Timing matters: association of persistent postoperative anemia with adverse outcomes in cardiac surgery Authors: Xidong Zhang, Yanhua Nie, Rui Zhang, Qiong Zhang, Xiaochen Wang Published: 2026 ABSTRACT ObjectiveTo investigate the dynamic impact of anemia at different perioperative timepoints (preoperative, postoperative day 1, 3, and 7) on adverse events in patients undergoing cardiac surgery.MethodsThis single-center retrospective cohort study included 315 adult patients who underwent cardiac surgery between January 2020 and June 2025. Anemia was defined according to World Health Organization (WHO) criteria. Baseline characteristics, perioperative indicators, and postoperative outcomes were compared between the preoperative anemia and non-anemia groups. Multivariable logistic regression models were used to assess the independent association of anemia at different timepoints with three pre‑specified primary outcomes: increased postoperative hospital stay, pulmonary infection, and low cardiac output syndrome (LCOS), after adjusting for potential confounders.ResultsResults: The prevalence of preoperative anemia was 48.6% (153/315). For the three pre‑specified primary outcomes, compared to the non‑anemia group, the preoperative anemia group had significantly higher rates of postoperative pulmonary infection (32.03% vs. 17.28%, p = 0.003) and increased postoperative hospital stays (median 10.0 [IQR 9.0–15.0] vs. 10.0 [7.0–12.0] days, p = 0.026). Multivariable analysis revealed that anemia on postoperative day 7 was an independent risk factor for increased postoperative hospital stays (OR = 3.033, 95% CI: 1.547–5.946, p = 0.001) and pulmonary infection (OR = 2.526, 95% CI: 1.235–5.167, p = 0.011). Furthermore, anemia on postoperative day 7 was a strong independent predictor of low cardiac output syndrome (LCOS) (OR = 8.494, 95% CI: 2.316–31.156, p = 0.001), whereas anemia on postoperative day 1 was associated with a lower risk of LCOS (OR = 0.146, 95% CI: 0.041–0.517, p = 0.003). RBC transfusion was associated with a reduced risk of adverse outcomes.ConclusionPostoperative anemia, particularly persistent anemia on postoperative day 7, is an independent risk factor for increased postoperative hospital stays, pulmonary infection, and LCOS in cardiac surgery patients. Dynamic monitoring of postoperative hemoglobin levels is crucial for risk stratification and individualized blood management. However, the modest sample size (n = 315) limits the precision of estimates for infrequent outcomes, and these findings therefore require validation in larger studies. Introduction Despite advances in surgical techniques and perioperative care, complications after cardiac surgery remain common and significantly affect patient outcomes [1, 2]. Anemia, a prevalent and modifiable risk factor, reduces blood oxygen-carrying capacity. During the high-demand stress of cardiac surgery, anemia can impair tissue and myocardial oxygen delivery, potentially exacerbating ischemia, compromising wound healing, and increasing cardiac workload [3, 4]. Nevertheless, existing evidence has important limitations. Most studies treat anemia as a static baseline condition, focusing on preoperative hemoglobin while overlooking its dynamic postoperative evolution [1]. Surgical blood loss, hemodilution, and inflammation drive significant hemoglobin declines, yet recovery trajectories vary widely, and the impact of anemia at different postoperative time points remains understudied. Moreover, associations between anemia and specific complications, such as low cardiac output syndrome (LCOS) and pulmonary infections, are often examined using univariate analyses without adequate confounder adjustment. This limits the ability to isolate anemia’s independent contribution [5]. The complex interplay between postoperative anemia and interventions such as red blood cell transfusion also requires further clarification. To address these gaps, this single‑center retrospective cohort study systematically investigated the independent impact of anemia assessed preoperatively and on postoperative days 1, 3, and 7 on adverse outcomes, including increased postoperative hospital stay, pulmonary infection, and low cardiac output syndrome. We also evaluated the role of red blood cell transfusion within this context, aiming to provide higher‑level evidence for developing more precise perioperative blood management strategies and for identifying modifiable risk factors to improve patient outcomes. Materials and methods Study design This was a single-center retrospective cohort study designed to investigate the impact of anemia status on adverse perioperative events in patients undergoing cardiac surgery. Study population Between January 2020 and June 2025, a total of 1352 adult cardiac surgery procedures were performed by the consistent operative team that served as the source population for this study. To establish a homogeneous cohort focused on evaluating the impact of anemia, we excluded patients who underwent minimally invasive surgical approaches or had specific complex cardiac pathologies, including congenital heart disease. From this candidate pool, 315 consecutive patients meeting all predefined inclusion and exclusion criteria were enrolled. By restricting the analysis to a single surgical team, we minimized variations in surgical technique and perioperative management, thereby enhancing the internal validity of our findings related to anemia. The inclusion criteria were as follows: (1) Age ≥ 18 years; (2) Undergoing cardiac surgery with or without cardiopulmonary bypass (including coronary artery bypass grafting(CABG), valve replacement, combined CABG and valve surgery, and transcatheter valve procedures); (3) Complete clinical data. The exclusion criteria were as follows: (1) Comorbid hematologic diseases, active bleeding, or coagulation dysfunction; (2) Undergoing emergency surgery; (3) Significant missing clinical data; (4) Patients undergoing transfemoral transcatheter aortic valve replacement (TF-TAVR); (5) Patients who underwent cardiac surgery via minimally invasive approaches (e.g., right anterior thoracotomy, left thoracotomy, or upper hemisternotomy); (6) Surgery for specific complex conditions including congenital heart disease, aortic dissection, or active infective endocarditis. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies. Grouping criteria Patients were divided into two groups based on preoperative hemoglobin (Hb) levels: the preoperative anemia group (n = 153) and the preoperative non-anemia group (n = 162). Anemia was defined according to World Health Organization (WHO) criteria: Hb < 130 g/L for men and Hb < 120 g/L for women [6, 7]. For postoperative time points, anemia was defined using a threshold commonly applied in cardiac surgery patients (Hb < 100 g/L), which accounts for acute hemodilution and surgical blood loss [8]. This differs from the preoperative definition, which followed the WHO criteria [6]. Data collection Patient baseline characteristics, perioperative metrics, and postoperative outcomes were collected from the electronic medical record system. Baseline data included age, sex, body mass index (BMI), smoking history, alcohol history, comorbidities (hypertension, diabetes, chronic kidney injury (CKD), fatty liver, chronic pneumonia, history of stroke), and cardiac function classification (NYHA). Perioperative metrics included surgical type, intraoperative blood loss, anesthesia duration, duration of mechanical ventilation, postoperative drainage volume, Hb levels on postoperative days 1, 3, and 7, volume of red blood cell transfusion, intensive care unit (ICU) length of stay, postoperative hospital stay, and complications (pulmonary infection, LCOS, cardio-cerebrovascular events, surgical site infection, 30-day readmission, death, etc.). The specific etiologies of preoperative anemia (e.g., iron, vitamin B12, or folate deficiency; anemia of chronic kidney disease) were not routinely and systematically screened for or collected as part of this retrospective study design. Statistical analysis Descriptive and univariate analyses Normally distributed continuous variables are presented as mean ± standard deviation and were compared using the Student’s t-test. Non-normally distributed continuous variables are presented as median (interquartile range, IQR) and were compared using the Mann-Whitney U test. Categorical variables are presented as frequency (percentage) and were compared using the χ² test. Multivariable regression and covariates Multivariate logistic regression analysis was employed to assess the association between anemia and postoperative adverse events, adjusting for covariates including age, sex, BMI, smoking history, alcohol history, comorbidities, cardiac function, surgical type, intraoperative blood loss, postoperative drainage volume, and red blood cell transfusion volume. A p-value < 0.05 was considered statistically significant. All analyses were performed using SPSS software version 27.0.Sample size considerations. Sample size considerations No a priori sample size calculation was performed due to the retrospective design. Precision of effect estimates was evaluated using 95% confidence interval widths (see Limitations). Model diagnostics and overfitting assessment Model fit was assessed using the Hosmer‑Lemeshow test (p > 0.05 indicating good fit) and discrimination using the AUC. To avoid overfitting, we required at least 10 events per variable (EPV) for primary analyses. For pulmonary infection (77 events) and prolonged hospital stay (157 events), all prespecified covariates were included (EPV > 10). For LCOS (30 events), we used a reduced model (age, sex, BMI, surgery type, RBC transfusion, POD7 anemia) achieving EPV ≈ 5, acceptable for exploratory analysis. All models converged. Ethics approval This study was approved by the Ethics Committee of the First Affiliated Hospital of USTC (Anhui Provincial Hospital). The approval number is [2025-RE-367]. Definition of outcome measures Postoperative pulmonary infection Postoperative pulmonary infection was defined by the presence of at least two of the following criteria: clinical signs/symptoms (e.g., new or worsening cough with fever ≥ 38.0 °C or pulmonary crackles), elevated white blood cell count (> 10 × 10⁹/L), new or progressive infiltrates on chest imaging, or identification of a pathogenic organism in respiratory cultures. Discharge decision process The decision to discharge a patient following cardiac surgery at our institution was based on a comprehensive assessment of medical stability alongside relevant social and logistical factors. Medical readiness was determined by the following criteria: stable vital signs without vasoactive support, no or only low-flow supplemental oxygen requirement, well-healed surgical sites without active infection or bleeding, stabilization of key postoperative laboratory and imaging findings (including electrocardiogram, echocardiogram, chest X-ray, and blood work), and the patient’s ability to eat independently and ambulate without major complications. Non-medical considerations, such as post-operative psychological/nutritional status, home support adequacy, and community resource coordination, were also integrated and could influence the discharge timeline. Red blood cell transfusion practice During the study period, there was no single, protocol-driven hemoglobin threshold for red blood cell (RBC) transfusion. Decisions were made individually by the clinical team. In stable postoperative patients, a hemoglobin level below 80 g/L was a common consideration for transfusion, while a level above 100 g/L in the absence of active bleeding or ischemia typically precluded it. The final decision integrated this value with the patient’s clinical trajectory, hemodynamic status, and evidence of oxygen delivery compromise. This reflects real-world practice but introduces the potential for indication bias. Study outcomes and prioritization This study had three primary outcomes, selected for their clinical relevance and expected association with perioperative anemia: increased postoperative hospital stays (exceeding the cohort median of 10 days), postoperative pulmonary infection (defined in Section "Postoperative pulmonary infection" ), and perioperative low cardiac output syndrome (LCOS) [9]. Secondary outcomes included ICU length of stay (hours), cardio-cerebrovascular adverse events (composite of stroke, myocardial infarction, or cardiac arrest), surgical site infection, 30-day readmission, and in-hospital mortality. All outcomes were pre-specified in the ethics committee-approved study protocol. Given the exploratory nature of this retrospective cohort, the primary analyses are hypothesis-generating, and the findings should be interpreted accordingly. Results Patient selection A total of 1,352 patients who underwent cardiac surgery by the same surgical team between January 2020 and June 2025 were screened. After applying the exclusion criteria sequentially, 315 patients were included in the final analysis. The patient selection process is illustrated in Fig. 1. Fig. 1Patient selection flow diagram Patient selection flow diagram Comparison of patient baseline characteristics This study included 315 patients who underwent cardiac surgery. Among them, 153 patients (48.6%) were in the preoperative anemia group and 162 patients (51.4%) were in the preoperative non-anemia group. Comparison of baseline characteristics between the two groups (Table 1) showed that patients in the preoperative anemia group were significantly older than those in the non-anemia group (66.17 ± 10.74 years vs. 61.99 ± 9.56 years, p = 0.0003) and had a significantly lower body mass index (BMI) (23.07 ± 3.23 kg/m² vs. 24.76 ± 3.37 kg/m², p < 0.001). Furthermore, the proportion of male patients was higher in the non-anemia group compared to the anemia group (67.90% vs. 57.52%, p = 0.057), a difference which only showed borderline significance (p = 0.057). The distribution of other baseline comorbidities, including smoking history, alcohol history, hypertension, diabetes, chronic kidney injury, fatty liver, chronic pneumonia, history of cerebral infarction, and NYHA functional class, showed no statistically significant differences between the two groups (all p > 0.05), indicating comparability between the groups regarding other potential confounding baseline factors. As expected by the grouping criteria, the preoperative hemoglobin (Hb) level was significantly lower in the anemia group compared to the non-anemia group (116.20 ± 13.31 g/L vs. 139.07 ± 9.87 g/L, p < 0.001), confirming the appropriateness of the grouping. Table 1Comparison of patient baseline characteristicsCharacteristicOverall (n = 315)Preoperative anemia group (n = 153)Preoperative non-anemia group (n = 162)Statisticp -valueAge (Mean ± SD)64.02 ± 10.7766.17 ± 10.7461.99 ± 9.56t = 3.660.0003Male, n (%)198(62.86%)88(57.52%)110(67.90%)χ²=3.630.057Body mass index (kg/m²) (Mean ± SD)24.13 ± 4.3023.07 ± 3.2324.76 ± 3.37t = 4.57< 0.001Smoking history, n (%)63(20.0%)25(16.34%)38(23.46%)χ²=2.490.115Alcohol history, n (%)50( 15.87%)22(14.38%)28(17.28%)χ²=0.510.473Hypertension, n (%)142(45.08%)72(47.06%)70(43.21%)χ²=0.490.485Diabetes, n (%)47(14.92%)21(13.73%)26(16.05%)χ²=0.330.567Chronic kidney injury, n (%)90(28.57%)50(32.68%)40(24.69%)χ²=2.460.117Fatty liver, n (%)67(21.27%)26(16.99%)41(25.31%)χ²=3.270.071Chronic pneumonia, n (%)150(47.62%)73(47.71%)77(47.53%)χ²=0.000.974History of cerebral infarction, n (%)174(55.24%)85(55.56%)89(54.94%)χ²=0.020.902NYHA classification, n (%)Class Ⅲ217(68.89%)106(69.74%)111(68.52%)χ²=0.020.880ClassⅣ97(31.11%)46(30.26%)51(31.48%)χ²=0.070.788Preoperative Hb (g/L) (Mean ± SD)125.05 ± 18.57116.20 ± 13.31139.07 ± 9.87t = 21.95< 0.001 Table 1Comparison of patient baseline characteristicsCharacteristicOverall (n = 315)Preoperative anemia group (n = 153)Preoperative non-anemia group (n = 162)Statisticp -valueAge (Mean ± SD)64.02 ± 10.7766.17 ± 10.7461.99 ± 9.56t = 3.660.0003Male, n (%)198(62.86%)88(57.52%)110(67.90%)χ²=3.630.057Body mass index (kg/m²) (Mean ± SD)24.13 ± 4.3023.07 ± 3.2324.76 ± 3.37t = 4.57< 0.001Smoking history, n (%)63(20.0%)25(16.34%)38(23.46%)χ²=2.490.115Alcohol history, n (%)50( 15.87%)22(14.38%)28(17.28%)χ²=0.510.473Hypertension, n (%)142(45.08%)72(47.06%)70(43.21%)χ²=0.490.485Diabetes, n (%)47(14.92%)21(13.73%)26(16.05%)χ²=0.330.567Chronic kidney injury, n (%)90(28.57%)50(32.68%)40(24.69%)χ²=2.460.117Fatty liver, n (%)67(21.27%)26(16.99%)41(25.31%)χ²=3.270.071Chronic pneumonia, n (%)150(47.62%)73(47.71%)77(47.53%)χ²=0.000.974History of cerebral infarction, n (%)174(55.24%)85(55.56%)89(54.94%)χ²=0.020.902NYHA classification, n (%)Class Ⅲ217(68.89%)106(69.74%)111(68.52%)χ²=0.020.880ClassⅣ97(31.11%)46(30.26%)51(31.48%)χ²=0.070.788Preoperative Hb (g/L) (Mean ± SD)125.05 ± 18.57116.20 ± 13.31139.07 ± 9.87t = 21.95< 0.001 Comparison of patient baseline characteristics Comparison of perioperative patient metrics A comparison of perioperative metrics is presented in Table 2. The distribution of surgical types, including isolated valve procedures, combined surgeries, off-pump CABG, and transcatheter interventions, did not differ significantly between the preoperative anemia and non-anemia groups (all p > 0.05). As anticipated and central to the study design, hemoglobin levels on postoperative days 1, 3, and 7 remained significantly lower in the preoperative anemia group compared to the non-anemia group (all p < 0.001). Analysis of procedural and postoperative course metrics using non-parametric tests (reported as median [IQR]) revealed the following: There were no significant between-group differences in intraoperative blood loss (200.0 [150–300] vs. 200.0 [200–300] ml, p = 0.107), anesthesia duration (325 [265–410] vs. 335 [300–435] min, p = 0.156), duration of mechanical ventilation (1 [1–1] vs. 1 [1–2] days, p = 0.060), postoperative day 1 drainage volume (350 [150–550] vs. 350 [200–500] ml, p = 0.469), or the volume of red blood cells transfused (4.0 [3.0–8.0] vs. 4.0 [3.0–6.0] units, p = 0.084). Significant differences were observed in key clinical outcomes. The preoperative anemia group had a substantially longer ICU length of stay (48[24–86] hours vs. 24 [24–72] hours, p < 0.001) and postoperative hospital stay (10.0 [9.0–15.0] days vs. 10.0 [7.0–12.0] days, p = 0.026). The incidence of postoperative pulmonary infection was significantly higher in the anemia group (32.03% vs. 17.28%, p = 0.003). Cardio-cerebrovascular adverse events occurred exclusively in the anemia group (4 [2.61%] vs. 0 [0%], p = 0.039). No significant differences were found between the groups in the rates of perioperative LCOS, surgical site infection, 30-day readmission, or mortality (all p > 0.05). Table 2Comparison of perioperative characteristics and outcomesMetricOverall (n = 315)Preoperative anemia group (n = 153)Preoperative non-anemia group (n = 162)Statisticp -valueSingle valve replacement (Aortic/mitral)66 (20.95%)28 (18.30%)38 (23.46%)χ²=1.320.250Aortic valve replacement + mitral valve replacement62 (19.68%)30 (19.61%)32 (19.75%)χ²=0.000.970Off-pump coronary artery bypass grafting67 (21.27%)30 (19.61%)37 (22.84%)χ²=0.530.468Off-pump CABG + valve replacement64 (20.32%)32 (20.92%)32 (19.75%)χ²=0.080.780TAVR71 (22.54%)33 (21.57%)38 (23.46%)χ²=0.160.686Intraoperative blood loss (ml), median (IQR)200.0[200 ~ 300]200.0[150 ~ 300]200.0[200 ~ 300]Z = − 1.6140.107Anesthesia duration (min), median (IQR)330[290 ~ 435]325 [265 ~ 410]335 [300 ~ 435]Z=−1.410.156Duration of mechanical ventilation (days), median (IQR)1 [1 ~ 1]1 [1 ~ 1]1 [1 ~ 2]Z=−1.890.060Postoperative day 1 drainage volume (ml), median (IQR)350 [200 ~ 550]350 [150 ~ 550]350 [200 ~ 500]Z=−0.720.469Postop day 1 Hb (g/L) (mean ± SD)107.51 ± 19.84100.40 ± 16.28114.23 ± 20.59t=6.59< 0.001Postop day 3 Hb (g/L) (mean ± SD)101.12 ± 18.3096.82 ± 17.08105.22 ± 18.52t=4.17< 0.001Postop day 7 Hb (g/L) (Mean ± SD)102.55 ± 16.1399.05 ± 14.41105.89 ± 17.00t=3.83< 0.001ICU length of stay (hours), median (IQR)24 [24 ~ 72]48 [24 ~ 86]24 [24 ~ 72]Z=−4.83< 0.001RBC transfusion volume (units), median (IQR)4.0 [3.0 ~ 6.0]4.0 [3.0 ~ 8.0]4.0 [3.0 ~ 6.0]Z=−1.730.084Postoperative pulmonary infection, n (%)77(24.44%)49(32.03%)28(17.28%)χ²=9.070.003Perioperative low cardiac output, n (%)30(9.52%)17(11.11%)13(8.02%)χ²=0.880.348Postoperative hospital stay (days), median (IQR)10.0 [8.0 ~ 14.0]10.0 [9.0 ~ 15.0]10.0 [7.0 ~ 12.0]Z=−2.220.026Cardio-cerebrovascular adverse events, n (%)4 (1.27%)4 (2.61%)0 (0.00%)χ²=4.280.039Surgical site infection/poor healing, n (%)16(5.08%)8(5.23%)8(4.94%)χ²=0.010.92030-day readmission, n (%)10(3.17%)6(3.92%)4(2.47%)χ²=0.620.431Mortality, n (%)22(6.98%)10(6.54%)12(7.41%)χ²=0.100.752TAVR, Transcatheter Aortic Valve Replacement; ICU, Intensive Care Unit; RBC, Red Blood Cell. *p-values for variables presented as median [IQR] were calculated using the Mann-Whitney U test; p-values for variables presented as mean ± SD were calculated using Student’s t-test. Categorical variables were compared using the χ² test Table 2Comparison of perioperative characteristics and outcomesMetricOverall (n = 315)Preoperative anemia group (n = 153)Preoperative non-anemia group (n = 162)Statisticp -valueSingle valve replacement (Aortic/mitral)66 (20.95%)28 (18.30%)38 (23.46%)χ²=1.320.250Aortic valve replacement + mitral valve replacement62 (19.68%)30 (19.61%)32 (19.75%)χ²=0.000.970Off-pump coronary artery bypass grafting67 (21.27%)30 (19.61%)37 (22.84%)χ²=0.530.468Off-pump CABG + valve replacement64 (20.32%)32 (20.92%)32 (19.75%)χ²=0.080.780TAVR71 (22.54%)33 (21.57%)38 (23.46%)χ²=0.160.686Intraoperative blood loss (ml), median (IQR)200.0[200 ~ 300]200.0[150 ~ 300]200.0[200 ~ 300]Z = − 1.6140.107Anesthesia duration (min), median (IQR)330[290 ~ 435]325 [265 ~ 410]335 [300 ~ 435]Z=−1.410.156Duration of mechanical ventilation (days), median (IQR)1 [1 ~ 1]1 [1 ~ 1]1 [1 ~ 2]Z=−1.890.060Postoperative day 1 drainage volume (ml), median (IQR)350 [200 ~ 550]350 [150 ~ 550]350 [200 ~ 500]Z=−0.720.469Postop day 1 Hb (g/L) (mean ± SD)107.51 ± 19.84100.40 ± 16.28114.23 ± 20.59t=6.59< 0.001Postop day 3 Hb (g/L) (mean ± SD)101.12 ± 18.3096.82 ± 17.08105.22 ± 18.52t=4.17< 0.001Postop day 7 Hb (g/L) (Mean ± SD)102.55 ± 16.1399.05 ± 14.41105.89 ± 17.00t=3.83< 0.001ICU length of stay (hours), median (IQR)24 [24 ~ 72]48 [24 ~ 86]24 [24 ~ 72]Z=−4.83< 0.001RBC transfusion volume (units), median (IQR)4.0 [3.0 ~ 6.0]4.0 [3.0 ~ 8.0]4.0 [3.0 ~ 6.0]Z=−1.730.084Postoperative pulmonary infection, n (%)77(24.44%)49(32.03%)28(17.28%)χ²=9.070.003Perioperative low cardiac output, n (%)30(9.52%)17(11.11%)13(8.02%)χ²=0.880.348Postoperative hospital stay (days), median (IQR)10.0 [8.0 ~ 14.0]10.0 [9.0 ~ 15.0]10.0 [7.0 ~ 12.0]Z=−2.220.026Cardio-cerebrovascular adverse events, n (%)4 (1.27%)4 (2.61%)0 (0.00%)χ²=4.280.039Surgical site infection/poor healing, n (%)16(5.08%)8(5.23%)8(4.94%)χ²=0.010.92030-day readmission, n (%)10(3.17%)6(3.92%)4(2.47%)χ²=0.620.431Mortality, n (%)22(6.98%)10(6.54%)12(7.41%)χ²=0.100.752TAVR, Transcatheter Aortic Valve Replacement; ICU, Intensive Care Unit; RBC, Red Blood Cell. *p-values for variables presented as median [IQR] were calculated using the Mann-Whitney U test; p-values for variables presented as mean ± SD were calculated using Student’s t-test. Categorical variables were compared using the χ² test Comparison of perioperative characteristics and outcomes TAVR, Transcatheter Aortic Valve Replacement; ICU, Intensive Care Unit; RBC, Red Blood Cell. *p-values for variables presented as median [IQR] were calculated using the Mann-Whitney U test; p-values for variables presented as mean ± SD were calculated using Student’s t-test. Categorical variables were compared using the χ² test Multivariate logistic regression analysis of anemia and postoperative adverse events For analyses involving postoperative day 7 anemia, the temporal relationship with outcomes such as pulmonary infection and prolonged hospital stay may be bidirectional. Therefore, these findings are exploratory and hypothesis‑generating. Causal interpretations are limited to preoperative and day 1 anemia for early outcomes, for example LCOS. Multivariate logistic regression analyses were performed with increased postoperative hospital stay, pulmonary infection, and perioperative LCOS as the dependent variables, respectively. Because of the limited sample size and the small number of events for some outcomes, all multivariable analyses are exploratory and their results require validation in larger cohorts. Increased postoperative hospital stay The results of the analysis (Table 3) showed that after adjusting for multiple covariates including age, sex, BMI, comorbidities, surgical type, and drainage volume, anemia on postoperative day 7 remained an independent risk factor for increased postoperative hospital stay (OR = 3.033, 95% CI: 1.547–5.946, p = 0.001). This indicates that patients who were still anemic on postoperative day 7 had a 3.03 times higher risk of a longer hospital stay compared to non-anemic patients. In contrast, preoperative anemia, anemia on postoperative day 1, and anemia on postoperative day 3 did not demonstrate independent predictive value (all p > 0.05). Red blood cell transfusion was identified as a protective factor (OR = 0.870, 95% CI: 0.765–0.989, p = 0.033), suggesting that transfusion was associated with a reduced risk of increased postoperative hospital stays. Table 3Impact of anemia on increased postoperative hospital stayVariableBS.E.Walddfp-valueOR95% Confidence intervalSex (male vs. female)-0.2020.3450.34410.5570.8170.4151.606Body mass index (kg/m²)0.010.0490.04610.8311.010.9181.112Smoking history (Yes vs. No)-0.5040.4961.0310.310.6040.2281.599Alcohol history (Yes vs. No)0.4910.5630.75910.3841.6340.5424.928Hypertension (Yes vs. No)-0.2390.3210.55510.4560.7870.4191.478Diabetes (Yes vs. No)-0.4640.4081.29210.2560.6290.2831.399Chronic kidney disease (Yes vs. No)-0.4770.3372.00510.1570.6210.3211.201Fatty liver (Yes vs. No)0.240.4180.3310.5661.2710.5612.882Chronic pneumonia (Yes vs. No)0.2850.3150.81810.3661.3290.7172.464History of cerebral infarction (Yes vs. No)-0.0330.3210.0110.9190.9680.5161.814NYHA cardiac function class (IV vs. III)-0.1880.3350.31510.5750.8290.431.598Surgical type (ordinal variable)-0.1340.1131.410.2370.8750.7011.092Postoperative day 1 drainage volume (mL)-0.0030.0016.4310.0110.9970.9960.999Red blood cell transfusion volume (U)-0.140.0654.54510.0330.870.7650.989Preoperative anemia (Yes vs. No)-0.4840.3162.34410.1260.6160.3321.145Anemia on postoperative day 1 (Yes vs. No)-0.1750.3620.23310.630.840.4131.708Anemia on postoperative day 3 (Yes vs. No)0.2330.3720.39310.5311.2630.6092.62Anemia on postoperative day 7 (Yes vs. No)1.110.34310.44310.0013.0331.5475.946Surgical type was coded as an ordinal variable: 1 = Single Valve Replacement (Aortic/Mitral), 2 = Aortic Valve Replacement + Mitral Valve Replacement, 3 = Off-pump Coronary Artery Bypass Grafting, 4 = Off-pump CABG + Valve Replacement, 5 = TAVR (Transcatheter Aortic Valve Replacement) Table 3Impact of anemia on increased postoperative hospital stayVariableBS.E.Walddfp-valueOR95% Confidence intervalSex (male vs. female)-0.2020.3450.34410.5570.8170.4151.606Body mass index (kg/m²)0.010.0490.04610.8311.010.9181.112Smoking history (Yes vs. No)-0.5040.4961.0310.310.6040.2281.599Alcohol history (Yes vs. No)0.4910.5630.75910.3841.6340.5424.928Hypertension (Yes vs. No)-0.2390.3210.55510.4560.7870.4191.478Diabetes (Yes vs. No)-0.4640.4081.29210.2560.6290.2831.399Chronic kidney disease (Yes vs. No)-0.4770.3372.00510.1570.6210.3211.201Fatty liver (Yes vs. No)0.240.4180.3310.5661.2710.5612.882Chronic pneumonia (Yes vs. No)0.2850.3150.81810.3661.3290.7172.464History of cerebral infarction (Yes vs. No)-0.0330.3210.0110.9190.9680.5161.814NYHA cardiac function class (IV vs. III)-0.1880.3350.31510.5750.8290.431.598Surgical type (ordinal variable)-0.1340.1131.410.2370.8750.7011.092Postoperative day 1 drainage volume (mL)-0.0030.0016.4310.0110.9970.9960.999Red blood cell transfusion volume (U)-0.140.0654.54510.0330.870.7650.989Preoperative anemia (Yes vs. No)-0.4840.3162.34410.1260.6160.3321.145Anemia on postoperative day 1 (Yes vs. No)-0.1750.3620.23310.630.840.4131.708Anemia on postoperative day 3 (Yes vs. No)0.2330.3720.39310.5311.2630.6092.62Anemia on postoperative day 7 (Yes vs. No)1.110.34310.44310.0013.0331.5475.946Surgical type was coded as an ordinal variable: 1 = Single Valve Replacement (Aortic/Mitral), 2 = Aortic Valve Replacement + Mitral Valve Replacement, 3 = Off-pump Coronary Artery Bypass Grafting, 4 = Off-pump CABG + Valve Replacement, 5 = TAVR (Transcatheter Aortic Valve Replacement) Impact of anemia on increased postoperative hospital stay Surgical type was coded as an ordinal variable: 1 = Single Valve Replacement (Aortic/Mitral), 2 = Aortic Valve Replacement + Mitral Valve Replacement, 3 = Off-pump Coronary Artery Bypass Grafting, 4 = Off-pump CABG + Valve Replacement, 5 = TAVR (Transcatheter Aortic Valve Replacement) Postoperative pulmonary infection For postoperative pulmonary infection, the multivariate analysis again highlighted the independent risk posed by anemia on postoperative day 7 (OR = 2.526, 95% CI: 1.235–5.167, p = 0.011). Anemia at other time points showed no significant association with the occurrence of pulmonary infection. Perioperative LCOS Given the small number of LCOS events (n = 30), these findings should be considered hypothesis-generating and interpreted with caution. Nonetheless, the analysis revealed a dynamic risk pattern (Table 4). The analysis revealed a dynamic risk pattern: the presence of anemia on postoperative day 7 (OR = 8.494, 95% CI: 2.316–31.156, p = 0.001) was an extremely significant independent risk factor for perioperative LCOS. It is particularly noteworthy that anemia on postoperative day 7 increased the risk of LCOS by more than 8-fold. In contrast, anemia on postoperative day 1 (OR ≈ 0.146, 95% CI: 0.041–0.517, p = 0.003) was identified as a statistically significant protective factor. Correspondingly, red blood cell transfusion again demonstrated a strong protective effect (OR = 0.836, 95% CI: 0.748–0.935, p = 0.002). Preoperative anemia and anemia on postoperative day 3 showed no significant association with LCOS. Table 4Impact of anemia on the occurrence of perioperative LCOSVariableBS.E.Walddfp-valueOR95% Confidence intervalSex (male vs. female)0.7060.5881.44310.232.0260.646.411Body mass index (kg/m²)0.0560.0750.55710.4551.0580.9131.225Smoking history (Yes vs. No)-0.5980.7510.63410.4260.550.1262.398Alcohol history (Yes vs. No)0.0850.7650.01210.9111.0890.2434.882Hypertension (Yes vs. No)-0.3520.5280.44310.5060.7040.251.981Diabetes (Yes vs. No)-0.1340.6460.04310.8360.8750.2463.103Chronic kidney disease (Yes vs. No)-0.7830.5072.38910.1220.4570.1691.234Fatty liver (Yes vs. No)-1.3350.6364.40510.0360.2630.0760.916Chronic pneumonia (Yes vs. No)0.4780.5010.90910.341.6130.6044.309History of cerebral infarction (Yes vs. No)0.1430.5140.07710.7811.1540.4213.158NYHA cardiac function class (IV vs. III)-0.3560.5320.44910.5030.70.2471.985Surgical type (ordinal variable)0.0940.1840.26210.6091.0990.7661.577Postoperative day 1 drainage volume (mL)0.0000.0010.05410.81710.9981.003Red blood cell transfusion volume (U)-0.1790.0579.80610.0020.8360.7480.935Preoperative anemia (Yes vs. No)-0.0290.5150.00310.9550.9720.3542.663Anemia on postoperative day 1 (Yes vs. No)-1.9270.6478.87810.0030.1460.0410.517Anemia on postoperative day 3 (Yes vs. No)0.7760.6491.4310.2322.1720.6097.75Anemia on postoperative day 7 (Yes vs. No)2.1390.66310.4110.0018.4942.31631.156Surgical type was coded as an ordinal variable: 1 = Single Valve Replacement (Aortic/Mitral), 2 = Aortic Valve Replacement + Mitral Valve Replacement, 3 = Off-pump Coronary Artery Bypass Grafting, 4 = Off-pump CABG + Valve Replacement, 5 = TAVR (Transcatheter Aortic Valve Replacement) Table 4Impact of anemia on the occurrence of perioperative LCOSVariableBS.E.Walddfp-valueOR95% Confidence intervalSex (male vs. female)0.7060.5881.44310.232.0260.646.411Body mass index (kg/m²)0.0560.0750.55710.4551.0580.9131.225Smoking history (Yes vs. No)-0.5980.7510.63410.4260.550.1262.398Alcohol history (Yes vs. No)0.0850.7650.01210.9111.0890.2434.882Hypertension (Yes vs. No)-0.3520.5280.44310.5060.7040.251.981Diabetes (Yes vs. No)-0.1340.6460.04310.8360.8750.2463.103Chronic kidney disease (Yes vs. No)-0.7830.5072.38910.1220.4570.1691.234Fatty liver (Yes vs. No)-1.3350.6364.40510.0360.2630.0760.916Chronic pneumonia (Yes vs. No)0.4780.5010.90910.341.6130.6044.309History of cerebral infarction (Yes vs. No)0.1430.5140.07710.7811.1540.4213.158NYHA cardiac function class (IV vs. III)-0.3560.5320.44910.5030.70.2471.985Surgical type (ordinal variable)0.0940.1840.26210.6091.0990.7661.577Postoperative day 1 drainage volume (mL)0.0000.0010.05410.81710.9981.003Red blood cell transfusion volume (U)-0.1790.0579.80610.0020.8360.7480.935Preoperative anemia (Yes vs. No)-0.0290.5150.00310.9550.9720.3542.663Anemia on postoperative day 1 (Yes vs. No)-1.9270.6478.87810.0030.1460.0410.517Anemia on postoperative day 3 (Yes vs. No)0.7760.6491.4310.2322.1720.6097.75Anemia on postoperative day 7 (Yes vs. No)2.1390.66310.4110.0018.4942.31631.156Surgical type was coded as an ordinal variable: 1 = Single Valve Replacement (Aortic/Mitral), 2 = Aortic Valve Replacement + Mitral Valve Replacement, 3 = Off-pump Coronary Artery Bypass Grafting, 4 = Off-pump CABG + Valve Replacement, 5 = TAVR (Transcatheter Aortic Valve Replacement) Impact of anemia on the occurrence of perioperative LCOS Surgical type was coded as an ordinal variable: 1 = Single Valve Replacement (Aortic/Mitral), 2 = Aortic Valve Replacement + Mitral Valve Replacement, 3 = Off-pump Coronary Artery Bypass Grafting, 4 = Off-pump CABG + Valve Replacement, 5 = TAVR (Transcatheter Aortic Valve Replacement) Model diagnostics The Hosmer‑Lemeshow test indicated good fit for all primary outcome models: pulmonary infection (χ²=6.32, df = 8, p = 0.61), prolonged hospital stay (χ²=8.15, df = 8, p = 0.42), and the reduced LCOS model (χ²=4.21, df = 8, p = 0.75). The full LCOS model also showed acceptable fit (χ²=5.87, df = 8, p = 0.66). Discrimination, measured by the AUC, was acceptable to good: pulmonary infection (AUC = 0.79, 95% CI: 0.72–0.86), prolonged hospital stay (AUC = 0.74, 95% CI: 0.68–0.80), and reduced LCOS model (AUC = 0.85, 95% CI: 0.76–0.94). Nagelkerke R² values were 0.21, 0.18, and 0.33, respectively. All models converged without issues. The reduced LCOS model confirmed that anemia on postoperative day 7 remained a strong independent predictor (OR = 7.92, 95% CI: 2.04–30.68, p = 0.003), consistent with the full model. Other postoperative adverse events In the multivariate analysis, no statistically significant associations were found between anemia status (preoperative, postoperative day 1, day 3, or day 7) and cardio-cerebrovascular adverse events, surgical site infection/poor wound healing, or 30-day readmission (all p > 0.05). It is noteworthy that although cardio-cerebrovascular events occurred exclusively in the preoperative anemia group (4 cases vs. 0 cases) in the unadjusted analysis (Table 2, p = 0.039), this association did not remain statistically significant after adjusting for multiple covariates including age, sex, BMI, comorbidities, and perioperative factors, likely due to the limited number of events (n = 4) and the comprehensive adjustment for confounding variables. Discussion This single-center retrospective cohort study systematically investigated the impact of anemia status at different perioperative time points on the prognosis of patients undergoing cardiac surgery. Our principal findings can be summarized as follows: First, the incidence of preoperative anemia was high (48.6%), and the preoperative anemia group had consistently and significantly lower hemoglobin levels on postoperative days 1, 3, and 7 compared to the non-anemia group, forming a “persistent anemia exposure window.” Second, preoperative anemia was significantly associated with increased incidences of postoperative pulmonary infection and cardio-cerebrovascular events, as well as increased postoperative hospital stays. Most critically, through multivariate analysis, we demonstrated that dynamic postoperative anemia, particularly anemia still present on postoperative day 7, is a strong independent risk factor for increased postoperative hospital stays, pulmonary infection, and LCOS, surpassing the static perspective focusing solely on baseline preoperative anemia. Furthermore, this study also revealed the complexity surrounding red blood cell transfusion being associated with a reduced risk of some adverse outcomes. From static to dynamic: the independent predictive value and central role of persistent postoperative anemia The core contribution of this study lies in shifting the research focus from a single baseline preoperative anemia assessment to the dynamic evolution of perioperative anemia, and in establishing the prominent prognostic value of anemia on postoperative day 7. Previous studies have often regarded anemia as a static baseline risk factor [10, 11]. However, our data indicate that postoperative anemia is a common and persistent state, and its impact exhibits significant time-dependency. Anemia on postoperative day 1 is nearly universal, primarily reflecting the acute effects of surgical trauma and hemodilution. In contrast, the persistence of anemia by postoperative day 7 strongly predicts an adverse patient recovery trajectory. Anemia persisting at this time point may reflect impaired hematopoietic reserve, ongoing occult blood loss, or a persistent inflammatory state. These factors collectively compromise tissue oxygenation and repair capacity. Ultimately, they lead to hindered recovery and a significantly increased risk of complications [8, 12]. Therefore, the hemoglobin level on postoperative day 7 could serve as an easily obtainable and effective “vital sign” for identifying high-risk patients with delayed recovery. The complex association between anemia and complications: time-dependency and mechanistic insights An important methodological consideration is that the comparisons of perioperative metrics reported herein, including hospital stay and transfusion volume, have been analyzed using non-parametric methods. This approach ensures the most appropriate handling of skewed data and strengthens the validity of the associated findings. This study provides an in-depth revelation of the complex and dynamic association between anemia and specific complications, particularly regarding LCOS. The complex dynamic association with LCOS The most critical finding is that anemia on postoperative day 7 is a very strong independent predictor of LCOS (OR = 8.494). This result has a solid pathophysiological basis: during the postoperative period, the myocardium is in a critical phase of stunning and repair, being highly sensitive to and having increased demand for oxygen supply [13, 14]. Persistent anemia at the end of the first postoperative week implies a prolonged state of relative myocardial oxygen supply insufficiency, which may severely impede effective repair, thereby inducing or exacerbating cardiac dysfunction. A finding warranting discussion is that anemia on postoperative day 1 was associated with a lower risk of LCOS (OR = 0.146). This paradoxical association should be interpreted with caution. It may reflect survivor bias, because patients with early fatal LCOS were excluded from the analysis. It may also reflect treatment bias, as early anemia may trigger more aggressive hemodynamic support. Alternatively, it could simply be due to chance resulting from multiple comparisons. Thus, this finding is hypothesis‑generating and not ready for clinical translation. Importantly, this finding does not imply that early anemia itself is protective; rather, it likely reflects more aggressive supportive care (e.g., early inotropic support, volume expansion) provided to patients with early hemodynamic compromise. In contrast, anemia persisting to day 7 showed a strong, consistent association with LCOS, which has a plausible biological basis [9, 15]. These findings lead us to a critical physiological insight: the duration of postoperative anemia carries greater prognostic significance than its absolute depth. The hemoglobin nadir observed at postoperative day 3, while physiologically notable, predominantly reflects an acute and anticipated confluence of hemodilution and the initial inflammatory surge. In contrast, persistent anemia at day 7 signifies a fundamentally different clinical scenario—a derailed recovery trajectory. By the seventh postoperative day, the early dilutional phase should typically be resolving. Consequently, the failure of hemoglobin to initiate a meaningful recovery points to sustained underlying pathologies, such as a persistent inflammatory milieu or inadequate hematopoietic reserve [16]. Thus, anemia at day 7 emerges as a dynamic biomarker of a patient’s physiological reserve and ongoing stress response [17]. Association with pulmonary infection and postoperative hospital stay The results of this study clearly show that both preoperative anemia and anemia on postoperative day 7 are independently associated with an increased risk of pulmonary infection. Anemia-induced tissue hypoxia impairs the function of alveolar macrophages and neutrophils, directly compromising the immune defense mechanisms of the respiratory tract [18]. A persistent state of postoperative tissue hypoxia makes patients more susceptible to nosocomial pneumonia, thereby delaying recovery. Furthermore, it should be noted that postoperative pulmonary infection in this study was defined as a comprehensive clinical diagnosis (see Section "Postoperative pulmonary infection"), and its incidence (24.44%) reflects the overall infection risk faced by patients after cardiac surgery. During the study period, a standard ventilator‑associated pneumonia prevention bundle was implemented for mechanically ventilated patients in our ICU as part of standard care. Against this backdrop of standardized, proactive prophylaxis, persistent anemia on postoperative day 7 remained a strong independent risk factor for pulmonary infection, a finding of particular significance. This suggests that anemia-induced tissue hypoxia and immunocompromise constitute a more fundamental, systemic pathway of infection susceptibility that extends beyond localized airway management [19]. Similarly, anemia on postoperative day 7, as an independent risk factor for increased postoperative hospital stays, comprehensively reflects the patient’s overall slow recovery speed. It indicates that the patient may be facing multiple challenges simultaneously, such as poor recovery of cardiac function, difficulty in controlling infection, and poor activity tolerance, serving as a comprehensive marker of hindered recovery [20]. Notably, our analysis revealed that patients with preoperative anemia experienced significantly longer stays in both the intensive care unit and the general ward. This pattern suggests that anemia is associated with a more complex and protracted recovery trajectory throughout the entire postoperative course, extending beyond the initial critical care phase. The higher rate (24.4%) reflects our sensitive diagnostic criteria and high-risk patient profile (e.g., older age, anemia, chronic lung disease). Red blood cell transfusion The observed association between RBC transfusion and reduced adverse outcomes should not be overinterpreted. This is a retrospective study with clear indication bias: sicker patients with more severe anemia are more likely to receive transfusions [12]. Despite adjustment, residual confounding likely persists. Thus, the apparent protective effect of transfusion is likely spurious or reflects unmeasured disease severity [21, 22]. Therefore, these results should not be interpreted as evidence that RBC transfusion provides therapeutic protection; they merely reflect the greater intensity of care directed toward sicker patients. No causal inference can be drawn; these findings are hypothesis‑generating and require prospective validation. Study limitations This study has several limitations. First, its single‑center, single‑team retrospective design limits generalizability and invites selection/information bias. Second, it lacked a pre‑registered analysis plan; to mitigate bias, outcomes and covariates were defined before data extraction. Third, the sample size (n = 315) was adequate for pulmonary infection but insufficient for less frequent events. LCOS had only 30 events (wide CI, OR 8.49 hypothesis‑generating); we used a reduced model (EPV ≈ 5) that needs validation. Even rarer outcomes (n = 4, n = 22) were underpowered. Fourth, multiple comparisons across time points increase the risk of chance findings; non‑primary and paradoxical associations are therefore hypothesis‑generating. Fifth, residual confounding (e.g., nutritional status) may persist. Detailed pulmonary function data were unavailable, and CPB time was not applicable to off‑pump or TAVR cases; anesthesia duration was used as a surrogate. Sixth, we lacked inflammatory biomarkers (CRP, IL‑6) and iron studies, so we could not distinguish inflammation‑driven anemia from iron deficiency; this limits mechanistic interpretation of persistent POD7 anemia as a biomarker of derailed recovery. Seventh, preoperative anemia etiology was not systematically assessed; observed anemia likely reflects chronic disease or heart failure, so associations may partly reflect underlying burden rather than low hemoglobin alone. Finally, the absence of a standardized transfusion protocol introduced indication bias, complicating interpretation of transfusion‑outcome relationships. Prospective, multicenter studies with pre‑registered protocols and standardized thresholds are needed. Conclusions The results of this study indicate that for patients undergoing cardiac surgery, the management strategy for perioperative anemia should transition from a single preoperative static assessment to dynamic monitoring and intervention throughout the entire perioperative period. Anemia persisting on postoperative day 7 was confirmed as a strong indicator for predicting increased postoperative hospital stay, pulmonary infection, and LCOS. It is recommended to consider this time point as a key node for postoperative risk stratification and clinical decision-making. In clinical practice, close attention should be paid to the dynamic trajectory of postoperative hemoglobin levels, and patients with persistent anemia should receive closer monitoring and comprehensive management. Future large-scale, prospective studies, particularly multicenter cohorts or randomized controlled trials, are needed to further validate the conclusions of this study and to deeply explore the effectiveness and optimal timing of non-transfusion treatment strategies for postoperative anemia.