Urban-rural disparities in out-of-hospital cardiac arrest outcomes: a nationwide Hungarian study

Abstract

Background: Out-of-hospital cardiac arrest (OHCA) outcomes often differ between urban and rural settings, but comprehensive nationwide data from Central-Eastern Europe using uniform data collection and modern confounding control remain limited. We investigated urban-rural disparities in OHCA outcomes in Hungary.

Methods: We analysed 130,258 OHCA cases (2018-2023) from the Hungarian National Ambulance Service registry, classified as urban (70.1 %) or rural (29.9 %) using national administrative categories. The primary outcome was on-scene return of spontaneous circulation (ROSC). We performed univariable and multivariable logistic regression, propensity score matching (PSM) and continuous response-time modeling using natural cubic splines.

Results: The overall ROSC rate was 9.1 % (urban: 9.4 %, rural: 8.3 %, p < 0.001). After PSM, urban location remained significantly associated with higher survival (OR = 1.26, 95 % CI 1.20-1.32, p < 0.001). EMS response times were significantly longer in rural areas (median 14.9 vs 9.8 min, p < 0.001). Urban survival advantage was most pronounced in cases with shockable rhythms (OR = 1.57, 95 % CI 1.43-1.72), medical-witnessed arrests (OR = 1.31, 95 % CI 1.20-1.42), and response times ≤8 min (OR = 1.59, 95 % CI 1.44-1.76).

Conclusions: Significant urban-rural disparities in OHCA on-scene ROSC persist even after accounting for patient and arrest characteristics. These findings highlight the need for targeted interventions to strengthen the Chain of Survival in rural communities.

Keywords: Cardiac arrest (CA); Cardiopulmonary resuscitation (CPR); Emergency medical services (EMS); Out-of-hospital cardiac arrest (OHCA); Urban-rural disparity.

Fig. 1

Standardised Mean Differences for Baseline Characteristics Before and After Matching in Urban and Rural OHCA Cases. The love plot presents the standardised mean differences (SMD) for key baseline characteristics before and after propensity score matching in urban and rural OHCA cohorts. Before matching, substantial imbalances were observed in multiple variables, including age, bystander-witnessed status, and EMS response times, reflecting marked demographic and structural differences between urban and rural cases. The dashed lines at ±0.1 represent the common threshold for acceptable balance after matching. Following matching, the balance of covariates improved substantially, with most SMDs reduced below the 0.1 threshold, indicating successful adjustment for baseline disparities. EMS response time showed residual imbalance after matching. This confirms that the matched cohorts were sufficiently comparable to isolate the independent effect of location on outcomes.

Fig. 2

Model Performance and Calibration in the Matched Cohort. (A) Receiver operating characteristic (ROC) curve demonstrating the discriminatory ability of the model in the matched cohort (AUC = 0.764). The blue line represents the ROC curve, with the grey shaded area showing the 95 % confidence interval. (B) The calibration plot compares the predicted versus observed survival probabilities, showing the relationship between model estimates and actual outcomes across urban (green), rural (red), and overall cohort (blue) predictions. The deviation from the ideal diagonal line (dashed) indicates some model miscalibration, particularly in the higher probability ranges (>0.7) where predictions slightly overestimate actual outcomes. Despite this visual difference, the overall calibration remains statistically acceptable as confirmed by the non-significant result of the Hosmer-Lemeshow test (p = 0.218). This pattern of calibration is common in rare-event prediction models and indicates that predicted probabilities may require minor adjustment for probabilities exceeding 0.7. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3

Impact of EMS Response Time Thresholds on Urban-Rural Survival Disparities. (A) Survival Rates. This panel demonstrates how survival rates differ between urban and rural settings across various EMS response time thresholds. Each threshold on the x-axis (5, 8, 10, 15, 20, and 30 min) creates two patient groups: those with 'fast' responses (≤ threshold, solid lines) and those with 'slow' responses (>threshold, dashed lines). The y-axis shows survival percentages. For urban settings (blue lines), patients receiving fast responses (solid blue) consistently show higher survival rates (10–23 %) compared to those with delayed responses (dashed blue, 7–10 %). Similarly, for rural settings (red lines), fast responses (solid red) yield better outcomes (8–12 %) than delayed responses (dashed red, 4–8 %). Notably, the survival advantage of fast response is more pronounced in urban areas, particularly at the earliest time points (≤5 min), where urban survival reaches approximately 23 % compared to 12 % in rural areas. (B) Urban-Rural Odds Ratios. This panel quantifies the urban survival advantage relative to rural areas, expressed as odds ratios (OR). The blue line represents the urban–rural OR for patients with 'fast' responses, whereas the orange line shows this relationship for 'slow' responses. For fast responses (blue line), we observe that at the 5-min threshold, urban patients have 2.37 times higher odds of survival than rural patients (OR = 2.37, 95 % CI 1.87–3.00). This advantage gradually diminishes as the threshold increases, reaching OR = 1.22 (95 % CI 1.15–1.31) at the 30-min threshold. For slow responses (orange line), a different pattern emerges, at early thresholds, the urban–rural difference is minimal (OR = 1.20 at 5 min), the urban advantage grows substantially at longer thresholds, reaching OR = 2.13 (95 % CI 1.78–2.54) at 30 min. These contrasting patterns suggest that urban settings maintain an advantage throughout all response times, but the nature of this advantage changes. For rapid responses, urban advantage likely stems from factors such as more experienced providers and better system coordination. For delayed responses, the advantage may reflect greater availability of advanced life support resources in urban areas that can sustain resuscitation efforts over longer periods. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4

Subgroup Analysis: Urban vs Rural Odds Ratios. Forest plot showing the odds ratios with 95 % confidence intervals for urban versus rural survival across different initial cardiac rhythms. The vertical dashed line at OR = 1.0 represents equal odds between settings, whereas OR > 1.0 represents better survival of urban-located OHCA. The highest urban advantage was observed in VF/VT (OR = 1.57) and bradycardia (OR = 1.03), whereas less pronounced in asystole (OR = 1.22) and PEA (OR = 1.09). Odds ratios with corresponding 95 % confidence intervals and p-values are shown on the right side.

Supplementary Fig. 1

Supplementary Fig. 2

Supplementary Fig. 3

Supplementary Fig. 4

Supplementary Fig. 5