Article Text

Download PDFPDF

Incidence, characteristics and risk factors for household and neighbourhood injury among young children in semiurban Ghana: a population-based household survey
  1. A Gyedu1,
  2. E K Nakua1,
  3. E Otupiri1,
  4. C Mock2,3,
  5. P Donkor1,
  6. B Ebel2,4,5
  1. 1Kwame Nkrumah University of Science and Technology, Kumasi, Ghana
  2. 2Harborview Injury Prevention & Research Center, University of Washington, Seattle, Washington, USA
  3. 3Departments of Surgery, Global Health and Epidemiology, University of Washington, Seattle, Washington, USA
  4. 4Departments of Pediatrics and Epidemiology, University of Washington, Seattle, Washington, USA
  5. 5Seattle Children's Hospital, Seattle, Washington, USA
  1. Correspondence to Dr Adam Gyedu, Department of Surgery, School of Medical Sciences, Kwame Nkrumah University of Science and Technology, Private Mailbag, University Post Office, Kumasi, Ghana; drgyedu{at}


Background There are few population-based studies on household child injury in African countries.

Objectives To determine the incidence, characteristics and risk factors of household and neighbourhood injury among children in semiurban communities in Kumasi, Ghana.

Methods We conducted a cross-sectional population-weighted survey of 200 randomly selected caregivers of children under 18, representing 6801 households. Caregivers were interviewed about moderate to severe childhood injuries occurring within the past 6 months, for which the child staying home from school or activity, and/or required medical care. Multivariable logistic regression was used to identify factors associated with injury risk.

Results Annual injury incidence was 593.5 injuries per 1000 children. Common causes of injury were falls (315.7 injuries per 1000 children), followed by cuts/lacerations and burns. Most injuries (93.8%) were of moderate severity. Children whose caregivers were hourly workers (AOR=1.97; 95% CI 1.06 to 3.68) had increased odds of sustaining an injury compared to those of unemployed caregivers. Girls had decreased odds of injury (AOR=0.59; 95% CI 0.39 to 0.91). Cooking outdoors (AOR=0.45; 95% CI 0.27 to 0.76) and presence of cabinet/cupboards (AOR=0.41; 95% CI 0.24 to 0.70) in the house were protective. Among children under 5 years of age, living in uncompleted accommodation was associated with higher odds of injury compared with living in a rented single room (AOR=3.67; 95% CI 1.17 to 11.48).

Conclusions The incidence of household and neighbourhood child injury is high in semiurban Kumasi. We identified several novel injury risk factors (hourly work, younger children) and protective factors (cooking outdoors, presence of cabinet/cupboards). These data may identify priorities for household injury prevention.

Statistics from


Injury is a major cause of death and disability worldwide,1 ,2 and the burden of injury falls heavily on children. Globally, injuries account for 13.5% of deaths among children aged 1–4 years.1 ,3 The burden of child injury disproportionately affects poor families living in low-income and middle-income countries (LMIC).4 In addition to those who die each year, millions more may develop disabilities exacting a profound toll on the individual, family and society.5

Most injury deaths among younger children occur in or around the home.6–8 To continue to make progress in global commitments to the Millennium Development Goal 4, progress must be made in the identification and control of injury hazards in the child's home environment.

The global burden of under-five deaths is increasingly concentrated in Sub-Saharan Africa (SSA).9 While there is recognition that effective injury prevention strategies are needed to reduce global morbidity and mortality for children, there are very few developing countries with population-based surveillance data on child injury, and even fewer studies that report on household child injury (HCI) in African countries.10–13 In Ghana, the burden of household injuries in children is not well described even though it is likely to be as high as in other developing countries.

The purpose of this population-based household study was to determine child injury incidence, describe causes of child injury and characterise potentially modifiable factors associated with HCI risk in semiurban communities in Ghana.


Survey design

We designed a cross-sectional population-based study of household risk factors and injury incidence among Ghanaian children in semiurban communities in Kumasi.

Study area

The study was conducted in the Asawase submetropolis of Kumasi, Ghana. Asawase is the Ghana site for the Family Health and Wealth Study (FWHS) within which our study was nested. The FHWS is a multicountry longitudinal open cohort study, operating across five African countries, that aims to examine how family size affects health and wealth. Asawase, a semiurban setting with characteristics typical of such settings in developing countries—limited social amenities, poor housing, densely populated and occupied mainly by people of lower socioeconomic status.

Most families in Asawase live in compound houses. Compound houses are large houses with rooms rented out to different families all of whom usually share the same basic utilities such as toilets and bathrooms. Very few multistorey compound houses exist in Asawase.

Many households do not have properly constructed kitchens. In some compound houses, there is a separate structure used by all families for cooking. Some cook outside their homes in the open and others do their cooking indoors or on an adjoining porch that has been enclosed and partially walled-off from sleeping areas.

For the purposes of the FHWS, Asawase was divided into four communities, and 40 electoral areas were randomly selected using population proportional to size sampling. Twenty families/households per electoral area were randomly selected from a sampling frame of households constructed for each of the 40 electoral areas. The households recruited for the injury study were randomly selected from the FWHS sampling frame. Asawase has a total population of 312 258.14

Study population

Using the prevalence of injury of 12.6% reported by Atak et al,15 a 95% CI and a margin of error of 10%, a sample size of 170 was calculated. We chose 200 to allow for incomplete entries.

For the parent study, 200 households containing children under 18 were randomly sampled from each of the four communities (figure 1). We designed an injury-specific survey module that was administered to 50 households randomly sampled from each community.

Figure 1

Study design and sampling framework.

Survey methods

Parents/caregivers were interviewed about the occurrence of childhood injury within the past 6 months and injury severity. Mock et al16 proposed a 1–3 month recall period for surveys of less severe injuries in developing countries, whereas for more severe or fatal injuries, they proposed a 1-year recall period. For this study, a recall period of 6 months was used as a compromise to capture injuries of all levels of severity. We used the injury severity classification proposed by the UNICEF Innocenti Research Centre.3 Potentially modifiable risk factors measured were generated from the 2008 Ghana Demographic and Health Survey household questionnaire.17 Our questionnaire was based in part on a prior injury questionnaire16 that had been used extensively in Ghana and for which the wording of questions had been well-worked out for the local culture and for which consistency on repeat visits was found to be acceptable.18–20 Questions were updated, modified and expanded for the current study. Figure 2 describes the theoretical model used to derive the survey questionnaire on household injury. We pretested the current study instrument at Krofofrom in Kumasi; Krofofrom has characteristics similar to that of the study area. Pretesting showed that the study tool was acceptable and easily understood. Enumerators collected demographic and household characteristics on ownership of consumable goods and physical characteristics of the household that were used to construct a wealth index using principal component analysis (PCA). Scores generated by the PCA were divided into quintiles, the first quintile representing the poorest households of the distribution. Oral informed consent was obtained from all respondents. The study was approved by the Institutional Review Boards of the Kwame Nkrumah University of Science and Technology and the University of Washington.

Figure 2

Conceptual framework.

Data analysis

Analyses used survey-sampling weights, adjusted for clustering by community and household.

Total household size for each of the four communities was estimated from the Ghana 2010 Housing and Population Census data prior to the start of the larger study. Household level weights (WHi) and household with child weights (WHCi) were estimated for each community. The overall weight (WTOTi) was obtained by multiplying WHi and WHCi. Stratified variables were compared using χ2 test. Since the survey recall period was 6 months, annualised incidence rates were calculated by multiplying injury incidence by 2. All analyses were done using survey weights in STATA V.11 (StataCorp, College Station, Texas, USA).

The primary outcome variable was presence or absence of any injury, defined as an injury sustained by a child under 18, occurring within 200 m of the house, that prevented the child from going to school or work or for which treatment was sought. We considered 200 m as the furthest distance that most children under five would get away from the house.

Multiple injuries sustained by a child were counted once for each mechanism of injury and separately for different mechanisms. For regression analysis, we considered whether the child had sustained any injury at all. We performed univariate logistic regression to assess the relationship between risk factors and injury occurrence using Wald's test. Multivariate logistic regression models incorporated variables that were independently significant in the univariate analysis (p≤0.05) as well as variables that were felt to be important or confounding (eg, age and sex of the child, caregiver level of education). The non-significant but important characteristics were controlled for in the backward regression model whose results are reported here. The fitted model was assessed using Hosmer and Lemeshow's goodness-of-fit test (p≤0.10). We used generalised linear mixed models to account for clustering by household and community.

Secondary multivariate regression models were performed for the subset of children under five.


Table 1 summarises the characteristics of the caregivers and households. We sampled 200 households with children under 18, representing a total 6801 households across the four communities. Majority (85.2%) of the caregivers were biological mothers of the children. The mean age of the respondents was 33.8±7 years. Basic education (66.4%) was the highest level most respondents attained and hourly work (68.4%) was the main occupation. A high proportion of the caregivers lived in a rented single room with their families while 21.8% lived in a rented flat. The main fuel used for cooking was charcoal and cooking was mostly done in the house.

Table 1

Caregiver and household characteristics

We considered cooking surfaces within 1 m from the floor as accessible to a child under five; in 69.7% of the households, the height of the stove or cooking surface was within reach of a child while 78.8% of the households had cabinets or cupboards for safe keeping of chemicals.

There were 637 children under 18 years in the sample, representing a total of 20 575 children in Asawase (table 2). Fifty-two per cent were females and 37.2% were within 1–4 years of age. The overall household injury rate among children was 29.7% in the past 6 months. Almost all the injuries (93.8%) were of moderate severity. There were no deaths.

Table 2

Child characteristics

The most common mechanism of injury was falls (53.2%) followed by cuts/lacerations (20.1%). Burns, assault and falling objects, as mechanisms, occurred with a proportion of 5.3%, 4.6% and 2.8%, respectively.

The annual incidence of injury among children was estimated as 593.5 injuries/1000 children/year. Falls contributed significantly to the annual burden of injury with an estimated incidence of 315.7 injuries/1000 children/year. Corresponding incidence for cuts/lacerations and burns were 119.2 and 31.2 injuries/1000 children/year, respectively. Males experienced a higher incidence (686.7 injuries/1000 children/year) compared with females (505.8 injuries/1000 children/year). The injury incidence rate included multiple injuries, suffered by 5.1% of the children (tables 3 and 4).

Table 3

Mechanism of injury, by child age group

Table 4

Mechanism of injury, by child sex

The multivariate logistic regression model (table 5) showed that children of caregivers who were hourly workers had 97% increased odds of injury compared with those of unemployed caregivers. A year's increase in caregiver's age was associated with a 5% decreased odds of injury to the child. The sex and age of the child were found to be associated with child injury: compared with infants, children 1–4 years and 5–9 years had 4.6 times and 3.1 times increased odds of injury, respectively. Female children had 41% decreased odds of injury compared with males. Moreover, cooking outdoors and presence of cabinet/cupboards in the house had a protective effect of 55% and 59%, respectively.

Table 5

Factors associated with household injury risk, children 0 through 18 years of age: univariate and multivariate regression models*

Table 6 shows a subgroup analysis among children under five. As the children grew older, they were at increased odds of sustaining an injury. Children of age 1–2 years (AOR=3.92; 95% CI 1.30 to 11.79) and those 3 years or older (AOR=6.37; 95% CI 1.80 to 22.56) were at increased odds of injury compared with infants. Children whose caregivers were employed (hourly work) had three times increased odds of injury compared with those of unemployed caregivers.

Table 6

Factors associated with household injury risk for children under 5 years of age: univariate and multivariate regression models*

Compared with children who lived in a rented single room, children who lived in uncompleted accommodation had 3.7 times increased odds of injury. Compared with those living in households where cooking was done indoors, children who lived in a household where cooking was done outdoors had 60% decreased odds of injury.


This study attempts to describe the burden of household injury among children under 18 in semiurban communities of the Kumasi metropolis, Ghana. It also tries to characterise potentially modifiable factors associated with HCI risk.

The annual incidence of household injury among children in semiurban Kumasi was estimated at 593.5 injuries/1000 children/year. Falls, cut/lacerations and burns accounted for almost 80% of HCI. The sex and age of the child, the primary caregiver's type of job, where cooking was done in the household and the availability of a cabinet in the house were found to significantly influence the odds of injury.

The study has some limitations. First, there was no way to independently validate the accuracy of information reported by caregivers. A recall period of 6 months was also used as a compromise to capture injuries of all levels of severity. This might introduce a recall bias into our findings with underestimates of incidence rates reported. A prior study in the same area estimated that using a 6-month recall period in comparison to a 1-month recall period results in a 75% decrease in reporting of minor injuries, 40% reduction in reporting of moderate injuries, but no change in reporting of major injuries.16 Second, participants may have provided socially desirable responses. This may explain why no information was recorded relating to injuries that lead to permanent disability or death to a child. Third, only information on injuries that occurred within 200 m of the house was recorded. Hence, motor vehicular and pedestrian injuries that occurred further away from the house were not reported. This may have led to the lower rates of injury reported among older children, who are more likely to be injured further from home. Fourth, the study was cross-sectional. Hence although we detected associations between injury occurrence and some risk factors, we cannot draw conclusions about causality.

Our estimate of HCI burden is about 10 times the 52.6 injuries/1000 children/year reported from the USA.21 The consequences of injuries to children (immediate effects of hospitalisation; disability on children; economic costs to patients, relatives and to society) are well known.5 ,22 We did not ask about the earnings/wages of caregivers in this study. It would thus be difficult to provide a fair estimate of the economic cost of child injury to the working parent in our study population. Fortunately, the overwhelming majority of HCI encountered in semiurban Kumasi were of moderate severity.

From our study, boys experienced a higher incidence of injury per year compared with girls. Culturally, boys are given relatively greater freedom to explore their environment.23 ,24 Studies show that with few exceptions, boys are more likely to sustain injury from all causes.25 ,26

As in previous reports,5 ,15 ,27 falls were the most common mechanisms of HCI, accounting for more than half of all injuries. Poor safety measures in and around the home may play a role. Of interest, none of the households surveyed was multistoried.

Cut/lacerations accounted for 20% of HCI and were more common among girls. The highest rates were among children of age 1–4 and 5–9 years. Lack of adequate supervision may play a role here as curious children may not sense the danger associated with sharp objects in general and kitchen knives in particular.

Fire-related burns have been reported as the second largest cause of injury-related child death in SSA, and its overall incidence of 8.7/100 000 has been reported as the highest worldwide.8 Burns as a mechanism of injury accounted for only 5.3% of HCI in semiurban Kumasi and significantly no injury-related death was reported.

As predictors of injury, children whose primary caregivers engaged in hourly paying jobs were at increased odds of injury compared with those of unemployed caregivers. Lack of quality supervision of children, involving appropriate attention, proximity and continuity,28 may explain this observation.29–32 Those engaged in hourly work are poorly paid and are unlikely to be able to contract the services of professional caregivers or pay for their children to stay in a school environment. Almost 70% of caregivers in semiurban Kumasi engaged in small scale trading mostly close to their homes. They tend to divide their attention between work or other activities and child supervision or temporarily leave children under the care of older children.

The average caregiver age was 33 years and a year decrease in caregiver age was associated with a 5% increased odds of HCI. Inexperience of the caregiver and poor attitudes towards hazards in and around the house, particularly to younger children, may explain the observation. Younger caregivers are also likely to be engaged in multitasks at all times leaving little room for child caregiving.

Children 1–4 years and those 5–9 years were at increased odds of injury compared to infants. A potential explanation is their curiosity to experiment and explore their surroundings coupled with an inability to perceive danger.33 ,34 The burden of injury is greater among them since they have more years ahead of them to be affected by disability.33 Ironically injuries in these children receive less attention from policymakers since the focus is more on under-five mortality from infectious diseases, an ongoing serious situation in LMICs.35 Majority (71%) of households in semiurban Kumasi lived in rented single rooms within compound houses. In half of all households, cooking was done on the enclosed adjoining porch. Cooking outdoors reduced the odds of HCI by 55% compared with cooking indoors. We believe cooking outdoors is protective compared with cooking indoors because it allows younger children more room to move about without encountering the cooking area. Future injury prevention work in Ghana should target design of cooking areas. The absence of a cabinet in the home was associated with increased odds of injury. Use of rooms for multiple functions, particularly absence of a separate kitchen, has been identified as a major risk for HCI.36–38

Most of the injuries were more commonly seen in children under five. The age of the child, the primary caregiver's type of job, where cooking was done and the availability of a cabinet in the house were all found to significantly influence their odds of HCI. In addition, children living in uncompleted accommodation had 3.7 times increased odds of household injury compared to those living in rented single rooms. In Ghana, due to lack of easily available credit, many buildings are constructed in stages, sometimes over years, as families put aside small amounts of money. Often times, people start living in these as soon as one part of the building is habitable. Similar findings have been reported in other SSA countries with informal settlements comprising apartment buildings with unprotected windows and precarious wiring or informally constructed shacks with restricted internal space.38 Such settlements expose children to more injury hazards as they have very limited safe recreational spaces.30 ,39 ,40


There is a high incidence of HCI in semiurban Kumasi that demands urgent action. This is particularly so given children's increasing exposure to risk of injuries, and possibly death, as a result of rapid urbanisation. The lack of reliable data, such as presented in our study, may contribute to the little attention paid to this public health issue.


There is the need to increase awareness about the high incidence of HCI in semiurban Kumasi. Providing community childcare facilities may relieve caregivers of their supervision role while engaged in economic activities. Ensuring that homeowners provide separate structures to serve as kitchens will likely reduce potential injury hazards that cooking indoors in the absence of a kitchen present to children. Finally, standards must be enforced to reduce the number of uncompleted houses within which some families live.

What is already known on the subject

  • The burden of child injury disproportionately affects poor families in low-income and middle-income countries (LMICs).

  • Little attention has been paid to child injuries in many African countries.

What this study adds

  • The estimated annual incidence of household child injury (HCI) in periurban Kumasi was 593.5 injuries/1000 children with falls accounting for more than 50% of HCI.

  • The modifiable risk factors of HCI in a LMIC; caregiver employment, cooking place, storage facilities and family residential accommodation.


We wish to express our sincere thanks to all caregivers who participated in the study.



  • Contributors AG, EKN, EO, CM and BE conceived of and refined the study design. AG and EO supervised collected all data. EKN and BE analysed the data. AG, EKN, PD and BE wrote the first draft of the paper. All coauthors provided significant input to the manuscript, revised it critically for important intellectual content and gave their final approval for the version to be published.

  • Funding This study was funded, in part, by a grant (D43-TW007267) from the Fogarty International Center, US National Institutes of Health. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

  • Competing interests None.

  • Ethics approval Institutional Review Boards of the Kwame Nkrumah University of Science and Technology and the University of Washington.

  • 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.