Article Text

Download PDFPDF

Keeping children safe at home: protocol for a matched case-control study of modifiable risk factors for poisoning
  1. Gosia Majsak-Newman1,
  2. Penny Benford2,
  3. Joanne Ablewhite2,
  4. Rose Clacy2,
  5. Frank Coffey3,
  6. Nicola Cooper4,
  7. Carol Coupland2,
  8. Mike Hayes5,
  9. Bryony Kay6,
  10. Elaine McColl7,
  11. Richard Reading1,
  12. Alex Sutton4,
  13. Jane Stewart8,
  14. Michael Craig Watson8,
  15. Denise Kendrick2
  1. 1Norfolk and Norwich University Hospitals NHS Foundation Trust, Norwich, UK
  2. 2Division of Primary Care, School of Medicine, Nottingham, UK
  3. 3DREEAM (Department of Research and Education in Emergency medicine Acute medicine and Major trauma), Nottingham University Hospitals' NHS Trust, UK
  4. 4Department of Health Sciences, University of Leicester, Leicester, UK
  5. 5Child Accident Prevention Trust, London, UK
  6. 6Emergency Department, United Bristol Healthcare Trust, Bristol Royal Infirmary, Bristol, UK
  7. 7Newcastle Clinical Trials Unit, The Medical School, Newcastle University, Newcastle upon Tyne, UK
  8. 8School of Health Sciences, University of Nottingham, Queen's Medical Centre, Nottingham, UK
  1. Correspondence to Gosia Majsak-Newman, Clinical Research & Trials Unit, Norfolk and Norwich University Hospitals NHS Foundation Trust, Norwich NR4 7UY, UK; gosia.majsak-newman{at}


Background Childhood unintentional and suspected poisonings are a serious public health problem. Evidence from systematic reviews demonstrates that home safety education in combination with safety equipment provision increases the safe storage of medicines and other products. There is lack of evidence that poisoning prevention practices reduce poisoning rates.

Objectives To estimate ORs for medically attended poisonings in children aged 0–4 years for items of safety equipment, home hazards and parental safety practices aimed at preventing poisoning, and to explore differential effects by child and family factors.

Design Multicentre case-control study in UK hospitals with validation of parent-reported exposures using home observations. Cases are aged 0–4 years with a medically attended poisoning occurring at home, matched on age and sex with community controls. Children attending hospital for other types of injury will serve as unmatched hospital controls. Matched analyses will use conditional logistic regression; unmatched analyses will use unconditional logistic regression to adjust for confounding variables. The study requires 266 poisoning cases and 1064 matched controls to detect an OR of 0.64 for safe storage of medicinal products and of 0.65 for non-medicinal products, with 80% power, a 5% significance level and a correlation between exposures in cases and controls of 0.1.

Main outcome measures Unintentional childhood poisoning.

Discussion This will be the largest study to date exploring modifiable risk factors for poisoning in young children. Findings will inform: policy makers developing poison prevention strategies, practitioners delivering poison prevention interventions, parents to reduce the risk of poisoning in their homes.

Statistics from

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.


Despite large reductions in childhood unintentional poisoning, mostly credited to legislation for use of child-resistant closures (CRCs) on medicines and cleaning products1 ,2 and formulation of safer medications,3 ,4 poisoning (including suspected poisoning) still remains a serious public health problem for young children.5 In 2004, unintentional poisonings claimed 45 000 lives of 0–19-year-olds worldwide, comprising 5% of injury-related deaths in the under 5-year-olds.5 Poisoning is the fifth leading cause of unintentional death for children under the age of 1 year in the USA6 and for children and adolescents in the EU.7 In Europe, 705 children aged under 5 years died from unintentional poisoning in 2011.8

Poisonings place a considerable burden on health services. Each year in the USA more than 1.2 million calls are received by poison control centres following an incident in the under 5-year-olds.9 In the UK each year more than 26 000 emergency department (ED) attendances10 and around 6000 hospital admissions result from poisonings in this age group.11 The estimated annual cost to the National Health Service (NHS) for poisoning injuries in children under the age of 15 years is more than £2 million.12 In the USA the lifetime cost of poisoning for children under the age of 15 years is around US$400 million,13 with an estimated average cost of US$1780 per poisoning case.5 In Victoria, Australia, the lifetime injury costs for poisonings in the 0–4-year-olds rank second among all lifetime injury costs in that age group.4

Children under the age of 5 years are particularly vulnerable to unintentional poisoning14 with 1–3-year-olds being at highest risk.15 ,16 This is thought to arise from increasing mobility, curiosity, mouthing, imitation and a lack of awareness of potential hazards,16–20 requiring closer supervision and more protective poison prevention practices.21 ,22 Most poisonings in this age group occur in the home.23 ,24 In high-income and middle-income countries poisonings commonly involve medicines, household cleaning products, cosmetics, pesticides and poisonous plants. In lower-income countries paraffin is a more common cause.14

Factors influencing risk for childhood poisoning include the child's age,15–18 ,25 ,26 gender,16 ,25 birth order,27 socioeconomic disadvantage,26–29 single parenthood,26 maternal age26 ,27 and maternal depression,26 in addition to presence, accessibility and unsafe storage of poisoning agents.5 ,30 ,31 Although the strongest evidence on reducing poisoning relates to use of CRCs,4 ,32 ,33 as CRCs are not ‘child proof’, other poison prevention practices still require evaluation.34 Systematic reviews have demonstrated home safety interventions increase the safe storage of medicines and cleaning products,34 ,35 but evidence that these poison prevention practices reduce poisoning rates is lacking.35

There are few well-designed observational studies demonstrating an association between safety behaviours and a reduction of childhood injuries, while adjusting for a wide range of possible confounding factors.34 Well-designed case-control studies in other areas of injury prevention have contributed to the development of important safety policies and legislation.36 ,37 To date, case-control studies have reported potentially modifiable risk factors, such as accessibility and unsafe storage of harmful substances, for unintentional poisoning.15 ,38–40 These studies have been limited by potential selection bias arising from use of hospital controls, small sample size, adjusting for a limited number of confounders, use of non-validated measures of exposures and using composite exposure measures, thus preventing estimation of the protective effects of specific safety behaviours.

This protocol describes a multicentre case-control study to investigate associations between a range of modifiable risk factors (safety equipment, home hazards and parental safety practices) and poisoning or suspected poisoning in young children. We will minimise selection bias by using community and hospital controls, analyses will be adjusted for a wide range of confounding factors and, where possible, parental reported exposures will be validated by home observations in a subset of study participants.



The primary objective is to estimate ORs for medically attended poisonings or suspected poisonings, resulting from exposure to medicinal, cleaning agents, garden chemicals and other household products commonly associated with poisoning incidents, occurring in children under 5 years of age for a range of modifiable risk factors including safety behaviours, safety equipment use and hazards, adjusting for a range of potential confounding factors. The secondary objective is to explore whether the effects of the above exposures vary by demographic and socioeconomic factors previously found to be associated with differential effectiveness of home safety interventions.41

Study design

This study is one of five multicentre matched case-control studies, using similar methods to study the following injury mechanisms: poisonings, scalds, falls from furniture, falls on one level and stair falls. The methods have been described in detail in the published protocol covering the falls studies.42 These five studies will be conducted simultaneously within EDs, inpatient wards and minor injuries units (MIUs) in four areas of the UK: Nottingham, Derby and Lincoln; Bristol; Norwich, Cromer and Great Yarmouth; and Newcastle upon Tyne and Gateshead. Recruitment began in June 2010.

Definitions of cases and controls

For the purposes of this study, cases will be defined as children aged 0–4 years attending an ED or MIU or admitted to hospital in participating NHS Trusts having, or suspected to have, inhaled, ingested, injected or absorbed a poison or suspected poison in the home or garden at the address where the child is registered with a family doctor/general practitioner (GP), henceforth to be referred to as the child's home or garden.

Children with intentional or suspected intentional poisoning, living in residential care, who have previously participated in the study as a poison case or who suffer fatal injuries will be excluded. Additionally, children who participate in one of the other four concurrent case-control studies within the preceding 12 months will not be eligible. Cases will be recruited to no more than two of the simultaneously ongoing case-control studies and will not be eligible for subsequent recruitment as controls.

Controls will be defined as children less than 5 years of age who did not present to one of the participating EDs or MIUs or were not admitted to hospital with a poisoning or suspected poisoning on the same date as the cases’ attendance to hospital. Children living in residential care will be excluded. Controls will be eligible for subsequent recruitment as a case or as a further control if the approach is made 12 months or more after their first recruitment, but may not be recruited more than twice to any of the five case-control studies.

Recruitment of cases and controls

Members of the healthcare team will assess eligibility for the study based on data from the medical records and will then approach those considered to be eligible at that point to invite them to take part in the study. Further assessment of eligibility will be undertaken during the initial approach and when parents return completed questionnaires.

Following introduction by clinical staff, participants will be recruited by the research team either face to face during their ED or MIU attendance or their hospital admission, or by telephone if the parent or guardian has given permission to be contacted via this route, or by postal recruitment following their attendance or admission. Telephone or postal contact will be initiated within 72 h of the attendance or admission. Participants who do not reply to the postal invitation will be sent one reminder 2 weeks after the original mailing.

There will be two types of controls: matched community controls and unmatched hospital controls. Community controls will be recruited within 4 months of the case injury from the register of the case's GP, and matched on age (within 4 months of the case's age) and sex. Staff at the GP's practice will identify and send postal invitations to the parents or guardians of 10 children from their age-sex register with the aim of recruiting an average of four controls per case. One reminder will be sent to people who have not replied to the postal invitation 2 weeks after the original mailing. If a practice is unwilling or unable to recruit controls, the practice geographically closest to the case's GP practice will be asked to identify potential controls.

Hospital controls will comprise four randomly selected cases already recruited to one of the concurrent case-control studies with a different mechanism of injury (scald, fall from furniture, fall on one level, or stairway fall). The process of recruitment of cases and controls has been fully described in the published protocol for the falls case-control studies.42

Definition of exposures

Questions about most exposures specific to poisonings will be asked in relation to the 24 h prior to the poisoning for cases and the 24 h prior to completing the questionnaire for controls. Questions for a small number of exposures were measured over the preceding week for cases and controls (as indicated below (†)). The exposures of interest are:

  1. Safety equipment use

    1. Safety gate across kitchen doorway*

    2. All medicines have child resistant caps or blister packs*

    3. All household products have child resistant caps*

    4. All medicines stored in locked medicine box, cupboard, cabinet, drawer or fridge*

    5. All household products stored in locked cupboard, cabinet, drawer or fridge*

  2. Home hazards

    1. Use of baby walker*

    2. Presence of item child could climb on to reach adult level surface/shelf/cupboard, etc

    3. Any medicines stored in container different from the one they originally came in*

    4. Any household products stored in container different from the one they originally came in*

  3. Safety behaviours

    1. All medicines stored safely*

    2. All household products stored safely*

    3. All medicines stored at adult eye level or above*

    4. All household products stored at adult eye level or above*

    5. Frequency of returning medicines to usual storage place immediately after use†

    6. Frequency of returning household products to usual storage place immediately after use†

    7. Teaching child safety rules about what to do/not do when sees cleaning products

    8. Teaching child safety rules about what to do/not do when medicines are on worktop

Exposures validated by home observation in a subset of cases and controls are starred in the list above. Use of baby walkers was only asked in families with children aged 0–12 months or 13–36 months, Safe storage was defined as stored at adult eye level (or above) or in drawers and cupboards with catches or locks, or if none stored in the house.

Definition of confounding variables

Potential confounding variables are the same as those which will be measured in the concurrent case-control studies.42 They include

  • sociodemographic and economic characteristics (eg, family size and structure, ethnic group, overcrowding, housing tenure, receipt of state-provided means-tested benefits, maternal age, time cared for outside the home, and place of out-of-home care), and

  • validated measures of child behaviour and temperament (the activity and high intensity pleasure subscales of the Infant, Early Child and Child Behaviour Questionnaires),22 ,43–46 parenting daily hassles (parenting tasks subscale),47 ,48 mental health,49 general health visual analogue scale,50 and Paediatric Quality of Life Inventory.51 ,52

Measurement of exposures and confounding variables

Three age-specific questionnaires (0–12 months, 13–36 months and 37–59 months) have been developed using lay research advisers and expert advice from the Child Accident Prevention Trust and from study team members. They have been piloted for face validity, comprehension, ease and time taken for completion with families of children attending EDs at participating NHS Trusts and families from children's centres in study centres. Wherever possible, these contain previously validated questions regarding exposures and potential confounders, as well as some measures which will be validated by home observations. Parents will be asked to complete the questionnaire appropriate to the participating child and in return, to increase response rates, will receive a £5 ($7.7, €6) gift voucher for use in local shops.53

Validation of exposure measurement

Parent-reported exposures will be validated by home visits to a sample of cases and controls to assess agreement between reported and observed safety behaviours, safety equipment use and home hazards. This has been described elsewhere.54

Sample size

To detect an OR of 1.59 (equivalent to an OR of 0.63 expressed as a protective association), based on at least 80% power, 5% significance level and a correlation between exposures in cases and controls of 0.1, 266 cases and 1064 controls will be required. This calculation is based on exposure prevalence estimated from the first 428 controls recruited to the study, taking account of missing data on exposures and requiring the largest sample size from prevalences of 21% (not putting cleaning products away immediately after use), 23% (not putting medicines away immediately after use), 27% (not storing all medicines safely), 29% (not putting all products away immediately after use), 55% (not storing all cleaning products safely) and 65% (not storing all products safely).


Prior to commencing statistical analysis, causal diagrams will be drawn depicting the interrelationships between variables and differentiating confounding variables from likely effect mediators (variables potentially in the causal relationship between exposure and outcome). Descriptive statistics for exposure and confounding variables will be generated separately for cases and controls. They will comprise frequencies and percentages for categorical variables and, depending on their distributions, means (and SDs) or medians (and IQRs) for continuous variables.

For the analysis of cases and matched controls, conditional logistic regression will be used to estimate ORs and 95% CIs for each exposure variable. Unadjusted and adjusted ORs will be presented. Adjustment will be made for deprivation (measured using the Index of Multiple Deprivation 2010 at super output area level)55 and distance from hospital. Models will also be adjusted for the minimal adjustment set of confounding variables identified from causal diagrams. Analysis of cases and hospital controls will use unconditional logistic regression to estimate unadjusted and adjusted ORs for each exposure. Since hospital controls are not individually matched to the cases, this analysis will adjust for age, sex, deprivation (measured using the Index of Multiple Deprivation 2010 at super output area level),55 distance from hospital and the other potential confounding variables as described for the matched analysis.

Differential effects by child and family factors found to be previously associated with effectiveness of home injury prevention interventions (child age, gender, ethnic group, housing tenure, parental unemployment and single parenthood)41 will be examined by the addition of interaction terms to regression models. Adjusted ORs will also be stratified by injury treatment (admitted, treated in ED or requiring secondary or primary care follow-up vs seen and examined but no treatment required). Multiple imputations will be used to take account of missing data.


To the best of our knowledge, this study will be the largest case-control study assessing a range of modifiable risk factors for unintentional childhood poisonings and suspected poisonings to date. We have attempted to address some of the potential limitations of previous studies. Two sources of controls, community and hospital controls, will be used as a previous study has demonstrated risk factors for child poisoning vary depending on the source of controls.15 To minimise response bias participant invitations, study information leaflets and questionnaires were developed with the help of a lay research adviser and were piloted with families who would have been eligible to participate in the study. Reminders to prompt non-responders and small monetary incentives for questionnaire completion will be used. By inviting participants to join the study soon after the injury we hope to obtain accurate reporting of exposures. We will also assess accuracy of self-reported exposures by home observations in a sample of study participants and compare this between cases and controls to assess differential misclassification of exposures. Our study questionnaires measure a wide range of confounding factors enabling analyses to adjust for these to minimise the risk of residual confounding and the impact of missing data will be assessed.

The findings from this rigorously conducted case-control study should be useful for policy makers and practitioners. They will be relevant for informing advice given in child health surveillance programmes, home safety assessments, schemes providing safety equipment, postinjury counselling and other child health promotion contacts.


The authors would like to acknowledge the contributions to the study design, methodology and comments on drafts of this paper from members of the Keeping Children Safe Study Group: Persephone Wynn, Clare Timblin, Ben Young from the School of Medicine, Division of Primary Care, Tower Building, University Park, Nottingham, NG7 2RD, UK; Philip Miller from Nottingham University Hospitals NHS Trust, Queen's Medical Centre Campus, Nottingham, NG7 2UH; Elizabeth ML Towner, Toity Deave, Trudy Goodenough from the Centre for Child & Adolescent Health, Health and Life Sciences, University of the West of England, Bristol, Oakfield House, Oakfield Grove, Bristol BS8 2BN, UK; Lisa McDaid from Norfolk and Norwich University Hospitals NHS Foundation Trust, Norwich NR4 7UY, UK; Adrian Hawkins, Paul Hindmarch from Institute of Health & Society, Baddiley-Clark Building, Newcastle University, Richardson Road, Newcastle upon Tyne, NE2 4AX, UK.



  • Funding This paper presents independent research commissioned by the National Institute for Health Research (NIHR) under its Programme Grants for Applied Research funding scheme (RP-PG-0407-10231). The views expressed in this paper are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health.

  • Competing interests None.

  • Ethics approval Ethical approval has been obtained from the Nottingham 1 Ethics Committee (reference number: 09/H0407/14). Approval has been provided by National Health Service research & development departments which deliver research governance to the participating hospitals and minor injury units.

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