The development of an instrument measuring unintentional injuries in young children in low-income settings to serve as an evaluation tool for a childhood home injury prevention program

Journal of Safety Research 36 (2005) 269 – 280 www.elsevier.com/locate/jsr

www.nsc.org

The development of an instrument measuring unintentional injuries in young children in low-income settings to serve as an evaluation tool for a childhood home injury prevention program E.R. Jordaan *, S. Atkins, A. van Niekerk, M. Seedat Received 21 January 2005; accepted 16 March 2005

Abstract Problem: Responding to the lack of standardized instrumentation, especially as applied in low-income contexts, the researchers decided that an instrument should be developed to serve as an evaluation tool for a childhood (unintentional) injury prevention program as well as a tool from which injury risks (poisonings, burns and falls) could be identified within households. Methods: Cross-sectional samples of 521 households in four low-income sites in Cape Town and Johannesburg, South Africa, participated in the three phases of the study. The generation of an item pool based on a table of specifications, subsequent scientific item reduction procedures, reviews from experts and pilot tests were used to develop the scales for measuring the injury risks. Results: The developed instrument complies with all the requirements for a valid and reliable measurement instrument. Conclusion: The instrument allows valid comparison of risks between communities, as well as before and after comparisons for an intervention program. Impact: The provision of this instrument may bring enormous benefits to research studies. It can also provide proactive, rather than reactive, information about injury risks before they develop into injuries and thus allows focusing of safety efforts for improvement of problematic areas in the households. D 2005 National Safety Council and Elsevier Ltd. All rights reserved. Keywords: Validity; Internal consistency reliability; Test-retest reliability measurement model

1. Introduction Child unintentional injuries from burns, falls, and poisonings are among the leading causes of unnatural death globally (Barss, Smith, Baker, & Mohan, 1998; Krug, 1999). This problem is increasingly evident in low- to middle-income populations in South Africa (Bradshaw, Bourne, & Nannan, 2003), where in 1998 more than threequarters of all injuries were unintentional (South Africa Demographic and Health Survey, 1998). These injuries are most frequently sustained at home, where children spend a significant part of their time (Child accident prevention Trust, 1989; Nomazamo Safety Promotion Fact Book, 1998). Unintentional injuries are reported to have a steep social gradient, with socio-economic status inversely related to injury (Diderichsen, LaFlamme, & Hallqvist, 1999; * Corresponding author. E-mail address: [email protected] (E.R. Jordaan).

Laflamme, 2001). In South Africa, many unintentional injuries are thought to occur in low-income neighborhoods and informal settlements, usually located on the periphery of the more resourced metropolitan centers. The interaction of certain combinations of personal, environmental, and equipment factors within these settings contribute to the high injury rates reported among children in and around homes in low-income neighborhoods (Barss et al., 1998; Nomzamo Safety Promotion Fact Book, 1998). And families with limited personal resources living at the economic margins of society have fewer opportunities to facilitate successful parenting, since few have the option to learn about injury prevention and home safety (Gielen, Wilson, Faden, Wissow, & Harvilchuck, 1995; Hapgood, Kendrick, & Marsh, 2000). In North America, Europe, and Australia, home visitation has been recognized as a potential good practice in childhood injury control (Bender, van Niekerk, Seedat, & Atkins, 2002). Home visitation programs (HVP) are

0022-4375/$ - see front matter D 2005 National Safety Council and Elsevier Ltd. All rights reserved. doi:10.1016/j.jsr.2005.03.006

270

E.R. Jordaan et al. / Journal of Safety Research 36 (2005) 269 – 280

generally seen as multi-method approaches, which usually entail educational, enforcement, and engineering components (Bender et al., 2002). A significant body of evidence has already been developed in support of the intervention’s application to promote other childhood health outcomes, particularly in the United States, with home visitation for childhood injury prevention a more recent phenomenon (Mohamed, 2001). Despite the increase of evaluated interventions in Europe, North America, and Australia, in South Africa there remains a lack of effective, replicable, and contextually congruent childhood injury prevention interventions (Van Niekerk & Duncan, 2002). In response to this lacuna, Atkins and colleagues investigated a home visitation program (Atkins, van Niekerk, Bender, & Seedat, 2002). They suggested that a home visitation program might be an effective intervention adaptable to a range of goals, if specifically designed to address the circumstances of the lives of these low-income parents with lower educational achievement. After their recommendation, a HVP was developed and piloted, with the intention of implementing, evaluating, and determining costs of such an intervention specifically for impoverished South African neighborhoods. The focus of the study was on burns, falls, and poison injuries. Standardized and tested instruments are especially important in the evaluation of intervention programs, such as the home visitation program (Dowswell, Towner, Simpson, & Jarvis, 1996; Towner & Dowswell, 2002). A literature review conducted on child injury risk measuring instruments revealed that few instruments have been tested and standardized. The review unearthed several checklists that could be used by parents to make their home safer (Safekids, undated; Vocational and Rehabilitation Research Institute, undated), however, few of these offered a comprehensive and well-documented basis for their checklists. Some well-documented survey instruments directed at children can, however, be identified from literature (Tymchuk, Lang, Dolyniuk, Berney-Ficklin, & Spitz, 1999; Tymchuk, Lang, Sewards, Lieberman, & Koo, 2003). These include the Framingham Safety Survey (Bass & Mehta, 1980; Bass, Mehta, Ostrovsky, & Halperin, 1985; Hansen, Wong, & Young, 1996). This instrument aimed to direct a service provider, such as emergency department or private practice health professional, to the correct information regarding home safety dangers. Although it is widely documented and reviewed, it is however not useful for identifying dangers within the home. An instrument standardized and reviewed for home use is the Home Accident Prevention Inventory (HAPI). The HAPI assessed the safety of the home environment including five categories: (a) fire and electrical hazards; (b) mechanical suffocation; (c) ingested object suffocation; (d) firearms; and (e) solid and liquid poisons. However, this instrument was again developed for a high income setting and would have to be greatly modified for application in the South African context. The elderly population has received more attention in terms of

instrument development, especially in terms of fall injuries in the home (Johnson, Cusick, & Chang, 2001; MacKenzie, Byles, & Higginbotham, 2002). Responding to the lack of standardized instrumentation, especially as applied in low-income contexts, the researchers decided that a new instrument should be developed to serve as an evaluation tool for a childhood injury prevention program as well as a tool from which injury risks could be identified within households. Data were collected in a sample of households from three low-income sites and the psychometric properties of the instrument were examined.

2. Research methods The following program of research was followed to develop a valid and reliable risk injury instrument: 1. Based on an extensive literature review, a table of specifications was first developed to ensure content validity of the instrument. 2. A thorough review of the literature pertaining to all constructs under study was done. 3. Previously used instruments to measure these constructs were also located. 4. A pool of items was identified. Existing instruments were adapted; items were deleted or added, making the instrument specific to the environment of the population. 5. A total of 45 items were originally generated from a variety of sources using the table of specifications. Both negatively and positively worded items were included. The items were screened for redundancy and clarity, leading to a reduced pool of items. 6. Experts in the field reviewed the set of items to maximize the content validity of the instrument. Steps 1 to 6 addressed the face and content validity of the instrument. 7. The instrument developed in steps 1 to 6 was administered to a sample of households from one Cape Town site (Nomzamo). 8. Ceiling (number of respondents indicating a positive risk) and floor (number of respondents indicating no risk) effects were investigated to identify the items not relevant to the population (Fleiss, 1981). 9. The test-retest reliability was determined and unreliable items were deleted (kappa <0.40). 10. The adapted instrument developed in steps 7 to 9 was administered to a sample of households from one Cape Town site (Nomzamo) and three Johannesburg sites (Slovo Park, Vlakfontein, and Eldorado). 11. The internal consistency reliability was measured. Among a few different internal consistency estimates, Cronbach (1951) was calculated for each dimension of

E.R. Jordaan et al. / Journal of Safety Research 36 (2005) 269 – 280

2.1. Description of the study sites

the measurement. The items with a high correlation with the total score were retained and aggregated into a score.

Nomzamo is an informal settlement located approximately 45 km east of Cape Town. The Nomzamo settlement comprises an estimated 8,008 residents and 1,820 households and can be characterized by its low-income context, where about 87% of households live below the subsistence level. Due to a lack of basic and health care facilities, young children living in Nomzamo may be more at risk for injury. The dominant spoken language in Nomzamo is Xhosa, followed by Sotho, and Afrikaans. Eldorado Park is situated in the South of Johannesburg next to Soweto and Lenasia and surrounded by three informal settlements (Joe Slovo Park, Mandela Square, Freedom Park). The area is made up of a mixture of middle- to low-income households. It’s comprised predominantly of people historically classified as ‘‘colored’’ in apartheid nomenclature and Afrikaans is the main spoken language. The population is about 300,000 to 350,000. It is now a more established, stable residential area dominated by low-cost, formal, small red brick housing, and, in some areas, council owned apartment blocks. Residents of surrounding informal settlements often use the facilities within Eldorado Park. Joe Slovo Park is situated in the south of Johannesburg between Eldorado Park and Nancefield Industrial Park. This is a predominantly Black African community and the spoken languages are Zulu, South Sotho, Tswana, and Tsonga. There are about 1,016 households with an average of about 4 to 5 people per house.

Steps 7 to 11 addressed the reliability and validity of the instrument. 12. We administer the instrument to the same sample of households from the Cape Town and Johannesburg sites as in step 10, after completing an intervention. 13. The responsiveness, which is the ability of a scale to detect change in the household’s score over time (i.e., the ability to measure the intervention effect) was calculated through comparison of pre-assessment scores with the post-assessment scores (Osborne, Hawthorne, Lew, & Gray, 2003). Steps 12 to 13 addressed the responsiveness of the instrument. 14. Again, the adapted instrument was administered to another sample of households from the Cape Town site and one of the Johannesburg sites (Slovo Park). 15. Test-retest reliabilities and internal consistency reliabilities were again calculated, with the focus on the falls scale and the paraffin appliances scale. 16. We arrive at a set of acceptable measures and a measurement model that provides a good fit to the data. 17. The implemented study will give us the opportunity to confirm the fit of the measurement model.

PRE-PILOT

PILOT 1 PRE-ASSESSMENT

Cape Town: Nomzamo 53

Cape Town: Nomzamo 85

PILOT 2

POST-ASSESSMENT

Intervention

Johannesburg: Eldorado Park 90 Vlakfontein 90 Slovo Park 90

Cape Town: Nomzamo 71 (30 C, 41 I) Johannesburg Eldorado Park 86 (42 C, 44 I) Vlakfontein 87 (43 C, 44 I) Slovo Park 84 (41 C, 43 I)

Ceiling/floor effects Test-retest Reliability

Internal consistency Reliability

271

Efficiency

C control households I intervention households

Fig. 1. Data and research methods.

Cape Town Nomzamo 57 Johannesburg:

Slovo Park 56

Falls and Paraffin Test-retest Reliability Internal consistency Reliability

272

E.R. Jordaan et al. / Journal of Safety Research 36 (2005) 269 – 280

Table 1A Pre-pilot: Test-retest reliability for Safety Practices (n = 53) Risk item

% safe % Observed Kappa 95% C1 houses agreement

Medicines are out reach Medicines in locked cupboards Household cleaners are kept in locked cupboard Alcohol out of reach Cosmetics out of reach Cigarettes/tobacco out of reach Matches are kept out of reach Child stays out of kitchen when cooking Fill paraffin stove/lamp Place stove safe place Place heater safe place Children’s toys kept safe place Matches lit safely Bath prepared safely Children Play safe place Get to school safely Pot handles safe direction Safe fire extinguishing Leave hot drinks in safe place Do not drink coffee with child on lap Electrical shock from appliance Fuses tripped in the house in last month

89 85 85

86.8 83.0 81.0

0.35* 0.00 to 0.81 0.29* 0.00 to 0.69 0.47 0.14 to 0.80

86 87 90** 89 57

77.6 84.9 80.0 86.8 54.7

0.59 0.61 0.59 0.00* 0.26*

0.38 0.30 0.38 0.00 0.04

to to to to to

0.79 0.91 0.80 0.01 0.49

63 67 61 71 69 77 83 82 91** 94** 96** 70

80.4 73.0 82.1 87.8 73.0 86.5 79.0 91.0 86.5 98.0 88.2 59.6

0.68 0.45 0.65 0.55 0.52 0.65 0.29* 0.66 NA NA 0.30* 0.34*

0.47 0.23 0.41 0.20 0.32 0.42 0.00 0.35

to to to to to to to to

0.88 0.68 0.89 0.89 0.72 0.89 0.61 0.97

96** 46

NA 82.7

NA 0.70

0.00 to 0.64 0.11 to 0.58

and South Sotho. The questionnaires were originally designed in English and then translated into the three main languages, Xhosa, Sotho, and Zulu. The questionnaires were translated by professional translators or members of the research organization and referred back to community members for appropriate language use. In Pilot 1, only the Xhosa and English versions were used. 2.4. Scale of measurements Most items were measured on a binary scale. Either the respondent was asked or the interviewer checked whether a specific risk was prevalent, whichever was more applicable. For example, the interviewer would check whether there were loose carpets and the respondent would be asked whether the stove was used as a heater as well. Some items were measured on a five-point Likert-type scale (Safety practices, for example ‘‘My child stays out of the kitchen while I am cooking’’). Most items had a response option of not applicable in addition to the other response options. A few items asking about injury experience were open-ended questions. 2.5. Pre-pilot: Cape town only

0.51 to 0.88

NA Too few responses to calculate. * These items are unreliable. ** These items have high safety in most households.

This phase was conducted in one area in Cape Town only. A random sample of 53 Nomzamo households was Table 1B Pre-pilot: Test-retest Reliability for Housecheck (n = 53)

Vlakfontein is situated further South of Johannesburg alongside the Golden Highway near the middle class community of Zackaryah Park. Vlakfontein Phase One has about 700 houses with an average of about five people per household. It comprises part of the Thembelihle community that chose to be relocated. They are currently enjoying services like water, sanitation, and electricity. It is a predominantly Black African community and the main spoken languages are Zulu, South Sotho, Tswana, and Sepedi. 2.2. Sampling, data and analysis An overview of the data and research methods is given in Fig. 1 and the details are discussed below. The study was conducted in three phases: A Pre-Pilot, Pilot 1 (pre-assessment and post-assessment), and Pilot 2. Whereas in Cape Town only one site participated, in Johannesburg three sites were recruited for participation in the study. Since our focus was on the injury risk measurement in the low-income household, the household was our unit of measurement. 2.3. Language and translations The languages spoken in the four communities are English, Xhosa, Sotho, Afrikaans, Tsonga, Zulu, Tswana,

Risk item

% safe Observed Kappa 95% C1 houses agreement

Electrical cords in walking areas Electricalds under furniture Frayed cords Knotted cords Cords attached with nails Outlets have no safety covers Outlets have no faceplates Heaters away from combustibles Candles away from combustibles Stove away from combustibles Type of iron electric Candle stable surface Heater stable surface Stove stable surface Paraffin lamp stable surface Stove in reach of children Heater in reach of children Flammable liquids in reach of children Cigarette lighter in reach of children Matches in reach of children Electricity outlet has waterproof cover Fence around home Sharp corner

88 81 88 83 68 36 34 48 45 69 79 44 31 78 63 79 86 94**

86.5 79.0 86.5 76.9 82.0 80.0 88.0 56.3 55.1 63.5 80.9 60.4 82.2 72.0 65.3 67.9 76.5 58.0

0.49 0.48 0.53 0.43 0.67 0.57 0.74 0.29* 0.29* 0.20* 0.32* 0.38* 0.66 0.24* 0.40 0.32* 0.60 0.20*

92**

58.8

0.23* 0.00 to 0.47

94** 43

58.0 69.8

0.13* 0.00 to 0.35 0.52 0.33 to 0.70

68 86

79.2 56.0

0.55 0.35 to 0.74 0.23* 0.00 to 0.86

* These items are unreliable. ** These items have high safety in most households.

0.19 0.23 0.23 0.15 0.47 0.33 0.55 0.08 0.08 0.00 0.00 0.16 0.47 0.00 0.17 0.09 0.41 0.00

to to to to to to to to to to to to to to to to to to

0.80 0.73 0.84 0.70 0.87 0.81 0.93 0.51 0.50 0.47 0.64 0.59 0.86 0.53 0.63 0.55 0.79 0.43

E.R. Jordaan et al. / Journal of Safety Research 36 (2005) 269 – 280

selected. The households were selected on the basis of maps that contained erf numbers for the area. The questionnaire was divided into three sections. The first section explored the household composition and demographics of the respondent. The second section explored safety attitudes and practices. And the third section required the interviewers to perform a housecheck in the home. The questionnaire was constructed using items gleaned from a literature search identifying child injury risk factors as well as existing questionnaires and checklists (Gielen, Wilson, Faden, Wissow, & Harvilchuck, 1995; Hapgood et al., 2000). A total of 22 safety practices items and 23 housecheck items were initially generated from a variety of sources using the table of specifications (Tables 1A and 1B). As suggested by Pedhazur and Schmelkin (1991), both positively and negatively worded items were included. Double barreled items, lengthy items, items with difficult vocabulary or multiple negatives, and ambiguous pronoun references were avoided (DeVellis, 2003). The instrument was subjected to expert review. The information was supplemented with focus group discussions in the community. The above process ensured high face validity and high content validity (appropriate coverage of the subject matter; DeVellis, 2003). The data from this pre-pilot were analyzed and the results were used to improve the questionnaire layout and the structure of questions and to eliminate some questions deemed to be unreliable. Repeated observations were made one week after the initial interviews. Each item was repeated and Kappa values calculated to measure the testretest reliability of the individual items. Floor and ceiling effects were examined to assess the feasibility of using the items considered for inclusion in the questionnaire. The refined questionnaire from this phase was then used in the first pilot phase. 2.6. Pilot 1: Cape Town and Johannesburg 2.6.1. Pre-assessment A stratified random sample of 85 households was selected from two geographic areas in Nomzamo, with approximately equal number of households from each of the two strata. The households were selected on the basis of maps that contained erf numbers for the areas. A stratified random sample of 90 households was selected from two geographic areas in Vlakfontein, Eldorado Park, and Jo Slovo Park, with 45 households from each stratum. The households were selected on the basis of maps that contained erf numbers for the areas. The data from this first pilot were analyzed and the results were used to improve the questionnaire. The internal consistency reliability was measured. Among a few different internal consistency estimates, Cronbach (1951) was calculated for each dimension of the measurement. The items with a high correlation with the total score were retained and

273

aggregated into a score. Risk scores were calculated for falls, burns, and poisonings. 2.6.2. Post-assessment An intervention was conducted in the households in one of the two strata from each of the four sites. The intervention consisted of five weekly visits to designated households by trained home visitors. The visitors visited the homes in pairs. At each visit, a different safety promotion topic was introduced, including burns, poisoning, falls, and traffic injury prevention, and child development as it relates to injury prevention. At each visit, the visitors discussed safety issues, performed a short injury risk checklist, demonstrated safety practices, and provided the household with a safety-related gift, such as paraffin safety caps. After an intervention was conducted, a postassessment was conducted at the same houses that participated in the pre-assessment. The post-assessment data were used to investigate the efficiency of the scales to detect small intervention effects of households receiving a short home visitation intervention or no intervention. Risk differences between households from the pre-assessment and post-assessment were calculated and tested for significant differences. Although the sample size was too small to indicate a significant change in the intervention effect, a change in the right direction was anticipated, that is, a positive mean change. The refined questionnaire from this phase was then used in the second pilot phase. 2.7. Pilot 2: Cape Town and Johannesburg This phase was conducted in one area in Cape Town and one area in Johannesburg only. A random sample of 57 Nomzamo households was selected and a random sample of 56 Slovo Park households was selected. The households were selected on the basis of maps that contained erf numbers for the area. A test-retest study was completed a week later. The refined questionnaire from this phase was then used in the implemented phase.

3. Results 3.1. Pre-pilot 3.1.1. General changes to the questionnaire Some items were moved from the general household section to the housecheck section. Rather than asking respondents about certain risks, interviewers made their own observations (e.g., ‘‘medicines are kept out of reach of my child,’’ ‘‘housecleaners are kept in a locked cupboard,’’ alcohol is kept out of reach’’). The frequency of floor and ceiling effects (Tables 1A and 1B) showed that some items

274

E.R. Jordaan et al. / Journal of Safety Research 36 (2005) 269 – 280

Table 2A Pilot 1 (pre-assessment): Internal consistency reliability (n = 355) site

Nomzamo Eldorado Slovo Park Vlakfontein a b c d

Fallsa

n

This This This This

score score score score

85 90 90 90 is is is is

0.65 0.25 0.38 0.62 an an an an

Poisonings

Burns

Negligent use of poisonous substancesb

Careless use of electrical appliancesc

Careless use of Paraffin and appliancesd

0.93 0.44 0.62 0.69

0.85 0.80 0.83 0.78

0.11 0.23 0.45 0.41

aggregate aggregate aggregate aggregate

of of of of

3 items. 12 items. 28 items. 6 items.

had no room for improvement with respect to safety, since they were already treated in a safe way in most (above 90%) of the houses. These items were not included in the questionnaire designed for pilot 1 (e.g., ‘‘tobacco are kept out of reach,’’ ‘‘has anyone received an electric shock?’’). The list of electrical appliances included all possible appliances, but not all of these appliances were used by the households, and thus a list of relevant appliances were constructed from the results of the pre-pilot (kettle, iron, fridge, TV, lamps, heater, and stove). The questionnaire was restructured appropriately to accommodate these changes. The matrix format of some of the questions did not appear to be appropriate for the level of education of the interviewers and were also changed. It became clear that not enough information was assessed about the risks of using paraffin appliances in these households, since many of these households use paraffin exclusively or in combination with electricity. Therefore, a section including questions specifically asking about the use of paraffin appliances was added to the questionnaire for pilot 1. 3.1.2. Test-retest reliability On the test-retest results, some items were shown to be unreliable by the low kappa values (kappa <0.4). Some of Table 2B Pilot 1 (pre- and post-assessment): Efficiency of scales to detect small intervention effects of households receiving a home visitation intervention or no intervention in Nomzamo n Poison score Negligent use of Control 30 poisonous substances Intervention 41 Burns score Careless use of electrical appliancese Careless use of paraffin and appliances

Control Intervention Control Intervention

30 41 30 41

Mean SD changea 1.77 2.49

0.10 0.95 0.23 0.34

4.70 7.05

2.75 4.16 1.17 1.32

t

Table 2C Pilot 1 (pre- and post-assessment): Efficiency of scales to detect small intervention effects of households receiving a home visitation intervention or no intervention in Eldorado Park n

Mean SD changea

Poison risk Negligent use of Control 42 0.43 poisonous substances Intervention 44 0.09

1.80 1.44

Burns risk Careless use of electrical appliances Careless use of paraffin and appliances

4.25 5.67 0.59 0.63

Control Intervention Control Intervention

42 44 42 44

0.05 0.93 0.26 0.30

t

p

0.96 0.338

0.81 0.417 0.25 0.799

SD standard deviation. a Mean change=pre-assessment score-post-assessment score (a positive score indicates a decline in risk score which implies an improvement in the risk profile).

the unreliable questions were ‘‘matches are kept out of reach,’’ ‘‘my child stays out of the kitchen when I am cooking,’’ ‘‘heaters out of reach,’’ and ‘‘stove in reach child’’ (Tables 1A and 1B). Some of these unreliable items were scrapped from the questionnaire (‘‘stove in reach of child’’) and some were modified (e.g., ‘‘where are children’s toys kept,’’ ‘‘where do children usually play when at home’’), and others were moved to the housecheck section to be assessed by the interviewers (e.g., medicines, cleaners, cosmetics, and alcohol are in locked cupboards). 3.2. Pilot 1 3.2.1. Internal consistency reliability First, the scales for falls, poisonings, and burns were assessed using the Nomzamo data. The data from the other three sites were then used to confirm the scales obtained from the Nomzamo data.

Table 2D Pilot 1 (pre- and post-assessment): Efficiency of scales to detect small intervention effects of households receiving a home visitation intervention or no intervention in Slovo Park n

p

Mean SD changea

Poison risk Negligent use of Control 41 poisonous substances Intervention 43

0.15 0.84

1.99 2.95

Burns risk Careless use of electrical appliances Careless use of paraffin and appliances

0.24 0.16 0.41 0.40

0.66 2.48 1.14 1.09

t

p

1.80 0.076*

0.52 0.607

1.04 0.303 0.36 0.721

SD standard deviation. a Mean change=pre-assessment score-post-assessment score (a positive score indicates a decline in risk score which implies an improvement in the risk profile).

Control Intervention Control Intervention

41 43 41 43

0.21 0.837 0.08 0.937

SD standard deviation. a Mean change=pre-assessment score-post-assessment score (a positive score indicates a decline in risk score which implies an improvement in the risk profile). * Significant at the 1% level.

E.R. Jordaan et al. / Journal of Safety Research 36 (2005) 269 – 280 Table 2E Pilot 1 (pre- and post-assessment): Efficiency of scales to detect small intervention effects of households receiving a home visitation intervention or no intervention in Vlakfontein N

Mean SD changea

Poison risk Negligent use of Control 43 0.67 poisonous substances Intervention 44 1.25

1.95 2.29

Burns risk Careless use of electrical appliances Careless use of paraffin and appliances

2.95 3.57 1.46 1.35

Control Intervention Control Intervention

43 44 43 44

0.51 0.68 0.65 0.16

t

p

1.26 0.211

0.24 0.809 1.63 0.106*

SD standard deviation. a Mean change=pre-assessment score-post-assessment score (a positive score indicates a decline in risk score which implies an improvement in the risk profile). * Approach significance (at 1%).

3.2.1.1. Falls. Initially the scale for falls contained 12 items. The prevalence of the 12 items varied from 2% (loose wiring) to 71% (sharp corners on tables). The three most prevalent risks relating to sharp corners on furniture were tables, chairs, and cupboards. After assessing the reliability of the 12 items, only 3 items included in the final scale focused on the risk of falls related to sharp corners on furniture in the house (Table 2A). Including these three items, the reliability estimate for Nomzamo was 0.65 for the falls scale. The excluded items were ‘‘electrical cords in walking areas,’’ ‘‘toys on a high shelf,’’ ‘‘sharp corners on tables,’’ ‘‘loose mats,’’ ‘‘loose paving,’’ ‘‘broken glass,’’ ‘‘loose wiring,’’ and ‘‘other sharp objects in the yard.’’

275

The collection of items from the falls scale does not seem to cover all the important falls risks in the homes, and they seem to be a grouping of ad hoc items put together, not forming a uni-dimensional scale. Too few items were considered for this scale, and more items should be investigated. The poor reliability estimates for two (Eldorado Park and Slovo Park) of the four areas also confirm that this scale is not adequate for use as a falls scale. A re-assessment of the literature will be conducted and more items will be included for assessment in the next phase of the study. 3.2.1.2. Poison. The scale for poison risk initially comprised of 19 items. The prevalence of the items varied from 9% (number of windows in the room with paraffin heater) to 40% (no child proof cap for paraffin container). The three most prevalent risks related to the negligent use of paraffin or paraffin appliances (container not labeled, stored within reach of child, and no childproof cap). After assessing the reliability of the 19 items, 12 items were retained for a score variable. All 12 items had a correlation above 0.6 with the total score. The overall reliability estimates for the poison risk scale were 0.93 for Nomzamo, 0.62 for Slovo Park, and 0.69 for Vlakfontein (Table 2A). The 12 items included in the scale indicated negligent use of poisonous substances (e.g., beauty products, medicines, paraffin, cleaning products, spirits and alcohol, rat poison). The items with a poor correlation with the total score (below 0.60) were excluded. The excluded items were ‘‘paraffin fill,’’ ‘‘number of windows,’’ ‘‘paraffin cap,’’ ‘‘cleaning products stored on same shelf as food,’’ and ‘‘loose/chipped/peeling paint.’’

Table 3A Pilot 2: Test-retest reliability for Paraffin scale (n = 113)

12 13 27 14 29 34 35 36 22 15 16 20 23 25 30

Items

code

%

% observed agreement

kappa

95% CI

Do you use the stove as a heater Where do you fill the par stove Who fills the par stoves/heater/lamp Stove is hot/warm when filling What do you use to fill the paraffin appliances What kind of container is used to store paraffin What colour is the paraffin in the container Is anything else ever stored in the paraffin container Do you use par for something else not yet mentioned Show me where you put your stove when cooking Is this a stable surface Do you use par for lighting Do you open windows or doors when you are using par stove/heater or lamp Do you use par appliances when people are sleeping From what age are children allowed to be alone in the same room as paraffin appliances

yes Inside Children under 10 Children over 10 yes Cup/bottle/jug Glass/cool drink bottle Cloudy/not see-through yes Yes Floor/crate/chairs/bricks no yes no

39.8 72.3 4.4* 6.2* 21.2 52.2 12.4 31.8 6.7* 24.8 9.8 34.9 45.7

86.1 78.0 96.5 90.3 88.5 88.5 88.5 91.6 94.2 88.5 90.7 85.0 100

0.70 0.38** 0.32** NA 0.57 0.77 0.33** 0.82 0.59 0.67 0.36** 0.71 1.00

0.55 to 0.84 0.17 to 0.59 0 to 0.80

Always/Usually/sometimes Less than 10 yrs

39.8 20.6

64.6 81.4

0.23** 0.42

0.05 to 0.41 0.20 to 0.64

NA not applicable/too few responses to calculate. * These items have low frequencies. ** These items are unreliable.

0.37 0.65 0.05 0.70 0.30 0.51 0.03 0.57 NA

to to to to to to to to

0.77 0.89 0.60 0.93 0.89 0.84 0.68 0.86

276

E.R. Jordaan et al. / Journal of Safety Research 36 (2005) 269 – 280

The final scale was aggregated into a score and labelled ‘‘negligent use of poisonous substances.’’ In general, the internal consistency seemed adequate for use as a reliable measuring instrument for poisonings. However, the internal consistency reliability for Eldorado Park is quite poor (Cronbach alpha=0.44, Table 2A). On closer inspection, we found that the poison risk profile for Eldorado Park was very different to the other sites, with generally fewer homes with poison risks, except for risks involving cleaning products. 3.2.1.3. Burns. The burn scale initially included 59 items (not shown). The prevalence of the items varied from 0% (‘‘electrical heater plugged in when not in use’’) to 72% (‘‘child in kitchen when cooking’’). The four most prevalent risks were ‘‘child in kitchen when cooking’’ (72%), ‘‘paraffin stove placed less than 30cm from combustibles’’ (41%), ‘‘electric lamps within 30cm from water’’ (45%), and ‘‘matches within reach of children’’ (41%). After assessing the reliability of the 59 items, 28 items were selected, indicating negligent use of electrical appliances, including placing the electrical equipment (kettle, iron, fridge, radio, TV, lamps, heater, and stove) within 30cm from a source of water, electrical cords hanging over counter tops, plugged in when not in use, and exposed wires. These items were aggregated into a

score and called ‘‘careless use of electrical appliances.’’ The reliability estimates for this scale were 0.85 Nomzamo, 0.80 for Eldorado Park, 0.83 for Slovo Park, and 0.78 for Vlakfontein (Table 2A). Those items with a poor correlation with the score were excluded. The excluded items were items about paraffin equipment, storing of matches, child protection when cooking or working with hot items, filling of paraffin equipment, and extinguishing of fires. Six items pertaining to the use of paraffin and paraffin appliances were aggregated into a separate score, and called ‘‘Careless use of paraffin and appliances’’. This included items about how the paraffin appliances are filled, where they are placed when in use, whether the surface they are used on is stable, and whether there is a long tablecloth under the appliance when in use. The overall reliability estimates for this scale were quite poor for all four areas (Table 2A). In general, the internal consistency for the scale, ‘‘careless use of electrical appliances,’’ seemed adequate for use as a reliable measuring instrument, but the paraffin and appliances scale needs some more refinement. 3.2.2. Efficiency Risk differences between households from the preassessment and post-assessment were calculated for the three risk outcomes ‘‘negligent use of poisonous substances,’’ ‘‘careless use of electrical appliances,’’ and ‘‘careless

Table 3B Pilot 2: Test-retest reliability for falls scale (n = 113)

46 43 40 45 38 42 39 56 57 37 4852 55 61 67 62 65 63 64

Items

code

%

% OG

kappa

95% CI

Is the child supervised in a baby walking ring How often do you leave your children (under 3) unattended on a high surface Does the child under age 3 play alone on the bed From what age are your children allowed to bath alone Does the child climb on any furniture or anything else during play Does any child climb on something and jump on the bed from there Does the child (under 10) run inside the house Check: Are there any uneven floor surfaces? Check: floors slippery Does the child climb on any furniture or anything else to reach something Does any furniture (cupboard, bed, chair, table) have sharp corners Check: Are there any lose carpets on the floor Check: High step to the house Child climb on roofs of buildings in yard Do children climb on gates inside/outside house Are there any objects in the yard child climb onto Check: Lose/uneven paving Check: Any lose wire lying around

Usually/sometimes/never Usually/sometimes/never

28.3 16.8

80.5 78.8

0.46 0.13**

0.27 to 0.65 0 to 0.35

Usually/sometimes/never Under 10 Usually/sometimes/never

34.5 78.4 56.6

67.3 79.1 77.0

0.24** 0.41 0.54

0.06 to 0.43 0.21 to 0.61 0.39 to 0.69

Usually/sometimes/never

30.1

78.8

0.46

0.28 to 0.64

Usually/sometimes/never yes yes Usually/sometimes/never

67.3 54.9 11.5 61.1

74.3 85.8 93.8 76.1

0.48 0.72 0.66 0.53

0.32 0.59 0.43 0.38

yes

69.0

78.8

0.56

0.41 to 0.71

yes yes Usually/sometimes/never Usually/sometimes/never yes yes yes

21.2 21.4 5.3* 33.6 24.1 22.3 22.3

78.8 84.8 97.3 77.9 87.4 86.6 77.7

0.27** 0.44** 0.65** 0.47** 0.65** 0.56** 0.21**

0.06 to 0.49 0.23 to 0.65 0.29 to 1.00 0.29 to 0.64 0.48 to 0.81 0.36 to 0.75 0.0 to 0.42

NA not applicable/too few responses to calculate. OG observed agreement. * These items have low frequencies. ** These items are unreliable.

to to to to

0.63 0.84 0.89 0.68

E.R. Jordaan et al. / Journal of Safety Research 36 (2005) 269 – 280

use of paraffin and appliances.’’ Although the sample size was too small to indicate a significant change in the intervention effect, a change in the right direction was anticipated, that is a positive mean change. Although the paraffin appliance scale has poor reliability, we are including this score in the efficiency analysis to assess its performance in an intervention phase (Tables 2B, 2C, 2D, 2E). The only significant change (at the 1% level) was found for the poison risk score (Slovo Park; t= 1.80, p=0.076). One of the risks was approaching significance (Vlakfontien, paraffin appliances risk score) and some were in the right direction (Nomzamo, Eldorado, Vlakfontein, careless use of electrical appliances). The internal consistency reliability results, as well as the efficiency results, show poor performance for the paraffin and appliances score (Tables 2A, 2B, 2C, 2D, 2E).

277

Another pilot study to refine the falls items as well as the paraffin items should be undertaken before the implemented study. 3.3. Pilot 2 3.3.1. Paraffin and Appliance scale The test-retest reliability results (kappa values) and the internal consistency reliability results (the correlation of each with the total score) are given in Tables 3A and 3C, respectively. The test-retest analysis shows that 3 of the 15 items had a very low prevalence, and five questions were unreliable (kappa value below 0.40). The internal consistency results show that six items had a negative correlation with the total score. The Cronbach alpha value for this scale is

Table 3C Pilot 2: Internal consistency reliability and change of wording for paraffin scale (n = 113) Pilot 2

Implemented phase

Wording of questions

code

r

Wording of Question

code

Changes

12

yes

0.17

yes

Same

inside

0.05

inside

children under 10 or children over 10 yes

0.23 0.35

Is the paraffin stove used as a heater Show me where the stove/heater/lamp is filled Who fills the par stoves/heater/lamp Stove is hot/warm when filled

yes

Show: more specific and all paraffin appliances Options more clear/Age more specific Same

cup/bottle/jug

0.11

cup/bottle

mention paraffin appliances

glass/cool drink bottle cloudy/not see-through yes

neg

no

0.28

Safety knowledge is assessed rather Safety knowledge is assessed rather same

yes

neg

floor/crate/ chairs/bricks no yes

0.18 neg neg

Do you open windows or doors when you are using par stove/heater or lamp Do you use par appliances when people are sleeping

no

0.02

always/usually/ sometimes

0.07

From what age are children allowed to be alone in the same room as paraffin appliances No question about candles

less than 10 yrs

neg

13 27 14 29 34 35 36 22

15 16 20

23

25

30

Do you use the stove as a heater Where do you fill the par stove Who fills the par stoves/heater/lamp Stove is hot/warm when filling What do you use to fill the paraffin appliances What kind of container is used to store paraffin What colour is the paraffin in the container Is anything else ever stored in the paraffin container Do you use par for something else not yet mentioned Show me where you put your stove when cooking Is this a stable surface Do you use par for lighting

No question about a tablecloth

neg

What is used to fill the stove/ lamp/heater Is it safe to store your paraffin in a cool drink bottle Is it safe to use paraffin that is see through Is anything else ever stored in the paraffin container Show me where the paraffin heater is placed when in use Show me where the paraffin stove is placed when cooking Show me where the paraffin lamp is placed when in use Are windows and doors always open when par appliances are used Are par appliances on when everyone in the house is sleeping Are children under 10 ever alone in the same room as par appliances Show me where your candles are places when in use Check: is there a long tablecloth under candles/par heater/stove/lamp

children under 18

no yes less than 30cm from combust unstable surface less than 30cm from combust unstable surface less than 30cm from combust unstable surface not open or no window

Ask about specific other par appliances used Same: more specific to interviewers re responses Ask more specific about use of lamp: more specific to interviewer re response Included option of no window available

yes

Use of appliances is changed to appliances on

yes

Age is more specific easier to respond

unstable surface less than 30cm from combust no

278

E.R. Jordaan et al. / Journal of Safety Research 36 (2005) 269 – 280

Table 3D Pilot 2: Internal consistency reliability and change of wording for falls scale (n = 113) Implemented phaseb

Pilot 2 a

Wording of questions

code

r

46

Usually/sometimes/ never Usually/sometimes/ never

neg

Usually/sometimes/ never Under 10

0.22

Is the child supervised in a baby walking ring How often do you leave your children (under 3) unattended on a high surface Does the child under age 3 play alone on the bed From what age are your children allowed to bath alone

43

40 45

38

Does the child climb on any furniture or anything else during play 42 Does any child climb on something and jump on the bed from there 39 Does the child (under 10) run inside the house 56 Check: Are there any uneven floor surfaces 57 Check: floors slippery 37 Does the child climb on any furniture or anything else to reach something 48 – 52 Does any furniture (cupboard, bed, chair, table) have sharp corners 55 Check: Are there any lose carpets on the floor 61 Check: High step to the house 67 Child climb on roofs of buildings in yard

0.07

0.21

Wording of Question

Do children under 5 bath alone

yes

Age is spelled out

Do children under 5 ever climb on surface higher than 50 cm

yes

Age is more specific

yes

Response Options easier for interviewer Interviewer cjecks

0.61

Usually/sometimes/ never

0.45

Usually/sometimes/ never yes

0.62

Does child under 10 run around inside house 0.13*** Check: any uneven floors?

yes Usually/sometimes/ never

0.04*** Check: floors slippery 0.68 Show where toys, sweets anything else are kept

yes

0.30

yes

0.06

Check: sharp corners on tables, chairs or cupboards

62

Do children climb on gates inside/outside house

Usually/sometimes/ never

65

Are there any objects in the yard child climb onto Check: Loose/uneven paving Check: Any loose wire lying around

yes

Check: any rugs and mats lose 0.11*** Check: High (over 30cm) step/stairs to house 0.13 Do children under 5 play outside without adult supervision 0.37 Check: Are there ay objects (over 50cm) that child under 10 an climb onto 0.28

yes yes

0.01*** Check: Loose/ uneven paving 0.11*** Check: Loose wires or cords

63 64

Usually/sometimes/ never

Changes Age spelled out

Usually/sometimes/ never

yes

code

Is the baby (1 year or younger) no supervised when in baby ring Is the child under age 3 ever yes unattended on surface higher 50cm

yes

The height of the surfaces are spelled out (taken from the literature)

yes same On high shelf In the kitchen near stove yes Interviewer checks

yes

same

yes

Height of step more specific Cover all play outside the house

yes

Cover all outside objects with height over 50cm/age spelled out

yes yes

same

yes

a

The correlation coefficient of each item with the total score. This column refers to the final questionnaire used in the implemented phase. *** These are 5 items referring to risks outside the house and/or risks pertaining to the house construction-might form a separate subscale (excluding these 5 items, Cronbach alpha=0.71). b

0.42, which is satisfactory. In Table 3C the authors make suggestions on how the questions can be altered to improve the reliability of this scale for the implemented phase: 1) A change of focus from an inactive response to an active response to show (where do you fill the paraffin appliances).

2) The relevant age of the children at risk is spelled out (who fills the paraffin appliance and are children under 10 ever alone in the same room as paraffin appliance). 3) Rather than asking about a risk that is time dependent, the safety knowledge is assessed (Is it safe to use paraffin that is transparent and what color is the paraffin in the container?).

E.R. Jordaan et al. / Journal of Safety Research 36 (2005) 269 – 280

4) All other paraffin appliances are asked about more specifically (show me where the paraffin heater is placed when in use, show me where the paraffin lamp is placed when in use). 5) Options for the interviewers are more clearly specified (show me where the paraffin stove is placed when cooking). 3.3.2. Falls scale The test-retest reliability results (kappa values) and the internal consistency reliability results (the correlation of each with the total score) are given in Tables 3B and 3D, respectively. One of the 15 items had a very low prevalence (‘‘Child climbs on roofs of buildings in yard’’). The testretest analysis shows that a number of questions were unreliable (kappa value below 0.40). The internal consistency results show that only one item had a negative correlation with the total score (‘‘Is the child supervised in a baby walking ring’’). The Cronbach alpha value for this scale is 0.64, which is acceptable. In Table 3D the authors make suggestions as to how the questions can be altered to improve the reliability of this scale for the implemented phase: 1) The sequence of the questions were changed so that the questions were in a chronological order according to children’s developmental stages: questions were first asked about the babies, then about children aged 3 years, and then about children aged 5 to 10 years of age. 2) The questions about the outside area are placed last, so that these can be checked just before leaving (lose/ uneven paving, lose wires/cords). 3) The risk heights for the surfaces and the age relevant to these heights are spelled out (is the child under 3 ever unattended on surfaces higher than 50cm, do the children under 5 bath alone).

4. Discussion During the three phases of the study, the measurement instrument has gone through extensive development. The scales we developed for measuring the injury risks (falls, burns, and poisonings) in a low socio-economic community comply with all the requirements for a valid and reliable measurement instrument. The risks pertaining to burns were measured using two separate scales, relating to the source of the risk, whether from a paraffin source (14 items) or from an electrical source (28 items). The two scores would ultimately be combined into a single risk score. This should be done in a way that gives equal weighting to the two sources of risk. The provision of this instrument may bring enormous benefits, not only to the HVP study, but also to other studies focused on assessing injury risks in low-income communities. It overcomes some serious limitations of traditional

279

measures such as: (a) It provides information about the risk problems of the households, and specifically about lowincome households with limited access to electricity; (b) It provides a means to evaluate risk exposures of households; and (c) It offers information about injury problems before they develop into accidents and injuries. Thus, they allow focusing of safety efforts for improvement of problematic areas in the households and the knowledge of these communities with limited resources. (d) It offers comprehensive information about injuries, and supplements the information obtained from hospital surveillance programs (WHO, 2002). In addition, it allows the valid comparison of risks between communities and serves as an evaluation tool for the HVP, which was the main motive for development of a measurement instrument. However, the development of an instrument is a continual process, one in which an end point is rarely achieved, but it is only successively approximated. Therefore, confirmatory analysis will be used to confirm the measurement model when the data for the implementation phase becomes available. The findings of this study should be interpreted in the context of the following limitations. First, the findings may not be generalizable to other communities. The selected communities have a low income, low electricity consumption and high paraffin consumption (South African Demographic and Health Survey, 1998). It was also clear from the analysis that the measurement instrument fared worse in Eldorado Park, which has a different risk profile for poisonings as well as burns, evidently due to the more formal housing nature of the area. Ninety-four percent of the houses in Eldorado Park use electricity for cooking, while in the other three areas only 40% use electricity for cooking, and most houses use paraffin for cooking (70%). Second, the results of this study were dependent upon sincere and honest responses of each caretaker. To maximize the integrity of the responses, voluntary participation in the survey, anonymity, and confidentiality were emphasized. Third, volunteers from the communities were used for data collection. They were trusted by the communities and could speak their home language. To ensure standardization of the survey, the supervisors checked the volunteers. The researchers have implemented the home intervention program (HVP) in two of the sites in an implementation phase of the study, using the measurement instrument discussed in this paper.

References Atkins, S., van Niekerk, A., Bender, S., & Seedat, M. (2002). Home visitation for childhood injury reduction: A review of the literature. Tygerberg’ Unisa Institute for Social and Health Sciences and Centre for Peace Action. Barss, P., Smith, G., Baker, S., & Mohan, D. (1998). Injury prevention: An international perspective. New York’ Oxford.

280

E.R. Jordaan et al. / Journal of Safety Research 36 (2005) 269 – 280

Bass, J. L., & Mehta, K. A. (1980). Developmentally oriented safety surveys: Reported parental and adolescent practices. Clinical Pediatrics, 19(5), 350 – 356. Bass, J. L., Mehta, K. A., Ostrovsky, M., & Halperin, S. F. (1985). Educating parents about injury prevention. Pediatric clinics of North America, 32(1), 233 – 242. Bender, S., van Niekerk, A., Seedat, M., & Atkins, S. (2002). Best practices for home visitation as a childhood injury prevention intervention, African Safety Promotion. A Journal of Injury and Violence Prevention, vol. 1 (1). (pp. 46 – 54). Tygerberg’ Medical Research Council. Bradshaw, D., Bourne, D., & Nannan, N. (2003). What are the leading causes of death among South African children? MRC Policy Brief, vol. 3. Tygerberg’ Medical Research Council. Child Accident Prevention Trust. (1989). Basic principles of child accident prevention. London’ Author. Cronbach, L. (1951). Coefficient alpha and the internal structure of tests. Psychometrica, 16, 297 – 334. DeVellis, R. F. (2003). Scale development: Theory and applications (2nd edR). Thousand Oaks, CA’ Sage Publications. Diderichsen, F., Laflamme, L., & Hallqvist, J. (1999). Understanding the mechanisms of social differences in injuries. In L. Laflamme, L. Svanstrom, & L. Schelp (Eds.), Safety Promotion Research. Stockholm’ Karolinska Institutet. Dowswell, T., Towner, E., Simpson, G., & Jarvis, S. N. (1996). Preventing childhood unintentional injuries-what works? A literature review. Injury Prevention, 2, 140 – 149. Fleiss, J. L. (1981). Statistical methods for rates and proportions (2nd edR). New York’ John Wiley and Sons. Gielen, A. C., Wilson, M. E. H., Faden, R. R., Wissow, L., & Harvilchuck, J. D. (1995). In-home injury prevention practices for infants and toddlers: The role of parental beliefs. Barriers, and Housing Quality, Health Education Quarterly, 22(1), 85 – 95. Hansen, L., Wong, D., & Young, P. C. (1996). Do the Framingham safety surveys improve injury prevention counseling during pediatric health supervision visits? The Journal of Pediatrics, 129(4), 494 – 498. Hapgood, R., Kendrick, D., & Marsh, P. (2000). How well do sociodemographic characteristics explain variation in childhood safety practices? Journal of Public Health Medicine, 22(3), 307 – 311. Johnson, M., Cusick, A., & Chang, S. (2001). Home-screen: a short scale to measure fall risk in the home. Public Health Nursing, 18(3), 169 – 177. Krug, E. (1999). Injury; A leading cause of the global burden of disease. Geneva’ World Health Organisation. Laflamme, L. (2001). Explaining socio-economic differences in injury risks. Injury Control and Safety Promotion, 8(3), 149 – 153. Mackenzie, L., Byles, J., & Higginbotham, N. (2002). Professional perceptions about home safety: cross-national validation of the home falls and accidents screening tool (HOME FAST). Journal of Allied Health, 31(1), 22 – 28. Mohamed, F. (2001). Considerations in designing a volunteer home visitation programme in low-income, at- risk communities: a review of the literature. Institute for Social and Health Sciences Monograph Series, 2(4), 32 – 63. The Nomzamo Safety Promotion Fact Book. (1998). MRC-Unisa Crime, Violence and Injury Lead Programme. The Nomzamo safety promotion fact book, unpublished. Osborne, R. H., Hawthorne, G., Lew, E. A., & Gray, L. C. (2003). Quality of life in the community-dwelling elderly: Validation of the Assessment of Quality of Life (AQoL) Instrument and comparison with the SF-36. Journal of Clinical Epidemiology, 56, 138 – 147. Pedhazur, E. J., & Schmelkin, L. P. (1991). Measurement, design and analysis: An integrated approach. Hillsdale, NJ’ Lawrence Erlbaum Associates. Safekids (undated). Childproofing basics. Retrieved 4 April, 2002, from www.Safekids.org/tier3_cd.cfm?content_item_id=326&folder_id=174

South African Demographic and Health Survey. (1998). Full report. Department of Health, Medical Research Council and Macro International. Cape Town’ Department of Health. Towner, E., & Dowswell, T. (2002). Community-based childhood injury prevention interventions: what works? Health Promotion International, 17, 273 – 284. Tymchuk, A. J., Lang, C. M., Dolyniuk, C. A., Berney-Ficklin, K., & Spitz, R. (1999). The home inventory of dangers and safety precautions-2: addressing critical needs for prescriptive assessment devices in child maltreatment and in healthcare. Child Abuse and Neglect, 23(1), 1 – 14. Tymchuk, A. J., Lang, C. M., Sewards, S. E., Lieberman, S., & Koo, S. (2003). Development and validation of the illustrated version of the home inventory for dangers and safety precautions: Continuing to address learning needs of parents in injury prevention. Journal of Family Violence, 18(4), 241 – 252. Van Niekerk, A., & Duncan, N. (2002). Editorial; A new journal for the promotion of injury prevention in africa. African safety promotion. A Journal of Injury and Violence Prevention, vol. 1(1). (pp. 1 – 4). Tygerberg’ Medical Research Council. Vocational and Rehabilitation Research Institute (undated). VRRI Home Safety Assessment. Retrieved 4 April, 2002, from www.vrri.org/hmfas.htm World Health Organisation (2002). Injuries and violence prevention annual report. Geneva’ Author. Esme´ Jordaan has an MSc in Statistics and has been working at the Biostatistics unit of the Medical Research Council (MRC) for the past 23 years. She is currently a specialist statistician, collaborating on projects in the field of epidemiology and medical and clinical biostatistics. More specifically, she has been collaborating with the Crime, Violence and Injury Unit of the MRC for the past 4 years, working in the field of safety research amongst other areas. She teaches various courses in Statistics of which two postgraduate courses in Statistics of the University of the Western Cape are currently the major ones and is currently co-promoter for two postgraduate students. She is affiliated to the South African Statistical Association and presents talks on Statistical aspects of her consultation work at their annual conference. She reviews papers for the South African Statistics Journal, Archives of Environmental Health and other journals. She is author or co-author for many proposals and publications, both nationally and internationally. Salla-Aija A. Atkins received a M.A. Psychological Research and B. Soc. Sci. (Hons) in Psychology from the University of Cape Town, and a B. A. in Psychology and Communications from University of South Africa. Atkins position is Scientific Officer, Primary Health Care, Directorate, University of Cape Town. Ashley van Niekerk is a Senior Scientist at the Crime, Violence and Injury Lead Programme (which is jointly coordinated by the ISHS and the Medical Research Council). He coordinates the identification, development, implementation, and evaluation of effective and low-cost crime, violence and injury prevention interventions appropriate for South African settings. He has a Masters Degree (Clinical Psychology) from University of the Western Cape, BA Honours (Psychology), University of the Western Cape, BA (Social Sciences & Humanities), University of Cape Town. Mohamed Seedat was appointed as Director of the National Presidential Lead Programme on Crime, Violence and Injury (CVI) focused on injury surveillance, the research-policy nexus, and the development of sustainable programmes for violence and injury prevention. Most recently (July 2002 – June 2003) Mohamed Seedat was a visiting professor at the prestigious Indian Institute of Technology, New Delhi, India. He holds a doctor of Philosophy degree from the University of Western Cape.