Predicting Septic Shock After Emergent Ureteral Stenting in Stone-Related Obstruction and Presumed Infections

Introduction

Classically, sepsis has been defined as excessive immune reaction attributable to an infectious etiology meeting criteria for systemic inflammatory response syndrome (SIRS). ¹ However, more recently, the Sepsis-3 Task Force has created the sequential organ failure assessment (SOFA) and quick sequential organ failure assessment (qSOFA) scores. ² Both scores have been validated in settings outside the intensive care unit (ICU). ^2,3 In addition, qSOFA is based exclusively on clinical parameters, being a simple and useful bedside tool for tracking and predicting a patient's risk of deterioration. The extent of their usefulness is still subject to controversy, as their ability to predict severe complications such as septic shock is still questioned. ^{4 –6}

Up to a third of cases of sepsis are attributable to urinary tract infections. ⁷ These usually are related to concomitant urinary obstruction, typically from ureteral calculi and are denominated as acute obstructive pyelonephritis (AOP). ^7,8 Standard of care in such cases is emergent urinary decompression through either percutaneous nephrostomy or indwelling ureteral stent placement and antibiotics. ⁹ Although these interventions have proven to be effective at reducing mortality rate, it is known that not all patients receive guideline-compliant interventions and, a subset of patients will continue to deteriorate and require escalation in therapy. ¹⁰ Even with adequate treatment, mortality rate may be as high as 10%. ^{11 –13} Accurate identification and stratification of patients at risk of infectious progression allows for health care systems to better allocate resources and provide improved care.

The applications and performance of qSOFA criteria have seen limited adoption and study in urology, where most studies still use SIRS criteria as an endpoint to represent sepsis and septic shock. ^{14 –16} Thus, current literature in urology is yet to determine the comparative performance of both scores in various settings.

Although the international consensus conference on definitions for septic shock recommended the use of SOFA or qSOFA criteria, at times, in lieu of SIRS criteria in the evaluation of patients at risk for sepsis, these scoring systems have not been applied systematically to sepsis of a urologic origin. ³ A recent study evaluation the risk of patients developing septic shock after percutaneous nephrolithotomy (PCNL) compared qSOFA and SIRS, finding qSOFA to be a stronger predictor of infectious and postoperative complications. ¹⁴

This study aims to compare the diagnostic and predictive performance of qSOFA and SIRS in patient deterioration to septic shock in patients presenting to the emergency department (ED) with obstructing ureteral stones and suspected infection.

Methods

With prior institutional review board approval, a retrospective chart review was performed identifying consecutive patients who underwent emergent retrograde ureteral stent placement between July 2016 and April 2020 at two sites (Massachusetts General Hospital [MGH] and University of Minnesota Medical Center [UMN]). Inclusion criteria were as follows: patients with unilateral obstruction from a renal pelvis or ureteral stone and documented suspicion for concomitant infection in the medical record who underwent decompression with a ureteral stent within 24 hours of presenting to the ED. Exclusion criteria included those with solitary kidneys, renal transplants, those with vasopressor support on admission, and bilateral ureteral obstruction. Suspicion for concurrent infection was based on clinical assessment (e.g., fever and general ill appearance), vital signs, and leukocytosis and pyuria.

Primary outcomes measured were identification of positive criteria for SIRS and/or qSOFA between presentation and time of stent placement, ICU admission, and continued vasopressor requirement beyond 1 hour after stent placement (defined as septic shock). SIRS criteria includes temperature <36°C or >38°C, heart rate >90 bpm, respiratory rate >20 breaths per minute and white blood cell count >12,000 or <4000 cells per mm³. A patient was considered to have positive SIRS criteria if two or more of the listed criteria were recorded concomitantly in the medical record between arrival to the ED and induction of anesthesia. qSOFA criteria includes Glasgow Coma Scale score <15 indicating altered mental status or documentation of altered mental status in nursing or medical notes, systolic blood pressure <100 mm Hg, and respiratory rate >22 breaths per minute.

A patient satisfied positive qSOFA criteria if two or more criteria were recorded concomitantly in the medical record between arrival to the ED and induction of anesthesia. Parameters used to determine SIRS and qSOFA criteria were collected for all study patients at the time of treatment and entered into the electronic medical record by nursing personnel. Septic shock is defined as the need for ICU-level care to initiate vasopressors to maintain mean arterial pressure >65 mm Hg with a suspected urinary tract infectious source for >1 hour after ureteral stent placement. Additional pre-, intra-, and postoperative data recorded were patient demographics, comorbidities, CT findings, culture data, and length of stay.

Statistical analysis was performed with R^® using the tidyverse, compareGroups, and expss packages. Data normality was assessed using the Shapiro–Wilk test, parametric testing for normally distributed variables was done using Student's t-test and analysis of variance with Tukey's post hoc, with results expressed in means and standard deviation. Nonparametric testing for non-normally distributed variables was done using Mann–Whitney U and Kruskall–Wallis tests, with results expressed in medians and interquartile ranges.

Chi-square and Fisher's exact were used to test discrete variables, with results expressed as frequencies and percentages. The McNemar test was performed to assess for differences in sensitivity and specificity between the two criteria. Univariate and multivariate logistic regressions were employed to determine odds ratios (ORs) and adjusted odds ratios (AORs), respectively. Omnibus' test of model coefficients and Hosmer and Lemeshow test of data fitness were employed to assess the model. A test probability of <0.05 was considered as statistically significant.

Results

A total of 187 patients were included for analysis. Site 1, MGH, contributed 143 (76%) patients, whereas site 2, UMN, contributed 44 (24%) patients. Mean patient age was 58.8 ± 18 years and women made up 60% of the patient population, whereas 40% were men. ED urine cultures were performed in 165 patients, of which 138 (83.6%) were positive. Blood cultures were performed in 100 patients, 43 (43%) were positive. Upper urinary tract cultures at the time of stent placement were performed in 147 patients, of these, 73 (51%) were positive. Of the 17 patients with negative lower urinary tract cultures, 14 (82%) patients had positive upper urinary tract cultures. Mean obstructing stone size was 6.8 mm ±4.2. ICU admission was required in 42 patients (22.5%), whereas 35 patients (18.7%) required vasopressor support. Mean ICU stay was 3.1 ± 4.4 days, whereas mean hospital stay was 3.4 ± 3.9 days. Table 1 summarizes baseline characteristics as well as for SIRS and qSOFA scores.

SIRS criteria were met by 103 patients (55.1%) of the entire cohort and 30 (85.7%) of the patients who met our defined criteria for septic shock. qSOFA criteria were met for 24 patients (12.8%) in the entire cohort and 18 (51.4%) of the patients who developed septic shock. Specificity for postoperative septic shock was significantly higher for qSOFA than for SIRS criteria (75% vs 29.1%, McNemar test p < 0.001). Sensitivity was higher for SIRS criteria than for qSOFA (85.7 vs 51.4%, McNemar test p < 0.01). qSOFA criteria had a significantly higher positive predictive value than SIRS (75% vs 29.1%), whereas negative predictive values were similar (89.6% vs 94%). These findings are summarized in Table 2.

Analysis employing response operator curves demonstrated superior performance of qSOFA criteria when compared with SIRS criteria for predicting septic shock. Although both SIRS and qSOFA had significant areas under the curve (AUC), qSOFA had a fair AUC of 0.750 (p = 0.001), whereas SIRS had a poor AUC of 0.659 (p = 0.008). qSOFA had an estimated 51.7% sensitivity with 0.1% false positive rates, whereas SIRS had an 82.8% sensitivity with 50.9% false positive rates. These findings are displayed in Figure 1.

OPEN IN VIEWER

FIG. 1.

ROC and AUC analysis of qSOFA and SIRS for septic shock. AUC = areas under the curve; qSOFA = quick sequential organ failure assessment; ROC = response operator curve; SIRS = systemic inflammatory response syndrome. Color images are available online.

Univariate logistic regression of SIRS and qSOFA for septic shock revealed significant predictive odds for qSOFA (OR 46 [0.25–228], p = 0.001) but not for SIRS (2.29 [.716–7.37], p = 0.162). Multivariate analysis further confirmed qSOFA's significant predictive odds for septic shock (AOR 60 [12.40–290.29], p = 0.001). Further analysis on the impact of altered mental status and its relationship with septic shock and qSOFA was explored though univariate and multivariate regressions. qSOFA and altered mental status had independent significant odds of septic shock in both uni- and multivariate analysis, with qSOFA being the strongest predictor. Correlation analysis of qSOFA and altered mental status reveal a significant correlation index of 0.379, p = 0.001. This model's summary is displayed in Table 3.

Discussion

In patients who present with ureteral obstruction caused by a ureteral stone with concomitant pyelonephritis, there is a significant risk of urosepsis and death. ¹² Prior studies have attempted to elucidate risk factors for progression and implement quality improvement protocols to expedite care in such patients. ^{17 –19}

SIRS criteria are still commonly employed as a surrogate or risk predictor for infectious complications, even though its relationship to risk of sepsis in urologic patients remains unclear. Prior studies have concluded SIRS high sensitivity comes with lower performance of low specificity and positive predictive values. Haas et al previously reported only 4 out of 44 SIRS + patients developed postoperative septic shock in patients undergoing ureteral stent placement or PCN for obstructing stones. ²⁰ Similarly, Lee et al reported only 17 out of 62 SIRS+ patients developed septic shock in AOP. ²¹ Literature comparing the accuracy of qSOFA and SIRS criteria in the context of stone disease is scarce. qSOFA was found to be better septic shock and postoperative complications in patients undergoing PCNL. ¹⁴

Our study demonstrated the following: except in sensitivity, qSOFA outperformed SIRS criteria for the evaluation/prediction of septic shock in the cohort of patients evaluated in this study (parameters described in detail earlier in results section). This finding is consistent with the small body of currently existing urologic literature on this topic. Fukushima et al performed a retrospective study comparing qSOFA, SOFA, and SIRS criteria in 141 consecutive patients with AOP with upper urinary tract calculi and demonstrated that qSOFA and SOFA had higher predictive accuracy than SIRS for predicting in-hospital mortality rate and ICU admission in this patient population. ²¹ The same hypothesis was tested by Pandey et al, and similar results were obtained. ¹⁵

A recent multi-institutional retrospective study comparing qSOFA and SIRS criteria in predicting postoperative sepsis in >300 patients who underwent PCNL demonstrated a greater specificity and predictive value for the qSOFA score. ¹⁴ It is important to note, however, that sensitivity is an important parameter to consider when screening patients. Especially in conditions where missed patients may have possibly fatal deterioration. In our study, SIRS has a significantly higher sensitivity, which may be of value when initially screening post-stented patients with concerns of concomitant infections. Thus, clinicians should aim to familiarize themselves with both scores and employ each to their strength.

Inconsistencies in performance in these scores may be related to their criteria's reflection of underlying physiopathologic processes. Some drivers of positive SIRS criteria, such as tachycardia, are common observations from obstructing ureteral stones, even in patients without systemic infection, and thus may explain SIRs' higher false positive rates. Thus, qSOFA criteria may be more closely related to processes reflecting underlying disruption to end-organ perfusion.

Our study has several limitations, mainly derived from its retrospective design lacking standardization of prospective trials. Although we had the benefit of a larger cohort from a total of two high-volume emergency rooms at tertiary referral centers in different regions of the country, there were likely subtle differences in pre- and postoperative care that could not be accounted for in this study. Although thorough chart review was performed on prior provider notes and nursing documentation to minimize error, the subjective nature of documentation in the electronic medical record may lead to study bias.

Future studies are important to confirm our findings and assess the value of the qSOFA criteria for predicting sepsis and septic shock after endourologic decompression of an infected upper tract obstruction, along with other urologic cases. Future studies should attempt to determine risk factors that may more strongly relate to septic shock to provide clinicians with a score that has a better performance.

Conclusion

SIRS and qSOFA scores may be useful to clinicians assessing patients undergoing ureteral stent placement for obstructing stone and concern of concomitant infection. Although SIRS has a significantly higher sensitivity, qSOFA is a better predictor of septic shock. Clinicians should be aware of the advantages of each score and employ them accordingly. Further larger prospective studies are needed to validate our findings and explore qSOFA's risk prediction in other urologic settings.