Original study Postoperative Urinary Retention: Risk Factors, Speed of Bladder Filling and Time of Catheterization: An Observational Study as Part of a Randomized Controlled Trial

If risk factors for postoperative urinary catheterization are known adverse events to the lower urinary tract may be prevented. Therefore, postoperative surgical patients were assessed for risk factors for urinary catheterization, for speed of bladder lling and for time till catheterization or spontaneous voiding. The individual maximum bladder capacity was used as threshold for urinary catheterization.


Abstract Background
If risk factors for postoperative urinary catheterization are known adverse events to the lower urinary tract may be prevented. Therefore, postoperative surgical patients were assessed for risk factors for urinary catheterization, for speed of bladder lling and for time till catheterization or spontaneous voiding. The individual maximum bladder capacity was used as threshold for urinary catheterization.

Methods
In this prospective observational study 936 general surgical patients were analyzed for risk factors for urinary catheterization. Patients were 18 years or older and were operated under general or spinal anesthesia without the need for an intra-operative indwelling urinary catheter. The maximum bladder capacity was measured at home by voiding in a calibrated bowl after a strong urge that could no longer be ignored. Postoperatively bladder volumes were hourly assessed with ultrasound. Patients were catheterized after reaching their maximum bladder capacity and being unable to void. Speed of bladder lling and time to catheterization were then calculated.
Catheterization or spontaneous voiding happened around 4 hours postoperatively.

Conclusion
Using the individual maximum bladder capacity, next to the other risk factors, identi es patients at risk for urinary catheterization. These factors should be taken in account whether or not to catheterize the patient to prevent unnecessary urinary catheterization. Considering urine production and maximum bladder capacities, the bladder should be scanned at least within 3 hours postoperatively to prevent overdistention and damage to the lower urinary tract.

Background
Post-Operative Urinary Retention (POUR) followed by urinary catheterization is a well-known and frequent complication after surgery under general or spinal anesthesia. [1][2][3][4] Since the introduction of routine bladder ultrasound the de nition of 'POUR necessitating urinary catheterization' has gradually changed.
This de nition changed from a time limit (= patient must have voided within a certain time period) into a volume limit (= scanned bladder volume in mL). In literature this volume limit can vary between 400 mL to 600 mL. 5,6 In an earlier study we demonstrated a large interindividual variation in maximum bladder volumes. This was independent of age, gender and BMI.

Aim of the study
To identify risk factors for urinary catheterization in a controlled setting. The strength of the risk factors may vary based on data which de nes catheterization need. To this end, we used the data from the RCT where the individual MBC was used as a threshold for urinary catheterization, rather than using a xed bladder volume limit. 1 Next to these risk factors, we calculated the speed of bladder lling and analyzed time to spontaneous voiding and catheterization. The results of these analysis should help health care providers in preventing adverse events to the LUT.

Type of study
This is an observational study analyzing risk factors for urinary catheterization as part of an RCT. 1

Participating patients
All patients provided written informed consent, including permission to use data for additional analysis. Included patients were at least 18 years of age and scheduled to undergo a surgical intervention under general or spinal anesthesia. Peroperatively, there was no anticipated need for an indwelling urinary catheter. Patients were informed and asked to participate during their visit at the pre-assessment anesthesia clinic (PAC). After approval and informed consent patients were asked to go to the restroom to assess the residual bladder volume by ultrasound. At home, the maximum bladder capacity was measured by postponing voiding as long as possible. When a strong urge occurred that no longer could be ignored, they had to void in a calibrated bowl (supplied by the hospital) to measure their maximum voided volume. They were asked to repeat this procedure three times at different moments during a week. The MBC was calculated as the largest voided volume at home minus the residual volume measured at the PAC. The MBC was recorded in the database.
Postoperatively, the bladder of each included patient was scanned every hour till the MBC was reached. At that moment, the patient was asked to void. When spontaneous voiding was not possible, urinary catheterization was performed by the nursing staff. A research assistant performed the bladderscan measurements using ultrasound (The BladderScan BVI 9400, Verathon, Bothell, WA, USA). The original aim was to evaluate the effect of using the MBC in prevention of urinary catheterization compared to a xed bladder volume limit of 500 mL.

Outcome
Pre-planned secondary outcome consisted of analysis of risk factors for urinary catheterization, based on the data from the RCT. 1 Only the data of the MBC group was used for analysis. These results were considered new data using a new de nition for POUR and urinary catheterization. Of the 893 patients in the MBC group who were analyzed in the original RCT for IPSS/QoL (international Prostate Symptoms Score/ Quality of Life score) ( Fig. 1), 43 patients were included with missing data but who could still be analyzed for risk factors (total 936 patients). Pre-and perioperative patient and procedural characteristics, prospectively collected in the original RCT, were considered as potential risk factors for the occurrence of urinary catheterization. 4,8,10 Potential risk factors were divided in unmodi able risk factors that could not be in uenced by anesthesiologists and surgeons, and in modi able risk factors that are under direct control of anesthesiologists (Table 1). Duration of surgery is not under direct control of anesthesiologists and is therefore considered not modi able. For developing possible prediction models, speed of bladder lling in milliliters per hour was calculated and time till catheterization or spontaneous voiding was assessed.

Statistical analysis
Categorical data are presented as counts and percentages. Continuous variables are presented as mean with SD or medians with interquartile ranges, depending on normality of data. For each potential risk factor, differences in the incidence of postoperative urinary catheterization were estimated using a univariate log-binominial regression model. In case of failure to converge, a "modi ed Poisson" approach was applied with robust error variances to estimate crude relative risks and con dence intervals. After univariate analysis of all potential risk factors, those with a P-value < 0.10 were included in the initial multivariable model. A backward elimination strategy was used to achieve the most suitable model to estimate the adjusted relative risks with the nal multivariable model, only including risk factors associated with postoperative urinary catheterization at a level of P < 0.05. In this, also rst order interactions were taken into consideration. A two-tailed P-value < 0.05 was considered to indicate statistical signi cance. All analyzes were performed using SAS software, version 9.4 (SAS institute, Inc, Cary, NC).

FULL MULTIVARIABLE ANALYSIS
In Figure B is displayed a full multivariable analysis for urinary catheterization in the MBC group including all potential risk factors with a level of p < 0.10 from univariate analysis. Using the backward elimination strategy, location of surgery and 'severe' IPSS were not identi ed as independent risk factors in the multivariable analysis.

FINAL MULTIVARIABLE ANALYSIS
In Figure C is displayed the nal multivariable model. Spinal anesthesia was the main modi able risk factor with a RR for hyperbaric bupivacaine of 8.1 and for articaine the RR was 3.1. The unmodi able risk factors MBC (RR 6.7), duration of surgery (RR 5.5), rst scan at PACU ≥ 250 mL (RR 2.1) and age ≥ 60 (RR 2.0) were identi ed as independent risk factors.

TIME OF VOIDING or CATHETERIZATION and SPEED of BLADDER FILLING
In Table 3 is calculated the time from the start of anesthesia till patients were able to void or were catheterized. The speed of bladder lling was estimated during this time period. This was done by subtracting the preoperative scanned bladder volume from the last scanned bladder volume before spontaneous voiding or catheterization. In both general and spinal anesthesia spontaneous voiding occurred after 280 min (4.5hrs). The scanned bladder volume was around 450 mL and the speed of bladder lling of 100 mL/u. Catheterization after general anesthesia happened signi cant later than after spinal anesthesia (352 ± 157 min versus 205 ± 74 min, p < 0.001). Spinal anesthesia patients who were catheterized (203 ± 94 mL/hour, p = 0.005) had a twice as high urine production than patients who voided spontaneously (107 ± 63 mL/hour).

Discussion
To our best knowledge this is the rst study that uses the individual maximum bladder capacity (MBC) to establish risk factors for urinary catheterization after general or spinal anesthesia. The most important modi able risk factor for postoperative urinary catheterization was spinal anesthesia. This is in accordance with literature. 1,2,8 The risk to be catheterized after using hyperbaric bupivacaine was eight times more, and after articaine three times more, compared to general anesthesia. There is scarce literature available comparing general anesthesia with spinal anesthesia and its association with urinary catheterization. [13][14][15][16] There are no recent studies about POUR or urinary catheterization after general anesthesia, let alone comparing their incidence with spinal anesthesia. Most studies about postoperative urinary catheterization are performed in orthopedic patients after spinal anesthesia. During spinal anesthesia the local anesthetics blocks the nerves necessary for spontaneous micturition (S2-S4). The spinal block has to regress till dermatome S3 before voluntary control over the external urethral sphincter has returned. Most patients are able to walk before spontaneous voiding is possible. For bupivacaïne not being able to void may last up to eight hours. 14 Therefore, if one wishes to modify and to reduce the risk for postoperative urinary catheterization (e.g. Day Care surgery), it may be justi ed to change the anesthesia technique. For example, using short acting local anesthetic for spinal anesthesia, or better if possible, use a regional technique (e.g. femoral or popliteal nerve block), or choose general anesthesia.
An MBC smaller than 500 mL was an unmodi able risk factor for urinary catheterization (RR 6.7). Bjerregaard et al, studying orthopedic patients after fast track hip or knee surgery, compared a threshold for POUR of 800 mL versus 500 mL. They found an incidence of 13.4% versus 32.2%. They concluded that a threshold of 800 mL can be set safely, without increasing urological complications. 18 Although their patient group consisted of 'older' patients, they did not know their MBC and their voiding history. A threshold of 800 mL may lead to complications in patients with smaller MBC's (e.g. <500 mL) or in patients with already complaints of the LUT. In general, a strict POUR protocol should be used to prevent bladder overdistention. When the MBC is known, the chance of urinary catheterization can be predicted and this may prevent unnecessary urinary catheterization, not too early and surely not too late.
Duration of surgery was in all analysis a strong unmodi able risk factor (RR 5.1), consistent with similar studies. 19-21 This could be due to the use of more anaesthetic drugs, longer unnoticed bladder lling or, in the case of using long acting spinal anesthesia, an impossibility to void for over eight hours. Shorter surgery time (for example a fast surgeon) can help to lower the incidence of urinary catheterization.
Not voiding before the start of surgery is considered a modi able risk factor for POUR followed by urinary catheterization. In the univariate analysis a preoperative bladder volume ≥ 150 mL was a signi cant risk factor. In the nal multivariable model this signi cance disappeared. Joelsson-Alm found in her prospective study about bladder distention in orthopedic surgery, that a higher preoperative bladder volume is a risk factor for POUR and urinary catheterization. 22 She concluded that encouraging patients to void before leaving for the operating theatre does not necessarily mean an empty bladder at the start of surgery. This is con rmed in our results. Patients who already had a considerable bladder lling at the start of surgery were at risk for large bladder volumes postoperatively. Measuring bladder volumes at the holding area and urging patients to void if needed, can prevent large postoperative bladder volumes.
Indeed, a postoperative bladder volume ≥ 250 mL after the rst scan at the PACU was an important unmodi able risk factor. 23-25 Also, having no urge to void does not mean an empty bladder, 71% of the patients who were catheterized had no urge to void.
In patients ≥ 60 years of age the incidence of urinary catheterization was 18.5%, compared to 5.7% when age < 60 years. Older age is a well-known unmodi able risk factor for postoperative urinary catheterization (RR 2.0) 1-4, 26-28 This could be due to higher IPSS scores in older patients. Or possibly to the different types of surgery performed in older patients: surgery on lower abdomen or lower extremity with longer operation times and the use of long acting spinal anesthesia, with or without the use of ephedrine/atropine. This effect was con rmed in the univariate analysis but disappeared in the nal multivariable analysis ( Figure A and C).
The modi able risk factor 'volume infused and taken orally' over one liter looked as it had a small risk reducing effect, but this was not signi cant (RR 0.7, p < 0.09). Patients had received on average 1.5 liter of uid at the time of voiding or catheterization. In literature, the amount of volume infused was often considered a modi able risk factor for urinary catheterization. [23][24][25] Recent studies showed that the amount of uids given or taken perioperatively is not a signi cant risk factor for urinary catheterization. 1,16,21 Possible modi able risk factors are drugs given peroperatively. 2,3,28 Opioids can have a dual effect on voiding, direct-by partially inhibiting the parasympathetic nerves that innervate the bladder, and indirectby decreasing the awareness of a full bladder and the sensation of urge. Our results could not con rm that piritramide had an effect on the incidence of urinary catheterization (RR 1.0, p = 0.91). We did not registered pain scores as they were titrated below a VAS of 4 (Visual Analogue Scale) following protocol.
Cardiovascular drugs may also have an effect on bladder function by interaction with the sympathetic and parasympathetic nerve system. For atropine and ephedrine this effect was not signi cant. However, the at home used anti-depressant drugs and diazepam did have a signi cant effect on POUR, although their numbers were relatively small. These patients need to be monitored closely.
To estimate the speed of bladder lling after surgery, the time from the start of anesthesia till catheterization or spontaneous voiding was calculated (Table 3). This had been done before by Kreutziger et al. 26 They studied time of voiding and catheterization in 86 patients after spinal anesthesia.
Catheterization happened after ± 200 minutes (2hrs and 20 min) and voiding after ± 270 minutes (3.5hrs), comparable with our results for spinal anesthesia. In our study catheterization happened much later after general anesthesia: only after ± 350 minutes (after almost 6hrs)! This difference in time to catheterization between spinal and general anesthesia can possibly be explained by the difference in speed of bladder lling. In patients who were catheterized, the speed of bladder lling during spinal anesthesia was ± 70 mL/hour more than during general anesthesia (203 mL/h versus 137 mL/h). The speed of bladder lling does not only depend on anesthesia technique but probably also depends on factors such as age, uid infused, antidiuretic hormone production, blood pressure, and is probably not linear. More studies are necessary to con rm our results. Still, with urine production and time of catheterization in mind: to prevent bladder overdistention it can be advised to scan the bladder at least within 3 hours (180 minutes) after the end of surgery. Some patients may then have reached their MBC, resulting in bladder volumes between 300 mL to 540 mL. This is a safe margin for urinary catheterization if the MBC is unknown, maybe too early, but surely not too late. A full bladder extended beyond the maximum bladder capacity for 2 to 3 hours can damage the detrusor muscle and should whenever possible be avoided. 29 In conclusion, in the present study we identi ed important independent risk factors for urinary catheterization. We used the individual maximum bladder capacity as the cut-off bladder volume limit for catheterization. The most important modi able risk factor was spinal anesthesia, and the most important unmodi able risk factors were an MBC < 500 mL, duration of surgery ≥ 60 minutes, the rst scan at the PACU ≥ 250 mL and age ≥ 60 years.