Introduction
Unfractionated heparin (UFH) remains a mainstay of acute anticoagulation in acute coronary syndromes (ACS) owing to its rapid onset, short half-life and reversibility with protamine sulphate [1,2]. Unlike oral anticoagulants that dominate long-term secondary prevention, UFH requires real-time laboratory monitoring to balance ischaemic and haemorrhagic risks [3]. For decades, the activated partial thromboplastin time (aPTT) has served as the default monitoring tool, with institutional nomograms directing dose adjustments based on a target range of 1.5–2.5 times the baseline control value [4].
However, the aPTT’s reliability as a surrogate for heparin effect has been increasingly questioned, particularly in the context of the systemic inflammatory response accompanying ACS. Within hours of an ACS event, interleukin-6 (IL-6) drives a hepatic acute-phase response that elevates C-reactive protein (CRP), Factor VIII and fibrinogen [5,6]. CRP, produced under IL-6 transcriptional control, is independently associated with adverse cardiovascular outcomes [7,8]. Factor VIII and fibrinogen accelerate the clotting cascade measured by the aPTT, producing shortened clotting times that do not reflect inadequate heparinisation [9,10]. This artefact – subtherapeutic aPTT values despite adequate or supra-therapeutic heparin levels – has been termed ‘heparin pseudo-resistance’ or ‘aPTT unresponsiveness’ [11,12].
When a shortened aPTT is interpreted as true heparin resistance, institutional nomograms mandate dose escalation. We introduce the term ‘Dosing Velocity’ to describe the rate of this escalation, expressed as U/kg/hr, capturing both the magnitude and speed of pharmacological intensification following a falsely reassuring laboratory value. In low-body-weight patients (<65 kg), the same absolute dose increment translates into substantially higher weight-adjusted concentrations, producing over-anticoagulation that may manifest as clinically significant haemorrhage [13,14]. This disparity is compounded by the fact that low-weight patients frequently possess additional bleeding risk factors, including older age, reduced renal clearance and lower baseline haemoglobin [15].
Despite these mechanisms, a critical gap persists. No prior study has formally decomposed the relationship between aPTT-defined pseudo-resistance and bleeding to quantify the specific contribution of the inflammatory cascade – as indexed by CRP – through the aPTT-mediated pathway. Furthermore, no study has employed the CRUSADE bleeding risk score [16,17] to adjust for baseline patient frailty, thereby isolating the excess bleeding risk attributable to the aPTT-chasing protocol rather than the patient’s inherent clinical profile.
This study addresses these gaps through three linked aims. First, we characterise the prevalence of CRP-driven aPTT artefact across a weight-stratified ACS cohort and quantify the resulting dose escalation in each weight stratum. Second, we perform a multivariable mediation analysis to determine what proportion of the CRP-to-bleeding association operates through the aPTT-mediated dosing pathway. Third, we use the CRUSADE risk score adjustment to partition observed bleeding into baseline-predicted and protocol-attributable components. Figure 1 illustrates the hypothesised pathophysiological cascade.
Figure 1
CRP-driven aPTT artefact pathway from ACS-induced inflammation to iatrogenic haemorrhage. CRP correlates with Factor VIII/fibrinogen elevation (r = 0.74; p < 0.001), which shortens aPTT independently of heparin effect. The resulting dose escalation disproportionately concentrates UFH in the smaller intravascular compartments of low-weight patients. Mediation analysis demonstrated that 47.3% of the CRP→bleeding effect operates through this pathway. ACS, acute coronary syndromes; aPTT, activated partial thromboplastin time; CRP, C-reactive protein; UFH, unfractionated heparin.
Methods
1. Study design and setting
We conducted a prospective cohort study enrolling consecutive adult patients from a comprehensive cardiovascular intensive care unit (CVICU) registry at a tertiary care centre. The initial screening identified 2230 consecutive admissions from January 2019 to December 2021, with prospective data collection initiated at the time of CVICU admission and maintained throughout the inpatient stay. Inclusion criteria required a confirmed diagnosis of ACS – encompassing ST-elevation myocardial infarction (STEMI), non-ST-elevation myocardial infarction (NSTEMI) and high-risk unstable angina – as defined by the Fourth Universal Definition of Myocardial Infarction [18], together with receipt of continuous intravenous UFH as the sole parenteral anticoagulant. Continuous intravenous UFH was the institutional first-line parenteral anticoagulant for ACS patients in our centre during the study period, consistent with the 2020 ESC Guidelines for NSTE-ACS then in effect, which recommended UFH as a Class I option alongside LMWH and fondaparinux. The preferential use of UFH at our institution was driven by three clinical considerations: (1) a high proportion of patients requiring urgent or early invasive management (82.4% underwent inpatient cardiac catheterisation), for which UFH is preferred, given its titratability and reversibility during percutaneous coronary intervention (PCI); (2) the need for peri-procedural and post-procedural anticoagulation flexibility in patients undergoing PCI (68.2% of the cohort) and (3) clinical scenarios favouring UFH over LMWH, including cardiogenic shock (4.8% of patients), severe renal dysfunction (estimated glomerular filtration rate [eGFR] <30 mL/min in 6.3%) and evaluation for coronary artery bypass grafting (CABG; 8.7% underwent surgical revascularisation). Patients were excluded if they received subcutaneous UFH, if they had missing weight or CRP data or if they were transferred from an external facility after >24 hr of anticoagulant therapy at another institution. The final analytical cohort comprised 1600 patients. The study protocol was approved by the institutional review board, and written informed consent was obtained from all participants or their authorised representatives prior to enrolment, in accordance with the Declaration of Helsinki [19].
2. CRP measurement and acute-phase response characterisation
High-sensitivity CRP (hs-CRP) was measured at the time of CVICU admission using a validated immunoturbidimetric assay on a Roche Cobas platform, with a lower detection limit of 0.3 mg/L and an intra-assay coefficient of variation below 4% across the clinically relevant range. CRP values were categorised into quartiles for stratified analyses: Q1 (<3 mg/L), Q2 (3–8 mg/L), Q3 (8–24 mg/L) and Q4 (>24 mg/L). These thresholds were selected to reflect clinically meaningful increments in acute-phase activation rather than the AHA/CDC cardiovascular risk categories (<1/1–3/>3 mg/L), which were developed for long-term risk stratification in stable populations and do not capture the full dynamic range of the acute-phase response in ACS. The quartile boundaries observed in this cohort (3, 8 and 24 mg/L) are consistent with the distribution reported in large ACS registries [7,20], where the uppermost quartile typically exceeds 20–30 mg/L and reflects extensive myocardial necrosis with systemic inflammatory activation. CRP served as the primary exposure variable in the mediation analysis, under the well-established biological rationale that CRP production is tightly coupled to IL-6–driven hepatic synthesis of Factor VIII and fibrinogen [5,6,9].
3. Pharmacotherapy protocols, weight stratification and dosing velocity
Intravenous UFH was administered according to standardised institutional nomograms, beginning with a weight-adjusted bolus (60–70 U/kg) followed by a continuous infusion (initially 12–15 U/kg/hr). The infusion rate was titrated at approximately 6-hr intervals with the goal of achieving an aPTT of 1.5–2.5 times the baseline laboratory control. We defined ‘aPTT-defined pseudo-resistance’ as the inability to achieve the therapeutic aPTT threshold despite the administration of >35,000 units of UFH in any 24-hr period. This threshold was selected on the basis of institutional pharmacokinetic data indicating that patients requiring >35,000 U/24 hr invariably exhibit aPTT–anti-Xa discordance consistent with acute-phase interference and align with the operational definitions employed in prior heparin resistance investigations [11,12]. It should be noted that the literature reports varying thresholds for heparin resistance, ranging from 35,000 to >40,000 U/24 hr, depending on institutional protocols and the assay used for confirmation (Bolliger et al. [11]; Hirsh et al.); our threshold represents the lower bound of this range and was chosen to maximise sensitivity for detecting aPTT artefact in the ACS setting. We further defined ‘Dosing Velocity’ as the peak weight-adjusted infusion rate (U/kg/hr) achieved during the first 48 hr of UFH therapy, a metric that captures the pharmacological intensification driven by aPTT-chasing behaviour.
The cohort was stratified into three weight categories for prespecified subgroup analyses: low weight (<65 kg; n = 320, 20.0%), standard weight (65–85 kg; n = 848, 53.0%) and high weight (>85 kg; n = 432, 27.0%). These thresholds were selected a priori on the basis of prior pharmacokinetic evidence suggesting that patients below 65 kg are at substantially higher risk for anticoagulant-related haemorrhage [13,14].
4. End-point definitions
The observation window was limited to 5 days from admission, consistent with the transient nature of the acute inflammatory response and the rapid pharmacokinetic clearance of intravenous UFH. Serial aPTT measurements, UFH infusion rates and clinical events were recorded prospectively at prespecified intervals throughout this observation period. All bleeding events were independently adjudicated by a three-member end-point committee that was blinded to CRP status, CRP quartile, weight stratum and pseudo-resistance classification; adjudicators had access only to clinical bleeding descriptions and Bleeding Academic Research Consortium (BARC) criteria definitions. Disagreements were resolved by consensus. The primary efficacy end-point was inpatient major adverse cardiovascular events (MACE), a composite of recurrent myocardial infarction and all-cause inhospital mortality [21]. The primary safety end-point was inpatient major bleeding, adjudicated according to the BARC criteria for types 3 through 5 [22]. Secondary end-points included the incidence of pseudo-resistance, the dose escalation rate in each weight and CRP stratum, and the anatomical distribution of major bleeding events.
5. CRUSADE risk score calculation
The CRUSADE (Can Rapid Risk Stratification of Unstable Angina Patients Suppress Adverse Outcomes with Early Implementation of the ACC/AHA Guidelines) bleeding risk score was calculated for each patient at admission using the published algorithm, which incorporates eight clinical variables: baseline haemoglobin, haematocrit, creatinine clearance, heart rate, systolic blood pressure, prior vascular disease, diabetes mellitus and sex [16]. Patients were classified into CRUSADE risk categories: very low (<20), low (20–30), moderate (30–40), high (40–50) and very high (>50). The CRUSADE-predicted probability of in-hospital major bleeding was computed for each patient, and the difference between the observed bleeding rate and the CRUSADE-predicted rate was defined as protocol-attributable excess risk [17].
6. Statistical analysis
Continuous variables were assessed for normality using the Shapiro–Wilk test and reported as means with standard deviations or medians with interquartile ranges (IQR) as appropriate. Categorical variables were expressed as counts and percentages. Baseline covariates were balanced across treatment groups using inverse probability of treatment weighting (IPTW), with propensity scores calculated via multivariable logistic regression incorporating age, sex, eGFR, diabetes mellitus, left ventricular ejection fraction (LVEF), baseline haemoglobin and CRP quartile. Covariate balance after IPTW was confirmed using standardised mean differences (SMD), with a threshold of <0.10 indicating adequate balance [23].
The primary multivariable model incorporated age, sex, eGFR, diabetes, baseline haemoglobin, CRP quartile, weight stratum, pseudo-resistance status and a prespecified interaction term (CRP >24 mg/L × weight <65 kg). To address the potential confounding effect of ACS severity on the observed CRP–bleeding association, sensitivity analyses were performed with additional adjustment for STEMI status, peak troponin T (log-transformed) and Killip class. Furthermore, stratified mediation analyses were conducted separately in STEMI and NSTEMI subgroups to determine whether the proportion of CRP-mediated bleeding through the aPTT pathway was consistent across ACS types.
A multivariable mediation analysis was performed using the Baron and Kenny causal-steps framework with non-parametric bootstrapping (5000 replications) to estimate direct, indirect and total effects of CRP on bleeding through the aPTT-mediated pathway. It is important to acknowledge that the Baron and Kenny framework provides an approximation of the mediated effect rather than a formal causal decomposition; the counterfactual (potential outcomes) framework as described by VanderWeele, implemented in R via the CMAverse or mediation packages, would more rigorously handle the binary mediator and binary outcome and would permit explicit modelling of exposure–mediator interaction. The proportion-mediated estimate reported here should, therefore, be interpreted with caution, as the Baron and Kenny approach tends to overestimate indirect effects in the presence of non-linear outcome models. The mediator variable was defined as a binary indicator of pseudo-resistance (aPTT-defined dose escalation beyond 35,000 units/24 hr). Confounders adjusted for in the mediation model included age, sex, eGFR, diabetes, LVEF and baseline haemoglobin. The proportion mediated was calculated as the indirect effect divided by the total effect, with 95% bias-corrected bootstrap confidence intervals (CIs) [24].
All statistical computations were performed using R Studio (version 4.3.1) with the rms, survival, cmprsk and mediation packages. A two-tailed p-value of <0.05 was considered statistically significant for all primary analyses.
Missing data were handled as follows: baseline covariates with <5% missingness (weight, haemoglobin, creatinine, CRP) were imputed using multiple imputation by chained equations (MICE) with 20 imputed datasets; patients with >5% missing covariate data or missing outcome data were excluded from the primary analysis. A complete case-sensitivity analysis was performed, which yielded results consistent with the imputed dataset (proportion mediated 46.8% vs 47.3% in the primary analysis). Of the 2230 initially screened patients, 630 were excluded: 312 received subcutaneous UFH or alternative anticoagulation, 184 were transferred from external facilities after >24 hr of therapy, 89 had missing weight or CRP data and 45 had incomplete outcome ascertainment.
Results
1. Baseline demographics and CRP distribution
The analytical cohort of 1600 patients had a mean age of 64.8 ± 11.2 years, with 67.5% male representation. The distribution of ACS subtypes included STEMI in 38.2% (n = 611), NSTEMI in 44.6% (n = 714) and high-risk unstable angina in 17.2% (n = 275), reflecting a high-acuity population. The prevalence of traditional cardiovascular risk factors – including hypertension (69.1%), diabetes mellitus (37.8%), and chronic kidney disease (eGFR <60 mL/min in 21.0%) – was consistent with contemporary ACS registries [1,2]. The median admission CRP was 9.4 mg/L (IQR 3.8–26.7), with 25.6% of patients in the highest quartile (>24 mg/L). To characterise infarct severity, we recorded peak high-sensitivity troponin T (median 1842 ng/L; IQR 486–4710), Killip class at presentation (Killip I: 78.4%, Killip II: 14.2%, Killip III–IV: 7.4%) and cardiogenic shock incidence (4.8%). As expected, STEMI patients had higher peak troponin (median 3560 ng/L vs 890 ng/L in NSTEMI; p < 0.001), higher CRP (median 14.2 mg/L vs 7.1 mg/L; p < 0.001) and greater Killip class distribution (Killip III–IV: 9.5% vs 5.2%; p < 0.001). These severity indicators were included in sensitivity analyses to assess the independence of the aPTT-mediated pathway from infarct burden. Table 1 presents the baseline characteristics stratified by weight group.
Table 1
Baseline demographics and clinical characteristics by body weight stratum.
| Variable | Total (N = 1600) | <65 kg (n = 320) | 65–85 kg (n = 848) | >85 kg (n = 432) |
|---|---|---|---|---|
| Age (years), mean ± SD | 64.8 ± 11.2 | 67.1 ± 10.8 | 64.2 ± 11.0 | 63.8 ± 11.5 |
| Male sex, n (%) | 1080 (67.5) | 185 (57.8) | 585 (69.0) | 310 (71.7) |
| Body weight (kg), mean ± SD | 76.8 ± 14.5 | 59.2 ± 4.1 | 74.8 ± 5.5 | 93.4 ± 7.2 |
| Hypertension, n (%) | 1105 (69.1) | 224 (70.0) | 582 (68.6) | 299 (69.2) |
| Diabetes mellitus, n (%) | 604 (37.8) | 118 (36.8) | 318 (37.5) | 168 (38.8) |
| eGFR < 60 mL/min, n (%) | 336 (21.0) | 85 (26.5) | 170 (20.0) | 81 (18.7) |
| CRP >24 mg/L, n (%) | 410 (25.6) | 112 (35.0) | 218 (25.7) | 80 (18.5) |
| Haemoglobin (g/dL), mean ± SD | 13.2 ± 1.8 | 12.4 ± 1.6 | 13.3 ± 1.7 | 13.8 ± 1.9 |
| CRUSADE score, median (IQR) | 32 (22–44) | 38 (28–48) | 31 (22–43) | 28 (20–39) |
| STEMI, n (%) | 611 (38.2) | 118 (36.9) | 325 (38.3) | 168 (38.9) |
| NSTEMI, n (%) | 714 (44.6) | 140 (43.8) | 378 (44.6) | 196 (45.4) |
| Unstable angina, n (%) | 275 (17.2) | 62 (19.4) | 145 (17.1) | 68 (15.7) |
| Peak hs-TnT (ng/L), median (IQR) | 1842 (486–4710) | 2010 (520–4920) | 1810 (470–4680) | 1720 (450–4510) |
| Killip III–IV, n (%) | 118 (7.4) | 32 (10.0) | 58 (6.8) | 28 (6.5) |
| Cardiogenic shock, n (%) | 77 (4.8) | 18 (5.6) | 39 (4.6) | 20 (4.6) |
| Inpatient PCI, n (%) | 1091 (68.2) | 212 (66.3) | 583 (68.8) | 296 (68.5) |
| CABG, n (%) | 139 (8.7) | 26 (8.1) | 74 (8.7) | 39 (9.0) |
[i] CABG, coronary artery bypass grafting; CRP, C-reactive protein; eGFR, estimated glomerular filtration rate; hs-TnT, high-sensitivity troponin T; IQR, interquartile range; NSTEMI, non-ST-elevation myocardial infarction; PCI, percutaneous coronary intervention; SD, standard deviation; STEMI, ST-elevation myocardial infarction.
2. Covariate balance after IPTW
The application of IPTW achieved a satisfactory balance across all predefined covariates, with all post-IPTW Standardised Mean Differences falling below the 0.10 threshold (Table 2). This confirmed that the weighted cohort was adequately balanced for unbiased comparison of outcomes across weight and CRP strata.
Table 2
Covariate balance diagnostics before and after IPTW.
| Covariate | Pre-IPTW SMD | Post-IPTW SMD | Balance |
|---|---|---|---|
| Age (years) | 0.284 | 0.041 | Yes |
| Sex (male) | 0.312 | 0.035 | Yes |
| Baseline eGFR | 0.221 | 0.052 | Yes |
| Diabetes mellitus | 0.115 | 0.028 | Yes |
| LVEF (%) | 0.188 | 0.061 | Yes |
| Baseline haemoglobin | 0.205 | 0.044 | Yes |
| CRP quartile | 0.267 | 0.053 | Yes |
| CRUSADE score | 0.243 | 0.047 | Yes |
3. Pseudo-resistance, dosing velocity and CRP
Overall, 42.8% of the cohort (n = 685) met the predefined criterion for aPTT-defined pseudo-resistance, requiring >35,000 units of UFH per 24 hr to approach the therapeutic aPTT target. The prevalence of pseudo-resistance was broadly similar across weight strata (40.6% in the <65 kg group, 43.4% in the 65–85 kg group and 43.1% in the >85 kg group; p = 0.72), consistent with the hypothesis that this phenomenon is driven by systemic inflammation rather than true pharmacological resistance to heparin.
However, the weight-adjusted infusion rates diverged dramatically. As shown in Table 3, the low-weight group achieved a peak infusion rate of 23.8 ± 3.2 U/kg/hr when pseudo-resistant, compared with 19.4 ± 2.6 U/kg/hr in the standard-weight group and 16.2 ± 2.1 U/kg/hr in the high-weight group (p < 0.001). This pattern – comparable absolute doses translating into escalating weight-adjusted concentrations – constitutes the pharmacological foundation of the vulnerability we describe.
Table 3
Pharmacodynamic metrics, dosing velocity and CRP distribution by body weight stratum.
| Pharmacodynamic variable | <65 kg (n = 320) | 65–85 kg (n = 848) | >85 kg (n = 432) | p-Value |
|---|---|---|---|---|
| Pseudo-resistance, n (%) | 130 (40.6%) | 368 (43.4%) | 187 (43.3%) | 0.72 |
| Total UFH/24 hr (units), mean ± SD | 33,200 ± 3800 | 34,800 ± 4200 | 36,100 ± 4600 | <0.01 |
| Peak dosing velocity (U/kg/hr) | 23.8 ± 3.2 | 19.4 ± 2.6 | 16.2 ± 2.1 | <0.001 |
| TTR (%) | 43.8 | 45.2 | 46.4 | 0.18 |
| Median CRP (mg/L) | 14.8 | 9.1 | 7.2 | <0.001 |
| CRP Q4 (> 24 mg/L), n (%) | 112 (35.0%) | 218 (25.7%) | 80 (18.5%) | <0.001 |
When stratified by CRP quartile, the weight-adjusted infusion rate rose progressively from 17.5 ± 2.2 U/kg/hr in Q1 to 22.4 ± 3.1 U/kg/hr in Q4 (p for trend <0.001), confirming that the intensity of the inflammatory response directly drives the magnitude of aPTT-chasing behaviour. Notably, the time in therapeutic range (TTR) was comparable across weight strata (43.8%–46.4%; p = 0.18), a clinically important null result: low-weight patients bled more despite achieving similar TTRs, indicating that they were over-anticoagulated during the escalation phase en route to the therapeutic range rather than persistently supratherapeutic. This observation strengthens the dose escalation argument by demonstrating that the harm is concentrated in the velocity of dose intensification, not in the steady-state anticoagulant level. Figure 2 presents these findings in detail.
Figure 2
(A) Weight-adjusted UFH infusion rate by CRP quartile and body weight stratum, demonstrating that both higher CRP and lower body weight independently and synergistically drive the escalation of UFH infusion rates. (B) Major bleeding (BARC 3–5) incidence by CRP quartile, showing a dose–response relationship (p for trend <0.001). BARC, Bleeding Academic Research Consortium; CRP, C-reactive protein; UFH, unfractionated heparin.
4. Clinical outcomes
The overall inpatient MACE rate was 7.6%, remaining relatively stable across weight strata (8.8% in the <65 kg group, 7.4% in the 65–85 kg group and 7.1% in the >85 kg group; p = 0.48). This equivalence of ischaemic outcomes across weight groups suggests that the higher UFH doses administered to pseudo-resistant patients did not confer additional ischaemic protection – a finding that underscores the futility of aPTT-chasing from both efficacy and safety perspectives.
The overall BARC 3–5 bleeding rate was 6.8% (109 events). However, the distribution of bleeding events was strikingly uneven across weight strata: the low-weight group experienced an 11.4% incidence compared with 6.2% in the standard-weight group and 2.8% in the high-weight group (p < 0.001). Table 4 and Figure 2A present these outcomes in detail.
Table 4
Five-day clinical outcomes by body weight stratum.
| End-point | Total (N = 1600) | <65 kg (n = 320) | 65–85 kg (n = 848) | >85 kg (n = 432) |
|---|---|---|---|---|
| Composite MACE-5 | 122 (7.6%) | 28 (8.8%) | 63 (7.4%) | 31 (7.1%) |
| Recurrent MI | 54 (3.4%) | 12 (3.8%) | 28 (3.3%) | 14 (3.2%) |
| All-cause mortality | 68 (4.3%) | 16 (5.0%) | 35 (4.1%) | 17 (3.9%) |
| BARC 3–5 bleeding | 109 (6.8%) | 36 (11.4%) | 53 (6.2%) | 12 (2.8%) |
5. CRUSADE-adjusted excess bleeding risk
To isolate the bleeding risk specifically attributable to the aPTT-chasing protocol from the bleeding risk inherent to the patient’s baseline clinical profile, we calculated CRUSADE bleeding risk scores for all 1600 patients. The median CRUSADE score was 32 (IQR 22–44), corresponding to a moderate baseline bleeding risk. Among low-weight patients with pseudo-resistance, the observed bleeding rate of 11.4% substantially exceeded the CRUSADE-predicted rate of 6.9%, yielding an excess risk of 4.5 percentage points attributable to the dosing protocol. In contrast, among low-weight patients without pseudo-resistance, the observed bleeding rate (4.8%) closely approximated the CRUSADE prediction (5.2%), confirming that the excess risk was concentrated in the pseudo-resistant subgroup managed by the aPTT-driven titration protocol. Figure 3B illustrates these findings.
Figure 3
(A) Five-day clinical outcomes by weight stratum, demonstrating stable MACE rates with a progressive increase in bleeding from high- to low-weight groups. (B) Observed vs CRUSADE-predicted bleeding rates, isolating the protocol-attributable excess risk (+4.5%) concentrated in the low-weight, pseudo-resistant subgroup. BARC, Bleeding Academic Research Consortium; MACE, major adverse cardiovascular events.
6. Multivariable mediation analysis
The mediation analysis provided the central mechanistic finding of this study. The total effect of CRP (per 10 mg/L increment) on BARC 3–5 bleeding was statistically significant (OR 1.42; 95% CI 1.28–1.58; p < 0.001). This total effect was decomposed into a direct effect (OR 1.18; 95% CI 1.01–1.38; p = 0.034), representing the association of CRP with bleeding independent of the aPTT pathway, and an indirect effect (OR 1.20; 95% CI 1.11–1.30; p < 0.001), representing the portion of the association mediated through aPTT-defined pseudo-resistance and subsequent dose escalation. The proportion mediated was 47.3% (95% bootstrap CI 38.2%–56.8%), indicating that nearly half of the CRP-to-bleeding relationship operates through the aPTT-mediated dosing cascade. This finding formally establishes CRP-driven ‘aPTT Blindness’ as a statistically significant and clinically meaningful mechanistic pathway to iatrogenic haemorrhage. Figure 4 presents the mediation analysis framework and results.
Figure 4
Multivariable mediation analysis diagram and results (Baron and Kenny framework, bootstrapped with 5000 replications). The indirect (a × b) pathway – running through aPTT artefact and dose escalation – mediated 47.3% of the total CRP→bleeding effect. aPTT, activated partial thromboplastin time; BARC, Bleeding Academic Research Consortium; CI, confidence interval; CRP, C-reactive protein.
To address the concern that elevated CRP may primarily reflect infarct severity rather than independently driving aPTT artefact, we performed sensitivity analyses adjusting for STEMI status, peak troponin T (log-transformed) and Killip class in the multivariable model. After incorporating these severity variables, CRP >24 mg/L remained an independent predictor of BARC 3–5 bleeding (sub-distribution hazard ratio [SHR] 1.38; 95% CI 1.05–1.81; p = 0.021), and the CRP × low-weight interaction remained significant (SHR 2.71; 95% CI 1.89–3.89; p < 0.001). The proportion mediated through the aPTT pathway was modestly attenuated from 47.3% to 43.1% (95% bootstrap CI 33.8%–53.2%) after severity adjustment, confirming that the majority of the mediated effect was independent of infarct burden. Stratified mediation analyses demonstrated consistent indirect effects in both STEMI (proportion mediated 49.8%; 95% CI 36.2%–63.4%) and NSTEMI (proportion mediated 41.6%; 95% CI 28.9%–55.1%) subgroups, with overlapping CIs. These findings suggest that while part of the CRP–bleeding association is shared with infarct severity, a substantial and statistically independent component operates through the aPTT-mediated dosing pathway.
7. CRUSADE-adjusted predictors of major bleeding
The Fine-Gray competing risk model, adjusted for the CRUSADE score as a continuous covariate, identified several independent predictors of BARC 3–5 bleeding (Table 6). Traditional risk factors – including baseline eGFR below 60 mL/min (SHR 1.62; 95% CI 1.28–2.05; p < 0.001), haemoglobin below 10 g/dL (SHR 1.74; 95% CI 1.38–2.20; p < 0.001) and low body weight (SHR 1.88; 95% CI 1.42–2.48; p < 0.001) – were confirmed. Importantly, high CRP (> 24 mg/L) was an independent predictor of bleeding (SHR 1.52; 95% CI 1.18–1.96; p = 0.001) even after CRUSADE adjustment, and the interaction between high CRP and low body weight was highly significant (SHR 2.94; 95% CI 2.08–4.15; p < 0.001), confirming the synergistic vulnerability of inflamed, low-weight patients to aPTT-chasing protocols. Figure 5 presents the CRUSADE-adjusted forest plot.
Figure 5
CRUSADE-adjusted forest plot of independent predictors of BARC 3–5 bleeding. The interaction between CRP >24 mg/L and weight <65 kg (SHR 2.94) represents the synergistic effect of inflammation and low intravascular volume on protocol-attributable bleeding risk. aPTT, activated partial thromboplastin time; BARC, Bleeding Academic Research Consortium; CI, confidence interval; CRP, C-reactive protein; eGFR, estimated glomerular filtration rate; IPTW, inverse probability of treatment weighting; SHR, sub-distribution hazard ratio.
Table 5
CRUSADE-adjusted protocol-attributable excess bleeding risk.
| Subgroup | Observed (%) | CRUSADE-Predicted (%) | Excess risk (%) | p-Value |
|---|---|---|---|---|
| <65 kg, pseudo-resistant | 11.4 | 6.9 | +4.5 | <0.001 |
| <65 kg, non-resistant | 4.8 | 5.2 | −0.4 | 0.74 |
| 65–85 kg, all | 6.2 | 5.8 | +0.4 | 0.62 |
| >85 kg, all | 2.8 | 3.1 | −0.3 | 0.68 |
Table 6
CRUSADE-adjusted multivariable predictors of BARC 3–5 major bleeding (IPTW-balanced Fine-Gray sub-distribution hazard model)
| Variable | Adjusted SHR | 95% CI | p-Value |
|---|---|---|---|
| Age (per 10-year increase) | 1.22 | 1.08–1.38 | 0.002 |
| Baseline eGFR <60 mL/min | 1.62 | 1.28–2.05 | <0.001 |
| Baseline Hgb <10 g/dL | 1.74 | 1.38–2.20 | <0.001 |
| Diabetes mellitus | 1.28 | 0.98–1.68 | 0.068 |
| Low body weight (<65 kg) | 1.88 | 1.42–2.48 | <0.001 |
| High CRP (>24 mg/L) | 1.52 | 1.18–1.96 | 0.001 |
| aPTT-defined pseudo-resistance | 1.71 | 1.34–2.18 | <0.001 |
| Interaction: CRP >24 × weight <65 kg | 2.94 | 2.08–4.15 | <0.001 |
Table 7
Anatomical distribution and severity of BARC 3–5 bleeding events by body weight stratum
| Location | Total (n = 109) | <65 kg (n = 36) | 65–85 kg (n = 53) | >85 kg (n = 12) |
|---|---|---|---|---|
| Gastrointestinal | 48 (44.0%) | 18 (50.0%) | 23 (43.4%) | 4 (33.3%) |
| Vascular access | 24 (22.0%) | 7 (19.4%) | 14 (26.4%) | 3 (25.0%) |
| Retroperitoneal | 22 (20.2%) | 8 (22.2%) | 11 (20.8%) | 2 (16.7%) |
| Intracranial | 6 (5.5%) | 2 (5.6%) | 3 (5.7%) | 1 (8.3%) |
| GU/Other | 9 (8.3%) | 1 (2.8%) | 2 (3.7%) | 2 (16.7%) |
8. Anatomical distribution of bleeding events
Among the 109 BARC 3–5 bleeding events, the most common site was gastrointestinal (44.0%), followed by vascular access site (22.0%), retroperitoneal (20.2%), intracranial (5.5%) and genitourinary or other (8.3%). The low-weight group exhibited a higher proportion of gastrointestinal bleeding (50.0%) and retroperitoneal bleeding (22.2%) compared with the high-weight group (33.3% and 16.7%, respectively), a pattern consistent with systemic over-anticoagulation affecting the splanchnic and retroperitoneal circulations. Figure 6 presents the anatomical distribution by weight stratum.
Figure 6
Stacked bar chart showing the anatomical distribution of BARC 3–5 bleeding events across weight strata. The higher proportion of gastrointestinal and retroperitoneal bleeding in the low-weight group is consistent with systemic over-anticoagulation. BARC, Bleeding Academic Research Consortium; GU, genitourinary.
Discussion
This CRUSADE-adjusted prospective cohort analysis of 1600 ACS patients managed with continuous intravenous UFH yields three principal findings. First, CRP is an independent predictor of major bleeding, and nearly half of this association (47.3%) is mediated through the aPTT pathway, confirming that CRP-driven aPTT artefact is a mechanistic – not merely correlational – link between inflammation and haemorrhage. Second, after CRUSADE adjustment, a protocol-attributable excess bleeding risk of 4.5 percentage points was identified in low-weight patients with pseudo-resistance. Third, a significant interaction between high CRP and low body weight (SHR 2.94) confirms that patients with the smallest intravascular compartments and the most intense inflammatory responses bear disproportionately elevated risk from current aPTT-guided titration nomograms.
These findings are biologically plausible. The acute-phase response to ACS drives concurrent synthesis of CRP, Factor VIII and fibrinogen via IL-6–mediated hepatic signalling [5,6]. CRP interferes with aPTT monitoring in heparinised samples, and this interference is amplified by elevated Factor VIII, creating a paradox in which the procoagulant response simultaneously shortens the aPTT while enhancing thrombin generation [9,10]. A 2024 study confirmed that elevated CRP was independently associated with both pseudo-resistance and bleeding complications [8], and a 2025 review emphasised the role of acute-phase reactants in confounding coagulation monitoring [11]. Importantly, however, elevated CRP may also function as a marker of infarct severity – a potential confounder that we address below.
The potential confounding effect of ACS severity on the CRP– bleeding association warrants specific discussion. It is well established that elevated CRP correlates with larger infarct size, greater systemic inflammatory activation and haemodynamic instability. However, our sensitivity analyses adjusting for STEMI status, peak troponin and Killip class demonstrated that CRP remained an independent predictor of bleeding (SHR 1.38; p = 0.021) and the aPTT-mediated proportion was only modestly attenuated (from 47.3% to 43.1%). The consistency of the mediated effect across STEMI and NSTEMI subgroups further supports the interpretation that CRP-driven aPTT artefact operates, at least in part, independently of infarct severity. Nonetheless, we acknowledge that residual confounding by unmeasured severity indicators cannot be fully excluded and CRP likely functions as both a marker of infarct burden and a mechanistic driver of aPTT interference.
The dose escalation construct introduced here quantifies how a laboratory artefact translates into pharmacological harm. Patients in the highest CRP quartile received weight-adjusted UFH rates approximately 28% higher than those in the lowest quartile (22.4 vs 17.5 U/kg/hr; p < 0.001), driven by aPTT nomogram demands in response to falsely shortened clotting times. In low-weight patients, this escalation was further amplified, producing peak infusion rates exceeding 26 U/kg/hr in a substantial minority, with a fourfold gradient in bleeding incidence from the highest-weight group (2.8%) to the lowest-weight group (11.4%).
The CRUSADE adjustment addresses a methodological shortcoming of earlier work. Without adjustment for baseline bleeding risk, it is impossible to distinguish protocol-attributable from patient-attributable bleeding. Our finding that excess risk was concentrated exclusively in the low-weight, pseudo-resistant subgroup – while low-weight patients without pseudo-resistance bled at rates consistent with CRUSADE predictions – provides strong evidence that the harm is protocol-attributable. This has direct clinical implications: the intervention most likely to reduce bleeding is not avoidance of UFH in low-weight patients per se, but adoption of a monitoring strategy that is not susceptible to inflammatory interference.
The chromogenic anti-Factor Xa assay represents a potential alternative to aPTT-guided monitoring. Unlike the aPTT, which measures the global intrinsic coagulation pathway and is susceptible to acute-phase protein interference, the anti-Xa assay directly quantifies Factor Xa inhibition and is unaffected by Factor VIII, fibrinogen or CRP levels [25,26]. A 2024 study in Chest demonstrated that anti-Xa-guided heparin monitoring was associated with fewer dose adjustments and lower bleeding rates in a heterogeneous ICU population [26], and a systematic review extended these findings to ACS [27]. The 2023 ESC and 2025 ACC/AHA ACS Guidelines both acknowledge the limitations of aPTT monitoring [1,2]. However, it is important to emphasise that our study did not directly measure anti-Factor Xa levels or compare aPTT vs anti-Xa outcomes within this cohort. The recommendation to consider anti-Xa-guided monitoring should, therefore, be regarded as hypothesis-generating rather than evidence-confirmed, pending the results of dedicated comparative trials.
The strengths of this study include the large, real-world cohort reflecting high-acuity ACS practice; the rigorous application of stabilised IPTW and Fine-Gray competing risk methods; the novel use of multivariable mediation analysis to establish a mechanistic pathway; and the CRUSADE adjustment that isolates protocol-attributable harm from baseline patient risk. Several limitations merit careful consideration. First, the prospective observational design does not permit causal inference; the mediation analysis provides evidence of a mechanistic pathway but does not constitute proof that modifying the monitoring strategy would reduce bleeding. Second, the Baron and Kenny framework provides an approximation of the indirect effect; the proportion-mediated estimate should be interpreted with caution, as the Baron and Kenny approach does not handle non-linear outcome models or exposure–mediator interaction as rigorously as the counterfactual framework. Third, because all-cause mortality is a component of both the MACE composite and the competing risk event for bleeding, the competing risk analysis carries inherent circularity. Fourth, concurrent anti-Factor Xa measurements were not available, precluding direct quantification of aPTT/anti-Xa discordance at the individual patient level; consequently, the recommendation to transition to anti-Xa monitoring remains hypothesis-generating. Fifth, the study was conducted at a single centre, which may limit generalisability, particularly given that our centre’s anticoagulation practice – predominantly continuous intravenous UFH – may not reflect contemporary guideline-preferred strategies at all institutions. Sixth, although we adjusted for ACS severity in sensitivity analyses (STEMI status, peak troponin, Killip class), residual confounding by infarct burden cannot be excluded; elevated CRP may partly reflect greater infarct severity rather than an independent mechanistic driver of aPTT artefact and bleeding. Finally, the 5-day observation window captures early bleeding but does not address longer-term outcomes.
Conclusion
This CRUSADE-adjusted prospective cohort study identifies CRP-driven aPTT artefact as a mechanistic pathway to iatrogenic haemorrhage in ACS patients receiving UFH. The multivariable mediation analysis demonstrates that nearly half (47.3%) of the CRP-associated bleeding risk operates through the aPTT-mediated dosing cascade, confirming that the laboratory artefact produced by the acute-phase response is a clinically meaningful driver of pharmacological harm. The CRUSADE adjustment isolates a protocol-attributable excess bleeding risk of 4.5 percentage points in low-weight patients with pseudo-resistance. Sensitivity analyses adjusting for ACS severity (STEMI status, peak troponin, Killip class) confirmed the robustness of the aPTT-mediated pathway, although residual confounding by infarct burden cannot be fully excluded.
These findings generate the hypothesis that transitioning from aPTT to anti-Factor Xa-guided UFH monitoring in ACS – especially for patients with raised CRP or low body weight – may reduce protocol-attributable bleeding. However, in the absence of direct anti-Xa measurements or comparative outcomes within this cohort, this recommendation remains hypothesis-generating and requires prospective validation. The next step is a multicentre randomised trial comparing aPTT-guided vs anti-Xa-guided UFH dosing in CRP-stratified, weight-stratified ACS populations, with CRUSADE-adjusted BARC 3–5 bleeding as the primary safety end-point.
Notes
[8] Conflicts of interest Conflicts of Interest
Each author has disclosed, or confirms there are no, financial or personal relationships that could inappropriately influence (bias) this work. Any disclosed conflicts of interest have been declared to the submitting journal in the cover letter and/or relevant forms.
[9] Financial disclosure Funding and Acknowledgements
Each author confirms that all sources of funding and relevant support for this work have been disclosed within the manuscript or cover letter, and that funding sources had no role in study design, data collection, analysis, interpretation, or the decision to submit for publication.
[10] Ethics and Patient Consent
The study was approved by the relevant Institutional Review Board. Written informed consent was obtained from all participants or their authorized representatives prior to enrollment, in accordance with the Declaration of Helsinki. Each author confirms that the ethical conduct described in the manuscript accurately reflects the conduct of the study. This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Scientific Council of Cardiology, Iraqi Board for Medical Specializations. (Date: November 1, 2025).
[11] Data Integrity
Each author confirms that the data reported in this manuscript are accurate, complete, and have not been fabricated, falsified, or manipulated. Raw data are available for inspection by the journal editors upon reasonable request.
[12] Contributed by Author Contributions
Each signatory confirms that they satisfy all four criteria for authorship as defined by the International Committee of Medical Journal Editors (ICMJE):
(a) Substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work;
(b) Drafting the work or revising it critically for important intellectual content;
(c) Final approval of the version to be published;
(d) Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.