Impact of social standing on injury prevention in a World Health Organization Safe Community—intervention outcome by household employment contract

K Lindqvist, T Timpka and N Karlsson

Department of Health and Society, Linköping University, S-581 85 Linköping, Sweden

Correspondence: Kent Lindqvist, Division of Preventive and Social Medicine, Department of Health and Society, Faculty of Health Sciences, Linköping University, S-581 85 Linköping, Sweden. E-mail: kenli{at}ihs.liu.se


    Abstract
 Top
 Abstract
 The WHO Safe Community...
 Materials and methods
 Results
 Discussion
 References
 
Background Although social inequality in health has been an argument for community-based injury prevention programmes, intervention outcomes with regard to differences in social standing have not been analysed. The objective of this study was to investigate rates of injuries treated in health-care among members of households at different levels of labour market integration before and after the implementation of a WHO Safe Community programme.

Methods A quasi-experimental design was used with pre- and post-implementation data collection covering the total populations <65 years of age during one year in the programme implementation municipality (population 41 000) and in a control municipality (population 26 000). Changes in injury rates were studied using prospective registration of all acute care episodes with regard to social standing in both areas during the study periods.

Results Male members of households categorized as not vocationally active displayed the highest pre-intervention injury rates. Also after the intervention, males in households classified as not vocationally active displayed notably elevated injury rates in both the control and study areas. Households in the study area in which the significant member was employed showed a post-intervention decrease in injury rate among both men (P < 0.001) and women (P < 0.01). No statistically significant change was observed in households in which the significant member was self-employed or not vocationally active. In the control area, only an aggregate-level decrease (P < 0.05) among members of households in which the significant member was employed was observed.

Conclusions The study displayed areas for improvement in the civic network-based WHO Safe Community model. Even though members of non-vocationally active households, in particular men, were at higher pre-intervention injury risk, they were not affected by the interventions. This fact has to be addressed when planning future community-based injury prevention programmes.


Accepted 9 December 2003

To handle the fact that unintentional injuries are growing into a global public health problem, injury prevention programmes have been introduced at national and local levels in many countries.1 Community-based programme development, building interventions into the existing networks in the civic environment and emphasizing broad participation, has long been known as a promising model in public health practice.2 For example, in Sweden, interventions in injury prevention based on local popular participation and inter-sectoral co-operation between authorities have been in use for more than two decades. The scope of the interventions includes empowerment of the socially weak and continuous tracking of high-risk environments and groups. Guided by participatory action research,3 the structure of these programmes has been developed by researchers, local politicians, civil servants, representatives of non-government organizations, and public health workers in co-operation.4,5 Injury prevention programmes based on self-governing civic networks of this type have been evaluated with respect to the programme processes,4 effects on the total injury incidence,6–8 and with respect to specific areas such as child injuries,9 sports injuries,10 and traffic injuries.11 Since 1989, the civic network model for community-based injury prevention has also been used as the basis for the WHO Safe Community accreditation. The accreditation criteria define the norms that communities have to fulfil to be eligible for Safe Community status (Table 1). In 2002, there were 70 Safe Communities, in both industrialized and developing countries.12 However, although awareness of social stratification is central to the definition of community-based programmes, the outcomes with regard to differences in social standing have not been analysed. This is an unsatisfactory situation, because previous studies have shown that different forms of social deprivation are associated with injury risk in different population groups. For instance, the risk of childhood injury has been reported to be associated with the income level in the residential area13 and the parents' educational level.14 For adults, the injury risk has been found to be related mainly to social class, and this association is stronger for men than for women.15,16


View this table:
[in this window]
[in a new window]
 
Table 1 The World Health Organization Safe Community criteria

 
The objective of the present study is to explore the WHO Safe Community model for community-based injury prevention with regard to relationships between social standing, gender, and post-intervention injury rates. In particular, the aim is to study, using a quasi-experimental design, rates of injuries treated in health-care among members of households at different levels of labour market integration. In the analyses a wider perspective on social standing than the purely economic is thus assumed. It is postulated that the relationship to the labour market is associated, first, with the phenomenon of social exclusion, i.e. that individuals may be disadvantaged not only economically, but also in terms of dimensions such as education and cultural resources.17 The assumption includes that there is a process of accumulating disadvantage, where different factors are important at different points in people's lives.18 Second, it is assumed that whether or not the significant adult in the household has an employment contract has a determining influence on the social capital, i.e. on the existing social networks, norms of reciprocity, and the potential the household members have to control and benefit from leisure time.19 The relationship to the labour market is classified using the Swedish Socio-economic Index (SEI),20 which has been employed since the early 1980s to represent social standing in most national databases and statistics. This index defines social standing primarily as being based on occupation. Children and young people are categorized to the SEI group to which their parents' household belongs.


    The WHO Safe Community model
 Top
 Abstract
 The WHO Safe Community...
 Materials and methods
 Results
 Discussion
 References
 
The injury prevention programme implemented in Motala municipality (population 41 000) in the western part of Östergötland County in Sweden is one of the original reference sites for the WHO Safe Community accreditation criteria. The aim of the community during the analysis stage (1983–1984), was to study the local epidemiology, to follow the economic consequences of injuries, and to analyse the local social structure and values. Stage two, the programme design and initiation (1985–1987), included organizing the management of interventions and setting local planning goals. The district Health Services Board, the Municipal Board, and political committees and management groups were approached to accept responsibility for the programme from its initial stages. The operational goal set for the programme was to reduce the total injury incidence in the municipality by 25% by the year 2000. It was also decided that the interventions would be focused on two risk populations (children and teenagers, and the elderly) and three risk environments (traffic, sports and recreation, and the workplace). According to the broad-participation strategy, self-regulatory local action groups were formed for each risk area, consisting of a group facilitator and representatives from the local organizations that managed injuries. The design evolved into a programme of action during the implementation stage (1987–1988). The intervention effects were to be mediated by a continuous reciprocal interaction between behavioural and environmental determinants.21 The stage was based on the involvement of a cross-section of the population, and both professionals and lay people were invited to participate in injury prevention. To set the agenda in the community, a general media campaign was initiated in 1987. Both passive interventions, in terms of modifying the physical environment, and active interventions aimed at behavioural changes through health education, were thus introduced into the local action groups by the facilitator. During the year of implementation, most groups decided to make changes in the physical environment, but corresponding safety education programmes were also started (Table 2). All interventions were based on majority decisions made by the local action groups.


View this table:
[in this window]
[in a new window]
 
Table 2 Organizational structure and interventions performed during the implementation phase

 

    Materials and methods
 Top
 Abstract
 The WHO Safe Community...
 Materials and methods
 Results
 Discussion
 References
 
A quasi-experimental design22 was used with pre- and post-implementation registrations covering the total populations <65 years of age in the programme implementation area and in a neighbouring control municipality (population 26 000) in Östergötland county. The pre-implementation study period covered the 52 weeks from 1 October 1983 to 30 September 1984. The post-implementation period covered 52 weeks from 1 January 1989 to 31 December 1989. Change in the morbidity rates following the intervention was studied using prospective registration of all acute care episodes during the study period. The study area had four health-care centres and a county annex hospital with a casualty department, whereas the control area shared the annex hospital and had two health-care centres, one of them with an emergency unit. The use of outpatient health-care services at these units was followed by recording data from all health-care contacts during both registration periods. Both the study and control areas are situated 50 km from Linköping University Hospital.

Validity and reliability tests
To identify confounding social trends, data regarding population age and gender mix, sites of residency, education, income, and employment were collected, retrospectively, from national registers (Statistics Sweden) for the study periods. Systematic differences between the areas concerning injured people seeking care outside the registration areas were obtained by analysing all visits to the emergency departments of Linköping University Hospital during September 1984. Fatal injuries, other than those recorded in the study, were researched by analysing local police records.

Data collection
Data were collected from two sources. For all patients contacting a health-care unit located in the study area during the study periods, a report form with the time of contact and standard personal data was filled in by staff at the care unit. The same form included whether unintentional injury was a possible reason for the contact. For the registered patients, specially trained nurses recorded an International Classification of Diseases, Eighth Revision (ICD-8) based diagnostic classification using medical records and discussions with physicians. The routine for data collection was tested in a pilot project. Before starting each study period, the staff at all relevant health-care units were carefully informed, and the routine was trained in practice for 2 weeks.

The SEI classification was used for the primary representation of the social standing of the individuals in the intervention and control areas. SEI data for all individuals in the intervention and control areas were collected from Statistics Sweden. Due to the changes in household structure in relation to retirement and ageing, individuals >65 years of age were excluded from the analyses.23 For the pre-implementation registration, SEI data originated in the census survey conducted in 1985; and for the post-implementation registration, the data originated in the census survey conducted in 1990. Considering that the WHO Safe Community model strongly relies on existing civic networks, the detailed SEI categories were used for coding individuals into three secondary categories based on the relation that the household had to the labour market; households in which the vocationally significant member was employed, households in which the vocationally significant member was an entrepreneur or self-employed, and households in which the adults not were vocationally active. This classification did not contain any vertical intra-group stratification, i.e. income or education differences within the social groups were not accounted for. The group ‘self-employed’ also included farmer's households. Temporarily unemployed individuals were coded according to their most recent occupation. The group ‘not vocationally active household’ comprised households in which the adult members had not been able to settle in the labour market, were on long-term sick leave, or had recently arrived in the country as immigrants. The group ‘not vocationally active household’ also comprised individuals who were institutionalized.

Statistical methods
Rates of injuries (expressed as per 100 person-years) were calculated by community (study and control municipality) for each study period (1983/84 and 1989), by socioeconomic group (employed, self-employed, and not vocationally active), by gender, as well as for women and men together. 95% CI were provided for injury rates. For individuals injured more than once, only the first episode was included in the calculations.

The differences in injury rates between 1989 and 1983/84 were computed with 95% CI and significance tests were performed. A P-value<0.05 was considered as statistically significant. All computations were performed using SPSS statistical software (release 11.5).


    Results
 Top
 Abstract
 The WHO Safe Community...
 Materials and methods
 Results
 Discussion
 References
 
Community characteristics
The age and gender mix in both areas were close to the national average and stable over the registration periods. Members of households in which the vocationally significant member was employed constituted the largest share of the population <65 years of age in both the intervention (82%) and control (81%) areas. The members of self-employed households represented 7% and 9%, respectively (Table 3). Members of households classified as not vocationally active constituted 10% of the population <65 years of age in both areas. The income levels in both areas were at 93% of the national average and remained stable between the registration periods. Between 49% and 51% of the total population in the intervention and control areas were gainfully employed during the registration periods. During both periods, the share of the population with more than compulsory school education was about 5% below the national average in both areas. Similarly, the share of urban residents remained between 79% and 82% in both areas. The distribution of employers was comparable between the areas and registration periods, the share employed by manufacturing industries (31–34%) was higher than the national average (21–20%).


View this table:
[in this window]
[in a new window]
 
Table 3 Populations below 65 years of age in the intervention and control areas displayed by sex and household relation to labour market

 
In 1983, the utilization of outpatient health-care services for reasons other than injuries was 27% higher in the intervention area than in the control area for males, and 17% for females. In 1989, the corresponding differences were 29% for males and 21% for females. Females generally made approximately 20% more outpatient visits to health-care units than males, both in the intervention and control areas during both registration periods. For adults, members of households categorized as not vocationally active displayed the highest overall frequency of outpatient visits. For these households, males displayed similar or higher care utilization rates when compared with females.

Quality of injury data
During the pre-implementation registration period, identity data were missing for less than 1% of the injured patients in the study and control areas. During the post-implementation registration, less than 0.5% of the injured patients in the study and control areas could not be identified in medical records. A lower share of all injured residents from the study area (11/422), 3% (95% CI: 1%, 5%), compared with the control area (28/253), 11% (95% CI: 7%, 15%), were found to have been directly provided with acute care outside the local health-care units during the month of the control study. Police records did not disclose previously unrecorded injuries.

Pre-intervention injury rates
Male members of not vocationally active households displayed the highest pre-intervention injury rate in both the control and study areas (Table 4). Also in households in which the vocationally significant member was employed, males showed higher injury rates than females. There was no difference between men and women among members of households in which the vocationally significant member was self-employed.


View this table:
[in this window]
[in a new window]
 
Table 4 Rate per 100 person-years (95% CI) of injured individuals in 1983/84 in intervention and control areas displayed by sex and household relation to labour market

 
Post-intervention injury rates
In the control area, an aggregate-level, post-intervention decrease in injury incidence (P = 0.02) was observed in households in which the vocationally significant member was employed (Table 5). In the study area, male members of not vocationally active households continued to display notably elevated injury rates after the intervention. Neither was there any change in injury rates observed in the study area for members of households in which the vocationally significant member was categorized as self-employed. For members of households in which the vocationally significant member was employed, the injury incidence had decreased both among men (P < 0.001) and women (P = 0.003).


View this table:
[in this window]
[in a new window]
 
Table 5 Rate per 100 person-years (95% CI) of injured individuals in 1989 and change in rates between 1989 and 1983/84 (95% CI) in intervention and control areas displayed by sex and household relation to labour market

 

    Discussion
 Top
 Abstract
 The WHO Safe Community...
 Materials and methods
 Results
 Discussion
 References
 
The objective of the present study was to analyse the WHO Safe Community model with regard to relationships between social standing, as defined by the employment contract held by the household's significant member, gender, and pre- and post-intervention injury rates. It was acknowledged that the WHO model emphasizes empowerment of the socially weak in safety promotion and continuous tracking of high-risk environments and groups. As expected, the results are in accordance with previous studies that have shown that socially disadvantaged groups are subject to a higher injury risk.24,25 In the present study, male members of households in which the adults were not vocationally active displayed the highest pre-intervention injury rates. But in contrast to what was expected, the results also displayed that the WHO Safe Community interventions, which in the main relied on local civic networks, failed to reach these disadvantaged households.

One possible explanation for these observations follows from the fact that the quality and quantity of the interaction in social networks vary between different social groups.26 The post-intervention injury rates observed in the present study may reflect the fact that the social networks available for households in which the adults are not vocationally active, and for self-employed households, have a different structure compared with households in which the adults are employees. Close social connections within groups in a community may in fact not necessarily be beneficial for all individuals in the population.27 In the present study, although for different reasons, the households in which the adults were not vocationally active and the households of self-employed adults were not exposed to safety information provided through traditional workplace sources, and these households can also be expected to participate less in local civic activities. Because of their high pre-intervention injury rates, this situation is especially negative for households in which the adults are not vocationally active. To reach this group, experience from previous successful programmes aimed at socially deprived groups can be employed.28–30 For example, in the evaluated programme, it is possible that engaging social workers, in home inspections, in family empowerment and educational programmes, was not fully exploited. However, when planning an extended professional outreach to members of households with small social networks, the benefit of targeting individuals at high risk must always be balanced against the possible intrusion on personal integrity. A first step could be to analyse the social networks available to these households to identify the organizations that are the most suitable to take the main responsibility for implementing the interventions.

A second explanation can be gender-related role behaviour, i.e. that in certain everyday settings men and women share behavioural patterns, whereas in other comparable settings they act according to different gender roles. One display of such gender-related behaviour is apparent in working life, where the distribution of occupations between men and women is highly differentiated. This differentiation has been suggested to have an influence on work-related illness and sick leave patterns.31 Such gender-related variation in risk exposure may also explain some of the differences in injury rates between men and women found in this study. For instance, it may explain the fact that while there were small pre-intervention gender differences in injury rates in self-employed households, there were significant differences in households in which the adults were employees. It can be assumed that the type of daily labour is more similar when comparing men and women in self-employed households, whereas the division of labour is more traditional in households in which the significant member is employed. It is thus possible that the higher injury rate among women in self-employed households is a consequence of a stressful general life situation, i.e. that the women take the main responsibility for the home and children, while also working in the family enterprise. However, although it may have had an influence on pre-intervention risks, gender segregation does not explain why the post-intervention injury rates among women in self-employed households remained constant. Therefore, before interventions are planned for women belonging to self-employed households, more knowledge is required both about direct injury risks, and the existing information and structural barriers that impede implementation of safety interventions.

At the aggregate level, the injury incidence in the control area only decreased for members of households in which the significant member was employed, and this decrease was significantly lower than in the study area. This observation suggests that the decrease in injury rates in the intervention area cannot be explained solely by general societal trends. A trend towards higher general utilization of health-care during the period of post-implementation registration was also observed in both areas, implying that the reduced injury rates cannot be explained by a lower availability of health-care services during the post-implementation registration. An evident weakness of the present study is that it is not based on a randomized design. For this reason, a set of confounders were prospectively followed in the study and control communities. Confounders with possible influence on the results that not were followed include alcohol consumption and number of immigrants. Retrospective analyses of the latter aspects using official statistics from each respective area did not display any evident differences between the areas or extraordinary rates. Another type of confounder that may have influenced the results is interventions performed at a national level that may have influenced the groups under study in different ways, e.g. general programmes on work-place safety. No adjustment for such programme effects were made in the present analyses. Nevertheless, although a randomized trial is the preferred benchmark test of community-based injury prevention, these observations correspond with previous studies which suggest that quasi-experimental approaches using cohorts can provide realistic appraisals of effectiveness.32

In conclusion, few studies have examined the impact of interventions in injury prevention in different social groups.33 Without such evidence, it remains difficult for safety promotion professionals to understand how to target social inequalities in injury rates. This study displayed important areas for improvement in the civic network-based WHO Safe Community model for safety promotion. The results showed that particularly non-vocationally active households not only were at higher pre-intervention injury risk, but they were also not reached by the interventions. This fact has to be addressed in planning future community-based injury prevention programmes. The revised programme models should include interventions that are informed by local injury data stratified not only by age and gender, but also by social standing. Only then can adequate outreach to high-risk groups be planned and implemented, taking into account the structure and qualities of local civic networks.


    Acknowledgments
 
This study was supported by grants from the Swedish National Rescue Services Board.


    References
 Top
 Abstract
 The WHO Safe Community...
 Materials and methods
 Results
 Discussion
 References
 
1 Lindqvist K, Timpka T, Schelp L, Åhlgren M. Evaluation of a home injury prevention program in a WHO Safe Community. Int J Consumer Product Safety 1999;6:25–32.

2 Rothman J. Planning and Organizing for Social Change: Action Principles from Social Research. New York: Columbia University Press, 1974.

3 Macaulay AC, Commanda LE, Freeman WL et al. Participatory research maximises community and lay involvement. BMJ 1999;319:774–78.[Free Full Text]

4 Lindqvist K, Timpka T, Schelp L. Ten years of experiences from a participatory community-based injury prevention program in Motala, Sweden. Public Health 1996;110:339–46.[ISI][Medline]

5 Swedish National Institute of Public Health (SNIPH). Strategies for Success. Report 1994:4. Stockholm, Sweden: SNIPH, 1994.

6 Ytterstad B. The Harstad Injury Prevention Study: Hospital-based Injury Recording and a Community-based Intervention. Thesis. Tromsö, Norway: Institute of Community Medicine, University of Tromsö, 1995.

7 Svanström L, Schelp L, Ekman R, Lindström Å. Falköping, Sweden, ten years after—still a safe community? Int J Consumer Product Safety 1996;1:1–7.

8 Timpka T, Lindqvist K, Schelp L, Åhlgren M. Community-based injury prevention: effects on health-care utilization. Int J Epidemiol 1999;28:502–08.[Abstract]

9 Lindqvist K, Timpka T, Schelp L, Risto O. Evaluation of a child safety program based on the WHO Safe Community model. Injury Prev 2002;8:23–26.[Abstract/Free Full Text]

10 Timpka T, Lindqvist K. Evidence-based prevention of acute injuries during physical exercise in a WHO Safe Community. Br J Sports Med 2001;35:20–27.[Abstract/Free Full Text]

11 Lindqvist K, Timpka T, Schelp L. Evaluation of inter-organizational traffic injury prevention in a WHO safe community. Accid Anal Prev 2001;33:599–607.[CrossRef][ISI][Medline]

12 Klang M, Andersson R, Lindqvist K. Safe Communities—the Application to Industrialized Countries. LCC Occasional Papers 5, Special Issue. Linköping, Sweden: Linköping Collaborating Centre, Linköping University, 1992.

13 Durkin MS, Davidson LL, Kuhn L, O'Connor P, Barlow B. Low-income neighbourhoods and the risk of severe paediatric injury: a small-area analysis in northern Manhattan. Am J Public Health 1994;84:587–92.[Abstract]

14 Petridou E, Kouri N, Trichopoulos D, Revinthi K, Skalkidis Y, Tong D. School injuries in Athens: socioeconomic and family risk factors. J Epidemiol Community Health 1994;48:490–91.[ISI][Medline]

15 Gijsbers van Wijk CMT, Kolk AM, van Den Bosch WJHM, van Den Hoogen HJM. Male and female health problems in general practice: the differential impact of social position and social roles. Soc Sci Med 1995;40:597–611.[CrossRef][ISI][Medline]

16 Whitehead M. The World Health Organization—WHO stimulates a commitment to tackling inequalities in health. BMJ 1995;310:1472.[Free Full Text]

17 Byrnes D. Social Exclusion. Buckingham: Open University Press, 1999.

18 Social Exclusion Unit. Social Exclusion Unit: What's it all about? http://www.cabinet.gov.uk/seu

19 Putman RD. Making Democracy Work. Princeton, New Jersey: Princeton University Press, 1993.

20 Socio-economic classification (SEI). Vol. 4. Reprinted 1984. Örebro, Sweden: Statistiska centralbyrån (SCB), 1982. (In Swedish).

21 Bandura A. Social Learning Theory. Englewood Cliffs, New Jersey: Prentice-Hall, 1977.

22 Cook TD, Campell DT. Quasi-experimentation. Boston: Houghton Mifflin, 1979.

23 Silventoinen K, Lahelma E. Health inequalities by education and age in four Nordic countries, 1986 and 1994. J Epidemiol Community Health 2002;56:253–58.[Abstract/Free Full Text]

24 Kelly SM, Miles-Doan R. Social inequality and injuries: do morbidity patterns differ from mortality? Soc Sci Med 1997;44:63–70.[CrossRef][ISI]

25 Laflamme L, Engström K. Socioeconomic differences in Swedish children and adolescents injured in road traffic accidents: cross-sectional study. BMJ 2002;324:396–97.[Free Full Text]

26 Lindström M. Social Participation, Social Capital and Socioeconomic Differences in Health-related Behaviours. Dissertation. Lund, Sweden: Lund University, Department of Community Medicine, 2000.

27 Portes A. Social Capital: Its origins and applications in modern sociology. Annu Rev Sociol 1998;24:1–24.[CrossRef][ISI]

28 Schwartz D, Grisso J, Miles C, Holmes J, Sutton R. An injury prevention program in an urban African-American community. Am J Public Health 1993;83:675–80.[Abstract]

29 Davidson LL, Durkin MS, Kuhn L, O'Connor P, Barlow B, Heagarty M. The impact of the safe kids/healthy neighbourhoods injury prevention program in Harlem. Am J Public Health 1994;84:580–86.[Abstract]

30 Kuhn L, Davidson LL, Durkin MS. Use of poisson regression and time series analysis for detecting changes over time in rates of child injury following a prevention program. Am J Epidemiol 1994;140:943–55.[Abstract]

31 Leijon M, Hensing G, Alexanderson K. Gender trends in sick-listing with musculoskeletal symptoms in a Swedish county during a period of rapid increase in sickness absence. Scand J Soc Med 1998;26:204–13.[ISI][Medline]

32 Kraus JF. Cohort studies in injury research. In: Rivara FP, Cummings P, Koepsell TD, Grossman DC, Maier RV. Injury Control. Cambridge: Cambridge University Press, 2001, pp. 129–38.

33 Dowswell T, Towner E. Social deprivation and the prevention of unintentional injury in childhood: a systematic review. Health Educ Res 2002;17:221–37.[Abstract/Free Full Text]