Scientific Papers

The role of endometrial scratching prior to in vitro fertilization: an updated systematic review and meta-analysis | Reproductive Biology and Endocrinology


The literature search yielded 879 potentially relevant reports (Fig. 1). Subsequently, the titles of these manuscripts were examined, resulting in 222 potentially eligible publications. The abstracts of these studies were then examined and eventually 96 manuscripts that could provide data to answer the research question were identified. The full text of these studies was examined thoroughly, resulting in the inclusion of 40 publications, that represent 39 RCTs [5, 8, 11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45, 47, 48] (one report [46] contained post-hoc analyses of a previously published RCT [41] (Table 1). It should also be noted that Liu et al., [29] included four groups in their study (intervention and no intervention during the follicular and the luteal phase of the cycle preceding IVF) and therefore, we analyzed the follicular and the luteal phase arms of the study separately. Characteristics of the reports included in the systematic review appear in Tables 1 and 2. Eligible studies were published between 2008 and 2022. Randomization method was reported in 34 of the publications included, while allocation concealment method was reported in 19 of the studies included (Table 1). Most studies did not state clearly if the participants or those involved in the analysis were blinded to the type of intervention. Only 3 studies were reported to be single-blind and 3 were reported to be double-blind. Financial support was declared in 20 studies (Table 1). The largest study published so far on this issue was by Lensen et al. in 2019 [5]. The risk of bias assessment of the eligible studies is presented in Table 3. Overall, 9 studies [11,12,13, 15, 16, 23, 27, 28, 37] were deemed to be at high risk of bias (Supplementary Figures 1 & 2).

Fig. 1
figure 1
Table 3 Risk of Bias assessment of included studies (using RoB-2)

Meta-analysis

Live birth

A significantly higher probability of live birth was present in embryo transfer cycles after endometrial scratching as compared to placebo/sham or no intervention (risk ratio-RR: 1.12, 95% CI: 1.05– 1.20; fixed effects model; heterogeneity: I2=46.30%, 28 studies, 29 datasets, 7425 patients; low certainty; NNT: 30) (Fig. 2). Publication bias did not seem to be present (p=0.727). A sensitivity analysis excluding studies at high risk of bias [13, 15, 16, 23, 27, 37] did not materially change the results obtained (RR: 1.13, 95% CI: 1.05-1.21; fixed effects model; heterogeneity: I2=29.87%, 22 studies, 23 datasets; moderate certainty; NNT: 28) (Supplementary Figure 3).

Fig. 2
figure 2

Forest plot presenting the risk ratio of live birth between women who had endometrial scratching prior to their embryo transfer and those who had a placebo/sham procedure or no intervention

Ongoing pregnancy

A higher, but not significantly so, probability of ongoing pregnancy was present in embryo transfer cycles after endometrial scratching as compared to placebo/sham or no intervention (RR: 1.07, 95% CI: 0.98– 1.18; fixed effects model; heterogeneity: I2=27.44%, 11 studies, 11 datasets, 4515 patient; low certainty) (Fig. 3). Publication bias did not seem to be present (p=0.494). A sensitivity analysis excluding studies at high risk of bias did not materially change the results obtained (RR: 1.07, 95% CI: 0.97-1.18; fixed effects model; heterogeneity: I2=0.00%, 8 studies, 8 datasets; moderate certainty) (Supplementary Figure 4).

Fig. 3
figure 3

Forest plot presenting the risk ratio of ongoing pregnancy between women who had endometrial scratching prior to their embryo transfer and those who had a placebo/sham procedure or no intervention

Clinical pregnancy

A significantly higher probability of clinical pregnancy was present in embryo transfer cycles after endometrial scratching as compared to placebo/sham or no intervention (RR: 1.12, 95% CI: 1.06– 1.18; fixed effects model; heterogeneity: I2=47.48%, 37 studies, 38 datasets, 8804 patients; low certainty; NNT: 27) (Fig. 4). Publication bias did not seem to be present (p=0.514). A sensitivity analysis excluding studies at high risk of bias did not materially change the results obtained (RR: 1.12, 95% CI: 1.05-1.19; fixed effects model; heterogeneity: I2=21.88%, 21 studies, 22 datasets; moderate certainty; NNT: 25) (Supplementary Figure 5).

Fig. 4
figure 4

Forest plot presenting the risk ratio of clinical pregnancy between women who had endometrial scratching prior to their embryo transfer and those who had a placebo/sham procedure or no intervention

Cumulative live birth

A higher, but not significantly so, probability of cumulative live birth was present in embryo transfer cycles after endometrial scratching as compared to placebo/sham or no intervention (RR: 1.11, 95% CI: 0.99–1.24; fixed effects model; heterogeneity: I2=0%, 2 studies, 1298 patients; very low certainty) (Supplementary Figure 6). Publication bias could not be assessed due to the small number of available studies.

Miscarriage

No significant difference in the probability of miscarriage was present in embryo transfer cycles after endometrial scratching as compared to placebo/sham or no intervention (RR: 0.89, 95% CI: 0.75–1.06; fixed effects model; heterogeneity: I2=0%, 24 studies, 25 datasets, 2568 patients; low certainty) (Supplementary Figure 7). Publication bias did not seem to be present (p=0.432).

Ectopic pregnancy

No significant difference in the probability of ectopic pregnancy was present in embryo transfer cycles after endometrial scratching as compared to placebo/sham or no intervention (RR: 1.02, 95% CI: 0.46– 2.27; fixed effects model; heterogeneity: I2=0%, 8 studies, 9 datasets, 1219 patients; very low certainty) (Supplementary Figure 8). Publication bias did not seem to be present (p=0.148).

Multiple pregnancy

No significant difference in the probability of multiple pregnancy was present in embryo transfer cycles after endometrial scratching as compared to placebo/sham or no intervention (RR: 1.11, 95% CI: 0.92–1.35; fixed effects model; heterogeneity: I2=25.68%, 17 studies, 18 datasets, 1974 patients; low certainty) (Supplementary Figure 9). Publication bias did not seem to be present (p=0.482).

Adverse events

Pain

Five studies [4, 8, 19, 36, 44] reported pain in the endometrial scratching group with VAS scores ranging from 3.5 to 6.4. Only one study (158 patients) provided VAS scores both in the endometrial scratching group and the control group (sham procedure) indicating higher pain scores (6.42, SD (2.35) vs 1.82, SD (1.52); P < 0.001) in women who had the endometrial scratching [19].

Bleeding

In patients allocated to endometrial scratching, bleeding was reported in a proportion of them in four studies [5, 8, 33, 42], while in further 8 studies [13, 19, 23, 29, 38, 39, 41, 43] no patients experienced bleeding after endometrial scratching The remaining studies did not report on this adverse event.

Infection

In patients allocated to endometrial scratching no infections were observed in 11 studies [5, 8, 13, 19, 23, 29, 38, 39, 41,42,43], while the remaining studies did not report on this adverse event.

Dizziness

In patients allocated to endometrial scratching, dizziness was not observed in 10 studies [8, 13, 19, 23, 29, 38, 39, 41,42,43] while in a single study [5] 7 out of 690 patients (~1%) who underwent endometrial scratching experienced this adverse event.

Fever

In patients allocated to endometrial scratching, fever was not observed in 10 studies [5, 13, 19, 23, 29, 39, 41,42,43] while in a single study [8] 3 out of 742 patients (0.6%) who underwent endometrial scratching experienced this adverse event.

Subgroup analyses

Type of instrument used to perform the endometrial injury

Pipelle-type catheters were used for endometrial scratching in 29 trials, while Novak curette was the tool of choice in 3 trials. A variety of other instruments were used for endometrial injury in the remaining studies (Table 2). The type of instrument used to perform endometrial scratching did not appear to be associated with the effect size observed (test for subgroup differences: p=0.13).

Timing of the endometrial injury

Endometrial scratching was performed during the cycle preceding IVF treatment in 33 RCTs (Table 2). In a single study, endometrial scratching was performed from day 3 of the cycle preceding embryo transfer until day 3 of the treatment cycle [5]. In 3 of the eligible RCTs, endometrial scratching was performed during the follicular phase of the cycle, while in further 3 RCTs it was performed on the day of oocyte retrieval (Table 2).

A subgroup analysis based on the time endometrial scratching was performed (in the preceding cycle, in the actual embryo transfer cycle or in either of the two) suggested significant difference between the subgroups (p=0.04) (Supplementary Figure 10). Studies in which the endometrial scratching was performed during the preceding cycle showed a pooled RR: 1.18 (95% CI:1.09-1.27; moderate certainty; NNT: 21), whereas studies in which the endometrial scratching was performed during the embryo transfer cycle showed a pooled RR: 0.87 (95% CI: 0.67-1.15; low certainty).

Single of double endometrial injury

Single or double endometrial scratching was performed in 34 and 5 of the eligible RCTs, respectively (Table 2). A subgroup analysis between studies with single and those with double endometrial scratching did not suggest a significant difference in the probability of live birth (p=0.27).

History of previous failed IVF cycles

A subgroup analysis according to whether the population evaluated in each study had experienced previous IVF failures or not suggested a significant difference between subgroups (p<0.001). The highest effect size was observed in studies which randomized patients with previous IVF failures (RR: 1.35, 95% CI: 1.20-1.53, fixed effects model, heterogeneity: I2=0.06%, 13 studies, 13 datasets, 2627 patients; moderate certainty; NNT: 14) (Supplementary Figure 11).

A further subgroup analysis according to the minimum number of previous IVF failures (0,1,2 and 3) also confirmed a significant difference between subgroups (p=0.04), with the largest effect size observed in studies that included patients with at least 3 failed IVF cycles (RR: 1.70, 95% CI: 1.14-2.54, fixed effects model; heterogeneity: I2=49.75%, 3 studies, 547 patients; low certainty; NNT: 12) (Supplementary Figure 12). Finally, a meta-regression performed using the minimum number of previous failed as an independent variable, suggested a positive significant association with the risk ratio of live birth in the included studies (coeff: 0.18, 9% CI: 0.06-0.31; p=0.004).



Source link