Acute Pyelonephritis: A Single-center Experience

Introduction

Acute pyelonephritis (APN) is a well-known disease, described extensively in literature. Earliest citing of the condition dates back to ancient Egypt, highlighting the destruction of the kidney parenchyma and its outcomes.[1] In spite of this long history, the nomenclature of APN is still controversial, and the semantic ambiguities can still cause confusion.

APN in the USA has an incidence as high as 250,000 cases per year and requires 100,000 hospitalizations every year.[2] In a recent large cohort in a community-based study, on average per year among 1000 patients, there were 152.7 lower respiratory tract infections (95% confidence interval [CI] 151.3–154.1) and 99.6 urinary tract infections (UTIs) (95% CI 98.4–100.8).[3] The absolute number of deaths in 28 days after the diagnosis of UTIs was still high compared with those after sepsis because of the higher incidence rate of UTIs. It was interesting that a high proportion of hospitalizations within 28 days were not on the day of diagnosis. This could reflect high underlying hospitalization rates in this cohort or could be attributable to exacerbation of underlying comorbidities by acute infections.[3] The considerable number of hospitalizations within 4 weeks of lower respiratory tract infections and UTIs with a cardiovascular cause of admission are particularly intriguing, as both infections have been found to exacerbate underlying cardiovascular comorbidity.[4] Hence, identification of the novel prognostic markers for this ancient disease is the need of hour to triage those with the risk of adverse outcomes.

It is essential to define the clinical, biochemical, and radiological profile of patients with APN to identify the early prognostic markers so as to prevent the morbidity and mortality. However, there are no Indian studies addressing these issues of incidence, prevalence, and risk factors. Our goal was to assess the relationship between the clinical laboratory data at presentation and adverse outcomes.

Materials and Methods

Results

We collected a total of 296 cases over a period of 2 years. Mean age of the patients was 53.85 ± 9.78 years. One hundred patients (33%) were admitted in Intensive Care Unit (ICU). Males outnumbered females, 195/296 (65.9%) and 101/296 (34.1%), respectively. Baseline clinical, laboratory, and microbiological characteristics are shown in Table 1. One hundred and forty-five patients (54.4%) were known diabetic. It was the first episode of APN in 190 (65.6%) patients, the remaining 106 (35.4%) had experienced previous episodes; recurrent UTI patients had diabetes in 66 (62%) patients, and chronic kidney disease (CKD) in another 30 (28%) patients.

Table 1 Baseline clinical, laboratory, and microbiological data

Hemoglobin of <10 g/dl was noted in 224 (75.7%) patients with 102 (34%) having <8 g/dl. Subgroup analysis showed about 67% of those with hemoglobin <10 g/dl had underlying CKD at presentation. Leukocytosis was evident in 237 patients (80%); mean leukocyte count was 21572.02 ± 8269.74. Leukocytes normalized in 5.6 ± 2.4 days.

Pyuria was noted in 100% of cases. Urinary culture was negative in 153 (51.7%) and positive in 143 (48.3%) cases. Most common organism isolated was Escherichia coli in 88 (29.7%), followed by Klebsiella pneumoniae in 16 (5.4%), pseudomonas in 16 (5.4%), Enterococcus in 13 (4.4%), and Proteus in 10 (3.4%) [Figure 1]. Antibiogram of the most common organism is shown in Table 2. There was no difference in the prevalence of organisms in inpatient department (IPD) and ICU except for isolation of Citrobacter in ICU; there was an increasing trend in antibiotic resistance to common antibiotics used in ICU as against those in IPD patients. E. coli was sensitive to cefotaxime in only 15%, piperacillin-tazobactam in 75% in IPD patients as against 56% in ICU (p = 0.04), and levofloxacin sensitivity noted only 37% in IPD and 24% in ICU (p = 0.03). There has been a rise in the incidence of Citrobacter, with sensitivity to cefotaxime only 20%, 66% in IPD and 35% in ICU (p = 0.04), levofloxacin only 12% sensitive in IPD and 10% in ICU. However, there was an alarming increase in resistance to imipenem in ICU patients with only 40% sensitivity as against 100% in IPD (p = 0.001).

Close

Figure 1

Microorganisms grown in urine culture

Export to PPT

Table 2 Antibiogram of most common organism in inpatient department and Intensive Care Unit patients

Serum creatinine more than 1.5 mg/dl at admission was seen in 285 (96.3%) patients [Table 1]; 120 (40%) of them had underlying CKD with diabetes mellitus (DM) being the most common cause. Mean serum creatinine at admission was 4.95 ± 2.70 mg/dl. One hundred and two out of 285 (35.4%) patients required dialysis after admission; out of which, 78 of them were dialysis free at discharge. Distribution of HbA1c% in the different range is presented in Figure 2.

Close

Figure 2

HbA1c distribution of patients studied

Export to PPT

Ninety cases (30%) were admitted in ICU. Multiple organ dysfunction syndrome (MODS) either at admission or during the course in hospital stay was seen in 94 (31.8%) patients. Eighty-seven (30%) patients with NEPN group and seven (60%) of EPN group had MODS at admission or during the course of hospital stay.

Risk factors categorizing the cases as complicated PN are presented in Table 3. Among the risk factors recognized for complicated PN, DM (54.4%) was the most common factor followed by renal calculi (14.4%), benign prostatic hyperplasia (6.7%), immunocompromised state (3.3%), stricture urethra and meatal stenosis (3.3%), and neurogenic bladder (2%) [Table 3]. In 153 patients with negative urine cultures, CT was positive for APN in 99.1%. Majority of these negative cases were attributed to the prior use of antibiotic therapy either at outside hospital or in-house use of antibiotic before sampling for cultures. Concordance between blood and urine cultures was only 20% in our study.

Table 3 Criteria categorizing patients with complicated pyelonephritis

Ultrasonography showed abnormality in 67 (22.6%) patients; 12 of the 296 patients submitted to CT (2%) showed EPN. Six out of twelve patients had Class II, four had Class III, one with Class I, and another with Class IV. Prevalence of E. coli increased in EPN patients to 10/12 (83.4%). Compared to NEPN patients, EPN patients had a higher incidence of DM, 54% and 90% (p = 0.015), respectively; higher risk of developing MODS (p = 0.003). Four out of twelve (33%) EPN patients died and 1/12 (8%) underwent nephrectomy [Table 4].

Table 4 Characteristics of emphysematous pyelonephritis and nonemphysematous pyelonephritis

Fifteen patients evaluated had a renal abscess on CT-kidney, ureter, and bladder. Subgroup analysis showed that there was no difference noted between patients with or without abscesses with regard to leukocytosis, duration of symptoms, urine cultures, and pyuria. Ultrasound examination showed abscess only in 2/15 cases. Patients with abscesses were more likely to be hospitalized for a longer time (20 ± 4.9 vs. 10 ± 3.3 days) and were more likely to be treated for a longer time. Two patients had renal vein thrombosis following APN on the same. Both of them had an extensive workup for prothrombotic state which was negative. Eventually, both had venous infarction of the affected kidneys with reduced renal function.

A total of 18 deaths were noted in 296 patients (6.1%). Hemoglobin <10 g/dl, serum creatinine at admission >1.5 mg/dl, HbA1c% >10%, and immunosuppression had statistically significant association with the development of MODS on univariate analysis, but on multivariate analysis, only hemoglobin, HbA1c%, and immunosuppression reached statistical significance. Even with attributable risk of mortality, only hemoglobin, HbA1c%, and immunosuppression reached statistical significance on multivariate analysis [Table 5].

Table 5 Risk factors associated with attributable risk of multiple organ dysfunction syndrome and death with acute pyelonephritis

Discussion

There has been a steady increase in the frequency of this disease in recent times, yet we do not have literature evidence regarding the risk factors, prognostic markers, and outcomes. Moreover, not all our patients had positive urine culture; and each episode of UTI adds to the renal damage and progression of CKD. It also precipitates or aggravates the underlying cardiovascular disease. Males outnumbered females in Indian studies in contrast to those Western studies [Table 6];[678] probably, reflecting the local gender differences in seeking hospital care. DM (54.4%) formed a single most important risk factor for categorizing as complicated PN. This emphasizes the fact that India being the capital of DM would likely to contribute more burden of PN in the future to the health-care system.

Table 6 Comparison of clinical and laboratory profile with other studies

Urine culture was positive in only 48% of cases in our study [Table 1], but large variation has been noted in different study groups depending on the local practices. In our study, a large number of culture negativity was attributed to prior antibiotic usage either at outside hospitals before referral or to the late collection of samples. Subgroup analysis of these culture negative groups revealed that they are at lower risk of developing complications such as MODS, renal abscess, progression of renal failure, and death. Prevalence of organisms isolated was almost similar to those reported in other studies with E. coli being the most common organism in both EPN and NEPN groups, followed by K. pneumoniae. The most important intriguing finding in our study was differences in sensitivity pattern between inpatient in wards and intensive care patients although prevalence of organisms was same, there was more trend toward resistance to commonly used antibiotics (piperacillin-tazobactam, cefoperazone-sulbactam, and levofloxacin) in intensive care group [Table 2]. This emphasizes us to draw stringent antibiotic policies in hospitals to prevent further development of resistance to the latest antibiotic armamentarium and also to establish local antibiotic sensitivity patterns periodically and change in the usage of antibiotics accordingly.

Although ultrasonography is the most commonly used noninvasive radiological modality in the management of patients with APN, decision-taking process in APN is limited, which explains why many authors consider it unnecessary in this clinical setting when the patient progresses satisfactorily with adequate treatment, as is usually the case.[910] Notwithstanding this, the role of ultrasonography in the initial management of complicated PN has been poorly studied, which explains the lack of a wide consensus on this aspect. A previous prospective study involving 245 patients with febrile UTI found that ultrasonography detected 19.2% of clinically relevant urological disorders.[9] This figure is very different to that seen in our study bearing in mind that 66% of patients were diagnosed with APN. Although these authors propose a prediction rule to select the patients with PN who could benefit from an ultrasound study, we believe that given the high morbidity and high prevalence of alterations requiring urgent instrumentation, an ultrasonographic study should be performed in all APN patients. Furthermore, ultrasound was poor in detecting complications such as abscess, in which case CT was considered superior by a large number of studies. Thus, CT scan forms an important tool in the management of PN patients, particularly when expected clinical response is not seen.

The information available to date about the prognosis for acute and complicated PN and its influencing factors is scarce and limited to a very few particular settings.[67811] In our study, 31.8% of the patients developed MODS. Among the risk factors to predict MODS, univariate analysis showed statistical significance for hemoglobin <10 g/dl, serum creatinine more than 1.5 mg/dl at admission, HbA1c% more than 10%, and immunosuppression; but on multivariate analysis, statistical significance reached only with hemoglobin, HbA1c and immunosuppression [Table 4]. This analysis of HbA1c% is peculiar to our study and has not been reported in any of the previous studies.

Mortality associated with complicated PN ranges between 3.2% and 9.8%.[67811] Hemoglobin <10 g/dl, HbA1c% more than 10%, and immunosuppression were the factors that reached statistical significance in both univariate and multivariate analysis in predicting the mortality. Most of these factors have been previously identified in various studies as factors of poor prognosis.[79101112] Our study is unique in estimating the prognostic significance of HbA1c% in predicting the mortality.

Thus, in addition to the traditional markers of adverse outcomes, HbA1c% adds to our armamentarium in recognizing those at-risk patients. It also emphasizes us to stick to the basic tenet of better blood sugar control to reduce the incidence of these avoidable complications.

Our study has several limitations such as it did not include those with postoperative and posttransplant patients. This was done to represent those community-acquired PN cases. We also did not report the follow-up of these cases which require further studies in larger settings.

Conclusion

The absence of infected urine does not rule out the diagnosis of APN in common clinical practice. Local antibiogram pattern is essential to prevent further resistances in organisms. HbA1c% adds to the predictive parameters to recognize at-risk patients to intensify the treatment and avoid complications.