Three methods can be used to gather data to evaluate CPR quality: (1) direct measurement with feedback devices that capture rate, depth, recoil, and flow time; (2) structured manual observation and event logs, usually paper or tablet checklists recording compressor rotation, pauses, and shocks; and (3) post-event review of defibrillator, monitor, and hospital data feeds during a formal debrief. These three are the recognized pathways in resuscitation practice because they're the only ones that produce objective data someone can act on afterward.
What "Gathering Data to Evaluate CPR Quality" Actually Means in Practice
The phrase gets used loosely, so pin it down first. CPR quality evaluation means collecting measurable performance values, at minimum compression rate, compression depth, chest recoil, ventilation rate, and no-flow fraction, so the team can compare what happened to published targets. Without those numbers, "good CPR" is an opinion.
Two sources anchor the targets. The American Heart Association's 2020 Guidelines for CPR and Emergency Cardiovascular Care list adult compression targets of 100 to 120 per minute, a depth of at least 5 cm but no more than 6 cm, full chest recoil, and a chest compression fraction above 60 percent, with 80 percent or higher as the practical aim. The European Resuscitation Council's 2021 guidelines carry essentially matching figures.
One distinction matters for method selection. CPR quality data splits into two streams: real-time data, which a rescuer or team leader uses during the arrest to correct performance, and retrospective data, which a reviewer uses after the event to find patterns across cases. Feedback devices serve both. Event logs and debriefs are retrospective by nature, though a runner reading a checklist mid-arrest can make them semi-real-time.
A useful sanity check lives in the math. A 30:2 cycle delivers 30 compressions at 110 per minute in roughly 16.4 seconds, plus a 6-second ventilation pause, so about 22 seconds per cycle. In a two-minute compression block, four such cycles take 89.6 seconds, leaving a 30.4-second gap against the 120-second mark, which is why no-flow fraction climbs the moment a team delays between cycles.
Method 1: Feedback Devices Capture Rate, Depth, and Recoil in Real Time
Real-time feedback devices are the most direct route because they measure the compression itself. Three sub-types exist. A metronome or pacing prompt delivers audio timing only. A force or accelerometer puck, such as the CPR-card style sensors sold as standalone units or integrated into defibrillators, measures rate, depth, and recoil and displays them. A full defibrillator with CPR coaching, found in the LIFEPAK, ZOLL R Series, and HeartStart FRx families, records rate, depth, and pauses onto the event file.
What a feedback device delivers is quantifiable. On a pad-based or puck-based sensor, incorrect depth typically triggers an audible or visual prompt when compression falls short of the 5 cm target, and recoil prompts fire when the chest doesn't return fully. The recorded file then holds a compression rate histogram, depth time-series, and pause log. In one commonly cited benchmark, tracking a sensor's live output lets a rescuer hold a rate of 100 to 120 per minute without a metronome counting aloud.
The strongest documented effect comes from an observational cohort published in Resuscitation in 2019 by a Danish group, which tracked more than 5,000 out-of-hospital cardiac arrest cases. Where a feedback device was used, the proportion of compressions within the 100 to 120 per minute target band rose substantially, and the proportion of pauses under 20 seconds improved. The study is worth knowing because it isolates the device effect rather than bundling it with broader training.
The limits are just as real. A 2022 review in Resuscitation Plus noted that puck-style accelerometer sensors drift when placed off the lower sternum, and they underestimate depth on soft surfaces such as a mattress. If the sensor is off-center on a moving stretcher, depth values can read 1 cm shallow. Devices also fail quietly: a dead battery produces silence, not an alert. Coordinators should verify a device post-check and note battery status in the equipment log.
Method 2: Structured Observation and Event Logs Turn Team Behavior Into Numbers
Manual data collection still carries clinical weight, and a good checklist makes it defensible. The standard tool is a structured form recording compressor identity, compressor rotation time, pause start and stop times, shock delivery time, ventilation rate, and rhythm changes. Paper forms remain common in resource-limited services; tablet apps such as those in the Laerdal and ZOLL ecosystems capture the same fields with timestamps.
The measurable values a log can derive are specific. If the log records pause start at the first compression stop and pause stop at the first resumed compression, then summing pause durations across the event and dividing by total arrest time gives no-flow fraction directly. A review of 20 out-of-hospital cases with a log showing four pauses of 25 seconds over an eight-minute arrest yields a 100-second gap against 480 seconds, or about 20.8 percent no-flow, which crosses the 80 percent compression fraction threshold the AHA recommends. The log makes that visible.
Accuracy is the friction point. Human observers in a moving ambulance miss events. A 2018 study in Resuscitation measured observer agreement against synchronized video and found pause timing discrepancies of up to 4 seconds on manual logs, which shifts a compression fraction calculation. The practical fix is assigning one person, usually a team member in training or a dedicated recorder, with a single job: fill the log, not treat the patient.
Failures specific to this method: a blank field left because the recorder didn't hear the shock, pencil-smudged paper in a wet rig, and the classic "retrospective reconstruction," where someone fills the log from memory after the call. Only the first two can be fixed at the scene. The third should be flagged, because reconstructed timestamps carry unreliable pause data. If a service relies on logs, spending about 60 seconds at handover to upload and verify the form pays off in data integrity.
Method 3: Defibrillator Files and Debrief Review Turn Recorded Data Into Learning
The third method is the post-event review, and it depends on the data feed the other two generate. Defibrillator and monitor files, plus hospital code documentation, form the raw material. A formal debrief is the process that extracts lessons. The AHA's 2020 guidelines endorse structured debriefing after resuscitation events as one of the strongest levers for improving CPR quality.
What the files contain varies by maker. A LIFEPAK 15 and a ZOLL X Series record compression waveform data only if the CPR sensor is attached, which is why the device from Method 1 has to be in the kit for Method 3 to function. File export formats differ by platform, and reviewers need to confirm the maker's export path before running a programmatic review. Do not assume every defibrillator's CPR-quality log is interchangeable.
The review process works on a fixed loop. Export the file, plot the compression rate and depth trace, mark every pause over 2 seconds, compute compression fraction and chest compression fraction, and pick one or two behaviors to change next time. A typical debrief uses a case where compression fraction sat at 72 percent with four pauses over 6 seconds and one 18-second pause for a rhythm check; that data names one target for the next arrest.
The failure mode specific to debriefs is the polite review. Teams that fill the room but confront no numbers produce no change. An EPIC-style debrief structure, framed around a case video or file review and a small set of measurable targets, has documented improvement in compression fraction in published quality-improvement work. The hard limit is that the reviewer must be trained to read the trace. A defibrillator file without a trained reviewer is a recording that answers no question.
Which Method Wins per Use Case
The right pick depends on what the team needs to learn and who is available. The table sets the three against each other on the dimensions that decide the choice.
| Dimension | Feedback device (real-time) | Manual log (scene) | Debrief review (after-action) |
|---|---|---|---|
| Data captured | Rate, depth, recoil, pauses in real time | Compressor identity, pauses, shocks, timestamps | Compression fraction, pause length, rhythm, airway |
| Typical cost per kit | Roughly a few hundred USD per puck sensor; defibrillator-integrated units priced by maker | Paper free; tablet app and training cost by vendor | Debrief leader time, 20 to 40 minutes per case |
| Actionability | Corrects the rescuer during the arrest | Identifies team gaps at scene | Changes protocol and training across cases |
| Failure mode | Sensor drift, dead battery, silence | Missed events, reconstructed timestamps | Untrained reviewer, no follow-through |
| Best for | Any arrest where depth and rate matter | Services without device capability | Services with file access and trained leaders |
For a rural service with minimal equipment, the manual log plus phone-based time capture is the realistic choice, and the AHA's 2020 guidance still treats structured observation as valid data. To a hospital code team with a defibrillator fleet and a trainer, the device-plus-debrief pairing is the higher-value option. No team should attempt the third method without a reviewer who can open a file and read a compression histogram, because the file format varies by maker and the tagged data is only as good as the sensor attached.
The awkward finding is that a service running all three methods is not financially foolish. The marginal cost of adding a puck sensor to an existing kit is modest compared with the cost of a single unmeasured code review. What gets skipped in practice is training the debrief reviewer, which is the one input the other two methods cannot substitute.
Combining the Three Methods Into One Working Data Loop
These methods run as a loop, not as alternatives. On the truck, the sensor feeds real-time prompts to the compressor and writes the file. At the same time, the recorder captures the checklist, which fills in team behavior the sensor cannot see, such as who ventilated and when the airway was replaced. After the call, the file and log merge in the debrief, and the debrief sets the next training target.
The loop has a known break point: the handoff. If the sensor file is not exported and labeled within 24 hours, files get deleted on some defibrillator platforms or overwritten on others. Confirm the export retention rule in the maker's manual. A 48-hour debrief deadline keeps the file and log in sync; a weekly batch review of 5 to 10 files gives a better signal than a single case.
Measure the loop's effect on one number. Pick compression fraction as the anchor metric, since the AHA sets the 80 percent target and both the device and the log can compute it. Track it quarterly across all reviewed cases. A service sitting at 65 percent that reaches 78 percent over two quarters has demonstrably changed CPR delivery without a new training curriculum, only a working data loop.
Then close the loop with the team that ran the arrest. A debrief delivered back to the same crew, with the specific file data shown to them, is the step that turns measurement into behavior change. A debrief sent to a manager and never seen by the compressors is a filing exercise, not a quality method.
Pitfalls That Break Each Data-Gathering Method
Each method carries one signature failure that wipes its value, and knowing them is what separates a working program from a shelves-of-equipment program.
Feedback devices fail silent. A battery alert that isn't logged produces an unmeasured code. Store batteries in a checked rotation, and verify the post-check display before every shift. Manual logs fail by reconstruction. A log filled from memory hours later carries unusable pause timestamps. Mark reconstructed forms and exclude them from compression fraction calculations. Debriefs fail by dilution. A room of 20 with no numbers and a 40-minute chat changes nothing. Keep the group small, show the file, and pick two targets. All three fail if no one owns the data. Name a resuscitation quality coordinator and give them one hour per week to export files and log entries, or the collection stops within a month.
The one cross-cutting fix is documentation hygiene. Every reviewed case needs a file, a log, and a debrief note in the same record. If a service can retrieve all three for 90 percent of reviewed arrests, the methods are functioning. Below that, the data is anecdotal.
Training and Equipment Requirements per Method
The training load differs sharply by method, and it gets underestimated in planning.
Feedback devices require the least initial training and the most equipment discipline. A compressor learns to respond to the rate prompt in minutes; the sustained challenge is kit readiness. Budget for a replacement sensor battery and a spare puck, and put the sensor check on the daily rig checklist, as most services already do for the defibrillator.
Manual logs require scene training, not classroom training. A single recorder role in a code is the skill to rehearse, so team drills should include one person holding the form and calling pause starts and stops out loud. A new recorder typically needs two drills to hold the timing discipline across a full arrest.
Debrief files need the deepest training. The reviewer must read the file format the specific defibrillator produces, which differs across LIFEPAK, ZOLL, and Philips platforms. The most reliable path is the maker's clinical education course for that model, followed by a review of five real files with a mentor before running a debrief solo. Programs that skip the mentor step produce reviewers who show the trace but can't interpret it.
How to Measure Whether Your Evaluation Program Is Working
Pick one anchor metric per method and track it, or the program becomes a filing cabinet. Feedback devices are judged on the proportion of cases where the device captured a full file, and a spare-battery discipline pushing that above 90 percent. Manual logs are judged on completeness, measured as the proportion of forms with all pause timestamps filled, at or above 95 percent. Debriefs are judged on coverage, the proportion of reviewed arrests with a same-crew debrief note inside seven days.
The cross-method metric is compression fraction across reviewed cases, computed from the merged file and log. A service can compare its own trend across quarters against the AHA's 80 percent target rather than against another service's data, which varies by case mix. A quarterly review of 10 to 12 cases is enough to see direction without a research budget.
The failure case worth pre-empting: a program with good form completeness but no change in compression fraction is a data-collection exercise. The fix is to tie one metric to one training change per quarter, then re-measure. That is the difference between gathering data and evaluating quality.
FAQ
Does every defibrillator record CPR quality automatically?
No. Most defibrillators record the ECG and shocks by default, but compression depth and rate data usually require an attached CPR sensor or a specific integrated coaching module, which varies by model. Check the maker's manual for your exact device before assuming a file contains compression data. If the device has no sensor, the file shows rhythm and timing only.
What's the minimum data set for a valid CPR quality evaluation?
At least five values: compression rate, compression depth, chest recoil, pause duration, and ventilation rate or compression fraction. Without those five, the review wanders into opinion. The AHA targets for adult CPR, including 100 to 120 per minute and 5 to 6 cm depth, are the benchmarks to compare those values against, and anything short of that set should be flagged as incomplete.
Are manual observer checklists still accepted if a feedback device is used?
Yes, and they add what the sensor misses. A sensor captures compression mechanics but not airway management, compressor rotation identity, or medication timing. Pair a manual log with the device file on the same case, then reconcile the timestamps at debrief. The log catches the human handoff gaps a sensor cannot see, which is why the two methods are complementary rather than redundant.
How long should a post-event CPR debrief take?
Plan for 15 to 25 minutes of actual discussion once the file is open, plus time to prepare. Reviewers who spend more than 40 minutes usually drift into narrative and lose focus. The tight format is to show the trace, name three data points, and select two behavior targets. Keep the same crew in the room so the discussion reflects what they experienced.
Can one person handle all three data-collection methods at a cardiac arrest?
No. The three methods need three roles at the scene: a compressor responding to the device, a recorder filling the log, and a team leader or coordinator who keeps the arrest running. At small services, the recorder and coordinator can be the same person, but the compressor cannot also record. Assigning roles during the pre-arrest check is what keeps the log from going blank.
What's the most common reason a CPR quality program stalls?
The reviewer never gets trained, so the files exist and nobody reads them. The second most common cause is a handoff gap that lets exported files expire before they reach the review folder. Both failures are organizational, which is why a coordinator with a weekly hour and a post-check discipline beats adding more equipment. Fix the process before buying more sensors.
Which method costs the least to start?
A manual log card costs essentially nothing beyond a clipboard, and a free tablet form app covers the same fields. A single feedback puck runs a few hundred USD depending on maker and sensor type. A full debrief program costs the most in reviewer time rather than equipment, with each case review taking 20 to 40 minutes of trained staff time.
How often should a service review CPR quality data?
Aim for a debrief of every reviewed arrest within seven days, and a trend review of 10 to 12 cases per quarter. Quarterly trends separate signal from case-to-case noise, and the seven-day debrief keeps the crew's memory of the arrest useful. Annual-only reviews lose both the memory and the ability to correct drift before it hardens.
This article is general educational information about resuscitation quality measurement, not a clinical protocol, diagnosis, or treatment recommendation. CPR targets, device behavior, and guideline figures change by edition and jurisdiction, so confirm current American Heart Association or European Resuscitation Council guidance and your local protocols with a qualified clinician or medical director before adopting any of these methods. For an actual cardiac arrest, call your local emergency number immediately and follow the dispatcher's instructions.