Introduction

Lumbar disc herniation (LDH) refers to the rupture of the annulus fibrosus of the intervertebral disc, leading to protrusion of the nucleus pulposus, compression of spinal nerves and cauda equina, and consequent inflammatory reactions, resulting in clinical symptoms such as pain and neurological dysfunction1. Low back pain is the leading cause of disability worldwide, and LDH is one of its most common contributors2, with a global prevalence of 7.62%3. In China, over 300 million individuals have lumbar spine disorders, and approximately 15.2% have been diagnosed with LDH. The peak incidence occurs between the ages of 30 and 50 years, and the prevalence is on the increase due to changing lifestyles and work patterns. The associated pain and functional impairment severely compromise patients’ quality of life4.

Current first-line treatments for LDH include conservative management and surgical interventions5,6. Studies report that symptoms in most patients improve with 6–12 weeks of conservative management7,8. Conservative approaches encompass bed rest, pharmacotherapy, exercise therapy, epidural injections, lumbar traction9,10,11. Surgical treatment is recommended for patients with severe pain that is unresponsive to conservative treatment, signs of cauda equina syndrome, or progressive motor weakness (Medical Research Council scale ≤ 3/5)12,13,14,15. Treatment selection—whether conservative or surgical—is influenced by multiple factors. While conservative management is safer and more suitable for mild or acute cases, its prolonged duration may be a disadvantage to patients with severe pain or disability. On the other hand, surgical intervention, though effective, raises safety concerns for some individuals16. In this context, informed treatment decision-making is critical. However, decision conflict —a psychological state of uncertainty when weighing risks, benefits, and personal values—may hinder optimal choices17,18,19,20.

The Health Belief Model is a core theoretical framework for understanding health behaviors. It posits that an individual’s health-related decisions and behavioral choices are collectively shaped by their perceptions of health threats (including perceptions of susceptibility and severity), as well as their perceptions of treatment benefits and barriers21,22. A study by Alshagrawi et al.23, which focused on COVID-19 vaccination decisions, revealed that perceptions of the severity of the virus and recognition of the protective efficacy of vaccines could significantly and positively predict individuals’ vaccination intentions and behaviors. In contrast, concerns about vaccine side effects reduced the certainty of vaccination decisions. This finding indicates that when individuals have a clearer understanding of disease threats and a more positive evaluation of treatment benefits, their goals regarding treatment decisions become more definite, and they are less likely to fall into the conflict of deciding “whether treatment is needed.” Furthermore, Khalilet al.24 validated this perspective in a study on prostate cancer prevention decisions: after health belief-based educational interventions significantly improved participants’ perceptions of cancer susceptibility and their beliefs in the benefits of screening, they not only showed a greater tendency to proactively choose screening behaviors but also exhibited a significant reduction in hesitation during decision-making processes, such as deciding “whether to undergo screening” and “when to initiate screening,” reflecting the alleviation of decision conflict. Additionally, a study by Taghikhah et al.25 found that patients with stronger health beliefs(HBs) made faster decisions on “whether to seek medical examination” after experiencing suspected symptoms and showed less preoccupation with secondary information such as “medical costs” and “accuracy of examinations” during the decision-making process. Conversely, patients with insufficient HBs ultimately had long-term decision delay. Based on the above evidence, we hypothesize that HBs may be negatively associated with treatment decision conflict (TDC) (Hypothesis 1).

Treatment expectations (TEX)—patients’ anticipated outcomes regarding cure likelihood, efficacy, and quality of life—also play a pivotal role in decision-making. Expectations significantly shape treatment preferences, adherence, and psychological states. A study reported that in patients with chronic disease, positive expectations correlate with reduced symptoms and improved quality of life26 Furthermore, a study by Smith27 showed that positive preoperative expectations in patients who underwent lumbar spine surgery were associated with lower post-surgical decision regret, directly supporting the critical role of expectation valence. Conversely, unmet expectations may trigger disappointment, depression, or decision-making dilemmas28. Studies propose that treatment expectations may act as a potential bridge in the correlational relationship between HBs and TDC (Fig. 1). Previous studies showed that patients often report selecting therapies that align with their perceived health gains shaped by belief-driven expectations29. This mediating role is further supported by multiple studies: Zhang found that patients with LDH with realistic positive expectations for conservative treatment had lower TDC when choosing between surgery and non-surgery30, while French reported in a systematic review of chronic low back pain that positive physical therapy expectations correlated with higher adherence and lower decision uncertainty31. These findings suggest that expectations may exhibit a mediating pattern in this correlational relationship (Hypothesis 2).

Fig. 1
Fig. 1
Full size image

Theoretical framework.

Although previous studies explored factors influencing TDC, the interplay between HBs, TEX, and TDC in LDH remains underexamined. This study aims to clarify the impact of HBs on TDC in patients with LDH, with TEX as a mediator. The findings may provide valuable insights and evidence to support informed decision-making for this population.

Methods

Study participants

This study recruited participants via convenience sampling for a cross-sectional questionnaire survey. The inclusion criteria were as follows: (1) diagnosis of LDH confirmed by a combination of clinical examination and appropriate imaging (lumbar MRI or CT scans)—specifically, patients must present with clinical symptoms (e.g., low back pain with lower limb radiculopathy, sensory loss, muscle weakness, or hyporeflexia) that correspond to the level of disc herniation identified on imaging; (2) age ≥ 18 years; and (3) voluntary participation. The exclusion criteria included individuals with communication barriers. The survey was anonymous, and all participants provided informed consent prior to completing the questionnaires. A total of 737 questionnaires were collected, with 30 questionnaires excluded as invalid (18 due to incomplete key scale items and 12 due to logical contradictions in responses). Finally, 707 valid responses were obtained (validity rate: 95.93%), and for individual missing scale items (missing rate < 1%), mean imputation by subscale was used to ensure data integrity.

Data collection

Demographic characteristics

Demographic data were collected using a structured questionnaire that covered eight parameters: (a) gender, (b) marital status, (c) age, (d) education level, (e) health insurance type, (f) employment status, (g) annual household income per capita, and (h) residential area.

Measurement of treatment decision conflict

The Decision Conflict Scale (DCS), developed by O’Connor et al.32 and cross-culturally adapted into Chinese by Li33, was used to assess treatment decision conflict. This 16-item scale comprises five subscales: Informed (3 items), Values Clarity (3 items), Support (3 items), Uncertainty (3 items), and Effective Decision-Making (4 items). Responses were rated on a 5-point Likert scale (0 = strongly agree to 4 = strongly disagree). The total scores are calculated by averaging item scores and multiplying them by 25 to yield a 0–100 scale, with higher scores indicating greater decision conflict. The Chinese version demonstrated excellent internal consistency (Cronbach’s α = 0.897).

Measurement of health beliefs

HBs were measured using the revised Chinese version of the Health Belief Scale (HBS). The original scale, translated by Yuan34, is a widely recognized tool for assessing HBs. Ji et al.35 conducted cross-cultural adaptation and validation, reporting strong reliability (test-retest reliability = 0.889, split-half reliability = 0.936) and internal consistency (Cronbach’s α = 0.967). The 48-item scale includes five subscales: Personal HBs, Perceived Ability to Implement, Perceived Control, Perceived Resource Utilization, and Perceived Threats. Items are rated on a 5-point Likert scale (1 = very weak to 5 = very strong), with higher total scores reflecting stronger HBs.

Measurement of treatment expectations

The Treatment Expectation Questionnaire (TEX-Q), developed by Shedden-Mora et al.36,37 at the University Medical Center Hamburg-Eppendorf (UKE), Germany, and adapted into Chinese by Yang38, was employed. This 15-item tool assesses six dimensions: Treatment Benefit, Positive Impact, Adverse Events, Negative Impact, Treatment Process, and Behavioral Control. Responses are scored on an 11-point Likert scale (0–10). We used the overall mean score of the TEX-Q after reversing the harm expectation subscales (items 7–11), with higher values indicating more positive overall treatment expectations. The Chinese version exhibited excellent reliability (Cronbach’s α = 0.903 overall; 0.842–0.924 for subscales; split-half reliability = 0.958).

Data analysis

Data analysis for this study was conducted using R software (version 4.3.3; R Core Team, 2024) on Windows 11 × 64 (build 26100). Descriptive statistics were used to summarize the key characteristics of patients with LDH. Continuous variables were presented as mean ± standard deviation (M ± SD) and categorical variables as frequencies and percentages. Independent sample t-tests or one-way ANOVA were used to examine the impact of different demographic characteristics on TDC. According to Hair et al.39, the correlation coefficient between individual items and the total scale score needs to be at least 0.5 for scale validity and appropriateness. Mediation analysis was conducted using the mediation package to examine the mediating effects40.

Ethics

This study was approved by the hospital ethics committee (approval number: 2025SZSYLCYJ0404) and complied with the ethical standards of the declaration of Helsinki. Written informed consent was obtained from each patient included in this study prior to participation.

Results

Participant characteristics

The sample comprised 707 consecutively recruited patients with LDH, of whom 63.083% were male and 36.917% were female. Regarding marital status, 68.741% were married while 31.259% were unmarried. Furthermore, regarding the educational background of the participants, 22.631% held a bachelor’s degree or higher, 11.598% had an associate degree, 36.21% completed high school or vocational secondary education, 19.095% had a junior high school education, and 10.467% had an elementary school education or below.

In terms of employment status, 53.890% were employed, 30.127% were retired, and 15.983% were unemployed. Regarding insurance coverage, 54.173% had BMI insurance, 28.854% had NCMS insurance, and 16.973% had URBMI insurance. Residential distribution showed that 59.689% lived in urban areas, while 40.311% resided in non-urban areas.

The time since the first onset of symptoms was distributed as follows: 28.713% within 3 months, 21.358% between 3 and 6 months, 28.854% between 6 and 12 months, and 21.075% over 1 year. Regarding the first medical consultation, 37.907% visited outpatient departments of tertiary hospitals, 34.795% chose to visit the outpatient departments of secondary or community hospitals, 13.861% were hospitalized in secondary hospitals, while 13.437% were hospitalized in tertiary hospitals.

Among all participants, 84.583% had previously sought medical care for lumbar disc herniation-related issues, while 15.417% had not. Regarding the annual income distribution of the participants (Table 1), 21.782% earned over ¥150,000 (approximately over $20,985; upper-class), 20.226% had an annual income between ¥80,000 and ¥150,000 ($11,216–$20,985; solid middle-class), 29.986% earned between ¥30,000 and ¥80,000 ($4,197–$11,216; categorized lower-middle-class), and 28.006% had an annual income of less than ¥30,000 (approximately less than $4,197; low-income).

Table 1 Descriptive statistics.
Full size table

Level of treatment decision conflict, health beliefs, and treatment expectations, and univariate analysis of treatment decision conflict

The scores of DCS, HBS, and TEX-Q of the recruited patients with LDH were 25.748 ± 14.241, 168.356 ± 33.543, and 6.797 ± 0.949, respectively (Table 2). The univariate analysis for TDC revealed that all general categorical variables included in this study showed significant between-group differences (p < 0.05), except for whether the participants had previously sought medical attention for LDH (i.e., consulted a doctor). Specifically, these significant variables included gender, marital status, education level, insurance type, employment status, annual income, residential area, time since symptom onset, and the hospital type for the first medical consultation. All variables demonstrating significant between-group differences were included as covariates in subsequent analyses (Table 3).

Table 2 Descriptive statistics.
Full size table
Table 3 Univariate analysis of decision conflict by demographic characteristics (ANOVA/t-tests).
Full size table

Correlations of treatment decision conflict, health beliefs, and treatment expectations

Correlation analysis revealed significant associations among the TDC, HBs, and TEX. HBs (M = 168.356, SD = 33.543) showed a negative correlation with TDCS (r = − 0.660, p < 0.001) and a positive correlation with TEX (r = 0.202, p < 0.001). TEX (M = 101.960, SD = 14.24) were negatively correlated with TDC (r = − 0.322, p < 0.001).

Additionally, age, symptom, operative treatment, having seen a doctor, the form of the first visit and the level of hospital demonstrated significant correlations with either treatment expectations or DCS. health belief were negatively correlated with TDC but positively correlated with treatment expectations. Notably, variables significantly associated with TDC and TEX were also included as covariates in subsequent analyses to control for potential confounding effects (Table 4).

Table 4 Means, standard deviations, and correlations with confidence intervals.
Full size table

In this study, common method bias (CMB) was not a significant concern, as the first common factor explained only 26.50% of the variance, which is below the critical threshold of 40% set by Harman’s single-factor test, indicating no substantial CMB.

Mediating role of treatment expectations on the relationships between health beliefs and treatment decision conflict

Regression analyses sequentially examined the correlational relationships among HBs, TEX, and TDC. The results demonstrated the following: (1) HBs were significantly and negatively associated with TDC (β = −0.555, p < 0.001), accounting for 50.1% of the variance in DCS; (2) 2) HBs were positively associated with TEX (β = 0.182, p < 0.001), explaining 22.5% of the variance in TEX; (3) Both HBs and TEX negatively predicted TDC with HBs showing a stronger effect (β = −0.595, p < 0.001) compared to TEX (β = −0.163, p < 0.001). Overall, they explained 52.1% of the variance in DCS. All models controlled for covariates in the analyses (Table 5).

Table 5 Regression coefficients.
Full size table

The mediation analysis revealed a statistically significant indirect effect of HBs on TDC through TEX (β = −0.013, S.E. = 0.005, 95% CI [− 0.024, − 0.006]). Simultaneously, the model demonstrated that the direct effect of HBs on TDC remained significant after controlling for TEX’s influence on TDC (β = −0.223, S.E.= 0.022, 95% CI [− 0.266, − 0.179]). These findings indicate that treatment expectations exhibit a partial mediating pattern in the correlational relationship between HBs and TDC, based on cross-sectional data. Although mediation was statistically significant, the small effect suggests other unmeasured factors may contribute to TDC (Table 6).

Table 6 Mediation effects.
Full size table

Discussion

This study analyzed 707 patients with LDH to explore the relationship between HBs, TEX, and TDC to assess the interactions among these variables and their impact on patients’ decision-making processes. Consistent with the findings of previous correlational studies, this study’s findings demonstrated that HBs are significantly negatively associated with TDC in patients with LDH21,22. HBs were found to be a significant negative predictor of TDC (β = −0.555, p < 0.001), accounting for 50.1% of the variance in TDC. These findings align with those of previous research21, indicating that stronger HBs are associated with lower feelings of conflict when facing medical decisions. HBs correlate with reduced perceived uncertainty, which may be attributed to patients reporting greater confidence in treatment options, in turn correlating with lower TDC22; however, temporal ordering cannot be established. This result underscores that strengthening patients’ HBs could be an effective strategy to reduce TDC in clinical practice. Additionally, this study found that patients’ perceptions of disease severity, treatment benefits, and treatment barriers significantly influenced their treatment choices. For example, patients who perceived their disease as severe and treatment benefits as significant were more likely to opt for aggressive treatment, whereas those who perceived more treatment barriers tended to choose conservative treatment or delay treatment. A study by Herrmann et al.41 also showed that disease conditions can influence patients’ treatment decisions.

Higher TEX were significantly associated with lower decision conflict (β = -0.163, p < 0.001),aligning with prior studies showing that higher expectations correlate with reduced decision conflict in chronic disease management25,26. This result further supports the importance of treatment expectations in the decision-making process, suggesting that clinicians should fully consider patients’ expectations when formulating treatment plans to reduce TDC; this is consistent with the findings of Oswald26. Treatment expectations influence treatment outcomes by affecting patients’ subjective feelings and behavioral responses. For instance, positive expectations can enhance treatment effects, producing a placebo effect, while negative expectations may lead to a nocebo effect that may hinder treatment efficacy28. Expectations not only influence short-term treatment outcomes but may also have long-term effects on patients’ recovery and quality of life. Research indicates that positive expectations can promote patient recovery and improve treatment satisfaction and adherence27. Therefore, treatment expectations play a significant role in medical decision-making dilemmas by influencing patients’ psychological states, treatment behaviors, and clinicians decision-making processes. Understanding this mechanism can help optimize treatment plans, improve patient satisfaction and treatment outcomes, and reduce ethical dilemmas and complexities in medical decision-making.

HBs exhibited a significant indirect associational pattern with TDC through TEX (β = −0.013, S.E. = 0.005, 95% CI [− 0.024, − 0.006]). Additionally, the model showed that after controlling for the effect of TEX on TDC, the direct effect of HBs on TDC remained significant (β = −0.223, S.E. = 0.022, 95% CI [− 0.266, − 0.179]). These results indicate that TEX show a partial mediating pattern in the correlational relationship between HBs and TDC. However, this pattern reflects observed associations in cross-sectional data and does not imply causation. HBs not only directly influence treatment decisions but also indirectly affect decisions by shaping patients’ TEX. Additionally, patients’ expectations regarding treatment efficacy, process, and behavioral control serve as a bridge between HBs and treatment decisions. HBs were significantly and positively associated with treatment expectations (β = 0.182, p < 0.001), accounting for 22.5% of the variance in treatment expectations in this cross-sectional sample. This suggests that stronger HBs are associated with higher expectations for treatment outcomes. HBs may enhance positive attitudes toward treatment, thereby raising expectations for treatment efficacy. This implies that clinicians should pay attention to patients’ HBs when formulating treatment plans to improve their expectations, potentially enhancing treatment adherence and outcomes.

This study highlights the importance of considering patients’ HBs and TEX in clinical practice. Healthcare professionals should actively assess patients’ HBs, understand their perceptions of disease and treatment, and use effective communication and education to help patients develop positive TEX, thereby facilitating decisions that promote recovery. Strengthening HBs fosters more rational treatment decisions, while positive TEX can influence patients’ HBs and decisions. A comprehensive understanding of the interactions among HBs, treatment decisions, and expectations is crucial for developing personalized treatment plans and interventions in clinical practice.

Study limitations

A key limitation of this study is the use of cross-sectional data to examine mediation, which precludes establishing temporal ordering of variables or definitive causal relationships. Moreover, mediation analyses in cross-sectional designs identifies correlational patterns rather than causal pathways; hence, the observed mediating role of TEX should be interpreted as an associational trend rather than a directional effect. Future longitudinal studies are needed to verify the temporal sequence of HBs, TEX, and TDC, which would strengthen inferences about potential causal relationships.

Furthermore, the study sample was primarily drawn from a specific region, limiting generalizability to other populations. Future research could expand the sample scope to improve generalizability. Additionally, the influence of HBs and TEX may vary across cultural contexts. Future studies should endeavor to stratify patients into cultural geographic subgroups and incorporate longitudinal designs to address temporal limitations. Longitudinal studies could further investigate the causal relationships among HBs, TEX, and TDC, providing new perspectives for medical decision-making support. Future studies should endeavor to stratify patients into cultural geographic subgroups.

Conclusion

By analyzing data from 707 participants, this study revealed the relationships between HBs, TEX, and TDC. The results showed that HBs were significantly negatively associated with TDC and positively associated with TEX in this cross-sectional sample. Treatment expectations partially mediated the relationship between HBs and TDC. These findings highlight that enhancing patients’ HBs and calibrating realistic TEX may be associated with reduced TDC in clinical practice, based on observed correlations; however, causal claims are not supported by cross-sectional data. Future research could explore other potential mediating variables and the stability of these relationships across different cultural contexts.

Clinicians should strengthen patients’ HBs through education and communication to reduce conflicts in medical decision-making. When formulating treatment plans, clinicians should consider patients’ expectations and use proactive communication and information provision to improve their expectations of treatment outcomes. Given patients’ diverse socioeconomic backgrounds and health conditions, individualized treatment plans should be developed to meet specific needs and reduce TDC. This study provides new insights into the relationships among HBs, TEX, and TDC and offers valuable guidance for clinical practice.