Abstract
Background
As the complexity of cardiac surgeries increases and patient selection criteria expand, the use of veno-arterial extracorporeal membrane oxygenation for high-risk patients has become more prevalent. Despite its critical role in sustaining life, postcardiotomy ECMO (PC-ECMO) is associated with high in-hospital mortality rates. Intensive care unit (ICU) capacity and staffing are crucial in determining patient outcomes. This study aimed to investigate the relationships among ICU capacity, staffing levels, and outcomes in PC-ECMO patients in China.
Methods
A multilevel cross-sectional analysis was conducted using data from 586 hospitals that participated in China’s National Quality Improvement Program in 2018. From these hospitals, we selected those that performed PC-ECMO procedures between April 2016 and December 2021. The novel ICU Capacity Comprehensive Index (ICUCCI) was calculated for each hospital, incorporating medical service capacity, technical ability, quality and safety, and service efficiency. ICU staffing was assessed by patient-to-bed, patient-to-physician, and patient-to-nurse ratios. The primary outcome was in-hospital mortality, with secondary outcomes including complications, length of stay (LOS), and hospitalization costs.
Results
A total of 102 hospitals, encompassing 2,601 patients, were included in the analysis. Higher ICUCCI values were associated with reduced in-hospital mortality (OR: 0.83, 95% CI: 0.70–0.97, P = 0.025) and fewer complications (OR: 0.82, 95% CI: 0.68–0.99, P = 0.046). However, higher ICUCCI values correlated with longer LOSs (IRR: 1.14, 95% CI: 1.06–1.22, P < 0.001) and increased hospitalization costs (IRR: 1.32, 95% CI: 1.24–1.40, P < 0.001). ICU staffing ratios, including patients per bed, physician, and nurse, were protective against mortality, with the ratio of patients per ICU bed showing the most pronounced effect (OR: 0.69, 95% CI: 0.55–0.87, P = 0.002). Increased staffing was also associated with longer LOS but did not affect overall complication rates or costs. The ratio of patients per ICU bed was linked to a greater risk of bloodstream infection (OR: 1.96, 95% CI: 1.14–3.46, P = 0.022).
Conclusions
This study highlights the critical role of ICU capacity and staffing levels in improving outcomes for patients receiving PC-ECMO. While higher ICU capacity and staffing are associated with reduced mortality, they also correlate with longer hospital stays and/or increased costs, suggesting the need for a balanced approach in resource allocation. Our findings underline the importance of optimizing ICU staffing ratios and enhancing healthcare equity to improve patient care across diverse healthcare institutions.
Supplementary Information
The online version contains supplementary material available at 10.1186/s13054-025-05443-2.
Keywords: ICU capacity, ICU staffing, Postcardiotomy-extracorporeal membrane oxygenation (PC-ECMO), Medical management, Quality control, Cross-sectional study
Background
The increasing complexity of cardiac surgeries and expanding indications have made high-risk patients eligible for procedures that were previously considered unsuitable [1, 2]. However, these patients face a significantly greater risk of failure to wean from cardiopulmonary bypass or the development of refractory cardiogenic shock postoperatively. In such cases, veno-arterial extracorporeal membrane oxygenation (VA-ECMO) has become a critical life-sustaining intervention [3]. Despite the increasing use of VA-ECMO, recent findings highlight persistently high in-hospital mortality rates among these patients [4, 5], emphasizing the pivotal role of hospital care and intensive care unit (ICU) management in determining outcomes [6].
ICUs are central to the management of patients receiving postcardiotomy extracorporeal membrane oxygenation (PC-ECMO), who require continuous monitoring, precise therapeutic adjustments, and multidisciplinary coordination [7]. The medical capacity, experience and workload of ICU staff are intrinsically linked to patient outcomes in such high-intensity environments [8, 9]. Our previous work through the ECMO Quality Improvement Action (EQIA) study revealed that ICU quality control indicators and volume parameters [10] are significantly associated with outcomes in veno-venal-ECMO-supported patients [11]. However, these studies also revealed key limitations, including that single quality indicators and simplistic volume metrics fail to fully capture the complexities of ICU management [12]. Building on this foundation, the present study introduced the multifaceted ICU Capacity Comprehensive Index (ICUCCI) and refined staffing parameters (volume of patients receiving PC-ECMO per ICU bed, physician, and nurse) to elucidate their relationships with PC-ECMO outcomes.
This research has significant implications for China’s healthcare system: First. Policy alignment: Our study is part of China’s National Clinical Specialty Capacity Evaluation Program, which aims to standardize and unify the assessment of clinical capabilities across healthcare institutions nationwide. For details about the National Clinical Specialty Capacity Evaluation and ICUCCI, please refer to the Supplement and sTable1, 2 and 3 & sFigure1, 2. Second. Hospital management optimization: From an administrative perspective, our data offer critical insights into the correlations among ICU staffing, ICU capacity, and the outcomes of patients receiving PC-ECMO. These findings provide an evidence-based foundation for optimizing medical resource allocation and advancing healthcare equity.
Methods
Study design and data sources
This multilevel cross-sectional study utilized data from 586 hospitals that participated in the 2018 National Quality Improvement Program survey in China [13]. The survey collected data on the number of registered ICU physicians and nurses, as well as ICU bed counts. From these institutions, we selected hospitals that performed PC-ECMO procedures between April 2016 and December 2021. The hospitalization details for each patient, including demographic information (e.g., age, sex), medical institution codes, diagnoses, surgical operation codes and dates, discharge status, length of stay (LOS), and associated costs, were obtained from standardized hospital discharge records submitted to the Hospital Quality Monitoring System (HQMS) [14]. The ICUCCI was calculated for each hospital using data from the National Health Commission’s integrated collection of real-world medical quality data, covering all secondary and tertiary public and private hospitals across China [15]. The collection of data was approved by the National Health Commission of China, and a waiver of informed consent was received from the Ethics Committee of Peking Union Medical College Hospital (PUMCH, ethics number I-23PJ1416). The study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.
Population, covariates, and outcome definitions
The target population consisted of patients who underwent cardiac surgery and required VA-ECMO support during or after surgery. The cohort was defined on the basis of the surgical and ECMO procedure dates recorded in the HQMS. Patients were categorized into three groups according to surgery type—coronary artery surgery, valve surgery, and other cardiovascular surgery—on the basis of standardized surgical codes. We excluded patients with more than one documented cardiac surgery or ECMO procedure and patients with duplicate records. Comorbidities and complications were determined on the basis of the admission status related to each diagnosis code. Comorbidities were classified through Elixhauser’s comorbidity classification method [16] and assigned in-hospital mortality risk scores on the basis of the van Walraven index [17]. The patients were further stratified into four risk groups [18]. For complications, we defined nine types on the basis of ECMO management protocols and expert opinions. The covariates included patient age, sex, surgery type, and baseline mortality risk. The primary outcome of interest was in-hospital mortality. The secondary outcomes included the occurrence of complications, LOS, and hospitalization costs. Hospitalization costs were adjusted according to the annual exchange rate between the Chinese yuan (RMB) and the U.S. dollar. See Supplemental sTables 4 & 5 for the standardized surgical codes, diagnostic codes and classifications for complications.
Exposure variables—ICU capacity comprehensive index (ICUCCI)
As one of the most populous countries in the world, China faces significant disparities in the distribution of healthcare resources across regions [19, 20]. In response, the National Health Commission introduced the “National Clinical Specialty Capacity Evaluation Program” in March 2024 to promote healthcare equity and improve the overall quality of medical services.
This program integrates real-world nationwide medical data through three major national datasets: the HQMS, the National Medical Quality Management and Control Information Network (NCIS), and the National Disease-Specific Quality Management and Control Platform. The data from these platforms are used to calculate the ICUCCI, which benchmarks hospitals nationwide, on a scale from 0 to 1, excluding personnel configurations. The ICUCCI encompasses four key dimensions—medical service capacity, technical ability, quality and safety, and service efficiency. This multifaceted approach provides an integrated measure of ICU capacity, offering a more robust and insightful evaluation of healthcare institutions.
Exposure variables—ICU staffing measurement (per bed, physician, and nurse)
ICU staffing was assessed by calculating the number of PC-ECMO patients per ICU bed, physician, and nurse at each hospital between 2016 and 2021. These indicators reflect the experience and workload of healthcare personnel involved in the postoperative management of patients receiving PC-ECMO [8, 21]. Our hypothesis was that lower staffing levels are associated with insufficient management experience, whereas higher staffing levels may initially improve patient outcomes [8]. However, beyond a certain threshold, excessively high staffing levels could lead to overwhelming workloads, potentially compromising the quality of care and worsening patient outcomes [8, 21].
Subgroup analysis
To assess the robustness of our findings, we performed subgroup analyses based on several factors, including age, emergency pathway admission, cardiopulmonary resuscitation (CPR) during hospitalization, LOS exceeding 90 days, surgery type, and the time period of patient inclusion (2016–2019 and 2020–2021). In addition, we also categorized these institutions into high an low volume centers based on the International Extracorporeal Life Support Organization (ELSO) standards (whether the annual ECMO case volume exceeds 20 cases) [22]. Another recent study from the ELSO registry revealed that the association between age and mortality or complications begins at 40 years of age for patients receiving VA-ECMO, prompting us to use the same age cutoff for stratification [23]. For patients whose LOS exceeds 90 days, we acknowledge that this may reflect long-term care needs, which could influence regression. Additionally, in the subgroup analysis, we excluded institutions with particularly low PC-ECMO case volumes (below the 25th percentile = 3 cases) to ensure the stability and reliability of our findings, as the outcomes from these institutions may not accurately represent institutional characteristics.
Statistical analysis
We performed descriptive statistical analyses to determine the ICUCCI, the number of ICU personnel and beds, the total number of patients receiving PC-ECMO, and ICU staffing parameters across institutions. At the institutional level, we calculated proportions for patient demographics, comorbidity severity, surgery type distribution, and clinical outcomes. The median (range or IQR) was used to describe the distribution.
Association analysis
For binary outcome variables, including in-hospital mortality and the occurrence of complications, we used multilevel random effects logistic regression models to estimate the associations between exposure variables and outcomes. The results are reported as odds ratios (ORs), which were calculated using the formula OR = (percentage/(100-percentage)). Each hospital was modeled as a cluster, accounting for random effects to capture institutional variations. When in-hospital mortality or complications were used as outcome variables, we adjusted for potential confounding variables using propensity scores. These scores were derived from separate logistic regression models that estimated the probability of mortality or the occurrence of complications on the basis of patient-specific characteristics.
For continuous outcome variables, including LOS and hospitalization costs, we employed zero-truncated negative binomial regression models to estimate the associations between exposure variables and outcomes. The exponentiated coefficients from these models were interpreted as incident rate ratios (IRRs). When analyzing LOS and hospitalization costs, we incorporated the same confounding variables as separate predictors in the model. Given that the LOS data and costs did not contain zero values, we opted to use a zero-truncated negative binomial regression model instead of a standard one, which could be biased due to the absence of zero counts. The statistical methods employed in this study are similar to those outlined in Aiken et al.’s multilevel cross-sectional study [21]. We used Python (3.8.5) for the statistical analyses.
Results
ICU capacity and staffing levels
Among the 586 surveyed hospitals, 102 performed PC-ECMO procedures and were included in the analysis. Among these, 16 hospitals were classified as high-volume centers, while 86 hospitals were classified as low-volume centers. The median ICUCCI was 0.39 (range 0.15–0.82) among these hospitals. Across these institutions, ICU resource levels varied significantly, with a median of 101.0 (range 16.0–616.0) ICU beds, 27.0 (range 4.0–185.0) ICU physicians, and 101.0 (range 4.0–687.0) ICU nurses per hospital. Between 2016 and 2021, a total of 2,601 patients across these institutions received PC-ECMO. The ICU staffing ratios varied considerably, with a median of 0.30 (range 0.03–3.99) patients receiving PC-ECMO per bed, 1.70 (range 0.06–18.5) patients per physician, and 0.25 (range 0.02–8.0) per nurse (Table 1, Fig. 1).
Table 1.
ICU Capacity and Staffing Levels
| Value | Median (Range) across hospital | |
|---|---|---|
| Hospital | 102 | / |
| ICU beds | 11,615 | 101.0 (16.0–616.0) |
| ICU physicians | 2805 | 27.0 (4.0–185.0) |
| ICU nurses | 11,405 | 101.0(4.0–687.0) |
| PC-ECMO / ICU beds | / | 0.30 (0.03–3.99) |
| PC-ECMO / ICU physicians | / | 1.70 (0.06–18.5) |
| PC-ECMO / ICU nurses | / | 0.25 (0.02–8.0) |
| ICUCCI | / | 0.39 (0.15–0.82) |
| Total ECMO volume per year | 1274 | 6.33(0.17–91.50) |
| High-volume center | 16 | 32.17(20.67–91.50) |
| Low-volume center | 86 | 4.75(0.17–19.67) |
ICUCCI ICU Capacity Comprehensive Index, PC-ECMO Postcardiotomy-Extracorporeal Membrane Oxygenation; High-volume center refers to hospitals performing more than 20 ECMO cases per year, while low-volume center refers to those performing 20 or fewer [22].
Fig. 1.
The Distribution of ICU Resources, The Implementation of PC-ECMO, ICUCCI and ICU Staffing Parameters across Different Hospitals. ICUCCI ICU Capacity Comprehensive Index, PC-ECMO Postcardiotomy-Extracorporeal Membrane Oxygenation, The violin plot illustrates the distribution of ICU resources, the implementation of PC-ECMO, the ICUCCI and ICU staffing parameters across different hospitals. The width of each plot at a given value reflects the data point density, whereas the central line represents the median value. Panel a depicts ICU resources (beds, physicians, and nurses) and the number of PC-ECMO implementations; Panel b shows staffing parameters and the ICUCCI, with each factor represented by a separate violin shape. The plot spans from the minimum to the maximum values, and outliers are displayed as individual points outside the whiskers
Characteristics of patients receiving PC-ECMO
Among the 2,601 patients included in this study, the median number of patients receiving PC-ECMO per hospital was 9 (IQR 3–24, range 1–324) across 102 hospitals. Most patients had a high baseline comorbidity risk at admission, with 72.3% (range 62–87.4% per hospital) of the patients having a van Walraven index ≥ 5. Among the cohort, 608 patients developed at least one of the nine predefined complications, resulting in a complication rate of 20.8% (range 0.5–33.3% per hospital). The three most common complications were acute kidney injury (AKI), systemic coagulation disorder, and bloodstream infection, which occurred in 232, 182, and 139 patients, respectively. At the hospital level, the median LOS for patients was 18.4 days (range 12.9–24.1 days), with a median hospitalization cost of $37,613 (range $30,131–$43,669) per patient. A total of 1,011 patients died during hospitalization, leading to a hospital-level in-hospital mortality rate of 47.2% (range 24.2–64.7%). For detailed patient characteristics, please refer to Table 2.
Table 2.
PC-ECMO Patient Characteristics
| Value | Median (Range) across hospital | |
|---|---|---|
| PC-ECMO patient | 2601 | 9 (1–324) |
| Characteristics of patient | ||
| Male | 1787 | 71.4% (62.8%-89.3%) |
| Female | 814 | 28.6% (10.7%-37.2%) |
| Adult | 2404 | 100.0% (97.4%-100%) |
| Child (under 18 years old) | 197 | 0.0% (0.0%-2.6%) |
| Emergency | 1327 | 64.2% (35.3%-81.0%) |
| Comorbidity, van Walraven Index van Walraven Index | ||
| < 0 points | 0 | 0.0% (0.0%-0.0%) |
| 0 points | 537 | 21.7% (9.6%-30.2%) |
| 1–4 points | 223 | 0.0% (0.0%-10.0%) |
| ≥ 5 points | 1841 | 72.3% (62.0%-87.4%) |
| Surgical type | ||
| Valve surgery | 709 | 11.0% (0.0%-33.3%) |
| Coronary artery surgery | 1109 | 37.5% (17.4%-63.5%) |
| Other surgeries | 783 | 35.7% (12.6%-55.4%) |
| Outcomes | ||
| Overall complications | 608 | 20.8% (0.5%-33.3%) |
| Acute kidney injury, AKI | 232 | 6.9% (0.0%-14.0%) |
| Hypoxic-ischemic encephalopathy | 98 | 0.0% (0.0%-3.8%) |
| Cerebral infarction | 62 | 0.0% (0.0%-1.3%) |
| Ischemia and arterial thrombosis | 61 | 0.0% (0.0%-3.2%) |
| Venous thrombosis | 70 | 0.0% (0.0%-3.3%) |
| Acute gastrointestinal disease | 27 | 0.0% (0.0%-0.0%) |
| Bloodstream Infection | 139 | 0.0% (0.0%-5.9%) |
| Intracerebral Hemorrhage | 42 | 0.0% (0.0%-0.0%) |
| Systemic Coagulation Disorder | 182 | 0.0% (0.0%-10.1%) |
| Length of Stay (LOS), day | / | 18.4 (12.9–24.1) |
| Hospitalization Cost, USD | / | 37,613 (30,131–43,669) |
| In-hospital mortality | 1011 | 47.2% (24.2%-64.7%) |
PC-ECMO Postcardiotomy-Extracorporeal Membrane Oxygenation
Associations between the ICUCCI and patient outcomes
In the overall cohort, the ICUCCI was found to be a protective factor against both in-hospital mortality (OR: 0.83, 95% CI: 0.70–0.97, P = 0.025) and complications (OR: 0.82, 95% CI: 0.68–0.99, P = 0.046). However, higher ICUCCI values were associated with longer hospital stays (IRR: 1.14, 95% CI: 1.06–1.22, P < 0.001) and increased hospitalization costs (IRR: 1.32, 95% CI: 1.24–1.40, P < 0.001) (Table 3a–d).
Table 3.
Association between ICU Capacity, ICU Staffing, and PC- ECMO Outcomes
| (a) | ||||||
|---|---|---|---|---|---|---|
| Patients | Odds | Unadjusted effects | p value | Adjusted effects | p value | |
| In-hospital mortality | 38.87% (1011/2601) | 0.636 | ||||
| ICUCCI | 0.748(0.638–0.877) | < 0.001 | 0.826(0.702–0.973) | 0.025 | ||
| PC-ECMO cases / ICU beds | 0.914(0.762–1.098) | 0.363 | 0.693(0.550–0.871) | 0.002 | ||
| PC-ECMO cases/ICU physicians | 0.849(0.719–1.002) | 0.058 | 0.712(0.586–0.864) | 0.001 | ||
| PC-ECMO cases/ICU nurses | 0.880(0.747–1.038) | 0.141 | 0.805(0.667–0.972) | 0.027 | ||
| (b) | ||||||
|---|---|---|---|---|---|---|
| Patients | Odds | Unadjusted effects | p value | Adjusted effects | p value | |
| Complication | 23.38% (608/2601) | 0.305 | ||||
| ICUCCI | 0.834(0.695–1.001) | 0.058 | 0.823(0.683–0.992) | 0.046 | ||
| PC-ECMO cases/ICU beds | 1.346(1.082–1.684) | 0.010 | 1.228(0.939–1.608) | 0.151 | ||
| PC-ECMO cases/ICU physicians | 1.084(0.894–1.317) | 0.439 | 0.980(0.783–1.227) | 0.909 | ||
| PC-ECMO cases/ICU nurses | 1.021(0.845–1.237) | 0.862 | 0.916(0.735–1.140) | 0.467 | ||
| (c) | ||||||
|---|---|---|---|---|---|---|
| Patients | Odds | Unadjusted effects | p value | Adjusted effects | p value | |
| LOS | – | – | ||||
| ICUCCI | 1.172(1.094–1.255) | < 0.001 | 1.138(1.062–1.220) | < 0.001 | ||
| PC-ECMO cases/ICU beds | 1.025(0.946–1.110) | 0.531 | 1.305(1.189–1.433) | < 0.001 | ||
| PC-ECMO cases/ICU physicians | 1.018(0.946–1.095) | 0.623 | 1.101(1.015–1.194) | 0.019 | ||
| PC-ECMO cases/ICU nurses | 1.001(0.931–1.076) | 0.968 | 1.097(1.012–1.190) | 0.023 | ||
| (d) | ||||||
|---|---|---|---|---|---|---|
| Patients | Odds | Unadjusted effects | p value | Adjusted effects | p value | |
| Hospital costs | – | – | ||||
| ICUCCI | 1.307(1.232–1.386) | < 0.001 | 1.319(1.241–1.402) | < 0.001 | ||
| PC-ECMO cases/ICU beds | 0.948(0.884–1.015) | 0.133 | 1.014(0.934–1.100) | 0.737 | ||
| PC-ECMO cases/ICU physicians | 0.965(0.906–1.027) | 0.272 | 0.974(0.908–1.045) | 0.469 | ||
| PC-ECMO cases/ICU nurses | 1.011(0.950–1.076) | 0.718 | 1.001(0.933–1.073) | 0.968 | ||
ICUCCI ICU Capacity Comprehensive Index, PC-ECMO Postcardiotomy-Extracorporeal Membrane Oxygenation, LOS length of stay
This table presents the relationships between exposure factors—ICUCCI and ICU staffing levels (ratios of patients receiving PC-ECMO to ICU beds, physicians, and nurses)—and four different outcomes: in-hospital mortality (a), complication occurrence (b), length of hospital stay (c), and hospitalization costs (d)
Subgroup analyses revealed a generally consistent protective effect or trend. For details, please refer to the Supplement (sFigure 3a, b).
Associations between ICU staffing and patient outcomes
Our analysis of ICU staffing indicators revealed a protective effect against in-hospital mortality. Specifically, higher ratios of patients receiving PC-ECMO to ICU beds (OR: 0.69, 95% CI: 0.55–0.87, P = 0.002), ICU physicians (OR: 0.71, 95% CI: 0.59–0.86, P = 0.001), and ICU nurses (OR: 0.81, 95% CI: 0.67–0.97, P = 0.027) were associated with reduced mortality risk. However, higher staffing levels were associated with longer hospital stays. Specifically, the ratios of patients receiving PC-ECMO to ICU beds (IRR: 1.31, 95% CI: 1.19–1.43, P < 0.001), ICU physicians (IRR: 1.10, 95% CI: 1.02–1.19, P = 0.019), and nurses (IRR: 1.10, 95% CI: 1.01–1.19, P = 0.023) were associated with an increased LOS. No clear correlation was found between staffing levels and overall complications or hospitalization costs. However, the ratio of patients receiving PC-ECMO to ICU beds was a risk factor for bloodstream infection (OR: 1.96, 95% CI: 1.14–3.46; P = 0.022) (Table 3 and sTable 6.b). Subgroup analyses revealed a generally consistent effect or trend of ICU staffing. However, in the subgroup that received in-hospital CPR, higher ratios of patients receiving PC-ECMO to ICU beds, ICU physicians, and ICU nurses were associated with increased complications. For details about the subgroup analyses, please refer to the Supplement (sFigure 4.a-f).
Discussion
Uneven distribution of ICU resources and the implementation of PC-ECMO
This study analyzed ICU capacity, ICU staffing, and their impacts on PC-ECMO outcomes across the Chinese hospitals included in our quality control survey. Our findings reinforce the uneven distribution of ICU resources and the implementation of ECMO, which is consistent with the finding of previous studies [20, 24]. However, unlike previous studies that focused on population or geographic density, we revealed these disparities at the institutional level. The number of ICU beds ranged from 16 to 616 per hospital, the number of ICU physicians ranged from 4 to 185, and the number of ICU nurses ranged from 4 to 687 per hospital. This underscores the need to balance resource allocation across healthcare institutions. Additionally, our study uniquely utilized the ICUCCI and specialized disease management staffing parameters, emphasizing deeper institutional differences.
Notably, the overall in-hospital mortality rate among patients receiving PC-ECMO in our cohort was 38.9%, which was significantly lower than the > 60% reported in the PELS-1 study and other studies [4–6]. Our data are reliable, as the HQMS is China’s national administrative platform where all hospitalized patient data are mandatorily uploaded under national regulations. This discrepancy may result from substantial variations in mortality rates across centers, as seen in the PELS-1 cohort (mortality rates ranging from 25 to 100%). Differences in patient selection, such as stricter ECMO eligibility criteria in China [24] and a lower complication rate in our cohort, may also contribute to the lower in-hospital mortality rate. However, as discussed in our nationwide ECMO cross-sectional study, limitations such as the lack of intraoperative variables, ECMO support duration, laboratory values, and process data (e.g., blood pressure and volume management) restricted our ability to fully interpret complications and mortality outcomes [11]. Furthermore, cultural factors such as the “dying at home” phenomenon may have contributed to an underestimation of in-hospital mortality, complicating direct comparisons with international studies [25].
Enhancing healthcare capacity and improving patient outcomes
The novel ICUCCI has been identified as a protective factor against both in-hospital mortality (OR: 0.83, 95% CI: 0.70–0.97, P = 0.025) and complications (OR: 0.82, 95% CI: 0.68–0.99, P = 0.046). Traditional methods typically rely on individual quality control indicators to assess ICU performance. These indicators include structural metrics such as the number of ICU beds or healthcare staff, process indicators such as Surviving Sepsis Campaign bundle compliance, and outcome indicators such as ICU mortality rates [11, 26–29]. However, these indicators have been criticized for being overly simplistic, lacking depth, and strongly correlating with hospital size [12]. As a result, they are considered inadequate for comprehensively assessing ICU capacity. Developed under the National Clinical Specialty Capacity Evaluation Program, the ICUCCI offers a holistic, multidimensional framework that integrates medical service capacity, technical proficiency, quality and safety standards, and service efficiency. These core evaluation dimensions align with the principles set by the Joint Commission on Accreditation of Healthcare Organizations (JCAHO) [30], a global leader in healthcare quality and patient safety, as well as the World Health Organization (WHO) Resolution WHA72.6 [31]. Moreover, as this indicator was developed under national administrative healthcare improvement program, it inherently possesses mandatory potential for nationwide implementation.
However, higher ICUCCI values were also associated with a longer LOS (IRR: 1.14, 95% CI: 1.06–1.22, P < 0.001) and increased hospitalization costs (IRR: 1.32, 95% CI: 1.24–1.40, P < 0.001). This could indicate the extended survival opportunities offered by hospitals with better capacity, which are equipped to manage complex patients for longer periods. Moreover, these hospitals may have conducted more intensive monitoring of patients (e.g., coagulation monitoring [32] and cerebral perfusion monitoring [33]), contributing to the higher costs observed. Thus, while improving the ICUCCI can increase patient survival, it also emphasizes the need for a balanced approach in resource allocation to prevent excessive consumption.
Optimizing PC-ECMO experience while maintaining quality control
In our analysis of ICU staffing, we identified a consistent protective effect of higher staffing levels against in-hospital mortality. Specifically, increased ratios of patients receiving PC-ECMO to ICU beds (OR: 0.69, 95% CI: 0.55–0.87, P = 0.002), ICU physicians (OR: 0.71, 95% CI: 0.59–0.86, P = 0.001), and ICU nurses (OR: 0.81, 95% CI: 0.67–0.97, P = 0.027) were significantly associated with reduced mortality risk. These findings align with those of previous studies emphasizing the critical role of adequate experience in optimizing patient outcomes [10, 11]. We also observed that higher staffing levels were associated with longer hospital stays, which could reflect extended survival opportunities.
Although staffing levels were not significantly correlated with overall complication rates or hospitalization costs, an interesting association emerged between the ratio of patients receiving PC-ECMO to ICU beds and the incidence of bloodstream infection. Regarding the statistical significance of the bed-related metric, we argued that the availability of ICU beds provided a more accurate measure than did the number of registered doctors and nurses, as the latter did not account for temporary staff adjustments or rotations. Although we did not further explore the time gap between catheter blood cultures and peripheral blood cultures to determine the proportion of catheter-related bloodstream infections [34], we believe that circulatory issues, rather than infections, are the primary concern for PC-ECMO patients. Thus, most bloodstream infections in this population were likely catheter related. This suggests that while higher staffing levels may increase survival rates, they could also increase the risk of infections, possibly due to heightened workloads or lapses in aseptic techniques. These findings highlight the need to balance patient load, medical staff experience, and process management.
Subgroup analyses generally supported the protective effect of staffing on mortality, with consistent trends despite limited statistical significance. However, among patients who received in-hospital CPR, higher staffing ratios were paradoxically associated with an increased risk of complications. This may be attributed to the complexity and resource-intensive nature of managing critically ill patients undergoing CPR, where increased staffing alone may lead to medical errors.
A key strength of our study is the use of staffing ratios as a measure of institutional expertise in specialized disease management and workload. While previous studies have relied on total case volume as a proxy for institutional experience [22], staffing ratios provide a more accurate reflection of workload and expertise [8]. By incorporating staffing metrics into our analysis, we gained deeper insights into the relationship between healthcare resources and patient care quality, particularly in high-acuity settings such as PC-ECMO management.
Limitations
First, unmeasured confounding factors, such as intraoperative conditions during cardiac surgery and the duration of ECMO support, were not accounted for in the analysis. These factors could significantly affect patient outcomes. Second, potential biases arising from the retrospective nature of data collection should be considered. However, our analysis demonstrated consistency across various subgroups, indicating that the results are robust (Supplementary sFigures 3 & 4). Third, the reliance on registered personnel data in this study may have underestimated the true workload of ICU staff, as it does not account for temporary staff adjustments. Fourth, defining the cohort by surgery and procedure date may have introduced discrepancies if ECMO support was initiated before surgery on the same day. However, these patients also reflect the capacity to manage patients requiring PC-ECMO. Furthermore, owing to the specific health administrative information system in China, the ICUCCI may not be directly applicable to healthcare systems in other countries. Nonetheless, the four key evaluation dimensions provide a framework that could be adapted for use in other healthcare systems or future international research. Future studies should prospectively explore ECMO management by investigating specific team sizes, qualifications, additional patient details, and developing related evaluation indicators.
Conclusion
In conclusion, our study underscores the significant role of ICU staffing levels and ICU capacity (as assessed through the ICUCCI) in influencing the outcomes of patients receiving PC-ECMO support. The findings indicate that enhancing ICU capacity, optimizing staffing ratios, and addressing regional disparities in healthcare resources have the potential to improve patient outcomes and foster greater equity in healthcare delivery across China.
Supplementary Information
Acknowledgements
The authors would like to thank all participants and staff in China National Critical Care Quality Control Center Group (China-NCCQC group) and China Critical Care Clinical Trials Group (CCCCTG). The China-NCCQC group consists of the following members: Yongjun Liu, Yan Kang, Jing Yan, Erzhen Chen, Bin Xiong, Bingyu Qin, Kejian Qian, Wei Fang, Mingyan Zhao, Xiaochun Ma, Xiangyou Yu, Jiandong Lin, Yi Yang, Feng Shen, Shusheng Li, Lina Zhang, Weidong Wu, Meili Duan, Linjun Wan, Xiaojun Yang, Jian Liu, Zhen Wang, Lei Xu, Zhenjie Hu, Congshan Yang, Longxiang Su, Xiang Zhou. The CCCCTG consists of the following members: Li Weng, Yan Chen, Shan Li, Jinmin Peng, Run Dong, Xiaoyun Hu, Wei Jiang, Chunyao Wang, Bin Du.
Abbreviations
- VA-ECMO
Veno-arterial extracorporeal membrane oxygenation
- PC-ECMO
Postcardiotomy extracorporeal membrane oxygenation
- ICU
Intensive care unit
- ICUCCI
ICU capacity comprehensive index
- LOS
Length of stay
- EQIA
ECMO quality improvement action
- HQMS
Hospital quality monitoring system
- RMB
Chinese yuan
- NCIS
National medical quality management and control information network
- CPR
Cardiopulmonary resuscitation
- ELSO
Extracorporeal life support organization
- ORs
Odds ratios
- IRRs
Incident rate ratios
- JCAHO
Joint commission on accreditation of healthcare organizations
- WHO
World health organization
Author contributions
XZ and TS designed the study. YHQ, XDM and ZL drafted the manuscript. WC, JQC, WP and ZYH carried out the data processing and statistical analysis. YJC, YWF, XDM and TS collected and provided the clinical data. LXS, XD, LW, YGL, HZJ and XZ reviewed the literature and revised the manuscript. All authors contributed to the article and approved the submitted version. All authors read and approved the final manuscript.
Funding
This work was supported by the Beijing Municipal Natural Science Foundation (L222019); CAMS Innovation Fund for Medical Sciences (CIFMS) (2024-I2M- C&T-C-002); National High Level Hospital Clinical Research Funding (2022-PUMCH-B-115&2022-PUMCH-D-005); and National Key R&D Program of China (2024YFF1207104).
Declarations
Ethics approval and consent to participate
This study was approved with a waiver of informed consent by the Ethics Committee of Peking Union Medical College Hospital (PUMCH, ethics number I-23PJ1416) as well as separate regulatory approvals at other sites.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Yehan Qiu, Xudong Ma and Zhe Li have contributed equally to this work.
Contributor Information
Ting Shu, Email: nctingting@126.com.
Xiang Zhou, Email: zx_pumc@163.com.
China Critical Care Clinical Trials Group (CCCCTG):
Li Weng, Yan Chen, Shan Li, Jinmin Peng, Run Dong, Xiaoyun Hu, Wei Jiang, Chunyao Wang, and Bin Du
China National Critical Care Quality Control Center Group (China-NCCQC group):
Yongjun Liu, Yan Kang, Jing Yan, Erzhen Chen, Bin Xiong, Bingyu Qin, Kejian Qian, Wei Fang, Mingyan Zhao, Xiaochun Ma, Xiangyou Yu, Jiandong Lin, Yi Yang, Feng Shen, Shusheng Li, Lina Zhang, Weidong Wu, Meili Duan, Linjun Wan, Xiaojun Yang, Jian Liu, Zhen Wang, Lei Xu, Zhenjie Hu, Congshan Yang, Longxiang Su, and Xiang Zhou
References
- 1.Sri D. Is cardiac surgery getting more complicated? Indian J Thorac Cardiovasc Surg. 2012;28(4):275–275. [Google Scholar]
- 2.Cleveland Clinic [Internet]. [cited 2025 Feb 10]. Cardiothoracic Surgical Outcomes: What Steady Improvements Add Up to Over Time. Available from: https://consultqd.clevelandclinic.org/cardiothoracic-surgical-outcomes-what-steady-improvements-add-up-to-over-time
- 3.Lorusso R, Whitman G, Milojevic M, Raffa G, McMullan DM, Boeken U, et al. 2020 EACTS/ELSO/STS/AATS expert consensus on post-cardiotomy extracorporeal life support in adult patients. J Thorac Cardiovasc Surg. 2021;161(4):1287–331. [DOI] [PubMed] [Google Scholar]
- 4.Biancari F, Dalén M, Fiore A, Ruggieri VG, Saeed D, Jónsson K, et al. Multicenter study on postcardiotomy venoarterial extracorporeal membrane oxygenation. J Thorac Cardiovasc Surg. 2020;159(5):1844-1854.e6. [DOI] [PubMed] [Google Scholar]
- 5.Biancari F, Perrotti A, Ruggieri VG, Mariscalco G, Dalén M, Dell’Aquila AM, et al. Five-year survival after post-cardiotomy veno-arterial extracorporeal membrane oxygenation. Eur Heart J Acute Cardiovasc Care. 2021;10(6):595–601. [DOI] [PubMed] [Google Scholar]
- 6.Mariani S, Heuts S, van Bussel BCT, Di Mauro M, Wiedemann D, Saeed D, et al. Patient and Management Variables Associated With Survival After Postcardiotomy Extracorporeal Membrane Oxygenation in Adults: The PELS-1 Multicenter Cohort Study. JAHA. 2023;12(14):e029609. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Koziol KJ, Isath A, Rao S, Gregory V, Ohira S, Van Diepen S, et al. Extracorporeal membrane oxygenation (VA-ECMO) in management of cardiogenic shock. JCM. 2023;12(17):5576. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Gershengorn HB, Harrison DA, Garland A, Wilcox ME, Rowan KM, Wunsch H. Association of intensive care unit patient-to-intensivist ratios with hospital mortality. JAMA Intern Med. 2017;177(3):388. [DOI] [PubMed] [Google Scholar]
- 9.Townsend SR, Phillips GS, Duseja R, Tefera L, Cruikshank D, Dickerson R, et al. Effects of compliance with the early management bundle (SEP-1) on mortality changes among medicare beneficiaries with sepsis. Chest. 2022;161(2):392–406. [DOI] [PubMed] [Google Scholar]
- 10.Barbaro RP, Odetola FO, Kidwell KM, Paden ML, Bartlett RH, Davis MM, et al. Association of hospital-level volume of extracorporeal membrane oxygenation cases and mortality analysis of the extracorporeal life support organization registry. Am J Respir Crit Care Med. 2015;191(8):894–901. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Cheng W, Chen J, Ma X, Sun J, Gao S, Wang Y, et al. Association between ICU quality and in-hospital mortality of V-V ECMO-supported patients—the ECMO quality improvement action (EQIA) study: a national cohort study in China from 2017 to 2019. Front Med. 2024;18(2):315–26. [DOI] [PubMed] [Google Scholar]
- 12.Verdonk F, Zacharowski K, Ahmed A, Orliaguet G, Pottecher J. A multifaceted approach to intensive care unit capacity. The Lancet Public Health. 2021;6(7): e448. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.He H, Ma X, Su L, Wang L, Guo Y, Shan G, et al. Effects of a national quality improvement program on ICUs in China: a controlled pre-post cohort study in 586 hospitals. Crit Care. 2020;24(1):73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Gao S, Yu Z, Ma X, Sun J, Ren A, Gao S, et al. Childhood-onset systemic lupus erythematosus in China, 2016–21: a nationwide study. Lancet Child & Adolescent Health. 2024;8(10):762–72. [DOI] [PubMed] [Google Scholar]
- 15.The General Office of the National Health Commission’s Notice on the Issuance of the National Clinical Specialty Capability Assessment Measures (Trial) [Internet]. [cited 2025 Feb 11]. Available from: http://www.nhc.gov.cn/yzygj/s7657/202403/890392a6cddb4c35aa7ad25e0392dbd1.shtml
- 16.Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi JC, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43(11):1130–9. [DOI] [PubMed] [Google Scholar]
- 17.van Walraven C, Austin PC, Jennings A, Quan H, Forster AJ. a modification of the elixhauser comorbidity measures into a point system for hospital death using administrative data. Med Care. 2009;47(6):626–33. [DOI] [PubMed] [Google Scholar]
- 18.Menendez ME, Neuhaus V, van Dijk NC, Ring D. The elixhauser comorbidity method outperforms the charlson index in predicting inpatient death after orthopaedic surgery. Clin Orthop Relat Res. 2014;472(9):2878–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Ni X, Li Z, Li X, Zhang X, Bai G, Liu Y, et al. Socioeconomic inequalities in cancer incidence and access to health services among children and adolescents in China: a cross-sectional study. The Lancet. 2022;400(10357):1020–32. [DOI] [PubMed] [Google Scholar]
- 20.Chen H, Xia Y, Qin Q, Cheng J, Xiong C. Spatial equity and factors associated with intensive care unit bed allocation in China. Arch Public Health. 2024;82(1):169. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Aiken LH, Simonetti M, Sloane DM, Cerón C, Soto P, Bravo D, et al. Hospital nurse staffing and patient outcomes in Chile: a multilevel cross-sectional study. Lancet Glob Health. 2021;9(8):e1145–53. [DOI] [PubMed] [Google Scholar]
- 22.Combes A, Brodie D, Bartlett R, Brochard L, Brower R, Conrad S, et al. position paper for the organization of extracorporeal membrane oxygenation programs for acute respiratory failure in adult patients. Am J Respir Crit Care Med. 2014;190(5):488–96. [DOI] [PubMed] [Google Scholar]
- 23.Fernando SM, MacLaren G, Barbaro RP, Mathew R, Munshi L, Madahar P, et al. Age and associated outcomes among patients receiving venoarterial extracorporeal membrane oxygenation–analysis of the extracorporeal life support organization registry. Intensive Care Med. 2023;49(12):1456–66. [DOI] [PubMed] [Google Scholar]
- 24.Cheng W, Ma XD, Su LX, He HW, Wang L, Tang B, et al. Cross-sectional study for the clinical application of extracorporeal membrane oxygenation in Mainland China, 2018. Crit Care. 2020;24(1):554. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Weng L, Hu Y, Sun Z, Yu C, Guo Y, Pei P, et al. Place of death and phenomenon of going home to die in Chinese adults: a prospective cohort study. Lancet Regional Health Western Pacific. 2022;18:100301. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Rui X, Dong F, Ma X, Su L, Shan G, Guo Y, et al. Quality metrics and outcomes among critically ill patients in China: results of the national clinical quality control indicators for critical care medicine survey 2015–2019. Chin Med J (Engl). 2022;135(9):1064–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Phua J, Kulkarni AP, Mizota T, Hashemian SMR, Lee WY, Permpikul C, et al. Critical care bed capacity in Asian countries and regions before and during the COVID-19 pandemic: an observational study. Lancet Regional Health Western Pacific. 2024;44:100982. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Wunsch H, Angus DC, Harrison DA, Collange O, Fowler R, Hoste EAJ, et al. Variation in critical care services across North America and Western Europe*. Crit Care Med. 2008;36(10):2787-e8. [DOI] [PubMed] [Google Scholar]
- 29.Arabi YM, Phua J, Koh Y, Du B, Faruq MO, Nishimura M, et al. Structure, organization, and delivery of critical care in Asian ICUs*. Crit Care Med. 2016;44(10):e940-8. [DOI] [PubMed] [Google Scholar]
- 30.Standards for Joint Commission Accreditation and Certification [Internet]. [cited 2025 Mar 27]. Available from: https://www.jointcommission.orghttps://www.jointcommission.org/standards/
- 31.Patient safety [Internet]. [cited 2025 Apr 3]. Available from: https://www.who.int/news-room/fact-sheets/detail/patient-safety
- 32.Pacheco-Reyes AF, Plata-Menchaca EP, Mera A, Bonilla C, Sosa M, Torrella P, et al. Coagulation management in patients requiring extracorporeal membrane oxygenation support: a comprehensive narrative review. Annals of Blood [Internet]. 2024;9:14. [Google Scholar]
- 33.Khanduja S, Kim J, Kang JK, Feng CY, Vogelsong MA, Geocadin RG, et al. Hypoxic-ischemic brain injury in ecmo: pathophysiology, neuromonitoring, and therapeutic opportunities. Cells. 2023;12(11):1546. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.O’Grady NP. Prevention of central line-associated bloodstream infections. N Engl J Med. 2023;389(12):1121–31. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.