Hospital-acquired acute kidney injury in very elderly men: clinical characteristics and short-term outcomes

Objectives We explored the risk factors for, and the clinical characteristics of, acute kidney injury (AKI), and the causes of death 28 days after such injury, in very elderly men. Methods This was a retrospective cohort study using data from the Geriatric Department of the Chinese PLA General Hospital. A total of 3464 elderly patients (≥ 75 years) were enrolled from January 2007 to December 2015. All patients were followed for 28 days or until death after AKI. Results In total, 668 patients (39.0%) developed AKI, and 623 men were included for the final analysis. The median age was 87 years. The 28-day mortality rate was 25.7%. The AKI etiologies were infections (39.6%), hypovolemia (23.8%), cardiovascular events (15.9%), nephrotoxicity (12.0%), and surgery (7.1%). Multiple organ dysfunction syndrome (46.4%) and pulmonary infection (22.5%) were the principal causes of death. Multivariate analysis revealed that time for AKI to develop (HR = 0.865; 95% CI 0.799–0.937; P < 0.001), low mean arterial pressure (HR = 0.970; 95% CI 0.958–0.981; P < 0.001), low serum prealbumin (HR = 0.924; 95% CI 0.894–0.955; P < 0.001) level, oliguria (HR = 2.261; 95% CI 1.424–3.590; P = 0.001), mechanical ventilation (HR = 1.492; 95% CI 1.047–2.124; P = 0.027), blood urea nitrogen (HR = 1.037; 95% CI 1.025–1.049; P < 0.001) level, magnesium (HR = 2.512; 95% CI 1.243–5.076; P = 0.010) level, and more severe AKI stages (stage 2: HR = 3.709; 95% CI 1.926–7.141; P < 0.001 and stage 3: HR = 5.660; 95% CI 2.990–10.717; P < 0.001) were independent risk factors for 28-day mortality. Conclusions The incidence of AKI increases significantly as age advanced. Identification of risk factors might lead to more intensive monitoring and early prevention, and might improve AKI patients’ outcomes in the very elderly.


Introduction
Acute kidney injury (AKI), previously termed as acute renal failure (ARF), has attracted increasing attention in recent decades because of its rising incidence and its adverse effect on patients. In the elderly, age-dependent structural and functional changes make an old kidney more sensitive to AKI [1,2]. An analysis showed that the incidence of AKI increases stepwise from 24.9 episodes (75-79 years) to 34.2 episodes (80-84 years) to 46.9 episodes (85 years and older) per 1000 patient-years, respectively [3]. Despite this, few studies have specifically focused on elderly patients with AKI.
AKI has been recognized as a surrogate marker of the severity of illness. This complex clinical disorder is closely associated with both short-and long-term mortality, the length of hospital stay, as well as a predictor of chronic kidney disease (CKD) [4][5][6][7]. However, focusing on the longterm outcomes of geriatric AKI individuals may often be difficult because they have significantly poor prognosis after hospital discharge, and survivors are often left with multiple chronic conditions [2,4]. The long-term survival rate might be reduced significantly; so short-term consequences, 1 3 such as in-hospital mortality, may be more important and practical.
AKI has become a huge medical burden in China [8]. The clinical features of elderly AKI patients in China are poorly known, and very few studies on risk factors predicting shortterm (28-day) outcomes and causes of death post-AKI have appeared. In the present study, we described the risk factors for mortality among AKI clinical characteristics, and the causes of death by 28 days post-AKI in very elderly patients.

Patients and methods
This was a retrospective cohort study performed in the Geriatric Department of the Chinese PLA General Hospital. We collected data on very elderly patients (≥ 75 years of age) treated from January 1, 2007, to December 31, 2015. All patients who developed AKI were enrolled. All patients who developed AKI were enrolled. Patients were divided into survivor and non-survivor groups by their outcomes within 28 days after AKI. We sought clinical risk factors for death and explored their effects.
AKI was diagnosed by reference (exclusively) to SCr level, thus by an SCr increase ≥ 0.3 mg/dl (≥ 26.5 µmol/L) within 48 h, or an increase to ≥ 1.5-fold the baseline value, known or presumed to have developed within the prior 7 days [9]. AKI severity was scored by reference to the difference between SCr baseline and peak values, using the KDIGO staging criteria: Stage 1, an increase ≥ 26.5 µmol/L, or a rise to ≥ 1.5-to 1.9-fold the baseline value; Stage 2, a rise to ≥ 2.0-to 2.9-fold the baseline value; Stage 3, a rise to ≥ 3.0-fold the baseline value, or to ≥ 353.6 μmol/L, or initiation of renal replacement therapy. We noted age, AKI etiology, gender, body mass index (BMI), co-morbidities (hypertension, chronic obstructive pulmonary disease [COPD], coronary disease, diabetes mellitus), recently performed surgical operations, need for dialysis, need for MV, and cause of death. MV was defined as ventilation for at least 24 h prior to AKI development. Oliguria was defined as urinary output < 400 ml/24 h. Urine output, mean arterial pressure (MAP), and the levels of SCr, blood urea nitrogen (BUN), uric acid, hemoglobin, serum albumin, serum prealbumin, serum magnesium, serum calcium and serum phosphate were also noted at the time of AKI diagnosis. The baseline SCr level was the most recent stable measure taken in the 1-3 months prior to admission with AKI [10,11]. Estimated glomerular filtration rates (eGFR) were calculate by the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) [12].
We excluded patients younger than 75 years, those with previously diagnosed CKD [13], those who stayed in the hospital for less than 48 h, those who had only one SCr or no SCr examination, those who had missing or incomplete medical history and those who had early death within 48 h after admission. We also excluded females, because we treated fewer females than males.

Statistical analysis
Continuous variables were presented as means ± standard deviations (SDs), or the median (25-75% interquartile range), depending on the variable distribution. Discrete variables are presented as counts or percentages. Statistical analyses were performed using SPSS version 17.0 for Windows (SPSS Inc., Chicago, IL, USA). Between-group comparisons were made using Student's t-tests or Mann-Whitney U-test. Correlations between potential risk factors and death were assessed using Pearson's Chi-squared or Fisher's exact tests. Survival curves were estimated by the Kaplan-Meier product-limit method and compared by the Mantel (log-rank) test. Prognostic factors of survival were identified by the use of the Cox proportional hazards regression model. A P value < 0.05 was considered to reflect statistical significance.

Baseline data of the enrolled patients
Among 1711 study patients, 668 (39.0%) developed AKI. Only 45 patients were excluded, resulting in 623 AKI patients suitable for analysis. The study flow chart is presented as Fig. 1. Table 1 shows the age-related incidence of AKI in the elderly patients.
According to the Kaplan-Meier plot, the survival curve of patients with no MV had a significantly better survival than patients with MV (Fig. 3b). Patients diagnosis AKI > 2 days had better survival than diagnosis AKI in 2 days (Fig. 3c). Patients with prealbumin level ≥ 181 g/L had better survival than patients with lower level (Fig. 3d). Patients with BUN level ≥ 12.8 mmol/L were significantly worse than the patients with lower BUN level (Fig. 3e). Patients with hypomagnesemia had better survival than patients with hypermagnesemia (Fig. 3f).

Discussion
We retrospectively studied very elderly patients with AKI over 28 days of follow-up to understand their characteristics, AKI etiologies, and hospital outcomes. To the best of our knowledge, no previous study has evaluated short-term mortality after AKI in the very elderly patients, especially standardized complete KDIGO criteria.
In the very elderly, the development of clinically relevant AKI is associated with age-dependent changes, which

3
are independent risk factors for AKI. Such changes render older patients more prone to medication-associated toxicities and ischemic injury. Furthermore, elderly individuals are very vulnerable to co-existing illness and are more likely to require surgery. Often, multi-drug combination therapy is required; this is a prime trigger of AKI. Previous research reported the incidence of AKI among very elderly patients as 23-40%, with most patients classified as stage 1 (RIFLE or AKIN criteria) [14][15][16]. We found that 39.0% of geriatric patients had AKI, in agreement with the literature [11].
As in other works [17,18], we found that infections (39.6%), hypovolemia (23.8%), cardiovascular events (15.9%), nephrotoxicity (12.0%) were the principal causes of AKI. Infections were the prime cause, almost always preceded by pneumonia. Elderly patients are particularly prone to pneumonia because of an impaired gag reflex, decreased mucociliary function, waning immunity, and increasing cardiopulmonary dysfunction [19]. In the very elderly, coexisting illnesses play major roles in triggering pneumonia, and they influence disease severity [20].
The in-hospital mortality of our elderly AKI patients (25.7%) was slightly higher than in previously published studies (16-20%) [11,21], but lower than the rate of 33-40% reported in previous reports on patients older than 75 years [17,22]. The wide variation in the mortality of elderly AKI stems from the population being studied, etiology to AKI, the ICU settings, the definition of AKI used and whether CKD is included during the studies. In our cohort, of the 172 stage 3 AKI patients, only nine patients required dialysis during follow-up, and 2 died within 28 days. The most common causes of death were pneumonia-associated MODS (49.4%) and pulmonary infections (22.5%). Other less frequent causes of death were respiratory failure (11.3%), cardiovascular events (10.6%), hemorrhagic shock (2.5%), ARF (0.6%) and other (3.1%). Immunity often wanes in the elderly. High comorbidity levels, age-related changes in organ function, and diminishing organ reserves render the elderly more vulnerable to pneumonia, causing AKI patients to progress to MODS [23]. In the present study, the second leading cause of death was pulmonary infection (22.5%), emphasizing that pneumonia caused either AKI or the lethal MODS.
Many studies have demonstrated the importance of age and male gender as risk factors for death in AKI [8,14,21]. In the present study, the median age was 87 years, and there was no significant difference of ages among survivors and non-survivors. This was probably because this is a geriatric hospital; almost health objects are male gender.
In the present study, there were some significant factors associated with death, such as lower MAP, oliguria and need for MV. The presentation of lower MAP in non-survivors is interesting. We suggest that this manifestation might result from the causes of AKI. In our study, 25.6% of the nonsurvivors were related to hypovolemia and it is expected that those who died with AKI might exhibit hypotension more frequent than those who survived. This suggests that the occurrence of hypotension, oliguria and MV is associated with a higher risk of death.
Malnutrition is a common problem in elderly patients. Several studies on the association between mortality and nutritional status have found that preexisting malnutrition is associated with poor outcomes in AKI patients [24,25]. In our study, the prealbumin levels of non-survivors were significantly lower than those of survivors; prealbumin levels of 181 g/L were strongly associated with a higher risk of 28-day mortality.
Previous research suggests that BUN is predictive for short-and long-term mortality independent of SCr level [26]. In patients with severe AKI who required dialysis, predialysis BUN level is predictive of 60-day mortality [27]. We found BUN level (≥ 12.8 mmol/L) to be independent predictors of short-term mortality. SCr was not associated with 28-day mortality after adjustment for other covariates.
The effects of dysmagnesemia on mortality remain controversial. Previous studies reported nearly twice mortality rates in hypomagnesemic compared to normomagnesemic patients [28][29][30]. However, another research reported no differences in mortality between hypomagnesemic and normomagnesemic patients [31]. In the present study, we did not find any significant effect of hypomagnesemia on patient mortality. But we find that hypermagnesemia was strongly associated with 28-day mortality.
Our study had some potential weaknesses. First, it was a single-center retrospective work, and the results may not be immediately applicable to other hospitalized patients. Second, we did not use the urine output criteria because these data were incomplete. Third, there were fewer female treated than males. So, biased results may be unavoidable. Fourth, the study excluded patients with CKD. As CKD is a potent risk factor for AKI, excluding patients with CKD in this analysis may have potentially unestimated the rate of AKI in the very elderly. Fifth, the follow-up duration was short. We did not study patients with less severe AKI (those who survived for 28 days or longer). We also did not explore the effects of AKI on kidney outcome.

Conclusion
There are important risk factors for death among very elderly patients with AKI that must be identified early to decrease mortality. Our findings support the view that the incidence of AKI increases significantly as age advanced. 28-day mortality is less severe than expected. Time for AKI to develop, low MAP, low prealbumin level, oliguria, requirement for MV, BUN level, magnesium level, more severe AKI stages are risk factors associated with short-term outcomes. Identification of these factors might lead to more intensive monitoring and early prevention, and might improve AKI patients' outcomes in the very elderly.

Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of interest.

Statement of human and animal rights
The study design was approved by the Clinical Ethics Committee of the Chinese PLA General Hospital.
Informed consent For this type of study, formal consent is not required.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.