The current study investigated a broad range of potential risk factors for unintentional poisoning in children aged 1–3 years, including three behavioural factors- child compliance, caregiver supervision and the nature of mother-child interaction. The methodology employed appropriate questionnaire, performance and observational methods to explore these factors. In addition, the study excluded passive poisoning events (i.e., not due to self-access by the child) from the case group and controlled for potential confounders and sources of bias by including control groups from different populations.
The univariate analysis identified 27 factors that were associated with poisoning risk in one or more case–control pairs. These factors came from all four domains assessed (child, mother, environment and mother-child interaction) and six of the 27 factors were eligible for all three multivariate models. The final models included the following factors as significant predictors for unintentional poisoning in children aged 1–3 years: maternal use of positive control, parenting stress, GHQ psychiatric caseness, supervision during risk taking activities and accessible medicinal substance storage locations in the bathroom.
Mothers of cases used more positive control than mothers in all three control groups during the observation of mother-child interactions. The presence of this factor in all three final models adds weight to this finding; however, this finding could also be related to the age differences between the case and three control groups. Mothers of children who had been poisoned had younger children who may have needed more help than the older children in the control groups. Interestingly, the differences between the groups involved positive and encouraging controlling of children. There was little evidence of more negative control in the poisoning or any other groups.
The caregiver-child relationship plays an important role in helping children develop the ability to regulate their behaviour . Research has shown that caregiver-child relationships characterised by a mutually responsive orientation can facilitate the process of self-regulation . In particular, positive interactions that are mutually responsive have been demonstrated to result in better compliance and increase the likelihood that maternal rules are internalised by the child . Child compliance and internalisation of maternal rules are two aspects of development that could potentially reduce a young child’s poisoning risk . Mothers in the poisoning group also exhibited more positive affect in both tasks than mothers in the other groups. This difference may reflect maternal behaviour used to encourage young children to complete tasks which may be less necessary with older children. In addition, the children in the poisoning group showed signs of less responsiveness and less dyadic cooperation than children in the control groups. This may explain further the univariate finding regarding more positive affect in mothers of poisoning cases, but also may indicate the lack of a mutually responsive orientation which has implications for the development of child self-regulatory behaviour.
Close interactions between caregivers and children may also facilitate caregiver knowledge regarding their child’s development. This knowledge can help caregiver’s anticipate their child’s risk of exposure to poisons in their environment and implement appropriate prevention strategies . Further research is warranted regarding the link between caregiver-child relationships, child self-regulation and unintentional poisoning risk.
Mothers in the poisoning group showed less parental stress than mothers in all three control groups. This measure of stress was not what might be predicted, particularly since the parental stress questionnaire was administered within 30 days of attendance at the Emergency Department for the poisoning event. It might be expected that a potentially life threatening event for their child would make caregivers more worried about their supervision skills; however, the results do not suggest this. Even if some of the caregivers had become more worried about parenting after the poisoning incident, their levels of parenting stress were still not as high as the control groups; parenting stress was high in the healthy group who had not had a recent episode relating to their child’s health and well-being.
In contrast, the mothers of children who had been poisoned had higher odds of being classified as psychologically distressed (based on the GHQ). This result is in line with a finding by Beautrais et al. . Beautrais and colleagues showed that children of mothers who were prescribed anti-depressants and tranquilisers had higher rates of poisoning. In this study, mothers in the poisoning group may have other sources of distress in their lives such that they did not see parenting as a particular problem. It is also possible that mothers in the poisoning group experienced less parenting stress as almost all of the children in the case group were first-born and had no siblings.
The low level of parenting stress reported by mothers in the poisoning group may have indirectly contributed to their children’s unintentional poisoning as it is possible this reflects the caregivers’ approach to supervision and poisons storage practices (i.e., more relaxed). Almost all of the poisonings in the current study occurred when, in most cases, the substance was located less than 1.4 metres from the ground in a temporary storage place, and the caregiver was in another room. Inadequate poisons storage practices and supervision may also reflect inaccurate caregiver perception of the level of development [24, 54]. The age of the case group and the circumstances of their poisoning events suggest that the caregivers were not yet aware that the children could or would try to access a substance stored in an unsafe manner.
Children in the poisoning group were supervised less closely during risk taking activities than children in the injury group. It is possible that this finding reflects supervision differences between boys and girls  as the poisoning group consisted mainly of boys. In addition to differences in how closely boys and girls are supervised, it may be that poisoning events occur when the supervision is not continuous, due to a distraction [34, 55] or lapse in attention . Descriptive data from eight of the nine poisoning events involving children in this study indicate that the poisoning events occurred when caregivers were not directly supervising their children (in another room) or were distracted.
The current study also explored the relationship between poisoning and a number of child developmental variables - fine and gross motor skills, language skills, compliance ability, positive and negative effect, independence, responsiveness and non-compliance. It found an association between poisoning and all of the developmental variables investigated except positive and negative effect for one or more case–control pairs. From the univariate analysis, children who were poisoned were more likely to score as less advanced in terms of fine and gross motor and language skills and be less compliant, less responsive and less independent in their behaviour compared to controls. In addition, gross motor and language skills were assessed using an age-standardised measure (based on the Denver II) and associated with poisoning for all three case–control pairs. Yet, none of the child developmental variables remained in the final models.
Age and sex differences were present between children in the case and three control groups in the study and it is possible that these differences may have influenced the factors identified as significant predictors of unintentional poisoning in children aged 1–3 years. Previous studies in NSW identified a link between age and sex and poisoning risk [4, 14] and this link was attributed to developmental differences. As children of the same age and sex can differ markedly in terms of their level of development, the current study attempted to control for the confounding effect of these two variables by enrolling equal numbers of males and females into the case and control groups for each single year of age and then matching them by age and sex. However, the desired sample size and age-sex composition was only achieved for the three control groups; the nine children in the case group were younger than the controls and predominantly male. Consequently, the number of children in each age-sex group differed between the case and three control groups and matching could not be done; instead, the age and sex differences were controlled during the analysis. Thus, the smaller sample size and lack of matching could have implications for the interpretation of findings in this study as well as the ability of the study to identify predictors for unintentional poisoning in children aged 1–3 years, such as aspects of development that may increase poisoning risk.
Response rates were very low for all four study groups, but the number of children recruited into the poisoning group was very small. Even though the composition of the case group is consistent with the results from the population studies in NSW that showed that children aged 1–2 years and males have the highest rates of poisoning among children aged 0–4 years , unintentional poisoning presentations represented a much rarer occurrence compared to the number of children eligible for the control groups. The difficulty in enrolling cases and controls may reflect the type of enrolment procedure used. The current study was required to employ an ‘opt-in’ enrolment procedure to gain ethics approval, yet this enrolment approach has been linked to poor response rates [57, 58]. In addition, the very low response rates in the control groups may have resulted in a potential source of bias. Compared to the three control groups, most mothers in the poisoning group had only one child, all had a university level degree or higher, a higher percentage worked full-time and none smoked. It is not known whether the observed socio-demographic differences reflect true differences between poisoning cases and other populations or were biases introduced as a result of the ‘opt-in’ enrolment procedure, which may have impacted results by affecting the types of participants who enrolled in the study . Nonetheless, the presence of controls from different populations may have reduced the effect of potential bias in the study  as the univariate and multivariate analyses revealed the same finding for more than one model. Thus, while bias may have potentially affected the results in the study, the consistency of findings across the three models indicates that the effect of any such bias was minimal.
Overall, the results of this study draw attention to the importance of the caregiver-child relationship and caregiver influences in poisoning risk. Most children are walking well by age 14 months  allowing them to move about their environment. However, protective aspects of development, the cognitive ability to remember safety rules and the self-regulatory ability to stop themselves from accessing hazards, lag behind. Thus, physical development enables children to interact with hazards in their environment before they are able to understand the effect of their actions . Given this imbalance, the caregiver-child-relationship may provide an avenue by which caregivers of children aged 1–3 years can raise their awareness of the rapid developmental changes occurring in their children. This information can then be used to reduce their children’s unintentional poisoning risk by implementing developmentally appropriate supervision and poisons storage practices to prevent exposure to hazardous substances.
This study could not be definitive, however, about the involvement of some of the caregiver factors due to the small number of poisoning cases and the age differences between cases and controls. Nevertheless, this study has helped to redefine the best approaches to study designs needed to understand more about caregiver involvement and the role of caregiver-child interactions in unintentional poisoning events. Larger samples of children who have experienced a poisoning are needed, along with control groups of younger children in order to avoid confounding of developmental differences.