Acute kidney injury is a major mediator of intra-abdominal pressure-related mortality in critically ill patients: a longitudinal analysis

Abstract

Introduction

Intra-abdominal hypertension (IAH) is a common complication in critically ill patients and is associated with increased mortality. While acute kidney injury (AKI) and respiratory impairment are also linked to IAH, their roles as mediators of mortality remain unclear. This study aimed to evaluate the associations between IAH and mortality, with a focus on AKI and pulmonary parameters as potential mediators.

Methods

This retrospective cohort study utilized the MIMIC-IV database and included adult patients admitted to the intensive care unit (ICU) with intra-abdominal pressure measurements. Patients with previous advanced kidney disease or early kidney replacement therapy (KRT) were excluded. Time-varying exposure to IAH, AKI status, and respiratory parameters were analyzed via marginal structural models (MSMs) and a mediational g-formula to assess the effects of mediation on mortality.

Results

Among the 555 patients, IAH was associated with mortality HR 2.20 (95% CI 1.54–2.56) and stage 3 AKI emerged as a significant mediator of IAH-associated mortality, accounting for almost half (41.5%, p < 0.001) of the excess mortality. KRT was associated with a protective effect, reducing the hazard ratio for mortality by 16.6%. Although the need for mechanical ventilation per se mediated a statistically significant but small effect of IAH on mortality (4.2%), no respiratory parameters, including driving pressure, demonstrated a significant mediating role.

Conclusion

Severe acute kidney injury (stage 3) is a key mediator of IAH-related mortality in critically ill patients, whereas KRT was associated with a protective effect. The absence of an important mediating role for respiratory parameters suggests that the relationship between IAH and mortality is driven primarily by renal mechanisms. However, pulmonary impairment may not have been fully captured by the variables studied, particularly in a retrospective study. Unmeasured aspects of pulmonary dysfunction could still contribute to mortality.

Supplementary Information

The online version contains supplementary material available at 10.1186/s44158-026-00350-6.

Introduction

Intra-abdominal hypertension (IAH) and its most severe form, abdominal compartment syndrome (ACS), are increasingly recognized as common and potentially life-threatening complications that significantly contribute to mortality in critically ill patients [1]. The prevalence of IAH in intensive care units (ICUs) can reach 50%. Among these patients, approximately one-third develop IAH after ICU admission [2]. In contrast to previous findings, intra-abdominal pathology is not a prerequisite for the development of IAH, and a broad patient population is affected, with an approximately equal distribution among medical, surgical, and trauma admissions [2, 3].

In the past, there was contradictory evidence regarding the clinical significance of IAH. For example, some ancillary studies have demonstrated that IAH is an independent predictor of mortality, whereas others have reported no such association [4, 5]. However, over the past 10 years, recent studies have consistently shown that IAH is independently associated with multiple organ dysfunction [6] and mortality in ICU patients [7], not only in patients with intra-abdominal pathologies such as acute pancreatitis [8] but also in patients with clinical conditions [9].

Acute kidney injury (AKI) and respiratory impairment are common complications closely associated with IAH [10, 11]. The kidneys are particularly susceptible to IAH, as it can cause increased renal venous pressure, decreased renal perfusion, and a reduction in the glomerular filtration rate [12]. Conversely, IAH may also impair the effectiveness of mechanical ventilation, alter respiratory mechanics, and disrupt gas exchange [11, 13–15]. However, given the high prevalence of AKI and pulmonary derangement among critically ill patients, accurately determining the true impact of IAH on the incidence of both AKI and respiratory impairment remains challenging. An additional complication in fully understanding the relationships among IAH, AKI, respiratory dynamics, and subsequent mortality is the time-varying nature of all involved variables, namely, intra-abdominal pressure, AKI stage, and respiratory parameters.

To the best of our knowledge, no study has evaluated intra-abdominal pressure as a time-varying exposure. Although a retrospective study assessed the impact of cumulative intra-abdominal pressure exposure on mortality [16], it did not evaluate outcomes related to AKI or the respiratory system. Additionally, many time-varying confounders were treated as fixed variables, which limits the interpretability of the findings.

In this study, we aimed to evaluate the association between IAH and in-hospital mortality, with a focus on AKI development and pulmonary parameters as potential mediators. We employed causal inference methods capable of handling time-varying exposures, mediators, and confounders.

Methodology

Study population

Data were extracted from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. The MIMIC-IV is a publicly accessible critical care repository comprising anonymized data from patients admitted to a single-center, tertiary care hospital (Beth Israel Deaconess Medical Center, Boston, MA, USA) between 2008 and 2019 [17]. Access to the database was granted following acceptance of the data use agreement and completion of “Protecting Human Subjects” training. This study was approved by the Institutional Review Boards of both the Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center, with a waiver of informed consent.

The study population included patients aged over 18 years with an ICU length of stay (LOS) exceeding 48 h and at least one measurement of intra-abdominal pressure during the ICU stay. The exclusion criteria included patients who underwent kidney replacement therapy (KRT) prior to or within the first 24 h of ICU admission, those with a baseline serum creatinine (sCr) level > 4 mg/dL (see baseline sCr definition below), and patients with three consecutive days of missing data, including measurements of intra-abdominal pressure. Only data from the first ICU admission for each patient were analyzed.

Data collection

The variables extracted from the MIMIC-IV database included both time-fixed and time-varying features. The time-fixed variables included sex; age; admission body weight; type of ICU admission (clinical vs. surgical); and baseline sCr. Baseline sCr was defined as the lowest sCr value during hospitalization if the estimated glomerular filtration rate (GFR), which was calculated via the simplified Modification of Diet in Renal Disease (MDRD) formula, exceeded 75 mL/min/1.73 m². For patients diagnosed with chronic kidney disease (CKD), the lowest sCr value during hospitalization was utilized; otherwise, baseline sCr was estimated assuming a GFR of 75 mL/min/1.73 m². Comorbidities were assessed via the Charlson Comorbidity Index (CCI), which categorizes comorbid conditions on the basis of International Classification of Diseases codes. The daily time-varying variables collected over the first 16 days of the ICU stay included the mean arterial blood pressure; need for invasive mechanical ventilation; arterial partial pressure of the oxygen/fraction of inspired oxygen (PaO₂/FiO₂) ratio (P/F ratio); tidal volume (Vt), positive end-expiratory pressure (PEEP), plateau pressure (Pplat), and 6-h urine output (UO); 24-h fluid balance; need for vasoactive support; and daily maximum serum creatinine. As suggested for the mediation analysis (see below), laboratory variables from the first day of the ICU stay were considered baseline measurements. The data were subsequently aggregated into 3-day intervals (5 periods); continuous variables were summarized by their means, ordinal variables (such as the AKI stage) were represented by the highest value within each interval, and binary variables (yes/no) were marked as present if the patient experienced the event at least once during the 3-day period. The data were grouped into 3-day blocks to maintain the positivity assumption of the marginal structural model (MSM).

IAH definition

IAH was defined as an intra-abdominal pressure ≥ 16 mm Hg. This threshold was selected on the basis of prior studies [15], including one utilizing the MIMIC-IV database [18], which demonstrated that values above this cutoff are associated with increased AKI incidence and represent Grade II IAH according to the World Society of the Abdominal Compartment Syndrome (WSACS). Additionally, a sensitivity analysis was performed using the standard definition of abdominal compartment syndrome (ACS), characterized by an intra-abdominal pressure > 20 mmHg [19].

AKI definition and respiratory mechanical variables

AKI was classified into stages on the basis of changes in sCr and UO over specific time frames. Stage 1 was characterized by a peak-to-baseline sCr increase of ≥ 0.3 mg/dL, a peak-to-baseline ratio of 1.5 to 1.9, or UO < 0.5 mL/kg/h for 6 h. Stage 2 was defined by a ratio of 2.0 to 2.9 or UO < 0.5 mL/kg/h for 12 h. Stage 3 included a ratio of ≥ 3.0, an sCr level of ≥ 4.0 mg/dL, UO < 0.3 mL/kg/h for 24 h, or anuria for ≥ 12 h. Ventilatory parameters were used to calculate respiratory mechanics variables: driving pressure = Pplat–PEEP and respiratory system compliance (Crs) = Vt/(Pplat–PEEP).

Outcome

The main outcome was mortality during the study period. We also evaluated AKI development in the first 16 days of the ICU stay (which was also evaluated as a potential mediator). Patients who were discharged alive from the ICU prior to day 16 were censored at the time of discharge (see below).

Statistical analysis

Continuous variables are expressed as the means (SDs) or medians (IQRs), and categorical variables are expressed as counts with percentages. For skewed data distributions, natural logarithmic transformation was performed as appropriate. Infeasible values were censored, and outliers in continuous variables were winsorized at the 99th percentile. The data for grouping into 3-day blocks included only validly available data. Mortality data were available for all patients.

To address the complexities of time-varying confounding, we employed MSM and the mediational g-formula rather than conventional regression techniques. In our study population, several variables—for example, non-renal Sequential Organ Failure Assessment (SOFA) score, fluid balance, AKI status, vasoactive support, and ventilatory status—act as time-varying confounders that are themselves influenced by prior exposure to IAH. Standard Cox proportional hazards models with time-dependent covariates are inadequate in this context, as they fail to account for this reciprocal relationship.

MSM analysis was conducted with inverse probability of treatment weighting (IPTW) to create a pseudopopulation in which exposure (IAH, severe AKI, or respiratory variables evaluated individually) was independent of confounders and mortality was the outcome. This MSM approach captured variability in exposure during the ICU stay, reflecting the fluctuating presence of exposure. To account for informative censoring due to ICU discharge, we also calculated the inverse probability of censoring weights (IPCWs). Marginal structural Cox models (MSCMs) were subsequently employed to estimate the causal association between IAH and mortality. Additionally, potential mediators (severe AKI and each respiratory variable—mechanical or oxygenation index) were evaluated against ICU mortality, adjusting for time-varying confounders. Because the variables exhibit complex interrelationships, making it challenging to ensure perfect model specification for all confounders, MSMs are a suitable choice given their flexibility in such scenarios in comparison with g-formulas [20].

We subsequently performed a mediation analysis in which severe AKI and each respiratory variable independently associated with mortality were potential mediators of the effects of IAH on mortality. The validity of mediation analysis relies on fundamental assumptions: (1) the absence of unmeasured confounding between the exposure (IAH) and the outcome (mortality); (2) no unmeasured confounding between the mediator (AKI stage 3) and the outcome; (3) correct temporal ordering, ensuring that exposure precedes the mediator, and the mediator precedes the outcome; and (4) correct model specification for both the mediator and outcome. While these assumptions are rigorous, we sought to minimize potential bias by including time-varying clinical markers in our models. However, we acknowledge that in the complex environment of the ICU, certain unmeasured factors may still influence these associations.

Prior to this mediation analysis, additional evaluations confirmed other assumptions necessary for mediation. The first assumption—that exposure (IAH) is associated with the final outcome mortality. The second assumption—that exposure (IAH) is linked to a potential mediator (AKI or respiratory variables)—and the third assumption—that possible mediators are related to mortality—were evaluated by performing MSM with each potential mediator as an exposure, followed by MSCMs to estimate causal effects between potential mediators and mortality, as explained above, accounting for time-varying confounding and IPCW.

Once these three assumptions were validated, mediation analyses were performed that incorporated time-varying exposures, mediators, and confounders. These analyses quantified the contribution of potential mediators to IAH-related mortality and estimated potential reductions in death rates if the mediator values reflected those observed in patients without IAH. Covariates included demographics, baseline comorbidities, ICU admission type (see above), and the aforementioned time-varying variables.

The recently developed survival mediational g-formula [21] allowed for time-varying mediation analysis suited to our data, accommodating the fluctuating severity and duration of IAH, as well as the dynamic mediator status and disease severity. SAS version 9.4 was used to generate the analysis dataset, and the R package ipw was utilized for MSM analyses. The meditational g-formula was implemented via the SAS macro [21]. Time-fixed confounders were included as predictors in all the models.

To evaluate the potential impact of selection bias and unmeasured confounders—arising from the restriction of our study to patients with available intra-abdominal pressure measurements—we performed a sensitivity analysis using the bounding method and summary measures [22]. By calculating the E-value [23] equivalents for selection and mediation, the magnitude of bias required to shift our observed estimates to the null is determined. The E-value principle was extended to mediation analysis [24], allowing us to estimate how robust our estimated natural direct and indirect effects are to potential violations. Standard errors were derived from 1000 bootstrap replications.

Results

Study population

According to the MIMIC-IV database, 50,920 adult patients were admitted to the ICU. Among these, 1944 patients had intra-abdominal pressure measurements during their initial ICU admission. The exclusion criteria included 212 patients with ICU stays shorter than 48 h, 351 patients who received KRT within the first 24 h of ICU admission, 165 patients whose baseline serum creatinine was ≥ 4 mg/dL or who were on maintenance KRT, and 661 patients whose data were missing for three or more consecutive days. Ultimately, 555 patients were included in the analysis. A flowchart detailing the inclusion and exclusion criteria is shown in Fig. 1, and Supplementary Table S1 displays the clinical differences between included patients and those who met the criteria but lacked intra-abdominal pressure measurements. Included patients presented with greater illness severity, characterized by higher SOFA scores, a higher incidence of AKI stage 3, impaired respiratory parameters, and increased mortality rates.

Fig. 1.

Participant flow charts for the MIMIC-IV cohort. ICU: intensive care unit; LOS: length of stay; KRT: kidney replacement therapy; sCr: serum creatinine; MIMIC-IV: Medical Information Mart for Intensive Care IV

Among these patients, 336 patients (60.5%) were male, with a mean age of 60.8 ± 15.0 years at ICU admission. Nearly half (n = 252; 45.4%) were initially admitted to a surgical ICU. The median nonrenal SOFA score at ICU admission was 7 (IQR, 4–10), and the median length of stay in the ICU was 8.0 (IQR, 4.0–14.9) days. During the ICU stay, almost all patients (n = 546; 98.4%) developed some stage of AKI, with 347 (62.5%) experiencing stage 3 AKI on at least 1 day. Additionally, 496 patients (89.4%) required mechanical ventilation. The overall mortality rate was 31.7% (n = 176). A complete description stratified by mortality status is provided in Table 1.

Table 1.

Patient demographics and characteristics at ICU admission according to the intra-abdominal pressure development of severe AKI (stage 2/3) during the first 16 days of ICU stay

	All patients (n = 555)	HI-IAP (n = 389)	Low-IAP (n = 166)	p
Age at ICU admission, years, mean ± SD	60.8 ± 15.0	59.8 ± 14.6	63.1 ± 15.9	0.009
Male sex, n (%)	334 (60.4)	247 (63.5)	87 (53.4)	0.03
Baseline sCr, mg/dL, median [IQR]	0.7 [0.5–1.0]	0.7 [0.5–1.0]	0.8 [0.5–1.0]	0.86
Charlson comorbidity index, median [IQR]	5 [3-7]	4 [2-7]	5 [3-7]	0.02
Previous conditions, n (%)
Diabetes mellitus	156 (28.2)	102 (26.2)	54 (33.0)	0.10
Chronic heart failure	115 (20.8)	71 (18.3)	44 (26.9)	0.02
Ischemic heart disease	200 (36.2)	147 (37.8)	53 (32.5)	0.24
Liver disease	184 (33.3)	142 (36.5)	42 (25.8)	0.01
COPD	151 (27.4)	101 (26.0)	50 (30.6)	0.36
Surgical admission, n(%)	251 (45.5)	174 (44.7)	77 (47.0)	0.57
Higher nonrenal SOFA, median [IQR]	7 [4-10]	7 [4-10]	6 [4-8]	0.01
Need of mechanical ventilation, n(%)	493 (89.3)	345 (88.7)	148 (90.8)	0.47
Need of vasoactive drugs, n(%)	400 (72.5)	281 (72.2)	119 (72.9)	0.85
AKI stage 3, n(%)	347 (62.5)	263 (67.6)	83 (49.4)	< 0.001
Respiratory parameters, [median, IQR]
Lowest PaO2/FiO2
Highest PEEP, cmH2O	135 [83–200]	126 [81–202]	133 [96–200]	0.11
Highest driving pressure, cm H2O	8 [5-13]	10 [5-12]	9 [5-12]	0.32
Lowest respiratory compliance, mL/cmH2O	10 [7-13]	11 [8-14]	9 [6-12]	0.03
	38 [38–53]	36 [27–50]	40 [30–56]	0.01
ICU LOS, days, median [IQR]	8.0 [4.0–14.9]	8.3 [4.0–15.6]	7.4 [3.9–13.3]	0.07
ICU mortality, n (%)	176 (31.7)	141 (36.2)	35 (21.1)	< 0.001

Binary time-varying variables (yes/no) were considered present if the patient experienced the event at least once during the ICU stay

ICU intensive care unit, eGFR estimated glomerular filtration rate, COPD chronic obstructive pulmonary disease, SOFA Sequential Organ Failure Assessment, PEEP positive end-expiratory pressure, LOS length of stay

IAH and mortality

To assess the association between IAH and mortality, we treated IAH as a time-dependent variable and employed MSMs. After adjusting for both fixed and time-varying covariates, we observed an adjusted hazard ratio (aHR) of 2.20 (95% CI 1.54–2.56) (see Fig. 2).

Fig. 2.

Adjusted cumulative survival according to the presence of intra-abdominal hypertension. The adjusted cumulative survival was calculated via baseline and time-varying confounder-adjusted survival curves via the marginal structural model. Hazard ratio: 1.98 (95% CI 1.54–2.56)

IAH is associated with stage 3 AKI and the need for KRT

We also investigated the association between IAH and AKI. No significant association was observed with AKI stages 1 and 2 (data not shown). However, IAH was significantly associated with stage 3 AKI (aHR 1.47; 95% CI 1.25–1.73). With respect to KRT, the association was even stronger, with an adjusted hazard ratio (aHR) of 2.22 (95% CI 1.69–2.87).

IAH and respiratory parameters

IAH was independently associated with the need for mechanical ventilation (aOR = 1.24; 95% CI 1.16–1.43). Among patients under mechanical ventilation (n = 496), the evaluated respiratory parameters (P/F ratio, PEEP, driving pressure, and compliance), including one parameter at each analysis and adjusting for the baseline and time-varying confounders, PaO2/FiO2, PEEP, and driving pressure, were independently associated with IAH (see Supplementary Table S2).

AKI stage 3 and KRT are associated with mortality

Although the association between AKI/KRT and mortality is well established, we conducted a formal mediation analysis using MSMs, as outlined in the methods section. AKI stage 3 and KRT were both significantly associated with mortality, with adjusted hazard ratios (aHR) of 3.31 (95% CI 2.49–4.41) and 2.24 (95% CI 1.61–3.11), respectively.

Respiratory parameters and mortality

Mechanical ventilation was independently associated with mortality (aHR 1.86; 95% CI 1.55–2.13). Prior to conducting the mediation analysis, we assessed the respiratory parameters linked to IAH that were also associated with the outcome in patients under mechanical ventilation. Among these parameters, PaO2/FiO2 and PEEP were significantly associated with mortality (see Supplementary Table S3).

AKI stage 3 and KRT mediates the majority of ICU mortality associated with IAH

In the initial analysis, we assessed whether mechanical ventilation, stage 3 AKI, and KRT act as significant mediators in the association between IAH and mortality. To achieve this, the G-formula was used to estimate the reduction in mortality disparity between patients with and without IAH under a hypothetical scenario in which all relevant mediators are balanced—that is, the frequency of mechanical ventilation, AKI stage 3, and KRT is comparable across groups. Figure 3 presents a directed acyclic graph outlining the primary causal pathways. In this simulated context, AKI stage 3 emerged as a key mediator of IAH-related mortality. The adjusted hazard ratio (aHR) for mortality in patients with IAH decreased from 2.20 to 1.65 (95% CI 1.42 to 1.77). These findings suggest that approximately 45.7% (p < 0.001) of the excess mortality associated with IAH can be attributed to the need for mechanical ventilation (4.2%) and mainly to the development of stage 3 AKI (41.5%). However, 54.3% of the mortality difference would remain, likely due to direct effects of IAH, other organ dysfunctions, or unmeasured mechanisms. We also examined the use of KRT as a secondary mediator following AKI development and its effect on IAH-related mortality. In this mediation model, KRT was associated with a 15.7% reduction (p < 0.001) in the estimated mortality hazard.

Fig. 3.

Simplified causal diagram illustrating the relationships among intra-abdominal hypertension (IAH), acute kidney injury (AKI) stage 3, kidney replacement therapy (KRT), mechanical ventilation (MV), and mortality for the first two time points. Baseline covariates (sex, age, comorbidities, type of ICU admission, and baseline renal function) are omitted, but they affect all variables in the graph. Time-varying confounders: nonrenal Sequential Organ Failure Assessment (SOFA) score; need for vasoactive support. AKI stage 3 accounts for 41.5% of IAH-related excess mortality, the need for MV per se accounts for 4.2%, and KRT has a protective effect, reducing IAH-associated mortality by 16%.

We further examined, among patients receiving mechanical ventilation (n = 496), whether respiratory parameters associated with both exposure and outcome—specifically PaO2/FiO2 and PEEP—mediated the relationship between IAH and mortality. No significant mediation effects were observed, with mediation percentages of − 0.3% (p = 0.66) and 0.5% (p = 0.52), respectively. This subpopulation analysis also confirmed the role of AKI stage as a primary mediator (39.3%) and KRT was associated with a 16.6% reduction in estimated mortality.

Sensitivity analysis

We performed two sensitivity analyses: the first to evaluate the potential impact of selection bias and the second to assess an alternative intra-abdominal pressure threshold > 20 mmHg. The first analysis indicated that selection bias could only account for the observed direct effect if an unmeasured confounder—such as unrecorded clinical indications for IAP monitoring—increased the risk of both IAH and mortality by 2.69-fold. Regarding the indirect effect, an association strength of 2.00 would be required for an unmeasured variable to negate the findings for both stage 3 AKI and mortality.

Using the 21 mmHg threshold, IAH remained significantly associated with mortality (aHR 3.41; 95% CI 2.59–4.23). While no significant mediation effect was observed for mechanical ventilation, AKI stage 3 persisted as a primary mediator, accounting for approximately 49.0% (p < 0.001) of the excess mortality associated with IAH. Conversely, KRT was associated with a protective mediated effect of 11.1%.

Discussion

In this study, using time-varying exposures, mediators, and confounders, we assessed the role of severe AKI and pulmonary impairment as mediators of IAH-related mortality in critically ill patients. Our main finding is that stage 3 AKI accounts for more than 40% of the mortality attributable to IAH, whereas KRT was associated with a protective effect, reducing AKI-related mortality by 16%. We also identified a small, although statistically significant, mediating effect of the need for mechanical ventilation. However, we found no ventilatory parameter that acted as a mediator between IAH and mortality.

Our findings align with several previous studies on the development of IAH. First, we confirmed that IAH is not exclusively a postoperative complication, as the proportions of patients admitted for medical and surgical reasons were similar [1]. Second, even after applying what is, to our knowledge, a novel approach in IAH research—a longitudinal, time-based analytical framework—we support other studies [1, 2, 25, 26] that IAH is associated with a markedly worse prognosis, effectively doubling the risk of early mortality. Third, although we did not observe a significant association between IAH and the development of less severe AKI (stages 1 and 2), our distinct methodological approach revealed that IAH was associated with a nearly 50% greater risk of developing stage 3 AKI.

The relationship between IAH and AKI remains somewhat controversial [3, 18, 27, 28]; however, most studies suggest that IAH is an important risk factor for AKI [29]. For example, using the same database used in the present study, Miao et al. [18] reported that patients with IAH at ICU admission had a 40% increased likelihood of developing AKI within 72 h. Because AKI-related complications can further worsen IAH—for example, through the development of a positive fluid balance—this relationship is likely bidirectional, an aspect that our methodological approach is capable of capturing (Fig. 3).

The main finding of our study is the identification of the critical role of stage 3 AKI as a mediator in the potential causal relationship between IAH and early mortality. While this is somewhat expected, given that the kidneys are among the organs most affected by IAH, several aspects of this causal association warrant further discussion. A key assumption in our causal diagram is that AKI arises as a complication of IAH, a premise strongly supported by existing pathophysiological and experimental evidence. Although the precise mechanisms by which IAH causes renal injury are not yet fully understood, multiple pathways have been proposed. First, increased intra-abdominal pressure can compress or obstruct the renal veins, leading to decreased renal arterial inflow and a reduction in the transcapillary hydrostatic pressure gradient essential for glomerular filtration. For example, Doty et al. [30] demonstrated that elevating renal venous pressure to 30 mmHg significantly reduced the glomerular filtration rate from 26 to 8 mL/min in pigs. Second, further increases in intra-abdominal pressure can impair cardiovascular function by decreasing cardiac preload, impairing myocardial contractility, and increasing afterload, all of which contribute to a decline in cardiac output [7]. On the other hand, we acknowledge that specific variables, such as fluid balance, are closely interconnected with both IAH and AKI [31], and as stated above, our causal analysis explicitly incorporates and adjusts for these relationships, considering how AKI impacts fluid balance and, conversely, how fluid balance may influence IAH.

Mortality mediation was observed only for stage 3 AKI. We speculate this may be explained by our time-varying assessment of renal function, which captures the temporal dynamics of the insult. By accounting for the duration of AKI, this model recognizes that milder stages are frequently transient—reflecting functional fluctuations rather than structural injury—and thus exert a diminished effect on mortality [32] compared to traditional fixed-variable analyses.

We also explored the potential mediating role of respiratory impairment in mortality associated with IAH. Although mechanical ventilation alone had a statistically significant yet modest mediating effect, its overall impact was limited. Among ventilated patients, several parameters potentially influenced by IAH were evaluated. Only the PaO2/FiO2 ratio and PEEP demonstrated associations consistent with being related to both IAH and mortality, but neither served as a significant mediator. The finding that respiratory parameters did not mediate the association between IAH and mortality warrants careful interpretation, given the established physiological impact of IAH on lung mechanics. This lack of observed mediation may stem from several factors. First, standard ventilatory variables may not fully capture the complex respiratory derangements induced by IAH. Unlike the AKI staging system, there is no standardized, validated severity scale for respiratory mechanics that could precisely quantify the degree of pulmonary insult. Furthermore, respiratory data are highly influenced by clinician-driven ventilator adjustments, which may mask the primary effect of IAH. Finally, the use of 3-day time blocks in our model may have overlooked transient or indirect respiratory changes that occur on a more granular temporal scale. Consequently, while a direct biological link is plausible, these data limitations likely hindered the detection of a significant mediatory effect in our analysis.

Our study has several limitations. First, as an observational, retrospective analysis, definitive causal relationships cannot be established. Although we mitigated bias from time-varying confounding by treating both the primary exposure and mediators as time-varying variables, the MSMs and mediational g-formula remain susceptible to violations of confounding assumptions and model misspecification. To address this, we adjusted for a comprehensive set of covariates reflecting dynamic illness severity; however, certain factors, such as the initiation of KRT, are largely influenced by clinical decision-making, making it challenging to fully quantify and adjust for all underlying variables. Another limitation is the selection bias introduced by including only patients whose intra-abdominal pressure was measured. While considering patients without intra-abdominal pressure measurements as having normal values, like imputation strategies used for variables such as bilirubin within severity scores such as SOFA [33], which might seem viable, prospective studies indicate that IAH may affect up to 50% of critically ill patients [1], and this imputation approach could introduce even greater bias. While we cannot entirely overcome this limitation, we performed a sensitivity analysis to quantify its potential impact. This analysis indicated that our main results would remain consistent even if unmeasured confounding factors contributing to selection bias—associated with a twofold increase in the risk of mortality or stage 3 AKI—were present, notwithstanding adjustments for severity markers such as non-renal SOFA score and organ support.

In conclusion, our analysis highlights the critical role of AKI, particularly stage 3 AKI, in mediating the increased mortality associated with IAH in critically ill patients, and KRT was associated with a reduction in the estimated mortality hazard. Second, our study did not identify a significant mediating role for respiratory parameters, suggesting that the relationship between IAH and mortality is driven primarily by renal mechanisms. Given the significant proportion of IAH-related mortality mediated by AKI, vigilant monitoring and implementation of practical strategies to prevent stage 3 AKI development, such as optimized fluid management and avoidance of nephrotoxic agents, are essential in this population.

Authors’ contributions

Conceptualization: A.B.L.; data curation: A.B.L., W.R.N.; formal analysis: A.B.L.; methodology: A.B.L.; writing—original draft: A.B.L., W.R.N.; writing—review and editing: A.B.L., W.R.N. All authors reviewed the manuscript.

Funding

None.

Data availability

The data supporting the study findings are available upon reasonable request after approval of a proposal from the corresponding author.

Declarations

Ethics approval and consent to participate

Not applicable.

Competing interests

The authors declare that they have no competing interests.