Preoperative positive urine nitrite and albumin-globulin ratio are independent risk factors for predicting postoperative fever after retrograde Intrarenal surgery based on a retrospective cohort

Over the last decades, the surgical treatment of kidney stones has changed with the development of technology. The use of ureteroscopy (URS) increases and has become the most common choice over time [1]. According to previous reports, the incidence of postoperative fever (POF) after ureteroscopic holmium laser lithotripsy is 2–28% [2]. POF is potentially serious because it may progress to urosepsis which leads to mortality [3]. Therefore, identifying high-risk POF individuals and strengthening the interventions are urgently needed.

Several studies reported the risk factors for POF after URS, and many of them focused on retrograde intrarenal surgery (RIRS) [4‐6], both were limited by the small sizes and the number of potential risk factors included. Besides, this question was not fully answered. Therefore, this paper aimed to determine risk factors for POF after RIRS. Additionally, we also developed a nomogram for predicting the risk of POF.

31 (2.8%) of 1095 cases had POF after the RIRS. Characteristics of included predictors and univariable logistic regression analysis results were displayed in supplementary Table 1. Then multivariate logistic regression analysis was performed enrolling these significant variables (P < 0.1). After the multivariable logistics regression and bootstrap resampling tests, time in RIRS ≥30mins (only the flexible scope use period) (OR: 2.16, adjusted 95%CI:1.01–4.62, adjusted P = 0.047), Preoperative positive urine culture (OR: 2.55, adjusted 95%CI: 1.01–6.42, adjusted P = 0.047), Preoperative positive urine nitrite (OR: 9.09, adjusted 95%CI: 2.99–27.64, adjusted P < 0.001), Albumin/globulin ratio (AGR) (OR: 0.14, adjusted 95%CI: 0.03–0.74, P = 0.020) were statistical significant and were regarded as independent risk factors for predicting POF (Table 1).

A nomogram based on multivariable logistic regression was built (Fig. 1). After 1000 times 10-fold internal validation, the average AUC was 0.835 with 95% CI (0.816, 0.870). There was no significant statistic in the Hosmer-Lemeshow test (Chi-Square = 8.29, DF = 8, P = 0.4056), which indicated that this nomogram had good goodness-of-fit. The calibration curve was shown in Fig. 2 and it demonstrated good agreement between observation and prediction in our POF predicting model. DCA indicated that the application of the POF predicting model could bring clinical net benefit when the probability is 1 to 27%, whereas the analysis does not detect any clinical net benefit after this threshold (27%) (Fig. 3).

URS including RIRS has become the most performed surgery in ureteral and kidney stone disease therapy [1]. Though the postoperative infectious complications rate is rare [15], they are potentially serious especially the occurrence of urosepsis. However, despite the risk factors reported by several studies, the question of predicting POF was not fully answered. At present, a single dose of antimicrobial prophylaxis 1 hour before surgery is recommended by AUA best practice statement [16]. However, the extended prophylaxis and adjusted postoperative use of antibiotics can be recommended for those individuals who are at high risk of POF. Thus, we tried to demonstrate independent risk factors for the POF and developed a nomogram. The probability derived from the nomogram could help identify high-risk patients.

The most significant predictor from our study was preoperative positive urine nitrite. Besides, to our knowledge, this was the first study reporting that preoperative positive urine nitrite was an independent risk factor for POF after URS, RIRS or PCNL. Nitrite is not normally present in urine unless uropathogens metabolite urinary nitrate. It can be detected if there are recordable and stable concentrations produced by bacteria incubated in the urine. A meta-analysis showed that positive nitrite can indicate urinary tract infection (UTI) with a probability of 99% in children of any age [17]. Similar high sensitivity was also reported in the studies enrolling elderly patients. More importantly, canceling urine cultures in negative leukocyte esterase and nitrite patients could reduce urine cultures by 40% and decreased inappropriate antibiotic use without compromise safety [18‐20]. Also, in the elder hospitalized patients, the diagnosis of a UTI through the presence of bacteriuria in urine culture should be treated with caution because the prevalence of asymptomatic bacteriuria is so high (15–35% in men and 25 to 50% in women) [18]. Therefore, urine nitrite might play a pivotal role in UTI. Unlike urinary culture, urine routine test containing nitrite index is standard practice before RIRS in our center and can obtain results quickly. Demonstrating this risk factor (preoperative positive urine nitrite) can help to decide whether to use antibiotics before the outcome of urine culture. Besides, it can also help to understand if it is necessary to undertake urine culture among patients with preoperative negative urine nitrite. Of course, this requires prospective randomized control trials (RCTs) to further explored.

Paralleling the literature, preoperative positive urine culture was found to be an independent risk factor associated with POF [6, 21‐23]. It was also identified as risk factors for infectious complications after PCNL [24]. In our clinical practice, urine cultures were standard obtained. Prophylaxis with a single dose of antibiotics 1 day before surgery was carried out if patients did not have infectious symptoms and negative urine culture. Otherwise, antibiotic treatment was used. Like the report by Southern et al. [6], we also recognize this exceeded the AUA guidelines which recommend that asymptomatic individuals did not need antibiotic therapy before surgery [11]. However, we thought this stratagem also contributed to the low rate of POF (2.8%).

Longer RIRS time (only the flexible scope period of use) was in our series, another risk factor independently associated with POF after RIRS. In RIRS surgery, longer operative time may not only mean a more complicated operation due to the stone burden but also can strengthen the effect of intrarenal pressure. Several previous studies also showed longer operation time was associated with postoperative infectious complications [5, 6]. Some predictors were not statistically significant might because they contribute to longer operation time together.

Additionally, AGR was found to be an independent protective factor for POF in our cohort. AGR was reported as a simple and valuable factor predicting the prognosis of several diseases, such as lung cancer [25], nasopharyngeal cancer [26], breast cancer [27], and so on. We assumed that albumin negatively affected the POF rate mainly in two ways. On the one hand, albumin is commonly considered as a marker to reflect nutritional status. And the nutritional status is associated with the immune response. Studies had reported that nutritional deficiency status (low albumin level) was associated with the frequency and severity of infection [28, 29]. On the other hand, albumin is likely to mobilize polyunsaturated fatty acids (PUFA) and aid in the formation of several anti-inflammatory lipids [30]. Therefore, low blood albumin levels may tend to pro-inflammatory status. In conclusion, albumin was expected to be negatively interrelated with both nutrition and inflammation, which partially contributed to the POF after RIRS.

After identifying independent risk factors for POF, we aimed to develop a nomogram. Nomogram helped us calculate how many points toward the probability of POF for each risk factor. Though this was not the first nomogram for predicting POF after RIRS, the previous study only enrolled 337 patients [5]. The DCA and excellent calibration in our study indicated that our model could confer net clinical benefit with threshold probabilities at 1–20%. And the POF rate reported in most studies was below 20% [2]. Thus, the nomogram which incorporated these four risk factors could serve as a convenient model for predicting POF among those individuals who undertook RIRS.

The retrospective design in our study was viewed as the main limitation, some valuable variables such as intrapelvic pressure could not be discussed in this study since almost all patients with RIRS received low flow irrigation with about 90 ml/min but it is still an issue worth discussing in the future studies. Our electronic medical records system helps us to track patients for an emergency visit, as well as readmission through other urologists, so it is less likely to lose POF patients based on our definition. Besides, though the patient loss rate met the requirements of statistical power, losing patients due to no record of surgery time was another limitation. Furthermore, we did not test the performance of previously published nomogram (337 patients) because the core independent risk factor C-reacting protein (CRP) in their study is not our routine tested index in our medical region. AGR and urine nitrite are two kinds of laboratory indicators that can be easily obtained in most institutions. Lastly, further large prospective e multicenter studies were needed to evaluate the validity of our model especially urine nitrite and AGR for the prediction of POF after RIRS.

In conclusion, the preoperative urine nitrite, AGR, RIRS time, and preoperative urine culture are found to be independent risk factors associated with POF after RIRS. Then we have developed a nomogram taking these factors into account that accurately predicted POF after RIRS. This model may be conveniently used to identify patients at high risk for POF after RIRS. It can guide both preoperative counseling and scheduling preoperative antibiotic use.