The finding, published in Frontiers in Medicine, does not mean that slowing down ambulances saves lives. Instead, the researchers argue, the interval from collapse to first medical contact may be acting as a hidden marker of who is salvageable in the first place, rather than a pure measure of treatment delay. The research team, led by Yixin Weng, Gaohaozhe Hu and Xuefeng Ma of the Department of Emergency at Cixi Emergency Center, examined every out-of-hospital cardiac arrest managed by their service between April 2023 and March 2026.

After exclusions for arrests witnessed by ambulance crews, cases where resuscitation was withheld, and records with incomplete timelines, 152 consecutive patients remained. Of these, 52, or 34.2 percent, achieved return of spontaneous circulation, known as ROSC, before hospital arrival. The outcome was defined strictly: a palpable pulse and measurable blood pressure sustained for at least twenty seconds without ongoing chest compressions, confirmed by the attending emergency physician.

What makes the study methodologically interesting is how the time intervals were constructed. Rather than relying on a single response-time figure, the team synchronized two independent data sources: the dispatch center's automatic logs of incoming calls, vehicle dispatches and scene arrivals, and the standardized prehospital record completed by the physician at the scene. From these they derived six distinct intervals, including the time from call to dispatch, dispatch to scene, scene to first medical contact, collapse to first medical contact, and collapse to the first chest compression.

When the arrest was unwitnessed, collapse time was estimated by the attending physician based on witness accounts and scene information. Discrepancies of more than one minute between sources were cross-checked against audio recordings of dispatch communications. The headline result was stark.

The median collapse-to-first-medical-contact interval was 20.0 minutes in the ROSC group versus 13.1 minutes in the non-ROSC group, a difference that was highly significant. By contrast, the conventional system metrics told no such story. EMS arrival times, call-to-dispatch times and dispatch-to-scene times were statistically indistinguishable between the two groups, hovering around ten to thirteen minutes in both.

In other words, the ambulances arrived equally fast for everyone; what differed was the clock that started at the moment of collapse, a clock that includes the patient-side delay of collapse recognition and call activation before the system is even engaged. To test whether this paradox survived statistical scrutiny, the researchers built a series of logistic regression models with a deliberately parsimonious design. Variables were chosen on prespecified causal grounds rather than by univariate significance screening, a strategy meant to avoid sparse-data bias in a cohort of this size.

The main model contained only the two core time intervals. A second model added age, and a third added the etiology of the arrest and the presence of a shockable initial rhythm, all treated as pre-exposure confounders because they precede the timing of medical contact. Post-exposure interventions such as intubation, intravenous access and adrenaline were deliberately excluded, since they lie on the causal pathway after first contact and adjusting for them would introduce overadjustment bias.

The association held. In the fully adjusted model, each additional minute of collapse-to-first-medical-contact time was associated with higher odds of prehospital ROSC, with an odds ratio of 1.188. Meanwhile, the collapse-to-CPR initiation interval showed a modest inverse association in the same model, with an odds ratio of 0.954, closer to the traditional expectation that longer no-flow periods hurt.

Bootstrap internal validation with 1,000 resamples yielded an optimism-corrected area under the receiver operating curve of 0.882, a calibration slope of 0.972 and a Brier score of 0.132, indicating that the model discriminated and calibrated well without obvious overfitting. As a single indicator, the collapse-to-FMC interval alone achieved an AUC of 0.737 with an optimal cutoff near fifteen minutes. The baseline characteristics offer the first clue to what is going on.

Patients who achieved ROSC were far more likely to have arrests of non-cardiac origin, 71.2 percent versus 22.0 percent, and more likely to present with a shockable rhythm such as ventricular fibrillation, 30.8 percent versus 16.0 percent. Both features are known to carry better prognoses. The authors interpret this as evidence that the ROSC group represents a distinct clinical phenotype with a greater probability of regaining a perfusing rhythm, and they suggest that resuscitation teams may invest longer on-scene efforts in patients they judge to be potentially salvageable, which mechanically lengthens the recorded intervals in those who succeed.

The subgroup analyses add nuance, though the authors are careful to label them exploratory because formal interaction tests did not reach significance. Among patients younger than 65, the collapse-to-FMC interval remained independently associated with ROSC after adjustment, with an odds ratio of 1.126, whereas in older patients no significant association appeared. Conversely, in the cardiac etiology subgroup, the collapse-to-CPR initiation interval became significantly associated with ROSC after adjustment, with an odds ratio of 1.567, a signal invisible in the unadjusted data.

The researchers speculate that some cardiac arrests unfold through a gradual agonal phase in which witnesses mistake abnormal breathing for continued life and delay calling for help, and in which the heart and brain may tolerate the hypoxic interval better than in a sudden arrhythmic collapse. This interpretation remains hypothetical. The authors are explicit about the limits of what can be concluded.

The outcome measured was any prehospital ROSC, including transient episodes that may have been followed by re-arrest, and no data on survival to discharge or neurological outcome were available. Collapse times for unwitnessed arrests depend on estimation and are vulnerable to recall bias, which could systematically inflate the interval in survivors. Resuscitation-duration bias, residual confounding by physician experience and arrest severity, and the modest single-center sample of 152 patients all temper the findings.

The study also cannot support any change to termination-of-resuscitation rules, which the authors note require validated multi-criteria frameworks rather than a single time interval. What the study does challenge is the reflexive equation of speed with quality in emergency medical services. Because dispatch-related intervals were identical between outcomes, a performance dashboard built solely on response times would have missed everything that distinguished the two groups, including the markedly higher rates of intubation, intravenous access, adrenaline administration and clinical pathway activation among ROSC patients.

The authors argue that prehospital quality improvement should weigh the timeliness of arrival against the quality and completeness of on-scene resuscitation, and that time metrics should be interpreted in the context of arrest etiology, patient age and initial rhythm rather than as universal targets. Larger, multicenter prospective studies with detailed time-event sequences and long-term follow-up will be needed to determine whether the paradox reflects case-mix, pathophysiology, or both.

Subject of Research: Association between prehospital time intervals and return of spontaneous circulation in out-of-hospital cardiac arrest

Article Title: Longer collapse-to-first-medical-contact interval is independently associated with ROSC in out-of-hospital cardiac arrest: a paradoxical finding that may reflect case-mix differences

Article References: Weng, Y., Hu, G., & Ma, X. (2026). Longer collapse-to-first-medical-contact interval is independently associated with ROSC in out-of-hospital cardiac arrest: a paradoxical finding that may reflect case-mix differences. Frontiers in Medicine, 13, 1989379. https://doi.org/10.3389/fmed.2026.1989379

Image Credits: AI Generated

DOI: 10.3389/fmed.2026.1989379

Keywords: out-of-hospital cardiac arrest, return of spontaneous circulation, prehospital time intervals, emergency medical services, first medical contact, cardiopulmonary resuscitation, shockable rhythm, case-mix, resuscitation outcomes, retrospective cohort study, EMS response time, cardiac arrest etiology