Acute kidney injury in elderly patients receiving invasive mechanical ventilation: early versus late onset

Abstract

Background

Acute kidney injury (AKI) is a severe complication in critical patients receiving invasive mechanical ventilation (MV). However, AKI which occurs in the first 48 h after MV (early AKI) and thus likely associated with the MV settings is probably different from AKI occurring following 48 h (late AKI). This study is aimed at exploring the incidence of early and late AKI in elderly patients receiving MV and identifying their different risk factors and outcomes.

Methods

This retrospective, observational, multicenter cohort study consecutively included 3271 elderly patients (≥ 75 years) receiving invasive MV at four medical centers of Chinese PLA General Hospital from 2008 to 2020. The diagnosis of AKI was made following the 2012 KDIGO criteria and categorized into early (≤ 48 h) or late (> 48 h-7 days) according to the time from MV.

Results

There were totally 1292 cases enrolled for the final analysis. Among them, 376 patients (29.1%) developed early AKI versus 132 (10.2%) developed late AKI. The 28-day mortality rates of the non-AKI, early AKI, and late AKI patients were 14.4, 46.8, and 61.4%, respectively. After 90 days, mortality rates of three groups were 33.2, 60.6, and 72.7%, respectively. Risk factors for early AKI included PaO₂/FIO_2, serum creatinine, hemoglobin, and positive end-expiratory pressure at the beginning of MV, while those for late AKI were PaO₂/FIO_2, serum creatinine, and hemoglobin. In the multivariable adjusted analysis, both early AKI (HR = 4.035; 95% CI = 3.166–5.142; P < 0.001) and late AKI (HR = 6.272; 95% CI = 4.654–8.453; P < 0.001) were related to the increased 28-day mortality relative to non-AKI. AKI was significantly related to 90-day mortality: early AKI (HR = 2.569; 95% CI = 2.142–3.082; P < 0.001) and late AKI (HR = 3.692; 95% CI = 2.890–4.716; P < 0.001).

Conclusions

AKI mostly develops in the initial 48 h following MV, which is related to the health and MV settings; while AKI occurring following 48 h is not associated with MV settings. Therefore, a strategy for kidney protection in patients with MV should take these differences into consideration.

Introduction

Invasive mechanical ventilation (MV) is a common yet high-risk intervention performed among critical patients [1, 2]. However, the life-saving tool can induce an increased peri-intubation complication rate [3, 4]. A prospective observational study including 2964 cases demonstrated that cardiovascular instability had the highest incidence rate after tracheal intubation [5]. However, the referenced study only focused on complications in 30 min following incubation initiation, and the short study time window might have missed other organ complications, including acute kidney injury (AKI) [6, 7].

AKI, previously termed acute renal failure, is characterized by a sudden decrease in renal function, primarily described in recent years using the widely accepted 2012 Kidney Disease Improving Global Outcomes (KDIGO) classification based on changes in the serum creatinine (SCr) level and/or urine output [8]. AKI is the frequent complication with various etiologies among critically ill patients receiving invasive MV. According to van den Akker et al. in their systematic review, the AKI risk increased by 3 times in patients receiving invasive MV, and various positive end-expiratory pressure (PEEP) or tidal volume settings did not influence renal function [9]. During the treatment of patients with invasive MV at our center, we found that these patients commonly developed AKI in the first 48 h of the post-intubation period, and the AKI incidence rate was higher than that of 15.5%–17.1% reported in previous studies [10, 11]. The impact of PEEP on AKI occurrence among such cases remains uncertain owing to either the design of the studies or a lack of clear AKI definitions.

It is vital to recognize AKI early and take intensive surveillance and treatments for limiting subsequent AKI or further disease progression, aiming to lower morbidity and mortality [12,13,14]. However, AKI that occurs in the initial 48 h post-MV (early AKI) and may thus be associated with MV settings can be different from AKI occurring following 48 h (late AKI). Therefore, this study aimed to investigate incidence and risk factors associated with AKI at different time intervals among critical elderly patients under MV and examine potential relationship of AKI with outcomes.

Methods

Study design

This retrospective, observational, multicenter cohort study consecutively included elderly patients (≥ 75 years) receiving invasive MV from four medical centers of the Chinese People’s Liberation Army (PLA) General Hospital between January 2008 and December 2020. Our study design was approved by Clinical Ethics Committee of the Chinese PLA General Hospital (number: S2023–725–01). Due to the observational retrospective nature, no informed consent was needed. Patient data were anonymous and deidentified. This study followed the Declaration of Helsinki.

Study patients

Critical elderly patients who developed the fatal cardiovascular, neurological or respiratory system impairment requiring in-hospital intubation were included from every center in our study period. Among them, those receiving ≥ 48-h ventilation or those having ≥ 2 measurements of SCr in the initial 48 h of MV were enrolled into this study. Patients developing chronic kidney disease (CKD) stage 4–5, patients undergoing nephrectomy, renal replacement therapy prior to MV, kidney transplantation, and patients without available SCr measurement or just one SCr examination, without available or insufficient medical history, hospitalized < 48 h and receiving MV post-operatively for general anesthesia were excluded from this study. Patients undergoing AKI before MV were also excluded to avoid the inclusion of patients suffering from constant kidney failure because this study was aimed at defining risk factors for de novo AKI (AKI developed in the first 48 h of MV).

Currently, no definite AKI definition is available for its diagnosis. Therefore, the existing 2012 KDIGO definition was the main diagnostic standard [15]. Early AKI was diagnosed with reference (exclusively) to the patient’s SCr level, to be a specific, by an SCr increase of ≥ 26.5 µmol/L within 2 days after MV; late AKI was diagnosed where as a ≥ 1.5-fold increase compared with the baseline value after 2 days (up to a week). The greatest SCr level following AKI was used to identify AKI stage in line with KDIGO criteria. Day 0 represented the MV starting day, and day 1 started at 8:00 a.m. on the following day. Difference in SCr represented the maximal change of SCr level on day 0, 1, or 2 [10]. Baseline SCr was the latest determination within the last 3 months [16]. If previous SCr measurement was unavailable, the minimum SCr level in the hospital stay was regarded as baseline SCr concentration [6].The definition of sepsis was made in line with third international consensus definitions for sepsis and septic shock (Sepsis-3) criteria [17]. If patients were hospitalized several times, only the first hospitalization was used in the analysis set.

Data extraction

Clinical information, which included demographic data (age, sex, body mass index [BMI]), underlying diseases (history of hypertension, coronary disease, cerebrovascular diseases, chronic obstructive pulmonary disease [COPD], diabetes mellitus and CKD), necessity of renal replacement therapy, mean arterial pressure and cause of MV, was extracted. In addition, laboratory data assessed on day 0 were uric acid, blood urea nitrogen (BUN), C-reactive protein, hemoglobin, serum albumin, and neutrophil–lymphocyte ratio. PEEP, partial pressure of oxygen in arterial blood (PaO₂), arterial carbon dioxide tension (PaCO₂), fraction of inspired oxygen (FIO₂), and PaO₂/FIO₂ were determined. Estimated glomerular filtration rate (eGFR) was identified with the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) method [18].

Statistical analysis

Continuous variables were shown to be mean ± standard deviation or median (25%–75% interquartile range) in accordance with data distribution. Discrete data were represented by counts or percentages. Student’s t-test or Mann–Whitney U test was employed to compare continuous data between groups, while one-way ANOVA or Kruskal–Wallis H test was applied to compare continuous data among 3 groups. Pearson’s Chi-square or Fisher’s exact test was applied to compare categorical data. Risk factors related to AKI were identified through forward step wise selection and logistic regression. The Cox proportional hazards regression model was applied to determine prognostic survival factors. Probabilities of survival were explored by Kaplan–Meier analysis, while log-rank test was adopted for comparing curves. R version 4.2.2 and SPSS 21.0 (SPSS Inc., Chicago, IL, USA) were used in statistical analysis. P < 0.05 represented significant differences.

Results

General characteristics of patients included in the study

From January 2008 to December 2020, totally 3,271 elderly patients underwent invasive MV. Among them, 1,979 patients were eliminated, the final study cohort included 1292 consecutive patients. Figure 1 displays a flowchart of this study. The median age of these enrolled patients was 89 years (85–92), with the majority (1171, 90.6%) being male. Hypertension (72.4%), coronary disease (71.1%), and cerebrovascular diseases (53.6%) were the most common comorbid conditions. Table 1 exhibits the baseline characteristics and outcomes of the study population.

Fig. 1

Flowchart of the patient inclusion and exclusion process. AKI, Acute kidney injury; SCr, serum creatinine; CKD, chronic kidney disease; RRT, Renal replacement therapy; MV, mechanical ventilation

Table 1 Characteristics of patients with AKI and without AKI

Baseline characteristics related to early and late AKI

In general, the 29.1% of cases (376/1292) developed AKI in the initial 48 h of MV (early AKI), and the severity distribution for early AKI was as follows, 31.4% at stage 1, 35.1% at stage 2, and 33.5% at stage 3. There were 8.8% of AKI cases needing renal replacement therapy. Another 10.2% patients (132/1292) had AKI from 48 h to 7 days after MV (late AKI), with patients at stages 1, 2, and 3 occupying 49.2%, 28.8%, and 22.0%. In addition, no patients required renal replacement therapy.

As shown in Table 1, early AKI and late AKI patients were compared, suggesting that age (88 vs. 89 years, P = 0.083), sex (90.7% vs. 90.2%, P = 0.855) and BMI (22.6 ± 2.7 vs. 23.0 ± 3.0 kg/m², P = 0.229) were of no significant difference. Similarly, differences between the two cohorts regarding preexisting comorbidities (hypertension P = 0.413, coronary disease P = 0.716, cerebrovascular diseases P = 0.092, diabetes mellitus P = 0.273, CKD P = 0.114) were not significant. Most early AKI patients had COPD (63.3% vs. 37.1%, P < 0.001). Difference in baseline median Scr level (78.0 vs. 90.0 µmol/L, P < 0.001) was significant between the two cohorts.

Clinical characteristics related to early and late AKI

Early AKI patients showed significantly decreased SCr (99.8 vs. 110.7 μmol/L, P < 0.001) and BUN (11.7 vs. 15.2 mmol/L, P < 0.001) contents as well as decreased uric acid level (313.3 vs. 354.0 μmol/L, P = 0.032) on day 0 than patients with late AKI. Early AKI patients also had higher albumin (34.1 ± 4.1 vs. 33.2 ± 4.5 g/L, P = 0.035) and hemoglobin (111 ± 21 vs. 101 ± 20 g/L, P < 0.001) levels when compared with late AKI cases. Mean arterial pressure (75 ± 22 vs. 79 ± 22 mmHg, P = 0.082), PO₂ (62.1 vs. 65.4 mmHg, P = 0.630), PCO₂ (44.7 vs. 45.5 mmHg, P = 0.579) and PaO₂/FiO₂ (129.1 vs. 133.6 mmHg, P = 0.644) levels were not significantly different. Relative to late AKI patients, early AKI patients frequently used a higher PEEP (1–3 cmH₂O: 13.3% vs. 10.6%; ≥ 4 cmH₂O: 26.1% vs. 12.9%, P = 0.003).

Relationship of AKI with mortality in elderly patients receiving MV

The 28-day mortality rates of non-AKI, early AKI, and late AKI patients were 14.4, 46.8, and 61.4%, respectively. After 90 days, the mortality rates of three groups were 33.2, 60.6, and 72.7%, respectively. Based on Kaplan–Meier curves, survival for non-AKI patients was superior to AKI patients, while that of early AKI patients was superior to late AKI patients (log rank P < 0.0001; Figs. 2, 3). Short-term mortality of AKI group exhibited a steady increasing trend with the increasing AKI severity (log rank P < 0.001; not shown).

Fig. 2

Kaplan–Meier survival curves for 28-day mortality in patients with early AKI, late AKI and without AKI during the time after mechanical ventilation (log-rank test: P < 0.0001). AKI, Acute kidney injury

Fig. 3

Kaplan–Meier survival curves for 90-day mortality in patients with early AKI, late AKI and without AKI during the time after mechanical ventilation (log-rank test: P < 0.0001). AKI, Acute kidney injury

Based on multivariate regression, AKI independently predicted the all-cause 28-day mortality (early AKI vs non-AKI: HR = 4.035; 95% confidence interval [CI]: 3.166–5.142; P < 0.001 and late AKI vs non-AKI: HR = 6.272; 95% CI: 4.654–8.453; P < 0.001), and 90-day mortality (early AKI vs non-AKI: HR = 2.569; 95%CI: 2.142–3.082; P < 0.001 and late AKI vs non-AKI: HR = 3.692; 95% CI: 2.890–4.716; P < 0.001) (Table 2).

Table 2 Multivariate Cox proportional hazard model analysis of risk factors for mortality

Risk factors related to early and late AKI

Multivariate logistic regression identified PaO₂/FIO₂ (odds ratio (OR) = 0.998; 95% CI:0.996–0.999; P = 0.045), SCr (OR = 1.029; 95% CI:1.023–1.035; P < 0.001), hemoglobin (OR = 1.011; 95% CI 1.004–1.017; P = 0.003), and PEEP (1–3 cmH₂O: OR = 1.735; 95% CI 1.143–2.635; P = 0.010; ≥ 4 cmH₂O: OR = 3.476; 95% CI 2.391–5.054; P < 0.001) on day 0 as risk factors related to early AKI in elderly MV patients (Table 3); while PaO₂/FIO₂ (OR = 0.997; 95% CI 0.994–0.999; P = 0.038), SCr (OR = 1.035; 95% CI 1.027–1.042; P < 0.001), and hemoglobin (OR = 0.986; 95% CI 0.976–0.996; P = 0.006) on day 0 were found to be risk factors related to late AKI (Table 3).

Table 3 Factors associated with AKI as indicated by logistic regression

Discussion

AKI occurring within and following 48 h of critical patients receiving MV was different in terms to risk factors and outcomes. Although elderly patients are related to a higher AKI risk, data on AKI prevalence or clinical implication under MV are lacking. This study provides novel findings by analyzing the relationship of AKI risk factors with mortality. Firstly, nearly 30% of inpatients developed AKI in the initial 48 h during MV, another 10.2% of patients developed AKI during 48 h–7 days after MV. Second, risk factors and clinical outcomes for early AKI were different from those late AKI. Actually, we found that PEEP was a risk factor for early rather than late AKI. Third, patients with stage 2 AKI exhibited a poorer prognosis, and those with stage 3 AKI showed the poorest prognostic outcome. Therefore, identifying AKI early plays a crucial role in initiating further assessment and treatment, thus avoiding subsequent kidney injury and decreasing mortality.

Clinical results identified hypercapnia, hypoxemia, great tidal volume and large PEEP level as risk factors associated with AKI occurrence among MV patients [3]. First, it has been indicated that mild hypoxemia in MV patients exerts no significant impacts on renal function. Severe hypoxemia (PaO₂ < 40 mmHg) can induce renal vasoconstriction, which causes low renal blood flow perfusion and decreased GFR, finally inducing ischemic injury of kidney tissues and cells [3]. Only 3% of patients had a PaO₂ < 40 mmHg at the initiation of MV, and one possible explanation might be that our enrolled patients were elderly with many underlying diseases, poor pulmonary function, and long-term oxygen therapy. Therefore, in our data, PaO₂ might not completely or objectively reflect actual respiratory function in the body. Second, hypercapnia independently influences kidney function. It can influence renal function even when the PaO₂ is normal or increases [19]. Therefore, compared to hypoxemia, hypercapnia is a more vital factor that can make critical impacts on renal function. Hypercapnia can directly or indirectly stimulate the release of norepinephrine and excite sympathetic nerves to cause renal vasoconstriction, inducing decreased GFR and renal blood flow. According to our results, 48.8% of patients had PaCO₂ > 50 mmHg, implying that when elderly patients suffered from AKI under MV, almost 50% were also hypercapnic. However, based on multivariate logistic regression, PaCO₂ did not independently predict the risk of the development of AKI. Third, injurious MV strategies can also impact the kidney by inducing abnormal hemodynamics, fluid reactive shock, and hypotension, and influencing renal perfusion by reducing GFR via declining cardiac output while promoting sympathetic and hormonal pathways. In addition, by manipulating permissive hypoxemia or hypercapnia, MV can also result in reduced GFR, renal hypoperfusion, or functional renal insufficiency [6]. Fourth, PEEP has been identified as a vital part of ventilatory strategy, which can maintain the opening of recruited alveoli and decrease intrapulmonary shunts to improve oxygenation. In addition, lung recruitment mediated by PEEP decreases alveolar overdistention. PEEP can reduce the repeated opening and the closing of alveoli in respiratory cycle, therefore avoiding pulmonary injury [20]. A systematic review and meta-analysis was conducted to assess the association of higher PEEP (median PEEP, 9 cm H₂O) as opposed to lower PEEP (median PEEP, 0 cm H₂O), with hospital mortality in adult intensive care unit patients undergoing MV for reasons other than acute respiratory distress syndrome (ARDS). These findings show that higher PEEP compared with lower PEEP was not associated with mortality in patients without ARDS receiving MV [21]. In our analysis, over three-fourths (75.1%) of patients had a zero PEEP, with 25% of the patients having a PEEP of ≥ 1 cm H₂O. However, invasive MV with PEEP was related to a poor prognosis.

Several studies have been made to examine the relationship of MV settings with AKI risk following different diagnostic criteria. Conflicting results have been reported. For example, Ottolina D et al. included totally 101 COVID-19 patients admitted to ICU in their retrospective observational study, and their median age was 61 (53–68) years. The authors reported that 38% had AKI (KDIGO AKI stage 2 or 3), and higher PEEP (median PEEP, 14.7 cm H₂O) patients showed a higher risk of developing AKI than low PEEP (median PEEP, 9.6 cm H₂O) patients [22]. Lombardi R et al. performed the prospective observational multicenter study in 494 ICUs from 32 countries including 3206 study subjects (average age, 58 years). Using the Scr-based KDIGO criteria, the authors found that 17.1% (547/3206) of early AKI occurred in the initial 48 h following MV, while 4.9% (157/3206) of late AKI occurred between days 3 and 7. Their results showed that AKI patients had a higher mortality risk than non-AKI cases. However, no independent association was found between PEEP and AKI, and only higher levels of peak inspiratory pressure and tidal volume were shown to independently predict the early AKI risk [11]. By contrast, Lombardi R et al. published another prospective study of 2783 adult critical patients receiving MV. The diagnosis of AKI was made following the Acute Kidney Injury Network (AKIN) criteria by an absolute elevation of SCr ≥ 26.4 μmol/L and/or a percentage elevation ≥ 50% in the initial 48 h following MV. These authors reported that 803 (28.8%) patients developed AKI following these two criteria, while just 15.5% of AKI cases fulfilled the diagnostic criteria based on the absolute elevation of SCr ≥ 26.4 μmol/L. However, tidal volume with/without PEEP was not independently correlated with AKI [10]. Similarly, McNicholas BA et al. found that there was no difference in PEEP or tidal volume in patients with AKI compared to patients without AKI (KDIGO criteria) [23]. The discrepancies among these studies might be caused by various definitions of AKI, specific study populations, ages, and clinical settings among studies.

In general, critical elderly patients are associated with an increased AKI risk during invasive MV. Therefore, it is crucial to protect the kidney when treating patients under invasive MV. Importantly, clinicians should learn the kidney pathophysiological characteristics, execute sepsis control, hemodynamic management, protective MV, and carefully monitor renal function, aiming to improve prognosis in early AKI patients with invasive MV.

Strengths of this study include the elderly age of the sample, the use of a consensus definition for AKI/CKD diagnosis, the early AKI (48-h time windows) and late AKI (7-day time windows) of the KDIGO guidelines and stages, and baseline SCr being available in the entire sample of included patients. However, this study still had the following limitations. Firstly, AKI was diagnosed in line with the elevation of SCr level, this could have resulted in an under estimation concerning the true incidence of AKI. Secondly, because of our specific patient population with a predominance of elderly males, it should be cautious when generalizing our results to other populations. Thirdly, specific key data were unavailable in the database, including ventilator settings, daily fluid balance or type, hospital stay length, MV duration, illness severity. Finally, as patient enrollment has been accrued in over a decade, management strategies of critical ill patients may have changed during this time interval. Therefore, the prognostic impact of early vs late AKI could have changed across the study time interval.

Conclusions

To conclude, early AKI can be frequently observed among elderly patients receiving MV, and approximately 30% of inpatients develop AKI in the initial 48 h after MV. Early AKI shows great difference from late AKI with regard to baseline data, risk factors and short-time outcomes.