Risk factors of mortality in patients with acute kidney injury: Retrospective cohort study with longitudinal follow-up

Abstract

BACKGROUND

Acute kidney injury (AKI) is a clinical syndrome that, even after recovery of normal renal function, increases the short-term risk of developing chronic kidney disease and may contribute to mortality in hospitalized patients.

AIM

To assess patient survival and identify risk factors for mortality.

METHODS

We conducted a retrospective cohort study with longitudinal follow-up of patients hospitalized between January 2002 and December 2015, who had an episode of AKI as defined by the Kidney Disease Improving Global Outcomes 2012 guidelines, with return to normal renal function with follow-up extending up to 5 years after discharge, with a median follow-up of 7 years. Patient survival was verified using civil status records and by telephoning patients or their relatives. Risk factors were assessed using univariate and multivariate survival analyses. Short-term mortality was assessed between 3 months and 1 year, medium-term between 1 year and 5 years, and long-term beyond 5 years.

RESULTS

A total of 214 patients were included, with mortality data available for 193. Of these, 18 patients (9.3%) died within the short term (mean: 9.22 ± 0.67 months), 18 patients (10.2%) died within the medium term (mean: 36.27 ± 2.73 months), and 26 patients (16.5%) died within the long term (mean: 103.6 ± 6.63 months). The overall mean survival was 159 months. The Kaplan-Meier survival curve showed overall survival rates of 79%, 68%, and 57% at 5 years, 10 years, and 14 years, respectively. Identified mortality risk factors included age over 65, hypertension, diabetes, vascular disease, high Charlson comorbidity index, potassium levels above 5.5 mmol/L, pre-renal AKI, and the short-term development of chronic kidney disease.

CONCLUSION

These findings highlight the need to identify patients at higher risk of mortality following an AKI episode.

Key Words: Acute kidney injury; Mortality; Chronic kidney disease; Risk factor; Survival

Core Tip: An article with a longitudinal follow-up study was conducted in a North African country for which data are lacking in the literature, with a long follow-up period of up to 18 years. We identified some mortality risk factors such as hypertension, age over 65 years, diabetes, vascular disease, high Charlson comorbidity index, potassium levels above 5.5 mmol/L, pre-renal acute kidney injury, and the short-term development of chronic kidney disease.

INTRODUCTION

Acute kidney injury (AKI) is a global health issue that occurs in both medical and surgical hospital departments. Its prevalence is estimated to range from 1% to 25%, depending on the study[1], affecting approximately 13.3 million people annually in 2015[2]. The incidence of AKI is increasing sharply each year, representing 5% of hospital admissions in Europe, and is estimated at 35.7% on the African continent[2,3]. AKI is a preventable and treatable condition, particularly when addressing specific etiologies such as obstructive AKI, dehydration, and hepatorenal syndrome, as supported by a 2024 review[4]. Despite this, AKI, especially in developing nations and African countries such as Tunisia, carries a high, frequently underestimated burden of morbidity and mortality[5].

AKI is a clinical and biological syndrome that can lead to preventable death, making timely intervention crucial. Even after full recovery of renal function, AKI presents short-, medium-, and long-term risks of developing chronic kidney disease (CKD)[6]. Several studies have estimated that the risk of developing new CKD is four times higher in individuals who have experienced an AKI episode compared to those who have not[7].

Mortality attributed to AKI was estimated at 1.7 million deaths worldwide, including 1.4 million in developing countries, with 20%-50% of deaths occurring in hospitals; this rate can reach up to 75% among patients admitted to intensive care units[2]. The short-term prognosis of AKI depends on the speed of management and the presence of associated multi-organ failure. Although long-term mortality has slightly decreased in recent years, it still seems to be influenced by age and the presence of comorbidities.

The prevention of AKI largely depends on identifying at-risk patients, avoiding nephrotoxic factors, and maintaining renal perfusion. Numerous studies and medical research have focused on the epidemiology, causes, and immediate prognosis of AKI, particularly in intensive care settings in developed countries, whereas relatively few studies have been conducted in developing countries, especially in Africa. Given the lack of long-term survival data in Africa—particularly Tunisia—and the limited data on patients with full renal recovery after AKI, this study aimed to evaluate survival outcomes and identify risk factors for mortality in this specific population.

MATERIALS AND METHODS

Study design and data collection

We retrospectively included patients with AKI who had been hospitalized in the Nephrology Department of our hospital between January 2002 and December 2015. The end of data collection was in January 2021. Inclusion criteria were age ≥ 15 years and full recovery of renal function, either at discharge or 3 months later. Exclusion criteria included incomplete data, allograft recipients, pre-existing CKD, and in-hospital mortality. For patients with at least 3 months of history, their baseline creatinine level was recorded; for others, a baseline was calculated using an estimated glomerular filtration rate (eGFR) of 75 mL/minute/m². AKI was defined according to the Kidney Disease Improving Global Outcomes (KDIGO) 2012 definition, and full recovery of renal function was defined as sustained recovery of the eGFR ≥ 60 mL/minute/1.73 m² on discharge and up to 3 months after discharge[8]. Anamnestic, clinical, therapeutic, and evolutionary data were collected using a standardized form. We collected demographic data; the medical history of patients, including diabetes mellitus, hypertension, and vascular disease; and all items of the Charlson comorbidity index (CCI)[9]. We recorded clinical and laboratory data. CKD was defined per KDIGO guidelines as a sustained reduction in eGFR (< 60 mL/minute/1.73 m²) for at least 3 months, based on multiple, validated laboratory assessments of creatinine[8]. The eGFR was calculated using the abbreviated Modification of Diet in Renal Disease formula. Serum creatinine was initially measured using the Kinetic Jaffé method, transitioning to an enzymatic method in 2019.

AKI severity was graded according to the 2012 KDIGO guidelines into stages 1, 2, and 3[8]. Based on clinical records, laboratory data, and ultrasound imaging, AKI etiologies were classified as pre-renal, renal, or post-renal, with a maximum follow-up period of 18 years.

The study’s main outcomes included: (1) Primary: Evaluation of all-cause mortality; and (2) Secondary: Short-, medium-, and long-term development of CKD.

Statistical analyses

All statistical analyses were performed using Social Science Statistical software version 23.0 (SPSS Inc., Chicago, IL, United States). Analysis of the distribution of quantitative variables was carried out using the Kolmogorov-Smirnov test. When the distributions were normal, the variables were expressed by their mean ± SD with (minimum and maximum). Otherwise, the variables were expressed by their median and (interquartile range Q1, Q3). Qualitative variables were expressed as simple frequency and relative n (%). To compare the means of independent variables, we used the Student’s t-test for unpaired variables. To study the association of risk factors and the onset of morbidity, the χ² test was used to compare the percentages. Kaplan-Meier survival analysis with the log-rank test was used to identify prognostic factors for survival. When conditions allowed, multivariate analysis using the Cox model was performed. In all statistical tests, the 5% threshold was used for statistical significance and missing data were excluded.

RESULTS

Study population

Between January 1, 2002, and December 31, 2015, 14989 patients were admitted to our Nephrology Unit. Of the 600 patients diagnosed with AKI, only 250 patients achieved full renal recovery, while the remaining patients either developed CKD by discharge or presented with pre-existing AKI. We finally included 214 patients. The flowchart of the study is illustrated in Figure 1.

Figure 1 A total of 14989 patients were hospitalized between 2002 and 2015 in the nephrology department, of whom 600 had acute kidney failure. Of the 600 patients, only 250 had regained normal renal function upon discharge and up to 3 months later. AKI: Acute kidney injury; CKD: Chronic kidney disease; eGFR: Estimated glomerular filtration rate.

Main population characteristics

Our population comprised 107 men and 107 women, the median age was 61 years (43-73.25), and 43.5% of patients were aged ≥ 65 years. The most common comorbid condition was hypertension in 80 patients (37.4%), followed by diabetes (32.2%) and cardiopathy (7.9%). CCI was calculated for all patients, and the most frequent score was 0% for 52.3% of patients. Among the global cohort, 31 patients presented with acute complications such as acute pulmonary edema, severe metabolic acidosis, and acute severe hyperkaliemia. All patients had a kidney ultrasound. Oliguria was the circumstance of discovery for 54.6% of patients. Median serum creatinine during the AKI episode was 3.8 mg/dL (minimum 1.26 mg/dL, maximum 34.6 mg/dL). Stage 3 KDIGO was the stage selected for 115 patients included in our study, 31 of whom required intermittent renal replacement therapy. During hospitalization, the median duration of the AKI episode was 9 days and 78.5% of patients presented with persistent AKI (episode duration > 7 days). The baseline characteristics of the study population are shown in Table 1.

Table 1 Demographic and clinical characteristics of patients with acute kidney injury between 2002 and 2015.

Characteristics	n (%)
Population	214
Age, (years) median (Q1-Q3)	61 (43-73.25)
Women	107 (50)
Comorbidities
Hypertension	80 (37.4)
Diabetes	69 (32.2)
Cardiovascular disease	38 (17.7)
Charlson comorbidity index
0	111 (52)
1-4	94 (44)
≥ 5	9 (4)
Etiology of AKI
Pre renal	114 (53.3)
Renal	83 (38.8)
Post renal	17 (7.9)
KDIGO classification
Stage 1	50 (23.4)
Stage 2	49 (22.9)
Stage 3	115 (53.7)
Dialysis required	31 (14.5)

AKI: Acute kidney injury; KDIGO: Kidney Disease Improving Global Outcomes.

Survival analysis of the population

Patient survival was assessed through the last follow-up, with data collection concluding on January 1, 2021. Survival was assessed for 193 patients. Mean overall survival was 159 months. The Kaplan-Meier survival curve is shown in Figure 2. According to the Kaplan-Meier survival curve, the overall survival rate at 5 years, 10 years, and 14 years was 79%, 68%, and 57%, respectively.

Figure 2 Survival curve of 193 patients whose data were collected.

Factors associated with mortality in our population

In our series, age ≥ 65 years was the sole factor statistically associated with early mortality. We observed a statistically significant association between early mortality and a history of hypertension, diabetes, vascular disease, and multiple comorbidities, as well as a high modified CCI. In our cohort, admission potassium levels > 5.5 mmol/L and pre-renal mechanisms were statistically linked to early mortality. In our study, the development of short-term CKD was associated with early mortality.

Prognostic predictive factors of survival were investigated by univariate survival analysis using the log-rank test and are summarized in Table 2.

Table 2 Univariate statistical predictive factors of survival after an acute kidney injury episode between 2002 and 2015.

Population characteristics		Mean overall survival in months
Age	≤ 65 years	196	P < 10-4a
	> 65 years	101
History	Yes	144	P = 0.001a
	No	186
Hypertension	Yes	128	P = 0.005a
	No	174
Diabetes	Yes	122	P = 0.007a
	No	172
Vascular disease	Yes	103	P = 0.007a
	No	167
Multitarget patients	Yes	126	P = 0.001a
	Non	177
Modified CCI	0-4	183	P < 10-4a
	5-6	121
	≥ 7	68
Kalaemia (mmol/L)	≤ 5.5	159	P = 0.009a
	> 5.5	128
AKI mechanism	Pre renal	138	P = 0.025a
	Renal or post renal	177
CKD development at ST	Yes	108	P = 0.019a
	No	167
MACE	Yes	170	P = 0.222
	No	155

^aP ≤ 0.05.

AKI: Acute kidney injury, CCI: Charlson comorbidity index; CKD: Chronic kidney disease; MACE: Major cardiovascular event; ST: Short term.

In multivariate analysis, only patient age in years was considered a significant risk factor for mortality (P = 10^-3) with a hazard ratio of 1.08 (1.05-1.1).

DISCUSSION

Main findings

The survival rate in our study was 79% at 5 years, with short-, medium-, and long-term mortality rates of 9.8%, 11%, and 17.2% respectively. In our series, only age ≥ 65 years was statistically associated with early mortality. We identified a statistically significant association between early mortality and a history of hypertension, diabetes, vascular disease, polytrauma, and a high modified CCI. Our study found that an admission potassium level > 5.5 mmol/L and prerenal mechanisms were statistically associated with early mortality. In our study, the development of short-term CKD was associated with early mortality. Age and elderly subjects were independent risk factors for mortality after an episode of AKI. A history of AKI was found to be an independent risk factor for mortality.

Comparison with the literature

Patient survival rates were consistent with previously published literature[10-12]. Contrary to our findings, other studies reported significantly higher mortality rates[13-15]. Based on a review of more than 17 studies conducted over roughly 5 years in sub-Saharan Africa, AKI in patients with pre-existing CKD was associated with a 32% average mortality rate[16].

In a prospective study of 425 patients hospitalized in the intensive care unit, those who achieved normal renal function recovery following an AKI episode experienced a significantly lower mortality rate compared to those with underlying CKD (46% vs 83%). These findings highlight that renal function recovery is a key predictor of improved survival[17]. Another study involving over 1700 patients found that those who regained normal renal function had mortality risks comparable to those who never experienced AKI[18].

This may explain the differences between our study and those published in the literature, since we included patients who had recovered normal renal function either at discharge or after 3 months. The causes of death most frequently described in the literature were cardiovascular disease, infectious pathologies, and cancer[19].

A 2020 review included several studies in which the authors did not identify sex as a risk factor for mortality after an episode of AKI[20]. Several studies indicate that women have a lower risk of hospital-acquired AKI and possess a natural resilience against ischemia during AKI episodes. Research in rats supports these findings, suggesting that sex hormones play a defining role; specifically, testosterone appears to increase the kidney’s susceptibility to ischemic damage[21,22].

Consistent with previous research, several studies indicate that advancing age is linked to higher mortality rates and poorer survival after an episode of AKI[11,14,15,21]. This aligns with established evidence highlighting a strong association between pre-existing comorbidities and increased mortality risk after an episode of AKI[11,19]. According to a 2021 review by Gameiro et al[20], patients with multiple comorbidities face a significantly higher risk of mortality following an AKI episode, even if they successfully regain normal renal function during their hospital stay.

Contrary to studies suggesting that diabetes is a good prognostic factor after AKI[12,19,23], our findings indicate that the presence of at least one cardiovascular pathology correlates with a poor prognosis, aligning with established literature[12,14,15,23].

In our population, a modified CCI ≥ 5 was associated with a fourfold increase in mortality risk compared to a score < 5, consistent with existing literature[19]. Furthermore, this modified CCI has been validated for predicting 10-year mortality in patients in the intensive care unit with AKI[24].

Patients with kalemia below 3.5 mmol/L had better survival than those with normal kalemia in our series. This contradicts the findings of Teo et al[12], who found no association between kalemia and mortality.

Although the classifications were not originally designed to predict the prognosis of patients with AKI, several studies have shown that the degree of AKI severity determined using the Risk, Injury, Failure, Loss of Kidney Function, Acute Kidney Injury Network, and KDIGO classifications strongly predicts increased mortality[12,15,19].

Recent reviews confirm these findings, highlighting that greater AKI severity, regardless of the classification used, is associated with an increased risk of mortality, with KDIGO stage III acting as a strong indicator of a poor prognosis[20,25].

Patients who underwent extra-renal purification (ERP) had the most severe AKI, since they were classified as KDIGO stage III, irrespective of creatinine concentration. The effect of ERP on patient prognosis has been the subject of several studies in recent decades. In a 2019 study by Teo et al[12] that included 422 patients, those who underwent dialysis were three times more likely to die than patients who did not undergo ERP. The data remain inconclusive regarding the superiority of ERP over conservative treatment, particularly regarding the optimal methodology and timing.

The duration of AKI has been widely studied as a predictor of long-term mortality. Coca et al[25] prospectively studied more than 35000 patients who had undergone non-cardiac surgery and found that the duration of AKI decreased long-term survival, irrespective of AKI severity.

While existing literature identifies AKI duration as a key mortality risk, our study found no such correlation. This discrepancy likely stems from our cohort’s lower comorbidity burden and the fact that all our patients achieved complete renal recovery within 3 months of discharge.

Numerous studies have validated that the development of CKD in the short term, following an episode of AKI, is associated with decreased patient survival[13,20].

Based on our latest literature review, this is the first African study to evaluate short-, medium-, and long-term outcomes after an episode of AKI with complete renal recovery. One strength of our study is the focus on patients who recovered normal renal function—an understudied group—while excluding those with known chronic renal failure to specifically assess outcomes in patients with complete recovery. However, these selection criteria limited the size of our sample. A second strength of our study was the long 18-year follow-up period. By tracking patients hospitalized as early as 2002 through 2021, we achieved a longitudinal scope rarely documented in existing literature. This allowed us to identify mortality risk factors after an episode of AKI, even in cases where renal function had fully recovered.

This study had several limitations, including a single-center, retrospective design and a notable loss to follow-up. The exclusion of these missing cases, who may have maintained normal renal function or developed CKD, could result in an overestimation or underestimation of the observed associations. Furthermore, the lack of a comparator group prevented us from determining the true extent to which AKI serves as an independent predictor of long-term renal function.

CONCLUSION

Our study confirmed the importance of studying the epidemiology, clinical, and biological characteristics, particularly the long-term progression of patients who have recovered normal renal function after an episode of AKI. Therefore, preventing AKI and diagnosing it early via novel biomarkers is crucial, as this facilitates rapid, targeted treatment to reduce complications and improve patient outcomes. Based on the totality of the results, it is reasonable to consider that AKI remains a real public health issue that requires close monitoring. These findings will help identify patients at risk of developing short-, medium-, and long-term complications, facilitating improved post-episode management through nephrological follow-up for at least 1 year, even after the recovery of renal function.