A comprehensive literature search was performed across multiple electronic databases, including PubMed, EMBASE, Cochrane Library, CNKI, VIP Database, and Wanfang Data, covering all publications up to April 10, 2024. The search strategy employed a combination of Medical Subject Headings (MeSH) terms and free-text keywords related to the interventions and the condition of interest. For English-language databases, terms such as “Compound Kushen Injection”, “Zoledronic Acid”, “Bone Metastasis”, “Bone Cancer Pain”, “Osseous Metastasis”, “Ostealgia”, and “Malignant Tumor” were used. Equivalent terms were utilized for Chinese-language databases. No language restrictions were applied to maximize the retrieval of relevant studies. Additionally, reference lists of all identified articles were manually searched to identify any additional eligible studies.

Studies were included if they met the following criteria: they were randomized controlled trials involving patients with pathologically confirmed malignant tumors and bone metastasis diagnosed through imaging modalities such as electron beam computed tomography (ECT) bone scan, computed tomography (CT), or magnetic resonance imaging (MRI), and exhibiting clinical symptoms of bone pain. The experimental group received CKI in combination with zoledronic acid, while the control group received zoledronic acid alone. The primary outcome measured was the overall pain relief rate, and secondary outcomes included the incidence of adverse events such as fever, rash, and gastrointestinal reactions. Studies were excluded if they were non-randomized controlled trials, observational studies, case reports, reviews, or editorials, or if they lacked primary outcome data or if valid information could not be extracted.

Study selection and data extraction were performed independently by two reviewers to minimize bias and errors. Initially, the titles and abstracts of all retrieved articles were screened to identify potentially eligible studies. Full-text articles were then obtained for studies that appeared to meet the inclusion criteria or when eligibility was unclear from the abstract. The full texts were assessed independently to determine final inclusion. Discrepancies between reviewers were resolved through discussion or, if necessary, consultation with a third reviewer to reach a consensus.

The methodological quality and risk of bias of the included studies were assessed independently by the two reviewers using the Cochrane Collaboration's Risk of Bias Tool (version 5.1.0). The assessment covered various domains, including selection bias (random sequence generation and allocation concealment), performance bias (blinding of participants and personnel), detection bias (blinding of outcome assessment), attrition bias (incomplete outcome data), reporting bias (selective reporting), and other potential sources of bias. Each domain was judged as having a low, high, or unclear risk of bias based on the information provided in the studies. Any disagreements in the assessment were resolved through discussion or by involving a third reviewer to ensure an objective evaluation.

Statistical analyses were performed using Review Manager software version 5.4. For dichotomous outcomes, odds ratios (ORs) with 95% confidence intervals (CIs) were calculated to measure the effect size. Heterogeneity among studies was assessed using the Chi-squared (χ²) test and quantified with the I² statistic. An I² value of 50% or less, combined with a P-value greater than 0.10, was considered to indicate low statistical heterogeneity, and a fixed-effect model was employed for the meta-analysis. If significant heterogeneity was detected (I² greater than 50% or P-value of 0.10 or less), a random-effects model was used to pool the data. Potential publication bias was evaluated through visual inspection of funnel plots when a sufficient number of studies (generally more than ten) were included in the analysis.

For the pharmacoeconomic evaluation, the target population consisted of individuals with bone metastases from malignant tumors who presented with bone pain symptoms, aligning with the patients included in the meta-analysis. The interventions compared in the cost-effectiveness analysis were as follows: the control group received zoledronic acid injection at a dose of 4 mg, administered intravenously once every 14 days for a total of two doses; the experimental group received the same regimen of zoledronic acid combined with CKI at a dose of 20 ml per day, administered intravenously for 28 consecutive days.

Given the short duration of the clinical studies included and the lack of long-term survival data, a decision tree model was employed for the cost-effectiveness evaluation. This model utilized the results of the meta-analysis to compare the short-term economic benefits of the two treatment regimens (9). The outcome measure was treatment effectiveness, coded as 1 for effectiveness and 0 for ineffectiveness. The study duration was 28 days. The model was constructed using TreeAge Pro 2019 software, with the specific structure illustrated in Figure 1.

The total effective rate of pain relief was used as the primary measure of treatment efficacy. Based on the forest plot results from the meta-analysis, a weighted calculation of the total effective rate for both groups was performed. Cost-effectiveness evaluation was conducted from the healthcare system's perspective, considering only direct medical costs. According to the meta-analysis results, if there was no statistically significant difference in the incidence of adverse reactions, the costs of managing adverse events were not included.

Drug prices were obtained from Yaozhi.com (https://www.yaozh.com/) and adjusted by ±20% for sensitivity analysis. CKI was administered in 5 ml vials, with a daily dosage of 4 vials; the usage range was varied between 2 and 6 vials for sensitivity analysis. The number of treatment days for zoledronic acid and CKI was also subjected to ±20% variation for sensitivity analysis.

Using the constructed decision tree model, the ICER was calculated through cost-effectiveness analysis and compared with the willingness-to-pay (WTP) threshold. If the ICER was lower than the WTP threshold, the combination of CKI and zoledronic acid was considered cost-effective; otherwise, it was not. Based on the 2023 national average disposable income (¥39,218) (10), ¥39,218 was set as the WTP threshold in this study.

Both one-way sensitivity analysis and probabilistic sensitivity analysis were conducted to test the robustness of the model. One-way sensitivity analysis calculated the effect of each parameter's variation on the ICER within the upper and lower limits of the pre-specified range. If the 95% CI for a parameter was known, the CI was used as the range; if not, a ±20% variation around the mean was used. The results were illustrated using a tornado diagram.

Probabilistic sensitivity analysis assessed the combined impact of all parameter variations on the cost-effectiveness of both treatment options. Based on the distribution characteristics of the parameters [Gamma distribution for cost data, Beta distribution for utility values and event rates, and Uniform distribution for drug administration days (11)], Monte Carlo simulation was performed 1,000 times. The results were displayed as cost-effectiveness scatter plots and cost-effectiveness acceptability curves.

After a comprehensive database search and full-text screening, 14 RCTs were included in the analysis the specific screening steps shown in Figure 2. These 14 studies involved a total of 1,269 patients, with 656 in the experimental group and 613 in the control group. A total of 704 males and 565 females were included, with ages ranging from 30 to 84 years. The main types of cancers involved included breast cancer, lung cancer, stomach cancer, nasopharyngeal carcinoma, among others. In all included studies, there were no statistically significant differences in age, gender, tumor type, and other basic conditions between the experimental and control groups, ensuring baseline comparability. Thirteen studies (12–24) reported the overall pain relief rate, seven studies (12–17, 25) reported the incidence of fever, eight studies (12, 13, 15–19, 25) reported gastrointestinal adverse reactions such as nausea and vomiting, and five studies (13, 16–18, 21) reported skin adverse reactions such as rash and itching.

Regarding randomization methods, seven studies reported using a random number table, one used a lottery method, and the rest did not specify the randomization process. None of the included studies mentioned whether blinding was implemented or whether allocation concealment was described, leading to an overall low-quality rating for these studies. The basic information of the included studies is presented in Table 1, and the risk of bias assessment results are shown in Figures 3, 4.

A total of 13 RCTs comprising 1,179 patients were included. The heterogeneity test showed P > 0.05 and I² = 0%, indicating no significant heterogeneity; thus, a fixed-effect model was used for the analysis. The meta-analysis demonstrated that the overall effect of pain relief for CKI combined with zoledronic acid had an OR of 3.43 (95% CI: 2.51–4.67, P < 0.0001), with a statistically significant difference. This suggests that the combination of CKI with zoledronic acid is more effective in treating bone metastatic cancer pain compared to zoledronic acid alone (Figure 5).

The heterogeneity tests for fever, gastrointestinal adverse reactions, and skin adverse reactions all resulted in P > 0.05 and I² = 0%, indicating no significant heterogeneity. Thus, a fixed-effect model was used for analysis. The combined OR values crossed the line of no effect, and the P-values exceeded 0.05, suggesting no statistically significant difference in the incidence of major adverse reactions between the two intervention groups. This indicates that CKI has a favorable safety profile (Figures 6–8).

A funnel plot was generated to analyze publication bias for the overall pain relief rate in the combination treatment of CKI and zoledronic acid for bone metastatic cancer pain. The results showed a roughly symmetrical distribution, indicating no significant publication bias in the current studies (Figure 9).

Based on the weights of each study from the forest plot (Figure 5), the pain relief rates were weighted to calculate the overall effectiveness rate. The results indicate that the overall effectiveness rate for the treatment group is 0.81, while for the control group it is 0.62. Thus, the effect values for the two groups are 0.81 and 0.62, respectively. The costs and relevant parameters required for the model are presented in Table 2.

The cost-effectiveness analysis results are shown in Table 3. The incremental effect is 0.19, and the incremental cost is ¥3,584, resulting in an ICER of ¥18,863.16, which is below the WTP threshold of ¥39,218. Therefore, using the per capita disposable income of 2023 as the WTP threshold, the combination of CKI and zoledronic acid is more economically viable compared to using zoledronic acid alone.

The tornado diagram (Figure 10) illustrates the results of the one-way sensitivity analysis. The three parameters with the greatest impact on the results are the overall effectiveness rate of the CKI group, the overall effectiveness rate of the control group, and the daily usage of CKI. Other factors, such as the duration of CKI usage, the unit price of CKI, the unit price of zoledronic acid, and the duration of zoledronic acid usage, have relatively minor effects on the results.

The probabilistic sensitivity analysis conducted through 1,000 rounds of Monte Carlo simulation indicates that a significant number of points in the incremental cost-effectiveness scatter plot cluster below the WTP threshold of ¥39,218 (Figure 11). This suggests that, at the payment threshold of ¥39,218, the combination of CKI and zoledronic acid provides greater health benefits compared to using zoledronic acid alone. The cost-effectiveness acceptability curve indicates that when the WTP threshold exceeds ¥18,863.16, the probability of the economic viability of the combination therapy gradually increases, further supporting the economic advantage of combining CKI with zoledronic acid (Figure 12).