Based on the search strategy developed, we obtained an initial total of 256 studies. The two researchers each perused the titles, abstracts, and full texts of all articles retrieved, and a total of 57 articles passed the initial screening. These 57 studies were again evaluated by the two researchers according to the inclusion and exclusion criteria developed for this Meta-analysis. Finally,12 RCTs were included in the meta-analysis (Fig 1) (Cassinelli et al., 2008; Jirarattanaphochai et al., 2008; Karst et al., 2003; Kelsaka et al., 2014; Kien et al., 2019; Kim et al., 2016; Pookarnjanamorakot et al., 2002; Raja et al., 2019; Reuben et al., 1997; Riest et al., 2008; Rowe et al., 1992; Siribumrungwong et al., 2015).

The search results for meta-analysis.

In this meta-analysis, a total of 12 articles that met the inclusion criteria were included (Cassinelli et al., 2008; Jirarattanaphochai et al., 2008; Karst et al., 2003; Kelsaka et al., 2014; Kien et al., 2019; Kim et al., 2016; Pookarnjanamorakot et al., 2002; Raja et al., 2019; Reuben et al., 1997; Riest et al., 2008; Rowe et al., 1992; Siribumrungwong et al., 2015). A total of 12 studies were included and investigated the efficacy of NSAIDs in preemptive analgesia for lumbar spine surgery. A total of eight studies used postoperative morphine consumption as the primary outcome measure (Cassinelli et al., 2008; Jirarattanaphochai et al., 2008; Kien et al., 2019; Pookarnjanamorakot et al., 2002; Reuben et al., 1997; Riest et al., 2008; Rowe et al., 1992; Siribumrungwong et al., 2015), with each morphine consumption divided into four subgroups based on postoperative time. Eleven other studies assessed the difference in efficacy between the two groups of patients by pain intensity rating, including VAS and NRS (Cassinelli et al., 2008; Karst et al., 2003; Kelsaka et al., 2014; Kien et al., 2019; Kim et al., 2016; Pookarnjanamorakot et al., 2002; Raja et al., 2019; Reuben et al., 1997; Riest et al., 2008; Rowe et al., 1992; Siribumrungwong et al., 2015), with seven studies using VAS as an outcome indicator (Cassinelli et al., 2008; Karst et al., 2003; Kelsaka et al., 2014; Kien et al., 2019; Kim et al., 2016; Pookarnjanamorakot et al., 2002; Rowe et al., 1992) and four studies using NRS as an outcome indicator (Raja et al., 2019; Reuben et al., 1997; Riest et al., 2008; Siribumrungwong et al., 2015). Three studies reported adverse events including pruritus, nausea and vomiting, dyspepsia, and constipation (Jirarattanaphochai et al., 2008; Raja et al., 2019; Siribumrungwong et al., 2015). The recording of morphine consumption and pain scores (VAS, NRS) at different time points postoperatively can reflect the effect of NSAIDs used perioperatively on the pain relief of patients undergoing lumbar spine surgery at different time points (short or long-term) postoperatively. The characteristics of the studies included in the meta-analysis are listed in Table 1.

The risk of bias assessment for the included 12 RCTs is shown in Fig. 2 (Cassinelli et al., 2008; Jirarattanaphochai et al., 2008; Karst et al., 2003; Kelsaka et al., 2014; Kien et al., 2019; Kim et al., 2016; Pookarnjanamorakot et al., 2002; Raja et al., 2019; Reuben et al., 1997; Riest et al., 2008; Rowe et al., 1992; Siribumrungwong et al., 2015). Nine studies used random sequence generation and gave specific grouping methods (Cassinelli et al., 2008; Jirarattanaphochai et al., 2008; Karst et al., 2003; Kien et al., 2019; Kim et al., 2016; Raja et al., 2019; Riest et al., 2008; Rowe et al., 1992; Siribumrungwong et al., 2015), and six studies used allocation concealment and gave specific allocation schemes (Jirarattanaphochai et al., 2008; Karst et al., 2003; Kien et al., 2019; Kim et al., 2016; Raja et al., 2019; Siribumrungwong et al., 2015), ten studies were blinded (Cassinelli et al., 2008; Jirarattanaphochai et al., 2008; Karst et al., 2003; Kelsaka et al., 2014; Pookarnjanamorakot et al., 2002; Raja et al., 2019; Reuben et al., 1997; Riest et al., 2008; Rowe et al., 1992; Siribumrungwong et al., 2015), and in 12 studies, selective reporting and other biases could not be accurately determined (Cassinelli et al., 2008; Jirarattanaphochai et al., 2008; Karst et al., 2003; Kelsaka et al., 2014; Kien et al., 2019; Kim et al., 2016; Pookarnjanamorakot et al., 2002; Raja et al., 2019; Reuben et al., 1997; Riest et al., 2008; Rowe et al., 1992; Siribumrungwong et al., 2015). The quality of the included studies was acceptable.

Risk of bias assessment graph.+, low risk; −, high risk; ?, unclear.

A total of 8 of the 12 included studies (598 patients) used postoperative morphine consumption as the primary outcome indicator (Cassinelli et al., 2008; Jirarattanaphochai et al., 2008; Kien et al., 2019; Pookarnjanamorakot et al., 2002; Reuben et al., 1997; Riest et al., 2008; Rowe et al., 1992; Siribumrungwong et al.,