Postoperative Urinary Catheterization Thresholds of 500 versus 800 ml after Fast-track Total Hip and Knee Arthroplasty: A Randomized, Open-label, Controlled Trial

POSTOPERATIVE urinary retention (POUR) may occur after almost any type of surgical procedure,¹  but is especially relevant in joint arthroplasty, in which a resulting urinary tract infection (UTI) may be a risk factor for periprosthetic infection.²  Hence, there is a focus on reducing unnecessary urinary catheterization by, for example, using bladder ultrasound and intermittent catheterization in appropriate patients.³  Consequently, the most used approach for perioperative bladder management in arthroplasty today is a catheterize-when-needed strategy, based on the patient’s bladder volume (BV) assessed by ultrasound bladder scanners.⁴  However, due to the lack of a precise, unambiguous definition of POUR, there are no evidence-based recommendations on the optimal catheterization threshold (CT) for treating POUR.⁵  Consequently, most authors recommend a BV of 500 to 600 ml as CT, mainly based on the assumption that large BVs may overstretch the bladder wall, potentially causing persistent voiding difficulties.^1,2,5–7  To date, only one randomized, controlled trial (RCT), including 1,840 patients from a mixed surgical population, has challenged this non–evidence-based CT.⁶  However, the use of an individualized CT in the index group and catheterization of index patients at BVs less than 500 ml means that no conclusions can be drawn about the use of an increased CT in general.⁶  Therefore, the present study is the first detailed, procedure-specific RCT to compare two BVs as CT in a group of unselected patients who underwent fast-track THA and TKA, based on the hypothesis that allowing transient BVs up to 800 ml would reduce the need for postoperative catheterization and the incidence of urological complications.

The study was approved by the Regional Ethics Committees of the Capital Region, Hillerød, Denmark (Reference number H-1-2014-024), and the Danish Data Protection Agency (Id. 02795; 30-1209) and was registered at clinicaltrials.gov on April 24, 2014 (Identifier: NCT02126813, principal investigator: Dr. Bjerregaard; https://www.clinicaltrials.gov/ct2/show/NCT02126813?term=POUR-RCT&rank=1).

This was a randomized, controlled, open-label trial performed in three Danish, high-volume, orthopedic departments, all of which used similar fast-track protocols for the perioperative setup, including early mobilization, multimodal opioid-sparing analgesia, and functional discharge criteria.⁸  Inclusion criteria were age greater than or equal to 18 yr; planned elective primary THA or TKA, and had provided informed, written consent. Exclusion criteria were previous cystectomy, use of preoperative urinary bladder catheterization, planned perioperative indwelling catheterization (by the operating surgeon or the anesthesiologist on duty), pregnancy, and/or had given birth within 6 months before surgery.

Screening for eligibility and information of possible participants was performed at preoperative information meetings 1 to 2 weeks before surgery. Informed consent was obtained either at the end of the information meetings or on the day of admission. Patients who did not attend the information meetings were informed on the day of admission, and consent was obtained thereafter.

Before the start of the study, the primary investigator prepared 800 sealed, opaque envelopes based upon a computer-generated, 1:1 ratio block randomization of numbers from 1 to 800 (block size of 40 and no stratification). Each study site started with 120 envelopes, receiving supplemental quantities according to enrollment rates.

The allocation of patients to either 500 or 800 ml CT was performed before surgery by a center subinvestigator by opening the next envelope in line. Each envelope contained a registration form, designed as a flowchart with explicit instructions for the perioperative bladder management and printed with the randomization number of the patient (corresponding to the number on the envelope). Names and social security numbers of the patients were entered in the registration forms. It was not possible for us to mask patients, caregivers, or any investigator to group assignment after allocation.

Surgery was performed according to the guidelines of the participating departments. Low-dose spinal anesthesia (SA) without the use of intrathecal opioids but supplemented with optional propofol sedation was considered the standard. General anesthesia (GA) was offered if requested by the patient. Volumes of intraoperative fluid administration were not standardized, but registered in the patient’s case report form, as were the type of surgery, the anesthetic technique, and whether intraoperative bleeding exceeded 500 ml.

For postoperative analgesia, a multimodal approach of opioid-sparing oral treatment combined with intraoperative local infiltration analgesia (in TKA) was used.⁸  Epidural analgesia or patient-controlled analgesia was not an option. After surgery, patients were observed in the postanesthesia care unit until fulfilling standard, predefined discharge criteria based on recommendations from the Danish Society of Anesthesiology and Intensive Care Medicine.⁹  In the ward, mobilization was initiated within 2 to 6 hours postoperatively.

In addition to standard baseline characteristics such as age and sex, we registered whether patients had any urological comorbidity (self-reported prostate hypertrophy, “overactive bladder,” prostate cancer, urinary incontinence, and/or previous prostatectomy/prostate resection), had any neurological comorbidity (multiple sclerosis, Parkinson disease, neurologic sequelae after a disc prolapse, diabetic neuropathy, and/or sequelae after a cerebral insult), or had previously undergone pelvic gynecological surgery (hysterectomy and/or prolapse surgery), as these factors have been proposed to increase the risk of POUR.^1,2,10  Also, we registered if the patient had a daily intake of diuretics.

Preoperatively, patients completed the International Prostate Symptom Score (IPSS) questionnaire (Danish version),¹¹  in which each of the seven questions yields a score between 0 (“symptom never present”) and 5 (“symptom always present”), thus giving a total score of 0 to 35. The IPSS questionnaire was repeated in a follow-up telephone interview at postoperative day (POD) 30, allowing calculation of pre- to postoperative differences in the IPSS (diff-IPSS), with a diff-IPSS greater than or equal to 1 indicating postoperative aggravation of lower urinary tract symptoms. From end of surgery and until patient’s first voluntary micturition, bladder scans were performed at 2-hour intervals by trained nurses, using ultrasound bladder scanners (BladderScan^® BVI 3000, 6100 and 9400; Verathon Medical Europe B.V., Netherlands). If a scanned BV was greater than 500 or 800 ml (depending on group assignment), the patient was encouraged to void but was catheterized if not succeeding. In case a patient was intermittently catheterized twice and still incapable of voluntary micturition, an indwelling catheter was inserted and left in place for 24 h. Any patient incapable of voluntary micturition, who had abdominal pain/discomfort and/or an urge to void, was catheterized despite the scanned BV.

The following was registered on the registration form, which followed the patient from before surgery until the first voluntary micturition: time for the last preoperative micturition; time, result, and consequence of each bladder scan; volumes of urine emptied by catheterization (if relevant); date and time of the first postoperative voluntary micturition; whether the patient had received an indwelling catheter (and why, if “yes”); and whether urinary catheterization was performed because of symptomatic POUR.

The conduction of the trial under everyday settings with registrations made by the clinical personnel resulted in intervals between bladder scans that sometimes exceeded 2 h. Consequently, a maximal delay of 1 h was accepted. If more than 3 h elapsed, the patient was kept in the per-protocol analysis if the next scan showed a BV less than the allocated CT. If the next scan showed a BV larger than the allocated CT or if the patient had voluntary micturition before a new scan was performed, the patient was excluded from the per-protocol analysis.

All patients were followed until discharge by the study site subinvestigator who registered any urological complications and/or diagnosed UTI. Follow-up was a telephone interview at POD30 by the study site subinvestigator asking whether the patient had had a UTI and/or had been readmitted. In case of confirmatory answers, further details such as “who diagnosed the UTI,” “where was the urinary test performed,” and “what was the reason for readmission” were asked. Finally, the patient was asked the seven questions of the IPSS questionnaire.

Our primary outcome was the number of patients needing postoperative catheterization before their first voluntary micturition. Also, we assessed catheterization frequencies after SA versus GA and the number of patients who received repeated intermittent catheterizations, who received indwelling catheterization, and who were catheterized due to symptomatic POUR.

Secondary outcomes were incidences of UTIs and urology-related readmissions within 30 days of surgery and number of patients with voiding difficulties commenced postoperatively, defined as a total diff-IPSS greater than or equal to 1. Incidences are reported individually for UTIs diagnosed by healthcare professionals and those not diagnosed by healthcare professionals. Regarding total diff-IPSS, we report both group medians (interquartile range [IQR]) and number of patients (%) with total diff-IPSS greater than or equal to 1. The last secondary outcome in the protocol was “time intervals between the last preoperative micturition; end of surgery; urinary catheterization (if relevant); and the first postoperative, voluntary micturition.” Intervals of relevance for this article were the time from the last preoperative to the first postoperative, voluntary micturition, and the time from end of surgery to the first intermittent catheterization (if relevant), as the former may tell us when to expect regain of normal bladder control, whereas the latter may prove useful in deciding when and how to perform postoperative bladder scans.

The incidences of UTIs, urology-related readmissions, and postoperative voiding difficulties (surrogated by the diff-IPSS) were considered safety measures. No additional systematic registration of adverse events was done, since all procedures were performed according to current clinical practice.

In a previous observational study of a similar population in the centers, we found that about 40% received postoperative urinary catheterization and that about 20% had emptied BVs between 500 and 800 ml by catheterization.¹²  To detect a conservative risk reduction of 30% (absolute risk reduction from 40 to 28%), with 90% power, 0.05 significance level (two-sided), and expected dropout rate of 5%, 338 patients were needed in each group. To accommodate unrecognized inaccuracies and the potential risk of type 2 errors in the assessment of secondary outcomes, we chose to include 400 patients in each group.

Based on histograms, probability plots, and the Kolmogorov–Smirnov goodness-of-fit test, none of our continuous variables were satisfactory normally distributed. Consequently, they are reported as medians with interquartile ranges (IQR) and compared between groups by the Mann–Whitney U test. IPSS and diff-IPSS were considered continuous variables. Categorical data are given as counts and group percentages and were compared between groups by the Fisher exact test. Furthermore, we calculated the risk difference (RD) and relative risk (RR) with corresponding 95% CI for all categorical outcomes.

All analyses were planned as per protocol. However, since we had to exclude data on 59 patients (7.4%) due to protocol violations (n = 53) and insufficient postoperative registrations (n = 6), we did a post hoc, modified intention-to-treat analysis in a population with inclusion of these 59 patients. The outcomes analyzed were the risk of catheterization, the incidence of UTI, and the total diff-IPSS. For the risk of catheterization, we used a “worst-case scenario,” assuming postoperative catheterization in all of the 26 excluded patients from the 800-ml group and no postoperative catheterization in all of the 33 excluded patients from the 500-ml group (fig. 1). Data analysis was performed by the primary investigator, using SAS version 9.3 (SAS Institute, Inc., USA), with P values less than 0.05 (two-sided) considered statistically significant.

One thousand two hundred ninety-six patients were assessed for eligibility between April 29, 2014, and May 28, 2015, of which 800 (62%) were randomly allocated to a postoperative CT of 500 or 800 ml (400 patients [50%] in each group). Three patients (0.4%) withdrew their consent after allocation (not allowing the use of their data), leaving 398 and 399 patients in the 500-ml and the 800-ml group, respectively. Furthermore, six patients (0.8%) did not receive the allocated intervention, six discontinued the intervention (0.8%), five were lost to follow-up (0.6%), and 59 (7.4%) were excluded from the per-protocol analysis, leaving 721 patients (90%) of whom 354 were assigned to the 500-ml group and 367 were assigned to the 800-ml group (fig. 1).

Baseline characteristics, relevant comorbidity, and intraoperative data are given for all consenting, allocated patients in tables 1 and 2, showing an acceptable balancing between groups. Excluded patients did not differ from those included in the per-protocol analysis (data not shown). The median length of stay was 2 days (IQR, 1 to 2) in both groups.

The results of the per-protocol analysis are shown in table 3. We found a significantly reduced proportion of catheterized patients in the 800-ml group (49 of 367 [13.4%]) compared to the 500-ml group (114 of 354 [32.2%]), giving an RR of 0.4 (95% CI, 0.3 to 0.6; P < 0.0001). Also the risk of repeated intermittent catheterization was significantly reduced in the 800-ml group (RR, 0.2 [0.1 to 0.6]; P = 0.0020). No difference was demonstrated for indwelling catheterization, but the number of incidences was minimal. Among the patients who had received SA, catheterization frequencies were 106 of 311 (34%) and 45 of 326 (14%) in the 500-ml and 800-ml group, respectively, whereas they were 8 of 43 (19%) and 4 of 41 (10%) among those who had received GA.

The number of UTIs diagnosed within 30 days postoperatively by a healthcare professional was similar in the two groups (7 of 354 [2%] and 8 of 367 [2%]; P = 1.0), and we found no significant differences between groups in the distributions of total diff-IPSS (P = 0.9787) or in the number of patients with total diff-IPSS greater than or equal to 1 (P = 0.4203). The diff-IPSS data from each of the seven symptom questions also showed no significant differences between groups (all P > 0.1000, see appendices 1 and 2). In addition to the UTIs shown in table 3, three patients (less than 1%) from the 500-ml group and four patients (1%) from the 800-ml group reported that they had had a UTI, which was not diagnosed by a healthcare professional. Only one patient had a urology-related readmission within the 30-day follow-up. This was a 74-yr-old man from the 800-ml group, catheterized once before his first postoperative voluntary micturition, who was discharged at POD2 but readmitted at POD4 with fever and bacteriuria. After the first voluntary micturition, he required repeated intermittent catheterizations and had suffered similar symptoms and treatment after a previous THA.

The time intervals between the last preoperative and the first postoperative voluntary micturition were similar in the two groups (P = 0.8201), whereas the intervals between the end of surgery and the first catheterization were shorter in the 500-ml group than in the 800-ml group (P = 0.0003) (table 3).

In the post hoc intention-to-treat analysis, 114 of 387 patients (29.5%) were catheterized in the 500-ml group compared to 75 of 393 patients (19.1%) in the 800-ml group (RD, 10.4% [4.4 to 16.4]; RR, 0.65 [0.50 to 0.84]; P = 0.0008). Nine patients (2%) in each group had a UTI (P = 1.0000) with no additional readmissions. Median total diff-IPSS was zero (IQRs, −2 to 1) in both groups (P = 0.6599) and proportions with total diff-IPSS greater than or equal to 1 were 116 of 379 (31%) in the 500-ml group versus 134 of 382 (35%) in the 800-ml group (RD, 4.5% [−2.2 to 11.1]; RR, 1.15 [0.93 to 1.41]; P = 0.1907).

Our results showed that a CT of 800 ml significantly reduced the risk of postoperative urinary catheterization in fast-track THA and TKA without increasing the risk of UTIs, urology-related readmissions, or lower urinary tract symptoms compared to a CT of 500 ml, and the results of our intention-to-treat analysis add further support to these findings.

One may argue that our hypothesis of a reduced catheterization frequency in the 800-ml group was a self-fulfilling prophecy, but since previous studies had reported an increased risk of persistent POUR at BVs greater than or equal to 500 or 700 ml,^13,14  we do not believe that the results were given beforehand. Despite data from previous RCTs suggesting that it is safe to allow transient BVs larger than 500 ml in some patients,^6,15  this large, procedure-specific trial is the first to provide high-quality evidence for the safe and beneficial use of a CT of 800 ml in an unselected group of patients undergoing fast-track THA and TKA.

Although it is still debated whether to recommend 500 or 600 ml as CT,^{1,2,4–7,10,16,17}  we chose 800 ml as CT in the intervention group, since a short lasting BV of up to 1,000 ml has been suggested to be non-harmful.^1,5,15  Among the 59 patients we excluded from the per-protocol analysis, 36 (61%) were not bladder scanned within the planned time intervals and five (8%) were not catheterized according to protocol (fig. 1). We chose not to exclude patients who were scanned later than 3 h from the last scan, if they had not reached a BV above their allocated CT, since no potential catheterization was “missed” in these patients. On the other hand, we did exclude 36 patients who might have reached their CT, but who were not catheterized due to a delayed/missing bladder scan, and therefore could bias our primary outcome. However, the majority of the excluded patients were not subsequently catheterized, and despite that they may have had BVs greater than 800 ml for some hours, their inclusion in the post hoc analysis did not increase the incidences of urological complications. Brouwer et al.⁶  randomized 1,840 mixed surgical patients to a standard CT of 500 ml or an individualized CT based on the maximal preoperative bladder capacity. In the intervention group, the mean CT was 611 ml (SD = 209 ml), which under the assumption of normally distributed data suggests a CT greater than 800 ml in about 15% of the patients.⁶  Therefore, both our post hoc analysis and the findings by Brouwer et al. support the safe use of CTs larger than 800 ml.

The catheterization frequency in the 500-ml group was slightly less than what we expected from our previous findings,¹²  indicating a positive effect of using a well-defined CT and regular bladder scans after THA and TKA, as suggested previously.^18,19  Catheterization was more frequent after SA than after GA in both groups, which in accordance with the current literature supports SA as a risk factor for POUR.^1,2,7,12  Considering that recent RCTs have suggested GA to have some advantages on recovery and patient satisfaction compared to SA in THA and TKA,^18,19  it may be appropriate to reconsider the recommendations for the optimal anesthetic technique. Thus, adequately powered RCTs are warranted to specifically compare GA and SA in causing POUR after fast-track THA and TKA.²⁰ 

The six patients catheterized because of symptomatic POUR were all allocated to the 800-ml group (table 3), suggesting that BVs between 500 and 800 ml may trigger a sensory threshold for the voiding reflex at times when voluntary motoric bladder control has not yet been regained in the majority of patients. However, for obvious reasons, these patients are easily identified and catheterized.

In contrast to what we hypothesized, the reduced catheterization rate did not reduce the incidence of UTIs (table 3), which may be explained by the low overall incidence of UTIs or because both urinary catheterization and urinary retention may predispose to UTI.²  The similar pre- to postoperative IPSS differences in the two groups (table 3 and appendix 1) suggest postoperative urinary tract symptoms to be independent of group assignment, thereby indicating no harmful effects of transitory BVs up to 800 ml.

Not surprisingly, the timespans from end of surgery to first bladder catheterization were significantly shorter in the 500-ml than in the 800-ml group, suggesting that if a CT of 800 ml is used, catheterization may first be needed later than 2 h postoperatively (table 3). More surprisingly, the timespans between the last preoperative and the first postoperative voluntary micturition were almost the same in the two groups (table 3), suggesting that recovery of bladder function may be independent of the maximal previous BV and whether or not the patient was catheterized before the first voluntary micturition.

The lack of a clear definition of POUR has resulted in great variance in reported incidences^1,16,20 ; thus, if POUR is defined by a BV (hence a CT), a low CT will result in a high incidence of POUR and a high catheterization frequency. Much effort has been made to compare indwelling versus intermittent catheterization^4,21–23  and to identify risk factors for POUR,^24–27  but the lack of consensus on how to define POUR, and when to catheterize, makes it difficult to compare outcomes between studies.²²  In fast-track THA and TKA, opioid-sparing analgesia and early mobilization may facilitate bladder function restoration, probably explaining why some of the “classic” risk factors for POUR seem without influence.^12,28  The results of this study may have widespread implications, not only sparing many patients the inconvenience of unnecessary urinary catheterization but also potentially saving resources in the health care system.^29,30 

This study is limited by its unblinded design. However, we prioritized that bladder scans and urinary catheterization were performed by the clinical personnel who normally perform these tasks, which made it practically impossible to mask patients, caregivers, or study site investigators. Consequently, masking of the primary investigator was considered disproportionally resourceful. Another limitation was the relatively large amount of protocol violations resulting in exclusion of 59 patients (7.4%) from the primary analysis. However, the results of our post hoc analysis did not suggest that the excluded patients had higher incidences of urological complications than those included in the per-protocol analysis. Anesthetic technique and intraoperative fluids administered were not standardized, but were well balanced between groups (table 2). Furthermore, we did not register whether patients were catheterized between their first voluntary micturition and the 30-day follow-up, hindering any conclusions on a possible relation between catheterization and UTI. Finally, this superiority study was statistically powered to show a risk difference in postoperative catheterization. Consequently, the insignificant differences in secondary outcomes may have been caused by type 2 errors. However, given the low incidence of urological complications, one may question the clinical relevance of such a potentially overlooked difference on our safety measures.

The strengths of our study are the 30-day follow-up on urological complications and the use of the IPSS to quantify relevant urinary tract symptoms after surgery. Also, the simple study design allowed for conducting the study under normal everyday clinical settings, thereby making standards for perioperative bladder management directly implementable.

In conclusion, this study is the first to provide evidence for the use of a BV of 800 ml as CT to reduce intermittent catheterization in an unselected group of patients undergoing fast-track THA and TKA and should serve as basis for future guidelines on perioperative bladder management in both arthroplasty and other surgical specialties. However, a very large trial will be needed to draw definitive conclusions on differences in complication rates.

The study was financially supported by the Lundbeck Foundation, Copenhagen, Denmark (grant number R25-A2702), which had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

Dr. Kehlet is a member of the Zimmer Biomet advisory board for Rapid Recovery Program, Dordrecht, The Netherlands. The other authors declare no competing interests.

Full protocol available from Dr. Bjerregaard: lars.stryhn.bjerregaard@regionh.dk. Raw data available from Dr. Bjerregaard: lars.stryhn.bjerregaard@regionh.dk.