Abstract
Background
Deficiencies in emergency department (ED) charting is a common international problem. While unintentional falls account for the largest proportion of injury related ED visits by youth, insufficient charting details result in more than one third of these falls being coded as “unspecified”. Non‐specific coding compromises the utility of injury surveillance data.
Objective
To re‐examine the ED charts of unspecified youth falls to determine the possibility of assigning more specific codes.
Methods
400 ED charts for youth (aged 0–19 years) treated at four EDs in an urban Canadian health region between 1997 and 1999 and coded as “Other or unspecified fall” (ICD‐9 E888) were randomly selected. A structured chart review was completed and a blinded nosologist recoded the cause of injury using the extracted data. Differences in coding specificity were compared with the original data, and logistic regression was undertaken to examine variables that predicted assignment of a specific E‐code.
Results
A more specific code was assigned to 46% of cases initially coded as unspecified. Of these, 73% were recoded as “Slips, trips, and stumbles” (E885), which still lacks the specificity required for injury prevention planning; 2% of charts had no fall documented. Multivariate analysis revealed that dichotomized injury severity (adjusted odds ratio (OR) = 1.75 (95% confidence interval, 1.11 to 2.78)), arrival at the ED by ambulance (adjusted OR = 5.41 (1.07 to 27.0)), and the availability of nurse's notes or triage forms, or both, in the chart (adjusted OR = 3.75 (2.17 to 6.45)) were the strongest predictors of a more specific E‐code assignment.
Conclusions
Deficiencies in both chart documentation and coding specificity contribute to the use of non‐specific E‐codes. More comprehensive triage coding, improved chart documentation, and alternative methods of data collection in the acute care setting are required to improve ED injury surveillance initiatives.
Keywords: accidental falls, emergency department, pediatrics, chart review, coding specificity
Injury is the leading cause of morbidity and mortality in Canadian children.1 Among youth (aged 0 to 19 years), unintentional falls are the leading cause of injury resulting in a visit to the emergency department (ED).1,2,3 ED data can be used for fall surveillance purposes; however, many jurisdictions do not have a unified and electronic ED information system for accurate data capture. When available, these data are coded according to the International Classification of Diseases, version 9 (ICD‐9), which includes a supplemental classification system for external causes of injury (E‐codes).4 Complete and specific E‐codes provide details of injury intent, mechanism, and circumstances, which can be used for injury surveillance. Previous analyses of youth falls resulting in an ED visit in an urban Canadian health region revealed that approximately one third of these falls are coded as “Other or unspecified fall” (ICD‐9 E888).5
Investigators from numerous international settings have similarly noted the high prevalence of pediatric fall related injuries with unspecific E‐codes.6,7,8,9,10,11,12 Non‐specific external cause of injury (E‐) coding compromises the ability of injury surveillance to contribute to the development of targeted youth fall prevention strategies.
Others have found that factors related to the patient, the injury, and the ED visit (for example, day and time of presentation, severity of injury) are associated with more complete capture of injury information and coding specificity.13,14 Much less is known, however, about the factors that predict coding specificity of falls in the pediatric population. The identification of characteristics that determine coding specificity has important implications for improving injury surveillance and hence for prevention programming.
If a substantial proportion of unspecified falls can be assigned a more specific E‐code, this may constitute an important improvement in our understanding of the epidemiology of injuries caused by falls in the pediatric population. Moreover, the re‐prioritization of fall prevention strategies may be possible if such reclassification attempts were successful. Thus the purpose of our study was to investigate the coding specificity of falls coded as E‐code 888 (“Other and unspecified falls”) among youth (aged 0 to 19) seen in the ED. Using data from a retrospective chart review, the objectives of this study were first, to determine the proportion of youth falls initially coded as “unspecified” that become “specified” (that is, were given a more specific E‐code) after a chart review; and second, to identify the variables that predicted whether or not a fall initially coded as “unspecified” could be given a more specific E‐code.
Methods
Study setting
During the study period, the Capital Health Region served a population of 870 000 (1996 census data) in Edmonton and the surrounding areas in Alberta, Canada, and a large northern Alberta referral population. This region has seven acute care hospitals serviced by full time emergency physicians, and a dedicated children's hospital in which the emergency department serves as the regional pediatric injury referral center.
Design
This retrospective chart review examined a sample of data pertaining to unspecified falls among youth identified from the Ambulatory Care Classification System (ACCS) database for the health region. Between 1997 and 1999, young people aged 0 to 19 years made 20 317 visits to an ED for a fall related injury (ICD‐9 E880–E886.9, E888). Approximately 35% of these visits (n = 7007) were coded as “Other and unspecified fall” (ICD‐9 E888).
Sampling sites
The number of fall related injuries (both specified and unspecified) resulting in an ED visit varied by the seven EDs in the health region. Four of these seven EDs were purposely selected for involvement in the study. These four sites were selected because they are known to assess and treat the largest number of youth injuries (personal communication, Capital Health Information Services, 2003). Altogether, there were 15 686 fall related visits to these four EDs, and 32% (n = 5072) of these visits were for falls coded as “Other and unspecified”.
Sampling charts for review
Using a table of random numbers, a weighted random sample was obtained from the 5072 charts for youth (aged 0 to 19 years) who visited one of the four selected EDs with an unspecified fall related injury. Weighting was based on the proportion of pediatric ED visits for a fall from each of the four sites to generate a total sample size of 400. Misclassified, duplicate, or multiple visit charts were randomly replaced until the specified number of charts was obtained from each site.
Data collection
A trained ED research nurse reviewed each patient chart to abstract information available to assign an E‐code to the injury and to collect any available contextual information surrounding the incident. To maximize reliability and minimize bias, the same ED research nurse conducted the chart review at all four sites. This nurse had research and clinical experience and was familiar with the ED system, documentation, and physicians' and nurses' writing and shorthand.
A standard data collection form was used. Information collected corresponded to a minimally required dataset and included:
Patient characteristics (age and sex);
ED visit characteristics (ED site, date and time of ED visit, means of arrival at ED, discharge status);
Fall and injury characteristics (injury severity1, body region of injury, nature of injury, and a narrative description of the cause and circumstance of the injury);
Chart characteristics (documents available in the chart that are used for coding the injury, including the triage form, nurse's notes, and physician's notes).
Repeat coding
An experienced medical record nosologist subsequently reviewed each data collection form and assigned an appropriate E‐code. The nosologist was unaware of the study hypothesis and was blinded to the initial E‐code on the chart.
Data analysis
A trained data entry clerk entered the data from the data collection forms into SPSS version 11.0. Fifteen percent (n = 60) of the charts were double entered to check for data entry accuracy. The outcome variable was the post‐chart review E‐code, specified or unspecified. The percentage of charts that were assigned a more specific E‐code was computed, as was a list of the specific E‐codes. Univariate analyses were undertaken to assess the relation between the outcome variable and potential predictor variables. Categorical variables are reported as counts and proportions, while continuous variables are reported as mean (SD). Differences are reported using an unadjusted odds ratio (OR) with 95% confidence intervals (CI).
Multivariable logistic regression analysis was carried out to calculate adjusted odds ratios (and 95% CI) and to determine which variables best predicted whether an “unspecified fall” would go on to be recoded as “specified” after the chart review. Independent predictor variables considered in the analysis included factors related to the patients (age and sex), the ED visit (day, time, and season of visit, means of arrival at the ED, and ED site), the injury (injury severity, discharge status), and the availability of chart documents.
Ethical considerations
Ethics approval for conducting the chart review was obtained from the Health Research Ethics Board (Panel B), University of Alberta. Administrative/operational approval was obtained from each participating ED. Only the ED research nurse (that is, the chart reviewer) had access to the patient charts. The study investigators only had access to data stripped of any patient identifying information.
Results
Sample
In all, there were seven missing charts, all of which were replaced by randomly sampled charts from the database; the overall sample was 400 charts.
Baseline characteristics
The mean age of patients was approximately 9 years and 58% were male (table 1). ED visits were roughly evenly distributed throughout the days of the week, with 29% occurring on the weekend. Visits were also evenly distributed by season. Most patients (57%) visited the ED between the hours of 4:00 pm and midnight, followed by 8:00 am to 3:59 pm (37%). The great majority of patients (97%) arrived at the ED by themselves or with their parents. The remaining 3% arrived by ambulance. Most of the patients (95%) were discharged after their ED visit, and only 2% were admitted to hospital (3% were missing discharge data). Thirty six percent of patients had an injury classified as severe.
Table 1 Baseline characteristics of sampled charts (n = 400).
| Variable | Frequency | Percent |
|---|---|---|
| (n = 400) | ||
| Demographics | ||
| Age at injury (years) (mean (SD)) | 8.91 (5.69) | |
| Sex | ||
| Male | 232 | 58.0% |
| Female | 167 | 41.8% |
| Missing | 1 | 0.25% |
| Characteristics of the ED visit | ||
| ED site | ||
| A | 68 | 17.0% |
| B | 67 | 16.8% |
| C | 152 | 38.0% |
| D | 113 | 28.3% |
| Day of the week of ED visit | ||
| Weekend (Sat–Sun) | 117 | 29.3% |
| Weekday (Mon–Fri) | 283 | 70.8% |
| Season of ED visit | ||
| Summer (June, July, August) | 91 | 22.8% |
| Autumn (September, October, November) | 112 | 28.0% |
| Winter (December, January, February) | 94 | 23.5% |
| Spring (March, April, May) | 103 | 25.8% |
| Time of ED visit | ||
| Midnight to 07:59 | 26 | 6.5% |
| 08:00 to 15:59 | 148 | 37.0% |
| 16:00 to 23:59 | 226 | 56.5% |
| Means of arrival at ED | ||
| Ambulance | 12 | 3.0% |
| Self/parent | 388 | 97.0% |
| Discharge status | ||
| Admitted to hospital | 9 | 2.3% |
| Treated and discharged | 381 | 95.3% |
| Missing | 10 | 2.5% |
| Severity of injury* | ||
| Severe | 142 | 35.5% |
| Not severe | 258 | 64.5% |
| ED charting | ||
| Triage form in chart | ||
| Yes | 101 | 25.3% |
| No | 299 | 74.8% |
| Nurse notes in chart | ||
| Yes | 221 | 55.3% |
| No | 179 | 44.8% |
| Physician notes in chart | ||
| Yes | 389 | 97.3% |
| No | 11 | 2.8% |
| Summary of chart documentation | ||
| Physician notes and nurse notes or triage | 262 | 65.5% |
| form | ||
| Only physician notes | 132 | 33.0% |
| Missing/incomplete data | 6 | 1.5% |
*Severe injuries were defined as an injury with any of the following characteristics: concussion or other head injury (including lacerations to the head); fracture; neck sprain/strain; requiring admission to hospital or transfer to a different hospital.
Totals may add up to more than 100% owing to rounding.
ED, emergency department.
A triage form was present in only 25% of charts; however, this varied by ED from a low of 3% to a high of 81%. Only 55% of charts contained ED nurse notes, while 97% of charts had notes from the ED physician. A summary measure of chart documentation showed that 66% of the charts had physician notes and nurse notes or a triage form, while 33% of charts only had physician notes.
Analysis of coding specificity
Overall, 46% of the charts contained sufficient documentation to allow reassignment of a more specific E‐code (table 2). Nearly three quarters (73%) of these specified charts were recoded as “Slips, trips, and stumbles” (E855). It was possible to identify a more specific type of fall for 43% of the charts, while another 3% of charts reported a fall but as a result of a strike or collision. For injury surveillance purposes, these injuries are typically not counted in falls.
Table 2 E‐codes determined from chart review of “other and unspecified falls”(E888).
| Category | E‐code | Description | n | Percent | |||
|---|---|---|---|---|---|---|---|
| Injuries with specified E‐codes | 184 | 46.4% | |||||
| Specified falls | E884.0 | Fall from playground equipment | 5 | 1.3% | |||
| E884.4 | Fall from bed | 1 | 0.3% | ||||
| E884.5 | Fall from other furniture | 3 | 0.8% | ||||
| E884.9 | Fall from one level to another | 8 | 2.0% | ||||
| E885 | Slips, trips, and stumbles | 134 | 33.5% | ||||
| E886.0 | Falls in sports | 13 | 3.3% | ||||
| E886.9 | Other and unspecified falls on the same level from collision, pushing, or shoving by or with other person | 8 | 2.0% | ||||
| Strike or collision followed by a fall | E826.0 | Pedestrian fall when struck by a pedal cyclist | 1 | 0.3% | |||
| E826.1 | Pedal cyclist injured from fall | 2 | 0.5% | ||||
| E848 | Accident involving toboggan, collision, fell off | 1 | 0.3% | ||||
| E917.9 | Striking against or struck accidentally by objects or persons | 7 | 1.8% | ||||
| E966 | Assault (followed by a fall) | 1 | 0.3% | ||||
| No fall documented on chart (false positive rate) | 7 | 1.8% | |||||
| Other or unspecified falls | E888 | Other or unspecified falls | 209 | 52.3% | |||
| Total | 400 | 100.0% | |||||
Two percent of charts had no fall documented (that is, they were false positives). The remaining 52% of charts remained coded as “Other or unspecified fall”.
Univariate analysis
For the purposes of the univariate (and multivariate) analysis, the outcome variable was dichotomized into specified E‐code v unspecified E‐code. The “specified” group included those injuries resulting from a “specified fall” and from a “strike or collision followed by a fall”. False positive cases (n = 7) were excluded from this analysis. Thus, of the remaining 393 charts, 47% had specified E‐codes and 53% had unspecified E‐codes.
Table 3 provides the results of the univariate analysis comparing charts assigned a more specific E‐code after the review with charts whose codes remained unspecified. Severity of injury factors (that is, means of arrival, discharge status, and injury severity) were significantly associated with assignment of a specific E‐code. More patients arriving by ambulance (OR = 5.95 (95% CI, 1.29 to 27.8)), admitted to hospital (OR = 9.37 (1.16 to 76.9), and presenting with severe injuries (OR = 1.63 (1.08 to 2.47)) were assigned a specific E‐code. Charts with more comprehensive documentation (for example, triage forms or nurse's notes in addition to physician notes) were also more commonly assigned a specific E‐code (OR = 2.56 (1.65 to 3.97)). Finally, charts from ED site “D” had decreased odds of being assigned a specific E‐code compared with the reference site (OR = 0.48 (0.26 to 0.89)).
Table 3 Univariate and multivariate analysis of predictor variables of E‐code specificity on chart review (n = 393).
| E‐code determined from chart review | Unadjusted OR (95% CI) | Adjusted OR (95% CI) | ||
|---|---|---|---|---|
| Specified E‐code (n = 184) | E‐code remained unspecified (n = 209) | |||
| Demographics | ||||
| Age (years) (mean (SD)) | 9.31 (5.80) | 8.59 (5.59) | 1.02 (0.99 to 1.06) | 1.03 (0.99 to 1.08) |
| Sex | 1.04 (0.70 to 1.56) | 1.02 (0.67 to 1.57) | ||
| Male | 59% | 58% | ||
| Female | 41% | 42% | ||
| Characteristics of the ED visit | ||||
| ED site | ||||
| A | 21% | 14% | 1.00 | 1.00 |
| B | 17% | 16% | 0.74 (0.37 to 1.47) | 1.17 (0.55 to 2.47) |
| C | 39% | 38% | 0.69 (0.38 to 1.23) | 1.98 (0.96 to 4.08) |
| D | 23% | 33% | 0.48 (0.26 to 0.89) | 0.93 (0.47 to 1.86) |
| Day of visit | 1.32 (0.85 to 2.04) | 1.28 (0.80 to 2.03) | ||
| Weekend | 32% | 26% | ||
| Weekday | 68% | 74% | ||
| Season of ED visit | ||||
| Summer | 26% | 20% | 1.00 | 1.00 |
| Autumn | 28% | 28% | 0.77 (0.44 to 1.35) | 0.82 (0.45 to 1.49) |
| Winter | 22% | 25% | 0.69 (0.38 to 1.24) | 0.74 (0.40 to 1.40) |
| Spring | 25% | 28% | 0.68 (0.38 to 1.20) | 0.81 (0.44 to 1.48) |
| Time of ED visit | ||||
| Midnight to 07:59 | 7% | 6% | 1.00 | 1.00 |
| 08:00 to 15:59 | 36% | 37% | 0.85 (0.36 to 2.01) | 1.00 (0.38 to 2.58) |
| 16:00 to 23:59 | 58% | 57% | 0.89 (0.38 to 2.07) | 1.17 (0.46 to 2.95) |
| Means of arrival | 5.95 (1.29 to 27.8) | 5.41 (1.07 to 27.0) | ||
| Ambulance | 5% | 1% | ||
| Self/parent | 95% | 99% | ||
| Discharge status* | 9.37 (1.16 to 76.9) | N/A‡ | ||
| Admitted to hospital | 4% | 0.5% | ||
| Treated and discharged | 96% | 99.5% | ||
| Severity of injury† | 1.63 (1.08 to 2.47) | 1.75 (1.11 to 2.78) | ||
| Severe | 42% | 31% | ||
| Not severe | 58% | 69% | ||
| Characteristics of the chart | ||||
| Documentation in chart | 2.56 (1.65 to 3.97) | 3.75 (2.17 to 6.45) | ||
| ED physician notes and nurse's notes or triage form | 77% | 57% | ||
| Only ED physician notes | 23% | 43% | ||
*Includes only those charts with discharge status information (n = 389).
†Severe injuries were defined as an injury with any of the following characteristics: concussion or other head injury (including lacerations to the head); fracture; neck sprain/strain; requiring admission to hospital or transfer to a different hospital.
‡Discharge status was not included in the multivariable logistic regression model because of high colinearity with severity of injury and because of the relatively small number of hospital admissions.
Totals may add up to more than 100% owing to rounding.
CI, confidence interval; ED, emergency department; OR, odds ratio.
Multivariable analysis
Before proceeding to the multivariable logistic regression analysis, independent variables were checked for colinearity. It was suspected a priori that discharge status and injury severity would be highly associated as admission to hospital was one criterion used for determining severity of injury. Therefore only injury severity was entered into the model.
To yield adjusted odds ratios, all other nine independent variables were entered into the model, despite their significance in the univariate analysis. This technique is consistent with a commonly cited 10% rule of thumb for the number of independent variables to include in a model.15 The adjusted odds ratios and 95% confidence intervals are included in table 3. As shown, patients presenting with severe injuries or who arrived at the ED by ambulance, and charts with more comprehensive documentation, remained the most significant predictors of being assigned a specific E‐code.
Discussion
This study, conducted in an urban Canadian health region, examined 400 charts randomly selected from youth ED visits for a fall that was coded as unspecified. Overall, while thorough evaluation of the chart improved the coding specificity, 52% remained unspecified after the chart review. Consistent with the results of the logistic regression, this highlights the lack of documentation of etiological information associated with injuries seen in EDs. This lack of documentation contributes to poor coding specificity of acute injuries in this health region, and presumably others. Moreover, as the “unspecified falls” categorization is commonly employed in the ED, this coding issue represents an important barrier for injury prevention practitioners in Canada and other countries.6,7,8,9,10,11,12
The vast majority of youth classified as having fallen did in fact fall; only 2% of reviewed charts had no evidence of a fall. It is notable that 46% of the remaining charts were assigned a more specific E‐code from the chart review. As evidenced from the logistic regression, severe injuries and charts with more documentation were more likely to be assigned a specific E‐code during the chart review. These findings suggest that busy nosologists also make errors when the etiological causes require extensive searching through non‐standardized charts. Consistent with the findings of others,6 the tendency for nosologists to overuse non‐specific E‐codes for pediatric falls appears largely to reflect deficiencies in chart documentation. While one possible solution is allocating more nosologist time per chart, the economic consequences of such an effort are probably unjustified. Furthermore, the vast majority (73%) of falls that were specified upon the chart review were coded to E885 (Slips, trips, and stumbles), a category that still lacks the specificity required for injury prevention planning.
There are numerous possible reasons for the lack of chart documentation and poor diagnostic specificity. First, all charts in this health region remain paper based. As such, chart legibility varies and even if the information is documented, coders and staff may be unable to identify the pertinent injury information. Second, these paper charts do not contain any prompting for specific information that would be crucial for improved coding specificity and a better understanding of the epidemiology of childhood injury. Without specific prompting using standardized template charts, it seems unreasonable to expect front line staff to report a set of mandatory injury specific variables. Previous research from this region showed a modest increase in charting pediatric injury information with education16; however, this increase may have been short lived. Finally, the use of known preventive injury strategies is also not collected routinely or in a standardized fashion. For example, prompted information on helmet use on the ED chart may increase coding specificity by helping nosologists distinguish sports related falls from general falls.
Relying on emergency physicians for precise and specific injury information seems fraught with non‐compliance, illegibility, and lack of standardization.17 Potential solutions for this problem are many and varied. The first opportunity to collect the necessary injury information exists at triage desks. Electronic prompting in triage, using one of the electronic triage programs,18 seems an ideal place to start. A second opportunity exists with electronic or paper based template charts, specifically designed for injury. Another option involves having physicians or triage nurses assign E‐codes, rather than or in addition to the nosologist.19,20 While this has been shown to be successful in certain locations, evaluations in other settings are still required before widespread application can occur. Finally, educational interventions seem less likely to work than mandated options.12 Educating the multitude of learners and staff in the ED—especially in the setting of ED overcrowding—seems an inefficient method to change behaviors. Until such time as some of these proposed interventions are introduced, improved coding of injury charts may not be possible.
Limitations
There are several limitations that require us to interpret these findings cautiously. First, the study was conducted in only one health region. Owing to potential differences in pediatric injury patients and EDs across regions, the generalizability of the results remains uncertain. It is known, however, that other regions similarly rely upon ED paper charts for pediatric fall injury surveillance. These sites would face the same challenges discussed here, including limited chart documentation, insufficient details for injury coding, and a need for real alternatives for injury surveillance data. Thus the study findings are likely to be highly relevant to those settings. Second, our study was conducted using years for which the ICD‐9 coding system was used. We recognize the ICD‐10 coding system has the potential for providing much more specific categories to code falls; however, the value of this enhanced coding system will be lost if the necessary injury specific information about the fall is not recorded in the first place. If information is not on the chart, the availability of highly specific codes becomes irrelevant. Finally, coding misclassification may have occurred; however, this bias was minimized by blinding the experienced nosologist to the initial E‐codes on the chart and by checking the reliability and consistency of coding throughout the data collection process.
Key points
Injury is the leading cause of death and disability in children.
Among youth aged 0 to 19 years, unintentional falls account for the largest proportion of injury related emergency department visits; however, one third of these falls are coded as “unspecified”.
Non‐specific external cause of injury (E‐) coding compromises the utility of injury surveillance to assist in the development of targeted youth injury prevention strategies.
Detailed chart review decreased the proportion of unspecified falls by 46%; however, 73% of these specified falls were recoded as “Slips, trips, stumbles” (E885), which still lacks the specificity required for prevention planning.
Deficiencies in both chart documentation and coding specificity contribute to the use of non‐specific E‐codes.
In order to further understand and prevent injuries, more comprehensive triage coding, improved chart documentation, and alternative methods of data collection in the acute care setting are required.
Conclusions
In a typical Canadian paper based emergency department without templated charts, detailed review of pediatric medical charts reduced the number of unspecified falls by 46%. Despite this apparent success, the resources required to reduce unspecified coding of pediatric falls on a broader scale are probably unjustified. Moreover, the improved recoding was limited to a general fall related injury code and would not provide the granular detail required by injury planners. If we are to have a better understanding of the injury pyramid21 and the causes of acute injury, improved chart documentation and alternate methods of data collection are required.
Acknowledgements
We would like to thank the Population Health Priorities program of the Population Health and Research Department at Capital Health and the Department of Emergency Medicine, University of Alberta for funding this research project. We would also like to thank the medical records staff at each participating hospital site for their contribution to this work. In addition, the assistance of Ms Kim Eden, University of Alberta Department of Emergency Medicine research nurse, and Anne Wiersma‐Olson, University of Alberta Hospital nosologist, was greatly appreciated.
BHR is supported by a Canada Research Chair from the Canadian Institutes of Health Research (CIHR). BH holds the recently appointed position of Professorship in Child Health and Wellness funded by the Alberta Children's Hospital Foundation, through the generous support of an anonymous donor and the Canadian National Railway Company.
Footnotes
1Severe injuries were defined as an injury with any of the following characteristics: concussion or other head injury (including lacerations to the head); fracture; neck sprain/strain; requiring admission to hospital or transfer to a different hospital.
References
- 1.Alberta Centre for Injury Control and Research (ACICR) Injury control facts for Canada and Alberta [monograph on the internet]. Edmonton, AB: ACICR; 2004 [cited 2005 Mar 9], Available from: http://www.med.ualberta.ca/acicr/pages/documents/InjuryControlFactsCD‐9‐04.pdf
- 2.Pless B, Millar W.Unintentional injuries in childhood: Results from Canadian Health Surveys [monograph on the internet]. Ottawa, ON: Health Canada; 2000 [cited 2005 Nov 21], Available from: http://www.phac‐aspc.gc.ca/dca‐dea/publications/pdf/unintentional_e.pdf
- 3.Predy G N, Lightfoot P, Edwards J.et alHow healthy are we? Health status in the Capital Health region – a technical report, 2004. Edmonton: Capital Health, 2005. (Canada)
- 4.World Health Organisation International classification of diseases 9th revision(ICD‐9): Clinical Modification (E‐code) section. Los Angeles: Practice Management Information Corporation, 1997
- 5.Kidsafe Connection, Stollery Children's Hospital, and Department of Population Health and Research Prevention of falls and fall related injury among youth in the Capital Health region: a review of the literature and local hospitalization and emergency department data. Edmonton: Capital Health, 2002. (Canada)
- 6.Langlois J A, Buechner J S, O'Connor E A.et al Improving the E‐coding of hospitalizations for injury: do hospital records contain adequate documentation? Am J Public Health 1995851261–1265. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.LeMier M, Cummings P, West T A. Accuracy of external cause of injury codes reported in Washington State hospital discharge records. Inj Prev 20017334–338. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Langley J, Stepheson S, Thorpe C.et al Accuracy of injury coding under ICD‐9 for New Zealand public hospital discharges. Inj Prev 20061258–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Agran P F, Anderson C, Winn D.et al Rates of pediatric injuries by 3‐month intervals for children 0 to 3 years of age. Pediatrics 200311683–692. [DOI] [PubMed] [Google Scholar]
- 10.Rivara F P, Alexander B, Johnston B.et al Population‐based study of fall injuries in children and adolescents resulting in hospitalization or death. Pediatrics 19939261–63. [PubMed] [Google Scholar]
- 11.Bawa H, Brussoni M, De Gagne D.et alFalls‐related injury data report 2001–2003. [monograph on the internet]. Vancouver, British Columbia: BC Injury Research and Prevention Unit; 2004 [cited 2006 Mar 6], Available from: http://www.injuryresearch.bc.ca/publications/reports/falls%20report%202004.pdf
- 12.Steenkamp M, Cripps R.Child injuries due to falls [monograph on the internet]. Adelaide, South Australia: Injury Research and Statistics Series (AIHW cat no, INJCAT 37);2001 [cited 2006 Mar 6]. Available from: http://www.nisu.flinders.edu.au/pubs/reports/2001/falls_injcat37.pdf
- 13.Adirim T A, Wright J L, Lee E.et al Injury surveillance in a pediatric emergency department. Am J Emerg Med 199917499–503. [DOI] [PubMed] [Google Scholar]
- 14.Macarthur C, Pless I B. Evaluation of the quality of an injury surveillance system. Am J Epidemiol 1999149586–592. [DOI] [PubMed] [Google Scholar]
- 15.Harrell F E, Lee K L, Mark D B. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 199615361–387. [DOI] [PubMed] [Google Scholar]
- 16.Voaklander D C, Cummings G, Policicchio C.et al A pilot intervention to improve the documentation of pediatric injuries. Can J Emerg Med 20002252–257. [DOI] [PubMed] [Google Scholar]
- 17.Cummings G E, Voaklander D C, Vincenten J.et al Emergency staff survey on their role in pediatric injury prevention education – a pilot study. J Emerg Med 200018299–303. [DOI] [PubMed] [Google Scholar]
- 18.Dong S L, Bullard M J, Meurer D P.et al Emergency triage: Comparing a novel computer triage program with standard triage. Acad Emerg Med 200512502–507. [DOI] [PubMed] [Google Scholar]
- 19.Bota G W, Therrien S, Rowe B H. A truncated E‐code system for injury surveillance in the emergency department: Description and clinimetric testing. Acad Emerg Med 19974291–296. [DOI] [PubMed] [Google Scholar]
- 20.Ribbeck B M, Runge J W, Thomason M H.et al Injury surveillance: a method for recording E codes for injured emergency department patients. Ann Emerg Med 19922137–40. [DOI] [PubMed] [Google Scholar]
- 21.Sahai V S, Ward M S, Zmijowskyj T.et al Quantifying the iceberg effect for injury: using comprehensive community health data. Can J Public Health 200596328–332. [DOI] [PMC free article] [PubMed] [Google Scholar]