SUBJECTS AND METHODS

The study was carried out between October 2000 and April 2001 in the Department of Hauts-de-Seine near Paris. This region was chosen because of increased injury prevention efforts, particularly in the second largest town of Boulogne (107 000 people), where a safe community program was started in 1997.^19,20

One hundred families from four towns (Boulogne-Billancourt, Chaville, Sévres, and Ville d’Avray), were selected by Mother and Child Protection services (PMI), a free nationwide health organisation. PMI’s usual criteria for services were used and included primipara, medical, psychological, and/or socioeconomic difficulties. Demographic data for these towns are summarized in table 1.

Families were randomized into two groups of 50; one family was lost to follow up in group 1. Group 1 received counselling, two pamphlets about domestic injuries and methods of prevention, emergency call numbers, and a safety kit. The kit included cupboard and drawer latches, door handle covers, table protection corners, electric outlet covers, a non-skid bathtub mat, a smoke detector, and a phone sticker with the number of the poison control centre.

Group 2 received counselling and pamphlets but did not receive the kit. Visits were conducted by health professionals (25 doctors, nurses, or auxiliary nurses) who received the same instructions for home visits and data collection. Qualitative and quantitative data, that is, process and impact, were collected by questionnaire and interview.

The first home visit occurred when a child reached the age of 6–9 months. Information about social status, highest school level of the respondent (generally the mother), physical environment at home, and safety behaviours were collected during this visit. A second visit was performed 6–8 weeks later by the same health professional.

Safety behaviour was measured at the first and second visits using a closed response questionnaire. The safe or dangerous character of each home situation was defined in an inventory at the end of the questionnaire. The definition of safe behaviours was related to appropriate answers to the questions in table 5. For instance: “Do you use electric outlet covers? (yes); Are medicines stored in locked or out-of-reach locations? (yes); Are there stray wires that might cause people to trip? (no) . . .”.

During the second visit, improvement in safe behaviour was assessed by increased use of safety devices. Three conditions were used in the analyses: first, safety improvement related to each type of home injury regardless of relationship to the kit; second, safety improvement for risks preventable by devices in the kit; and third, safety improvement for risks not related to devices in the kit.

Global satisfaction with the intervention was assessed among families through an open questionnaire during the second visit and focus groups assessed the global benefit of the kit, negative and positive points, and receptivity of the families.

Filemaker Pro software (version 4.1) was used for data capture. Statistical analyses were performed by SPSS software (version 8.0). The χ² test was used for comparison of qualitative data and the Fisher’s exact test was used in the case of small sized samples. Quantitative data were compared by Student’s t test. A p value less than 5% was considered as significant.

RESULTS

Quantitative evaluation in all families

The two groups were comparable in age, gender, and employment of the respondent, type of family, type of dwelling, number of children in the family, and highest school level (tables 1 and 2). Between the first and the second visit, the percentage of safety improvement was significantly higher in the group with the kit (group 1). Comparisons between the two groups and between the first and the second visits were analysed for each type of risk (table 3). Improvement was related to the risk of falls (p<0.02), fire and burns (p<0.001), poisoning (p<0.01), and suffocation (p<0.001).

The use of devices provided in the kit was significantly different between the two groups: 64.4 % safety improvement in group 1 versus 41.2 % in group 2 (p<0.01) (table 4). The relative risk (RR) of safety changes between groups was 1.56 (95% confidence interval (CI) 1.35 to 1.80). Improvements were also observed for changes unrelated to the devices provided: 31.2 % in group 1 versus 20.2 % in group 2 (p<0.05) (table 5). Several categories had higher improvement in group 2, yet the RR between groups was 1.54 (95% CI 1.22 to 1.93).

Quantitative evaluation in particular situations

In single parent families, there was a significant difference between the groups for use of devices provided in the kit: 90.0% in group 1 versus 44.4% in group 2 (p<0.007). No difference was noted for devices not provided. There also was a significant difference in use of provided devices for families where the respondents’ level of education was low (no university studies): 61.5% in group 1 versus 36.5% in group 2 (p<0.01).

In families with at least two children (25 in group 1, 29 in group 2), no significant difference was observed between groups for any type of prevented risk.

Qualitative evaluation

The receptivity and interest of families was considered good. At the second visit, some parents had made safety improvements in areas not addressed at the first visit, such as curtain strings or furniture under windows. More than nine in 10 families declared they would participate if such a program was implemented again. They also claimed considerable need for training, particularly concerning child injury morbidity, risks according to stage of child development, and means of prevention. Training in optimal use of the kit and what to do in case of emergency were sought by most parents.

Focus group meetings with health professionals were relevant for both process and impact assessment. Kit delivery at home was generally considered as propitious for discussion and for the effectiveness of the intervention. It heightened awareness of potential domestic dangers and prompted reflection on safety. The questionnaire was also positively perceived and was considered as a vehicle to provide information.

Some devices required assistance from health professionals for installation and use, particularly those requiring a screwdriver. This happened only during the second visit and had no influence on results. The most appreciated devices were door handle covers, electric outlet covers, cupboard and drawer latches, pamphlets, and emergency phone numbers. Devices that were rarely used or unsuited to the home included a screw hook, an inflated tap cover, and a refrigerator closing strip. Table corner covers and a universal door latch for electrical appliances were judged as too fragile. Health professionals proposed the inclusion of new devices such as window chains and curtain string rollers. All health professionals expressed the desire for the program to continue so they could incorporate it into their daily practice

DISCUSSION

Four minimal conditions, identified in the literature, are required to guarantee the success of such a project:

1. Kit delivery and counselling must be done in the home.

2. Safety devices must be free and easy to use.

3. Families may need help to install devices.

4. Several home visits must be planned for follow through.

Repeated home visits and subsidized safety devices promote a “therapeutic alliance” especially for groups at high risk of injury.²¹ The present study was conceived in this spirit following the example of the Quebec study. City selection was related to practical considerations for project implementation with adequate staff. Family selection based on PMI recruitment criteria theoretically targeted low socioeconomic populations. However, nearly half of the families in both groups had a high education level, two thirds had a wage earning job, and the unemployment rate was quite low. But other criteria for PMI recruitment, such as psychological difficulties or isolation, are also related to home injury risk. Thus our pilot study included several differently defined high risk groups.

A possible difficulty in the interpretation of results is the use of global tests performed on a sum of questions. This unweighted sum of risk situations may be questionable. Additionally, some families are probably more at risk than others, and some risks are linked. But statistical tests performed for each question individually lack power so comparisons are rarely significant. Since the aim of the study was to test the impact of the kit and its effect on safety behaviours as a whole, we believe global tests are relevant.

Other studies have also shown positive results for a combined approach of free home counselling and safety devices. Six year follow up data of a fire prevention program in Oklahoma City (USA) indicate an 81% reduction in fire related injuries among the targeted population versus 7% in the rest of the town.²²

Data on the financial aspects of this kind of intervention are rare. The same program among low socioeconomic groups in London is being assessed.²³ A cost effectiveness analysis of the preventive impact of home visits in a multicentre survey in Canada indicates a savings of 372 Canadian dollars per avoided injury.²⁴

Owning prevention tools in not always sufficient. The adoption of safety conscious behaviour, even among higher socioeconomic groups, is not automatic.²⁵ Home visits, free safety devices, and a pedagogic approach are all needed.²⁶

A recent review of 22 randomized studies targeting child safety practices indicates that most efficient interventions are those combining health education, behaviour change strategies, and “reinforcement”.²⁷

Our results reflect the same trends as those found in Quebec: the relative risk between groups in terms of safety changes related to provided devices was 1.86 in Quebec and 1.56 in France. Concerning devices not provided in the kit, the relative risk was significantly different from 1, but with a larger difference (2.55 in Quebec, 1.54 in France). However four years separated the control group’s visits and those of the experimental group in Quebec.

In situations supposedly of high risk—single parent families, low level of education—kit safety device results are superimposed on the whole sample, resulting in a highly significant difference between groups. This is not the case for devices not provided and we conclude that free safety devices are probably crucial for implementation of home safety changes. We also observed that the impact of the kit was lower for families with at least two children, probably because parents were already better informed. But the sample is small and the results may be difficult to extrapolate. As in the study by Leduc,¹⁸ we observed in the experimental group a “multiplying effect” of additional safety improvements not related to the provided devices and not explicitly discussed at the first visit. It appears the safety kit increases the global sensitivity of families to prevention.

Sometimes health professionals installed or helped families to install devices during the second visit. This a reality in the field and must be taken into account in the planning of an effective program. An additional reality is that visits to some dangerous dwellings often prompted the installation of a smoke detector.

The dangerous or safe nature of each home situation was defined at the end of the questionnaire; this may reduce the risk of interviewer bias linked to both subjective assessment of items and knowledge of group assignment (control or experimental).

The success of this region-wide pilot study was probably facilitated by previous injury awareness campaigns and by the multidisciplinary safe community program in Boulogne-Billancourt.^19,20 To promote this type of intervention on a large scale requires support at the local level. A cartographic analysis of injuries, similar to that of Boulogne-Billancourt (to be published), may also facilitate the targeting of families in high risk areas.

While it is probable that this type of program results in injury reduction, we did not measure its effect on injury occurrence. Before nationwide implementation, the impact of the program must be assessed in terms of reduced morbidity. A cost outcome analysis must also be conducted in addition to addressing ethical issues concerning populations who do not receive the kit.