Globally more than 35 000 children die each year from poisoning,1 making it the fifth most common fatal injury in children under 5. Unintentional poisoning is a particular concern in low and middle income countries, which account for 94% of all poisoning deaths.2 In higher income countries, unintentional poisoning is responsible for considerable morbidity and healthcare resource use.3^–7 Such poisonings mostly occur within the home,8^–10 with the highest rates reported in 1–3 year olds.510 Medicines are responsible for about two-thirds of cases, with household products accounting for the remainder.511

Although legislating for child-resistant closures (CRCs) and packaging for medicines has resulted in a reduction in childhood poisoning deaths12^–14 and hospital admissions,15 these are not necessarily “child-proof”. Some CRCs are difficult for adults to open, and may not be closed properly, or products may be transferred to containers without CRCs. Even when CRCs are properly closed, some children can open them.16 It is possible that the effectiveness of CRCs and other child-resistant packaging could be enhanced by safer practices for storing and using substances,10 and further evaluation of the impact of such strategies is required.

Self-reported practices for preventing unintentional poisoning are commonly used to explore poisoning risk and to evaluate poison prevention interventions. There is concern that self reports may overestimate safe practices, and some evidence to support this in studies assessing home safety practices.17^–19 Few studies have validated self-reported practices for preventing unintentional poisoning. One study validated reported possession of ipecac and found a high sensitivity, specificity, and negative predictive value and a relatively high positive predictive value.17 A second study validated reported storage of all medicines combined and all household products combined in kitchens and bathrooms and found that the sensitivity and negative predictive value varied between medicines and cleaning products and between rooms, but all had high specificities and positive predictive values.20 No studies to date have examined the validity of self-reported storage of specific medicines or cleaning products. As part of a larger study exploring practices for preventing unintentional poisoning in families with pre-school children, we validated self-reported possession and storage of a range of substances within the home.

METHODS

All health visitors in Nottingham (n = 149) were asked to express interest in participating in the study exploring poison prevention practices. Fifty-six (38%) expressed interest, and they were stratified into three groups based on the Townsend deprivation score21 of the general practice to which they were attached. Health visitors were randomly chosen within strata using random numbers and were invited to participate.

A questionnaire was designed to measure sociodemographic and economic characteristics, possession and storage of a range of medicines and cleaning products, parental perceptions of the harmfulness of a substance on ingestion, preventive strategies, and actions parents may take if a child ingested a substance. The questions relating to medicines used common terms such as “painkillers for children”, “antibiotics”, “iron tablets”, and “cough mixture”. We did not use generic or brand names, as there are numerous brand names for many medicines and we felt parents were more likely to know the common term than the generic or brand name. The questionnaire was piloted on 60 families not participating in the study. Twenty-three health visitors sent a postal questionnaire to parents (or carers; hereafter referred to as parents) of children aged 12–35 months (n = 1259) on their caseloads, between May 2005 and August 2006. Parents returned the questionnaire directly to the researchers and were asked to provide contact details if they were willing to participate in further research. A total of 763 questionnaires were returned, of which 170 were returned between May and August 2006, and 43 of these 170 (25%) provided contact details. They were contacted by telephone or letter and were not informed that the purpose of the visit was to validate the previously completed questionnaire, but were told that a home safety survey about childhood poisonings was being carried out. The visits were organized as soon as possible after the questionnaire was returned.

A checklist was designed for completion during the home visit which included possession and storage of medicines (painkillers for adults and children, antibiotics, cough mixtures, and iron tablets) and household products (bleach, toilet cleaner, essential oils, ant or rat killer, dishwasher tablets, and turpentine). These substances were chosen to include common agents, including prescribed and over-the-counter drugs, those with several possible storage locations, and substances of varying toxicity. The checklist and home visit was piloted on six families.

At the home visit, parents were asked to lead the researcher to the locations of each substance, described using the common terms given above. The researcher was medically qualified and was able to distinguish which medicines they were being shown by referring to the drug name. Some rooms were not observed if the parents requested this—for example, because a child was sleeping in the room or because showing the researcher a garage or garden shed would require going outside. In such situations, these cases have been excluded from the analysis. The visits took 10–30 min, and the researcher was blinded to the self-reported data on the questionnaire.

To detect a sensitivity of 80% for safe storage, with 95% CI from 60% to 100%, and estimated safe storage of medicines of 92% and cleaning products of 67%,20 17 home observations were required for storage of medicines and 30 for storage of cleaning products.

Data were analyzed using Stata V9. Comparisons between characteristics of families who had a home visit and those who only completed the questionnaire were made using random effects logistic regression to account for clustering by health visitor. Sensitivities, specificities, and predictive values were calculated using home observations as the “gold standard”. These were calculated for (a) possession of substances, (b) safe storage of substances, defined as stored at adult eye level or above or in locked cupboards or drawers,20 and (c) low exposure to substances, defined as storing the substance safely or not possessing it. Cases with missing data were excluded from analyses (fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Summary of definitions.

Ethical approval was obtained from Nottingham Research Ethics Committee (reference number P1080301).

RESULTS

The response rate to the postal questionnaire was 61% (763/1259). Thirty home visits were undertaken between June and August 2006. The median length of time between researchers receiving a completed questionnaire and undertaking the home visit was 27 days (interquartile range 15–34). Table 1 reports the characteristics of families participating in the home observations and families completing the questionnaire that did not have a home observation. Characteristics did not differ significantly between these two groups of families.

Table 2 reports sensitivity, specificity, and predictive value for self-reported possession of substances. Most families were observed to have pain killers, cough medicine, bleach, dishwasher tablets, and toilet cleaner. Few families had iron tablets or antibiotics. The sensitivity was high for most substances, except for iron tablets, antibiotics, and essential oils. When nearly all families had medicines or household products, 95% CIs were very wide around specificities and negative predictive values, making these estimates unreliable. Positive predictive values were high, except for antibiotics and iron tablets, which few families possessed on observation.

Table 3 reports sensitivity, specificity, and predictive value for self-reported storage of substances. Most families were observed to store pain killers, cough medicines, essential oils, turpentine, and ant or rat killer safely. Iron tablets, antibiotics, bleach, dishwasher tablets, and toilet cleaner were less likely to be stored safely. Sensitivities were high for all substances, with high or relatively high positive predictive values. Again, when few families did not store substances safely, 95% CIs were very wide around specificities and negative predictive values, making estimates unreliable. Specificities were lower than sensitivities for all household products, but negative predictive values were high for household products.

Table 4 reports sensitivity, specificity, and predictive value for low exposure to substances. Most families were observed to have low exposure to medicines, essential oils, turpentine, and ant or rat killer. Between one-third and one-half of families were observed to have high exposure to bleach, dishwasher tablets, and toilet cleaner. Sensitivities and positive predictive values were high for virtually all substances. Specificities and negative predictive values were low, with wide 95% CIs for all medicines and for essential oils, as very few families were observed to have high exposure. Sensitivities exceeded specificities for all substances. Negative predictive values for bleach, dishwasher tablets, and toilet cleaner were high.

DISCUSSION

Most families have a wide range of substances of varying toxicity in the home. Most store medicines safely or do not possess certain medicines, reducing their children’s exposure to such substances. Household products such as bleach, dishwasher tablets, and toilet cleaner are less likely to be stored safely, increasing children’s exposure to these substances.

The sensitivity, specificity, and predictive value of self-reported possession of substances, safe storage of substances, and exposure to substances varies between substances. This suggests that studies measuring practices for preventing unintentional poisoning should ask specific questions about different potentially hazardous substances, rather than asking questions about possession or storage of all medicines or all household products.

Self-reported possession and safe storage of substances are useful for identifying those families who do possess substances and store them safely. The high positive predictive values suggest that there is little over-reporting of safe storage for most medicines or bleach. This is consistent with the findings of other validation studies, which found high positive predictive values for safe storage of medicines and cleaning products20 and relatively high positive predictive value for possession of ipecac syrup.17 We found that combining data on possession and storage to assess exposure to substances produces high sensitivities and positive predictive values for virtually all substances, and (when there are sufficient data) produces high negative predictive values.

The low sensitivities and positive predictive values for possession of iron tablets and antibiotics suggest that families find it difficult to accurately report possession of these substances. Possible explanations for this include lack of awareness about the contents of medicines (eg, antibiotics used to treat acne or whether multivitamins contain iron) or failure to dispose of medicines once the course of treatment is finished (as opposed to pain killers or cough medicines which are kept for treating new occurrences of symptoms).

Over-reporting of safe storage was most common for dishwasher tablets and toilet cleaner. This may have occurred because these substances are used frequently and families may have reported their “usual” place of storage, despite being asked where the products were on the day of questionnaire completion. Our findings suggest that most families who report storing dishwasher tablets or toilet cleaner unsafely will be correctly identified, but care should be taken in interpreting self-reported safe storage of these substances.

Limitations of the study

There are several ways in which bias could arise in a study such as this. A response bias could have occurred whereby families who responded to the questionnaire differed in terms of poison prevention practices from non-responders. However, there are difficulties in assessing whether this is the case, as our sample covers four local authority areas for which 2001 census data are available, but census data are not available for these areas combined. Families from a black or minority ethnic group do not appear to be under-represented in our sample compared with census data for the four local authorities (range 3.8–15.1%), nor do single-parent families (range 4.5–9.9%), non-owner occupiers (range 18.8–50.0%), or families living in overcrowded conditions (range 3.1–9.0%). Our sample has a higher proportion of respondents with a degree (range 17.5–30.4%) and a lower proportion without access to a car (range 16.8–44.9%).22^–25 Although it is therefore possible that non-responders to the questionnaire may have different poison prevention practices from responders, we have previously found that families with children aged 3–12 months in Nottingham who responded to a safety questionnaire did not differ from non-responders in terms of storage practices for medicines or cleaning products.26

The second possible source of response bias is that families who agreed to the home visit may differ from those who only completed the questionnaire. Although we did not find any significant differences between sociodemographic and economic characteristics between these two groups, these analyses will have had limited power. A larger study in a similar population in Nottingham also failed to find differences in self-reported storage of medicines and household products among families agreeing to a home observation and those who only completed a questionnaire.20 Although this is reassuring, the question of whether those who agree to a home visit report poison prevention practices more accurately than those who only complete a questionnaire remains unknown.

We were unable to produce reliable estimates of the specificity and negative predictive value for some substances because of very small numbers of families not having, or not storing, them safely. Further work on a larger sample is required to estimate specificities and negative predictive values for these substances. In addition, some parents requested that we did not observe storage of substances in garages or sheds (eg, turpentine or ant or rat killer). This precluded calculation of their sensitivity, specificity, and predictive value because of small numbers. It is possible that storage practices may have changed between completion of the questionnaire and the home visit taking place, particularly for those prescribed as a course (eg, antibiotics or iron tablets), rather than for long-term, or repeated intermittent use (eg, pain killers). This may partly explain the discrepancy between observed and reported possession of iron tablets. Interestingly, this does not appear to apply to antibiotics, as all parents reporting antibiotics on the questionnaire were observed to have them on the home visit.

Finally, although the researcher was blinded to the responses on the questionnaire, the parent may have been able to remember their responses and this may have influenced the substances and storage places shown to the researcher on the home visit. This is unfortunately a limitation of all validation studies using home observations to validate self-reported safety practices.

IMPLICATIONS FOR INJURY PREVENTION PRACTICE

Researchers studying poison prevention practices should ask specific questions about possession and storage of different substances. Combining data on possession and storage to assess exposure to substances produces high sensitivities and positive predictive values for virtually all substances, suggesting that this method may be sufficiently accurate to identify families with safer poison prevention practices.

• Download figure
• Open in new tab
• Download powerpoint