Pool Fencing Cochrane Review

Pool fencing for preventing drowning in children

Thompson DC, Rivara FP

This review should be cited as: Thompson DC, Rivara FP. Pool fencing for preventing drowning in children (Cochrane Review). In: The Cochrane Library, Issue 2, 2001. Oxford: Update Software.

A substantive amendment to this systematic review was last made on 19 September 1997. Cochrane reviews are regularly checked and updated if necessary.

Background: In most industrialized countries, drowning ranks second or third behind motor vehicles and fires as a cause of unintentional injury deaths to children under the age of 15. Death rates from drowning are highest in children less than five years old. Pool fencing is a passive environmental intervention designed to reduce unintended access to swimming pools and thus prevent drowning in the preschool age group. Because of the magnitude of the problem and the potential effectiveness of fencing we decided to evaluate the effect of pool fencing as a drowning prevention strategy for young children.

Objectives: To determine if pool fencing prevents drowning in young children .

Search strategy: We used Cochrane Collaboration search strategy of electronic databases, searched reference lists of past reviews and review articles, Cochrane International Register of RCT's, studies from government agencies in the United States and Australia, and contacted colleagues from International Society for Child and Adolescent Injury Prevention, World Injury Network, and CDC funded Injury Control and Research Centers.

Selection criteria: In order to be selected a study had to be designed to evaluate pool fencing in a defined population and provide relevant and interpretable data which objectively measured the risk of drowning or near drowning or provided rates of these outcomes in fenced and unfenced pools. The completed studies meeting selection criteria employed a case-control design. No randomized controlled studies have been identified.

Data collection and analysis: Three published studies met selection criteria. Data were extracted by two reviewers using standard abstract form. Odds ratios with 95% CI, and incidence rates, were calculated for drowning and near-drowning. Attributable Risk percent (AR%) was calculated to report the reduction in drowning due to pool fencing.

Main results: Case control studies which evaluate pool fencing interventions indicate that pool fencing significantly reduces the risk of drowning. Odds ratio for the risk of drowning or near drowning in a fenced pool compared to an unfenced pool is 0.27 95%CI (0.16, 0.47). Isolation fencing (enclosing pool only) is superior to perimeter fencing (enclosing property and pool) because perimeter fencing allows access to the pool area through the house. Odds ratio for the risk of drowning in a pool with isolation fencing compared to a pool with three sided fencing is 0.17 95%CI (0.07, 0.44)

Reviewers' conclusions: Pool fences should have a dynamic and secure gate and isolate (i.e., four-sided fencing) the pool from the house. Legislation should require isolation fencing with secure, self-latching gates for all pools, public, semi-public and private.

Background

In most industrialized countries, drowning ranks second or third behind motor vehicles and fires as a cause of unintentional injury deaths to children under the age of 15. Death rates from drowning are highest in children less than five years old. Among children, the most important risk factors for drowning are age and, to some degree, location. For infants, bathtub drowning poses the greatest hazard. Once children attain mobility as toddlers (ages 1-4), swimming pools pose the greatest risk of immersion injury. Rates vary by area and are higher in locations where the weather is hot and exposure to swimming pools is the greatest. For example, the rate for children ages 0-4 in the United States is 3.9 per 100,000 but 9.4 per 100,000 in Arizona. (U.S. Mortality Data,) Preschool drowning rate in Australia is 8.2 per 100,000 and varies from 4.69 in the Australian Capital Territory (Canberra) to 15.7 per 100,000 in Queensland. (Pearn 1978a)

It has been estimated that for each childhood drowning fatality, about four children are hospitalized (Wintemute 1990) and 14 are seen in the emergency department and released (Spyker 1985). Geddis 1984 estimated that there were 10 "near misses", that is children sustaining immersion who were rapidly rescued, for each child seen in the emergency department. However, among those sustaining immersion and losing consciousness, the morality rate is as high as 50%. The outcome for most children with immersion is determined by their status on arrival to the emergency department; medical and ICU care once admitted appear to have relatively little impact on outcome.

Pool fencing is a passive environmental intervention designed to reduce unintended access to swimming pools and thus prevent drowning in the preschool age group.

Over the last decade many communities, particularly in Australia, have passed laws requiring the fencing of private and public pools. In fact, much of the evidence that pool fencing reduces the risk of drowning among children comes from studies examining before and after rates of drowning for fenced and unfenced pools. Even after some of the pioneering studies had been done, the favorable attitude toward pool fencing did not necessarily translate into actual changes in pool fencing.(Fergusson 1983, Nixon, 1986, Wintemute 1990a, Choo 1995)

More recently, studies have examined whether the type of fence surrounding a pool makes a difference. Comparison studies of perimeter fencing (property barrier) versus isolation fencing (around immediate pool area) show isolation fencing to be much more effective in reducing the risk of drowning.(Wintemute 1990). Adding to this finding are studies examining children's ability to climb certain types of fences. (Nixon 1979; Rabinovich 1994) In particular, chain-link fences, while allowing visibility of the pool area, are the most easily scaled by children as young as two years old. Ornamental iron bar fences were advocated as a better barrier, with reduced 'climb ability' while retaining the visibility factor. Fence height makes little difference if the child is able to climb; one study showed the median time for four-year-olds to climb a five-foot fence was 17 seconds. (Nixon 1979). For a fence that is not scaleable, the most important element of fencing is a secure, self-closing gate. Taking these and other finding into account, the U.S. Consumer Product Safety Commission has compiled a list of minimum recommendations (US-CPSC 1991) for residential pool fencing. Recent legislation passed in Seattle, Washington, USA requires a fence height of at least five feet, and an inter-bar spacing to be no more than four inches (Quan 1990).

Because of the magnitude of the problem and the potential effectiveness of fencing we decided to evaluate the effect of pool fencing as a drowning prevention strategy for young children.

Objectives

The specific aims were:

1. Comparison of drowning and near-drowning rates for fenced and unfenced pools.
2. Comparison of drowning rates for specific fencing types (isolation vs perimeter).
3. Calculation of attributable risk percent (AR%) to quantify the reduction in drowning attributed to pool fencing.

Criteria for considering studies for this review

Types of studies

Studies eligible for review are controlled studies which evaluate the effectiveness of pool fencing using some type of comparison group, provide relevant and interpretable data which objectively measure the risk of drowning or near drowning. The comparison could be to another group, whether it be part of an RCT, a controlled trial, a case-control study, a cohort study or an ecological study. Ecological studies either involve comparison across communities (ecological group study), or over time within a community as in a time series study or a before-after study (ecological time study). An ecological mixed study examines within-group changes and between-group changes. For example, such a study might compare drowning rates before and after passage of fencing legislation in State A, and also compare these data to mortality in State B, which has no such law during the entire period being examined.

Types of participants

Children 14 years of age or younger who are exposed to swimming pools.

Types of intervention

Perimeter (three sided) and isolation (four sided) pool fencing.

Types of outcome measures

1. Drowning (submersion with fatal outcome)
2. Near drowning (non-fatal submersion resulting in treatment in a hospital or emergency department).

Search strategy for identification of studies

See: Collaborative Review Group search strategy.

The overall goal of the search strategy was to be as inclusive as possible while keeping the number of irrelevant articles to a minimum. The first step was to use known review articles, augmented by reviews found in MEDLINE, to create a core reference bibliography. The second step was to search MEDLINE from 1966 to the present, using drowning or near-drowning as subject headings and adding interventions identified in the research question as key words. Results were compared to the core bibliography to see if key studies had been picked up. Next, other relevant databases were identified and searched (for example, EMBASE, ERIC, CINAHL), tailoring search terms to fit the requirements of each database. Organizations, agencies, and researchers knowledgeable in drowning prevention were contacted and their leads followed up. At all stages, references listed in studies selected for review were checked and new ones added to the project bibliography. Finally, the Cochrane Collaboration's "optimal search strategy" was run over both the MEDLINE and EMBASE databases to ensure that no indexed controlled study had been overlooked. We searched the following electronic databases: MEDLINE, ERIC, CINAHL, PSYClit, DIALOG which includes EMBASE, NTIS, Sport, BIOS, dissertation abstracts. Additionally we searched reference lists of past reviews and review articles, the Cochrane International Register of RCT's, studies from Consumer Products Safety Commission, National Spa and Pool Institute, National Safety Council, abstracts of Third International Injury Conference, Melbourne Australia, contacted colleagues from International Society for Child and Adolescent Injury Prevention, World Injury Network, and CDC funded Injury Control and Research Centers.

Methods of the review

The results of the search were screened by two independent reviewers and articles evaluating the effectiveness of pool fencing using a comparison group were selected for full review. The quality of these studies passing the screening criteria was then evaluated by these independent reviewers. In order to be included in this review, we required that studies have complete outcome/case ascertainment, accurate exposure measurement, appropriate selection of a comparison/control group and elimination or control of factors such as selection bias, observation bias, and confounding. For a cohort study ascertainment of exposure and outcome should be the same for all members of the cohort. Case control studies should have equal ascertainment of the exposure for case and control groups. Additionally, controls should be selected from the same population from which the cases were derived. Methodologically acceptable ecological studies require that ascertainment of exposure, outcome, and measurement of potential confounders be the same for all members of the population. Studies which did not meet methodological criteria were excluded from the review.

Description of studies

See: Tables of studies

Pitt, 1991

Study Design: Case control study, population based.
Intervention : Fencing around domestic pools
Population: All children under 14 years of age who were treated for an immersion injury at Mater Children's Hospital in S. Brisbane, Australia from 1984 to 1989 were potential cases (n=139). Death certificate records and autopsy files of the Institute of Forensic Pathology were reviewed to identify children who died at the scene. The population based control group was identified by a random sample telephone survey to identify households with swimming pools. Both case and control groups were interviewed in person to determine pool fencing characteristics.
Outcomes: Relative risk of drowning and near drowning in fenced compared to unfenced pools.
Results: Significantly decreased risk of drowning in fenced pool compared to unfenced pool (OR=0.27, 0.15-0.49). Similar, significantly decreased risks were found when stratified by pool type (in ground, above-ground, spa). 83% of all immersions occurred in children less than three years of age. All analyses were restricted to children who had unintended access to the pool. Characteristics of the fence (three or four sided) and pool gate were well defined.
Conclusions: The authors concluded that four sided pool fences were an effective method of preventing drowning and near drowning in children. Gates with child resistant latches and automatic closers are essential to limit access to pools.

Intergovermental, WA

Study Design: Case control study, population based.
Intervention: Pool fencing, comparing, isolation (four-sided) and perimeter (three-sided) fencing.
Population: Cases were preschool children aged 0-4 years who drowned in fenced private swimming pools in the metropolitan area of Perth, Australia, between 1975 and Feb. 1988. The comparison (control) group were fenced pools in households with children in the same age group. Controls were identified by a household survey in 1988.
Outcome: Estimated incidence rates for drowning for specific fencing types and attributable risk percent (AR%) for pool fencing.
Results: There was a significantly decreased risk of drowning in isolation-fenced pools compared to perimeter-fenced pools (OR=0.16, 95%CI (0.05-0.49). The attributable risk percent (AR%) calculation. indicated an 84% reduction in drownings attributed to pool fencing. Estimates of incidence rates were obtained by dividing the number of drowning cases by the estimated pool years at risk. The estimated incidence rate was 1.3 deaths per 10,000 pool years for pools with isolation fencing compared to 8.1 deaths per 10,000 pool years where perimeter fencing was used.
Conclusions: Revise the existing legislation to require isolation fencing around all existing pools and provide for annual inspections and educational campaigns.

Fergusson, 1984

Study Design: Case control study
Intervention: Pool fencing vs no fencing. Type of fencing and type of gate not specified.
Population: Cases were 60 pool drownings between 1973-1981 where the fencing status was known. The proportion of fenced pools in the Christchurch, New Zealand area was obtained from two surveys conducted by the Christchurch Child Development Study in 1980 and 1982.
Outcome: The risk of drowning in unfenced pools and the attributable risk percent (AR%) for pool fencing
Results: The protective effect of pool fencing was OR=0.29 95%CI (0.15,0.57) if 46% of pools in the community were fenced and OR=0.49 95%CI(0.24,0.97) if 34% of the pools were fenced. Attributable risk percent calculated using Bayes' Theorem ranged from 41% to 67% reduction in drowning attributed to pool fencing.
Conclusions: Universal pool fencing could prevent a substantial number of pool drownings.

Methodological quality

See: Table of included studies

Major problems encountered in all three of the studies revolve around selection of an appropriate control/comparison group and measuring the exposure to pools by children in various age groups. The Western Australia study (Intergovermental, WA) obtained data for the control/comparison group from a survey conducted in 1988, the end of the data collection period for the cases. It is not known if the distribution of the two fencing types (perimeter and isolation) for pools in households with young children changed over the 13 year period. Pitt, 1991 did not specify whether the identified control households contained children, nor did he document their ages. Fergusson, 1984 did not provide the age group distribution of the drowning cases or the control group. The control group for this study was appropriately obtained from a survey which identified the distribution of fenced and unfenced pools in the community; however, the proportion of children exposed to unfenced pools is unknown. Pitt, 1991 did limit his case group to children with unintended access to pools; however, there was not similar exposure information for the control group. One case control study, (Present, 1987) was excluded on methodological grounds-see Table of excluded studies.

None of the studies adjusted for the possible confounding influence of parental and child behavior. It is possible that parents who are more safety conscious would fence their pools and supervise their children more carefully.

Results

• List of comparisons
  

This review consists of case control studies since no randomized control or cohort studies were found. These three case control studies seem to indicate that pool fencing can significantly reduce the risk of drowning. Summary odds ratio for the risk of drowning and near drowning in a fenced pool compared to an unfenced pool is 0.27 95%CI (0.16, 0.47). Isolation fencing (enclosing pool on four sides) is superior to perimeter fencing (three sided fencing) because perimeter fencing allows access to the pool area through the house. Odds ratio for the risk of drowning in a pool with isolation fencing compared to a pool with three sided fencing is 0.17 95%CI(0.07, 0.44)

See Tables and Figures

Discussion

Case control studies are well suited to evaluate pool fencing as an intervention to prevent pool drowning in children. Drowning is a rare outcome which makes cohort or interrupted times series designs difficult. The only randomized controlled trial (RCT ) design which is feasible would be an RCT evaluating isolation vs. perimeter fencing; however, this would be logistically difficult to carry out.

Measuring exposure is difficult and requires collecting information on the number of pools in the population and the age distribution of the population exposed to pools. Pearn has proposed an objective method called the Swimming Pool Drowning Index (SPDI), to evaluate the effect of fencing legislation changes and educational initiatives. This formula adjusts for population distribution and home pool installation changes over time The SPDI is calculated as follows: (No. of pool fatalities/No. of children at risk) x (No. of private pools/No. of private dwellings) (IIntergovermental, WA). Due to the small number of drownings in a given year, the index is more stable if calculated over a multi-year period of time; cases should be grouped in 5 year segments to obtain stable rates for comparison. More complete data collection systems are needed to calculate a reliable SPDI. Since drownings are an infrequent occurrence the addition of near-drowning cases would make evaluation of interventions easier. However, in most areas reporting of near-drownings is incomplete. The direction of this ascertainment bias would depend on the distribution of the unreported near drowning cases between fenced and unfenced pools.

Reviewers' conclusions

Implications for practice

Isolation fencing with dynamic self latching gates is an effective environmental intervention which reduces unintended access to pools and reduces the risk of drowning for preschool children. Legislation accompanied by educational campaigns should be implemented for all public, semi- private and private swimming pools.

Implications for research

Additional case control studies are needed to provide a more precise estimate of the protective effect of fencing. The study design should use pools as the unit of analysis. Pools in which a young child drowns would be considered cases and other pools where no drowning occurred would be considered controls. Information would be collected on exposure to pools for children of various ages in the case and control groups. This would allow for controlling for the degree to which each pool is exposed to a young child in the home on the owner's property.

Studies examining fencing enforcement might allude to better fencing legislation. Specifically, a study comparing types of fencing legislation and their policies (fines, periodic inspections, etc.) would be important in determining the most effective and practical means of enforcement for a given community. Updating and maintenance of existing databases of drownings, near-drownings, number of private and public pools, fencing types, and regulations is the most important element in ascertaining the effectiveness of not only fencing type but also any other intervention.

Acknowledgements

Matthew Patterson, MPH
Ann Zavitkosky, MS
Chris Beahler, MLS
Linda Quan, MD

Potential conflict of interest

None known.

References

References to studies included in this review

Fergusson, 1984 (published data only)

Fergusson, 1994 Fergusson DM and Horwood LJ. Risks of drowning in fenced and unfenced domestic swimming pools. N Z Med J 1984;97(767):777-9.

Intergovermental, WA (published data only)

Intergov-WA, Intergovernmental Working Party on Swimming Pool Safety. Preschool drowning in private swimming pools. Health Department of Western Australia 1988.

Pitt, 1991 (published data only)

Pitt, 1991 Pitt, W R and Balanda, K P. Childhood drowning and near-drowning in Brisbane: the contribution of domestic pools. Med J Aust 1991;154(10):661-5.

* indicates the major publication for the study

References to studies excluded from this review

Milliner, 1980

Milliner, 1980 Milliner, N, Pearn, J, and Guard, R. Will fenced pools save lives? A 10-year study from Mulgrave Shire, Queensland. Med J Aust 1980;2(9):510-1.

Nixon, 1986

Nixon, J, Pearn, J, Wilkey, I, and Corcoran, A. Fifteen years of child drowning-a 1967-1981 analysis of all fatal cases from the Brisbane Drowning Study and an 11 year study of consecutive near-drowning cases. Accid Anal Prev;18(3):199-203.

Pearn, 1979

Pearn, J H, Wong, R Y, Brown, J 3d, Ching, Y C, Bart, R Jr, and Hammar, S. Drowning and near-drowning involving children: a five-year total population study from the City and County of Honolulu. Am J Public Health 1979;69(5):450-4.

Present, 1987

Present, P. Child drowning study: a report on the epidemiology of drownings in residential pools to children under age five. U.S. Consumer Product Safety Commission, Directorate for Epidemiology, Division of Hazard Analysis 1987 September.

Additional references

Choo 1995

Choo S, Teo WC, and Nixon J. Publicly expressed attitudes to pool fencing in letters to the editor in Brisbane newspapers. In: Nixon J, Pearn J, Oldenburg B, Pitt WR, editor(s). Review of countermeasures to reduce drowning, near drowning and spinal injuries from diving into shallow water. Commonwealth Department of Human Services and Health, Canberra ACT, 1995:Monograph series no 2, 69-71.

Fergusson 1983

Fergusson, DM, Horwood, LJ, Shannon, FT. The safety standards of domestic swimming pools 1980-1982. New Zealand Medical Journal, 1983;96, 93-95.

Geddis 1984

Geddis DC. The exposure of pre-school children to water hazards and the incidence of potential drowning accidents. NZ Med J. 1984;97:223-226.

Harris 1992

Harris, AH, Warchivker, I, de Clerk, NJ. Isolation fencing and the prevention of toddler drowning in backyard pools: an economic analysis. Economics Programme, Murdoch University 1992.

Nixon 1979

Nixon, JW, Pearn, JH, Petrie, GM. Childproof safety barriers. Australian Paediatric Journal, 1979;15(4).

Quan 1990

Quan L, Gomez A. Swimming pool safety: an effective submersion prevention. J Environ Health, 1990;52(6):344-346.

Rabinovich 1994

Rabinovich, BA, Lerner, ND, Huey, RW. Young children's ability to climb fences. Human Factors 1994;36(4).

Spyker 1985

Spyker DA. Submersion injury: epidemiology, prevention and management. Pediatr Clin North Am. 1985;32:113-125.

U.S. Mortality Data,

National Center for Health Statistics (NCHS), see CDC_WONDERAcdc3. cdc.

US-CPSC 1991

U. S. Consumer Product Safety Commission. CPSC staff recommendations for barriers for residential swimming pools, spas, and hot tubs. Washington D.C.

Wintemute 1990

Wintemute, GJ. Childhood drowning and near-drowning in the United States. American Journal of Disease in Childhood, 1990;144(6).

Wintemute 1990a

Wintemute GJ & Wright MA. Swimming pool owners’ opinions of strategies for prevention of drowning. Pediatrics. 1990;85(1).

Cover sheet

Sources of support

Extramural sources of support to the review

• Centers for Disease Control USA
  

Intramural sources of support to the review

• Harborview Injury Prevention & Research Center, University of Washington, USA
  

Synopsis

To be added.

Comments and criticisms

Relevance of the review to developing countries

Summary of comments and criticisms

Potential relevance of this review to developing countries.

I wonder if the authors have considered the relevance of protecting exposed water to safety in developing countries. Myaux et al. in Bull WHO 1997; 75:533-539 examine the effect of flood control embankment of total mortality in children 1-5 years. This included direct protection (death rates from accidental drowning) and indirect (death rates from infectious diseases).

I certify that I have no affiliations with or involvement in any organisation or entity with a direct financial interest in the subject matter of my criticisms

Reviewer's reply

A new review is currently underway, entitled 'Interventions for the prevention of drowning', which will include studies of measures to physically separate people from water and engineering modifications to aid exit from areas of water. The citation mentioned has been forwarded to the Reviewer who is preparing this review.

Contributors to comment

Paul Garner
Frances Bunn

Keywords

*Accident Prevention; Child, Preschool; Drowning/*prevention & control; Human; Swimming; *Swimming Pools;

Tables & Graphs

• MetaView graphs
  The figures and graphs in Cochrane Reviews display the Peto Odds Ratio and the Weighted Mean Difference by default. These are not always the methods used by reviewers when combining data in their review. You should check the text of the review for a description of the statistical methods used.
• List of comparisons
• Table of included studies
• Table of excluded studies
• Table of ongoing studies

List of comparisons

Fig 01 FENCING VS NO FENCING - ALL POOL TYPES

01.01.00 drowning & near drowning

Fig 02 FENCING VS. NO FENCING - IN-GROUND POOLS

02.01.00 Drowning and near-drowning

Fig 03 FENCING VS. NON-FENCING - ABOVE GROUND POOLS

03.01.00 Drowning and near-drowning

Fig 04 ISOLATION VS. PERIMETER FENCING

04.01.00 Drowning

Fig 05 FENCING VS NO FENCING-ALL POOL TYPES

05.01.00 drowning

Tables of other data

Tables of other data are not available for this review

Additional tables

Additional tables are not available for this review

Table of included studies

Table of excluded studies

Table of ongoing studies

A table of ongoing studies is not available for this review

The Cochrane Library