The present study was registered on PROSPERO (registration # CRD42022358981), an international database of prospectively registered systematic reviews (National Institute for Health Research) (14). The systematic review was drafted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (15).

Using a stated search strategy (Table 1), two investigators (VC, GLa) individually screened the main databases (PubMed/Medline, Scopus, Web of Science, and Cochrane) with combined keywords. MeSH headings and terms used were “Pyeloplasty” AND “Infants” (Supplementary material S1). Studies published from 1984 to August 2022 in English language were included. The list of references was screened as well to detect possible pertinent cross references. Case reports, opinion articles, and reviews were excluded. All comparative studies reporting the outcomes of OP compared to those of MIS (i.e., LAP and/or RALP) to treat UPJO in infants or patients <15 kg were included. The full text of theoretically suitable papers was retrieved and individually assessed for eligibility by the same two authors. Any divergence over the entitlement of papers was solved through a further debate with a third author (GLi).

The studies comparing OP vs. MIS for the Anderson–Hynes pyeloplasty in infants were included in the meta-analysis. The exclusion criteria are as follows: •

treatment other than Anderson–Hynes pyeloplasty, retroperitoneoscopy, OTAP, recurrent UPJO, secondary UPJO, patients >1 year; and

Categorical variable rates were compared with Pearson's χ² test or two-tailed Fisher’s exact probability test. When median and range were reported, mean ± SD were valued (16).

The meta-analysis was managed with RevMan 5.4 (17). The random effects model was selected. The risk ratio (RR) was assessed for categorical variables. Differently, mean differences (MD) were preferred in the case of continuous variables. Both results were reported with 95% confidence intervals (CIs). Data were expressed as mean ± SD. I² values were used to judge homogeneity and quantify the dispersion of effect sizes. Biases among the papers included were evaluated with the funnel plot. Quantitative and demographic data were compared using Fisher's exact test and expressed as number, percentage, or mean ± SD using the RR and 95% CI. A p < 0.05 was considered significant.

Two authors (DDR and MEM) assessed the risk of bias for individual studies. This assessment was achieved with a methodological index for nonrandomized studies (MINORS) (18). Dissimilarities between the two authors (DDR and MEM) were solved through a discussion with a third author (GLi). The score for this index ranges between 0 and 24 points. The “gold standard” cutoff was 19.8 points. With regard to the quality of each outcome, we graded the quality of evidence, thanks to the Grading of Recommendations Assessment, Development and Evaluation (GRADE) methodology (19). The quality of evidence was graded as high, moderate, low, and very low in all results. Observational studies were assessed as low quality of evidence. The quality of evidence was further reduced in the case of risk of bias, inconsistency, indirectness imprecision, and publication bias. MINORS was adopted to judge the risk of bias in observational papers. Inconsistency was determined according to heterogeneity, and I² value was used to evaluate heterogeneity. As established in Cochrane guidelines, heterogeneity was assessed as low, moderate, substantial, and considerable when I² values were 0–40, 30–60, 50–90, and 75%–100%, respectively (20). If a score overlapped two groups, we inserted a mixed inconsistency (e.g., low/moderate) in our GRADE table. Finally, imprecision was evaluated with optimal information size (OIS) based on 25% relative risk reduction, 0.05 a-error, and 0.20 b-error (21).

The initial review retrieved 811 studies from databases using keywords “Pyeloplasty” AND “Infants.” Thanks to the screening of all these titles and abstracts, we selected 70 papers focusing on infants or children <15 kg of weight with UPJO. Among these, only 13 publications were comparative studies between OP vs. MIS (1, 13, 22–32) (Figure 1).

Six studies (1, 22, 24, 25, 27, 31) described OP vs. LAP, and one was excluded (31) due to incomplete data. Three papers treating OP vs. RALP (26, 30, 32) were included. Two studies compared OP vs. LAP + RALP (13, 28); one was excluded (28) due to incomplete information. Two papers (23, 29) treating RALP vs. LAP were excluded.

Nine articles were included in the meta-analysis, eight retrospective studies (1, 22, 24–27, 30, 32) and one prospective study (13).

The total number of pyeloplasties performed in infants and patients <15 kg of weight was 3,145, with 2,859 (90.9%) OP and 286 (9.1%) MIS, comprising 145 (50.7%) LAP and 141 (49.3%) RALP. The M/F ratio was 3:1, with 2,301M (73.2%) and 844F (26.8%). Of 2,859 OP patients, 2,092 were males (73.2%) and 767 were females (26.8%); of 286 in the MIS group, 209 were males (73.1%) and 77 were females (26.9%), with no differences between OP and MIS groups [p = ns, RR: 1.001, 95% CI: (0.93–1.08), Table 2].

The side of the kidney affected by UPJO has been reported in seven papers (1, 22, 24–26, 30, 32), with 462 patients (279 OP and 183 MIS). Of 279 OP patients, 162 had left kidneys affected (58.1%) and 117 had right kidneys affected (41.9%). Of 183 in the MIS group, 101 had left kidneys affected (55.2%) and 82 had right renal units affected (45%), with no differences between the two groups [p = ns, RR: 1.052, 95% CI (0.89–1.24), Table 2].

The mean age at procedure has been reported in nine papers (1, 13, 22, 24–27, 30, 32), with no difference between OP infants (4.9 ± 1.4 months) and MIS patients [5.8 ± 2.2 months; p = ns, MD −0.9, 95% CI: (−2.21–0.22), Table 2 and Figure 2].

The weight at surgery has been reported in six of the included papers (1, 22, 24, 26, 30, 32). Weight was not significantly different among the two groups: 6.4 ± 1.4 kg in OP vs. 6.9 ± 1.4 in MIS [p = ns, MD −0.71, 95% CI (−1.47–0.06), Table 2 and Figure 3].

Operative time (OT) has been reported in seven studies (1, 24, 26, 32, 25, 13, 32). OT was significantly lower in OP than that in MIS [129.4 ± 24.1 vs. 144.0 ± 32.3 min, respectively; p = 0.0004, MD: −18.19, 95% CI: (−28.35, −8.04), Figure 4].

Nine papers (1, 13, 22, 24–27, 30, 32) have shown an increased LOS in OP compared to that in MIS [3.2 ± 1.9 vs. 2.2 ± 0.9 days, respectively; p = 0.01, MD: 0.76, 95% CI: (0.16–1.36), Figure 5].

Eight studies have been reported on postoperative complications (1, 13, 22, 24–26, 30, 32), such as urinary leakage, urinary infection, and bleeding. The incidence of complications was comparable between the two groups: 10.0 ± 12.9% in OP (32/319 patients) vs. 10.9 ± 11.6% in MIS procedures [27/248 patients; p = ns, RR: 0.95, 95% CI (0.45–2.01), Figure 6].

The length of postoperative follow-up has been reported in five papers (13, 25, 26, 30, 32). OP presented a longer but not significant follow-up than MIS [21.5 ± 8.1 months vs. 13.9 ± 4.7 months, respectively; p = ns; MD: 7.03, 95% CI: (−2.23–16.29), Table 2 and Figure 7].

The failure rate has been mentioned in seven papers (1, 13, 22, 24–26, 30). In most of the studies included, success has been defined as resolutions of symptoms and improved ultrasonographic or renographic parameters at the follow-up. The failure rate was not different in OP (5.2 ± 3.5%, 16/308 patients) and MIS [4.2 ± 3.3%, 10/238 patients; p = ns; RR: 1.28, 95% CI: (0.58–2.82), Figure 8].

Only the paper by Andolfi et al. (13) further compared LAP vs. RALP. No conversion was reported in both groups. Furthermore, the incidence of complications and the success rates were similar for both procedures.

We further screened these nine papers included in the meta-analysis with regard to the outcomes between left surgery and right surgery. However, no data were reported on operative time, the length of hospital stay, the incidence of complications, and the failure rate regarding the side of the procedure.

There are several limitations of the present study. As reported above, all but one studies were retrospective, which may lead to select bias. None of the papers provided sample size calculations. As expected, a blinded evaluation of objective endpoints was not possible. Moreover, the outcomes of MIS were strictly dependent on procedural volume. High-volume centers presented perioperative outcomes that were equivalent to or better than those of OP, different from low-volume centers (28). Furthermore, none of the studies have reported with regard to the loss of follow-up. As a consequence, in our meta-analysis, none of the studies reached the gold standard cutoff on MINORS of 19.8 out of 24 (Table 3).

According to the GRADE methodology, the quality of evidence of the meta-analysis was low regarding the length of hospital stay, the incidence of postoperative complications, and the failure of the surgical procedure (Table 4). Both the reduced number of MIS infants and the considerable heterogeneity of the data could generate possible bias.

However, when assessed in duplicate by two authors (DDR and MEM) using A Measurement Tool to Assess Systematic Reviews (AMSTAR) (43), the present study received an honest score (Supplementary material S2).

The PRISMA checklist was finally fulfilled (Supplementary material S3).