Article Text


Predictors of unintentional poisoning among children under 5 years of age in Karachi: a matched case–control study
  1. Bilal Ahmed1,
  2. Zafar Fatmi1,
  3. Amna R Siddiqui1,
  4. Abdul L Sheikh2
  1. 1Department of Community Health Sciences, Aga Khan University, Karachi, Pakistan
  2. 2Department of Pharmacy, Aga Khan University, Karachi, Pakistan
  1. Correspondence to Bilal Ahmed, Department of Medicine, Aga Khan University, Stadium Road, PO Box 3500, Karachi 74800, Pakistan; bilal.ahmed{at}


Objective Poisoning is the fourth leading cause of unintentional injury and a common paediatric emergency in children under 5 years of age. The objective of this study was to determine the factors associated with unintentional poisoning among children under 5 years of age presenting to emergency rooms at tertiary care hospitals in Karachi, Pakistan.

Setting Children were recruited from the emergency rooms of the three biggest tertiary care hospitals in Karachi.

Design A matched case–control study was conducted on 120 cases and 360 controls, with matching done on gender and age. Parents were interviewed using a structured questionnaire containing information on sociodemographic factors, the child's behaviour, and the storage practices of hazardous substances of caregivers in the homes. Conditional logistic regression was performed to analyse the data.

Results Accessibility to hazardous chemicals and medicines due to unsafe storage (adj mOR=5.6, 95% CI 1.9 to 16.7), child's behaviour reported as usually aggressive (adj mOR=8.2, 95% CI 4.6 to 16.1), storage of kerosene oil and petrol in soft drink bottles (adj mOR=3.8, 95% CI 2.0 to 7.3), low socioeconomic status (adj mOR=9.2, 95% CI 2.8 to 30.1), low level of mother's education (adj mOR=4.2, 95% CI 1.8 to 9.6), and history of previous poisoning (adj mOR=8.6, 95% CI 1.7 to 43.5) were independently related to unintentional poisoning.

Conclusion The practice of storing kerosene and petroleum in soft drink bottles and the easy accessibility of chemicals and medicines are potentially modifiable. Health messages focusing on the safe storage of chemicals and medicines and the use of child resistant containers may play a key role in decreasing the burden of childhood poisoning in Karachi, Pakistan.

Statistics from


Poisoning is the fourth leading cause of unintentional injury among children.1 2 The majority of poisoning cases occur among preschool children under 5 years of age.3–5 An estimated 86 000 childhood poisoning incidents were treated in US hospital emergency departments in 2004, amounting to 429.4 poisonings per 100 000 children.3 Studies have identified kerosene, petrol, medicines, insecticides, and household cleaning products as major hazards for poisoning incidents among young children.6–8 Kerosene oil poisoning is commonly reported from developing countries as kerosene is a major source of fuel for lighting and power generation.8–10 Poisoning can have long term psychological and physical consequences for children and may result in large societal costs. In low and middle income countries, poisoning accounts for 10% of the total burden of unintentional injuries, and 6% of disability adjusted life years.11

A study from Bangladesh estimated an incidence of 11 poisoning cases per 100 000 per year.12 A population based study from Pakistan reported an annual incidence of unintentional poisoning cases of 4.3%.13 A multi-country global childhood unintentional injury surveillance observed 4% of poisoning cases out of 1559 childhood injuries.4 A systematic review from south Asia estimated case–fatality ratios ranging between 0–11%, with a median of 2.4%.2

Pakistan is a developing and low income country with more than 24 million children under 5 years of age, and is highly vulnerable for such incidents.14 There are few descriptive studies available from Pakistan on unintentional childhood poisoning.6–8 However, there are no analytical epidemiological studies that have focused on factors within the household that are associated with poisoning among young children. The objective of this study was to determine the factors associated with poisoning among children under 5 years of age presenting to the emergency rooms (ERs) of tertiary care hospitals in Karachi. We hypothesised that, among unintentional poisoning cases in children under 5, the accessibility of hazardous substances (chemicals and medicines) is twice that of controls.


This study was conducted in the ERs of three large tertiary care hospitals in Karachi: the Aga Khan University Hospital (AKUH), the Civil Hospital Karachi (CHK), and the National Institute of Child Health (NICH). AKUH, a private hospital, caters for 12 000 paediatric patients annually, whereas CHK and NICH are public hospitals, and receive approximately 150 paediatric patients in the ER daily. These hospitals provide services for the upper, middle and lower socioeconomic classes of the Karachi population. The study data were collected from August 2008 to March 2009.

The investigation was a matched case–control study. All consecutive poisoning cases were enrolled in the study, including gravely ill children or those who died either in the ER or before reaching the hospitals during the study period. Cases were defined as children under 5 years of age with oral ingestion of any noxious substances. They were recruited into the study after a definite diagnosis of accidental intake of poisoning was made by the attending physician of the respective hospitals. The noxious agents considered for poisoning in this study were any substance that had the potential for toxic effects and included medicines, insecticides, pesticides, petroleum products, household chemicals, and cosmetics. Children admitted for food poisoning, adverse drug reactions, and poisoning with animal venom were excluded.

Controls were selected from the ER of the same hospitals and comprised children with complaints other than poisoning. Controls were matched for age (±6 months) and sex, as these were the known confounders.15 Three control children per case visiting the ER of the same hospital within 48 h of case identification were enrolled for the study. Children with symptoms of chronic illness such as known cases of cardiac disease, renal failure, chronic pulmonary disease, cancer patients, and road traffic accidents were excluded from the study sample. Children brought from nearby areas outside Karachi were not included in the study.

Caregivers were interviewed using a structured questionnaire. The interview was conducted in the Urdu language by trained undergraduate medical students. As a quality control measure, the study hypothesis was not shared with the interviewers.

The study was approved by the Ethical Review Committee of Aga Khan University. Written informed consent was obtained from each patient's caregiver and an explanation of the purposes of the research was provided to them.

Information on the sociodemographic characteristics of the child and caregiver, and the storage practices for medicines and chemicals in their household, were obtained. The economic status of the child's family/parents was determined using family income, household structure, and ownership. Information on income was gathered and categorised into low, middle, and upper income groups. Information about the age of the parents (in years), their mother tongue (taken as a proxy to ethnicity), parental education on the basis of number of schooling years completed, the occupation of the parents, and family type (nuclear or extended) was also obtained. Respondents were asked about child behaviour and whether it was either aggressive or not. All the questionnaires were initially developed in English, and translated into Urdu. The back translation method was used to ensure conceptual equivalency and consistency between the English and Urdu languages. Six undergraduate medical students were recruited and trained for data collection, obtaining informed consent, and maintaining privacy during the interviews. The questionnaires were pretested on 5% of the sample.

Statistical analysis

The probability of exposure, caregiver practices of unsafe storage of hazardous substances, among controls was ascertained through the literature and was between 30–50%. Keeping the exposure at 0.50 with the correlation coefficient of 0.60 for exposure between matched case and control patients, a sample of 106 cases with three matched controls were needed to achieve a power of 80% for an odds ratio (OR) of 2.5 at a significance level of 0.05. The sample size was inflated 12% for errors and non-response. Thus a total number of 120 cases and 360 controls were required to achieve the objectives of the study. The NCSS-PASS was used for sample size calculation.16

Analysis was carried out using the Statistical Package for Social Sciences (SPSS), version 11.5. Initially descriptive statistics and frequencies were generated. Matched sets were created in the datasets of each case with three controls. Data were analysed using the conditional logistic regression to consider matching factor.

In inferential statistics, continuous variables were checked for their linearity, by doing quartile analysis. Dummy variables were created for variables with more than two categories and the reference group for each variable was defined as the category with the minimal risk for poisoning using previous studies.

Accurate data on the ages of the children (in months) were collected, and later divided into categories at the time of analysis. Accessibility to hazardous substances was taken as the main exposure variable. It was ascertained using the variables of the height of storage of chemicals and medicines separately. A storage height of chemicals and medicines of 2 m or above was considered safe for children. Similarly, the storage of chemicals and medicines under lock and key was considered safe. A composite variable was generated, with 0=safe storage of both chemicals and medicines, 1=either one stored safely (chemicals or medicines), 2=both chemicals and medicines stored unsafe.

A wealth index (WI) was developed to obtain a robust measurement of socioeconomic status (SES). WI is based on household possessions using proportionate weighting of items.17 18 For example, if 40% of households in a sample of 100 owned a television, then a television would get a weight of 60% (100–40). Weights for each item were summed into a linear index and households were allocated a final score. Since the WI was a composite measure, we used tertiles of the WI for analysis of SES.

Multicolinearity was checked among all the independent variables. A univariate conditional logistic regression analysis was conducted to assess the (crude) association of each independent factor with poisoning. Biological significance and a value of p≤0.25 were considered as criteria for a variable to be significant in univariate analysis.

Biological plausible interactions among variables and confounding were also checked.

Multivariable conditional logistic regression analysis was done and adjusted matched odds ratios (mORs) were calculated.


The data included 120 cases and 360 matched controls. No eligible case or control refused to participate.

Table 1 shows descriptive statistics of only poisoning cases. Incidental uptake of medicines was the most common type of poisoning, followed by kerosene oil ingestion. Similarly, the majority of the incidents of poisoning occurred in the bedroom followed by the drawing/dining room, courtyard and kitchen. Most of the cases were kept under observation for <6 h. Among all the subjects enrolled during the study period, only one1 case—who had ingested medicine (warfarin)—died.

Table 1

Distribution of unintentional poisoning characteristics among children under 5 years of age at tertiary care public and private hospitals in Karachi (n=120)

Distribution of all cases by type of poisoning and gender is presented in table 2. The majority of poisoning incidents occurred among children 1–2 years of age.

Table 2

Distribution of cases of unintentional poisoning among children under 5 years of age by type of poisoning and gender at tertiary care public and private hospitals in Karachi (n=120)

Socioeconomic, demographic, and behaviour characteristics of the children and caregiver characteristics of cases and matched controls are presented in table 3. The ages of both the cases and the control patients were same (p=0.98). Among the cases, medicinal users at home, history of previous poisoning incident, and child behaviour reported as aggressive were more prevalent than among controls. Similarly, the proportion of parents living apart and living in a nuclear family system was more in cases than controls. The storage practices of medicines and chemicals showed that the proportion of unsafe storage of chemicals that is not at a height >2 m is greater among cases than controls (30.8% vs 16.7%). Similarly, more households of cases (80%) stored unlocked chemicals compared to controls (56.1%). Among the cases the safe storage of medicine at a height >2 m was 64.2% compared with 75.3% for the controls. Only 8.3% of the caregivers of cases reported that they kept medicines locked, compared to 23.6% of the controls (table 4). However, when these variables were combined to make a composite variable it showed that 14.2% of the caregivers of cases reported storing both chemicals and medicine unsafely, compared to 5.8% of controls.

Table 3

Characteristics of 120 unintentionally poisoned children and 360 matched controls in tertiary care public and private hospitals in Karachi

Table 4

Storage practices of medicines and chemicals in household

Results of unadjusted analysis of factors associated with poisoning are also presented in table 3. Unsafe storage of both chemicals and medicines was 3.5 times more likely to be associated with unintentional poisoning. Poisoning was significantly associated with storage of kerosene and petroleum oils in soft drink bottles (mOR=4.3, 95% CI 2.6 to 7.1), and child's behaviour reported as aggressive by parents (mOR=6.6, 95% CI 3.9 to 11.3). Both previous poisoning and medicine users at home were 5 and 1.5 times more likely to be associated with poisoning, respectively.

Mother's education level <10 years (mOR=2.4, 95% CI 1.2 to 4.8) and no formal education (mOR=3.8, 95% CI 1.9 to 7.3) were significantly associated with childhood poisoning.

Multivariable conditional logistic regression was performed to determine factors independently associated with poisoning, which are presented in table 5.

Table 5

Multivariable conditional logistic regression analysis of potential factors for unintentional poisoning among children under 5 years of age in tertiary care public and private hospitals in Karachi

Accessibility to hazardous chemicals and medicines was significantly associated with poisoning. Analysis showed that unsafe storage of either chemicals or medicines occurred 1.5 times more in cases compared to controls. Similarly, unsafe storage of both chemicals and medicines occurred 5.6 times more in cases compared to controls (mORadj=5.6, 95% CI 1.9 to 16.7). Storage of kerosene and petroleum in soft drink bottles occurred 3.8 times more in cases compared to controls (mORadj=3.8, 95% CI 2.0 to 7.3). Cases were 8.2 times more likely to be aggressive compared to controls (mORadj=8.2, 95% CI 4.2 to 16.1). Similarly, the reporting of a previous poisoning incidence was 8.6 times more in cases compared to controls (mORadj=8.6, 95% CI 1.7 to 43.5).

The low level of mother's education was strongly associated with poisoning in children. Multivariable analysis determined that mothers educated for <10 years were twice as likely to be among cases compared to controls (mORadj=2.2, 95% CI 0.9 to 5.2). Similarly, those mothers with no formal education were four times more likely to be among cases compared to controls (mORadj=4.2, 95% CI 1.8 to 9.6).


To the best of our knowledge this is the first epidemiological study which has focused on factors associated with unintentional poisoning from south Asia. The analytical design and the measures taken to minimise potential biases add to the validity of the results and helps in understanding the relation between the factors and unintentional poisoning. The findings of the study are important for designing prevention interventions because the majority of these factors are potentially modifiable. Previous studies conducted elsewhere (other than in south Asia) have identified younger parents (≤21 years), poor socioeconomic status, children not living with two parents, and previous poisoning as important determinants of childhood poisoning.10 19–21

The main exposure variable— that is, accessibility to unsafe stored hazardous chemicals and medicines—was strongly associated with poisoning incidences in multivariable analysis. Moreover, storage of kerosene and petroleum in soft drink bottles is a very common practice in Pakistan and elsewhere in south Asia. These hydrocarbons are the major source of fuel for power generation for most developing countries including south Asia. This study investigated the accessibility of hazardous substances to children in households, which has not been previously evaluated in the setting of a developing country. A study from the Netherlands showed that 50.1% of toddlers were exposed to the unsafe storage of poisonous products inside the home.22 Educational solutions such as safe storage have not been shown to be solely effective. A recent systematic review indicated that home safety education and the provision of safety equipment improves poison prevention practices; however, there was a lack of evidence that poisoning rates were affected.23 Therefore, a holistic mechanical environmental solution such as the use of child resistant containers (CRC) should be adapted.

Childhood behaviour reported as aggressive was identified as another important determinant in our study, a finding consistent with previous studies.24–28

The occurrence of a previous poisoning incident has been theorised as an important risk factor in previous studies.19 21 It has been estimated that 30% of young children who experience one poisoning episode will go on to have at least one further such incident before the age of 6 years.29 Therefore, such children should be specially protected and their parents need aggressive counselling by the health providers.

Low socioeconomic status was independently associated with childhood poisoning, a finding consistent with previous studies.13 20 30 31 Similarly, low level of mother's education was also independently associated with unintentional poisoning. Studies from developing countries have indicated that the occurrence of unintentional childhood injuries is more closely related to maternal education than any other socioeconomic factor.4 32

Kerosene oil and medicine are the major substances responsible for unintentional poisoning among children in this study. Descriptive studies from south Asia have reported an increasing trend of medication poisoning, due to medicines being dispensed in non-child resistant containers.32 However, the introduction of child resistant containers and packaging for medicines and household chemicals has been shown to be effective in the USA and UK.33 34

We did not find a significant association between parents living apart and unintentional poisoning in children, a finding in agreement with another study from Thailand.35 However, a study conducted in Greece by Petridou et al showed a greater risk of unintentional poisoning associated with children not living with both parents.19 Probably the social network operating among communities in south Asia is sufficiently strong enough to provide a good protection against childhood poisoning when parents have separated.

When interpreting the results, it should be noted that ascertainment of exposure and other covariates in this study were based on self reported data provided by the caregivers. Such data have been considered to be generally reliable and valid, as injury and particularly poisoning is a defined event and usually well remembered.

Differential recall is the inherent limitation of a case–control study; however, large ORs cannot be explained on the element of differential misclassification alone. In order to prevent this, the questions were related to only the last 5 years of recall. Nevertheless, it may be particularly pronounced in parents of cases reporting child behaviour as aggressive, a history of previous poisoning, and sibling poisoning.

The possibility of selection bias cannot be ruled out as this is a facility based study. Additionally, we cannot overlook the likelihood of interviewer bias because it is possible that the interviewers asked questions on behavioural issues, such as aggressiveness, in a way that was different for the caregivers of cases compared with the controls. Since the interviewers could not be blinded for case–control status in this study, they were trained to treat both groups equally in order to minimise this bias.


The accessibility of hazardous chemicals and medicines due to unsafe storage, reported child behaviour as aggressiveness, history of previous poisoning, lower education level of the mother, storing of kerosene and petroleum in soft drink bottles, and low socioeconomic status were associated with increased risk of unintentional poisoning among children under 5 years of age living in Karachi. The practices of storing chemicals and medicines are potentially modifiable and efforts to reduce the prevalence of unsafe storage could have the benefit of reducing the incidence of unintentional poisoning in this population.

Interventions such as parental education to increase awareness of the need for safe storage of hazardous chemicals and medicines, and familiarity about the utilisation of poison information centres when a poisoning incident has occurred, should be developed among the general population. Similarly, the regulation of child resistant packaging for hazardous chemicals and medicines will also help in curtailing the incidence of unintentional poisoning among young children.

What is already known on this subject

  • Unintentional childhood poisoning accounts for 4.3% of unintentional injuries in Pakistan.

  • Studies from Greece, Malaysia and Thailand have identified low socioeconomic status, number of siblings greater than three, and unsafe storage as major determinants.

What this study adds

  • Unsafe storage of household chemicals and medicines, storage of kerosene oil in soft drink bottles, low level of education of the mother, the child's aggressive behaviour, and low socioeconomic status are the independent factors associated with unintentional childhood poising.


View Abstract


  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval This study was conducted with the approval of the Ethical Review Committee, The Aga Khan University Hospital, Karachi.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request permissions

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.