Halloween: Is it the most dangerous day for kids?
Analysis of the 56 million ePCR records in the NEMSIS national dataset shows days with the most traumatic pediatric cardiac arrests in 2013 and 2014
On Halloween, in most U.S. communities, children spill into their neighborhoods, running from house to house, to collect candy from their neighbors.
Many injury prevention campaigns, rightfully, remind children and parents of the dangers of being unseen to motor vehicle drivers and the potential for children to be killed or injured on Halloween. Nearly 20 years ago the CDC examined the Fatal Analysis Reporting System (FARS) data to look at childhood pedestrian deaths during Halloween.
In the study period, 1975 to 1996, "from 4 p.m. through 10 p.m. on October 31, a total of 89 deaths occurred among pedestrians aged 5-14 years, compared with 8,846 on all other evenings. Overall, among children aged 5-14 years, an average of four deaths occurred on Halloween during these hours each year, compared with an average of one death during these hours on every other day of the year."
A widely-circulated SafeKids.org infographic states, "Twice as many kids are killed while walking on Halloween than any other day of the year." A reference is not cited.
But is it true?
EMS and other health care providers have anecdotal notions about when injuries and fatalities are more likely to occur. For example, a full moon might be cause for dread in emergency department staff.
National ePCR Safety data
In many instances, a child struck by a vehicle will lead to an EMS activation, ambulance response and creation of an electronic patient care report (ePCR). Tens of millions of U.S. ePCR records are available in the National EMS Information System (NEMSIS).
"Any EMS field provider can ask a question and investigate more than 56 million ePCR records with just a few clicks," said Nick Nudell, chief data officer of the Paramedic Foundation.
The NEMSIS data set spans all of 2013, 2014 and 2015 year-to-date submitted ePCR records.
"The first step is to pose the question," said Nudell. "For example, we want to see if the NEMSIS data supports the previous finding that pediatric fatalities are more common on Halloween."
Once the question is clarified any EMS field provider, educator or leader can access the NEMSIS data through the NEMSIS Cube, an online tool for selecting data elements, filtering those elements, and reporting the data.
"There are a lot of records available for each patient age group, based on the age inputted into the ePCR by the EMS provider," said Nudell. "Unfortunately 7 million ePCR records do not have an age recorded and can't be included."
The remaining 49 million records were filtered for traumatic cardiac arrests. The NEMSIS data contains 24,145 traumatic cardiac arrest records, with a patient age recorded.
The next step is to filter for traumatic cardiac arrest patients under the age of 20. The result was 2,409 records, which is only 0.005 percent of all calls with a recorded age.
An additional filter was applied to evaluate if Halloween is more deadly for kids than any other day of the week.
To look at the trends by day of the week and to evaluate whether Halloween is really as deadly as some have said, the data was limited to 2013 and 2014. The number of traumatic arrests, by age group, was fairly similar in each year.
The data was exported from the NEMSIS EMS Data Cube and imported into Microsoft Excel for analysis. The data, plotted into a chart, shows the pediatric traumatic cardiac arrests for both 2013 and 2014 (y-axis) on each day of the year (x-axis).
"Of course there are other causes of pediatric traumatic cardiac arrest — falls, gunshots, football collisions, but we want to see if there is a spike or above normal number of deaths on October 31," Nudell said.
Explore the NEMSIS data
The chart or plot of deaths over time has five lines. The up and down lines connect the daily deaths by traumatic arrest. As expected, some days are high and others are low. The lowest green line is the average number of pediatric traumatic cardiac arrests deaths per day.
The two red lines, above the green line, are more important and help identify days with an out of the ordinary or unexpected number of deaths.
"The red lines indicated the standard deviation around the mean. Roughly two-thirds of all days are within the 2nd standard deviation," Nudell said. "Any day that is above the other red line, which is the 3rd standard deviation, is significant."
Any day less than the lower red line is within what is expected as normal. Days above the lower red line are worth considering as special days. "Halloween does not stand out, in fact, it has fewer cases (traumatic cardiac arrest) than most other days," Nudell said.
Significant days According to NEMSIS
Several days clearly stand above or outside the 3rd standard deviation.
"The next step for a medic investigating this question is to ask, 'What is special about April 25, 2014 or May 18, 2013; or June 13 and 15, 2014; or Aug. 10, 2014; or Aug. 30, 2013?' Those days clearly stand out," Nudell said.
Remember the NEMSIS data is a national collection of 56-plus million ePCR records. Any analysis might lead to more questions than answers. Some of those questions might be:
- What is significant about the dates we discovered?
- Were there any MCIs that would have contributed to these numbers?
- Are the right filters being applied?
- How would chart shift if only kids age 0-10 we considered?
- How would the chart shift we filtered for car accidents only?
What are your NEMSIS data queries?
Do you have a question you want to explore with the NEMSIS data? Share your question in the comments and what you found with the data. Consider submitting the results of your exploration to firstname.lastname@example.org for an upcoming article.
This article was originally posted on Oct. 30. 2015. It has been updated.