of low birth weight and infant
mortality: A social area analysis. Annals of the New
York Academy of Sciences, 218, 87-107.
Syme, S. L., and Berkman, L. F. (1976). Social class,
susceptibility, and sickness. American Journal of
Epidemiology, 104, 1-8.
Webber, R. J. (1975). Liverpool Social Area Study. 1971
Data: Final Report. Technical Paper 14. Planning
Research Applications Group: London.
Alex Scott-Samuel