Clinical features,outcomes and predictors of drug resistance and complications in patients admitted with pyelonephritis

Abstract

Drug resistance and the presence of structural complications have significant implications for the treatment of acute pyelonephritis. We aimed to examine the predictors of drug resistance and complications in a retrospective cohort of patients admitted with pyelonephritis. 188 patients were included in this study. Patients who had had a urinary catheterization in the previous month and who lived outside the district in which the hospital was located were more likely to have ESBL infections. Carbapenem resistance was associated with recent urinary catheterization, a positive urine nitrate test, hypotension requiring vasopressors and the need for intensive care. A history of flank pain, urea level >13.3 mmol/L, a differential neutrophil count >75% and a urinalysis with >1000 leucocytes per high power field was associated with an increased risk of complications. A score derived from these variables to predict structural complications of infection had a sensitivity of 77.8% and a specificity of 67.1%

Keywords

Pyelonephritis carbapenem resistance ESBL renal abscess

Introduction

Acute Pyelonephritis is one of the most common reasons for in-patient admission in India and around the world. The treatment of pyelonephritis without complications requires a short course of antibiotics¹ and has a good prognosis. Structural complications of the infection, including hydronephrosis, renal abscess and emphysematous pyelonephritis may require a longer course of antibiotics with or without an invasive procedure for management. Drug resistance is widely prevalent in India and increasing owing to widespread misusage. Common Gram-negative urinary pathogens are increasingly found to have extended spectrum beta-lactamase producing bacteria (ESBL)² and reports of carbapenem-resistant organisms (CRO) are also increasing.³

Predicting beta-lactam and carbapenem resistance in India has very significant implication in clinical practice. The early initiation of appropriate antibiotics obviously improves outcome, making the choice of initial empiric antibiotics crucial.⁴ Predictors of ESBL and CRO infections may therefore help in clinical decision making at presentation.

Structural complications may require percutaneous nephrostomy or abscess drainage. Current guidelines suggest that imaging is not required in the majority of patients who present with pyelonephritis.⁵ In a resource-limited setting like India where patient follow-up may be poor, predicting the risk of structural complications should also help decide which patients require imaging at presentation.

Methods

Ours was a retrospective cohort study over a period of 18 months (June 2018-December 2019) among patients admitted with a diagnosis of pyelonephritis. The study was approved by the Institutional Research Board of the institution (IRB Min. No.13436), and was conducted in a general medical unit of a medical college and tertiary care hospital in south India.

Consecutive adults (>18 years old) were included if a diagnosis of acute pyelonephritis was recorded in our computerized medical records.

Clinical features, predisposing factors and comorbidity, laboratory results, clinical course and outcome were obtained. Data on blood or urine cultures and antimicrobial susceptibility were included, as were structural complications as defined above.

Consecutive patients were chosen to avoid ascertainment and selection bias. Data were collected in a specially-designed case record form from electronic medical records.

Sample size was calculated assuming using the formula 4 pq/d2 where p = a complication rate of 30 percent (obtained from a previous study),⁶ q = (100-p) and d = precision of + /- 7% (which assumes that the actual rate may lie between 23% and 37%).⁷ The sample size obtained was 172.

For baseline variables, continuous data were described with mean and standard deviation, while for categorical variables frequencies and percentages were obtained. Patients were compared between groups (ESBL infections vs. non-ESBL, CRO infections vs. non-CRO, complicated vs. uncomplicated infections). Student's t-test was used to assess the statistical significance of associations for continuous variables. Chi-squared test was employed to assess the statistical significance of categorical data. Logistic regression was used to adjust for common confounders. Unadjusted and adjusted odds ratios and their 95% confidence intervals were calculated. We attempted to develop a score to predict patients who have complications of pyelonephritis using the odds ratios of significant risk factors identified on multivariable analysis. A receiver-operator-characteristic curve was drawn to identify the optimal threshold to predict the complications of acute pyelonephritis. All statistical analysis was done using SPSS 16.0.

Results

We included 181 patients in our study, whose mean age was 57.5 years and of whom 61% were women. Most (81.2%) had fever while <33% had flank pain or lower urinary tract symptoms. A systemic inflammatory response syndrome (SIRS) was found in 76% and 44% had renal angle tenderness. Nearly 70% had co-existing diabetes mellitus; 21% chronic kidney disease and 23% urological abnormalities. Of the women, 64% were post-menopausal, while 14% had a history of recent urinary catheterization and 14% a history of recent urinary tract infection. Those with diabetes had poor glycemic control with a mean HbA1c of 8.7%. (Table 1)

Table 1.

Demographic and clinical characteristics of included patients.

	N (%)	Mean (SD)
Demographic features
Age (years)		57.8(15.6)
Sex- Female	110(60.8)
Residence within the district	101(57.1)
Clinical features
Fever	147(81.2)
Presence of lower urinary tract symptoms	47 (26.3)
Flank or lower back pain	53(29.8)
Vomiting	66(37.1)
Altered sensorium	26(14.5)
Heart rate (/min)		103.6(23.1)
Systolic blood pressure (mmHg)		121.2(30.3)
Respiratory rate (/min)		25.3(7.4)
Renal angle tenderness	79(43.6)
qSOFA score		1.1(0.76)
Presence of SIRS	132(76.3)
Co-morbid illnesses and predisposing factors
Diabetes Mellitus	126(69.6)
Hypertension	103(56.9)
Chronic Kidney disease	38(21)
Urological disease	42(23.2)
Recent catheterization	25(13.9)
Post-menopausal (N = 80)	51(63.8)
Prior urinary tract infection	25(13.8)
Treatment details
Antibiotic used
Meropenem	111(68.9)
Piperacillin + Tazobactem	8(5)
Colistin	7(4.3)
Linezolid	9(5.6)
Other	26(16.1)
Duration of antibiotics in days		17.4(12.6)
ICU care required	26(14.4)
Procedure required	21(11.7)
Outcomes
Alive and well at discharge	158(87.8)
Alive at post-discharge OPD follow-up	127(70.2)
Follow up at 1 year	42(23.2)

Common abnormalities described on sonography were bulky kidneys (18.8%), hydronephrosis (16.6%) and renal abscesses (9.4%). The 181 subjects had 234 urine and or blood cultures. Among these 40.1% showed no growth. Urine cultures were positive in 63.2% and blood cultures in 10.7%. The common causative organisms identified were E. Coli (52.3%), Enterococcus spp (12.8%) and Klebsiella spp (7.3%). E. Coli infections were ESBL in 60.9% and CRO in 7.3%. Renal dysfunction (mean Creatinine 194.5µmol/L) was common.

Overall, 29% were classified as having structural complications on imaging; 14% required ICU care while 11% required a surgical or radiological procedure such as a percutaneous nephrostomy or abscess drainage. In-hospital mortality was 12.2% and 70% attended at least one out-patient follow up visit. (Table 2)

Table 2.

Laboratory, imaging and microbiology results of included patients.

Laboratory investigations and imaging
Haemoglobin (g/L)		107.6(23.7)
Total WBC count (/cumm)		14582(7606)
Neutrophil percentage		79.9(11.5)
Creatinine (µmol/L)		194.5(159.2)
Bicarbonate (mmol/L)		17.2(4.7)
Albumin (g/dL)		3.2(0.7)
HbA1c (%)		8.7(2.4)
Ultrasound findings
Bulky kidneys	34(18.8)
Hydronephrosis	30(16.6)
Renal abscess	17(9.4)
Emphysematous changes	5(2.8)
Urine culture positive	110(63.2)
E.Coli	57(52.3)
Enterococcus	14(12.8)
Mixed	20(18.3)
Klebsiella	8(7.3)
Other	10(9.2)
Blood culture positive	52(30.4)
Infection with ESBL organisms among gram negative infections(n = 111)	82 (73.9)
Infection with carbapenem resistant organisms among gram negative infections(n = 107)	18(16.4)
Complication present	49(29.3)

A qSOFA score of >2 and residence outside the district in which the hospital is located were significantly associated with ESBL infection on both bivariate and multivariate analysis. Recent urinary catheterization, the need for vasopressor support, ventilation or intensive care, a diagnosis of malignancy and a positive urine nitrate test were significantly associated with CRO infections on bivariate analysis. Patients with a history of flank pain and those diagnosed with a complication of infection (renal abscess, emphysematous infection or hydronephrosis) were less likely to have a carbapenem-resistant organism in their blood or urine cultures. Multivariate analysis by forward conditional logistic regression revealed that only the associations with recent catheterization, the need for intensive care, a positive nitrate test and a diagnosis of malignancy remained significant.

The presence of flank pain on history, differential neutrophil count >75%, urea level >13.3 mmol/L and a urine leucocyte count >1000 per high power field predicted the presence of complications on bivariate and multivariate analysis. (Table 3)

Table 3.

Factors associated with ESBL and CRO infections.

	ESBL infections N(%)	Non-ESBL infections*N(%)	Bi-variate analysis OR (95% CI); p value	Multi-variate analysis** Adjusted OR (95%CI); p value
Residence outside the district	41(50.6)	5(17.9)	4.7(1.6-13.6); 0.002	7.5(2.0-27.7); 0.003
qSOFA score≥2	23(31.1)	3(11.1)	3.6(1.0-13.2);0.042	5.4(1.1-26.5);0.036

	CRO infections N(%)	Non-CRO infections * N(%)	OR (95%CI); p value	Multivariate analysis *** Adjusted OR (95%CI); p value
History of flank pain	1(5.9)	32(34.8)	0.1(0.0-0.9); 0.017	NS
Recent urinary catheterization	7(38.9)	9(9.9)	5.8(1.8-18.7); 0.001	7.1(1.2-43.4); 0.034
Malignancy	3(16.7)	4(4.3)	4.4(0.9-21.7)0.05	13.77(1.3-142.8); 0.028
Hypotension requiring vasopressors	5(27.8)	7(7.6)	4.7(1.3-16.9); 0.012	NS
Ventilation required	4(22.2)	4(4.3)	6.3(1.4-28.1);0.008	NS
Urine nitrate positive	7(43.8)	11(13.4)	5.0(1.6-16.2); 0.004	7.7(1.8-33.3);0.006
Intensive care	5(27.8)	4(4.3)	8.6(2.0-35.6); 0.001	18.1(2.7-120.4);0.003
Presence of complications of infection	1(5.6)	33(38.4)	0.1(0.0-0.7); 0.007	NS
Alive and well at discharge	14(77.8)	86(93.5)	0.2(0.1-1.0); 0.034	NS

*All comparisons were made to non-ESBL gram-negative infections

** Multivariate model- Adjusted for age and sex and all factors which were significant on bivariate analysis

Factors which were not significantly associated with ESBL or CRO infections: Age, sex, fever, dysuria, urinary frequency, presence of SIRS, qSOFA score, SOFA score>2, vital signs at presentation, renal angle or abdominal tenderness, comorbid illnesses (diabetes, hypertension, urological disease), post-menopausal status, prior urinary tract infection, ultrasound findings, haemoglobin, total and differential WBC count, platelet count, renal function tests, total bilirubin, electrolytes, presence of ultrasound findings, HbA1c

A score to predict the presence of complications of pyelonephritis was derived by giving each predictive factor a score equal to their adjusted odds ratio, rounded off to the nearest whole number (Table 4). The presence of flank pain, neutrophilia, high blood urea and pyuria suggested the presence of a complication. The receiver-operator-characteristic curve using a combined score showed an area under the curve of 0.77. The highest possible score is 19. The optimal threshold for predicting complications of infection was a cut-off of 12 which had a 77.8% sensitivity and 67.1% specificity.

Table 4.

Predictive factors for complications of pyelonephritis.

Characteristic	ComplicationsN(%)	No complicationsN(%)	Bi-variate analysisOR(95% CI); p value	Multivariate analysis*Adjusted OR(95%CI);p	Score
Presence of flank pain	23(46.9)	28(24.1)	2.8(1.4-5.6); 0.004	3.0(1.2-7.9); 0.024	3
Neutrophil percentage >75	39(86.7)	71(67.0)	3.2(1.2-8.3); 0.01	10.5(2.1-51.7); 0.004	10
Urea >13.3 mmol/L	17(40.5)	20(20.2)	2.7(1.2-5.9); 0.012	2.7(1.0-6.9); 0.046	3
Urine leucocyte count >1000/hpf or “numerous”	12(27.3)	14(12.2)	2.7(1.1-6.4); 0.02	3.4(1.2-10.0); 0.024	3

*Adjusted for all variables which were showed significant association on bivariate analysis.

Variables that were not significantly associated: Age, sex, residence, fever, dysuria, urinary frequency, presence of SIRS, qSOFA score, SOFA score>2, vital signs at presentation, renal angle or abdominal tenderness, comorbid illnesses (diabetes, hypertension, urological disease), post-menopausal status, prior urinary tract infection, ultrasound findings, haemoglobin, total WBC count, platelet count, total bilirubin, electrolytes, presence of ultrasound findings, HbA1c, vasopressor use, ICU admission.

Discussion

Our retrospective cohort of patients had similar clinical and laboratory findings and outcome are similar to those in previous studies.^8,9,10 Beta-lactam resistance in community-acquired infections is already common across India^8,9 Carbapenem resistance infection is also increasingly reported.^11,12 Patients who travel from further away are more likely to have already received antibiotics prior to presentation to a tertiary care centre, thus explaining more frequent ESBL isolates, an are therefore less likely to respond to empiric antibiotic treatment. They also had a higher qSOFA score. Other risk factors for ESBL infections not identified in our study include prior antibiotic use, prior hospitalization and a previous episode of urinary infection.¹³ It is no surprise that catheter associated urinary tract infections are at a higher risk of being caused by drug-resistant organisms.¹² A diagnosis of malignancy also increased the risk of CRO and is likely to be due to their frequent contact with the healthcare system. Carbapenem resistance was associated with increased risk of admission to intensive care. These patient's blood pressure and qSOFA scores at presentation did not differ significantly; thus, it may be inferred that their subsequent clinical deterioration was the result of a delay in the administration of appropriate antibiotic. Risk factors for Carbapenem resistance documented in other studies but not seen in ours include previous antibiotic or carbapenem use, length of stay in hospital and dialysis.¹⁴

The predictive factors of flank pain, azotemia, urine leucocyte count and blood neutrophil percentage are simple clinical and laboratory features are readily available in most primary and secondary care centres. While individually they do not have very high discriminatory power, together they can be used in a resource-limited setting as predictors of structural complications.¹⁵

The retrospective nature of our study inevitably involves missing some patient data. Moreover, our study, conducted at a tertiary care centre, may not be generalizable to primary and secondary care settings where the percentage of pyelonephritis caused by drug-resistant pathogens and those with complications will be much lower.

Conclusion

In our setting, ESBL and CRO infections account for 79% and 18% of acute pyelonephritis respectively. Nearly 29% develop complications. A history of urinary catherization in the previous month is a risk factor for CRO infections. Patients with infection by drug resistance organisms tend to be sicker on admission and require intensive care more frequently, probably due to a delay before receiving appropriate antibiotics. The presence of flank pain, high serum urea, leucocyturia and neutrophilia may suggest renal complications prompting the need for renal imaging.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article

ORCID iDs

Amita Jacob

Rani Diana Sahni

Tina George

References

Rudrabhatla

Deepanjali

Mandal

, et al. Stopping the effective non-fluoroquinolone antibiotics at day 7 vs continuing until day 14 in adults with acute pyelonephritis requiring hospitalization: a randomized non-inferiority trial. PLoS One 2018; 13: e0197302.

Kumar

Chatterjee

Deka

, et al. Increased isolation of extended-Spectrum Beta-lactamase-producing Escherichia coli from community-onset urinary tract infection cases in uttarakhand, India. Cureus 2021; 13: e13837.

Jaggi

Chatterjee

Singh

, et al. Carbapenem resistance in Escherichia coli and Klebsiella pneumoniae among Indian and international patients in North India. Acta Microbiol Immunol Hung 2019; 66: 367–376.

Liu

Fielding-Singh

Greene

, et al. The timing of early antibiotics and hospital mortality in sepsis. Am J Respir Crit Care Med 2017; 196: 856–863.

Nikolaidis

Dogra

Goldfarb

, et al. ACR Appropriateness criteria® acute pyelonephritis. J Am Coll Radiol 2018; 15: S232–S239.

Chen

K-C

Hung

S-W

Seow

V-K

, et al. The role of emergency ultrasound for evaluating acute pyelonephritis in the ED. Am J Emerg Med 2011; 29: 721–724.

Lemeshow

Hosmer

Klar

, et al. Adequacy of Sample Size in Health Studies. John Wiley and Sons, 1990.

Gopal

Wilson

Viggeswarpu

, et al. Clinical profile and predictors of outcomes in older inpatients with pyelonephritis in a tertiary care hospital in southern India. J Clin Diagn Res JCDR 2015; 9: OC05.

Kumar

Ramachandran

Mete

, et al. Acute pyelonephritis in diabetes mellitus: single center experience. Indian J Nephrol 2014; 24: 367.

10.

Umesha

Shivaprasad

Rajiv

, et al. Acute pyelonephritis: a single-center experience. Indian J Nephrol 2018; 28: 454.

11.

Sekar

Srivani

Amudhan

, et al. Carbapenem resistance in a rural part of southern India: escherichia coli versus Klebsiella spp. Indian J Med Res 2016; 144: 781–783.

12.

Milan

Ivan

. Catheter-associated and nosocomial urinary tract infections: antibiotic resistance and influence on commonly used antimicrobial therapy. Int Urol Nephrol 2008; 41: 461.

13.

Larramendy

Deglaire

Dusollier

, , et al. Risk factors of extended-Spectrum Beta-lactamases-producing Escherichia coli community acquired urinary tract infections: a systematic review. Infect Drug Resist 2020; 13: 3945–3955.

14.

Palacios-Baena

Giannella

Manissero

, , et al. Risk factors for carbapenem-resistant Gram-negative bacterial infections: a systematic review. Clin Microbiol Infect. 2021; 27: 228–235.

15.

Kim

Seo

M-R

Kim

S-J

, et al. Usefulness of blood cultures and radiologic imaging studies in the management of patients with community-acquired acute pyelonephritis. Infect Chemother 2017; 49: 22.