The three correct methods for gathering CPR quality data are capnography, visual observation, and feedback devices. Workshops and focus groups are training or research formats, not data-collection instruments for CPR performance. The three valid methods differ in objectivity, timing, and what they can actually measure, and each carries distinct blind spots that determine when it should be trusted.
How each method captures CPR quality
Capnography measures exhaled carbon dioxide (ETCO2) continuously through a sensor in the airway circuit. During CPR, pulmonary blood flow is the main determinant of CO2 delivery to the lungs, so ETCO2 functions as a real-time proxy for cardiac output generated by compressions. A sudden rise in ETCO2 during resuscitation suggests return of spontaneous circulation (ROSC). Capnography also captures ventilation rate and over-ventilation, which directly affect intrathoracic pressure and venous return. The mechanism is physiologic rather than mechanical: it reports the downstream effect of compressions on circulation.
Visual observation uses a trained person or a checklist to score rate, depth, recoil, and hand position. It requires no equipment on the patient, which makes it the fallback method in any setting. Its weakness is inherent: a human observer estimating 100 to 120 compressions per minute and 5 to 6 centimeters of depth by eye is working at the edge of perceptual accuracy. Observation captures what the compressor does, not what the patient receives.
Feedback devices are sensor-based systems built into defibrillators, CPR pads, or standalone units. They measure acceleration or force at the sternum and display real-time prompts for rate and depth. Some calculate a chest compression fraction (CCF) and generate a post-event report. The device closes the loop between performance and correction within seconds, which observation cannot do and capnography does only indirectly.
What the evidence says about each method's accuracy
Feedback devices outperform human estimation on the variables they measure. Studies using device-recorded data consistently find that rescuers without feedback compress too fast and too shallow, and that rate and depth improve when auditory or visual prompts are active. The devices are most accurate for rate; depth measurement is sensitive to hand position and chest compliance, so a device can read adequate depth while actual compressions are less effective.
Visual observation is reliable for gross errors and unreliable for fine ones. Observers detect absent recoil, wrong hand placement, and obvious pauses well. They systematically overestimate depth and underestimate rate. A checklist improves consistency between observers but does not fix the perceptual ceiling. This matters for training and team debriefing, where observation is the practical option.
Capnography is the most physiologically meaningful measure and the least specific for any single CPR component. A low ETCO2 can reflect poor compressions, low systemic perfusion, severe acidosis, or a combination. Interpretation requires the trend, not the single value. The professional caveat: an ETCO2 that does not rise despite apparently adequate compressions is a signal to reassess the whole resuscitation, not proof that the compressor is failing.
Quantified context: thresholds and targets
Current adult CPR targets set compression rate at 100 to 120 per minute, depth at 5 to 6 centimeters (about 2 to 2.4 inches), full recoil between compressions, and a chest compression fraction above 60 percent, with 80 percent or higher as a quality goal. Ventilation is targeted at roughly 10 breaths per minute with minimal interruption. Feedback devices typically display these thresholds directly.
ETCO2 interpretation during CPR uses trends rather than fixed cutoffs. Persistently very low values (commonly cited below 10 mmHg) after adequate ventilation and compressions suggest poor perfusion and a low likelihood of survival; a value that rises abruptly by a meaningful margin suggests ROSC. Reconcile the units: centimeters for depth in most international guidance, inches in US practice, and millimeters of mercury for ETCO2. A single number rarely changes management; the trajectory across minutes does.
Timeline table: when each method delivers usable data
| Phase | Capnography | Visual observation | Feedback device |
|---|---|---|---|
| First 30 seconds | Waveform appears once the airway is secured | Immediate, no setup | Immediate if pads or sensor already placed |
| 1–2 minutes | ETCO2 trend establishes perfusion baseline | Observer scores rate and depth by eye | Real-time rate and depth prompts active |
| Intra-event | Detects ROSC rise and over-ventilation | Captures pauses and hand position | Flags compression fraction and pauses |
| Post-event | Waveform review in debrief | Checklist review in debrief | Downloadable report with minute-level data |
| Training use | Limited simulator value | Core method for skill assessment | Standard for objective skill scoring |
The practical read: observation starts fastest, capnography becomes most informative after a minute or two, and feedback devices produce the most reliable post-event record.
Comparison table: strengths and limits side by side
| Method | What it measures | Objectivity | Main blind spot |
|---|---|---|---|
| Capnography | Circulatory effect of compressions and ventilation | High for physiology | Cannot isolate compression quality from perfusion |
| Visual observation | Visible technique and timing | Low to moderate | Overestimates depth, misses fine rate errors |
| Feedback device | Rate, depth, recoil, compression fraction | High for mechanical metrics | Depth readings affected by position and compliance |
| Workshops | Skills practice and teaching | Not a measurement method | No patient data collected |
| Focus groups | Perceptions and team experience | Not a measurement method | Recall bias, no performance data |
The two distractors are excluded for the same reason: they produce opinions and practice opportunities, not measurement of actual CPR performance. A workshop can include feedback devices, but the workshop itself is not the data source.
Edge cases and the counterintuitive detail
The counterintuitive finding is that better device compliance does not automatically mean better resuscitation. Rescuers who chase a device's depth prompt can lean on the chest and fail to allow full recoil, which impairs venous return. Rate prompts can also fragment into mechanical, stop-start compressions. The device measures the component it was designed to measure; it does not integrate the gesture the way a patient's circulation does. That is why capnography and feedback data are complementary rather than redundant.
Pediatric and neonatal resuscitation shift the thresholds and the interpretation. Depth targets are proportionally different, ventilation rates are higher, and ETCO2 kinetics differ with smaller lungs and congenital heart disease. Use age-specific guidance rather than scaling adult numbers.
Mechanical CPR devices and manual CPR produce different ETCO2 patterns, so device-specific experience matters when interpreting trends. Similarly, a patient with severe metabolic acidosis or a pulmonary embolism may generate misleading ETCO2 values regardless of compression quality.
Red flags when interpreting CPR quality data: ETCO2 flat or falling despite good compressions: reassess airway, ventilation, and reversible causes. Device shows good depth but waveform and perfusion remain poor: suspect recoil failure or poor hand position. Observation scores look clean but compression fraction is below 60 percent: pauses are being missed by human eyes. Ventilation rate above about 10 per minute with falling ETCO2: over-ventilation is likely impairing hemodynamics. A single metric presented as proof of quality without corroborating data: treat as incomplete evidence.
Practical actions and thresholds
A layered approach is the defensible one. Use feedback devices as the default intra-event measurement when available, because they give actionable rate and depth data in real time. Add capnography whenever an advanced airway is in place, and read it as a trend against a baseline established in the first minute or two. Use structured visual observation when equipment is absent and as a training and debriefing tool, with a checklist to standardize scoring.
For programs evaluating resuscitation performance: Collect device reports after every event and review compression fraction against the 60 percent floor and 80 percent target. Pair ETCO2 trends with device metrics in debriefs to distinguish technique from physiology. Train observers with a validated checklist and calibrate them against device data periodically. Set a review threshold: any event with compression fraction below 60 percent or a sustained ETCO2 trend below 10 mmHg triggers case review. Treat workshops and focus groups as education and qualitative research, and label them as such in any quality report.
Limitations and what the data cannot tell you
No method directly measures coronary perfusion pressure at the bedside. Devices measure surrogate mechanics, capnography measures a downstream physiologic signal, and observation measures visible behavior. All three are proxies, and none proves that a given compression generated adequate blood flow in a specific patient. Device accuracy varies by model and by patient anatomy, and observational accuracy varies by observer training.
The three-method answer is therefore a starting point for measurement design, not a ranking. The strongest evaluations triangulate all three where feasible, and state clearly which metric supports which conclusion. Where evidence for a specific device or threshold is limited, the defensible position is to treat the data as directional and to avoid claims of certainty.
This article is general educational information about CPR quality measurement, not a clinical protocol, diagnosis, or treatment plan. It does not replace current resuscitation guidelines or training from a qualified instructor. In an actual emergency, follow local emergency services and your certified training.