More than 400,000 hysterectomies were performed in 2010, making it the most common major gynecologic surgery in the United States.1 Indwelling Foley catheters, customarily placed at the time of hysterectomy, are a known risk factor for urinary tract infection (UTI).2,3 Reducing catheter-associated UTI is a patient safety issue and a quality improvement measure for institutions; it is recommended that catheters be removed as soon as they are no longer needed or indicated.4
Early catheter removal postoperatively is a potential strategy to reduce the rates of UTI.5–7 A meta-analysis showed that same-day catheter removal decreased the risk of UTI and increased the risk of urinary retention.8 Joshi et al9 have demonstrated that early catheter removal leads to decreased pain perception and that rates of recatheterization were not significantly higher. However, because of differences in protocols and relatively small numbers of patients, the magnitude of UTI risk reduction is not well described among patients who had the Foley catheter removed on the same day as their hysterectomy versus on postoperative day 1 or later. The aims of this study were to examine the effect of length of catheterization on the rate of UTI after hysterectomy by specifically examining odds of UTI relative to the number of days of catheterization and to determine other patient and perioperative factors that may be independently associated with UTI.
MATERIALS AND METHODS
This was a retrospective case-control study using data collected by the Michigan Surgical Quality Collaborative (MSQC), a group of 52 academic and community hospitals that voluntarily abstract data regarding perioperative surgical care for gynecologic, vascular, and general surgery cases. Funding for the collaborative is provided by Blue Cross Blue Shield of Michigan/Blue Care Network. Surgical clinical quality reviewers at each clinical site, employed by the respective hospitals, abstract demographics and preoperative, intraoperative, and postoperative details of care. A standardized collection method is used for chart abstraction at each hospital in the collaborative. Patients from all insurance payers are included in the case sampling. At each MSQC hospital, the first 25 cases of an 8-day cycle are eligible for chart abstraction using a standardized collection method. The 8-day cycle is used so that each cycle starts on a new day of the week and promotes sampling of surgeons who usually operate on certain days. Data are validated by routine site visits, conference calls, and internal audits. This study was designated “not regulated” by The University of Michigan Institutional Review Board (HUM00073978).
Hysterectomies performed from January 2, 2013, to July 2, 2014, were included. In the collaborative, data such as insurance status and indication for procedure were included and abstracted at this time. Before that, this information was not available and that is why data before this time was not included. Data including patient demographics, type of hysterectomy, appropriate operative antibiotic prophylaxis, concomitant procedures (including cystoscopy, pelvic organ prolapse [POP] repair, or sling), intraoperative complications, surgical time, length of catheterization, and UTI were analyzed. Appropriate antibiotic prophylaxis has been defined in previous MSQC studies as a β-lactam, appropriate alternative regimen, or appropriate antibiotic plus additional coverage, per the American Congress of Obstetricians and Gynecologists and Joint Commission Surgical Care Improvement Project guidelines.10–12 Cases were defined as those patients with postoperative UTI. Urinary tract infection was defined as either the presence of symptoms and a positive culture (>105) or a positive urinalysis and low colony count culture (≥103 and <105) within 30 days of hysterectomy. A positive urinalysis is defined in the database as a dipstick positive for leukocyte esterase and/or nitrite, 10 white blood cell counts per cubic millimeter or higher of unspun urine or 3 white blood cell counts per high-power field or higher of spun urine, or microorganisms seen on Gram stain of unspun urine. Cases were grouped by length of exposure to the indwelling Foley catheter into the following mutually exclusive groups: (1) low exposure—no catheter was placed or the catheter was removed the day of the hysterectomy (postoperative day 0), (2) intermediate exposure—the catheter was removed on postoperative day 1, (3) high exposure—the catheter was removed on postoperative day 2 or later but before hospital discharge, and (4) highest exposure—the patient was discharged home with an indwelling catheter. All combinations of catheter exposure and subject length of stay were reviewed for clinical plausibility and temporality. Patients with missing or implausible information regarding Foley catheter placement or discontinuation were excluded.
Rates of UTI were calculated for each catheter exposure group and compared using a χ2 test. Bivariate analysis was used to identify whether the following variables were independently associated with UTI after hysterectomy: age; race; primary insurance; body mass index; American Society of Anesthesiology classification; preoperative diagnosis of diabetes (including both medically and diet-controlled type I and II); smoking status; functional status classification; diagnosis of gynecologic cancer based on International Classification of Diseases-9 diagnosis codes; surgical time; estimated blood loss; length of stay; route of hysterectomy (abdominal, laparoscopic-assisted vaginal, laparoscopic, robotic, and vaginal); concomitant procedures such as cystoscopy, sling, and POP repair; perioperative blood transfusion (including preoperative, intraoperative, and postoperative); intraoperative injury to bowel, bladder, or ureter; and any other surgical site infection, readmission, and reoperation. To capture menopausal status, a sensitivity analysis was performed to identify the clinically meaningful age cut point relative to UTI of 50 years. Comparisons between patients with UTI and no UTI were performed using χ2 tests for categorical variables or Fisher exact test in the case of small cell sizes. For continuous variables, normality was assessed with Shapiro-Wilk test and skew and kurtosis. Normally distributed variables were compared using Student t test, and nonnormally distributed variables were compared using nonparametric Wilcoxon-Mann-Whitney tests.
Variables independently associated with UTI after hysterectomy from bivariate analyses were considered for further analysis using stepwise logistic regression. Multivariable hierarchical logistic regression models accounting for hospital variation were developed to identify independent risk factors for UTI. Variables that were excluded from initial stepwise regression but deemed clinically relevant were appended in the final model. Spearman rank and Pearson correlation matrices were used in the final model to assess for collinearity and confounding among candidate covariates. Observed and model-predicted rates were assigned by decile, quintile, and quartile to assess goodness of fit. Model fit was tested with Pearson residuals, with strong fit being close to 1, and calculation of the C-statistic for appropriate concordance. A clinically relevant interaction term was tested in the final model to account for conditional odds of UTI and reflects variation in practice patterns with concomitant procedures during hysterectomy. The interaction term accounted for both vaginal route of hysterectomy and concomitant performance of POP repair and/or midurethral sling to quantify the adjusted odds of developing UTI given these 2 factors that could be dependent on one another. All statistical analyses were generated using STATA v13 (StataCorp LP, College Station, Tex) and SAS Version 9.4 (SAS Institute, Inc, Cary, NC) with P < 0.05 considered significant for all analyses.
Data on 12,098 hysterectomies were available. A total of 10,354 hysterectomies had complete data for analysis; 2.3% (n = 222) of these cases were complicated by UTI. There were 1744 cases (14.4%) excluded from the analysis because of incomplete data, and among these subjects, UTI complicated 1.0% (n = 17). Each subject was grouped by catheter exposure: low exposure (no catheter placed or catheter removed postoperative day 0) (n = 2915), intermediate exposure (catheter removed postoperative day 1) (n = 6297), high exposure (catheter removed postoperative day 2 or later) (n = 802), and highest exposure (patient discharged home with indwelling Foley catheter) (n = 340). The UTI rate for the low exposure group was 1.3% (n = 37/2915), intermediate exposure was 2.1% (n = 130/6297), high exposure was 4.1% (n = 33/802), and highest exposure was 6.5% (n = 22/340), with P < 0.0001 for analysis across all groups.
Bivariate analysis was performed for demographics, medical comorbidities, and perioperative variables relative to UTI prevalence (Table 1). With regard to demographic characteristics, compared with those without UTI, women with UTI were more likely to be older than 50 years, have diabetes, be functionally dependent, and have a diagnosis of gynecologic cancer. With regard to surgical factors, subjects with a postoperative UTI were more likely to have another surgical site infection, intraoperative bladder or ureteral injury, longer surgical time by nearly 30 minutes, a greater estimated blood loss, and a postoperative readmission. Compared with other hysterectomy routes, women who had a vaginal hysterectomy were significantly more likely to have a postoperative UTI, and women with robotic hysterectomy were less likely to have a UTI (Table 2). Subjects who had concomitant POP repair or midurethral sling were more likely to have a UTI, but those who had cystoscopy alone were not (Table 2).
Independent risk factors for UTI were identified using multivariable hierarchical logistic regression (Table 3). The model calibration was measured with a C-statistic of 0.71, and Pearson residual was 0.89. Patients in the high and highest catheter exposure groups had greater odds of postoperative UTI compared with those in the low exposure group. Dependent functional status was also independently associated with greater odds of UTI. The relationship between a vaginal route of hysterectomy and a concomitant performance of POP repair and/or sling was described with an interaction term in the final model. The conditional adjusted odds ratios (ORs), shown in Figure 1, quantify how the odds of postoperative UTI are associated with concomitant procedures (POP repair, but not route of hysterectomy). Among hysterectomies that had POP repair and/or sling, those with a vaginal approach had 2.13 greater odds of UTI compared with those with an abdominal or laparoscopic/robotic approach. Among hysterectomies with a vaginal approach, those who included POP repair and/or sling as a concomitant procedure had 2.58 greater odds of UTI than vaginal hysterectomies without those concomitant procedures. Vaginal hysterectomy alone was not associated with increased odds of UTI compared with abdominal hysterectomy (OR = 0.80, Fig. 1). An abdominal or laparoscopic hysterectomy done with a POP repair or sling was not associated with a change in the odds of UTI compared with abdominal or laparoscopic hysterectomy alone (OR = 0.97, Fig. 1).
In this study, we found that longer catheter time increased the odds of UTI, with the highest rates of UTI occurring in those who had the longest catheter exposure. Although there is the risk of recatheterization to consider with early catheter removal, this demonstrable increase in odds of UTI in a large database of patients undergoing hysterectomy supports early removal protocols. Patients with a postoperative UTI in this study were more likely to have additional surgical site infections and to be readmitted. Efforts to reduce complications are a primary target to reduce readmissions;13 therefore, reducing postoperative UTI rates is an important quality issue.
The finding that intraoperative bladder and ureteral injury were related to higher rates of UTI is expected, because these groups often retain their catheters for extended periods postoperatively. Also in our model, longer surgical time was independently associated with postoperative UTI. This may be related to the complexity of the surgery, but operative time could be another modifiable risk factor for UTI; it has also been noted in the literature as a risk factor for complications after hysterectomy.14 In addition, the association of dependent functional status with postoperative UTI identifies such patients as a particularly high-risk group that may be worthy of focused study on strategies to mitigate risk. The performance of concomitant cystoscopy alone did not increase the rate of UTI; thus, institutions considering implementation of universal cystoscopy policies at the time of hysterectomy may not need to be concerned about an adverse effect of such a policy on their UTI rate.
Urinary tract infection is a known complication of pelvic reconstructive surgery.15,16 We used an interaction term to account for the individual roles that route of hysterectomy and concomitant prolapse or incontinence procedures play in the odds of UTI. Our analysis indicates that a vaginal hysterectomy does not, on its own, increase the odds of UTI. Likewise, in our analysis, the concomitant performance of pelvic reconstruction at the time of hysterectomy did not, on its own, drive up the odds of UTI. Rather, there was an interaction between vaginal route of hysterectomy and pelvic reconstruction that drove up the odds of UTI. The conditional adjusted ORs provide evidence of the dependency of concomitant procedures on the odds of the outcome of UTI.
Our study has several limitations. We were able to determine the day of catheter removal, but the time of removal was not recorded, and we could not quantify hours of catheter exposure. Only patients with an indwelling Foley catheter are captured in this analysis; those who performed intermittent catheterization were not identified in the data collection protocol. The performance of a void trial after surgery is not abstracted in the database, and therefore, repeated catheterization attempts after failed voiding trials were not captured but could also increase the risk of UTI. It is also possible that we may not have identified all UTIs. The nurse abstractor at each institution reviews the medical record for events occurring within 30 days of surgery and will attempt to contact the patients if no routine follow-up is available, but it is possible that postoperative complications treated at another hospital system were not known. Lastly, it was necessary to exclude 14% of the patients who underwent hysterectomy because of incomplete data on whether a catheter was used or the length of catheterization. As a result, we could not assign these patients to an exposure group and therefore excluded these women from the analysis. In the majority of these cases, there was no documentation of intraoperative or postoperative use of a Foley catheter; therefore, we suspect that no indwelling catheter was ever placed. If so, these cases represent an ultralow exposure group. This would be consistent with the even lower UTI rate among these patients of 1.0%. Despite these missing patients, our overall UTI rate was similar to previous studies.1
Our study is strengthened by the large number of included patients, which allows us to identify important associations despite a low disease occurrence. In addition, the database captures 30-day postoperative day data, does not rely on billing codes, and has well-defined specific criteria for symptomatic UTI.
In summary, this analysis supports development of protocols that shorten the length of time a catheter is used to decrease the rate of postoperative UTI after hysterectomy.17 Consideration should be given to removal of an indwelling Foley catheter on postoperative day 0 when clinically feasible.
The authors thank Sarah Block for manuscript assistance and Ryan Jakubowski for graphic design.
1. Wright JD, Herzog TJ, Tsui J, et al. Nationwide trends in the performance of inpatient hysterectomy
in the United States. Obstet Gynecol
2013;122(2 Pt 1):233–241.
2. Lake AG, McPencow AM, Dick-Biascoechea MA, et al. Surgical site infection after hysterectomy
. Am J Obstet Gynecol
2013;209(5):490 e491–490 e499.
3. Dieter AA, Amundsen CL, Edenfield AL, et al. Oral antibiotics to prevent postoperative urinary tract infection
: a randomized controlled trial. Obstet Gynecol
4. Hooton TM, Bradley SF, Cardenas DD, et al. Diagnosis, prevention, and treatment of catheter
-associated urinary tract infection
in adults: 2009 International Clinical Practice Guidelines from the Infectious Diseases Society of America. Clin Infect Dis
5. Ahmed MR, Sayed Ahmed WA, Atwa KA, et al. Timing of urinary catheter
removal after uncomplicated total abdominal hysterectomy
: a prospective randomized trial. Eur J Obstet Gynecol Reprod Biol
6. Chai J, Pun TC. A prospective randomized trial to compare immediate and 24-hour delayed catheter
removal following total abdominal hysterectomy
. Acta Obstet Gynecol Scand
7. Dobbs SP, Jackson SR, Wilson AM, et al. A prospective, randomized trial comparing continuous bladder drainage with catheterization at abdominal hysterectomy
. Br J Urol
8. Zhang P, Hu WL, Cheng B, et al. A systematic review and meta-analysis comparing immediate and delayed catheter
removal following uncomplicated hysterectomy
. Int Urogynecol J
9. Joshi B, Aggarwal N, Chopra S, et al. A prospective randomized controlled comparison of immediate versus late removal of urinary catheter
after abdominal hysterectomy
. J Midlife Health
10. Uppal S, Harris J, Al-Niaimi A, et al. Prophylactic antibiotic choice and risk of surgical site infection after hysterectomy
. Obstet Gynecol
11. ACOG Committee on Practice Bulletins—Gynecology. ACOG practice bulletin No. 104: antibiotic prophylaxis for gynecologic procedures. Obstet Gynecol
12. The Joint Commission. Specifications Manual for National Hospital Inpatient Quality Measures. The Joint Commission. Accessed October, 2015.
13. Dessources K, Hou JY, Tergas AI, et al. Factors associated with 30-day hospital readmission after hysterectomy
. Obstet Gynecol
14. Catanzarite T, Saha S, Pilecki MA, et al. Longer operative time during benign laparoscopic and robotic hysterectomy
is associated with increased 30-day perioperative complications. J Minim Invasive Gynecol
15. Sutkin G, Alperin M, Meyn L, et al. Symptomatic urinary tract infections after surgery for prolapse and/or incontinence. Int Urogynecol J
16. Grimes CL, Lukacz ES. Urinary tract infections. Female Pelvic Med Reconstr Surg
17. Hakvoort RA, Thijs SD, Bouwmeester FW, et al. Comparing clean intermittent catheterisation and transurethral indwelling catheterisation for incomplete voiding after vaginal prolapse surgery: a multicentre randomised trial. BJOG
Keywords:Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.
catheter; hysterectomy; urinary tract infection; postoperative; risk factors