CARIS Research Update: What We’re Learning About Contact Obstacle Incidents
- Arielle Pechette Markley, DVM, DACVSMR
- 13 hours ago
- 6 min read
Written by:
Dr. Arielle Pechette Markley, Director of Research, Red Sage Integrative Veterinary Partners
Dr. Abigail Shoben, Associate Professor, Biostatistics, The Ohio State University College of Public Health
Recent discussions have renewed focus on the safety of contact obstacles in agility, particularly the dogwalk. These conversations are important, but meaningful progress depends on moving beyond individual events toward objective, population-level data.
As part of the Coalition for Agility Reporting on Incidents and Safety (CARIS) initiative, we conducted a prospective, multi-organization data collection effort designed to better understand how often incidents occur and what they look like across competition.
This post provides an executive summary of early findings from that work.
Study Overview
Between April and November 2025, we collected incident data from agility trials across multiple organizations using:
End-of-trial reporting (total run counts)
Real-time incident reporting by judges
An incident was defined as an unexpected exit from a contact obstacle (dog walk, A-frame, or teeter) occurring during performance. Unexpected exits included events such as jumping or falling from the obstacle prior to completion, exiting from a height above the contact zone consistent with a fall, or marked loss of control (e.g., stumbling or faceplanting) on exit. Controlled performance errors, such as missed contact zones without loss of balance or control, were not classified as incidents.
After data cleaning and matching:
1,385 trials contributed usable run data
1,193 incident reports were analyzed
Data spanned observations from more than 300,000 observed runs on each obstacle (320K dogwalk, 340K teeter, 418K Aframe).
To our knowledge, this represents one of the largest coordinated prospective datasets examining contact obstacle incidents in agility.
Key Finding #1: Incident Rates Are Low, but Not Equal Across Obstacles
When standardized to runs, incident rates differed by obstacle:
Dogwalk: 2.10 incidents per 1,000 runs
Teeter: 1.04 per 1,000 runs
A-frame: 0.40 per 1,000 runs
This means:
Roughly 1 incident every ~475 dogwalk performances*
Roughly 1 incident every ~960 teeter performances
Roughly 1 incident every ~2,500 A-frame performances
*technically this is per run involving each obstacle (e.g., 1 incident per 475 runs involving a dogwalk). We do not know for sure that the obstacle was attempted once and only once on each reported run involving the obstacle.
The relative ordering (dogwalk > teeter > A-frame) was consistent across organizations.
For some perspective on 1000 runs, 1000 runs is equivalent to:
One dog competing 25 weekends a year with 4 runs involving the contact obstacles per weekend over 10 years = 1000 runs for that dog in competition
Attending five trial days with 200 runs involving contact obstacles = observing 1000 runs
Attending 20 trial days with 50 runs involving contact obstacles = observing 1000 runs
Additional thoughts about risk perception, particularly on the 1 in 1000 scale, are available here.
Key Finding #2: Most Incidents Do Not Result in Apparent Injury
The vast majority of reported incidents were minor at the time of occurrence:
91.5%: No apparent injury; dog continued the run
7.6%: No visible injury, but handler elected to stop the run
<1%: Apparent injury (e.g., limping, bleeding)
0 cases required immediate veterinary transport
When standardized to total runs, the observed rate of apparent injury was extremely low:
Dogwalk: ~0.03 injuries per 1,000 runs (equivalent to 1 in 33,300 runs)
Teeter: 0 per 1,000 runs (no observed cases)
A-frame: ~0.002 per 1,000 runs (equivalent to 1 in 500,000 runs)
These values reflect injuries identified at the time of the incident only and should be interpreted with caution. The study did not include follow-up, so delayed or subclinical injuries are not captured. Additionally, estimates are sensitive to small numbers, particularly for the A-frame and teeter.
Key Finding #3: Patterns Are Consistent Across Organizations
Despite differences in rules, course design, and populations:
The relative risk by obstacle was consistent
Incident reporting rates aligned closely between submitted and received data
This supports the reliability and generalizability of the dataset
This consistency strengthens confidence that these findings reflect true underlying patterns, not isolated or organization-specific effects.
What Do These Incidents Actually Look Like?
Across obstacles, most incidents were not contact-zone errors.
Instead, they were characterized by:
Dogwalk:
Exiting from the middle plank or during ascent (>70%)
A-frame:
Exiting high above the contact zone or on ascent (~70%)
Teeter:
Leaving the plank before or around the tip (~85%)
How Do These Findings Compare to Previous Research?
Interpreting agility safety data requires careful attention to what is being measured (incidents vs. injuries) and how the data are collected (prospective vs. retrospective). These differences strongly influence reported rates and perceived risk.
Most prior studies have relied on retrospective handler surveys, which:
Primarily capture injuries, not all incidents
Are subject to recall bias
May overrepresent more severe or memorable events
Comparison to Dogwalk-Specific Data
A recent retrospective study of Finnish agility dogs (Inkilä et. al., 2025) reported:
2.6 falls per 1,000 dogwalk performances
0.6 injuries per 1,000 dogwalk performances
Approximately 5% of incidents resulting in injury
At first glance, these injury rates appear higher than those observed in the current dataset. However, several key differences explain this:
Prospective vs. retrospective design: Our study captures incidents in real time, while retrospective surveys rely on memory and may preferentially capture more severe events.
Definition of events: Our dataset includes a broad range of unexpected incidents, many of which are minor. The Finnish study focuses more specifically on falls and injuries.
Injury assessment: Our data reflect apparent injury at the time of the event only. Retrospective studies may include delayed or subsequently diagnosed injuries.
Despite these differences, there is important alignment:
The observed dogwalk incident rate (~2.1 per 1,000 runs) is similar in magnitude to previously reported fall rates
Both datasets indicate that injury occurs in a minority of events, though the proportion varies by methodology
Comparison to Early Judge-Reported Data (UKI)
Preliminary judge-reported data collected by UKI found:
~3.2 incidents per 1,000 runs
No reported injuries during the collection period
This is slightly higher than the rate observed in the current dataset but remains within a similar range.
Differences to consider:
UKI data included only standard runs, not games classes
Incident definitions were broad, including voluntary exits
Injury was not a primary outcome
Notably, both datasets demonstrate:
Low overall incident rates on a per-run basis
Rare immediate injury
Similar patterns in the types of events being captured
Comparison to Other Obstacles
Tunnel-related incident rates have been reported by Ford, et. al. at approximately:
~15 incidents per 1,000 tunnel performances
This is substantially higher than contact obstacle incident rates observed in this study and would likely be even higher on a per run basis. Even averaging just two tunnels per run, this rate would be approximately 30 tunnel incidents per 1000 runs. In contrast, the rates observed in this study were:
Dogwalk: ~2.1 per 1,000 runs
Teeter: ~1.0 per 1,000 runs
A-frame: ~0.4 per 1,000 runs
This contrast highlights an important distinction:
Some obstacles may have higher frequency of incidents
Others may carry greater perceived consequence
Frequency of events and severity of concern are not always aligned.
Take-Home Context
Across datasets:
Dogwalk incident/fall rates are consistently in the range of ~2–3 per 1,000 performances
Injury rates are substantially lower than incident rates
Reported rates are highly dependent on study design and definitions
These findings reinforce the importance of:
Prospective, standardized data collection
Clear distinction between incidents and injuries
Continued focus on understanding the mechanisms underlying these events
Connecting Back to Dogwalk Safety
In our previous post👉 The Dog Walk in Agility – Part 2
we discussed how:
The dogwalk combines speed, elevation, and a narrow surface
Small changes in stride or timing can have large consequences
These new data support that framework:
The dogwalk has the highest incident rate
Most events involve early exits or loss of balance
However, serious injury appears uncommon at the time of the event
This reinforces an important point: The dogwalk is not uniquely dangerous, but it is uniquely demanding.
Important Limitations
As with any field-based study:
“Incident” reporting may vary between judges
Some events may be missed or underreported
Only immediate outcomes were captured (no follow-up injury data)
Rates are sensitive to small numbers, particularly for injury
These findings should be interpreted as descriptive and foundational, not definitive.
What Comes Next
This is just the first step.
Ongoing analyses will focus on:
Detailed incident characteristics
Contributing factors (course, environment, performance variables)
Video analysis of dogwalk falls
Integration with performance and footfall data
The goal is to move from:
“How often does this happen?”to“Why does it happen and how can we reduce risk?”
Bottom Line
Contact obstacle incidents are rare on a per-run basis
Most do not result in apparent injury
The dogwalk has the highest incident rate but it is still relatively low
Patterns are consistent across organizations
Objective, prospective data is now available to guide evidence-based decisions
A Note on Data Sharing
These results are part of an ongoing research project. They are being shared in summary form to inform the community while full analyses are prepared for peer-reviewed publication.
Want to stay up-to-date on the latest research? Join our email list here.