PubMed, Embase, Web of Science and Cochrane Library databases were searched between 1st January 1990 and 1st March 2023. We developed search strategies for each database according to the principle of PICOS (Population, Intervention, Comparison, Outcome, Study design), and the references of the identified studies were checked for potential eligibility (Supplementary Table S1). Specifically, the search strategy for PubMed was: (alpha blocker OR alpha blockade OR alpha antagonist OR prazosin OR phenoxybenzamine OR doxazosin OR terazosin OR alfuzosin OR silodosin OR tamsulosin) AND (urinary retention OR voiding difficulty). The keywords regarding orthopaedic procedures were deliberately omitted during the initial search to avoid the exclusion of potentially relevant studies, given the diversity of orthopaedic procedures.

Eligible studies were identified using the following inclusion criteria: 1) patients following orthopaedic surgery as the interested population; 2) alpha blocker as the intervention; 3) placebo or no treatment as the comparison; 4) incidence of POUR as the primary outcome; 5) with comparative study design.

We excluded non-English language reports, in vitro studies, case reports, brief reports, conference abstracts/posters or reviews. After removing duplicates, titles and abstracts were reviewed to identify eligible papers. Full texts were assessed to determine the final list of publications eligible for inclusion in the study.

After the final list of included studies was set, we extracted the following information: year of publication, patient age, definition of POUR, operative information, administration of drugs and study design. The primary outcome of interest was the incidence of POUR. Complications were extracted as secondary outcomes.

The quality of the included studies was assessed independently by two authors. The modified Jadad Scale was employed for RCTs (Moher et al., 1996), and the Newcastle-Ottawa Scale for cohort studies (Stang, 2010). The publication bias was estimated using the funnel plot and Harbord’s test (Harbord et al., 2006). In case of disagreement, a third senior doctor was consulted.

Statistical analysis was performed using RevMan software (version 5.3, Cochrane Collaboration, Oxford, the United Kingdom), R software (version 4.1.3, R Foundation for Statistical Computing, Vienna, Austria) and Trial Sequential Analysis (TSA) software (version 0.9.5.10 beta, Copenhagen Trial Unit, Copenhagen, Denmark), with p < 0.05 as the threshold for statistical significance. Heterogeneity among the studies was evaluated through the I² statistic and Q test. Generally, a fixed-effects model is employed to calculate pooled results if I² is less than 50% and the p-value for the Q test exceeds 0.05. Conversely, a random-effects is employed if I² exceeds 50% or the p-value for the Q test is below 0.05. In this meta-analysis, we preferred the random-effects model to the fixed-effects model due to the inclusion of various alpha blockers and diverse dosing regimens. We postulated that the effectiveness of these interventions might exhibit variability across studies. Thus, a random-effects model was prespecified (Borenstein et al., 2010; Nikolakopoulou et al., 2014). The pooled rate of POUR or complications was calculated using the Mantel-Haenszel (M-H) method with a 95% confidence interval (CI). To ensure the robustness of the primary outcome of POUR, we evaluated the risk of false positives or false negatives using a post hoc TSA. Sequential boundaries were set according to a type I error of 5% and a power of 80%. The required information size (RIS) was calculated to determine whether the sample size was adequate to reach a reliable conclusion. The cumulative curve of the Z score was plotted. Sensitivity analysis was conducted using the leave-one-out analysis. Subgroup analysis was performed stratified by the type of operation.

A total of 3,587 studies were identified at the initial search. After removing duplicates, 1,544 records were screened by titles and abstracts. Of these, 36 papers were assessed for eligibility by reading the full texts. Finally, seven studies (six RCTs and one retrospective cohort study) were included in the analysis (Figure 1) (Petersen et al., 1991; Basheer et al., 2017; Schubert et al., 2019; Schubert et al., 2020; Choi et al., 2021; Ding et al., 2022; Rughani et al., 2022). Among these studies, five focused on patients undergoing arthroplasty, while two investigated cases undergoing spine surgery. Six out of the seven studies reported the proportion of patients with a history of benign prostatic hyperplasia or urinary retention, with the cohort study exhibiting the highest proportion in this regard. Tamsulosin was the most commonly used drug. The details of the included studies were summarized in Table 1.

All RCTs had a Jadad score higher than four, indicating that they were of high quality. The Newcastle-Ottawa rank also revealed the high quality of the cohort study. The details of the assessment were summarized in Table 2. The funnel plot did not show the concerns of possible publication bias (Figure 2), which was also consistent with the formal test (Harbord’s test, p = 0.145).

A total of 748 patients (Male: 95.6% VS. Female: 4.4%) received alpha blockers (Intervention group), while 859 patients (Male: 96.2% VS. Female: 3.8%) received a placebo or no intervention (Control group). Overall, there were 294 episodes of POUR in 1,607 patients (18.3%). The pooled analysis showed no significant difference in the incidence of POUR between the two groups with moderate heterogeneity (Intervention group: 126/748 [16.8%] VS. Control group: 168/859 [19.6%]; RR = 0.75; 95% CI 0.51 to 1.09; p = 0.130; Heterogeneity: I² = 67.1%; p = 0.006) (Table 3; Figure 3). The sensitivity analysis using the leave-one-out analysis revealed that the exclusion of the study by Schubert et al. (2020) (Schubert et al., 2020), representing the only retrospective cohort study in this meta-analysis, demonstrated a statistically significant impact on the pooled results (Figure 4).

Considering the level of evidence of each study, we subsequently analyzed the data only from RCTs. This analysis reflected that the pooled incidence of POUR was lower in the Intervention group than that in the Control group (Intervention group: 84/519 [16.2%] VS. Control group: 126/529 [23.8%]; RR = 0.67; 95% CI 0.47 to 0.95; p = 0.03; Heterogeneity: I² = 50%; p = 0.08) (Figure 5). However, the post hoc TSA showed that the cumulative curve of the Z score did not cross the sequential boundary nor the RIS, suggesting that the total sample size of the existing evidence may be insufficient to draw conclusive results (Figure 6).

The included studies were stratified into two subgroups: the Arthroplasty subgroup and the Spine surgery subgroup. In the Arthroplasty subgroup, the incidence of POUR was 19.2% in the Intervention group and 23.2% in the Control group, respectively. Two studies conducted a comparative assessment of knee and hip arthroplasty. In the study by Petersen et al., the incidence of POUR was comparable (Knee arthroplasty: 41% VS. Hip arthroplasty: 48%) (Petersen et al., 1991). Choi et al. reported that hip arthroplasty was not a risk factor for POUR as compared with knee arthroplasty (OR = 0.821, p = 0.711) (Choi et al., 2021).

In the Spine surgery subgroup, POUR was documented at a rate of 13.3% in the Intervention group and 12.8% in the Control group, respectively. The main surgical level was the lumbar region. The relationship between the incidence of POUR and the surgical levels remained unexplored in the two studies (Basheer et al., 2017; Rughani et al., 2022).

Prophylactic administration of alpha blockers did not show a significant protective effect in either subgroup (Arthroplasty subgroup: RR = 0.64, 95% CI 0.39 to 1.06; Spine surgery subgroup: RR = 1.03, 95% CI 0.69–1.55). Little heterogeneity was observed between the two subgroups (p = 0.15) (Figure 7).

Five studies reported the data of complications and the pooled analysis showed that the administration of alpha blockers would introduce a higher risk of complications (88/651 [13.5%] VS. 56/766 [7.3%]; RR = 1.73; 95% CI 1.27 to 2.37; p = 0.0005; Heterogeneity: I² = 0%; p = 0.69) (Figure 8) (Petersen et al., 1991; Schubert et al., 2019; Schubert et al., 2020; Ding et al., 2022; Rughani et al., 2022). UTI was the most commonly described and concerned complication. The rate of UTI was similar between the two groups (23/651 [3.5%] VS. 18/766 [2.3%]; RR = 1.65; 95% CI 0.92 to 2.94; p = 0.09; Heterogeneity: I² = 0; p = 0.55) (Figure 9).

Our research holds clinical importance as it seeks to investigate the efficacy of pharmacologic interventions against POUR in orthopaedic patients. However, our findings showed a lack of statistically significant difference in the incidence of POUR when comparing cases with or without the administration of alpha blockers. These results differed from previous studies and meta-analyses in other clinical disciplines. Ghuman et al. conducted a meta-analysis of 15 RCTs in general surgery and urologic surgery and found that the prophylactic use of alpha blockers could yield a 17.1% absolute reduction in POUR risk (Ghuman et al., 2018). Another meta-analysis encompassing five RCTs revealed a reduction of 20.6% in the risk of POUR among patients after hernia repair following the administration of alpha blockers (Clancy et al., 2018). The discrepancy between our results and prior findings may be attributed to multiple factors. First, orthopaedic patients have an elevated inherent susceptibility to POUR owing to multiple factors, such as advanced age demographics, frequent use of neuraxial anaesthesia, and routine employment of indwelling urinary catheters (Schubert et al., 2019; Leitner et al., 2021). As a result, the perceived effectiveness of alpha blockers might not manifest to the same extent as witnessed in other disciplines. Second, orthopaedic procedures are typically major surgeries (Hoogeboom et al., 2014; Dietz et al., 2019), which pose a challenge for cautious perioperative fluid management (Heming et al., 2020). Arthroplasty and spine surgery often entails substantial intraoperative fluid administration (exceeding 1,000 mL), which is also linked to a potential risk of POUR (Lee et al., 2017; Aiyer et al., 2018; Halawi et al., 2019; Ongaigui et al., 2020). Third, our post hoc TSA to evaluate the robustness of our findings provided a more conservative estimation. It could be inferred that our aggregated results of RCTs might be false positive due to a relatively small sample size. Thus, more high-level evidence is required to determine the true efficacy of alpha blockers against POUR in orthopaedic patients.

Only two kinds of orthopaedic procedures, arthroplasty and spine surgery, were included in our review despite a thorough investigation across four main databases. In terms of lower-limb arthroplasty, our subgroup data indicated that the efficacy of alpha blockers seemed to be consistent between knee and hip procedures. However, it should be noted that the incidence of POUR might not be close between these two procedures. Abdul-Muhsin et al. reported that knee arthroplasty had a higher prevalence of POUR than hip arthroplasty (Abdul-Muhsin et al., 2020). As for the spine surgery, the majority of cases in our review pertained to lumbar-level interventions, which was reported to be associated with an elevated susceptibility to POUR as compared with surgeries targeting the cervical or thoracic regions (Altschul et al., 2017). Due to the limited available data, our review was unable to explore potential correlations between surgical levels and drug efficacy.

Information regarding the use of alpha blockers to prevent POUR in other orthopaedic surgeries is lacking. Several studies have identified the risk factors and preventive regimens for patients following different orthopaedic surgeries. Tobu et al. found that patients with femoral neck fractures who had dementia and/or delirium had a tenfold higher risk of developing POUR (Tobu et al., 2014). Higashikawa et al. also found that cognitive impairment in female patients with proximal hip fractures was associated with a higher likelihood of POUR (Higashikawa et al., 2019). The authors recommended necessary nursing care in activities of daily living and neurofunctional assistance to mitigate the risk. A history of benign prostatic hyperplasia or neurogenic bladder is also a well-established risk factor. A meta-analysis showed that patients with such a condition were associated with a 3.8-fold risk of developing POUR after spine surgery (Chang et al., 2021). In our meta-analysis, a strategic exclusion of patients with a history of benign prostatic hyperplasia was instituted across most RCTs to mitigate potential bias. However, this discerning protocol also attenuated the generalizability of the pooled results. Further studies are appealed to explore the strategy against POUR in patients at higher risk, such as those with benign prostatic hyperplasia or neurogenic bladder. Other risk factors including spinal/epidural anaesthesia, advanced age and excessive fluid administration should also be considered in orthopaedic surgeries (Cha et al., 2020). Santini et al. adopted the International Prostate Symptom Score to assess the risk of POUR and found that a score above 18 was strongly related to a predisposition to POUR (Santini et al., 2019). However, to our knowledge, few scoring systems for POUR in orthopaedic surgeries have been widely used with a well-validated prediction (Bracey et al., 2022).

Our research found that the administration of alpha blockers was related to a higher risk of complications while the incidence of UTI was similar between the two groups. Other reported complications included dizziness and vomiting, which might result from vasodilation caused by the drugs. Previous studies have also raised concerns about alpha blockers, such as the carcinogenicity of phenoxybenzamine and the cardiovascular effects of prazosin (Hasford et al., 1991; Suarez-Torres et al., 2021). In our included studies, Schubert et al. reported two cases of tamsulosin-induced floppy iris syndrome (Schubert et al., 2019). Bell et al. found that the risk of floppy iris syndrome in patients after cataract surgery was 2.3 times higher when tamsulosin was administered (Bell et al., 2009), and the authors attributed this complication to the mechanism that the alpha-1A adrenergic receptors were also present in the dilator smooth muscle of the iris, and thus tamsulosin might compromise mydriasis. Surgeons should pay attention to our assembled data and inform patients of the associated complications.

We noted several limitations in our meta-analysis. First, the possibility of missing relevant studies cannot be completely avoided, which may introduce bias into our methodology. Second, one of the included studies utilized a cohort study design, which may affect the level of evidence of our pooled data. However, we performed a leave-one-out analysis and calculated the pooled results after excluding the cohort study. We used the TSA to estimate the robustness of our conclusion, which suggested that our results of RCTs might not possess full sufficiency and conclusiveness. Third, we included all the studies with multiple alpha blockers as an intervention and assumed that the efficacy of alpha blockers was similar. We used the random-effects model to synthesize the data more conservatively. However, we still noticed that the pooled results reflected a moderate degree of heterogeneity, despite performing a subgroup analysis. Fourth, the data on POUR that were pooled were extracted directly from the articles and we found that the definitions of POUR in each study were similar but not identical, which might bring quantitative bias into the data. Fifth, only two types of orthopaedic surgeries were included, which might jeopardize the generalizability of the conclusion. The distribution of surgical subtypes between the intervention and control groups might also introduce bias.