Following the PRISMA statement, a systematic literature search of PubMed, Embase, Cochrane Library, Web of Science, CINAHL, CNKI and the Wanfang and VIP databases was performed. The search period was from inception to July 30, 2022. The search terms included “lumbar disc herniation or recurrent lumbar disc herniation”, “percutaneous transforaminal endoscopic discectomy or percutaneous endoscopic lumbar discectomy” and “risk factor”. In addition, we manually retrieved the target literature by reading the relevant systematic reviews and references of the included studies. This meta-analysis did not register online, and the protocol was not prepared.

Inclusion criteria: (1) Chinese and English studies published in peer-reviewed journals; (2) the subjects were diagnosed with lumbar disc herniation based on MRI, and the recurrence of lumbar disc herniation after transforaminal endoscopy was not limited to age, gender and duration of illness; (3) information on risk factors for relapsed populations was reported in the literature; (4) case-control studies and cohort studies. Exclusion criteria: (1) non-population study; (2) conference articles, case reports, systematic reviews and other research types; (3) outcome information was insufficient, and data analysis could not be performed; (4) literature research repeated reports; (5) the researchers could not obtain the complete article research.

Two reviewers independently reviewed each article's abstracts and full text according to the inclusion of exclusion criteria. For disagreements between the two reviewers, a third reviewer was recruited for discussion until consensus was achieved. After literature screening, two reviewers independently respectively extracted the following information: documentary information, the demographic characteristics of research objects and the influencing factors of recurrent lumbar disc herniation.

The Newcastle–Ottawa Scale (NOS) was used to evaluate the quality of observational studies. The scale was evaluated from eight aspects: the representativeness of the study population, the comparability between groups, the adequacy of the evaluation of the results, the adequacy of the follow-up time and the integrity of the follow-up. The total possible score was nine points; seven points and above indicated high-quality literature, and five points and below indicated low-quality literature.

Revman 5.3 was used to evaluate the quality of diagnostic tests and draw the risk map of bias. The effect size of the count data was expressed by the odds ratio (OR), and the 95% confidence interval (CI) was used to estimate the interval range of the effect size. The heterogeneity test was used to determine the size of heterogeneity by the test of I² or Q. If I² < 50% or p > 0.1, the included literature was considered homogeneous; if I² > 50% or p ≤ 0.1, the included studies were considered largely heterogeneous, and a subgroup meta-analysis was used to explore the source of heterogeneity further. P < 0.05 indicated that the difference was statistically significant.

After a systematic retrieval and screening of the target literature, 13 studies were included in this study (10–22); the literature screening flow chart is shown in Figure 1. The 13 studies were published from 2007 to 2022, involving 3,393 patients with lumbar disc herniation treated by transforaminal endoscopy, of which 323 were recurrent lumbar disc herniation, and 3,070 patients had no recurrence after treatment. The average age of the patients included in the study was 39.32–63.7 years old, mainly male. The NOS score of the literature quality was 6–9 (mean: 7.5; median: 8.0). Detailed basic characteristics of the included studies and literature quality evaluation results are shown in Table 1.

Eight, ten and seven studies reported the results of age, gender and BMI on the recurrence of lumbar disc herniation after percutaneous transforaminal endoscopic treatment, respectively. Regarding age, 62 patients with recurrence were <60 years old, and 79 patients were ≥60 years old. Patients <60 years old had a lower risk of recurrence after treatment (OR = 0.44; 95% CI: 0.23, 0.87). Age ≥60 years was a risk factor for recurrence; the risk of recurrence after treatment in this population was 2.23-fold (95% CI: 1.13, 4.41), as shown in Figure 2. In terms of gender, there were no significant differences in recurrence between males (OR = 1.01; 95% CI: 0.75, 1.35) and females (OR = 0.97; 95% CI: 0.72, 1.30). Patients with a higher BMI had a higher risk of relapse after treatment, and the risk of recurrent lumbar disc herniation was 2.89 (95% CI: 1.23, 6.80) for BMI ≥25 and 0.41 (95% CI: 0.18, 0.95) for BMI <25, as shown in Figure 3.

The effects of smoking vs. alcohol consumption on recurrent lumbar disc herniation were reported in six and five studies, respectively. According to the heterogeneity evaluation results (I² = 54%, I² = 0%), the random-effect and fixed-effect models were used to analyse the effect of smoking and drinking, respectively. Meta-analysis showed that there were no statistically significant effects of smoking (OR = 1.41; 95% CI: 0.81, 2.44) and alcohol consumption (OR = 1.43; 95% CI: 0.99, 2.05) on the occurrence of recurrent lumbar disc herniation in patients. However, there was an increased risk of recurrence between smoking and alcohol consumption.

All seven studies reported the effect of hypertension vs. diabetes on recurrent lumbar disc herniation. Heterogeneity evaluation showed good homogeneity among the included studies (I² = 0%, I² = 19%), and a fixed-effects model was used for systematic evaluation. A systematic review showed that hypertension (OR = 1.05; 95% CI: 0.70, 1.58; P = 0.80) might increase the risk of recurrence in patients, but the difference was not statistically significant, as shown in Figure 4A. Diabetes mellitus significantly increased the risk of recurrent lumbar disc herniation (OR = 1.73; 95% CI: 1.18, 2.55) compared with that of the non-recurrent population, as shown in Figure 4B.

The effects of the disease duration, type of herniation and Pfirrmann grade on recurrent lumbar disc herniation were reported in three, six and seven studies, respectively. Meta-analysis suggested that if the disease duration was ≥4 years, the OR of recurrent lumbar disc herniation after treatment was 2.93 (95% CI: 1.58, 5.43), as shown in Figure 5. There was no statistically significant effect of protrusion, prolapse or mobilisation on recurrence. Patients with Pfirrmann grades 3–4 had a much higher risk of recurrence after treatment than those without recurrence, with an OR of 3.10 (95% CI: 2.18, 4.40), as shown in Figure 6.

Eight, three and three studies reported results on the effect of the surgical level of the lumbar disc, incomplete removal of the nucleus pulposus and intraoperative fibrous damage on recurrent lumbar disc herniation in patients, respectively. The heterogeneity evaluation results showed some heterogeneity among the included studies, and the pooled effect size was calculated using a random-effects model. The results of the systematic review showed that there were no statistically significant effects of surgical levels L3–4 (OR = 0.85; 95% CI: 0.34, 2.13), L4–5 (OR = 1.52; 95% CI: 0.76, 3.04) and L5–S1 (OR = 0.72; 95% CI: 0.39, 1.33) on the occurrence of recurrent lumbar disc herniation in patients. In addition, incomplete nucleus pulposus removal (OR = 3.26; 95% CI: 1.69, 6.27) and intraoperative fibrous damage (OR = 3.18; 95% CI: 1.56, 6.50) were associated with an increased risk of post-treatment recurrence, as shown in Figure 7.