Hospital-Acquired and Community-Acquired Acute Renal Failure in Hospitalized Chinese: A Ten-Year Review

Abstract

Objectives. To investigate the difference between hospital-acquired acute renal failure (HA-ARF) and community-acquired acute renal failure (CA-ARF) in hospitalized Chinese. Methods. The diagnosis of ARF in Peking University Third Hospital from January 1994 to December 2003 was reconfirmed and subdivided into AC-ARF and HA-ARF. Data of epidemiology, etiology, prognosis, and associated factors were analyzed. Single-variable analysis and multivariate logistic regression analyses were performed to investigate the correlation between clinical features and prognosis respectively. Results among 205 reconfirmed CA-ARF had a predominance of 59.5%, but HA-ARF demonstrated an increase by 1.06 during the last five years (p = 0.003). In all, 70.5% CA-ARF was diagnosed in internal medicine with 45.9% in department of nephrology, whereas 59.1% HA-ARF was diagnosed in surgical department with 51.8% in ICU. Distribution difference among departments was significant (p < 0.01). Further, 90.2% CA-ARF was associated with a single factor, while 36.1% of HA-ARF had two or more causes (p < 0.01). Also, 26.5% HA-ARF and 18.9% CA-ARF was drug-associated (p > 0.05) while 24.1% HA-ARF and 12.3% CA-ARF was infection-associated (p < 0.01). HA-ARF vs. CA-ARF was 62.7% vs. 23.0% in mortality (p < 0.01), 0.54 ± 0.24 vs. 0.27 ± 0.18 in ATI-ISS index (p < 0.01) and 19.6 ± 4.9 vs. 15.7 ± 5.6 in APACHE II scores (p < 0.01). MODS and SIRS were common independent predictors with oliguria for HA-ARF and advanced age for CA-ARF, respectively. Conclusions. In hospitalized Chinese during the last ten years, CA-ARF was still predominant with simpler cause and lower mortality, whereas HA-ARF was increasing with more complicated cause and higher mortality.

Keywords:

• HA-ARF
• CA-ARF
• etiology
• epidemiology
• prognosis

INTRODUCTION

Acute renal failure (ARF) is a common and critical clinical syndrome found in both the hospital and the community.[1–3] The differentiation between hospital-acquired ARF (HA-ARF) and community-acquired ARF (CA-ARF) is important in determining the prognosis, and knowledge of their characteristics may allow better management of such patients,[4–8] but the report on the difference between HA-ARF and CA-ARF was dearth and results varied. Kaufman[9] in Boston Veterans Affairs Medical Center reported that about 1% of hospital admissions suffered from CA-ARF, of which 70% had prerenal azotemia with the lowest mortality (7%), 11% had intrinsic ARF with the highest mortality (55%), and 17% had postrenal obstruction. Drug-induced nephrotoxicity and infection-related causes were the most common underlying etiologies, but ischemic acute tubular necrosis accounted for only 1%. Feest et al.[10] found that in England, 172 per million adults developed severe CA-ARF (serum creatinine concentration > 500 mumol/ l), which was at least twice as high as the incidence reported from renal unit based studies. Liano[11] in Spain reported that incidence of ARF in Madrid was 209 per million population, of which 48% was CA-ARF. As to HA-ARF, Hou et al.[12] reported that in the northern United States in the late 1970s, HA-ARF was 4.9% of admissions and mortality rate was 64%. When the study was repeated about two decades later, the prevalence had increased to 7.2%.[13] In both studies, reduced renal perfusion is the leading cause of HA-ARF (∼40%) with the drug-induced increasing from 8% to 16%, contrast-induced remaining at 11%, and the postoperative decreasing from 18% to 9%.[12],[13] Obialo founded in hospitalized African Americans, the incidence of CA-ARF was 3.5 times greater than that of HA-ARF. Prerenal ARF was more common among CA-ARF, but intra-renal ARF was more common among HA-ARF. HA-ARF was twice as likely to require dialysis and had a relative risk of 2.5 for shortened survival. The mortality was high in younger patients with CA-ARF and in older patients with HA-ARF.[6]

In China, more and more ARF have been diagnosed.[14] A previous study supposed that the number of patients diagnosed with ARF had increased in hospitalized Chinese during the past ten years, with HA-ARF as the major source and infections, drugs, and operations as the leading causes.[15] This study was designed to determine the difference between HA-ARF and CA-ARF in hospitalized Chinese during the last ten years.

SUBJECTS AND METHODS

Study Subjects

This paper is a computer-assisted retrospective review of adult (age 18 years or older) patients diagnosed with ARF from ICD-9 codes from January 1994 to December 2003 in Peking University Third Hospital, a tertiary comprehensive care center with 1300 beds, approximately 6000–6500 daily adult outpatients, and 23,000–25,000 annual admissions to medical/surgical services. Two nephrologists independently collected the clinical and biochemical data of the patients diagnosed with ARF within 24 hours after hospitalization or after hospital diagnosis. The reconfirmation of ARF was performed using the following inclusion criteria: a rise within hours or days in serum creatinine level to 176 μmol/ l (2.0 mg/dL) or above if no prior history of renal dysfunction was available, or by more than a 50% increase in creatinine if baseline was above 132 μmol/ l (1.5 mg/dL). The end point of the investigation was death in hospital or discharge. If a patient was admitted or diagnosed with ARF more than once during the study period, all admissions were included.

ARF Classification

CA-ARF was defined as ARF developing outside the hospital[9–11] while HA-ARF was defined as ARF developing during hospitalization for non-renal related problems such as operations, interventions, drugs, etc.[4],[12],[13] Pre-renal ARF was defined as ARF with two of the following:

• urine osmotic pressure more than 600 mosm/ l;

• urine sodium less than 20 mmol/ l;

• ratio of urine/serum creatinine more than 40;

• fractional excretion of sodium less than 1% or renal failure index less than 1.0 mmol/ l.

Intra-renal ARF was defined as ARF with two of the following:

• urine osmotic pressure less than 300 mOsm/ l;

• urine sodium more than 40 mmol/ l;

• ratio of urine/serum creatinine less than 20;

• fractional excretion of sodium greater than 2% or renal failure index greater than 1.0 mmol/ l.

Pre-existing renal insufficiency was defined as a documented creatinine level of 132.6 μmol/ l (1.5 mg/dL) or higher in previous 12 months.

Criterion and the Prognostic Models

Systemic inflammatory response syndrome (SIRS)[16],[17] and multiple organ dysfunction syndrome (MODS)[16],[18] were defined as in the literature. The acute physiology and chronic health evaluation (APACHE II) is the sum of three components[19]: an acute physiology score (APS), the Glasgow coma scale (GCS), and an age-related score and chronic health evaluation score (CHE). As our data collection was retrospective, GCS was absent.

Acute tubular necrosis – individual severity score (ATN-ISS) is a prognostic evaluation of ARF that is calculated through a regression equation[20]:where male is 1 and female is 0; hypotension means a systolic blood pressure lower than 100 mmHg for more than 10 hours regardless of the use of vasoactive drugs; jaundice means a concentration of bilirubin higher than 2 mg/dL; coma means deep unconsciousness; assisted respiration means need of mechanical support; and 0.21 is the equation constant. Each clinical variable takes a value of 1 or 0 depending respectively on its presence or absence, with the exception of age, which takes the value of the patient's decade.

Prognosis

The outcome of patients was divided into three categories, according to literature[21]:

1. Recovery, including patients in whom hemodialysis was discontinued due to serum creatinine returning to normal or closing to baseline;

2. Treatment failure, including patients who were dependant on renal replacement therapy or whose serum creatinine remained higher than the initial level when ARF was diagnosed; and

3. Death, patients died in hospital.

Statistic Analysis

All continuous variables are presented as mean ± standard deviation (SD). Univariate analysis was conducted with two independent-sample student's t-test for comparisons of continuous variables, U-test for nonparametric comparisons, and chi-square for dichotomous variables. Logistic regression analysis was used to determine predictors for hospital mortality, and Kendall was used for correlation analysis. All analyses were performed with the statistical software package SPSS 11.0. Significance was accepted for a two-sided p value of less than 0.05.

RESULTS

Epidemiological and Laboratory Analysis

In all, 211 ARF-diagnosed cases were reconfirmed, with six cases deleted due to incomplete data. Of the 205 cases, CA-ARF had a predominance of 59.5%. The HA-ARF ratio, however, doubled in the later five years (p = 0.003). The creatinine levels at the diagnosis were consistent for both ARF and in different periods (see Table 1). HA-ARF had more advanced average age, higher serum BUN, more common oliguria, and longer hospital stays (p < 0.001). There was no significant difference in gender or peak serum creatinine levels between the two groups (see Table 2).

Table 1 The ratio and creatinine at diagnosis of ARF by years

Years	ARF (n)	HA-ARF (n, %)	CA-ARF (n, %)	HA-ARF creatinine at diagnosis	CA-ARF creatinine at diagnosis
94–98	49	11 (22.4%)	38 (77.6%)	365.2 ± 98.4 μmol/ l	349.5 ± 106.2 μmol/ l
99–03	156	72 (46.2%)	84 (53.8%)	348.3 ± 117.0 μmol/ l	341.5 ± 113.6 μmol/ l
Total	205	83 (40.5%)	122 (59.5%)

*p < 0.05, 94–98 vs. 99–03.

Table 2 Epidemiology, laboratory, and clinics of ARF

	HA-ARF (n = 83)	CA-ARF (n = 122)
Gender (men/women)	50/33	71/51
Age (year)	61.5 ± 16.3	51.1 ± 20.3^†
Scr (μmol/ l)	593.0 ± 384.1	501.6 ± 240.7
BUN (mmol/ l)	32.4 ± 13.0	27.8 ± 13.7
Oliguria (n,%)	59 (71.1%)	67 (54.9%)
Hospital stay (day)	51.4 ± 40.8	25.3 ± 13.7^†

*p < 0.005, ^†p < 0.001, HA-ARF vs. CA-ARF.

Division Distribution

All told, 59% of HA-ARF was diagnosed in the surgical department, while 70.5% of CA-ARF was diagnosed in internal medicine (p < 0.001). Further study showed only 4.8% HA-ARF was from renal division and 51.8% from ICU, while 45.9% of CA-ARF occurred in renal division and only 10.7% in ICU (p < 0.001; see Table 3).

Table 3 Division distribution of ARF (n, %)

	HA-ARF (n = 83)	CA-ARF (n = 122)
Internal medicine	34 (41%)	86 (70.5%)
Renal division	4 (4.8%)	56 (45.9%)
Surgical department	49 (59.0%)	36 (29.5%)
ICU	43 (51.8%)	13 (10.7%)

*p < 0.001, HA-ARF vs. CA-ARF.

Pathophysiology

Pre-renal and intra-renal was 27.7% and 72.3% in HA-ARF vs. 18.9% and 69.7% in CA-ARF (p > 0.05), respectively. It was found that 36.1% of HA-ARF had two or more causes, while 90.2% of CA-ARF was associated with one cause (p = 0.01); 26.5% HA-ARF and 18.9% CA-ARF were drug-associated (p = 0.194) while 24.1% HA-ARF and 12.3% CA-ARF were infection-associated (p = 0.028). Post-renal ARF only occurred in the group of CA-ARF with the ratio of 11.4%, while the postoperative ARF only occurred in the group of HA-ARF with a ratio of 25.3% (see Table 4).

Table 4 Pathophysiology and etiology of ARF (n, %)

	HA-ARF (n = 83)	CA-ARF (n = 122)
Pre-renal	23 (27.7%)	23 (18.9%)
Intra-renal	60 (72.3%)	85 (69.7%)
Post-renal	0	14 (11.4%)
Multiple factors	30 (36.1%)	12 (9.8%)
Drug-induced	22 (26.5%)	23 (18.9%)
Infection-related	20 (24.1%)	15 (12.3%)
Operations	21 (25.3%)	0

*p < 0.05, HA-ARF vs. CA-ARF.

Outcome and Severity Estimate

HA-ARF vs. CA-ARF was 62.7% vs. 23.0% in mortality (p < 0.001), 19.6 ± 4.9 vs. 15.7 ± 5.6 in APACHE II score (p < 0.001), 0.54 ± 0.24 vs. 0.27 ± 0.18 in ATN-ISS index (p < 0.001), 2.4 ± 1.0 vs. 0.5 ± 0.9 in number of MODS (p < 0.001), 55.4% vs. 4.9% in mechanic ventilation (p < 0.001), and 47.0% vs. 41.8% in utilization ratio of renal replacement therapy (RRT) (p > 0.05; see Table 5).

Table 5 Outcome, severity, and therapy in ARF

	HA-ARF (n = 83)	CA-ARF (n = 122)
Mortality (n, %)	52 (62.7%)	28 (23.0%)^†
Recovery (n, %)	17 (20.5%)	82 (67.2%)^†
SIRS (n, %)	51 (61.4%)	33 (27.0%)^†
Number of failed organs	2.4 ± 1.0	0.5 ± 0.9^†
ATN-ISS	0.54 ± 0.24	0.27 ± 0.18^†
APACHEII	19.6 ± 4.9	15.7 ± 5.6
Mechanic ventilation (n, %)	46 (55.4%)	6 (4.9%)^†
RRT utilization (n, %)	39 (47.0%)	51 (41.8%)

*p < 0.05, ^†p < 0.001, HA-ARF vs. CA-ARF.

Correlation Analysis

Univariate analysis showed that ICU stays, oliguria, low serum albumin, mechanic ventilation, MODS, and SIRS correlated with the prognosis of HA-ARF (p < 0.05), and advance age, hospital stays, serum BUN, serum bilirubin, MODS, and SIRS correlated with the prognosis of CA-ARF (p < 0.05). RRT had a significant relation with the prognosis of CA-ARF but not HA-ARF (see Table 6). Multiple regression analysis identified that SIRS and MODS were common independent predictors with oliguria for HA-ARF and advance ages for CA-ARF (p < 0.05; see Table 7).

Table 6 Univariate analysis of risk factor

	HA-ARF		CA-ARF
	Non-survivors (n = 52)	Survivors (n = 31)	Non-survivors (n = 28)	Survivors (n = 94)
Age	63.8 ± 16.4	57.5 ± 16.0	66.9 ± 13.8	46.5 ± 19.5^†
Hospital stay	46.1 ± 43.3	61.4 ± 61.3	14.6 ± 14.5	28.3 ± 25.0
ICU	15	7^†	38	19
Oliguria	43	16	19	48
Peak serum creatinine	490.0 ± 231.1	512.3 ± 258.7	509.3 ± 275.8	619.6 ± 407.0
BUN	33.8 ± 13.3	30.5 ± 12.7	34.4 ± 13.0	25.7 ± 13.2
Bilirubin	63.7 ± 123.9	49.7 ± 83.2	69.1 ± 149.6	22.5 ± 85.3
Albumin	32.6 ± 7.3	36.4 ± 5.3	32.9 ± 6.7	33.9 ± 8.0
Number of failed organs	1.7 ± 0.9	0.7 ± 0.8^†	1.4 ± 1.3	0.2 ± 0.5^†
SIRS	41	10^†	17	16^†
Mechanic ventilation	34	12	3	3
RRT	28	11	12	39

*p < 0.05, ^†p < 0.001, non-survivors vs. survivors.

Table 7 Multivariate analysis of risk factors for mortality

	Regressioncoefficient	Standard error	OR	p
HA-ARF
Oliguria	1.72	0.74	5.56	0.020
SIRS	1.51	0.67	4.53	0.024
Number of failed organs	0.98	0.43	2.67	0.021
CA-ARF
Age	0.92	0.03	1.10	0.001
SIRS	1.64	0.73	5.17	0.024
Number of failed organs	1.62	0.46	5.07	<0.001

DISCUSSION

In the current study, it was found that CA-ARF was still the major component of ARF in hospitalized Chinese, but the predominance over HA-ARF was approximately 1.5 times, not as high as 3.5 times as reported elsewhere[6]. HA-ARF ratio doubled in last five years, which was similar in rising tendency, but did not reach the extend of 5–10 times more than CA-ARF as reported elsewhere.[5] Because the creatinine level at the diagnosis was consistent for both HA-ARF and CA-ARF and in both the former and the later five years, the change of ratio between CA-ARF and HA-ARF cannot be attributed to the criteria for different ARF or in different period. Feest[10] focused on patients with a creatinine level greater than 500 μmol/ l, Kaufman[9] and Obialo[6] included patients with creatinine more than 176 μmol/ l, and the present patients were diagnosed with the mean creatinine level of 340–360 μmol/ l. The difference of the ratio among the studies may be due to the inclusion criteria.

In agreement with other reports,[7],[22–24] the present study also observed an elder predominance in group of HA-ARF, and one-third of HA-ARF had more than one cause. What attracted attention was that drugs, operations, and infections together accounted for nearly three-quarters of the causes, and more than 25% of HA-ARF was postoperative. This indicated that the cause of HA-ARF converged and was highly related with treatments. In the studies by Hou and Nash,[12],[13] drug-induced HA-ARF and postoperative HA-ARF jointly accounted for only 25% of HA-ARF in different periods, and postoperative HA-ARF decreased from 18% in 1970s to 9% in 1990s. Combined with the fact that more than 40% came from internal medicine, it may also well be supposed that HA-ARF was a hospital-wide problem.

In this study, it was also founded that 50 was the average age of patients with CA-ARF, and more than 90% CA-ARF was single cause-related. The underlying diseases were not investigated due to the inadequate data, but many middle-aged patients developed CA-ARF from drugs, poisoning, trauma, rhabdomyolysis, or infections (not shown because of the small samples). This indicated that the cause of CA-ARF was various and separate. Given that CA-ARF was relatively more frequent and younger, and that drug-induced ARF in CA-ARF was nearly as high as in HA-ARF, it was suggested that the rational use of drugs, especially nephrotoxic drugs, was not only important within the hospital but also outside the hospital.

When the physiologic characteristics of ARF were divided into the categories of pre-renal, intrarenal, and post-renal causes, it was found that 70% of CA-ARF was prerenal azotemia,[9] 40% of HA-ARF had reduced renal perfusion, and pre-renal azotemia and ischemic tubular necrosis occurred accounted for 70% of the ARF cases in their study.[13] In this study, fewer than 20% of patients were pre-renal ARF and more than 70% were intrarenal ARF, both in CA-ARF and HA-ARF. Although there were not enough pathological data to diagnose acute renal necrosis, the relatively low percentage of pre-renal and high percentage of intrarenal might be supposed to be related to a delayed or missed diagnosis of ARF, as most of the cases were not diagnosed until creatinine had elevated to up to 350 μmol/ l. This again accentuates the urgency to develop a consensus criterion for ARF and update the concept among doctors in departments.

In the present study, the mortality rate was 62.7% in HA-ARF and 23% in CA-ARF. Comparing with CA-ARF, patients with HA-ARF had more risks, such as older age, oliguria, SIRS, MODS, mechanic ventilation, and higher severity indexes such as APACHE II score and ATN-ISS score, which were similar with other reports.[2],[16],[19],[25],[26] Here it is emphasized that although the data that showed patients with HA-ARF were more serious and mechanic ventilation-dependent, there was no significant difference in the utilization ratio of renal replacement therapy (RRT) from CA-ARF. Those patients might not have indications for RRT, but unawareness of RRT among doctors could not be excluded as continuous RRT (CRRT) was not widely put into use in our hospital until three years ago. By the way, the prognosis of patients with AC-ARF but not HA-ARF showed a relationship with hemodialysis and hemofiltration in this study, indicating that renal replacement therapy (RRT) was an effective means for simple renal failure but not for complicated renal failure, especially with MODS. It has been widely accepted that the outcome of ARF is primarily dependent on severity of the diseases.[27–29]

The prognosis of HA-ARF and CA-ARF were analyzed with univariate analysis and multivariable analysis. It was found that MODS and SIRS were independent predictors for both HA-ARF and CA-ARF, which was similar with previous reports.[7],[15],[16],[28],[30] The incidence of oliguria was higher in HA-ARF, and oliguria was the independent predictors of HA-ARF. This illustrated that the early detection of oliguria and timely RRT may improve the outcomes of patients with HA-ARF, which was proven previously.[16],[31] Although the average age of CA-ARF was younger, the age gap between the survivor group and non-survivor group of CA-ARF was larger. This suggested that older patients with CA-ARF were more susceptible to death, and age was more important for the prognosis of CA-ARF.

CONCLUSION

In hospitalized Chinese during the last ten years, CA-ARF was still predominant with a simpler cause and lower mortality, but HA-ARF was increasing with more complicated causes and higher mortality.