Objective: To assess (a) the effect of home safety education and the provision of safety equipment on poison-prevention practices and poisoning rates, and (b) whether the effect of interventions differs by social group.
Data sources: Medline, Embase, Cinahl, ASSIA, Psychinfo, Web of Science, plus other electronic sources and hand searching of conference abstracts and reference lists. Authors of included studies were asked to supply individual participant data.
Review methods: Randomised controlled trials, non-randomised controlled trials and controlled before-and-after studies, with participants aged ⩽19 years, providing home safety education with or without free or subsidised safety equipment and reporting poison-prevention practices or poisoning incidents were included. Pooled odds ratios and pooled rate ratios were estimated, and meta-regression estimated intervention effects by child age, gender and social variables.
Results: Home safety interventions increased safe storage of medicines (OR 1.57, 95% CI 1.22 to 2.02) and cleaning products (OR 1.63, 95% CI 1.22 to 2.17), the possession of syrup of ipecac (OR 3.34, 95% CI 1.50 to 7.41), and having poison control centre numbers accessible (OR 3.67, 95% CI 1.84 to 7.33). There was a lack of evidence on poisoning rates (rate ratio 1.03, 95% CI 0.78 to 1.36) and no consistent evidence that intervention effects differed by child age, gender or social group.
Conclusions: Home safety education and the provision of safety equipment improve poison-prevention practices, but the impact on poisoning rates is unclear. Such interventions are unlikely to widen inequalities in childhood poisoning-prevention practices.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Poisoning is the fifth most common fatal injury in children under 5 worldwide.1 Poisoning resulted in more than 31 000 emergency department attendances in the UK in 20022 and more than 6000 hospital admissions in children aged under 15 years in England and Wales in 2005/6.3
Medicines account for more than 50%, and household products for around one-third, of medically attended poisonings.4 Young children are at particular risk, with those under 2 being at greater risk from household products than medicines, and the reverse for children over 2.5 Boys are at greater risk of poisoning than girls until adolescence.1 2 6–8 There are steep social gradients in childhood poisoning mortality9 and morbidity,4 10 and parental unemployment5 and single parenthood11 are associated with a higher risk of childhood poisoning. Poison-prevention practices also vary by social group, with families with only one child, those with mothers in employment, those with parents with higher levels of education, and those from ethnic minority groups being less likely to store potentially poisonous substances safely at home.12 13
There is a wide range of strategies for preventing poisoning including child-resistant packaging,14 locks on cabinets or drawers, storage out of reach, not transferring substances from original containers, replacing in the usual storage place immediately after use,15 safely disposing of unwanted medicines,15 using substances containing bittering agents,16 educating children about the dangers of substances,17 using poison control centres (PCCs),15 18 and, in some countries, emetics have been used to induce vomiting after a poisoning.19
Most strategies rely to a greater or lesser degree on educating parents, yet few systematic reviews have specifically investigated the effect of educational interventions on child poison prevention. To address this lack of evidence, we have undertaken a systematic review and meta-analysis evaluating the effect of home safety education and the provision of safety equipment on poison-prevention practices and poisoning rates. In addition, in view of the social inequalities in poisoning, we have investigated whether the effect of such interventions differs by social group.
The findings reported in this paper comprise part of a Cochrane systematic review and meta-analysis covering home safety education and safety equipment for the prevention of all injuries in childhood.20
We searched a range of sources as shown in table 1. We contacted authors of included studies, and surveyed Sure Start and Home Start schemes and Fire and Rescue Services in the UK, the Royal Society for the Prevention of Accidents, National Children’s Homes charity and the Community Practitioners and Health Visitors Association to ask about relevant ongoing or unpublished studies. The search strategy is described elsewhere.20 Articles in all languages were retrieved and translated as necessary.
Selection and study characteristics
We included randomised controlled trials (RCTs), non-RCTs (e.g. quasi-randomised controlled trials) and controlled before-and-after studies (CBAs), with participants aged ⩽19 years, which provided home safety education with or without free, low-cost or discounted safety equipment and reported poison-prevention practices or self-reported or medically attended poisoning. The practices selected as outcomes for analysis were: safe storage of medicines (defined as stored at adult eye level or above, in locked cupboards/drawers/cabinets or inaccessible to child), safe storage of cleaning products (defined as above), possession of syrup of ipecac (to induce vomiting), and having the PCC number in an accessible place. We excluded studies that evaluated PCCs using contact rates as their outcome, as increased contact rates could represent greater awareness and use of the centre or a higher incidence of poisoning. Community-based trials with multi-faceted interventions were eligible if they included individual or group home safety education.
Quality was assessed independently by two researchers, blind to author, institution and study results. We used blinding of outcome assessment and completeness of follow-up for all designs, allocation concealment for RCTs, and assessment of the distribution of potential confounders for non-randomised studies. Quality assessment was compared between two researchers for the first 41 studies for all home safety and injury outcomes (see Results) and subsequently all studies were assessed by two researchers, with disagreements resolved by referral to a third researcher.
Titles and abstracts were assessed for inclusion independently by two researchers, with disagreements resolved by referral to a third researcher. Articles without abstracts were retrieved in full. To maximise information available for analyses, we attempted to obtain individual participant data (IPD) from authors, and, where obtained, datasets were formatted across studies to ensure uniformity. Where IPD were unavailable, data were extracted from published articles, on to standard extraction forms independently by two researchers, blind to author and institution. Data extraction was compared between researchers for the first 53 studies covering all home safety and injury outcomes, with a high level of agreement found (see Results), hence data were extracted on the remaining studies by one researcher.
Quantitative data synthesis
Meta-analyses were undertaken where three or more studies reported the same outcome. Pooled odds ratios (ORs) were estimated from summary binary data using a random effects model, or a fixed effect model if there were fewer than five studies without significant between-study heterogeneity. For rate outcomes, study-specific effect sizes and standard errors were estimated from summary data using Poisson regression and entered into the meta-analysis. Where necessary, studies with a clustered design were adjusted by dividing the number of participants with an outcome and the total number of participants in each arm by the design effect, and the number of events and person-years by the variance inflation factor. Intraclass correlation coefficients (ICCs) and coefficients of variation were calculated from IPD or obtained from published data. Where coefficients of variation were unavailable, we used a value of 0.25.21 Analyses were undertaken using Stata V8.
The effect of interventions by gender, ethnic group, single parenthood and parental unemployment was examined using IPD and study summary level data simultaneously in a novel meta-regression model.22 The effect of age was assessed using only IPD. Analyses were undertaken using WinBUGS V1.4.1.
Cases with missing values were excluded from analyses. Heterogeneity was assessed by χ2 tests and the I2 statistic and explored by meta-regression (described above) and by a priori specified subgroup analyses (the provision of safety equipment, follow-up period, setting, study design and study quality). Where meta-analyses included seven or more studies, publication bias was assessed using funnel plots and Egger’s test. The trim-and-fill method23 estimated the likely effect of any statistically significant publication bias. Sensitivity analyses assessed the robustness of meta-analysis findings with respect to the individual contribution of each study and the size of the ICC for clustered studies.
Ethics approval was obtained from Nottingham Local Research Ethics Committee (reference number 04/Q2404/25).
Trial flow and study characteristics
Figure 1 shows the process of study identification and selection. Eighteen studies were included in at least one of the meta-analyses, 14 of which were RCTs,24–37 two were non-RCTs,38 39 and two were CBAs.40 41 IPD were obtained for nine studies.24 25 30–32 36 37 39 40Table 2 shows the characteristics of included studies. In 15 studies, the intervention included the provision of free or subsidised safety equipment.24–26 28–30 32–39 41 In nine studies, the intervention was provided mainly in a clinical setting.24 27 28 30–33 37 39 In eight studies, the intervention was provided by health professionals,24 27 30 33 35–37 39 in three by researchers,25 29 38 in two studies the intervention comprised part of a multifaceted community injury prevention programme,40 41 and the remaining studies had interventions comprising a video-tape,28 computer-generated advice,31 a mailed intervention,34 and interventions provided by lay workers32 or educational paraprofessionals.26
Reliability of data extraction and quality assessment
Nearly all the data items extracted for the first 53 studies covering all home safety and injury outcomes (1386/1424; 97.3%) were identical. High levels of agreement were found for satisfying inclusion criteria overall (κ = 1.00), and for inclusion criteria relating to study design (κ = 1.00), age of study participants (κ = 1.00), interventions (κ = 0.85) and outcomes (κ = 0.84). Agreement on study quality was relatively low for some quality markers (RCTs: allocation concealment κ = 0.53; blinding of outcome assessment κ = 0.92; completeness of follow-up κ = 0.92. Non-RCTs: blinding of outcome assessment κ = 0.25; completeness of follow-up κ = 0.23; distribution of confounders κ = 0.69).
Quantitative data synthesis
Effect of home safety interventions
Home safety interventions were effective in increasing safe storage of medicines (fig 2; OR 1.57, 95% CI 1.22 to 2.02) and cleaning products (fig 3; OR 1.63, 95% CI 1.22 to 2.17), the possession of syrup of ipecac (fig 4; OR 3.34, 95% CI 1.50 to 7.41), and having the PCC number accessible (fig 5; OR 3.67, 95% CI 1.84 to 7.33). The number needed to treat to result in one additional family behaving safely was 14 (95% CI 10 to 29) for storage of medicines, 9 (95% CI 6 to 21) for storage of cleaning products, 4 (95% CI 2 to 11) for having syrup of ipecac, and 3 (95% CI 2 to 7) for having the PCC number accessible.
Effect sizes varied significantly between studies for storage of cleaning products, possession of ipecac, and having the PCC number accessible. They tended to be greater when free or subsidised safety equipment was provided with education than when the intervention comprised education alone (eg, safe storage of cleaning products: OR 1.90, 95% CI 1.25 to 2.89 (equipment and education) vs OR 1.12, 95% CI 0.89 to 1.41 (education alone); possession of ipecac: OR 10.41, 95% CI 2.40 to 45.09 (equipment and education) vs OR 1.77, 95% CI 1.08 to 2.91 (education alone)). Effect sizes also tended to be larger when interventions were delivered at home rather than in a clinical setting (eg, storage of cleaning products: OR 2.31, 95% CI 1.00 to 5.32 (home) vs OR 1.29, 95% CI 1.09 to 1.53 (clinical setting); possession of ipecac: OR 5.45, 95% CI 1.22 to 24.32 (home) vs OR 2.02, 95% CI 1.08 to 3.75 (clinical setting); having the PCC number accessible: OR 5.62, 95% CI 1.30 to 24.37 (home) vs OR 2.88, 95% CI 1.23 to 6.73 (clinical setting)). There was no consistent relationship between effect sizes and length of follow-up.
There was a lack of evidence that home safety interventions were effective in reducing rates of poisoning (rate ratio 1.03, 95% CI 0.78 to 1.36) (fig 6). This analysis included only three studies with 5980 person-years in the combined intervention arms and 6137 person-years in the combined control arms.
Effect of study quality and adjusting for clustering
The findings were robust to excluding non-randomised designs (figs 3–5) and to excluding each study in turn for safe storage of cleaning products, possession of ipecac, and having the PCC number accessible. Effect sizes varied by study quality only for adequate allocation concealment for having the PCC number accessible (OR 1.59, 95% CI 0.95 to 2.68 (adequate concealment) vs OR 7.02, 95% CI 3.18 to 15.50 (inadequate/unclear concealment)). All effect sizes were robust to assumptions about the size of the ICC.
There was evidence of publication bias only for safe storage of cleaning products (Egger’s test: regression coefficient 2.01 (SE 0.70), p = 0.019). Use of the trim-and-fill technique resulted in a decrease in the effect size from 1.63 (95% CI 1.22 to 2.17) to 1.17 (95% CI 0.87 to 1.57).
Does the effect of home safety interventions vary by child age, gender or social group?
Interventions were more effective in increasing having the PCC number accessible among families with at least one unemployed parent than among families with both parents employed (OR 7.59 vs 1.47, respectively; ratio of ORs: 5.17, 95% credible interval 1.29 to 20.55). The effect of the interventions did not appear to vary by any of the other covariates.
Home safety education that includes poison prevention, especially where cupboard locks, ipecac and PCC number stickers are provided free or at low cost, is effective in increasing safe storage of medicines and cleaning products, the possession of ipecac, and having the PCC number accessible. We did not find any studies that reported outcomes such as transferring substances from original containers, putting substances away immediately after use, disposing of unwanted medicines safely, or use of bittering agents. Importantly, we have also shown that such interventions are unlikely to widen existing inequalities in childhood poison-prevention practices.
Strengths and weaknesses of the study
This is the largest and most comprehensive systematic review and meta-analysis in this area to date. It is the first meta-analysis in this field to obtain IPD and to use these, via a novel statistical model, to quantify the effect of home safety interventions by child age, gender and a range of variables describing social group.
Significant heterogeneity was found among effect sizes for three of the meta-analyses. In two of these, the heterogeneity related to the extent to which an intervention was effective, not whether it was, or was not, effective. Whether safety equipment was provided and the setting in which the intervention was delivered explained some of the heterogeneity for some outcomes. The meta-regression analyses suggest that some heterogeneity in the PCC analysis may also be explained by parental unemployment. It is also possible that differences in the content and delivery of the interventions may explain some of the observed heterogeneity.
There was evidence of a publication bias for safe storage of cleaning products, suggesting that we may have failed to find and include some small studies with negative findings. Alternatively this finding could be a type 1 error arising from multiple significance testing. The results relating to storage of cleaning products should therefore be interpreted with some caution.
It has never been recommended in the UK for parents to give syrup of ipecac to children to induce vomiting after ingestion of poisonous substances. In recent years, the American Academy of Pediatrics has stopped recommending its use in the USA.15 19 Our findings on ipecac possession may therefore only be relevant to countries that still recommend that parents keep ipecac at home for such use.
Few of the studies included in our meta-analysis reported data on poisonings. It is likely that our analyses were underpowered to detect anything but large (and possibly implausible) treatment effects. This is confirmed by post hoc power calculations using simulation with 400 iterations,42 which suggest that 21 500 person-years per arm are required to have 80% power to detect a 20% reduction in the rate of poisoning. The 95% CI for poisonings shows that plausible treatment effects range from a 22% reduction in poisonings to a 36% increase.
Although the vast majority of poisoning deaths occur in low and middle income countries,43 all of the studies included in our review report interventions implemented in higher income countries. The transferability of such interventions to low and middle income countries remains to be evaluated, as do interventions targeting the specific poisoning problems faced by these countries. As such, our findings may not be generalisable beyond higher income countries.
Comparisons with previous research
We have only been able to find one previous meta-analysis44 and one narrative systematic review45 of childhood poison-prevention interventions. The meta-analysis focused on home safety education provided in a clinical setting and estimated a pooled OR from three RCTs for safe storage of cleaning products of 1.8 (confidence intervals not reported). The authors concluded that clinical counselling has little effect on most home safety practices designed to childproof the home. Our results are substantially more positive than these, as we found significant effects for all four poison-prevention practices that we examined, with ORs ranging from 1.6 to 3.7. This is likely to reflect our inclusion of a larger number of studies, reporting a greater range of outcomes. The narrative review focused on community-based poison-prevention programmes, and found only four studies that reported injury outcomes. Only one study found a significant effect, which was a reduction in kerosene ingestion after a container-distribution programme.46 The authors concluded that there is insufficient evidence on the effectiveness of community-based poison-prevention programmes, and our findings are consistent with these.45
What is already known on this topic
Poisoning causes considerable mortality in low and middle income countries and considerable morbidity and healthcare resource use in higher income countries.
There are steep social gradients in poisoning mortality and morbidity.
Most poison-prevention strategies rely on parental education to a greater or lesser extent, but there have been few systematic reviews and meta-analyses of educational approaches to poison prevention.
There is little evidence assessing whether poison-prevention interventions differ by social group.
What this study adds
Home safety education and the provision of safety equipment improves a range of poison-prevention practices, but its impact on poisoning is unclear.
There was no consistent evidence that educational interventions are likely to widen inequalities in poison-prevention practices.
Further research is needed to evaluate poison-prevention interventions.
There is little work with which to compare our findings in relation to effects by demographic and social variables. Three studies included in our review reported subgroup analyses comparing the treatment effect by a range of factors.24 36 37 Only one found evidence of a differential effect, showing a greater effect of the intervention in single-parent families, smaller families and those with lower levels of education.36 This is consistent with our findings that, for most outcomes, there was no differential effect for a range of demographic and social variables, and, when a differential effect was found, the effect was greater in the group considered to be at higher risk.
Explanation of our findings
Although the interventions we reviewed improved poison-prevention practices, we were unable to show a reduction in poisoning rates. There are several possible explanations for this. Firstly, the three studies reporting poisoning rates that we included in our meta-analysis had 5980 person-years in the combined intervention arms and 6137 person-years in the combined control arms. Our post hoc power calculation showed that our analyses were substantially underpowered to detect a 20% reduction in the poisoning rate. Such a reduction may also be over optimistic; six of the eight studies in the meta-analysis of safe storage of medicines had absolute differences in the percentage of medicines stored safely of less than 10% between treatment groups, as did eight of the 11 in the meta-analysis of safe storage of cleaning products. Secondly, the studies included in our meta-analysis of poisoning rates combined all poisonings in one outcome measure, which may have included poisonings that could not plausibly be prevented by the intervention. For example, cupboard locks fitted to kitchen and bathroom cabinets cannot prevent poisoning from medicines stored in a parent’s handbag or garden chemicals stored in an outside shed. This lack of specificity in outcome definition will have limited the potential for showing a positive effect of the intervention. Thirdly, the provision of poison-prevention education and equipment may not reduce childhood poisoning rates. Our findings suggest that there is an absence of evidence in this area, rather than evidence of absence of an effect.
We did not find consistent evidence that home safety education is likely to widen inequalities in poison-prevention practices, which may reflect the provision of free or subsidised locks, PCC number stickers, and syrup of ipecac in many studies, which will have reduced the cost of obtaining these items for parents.
Implications for research and practice
Further research is required to address the question of whether poison-prevention interventions reduce poisoning rates. Very large RCTs or multiple smaller randomised trials that are sufficiently clinically homogeneous to meta-analyse would provide us with the best evidence. Such trials would be extremely expensive, logistically challenging, and, in the current research climate, unlikely to be funded. Furthermore, the UK government has recently agreed to provide £18 million for home safety equipment schemes as part of the Children’s Plan.47 This is likely to reduce uncertainty about the effectiveness of such interventions among health and social care practitioners and families, creating ethical issues and limiting the feasibility of conducting such trials. Under these circumstances, rigorous observational studies that control or adjust for a wide range of confounding factors may be the more feasible alternative.
Where child health and social care providers already provide poison-prevention education and equipment, they should continue to do so, as we have not shown that these are not effective in preventing childhood poisoning. Further research on the effectiveness of poison-prevention education and equipment is needed so that decisions about poison-prevention programmes can be evidence based.
We thank the researchers who shared their individual participant data with us. This work was undertaken by the Universities of Nottingham and Leicester. Funding was provided by the Department of Health. The views expressed in the publication are those of the authors and not necessarily those of the Department of Health. This paper is based on a Cochrane Review published in The Cochrane Library 2007, Issue 1 (see www.thecochranelibrary.com for information). Cochrane Reviews are regularly updated as new evidence emerges and in response to feedback, and The Cochrane Library should be consulted for the most recent version of the review.
Competing interests: DK, CC, MW and AW undertook some of the trials included in this review. There are no other competing interests.