Maine Lumbar Spine Study Despite the RCT being considered the most valid study design, the Maine lumbar spine study (MLSS), a prospective cohort study, provides some of the best long-term data comparing surgical and nonoperative treatment of sciatica caused by lumbar disc herniation (Table 2).10-13 The MLSS enrolled 507 (235 surgical and 272 nonoperative) patients from the practices of 25 surgeons and 5 occupational medicine specialists in Maine. Patients were enrolled if they suffered from sciatica, defined as pain radiating to below the knee, though confirmatory imaging studies were not required. Given that it was an observational study, treatment was determined by the treating physician and the patient. The primary outcome measure was self-reported improvement in the predominant symptom (leg pain or back pain). The exact definition of “improvement” varied among the 3 reports (1, 5, and 10-year follow-up), with patients reporting that their predominant symptom was “much better” or “completely gone” qualifying as improved in the 1-year report, whereas those answering “better” were also included in the “improved” group at 5 and 10 years. Many other outcome measures, including back and leg pain frequency and bothersomeness, sciatica frequency and bothersomeness indexes, Roland disability scale, SF-36 scores, and work status, were also recorded. Results were reported at 1, 5, and 10 years, and statistical modeling techniques were used to control for the significant baseline differences between the 2 groups. Substantial crossover occurred, with 15% of patients who initially chose nonoperative treatment undergoing surgery within 3 months, and 25% of the remaining nonoperative patients undergoing surgery between 3 and 120 months. The authors addressed this by assigning patients who crossed over within the first 3 months to the surgery group, while analyzing those who subsequently crossed over after 3 months with the nonoperative group.
缅因州腰椎研究尽管RCT被认为是最为有效（可信度）的研究设计，缅因州腰椎研究（MLSS），一项前瞻性队列研究，针对腰椎间盘突出导致坐骨神经痛的患者，也为手术与非手术治疗的比较提供了一些长期随访的数据（表2）[10-13]。MLSS纳入了507例患者（235例手术，272例非手术），这些患者来自缅因州的25位外科医生和5为职业病学专家的门诊。如果患者存在坐骨神经痛，一般指膝部以下的放射痛，即使没有获得确切的影像学证据都纳入研究。由于这是一项观察研究，治疗方式取决于主治的医师和患者的意愿。评估结果的主要指标为主要症状（腿痛或腰痛）改善程度的自我评价。对于“改善”的确切定义在3个报告（1，5，10年随访）中存在差异，在1年随访的研究中，患者认为其主要症状“好得多”或“完全消失”则归为“改善”，而5年和10年随访的报告中，患者称其“较好”也归为“改善”。很多其他的治疗结果评价包括背痛和腿痛的频率和焦虑，坐骨神经痛的频率和焦虑指数，Roland功能障碍等级，SF-36评分，以及工作状况都记录在案。1、5、10年都报道了相关的结果，应用统计模型方法以控制两组间显著的基线差异。两组间治疗方式的变换较多，15%的患者起初选择非手术治疗3个月内进行了手术治疗，余下的非手术治疗患者25%在3-120个月之间进行了手术治疗。作者将3个月内变换治疗方式进行手术的患者直接分配到手术组，而对3个月以后非手术治疗组变换治疗方式进行手术治疗的患者进行分析。
The MLSS demonstrated that surgical patients were significantly more likely to report improvement in their predominant symptom compared with the nonoperative patients at 1 and 5 years (71% vs 43%, P < 0.001 at 1 year; 70% vs 56%, P < 0.001 at 5 years). By 10 years, the difference on this outcome measure was no longer significant (69% of the surgical patients reported improvement vs 61% for the nonoperative patients, P = 0.2). However, if only patients who answered that their predominant symptom was “much better” or “completely gone” were included in the improved category (as was the case for the 1-year results), the surgical group continued to have significantly better results at 10 years compared with the nonoperative group (56% “definitely improved” with surgery vs 40% nonoperative, P = .006). The surgical group also had significantly better results on most secondary outcome measures at all follow-up times, including low back pain improvement, leg pain improvement, sciatica frequency and bothersomeness indexes, and the Modified Roland Scale. The proportion of patients returning to work and receiving disability compensation was similar for the 2 treatment groups at all follow-up times. The timing of improvement varied between the 2 groups, with the surgery group reaching maximal improvement on the Modified Roland Scale within the first year, whereas the nonoperative group continued to make small gains on this outcome between 2 and 10 years. By 10 years, 25% of surgical patients had undergone at least 1 additional spine operation, and a similar percentage of patients in the nonoperative group underwent surgery between 3 months and 10 years. In summary, the surgery patients had greater improvement of pain, function, and satisfaction, whereas there were no differences in work status between the 2 groups. MLSS显示，1年和5年时，与非手术的患者相比，手术患者报告其主要症状的改善更为显著（1年时，71%vs43%，P<0.001；5年时，70%vs56%，p<0.001）。10年时，治疗结果的这一评价指标不再有明显的差异（69%的手术患者报告改善，而非手术患者为61%，p=0.2）。然而，如果针对主要症状的改善情况，只将回答“好很多”或“完全消失”的患者归入“改善”（与1年时结果评价类似），则手术组与非手术组相比，在10年仍然具有更好的结果（手术组56%的患者“明确地改善”，而非手术组则为40%，p=0.006）。在所有的随访时间点，手术组在很多次要的结果评价指标上也有明显更好的结果，包括腰痛的改善、腿痛的改善、坐骨神经痛的频率、焦虑指数、改良的Roland等级等。在所有时间点，两组患者恢复工作的比率和接受劳动能力丧失补偿的比率都相似。2组患者病情改善的时间差异较大，手术组改良Roland等级达到最大改善的时间出现在第一年内，而非手术组接近这一结果的时间要延迟至2-10之间。10年时，25%的患者经历了至少1次额外的脊柱手术，而非手术组3个月至10年间进行手术的患者也有类似的百分比。总的来说，手术的患者疼痛、功能和满意度等的改善更大，而两组间的工作状况则没有明显的差异。
The MLSS was the first large-scale study to compare surgical and nonoperative outcomes for lumbar disc herniation. Its strengths include its size, prospective nature, multicenter involvement, long-term follow-up with relatively low attrition, and use of validated outcome measures. However, its limitations must be considered. Its observational design contributed to marked baseline differences between the surgical and nonoperative groups, with the surgical group generally having worse symptoms and fewer workers’ compensation patients than the non-operative group. Although statistical modeling could control for these measured baseline differences, the potential for confounding by unmeasured variables existed. Additionally, specific radiographic findings correlating with clinical findings were not required as inclusion criteria, so patients without actual disc herniations may have been included. There was a substantial crossover from initial nonoperative treatment to surgery after 3 months (25% of nonoperative patients), and these patients were included in the nonoperative group for analysis. A 10-year as-treated analysis was performed that showed no significant differences between the two as-treated groups, with the exception of greater improvement on the Roland score for the surgical group. This suggests that the benefit of surgery was not underestimated by including patients who underwent surgery beyond 3 months in the nonoperative group. In fact, patients who initially chose nonoperative treatment and subsequently underwent surgery had the worst results of all, with only 40% of these patients reporting improvement in their predominant symptom at 10 years. A final concern was the use of mail-in questionnaires rather than actual clinical follow-up as this precluded repeat physical examinations. Despite these limitations, the MLSS currently offers the best long-term follow-up data comparing surgery with nonoperative treatment for lumbar disc herniation.
MLSS是第一个比较腰椎间盘突出症手术与非手术治疗结果的大样本研究。其优势主要体现在样本量、前瞻性、多中心、长期随访且缺失相对较小、应用了有效的结果评价指标。然而，也应该认识到它的局限性，其观察性的设计使得手术与非手术组的基线存在显著的差异，手术组通常症状更严重，且与非手术组相比接受职工补偿的患者更少。虽然应用统计模型能控制这些基线评估上的差异，但仍然存在一些无法测定的变量导致潜在的混杂。此外，与临床表现相符的特异性的影像学表现并没有将其作为纳入标准，因此，没有确切的椎间盘突出的患者也纳入了本研究。最初选择非手术治疗的患者有较大一部分（25%非手术治疗的患者）3个月后变换治疗方式，进行手术治疗，这些患者仍被归入非手术组进行分析。10年时进行的接受治疗分析显示，除了手术组Roland等级有较大的改善外，两个接受治疗组的差异没有统计学意义，这表明非手术组的患者3个月后进行手术治疗没有低估手术的效力。事实上，最初选择非手术治疗后来进行手术的患者在所有患者中治疗结果最差，10年时这些患者仅有40%报告其主要症状“改善”。最后一点顾虑则是该研究通过邮寄调查问卷而不是进行实际的临床随访，这样并没有进行反复的体格检查。尽管有如此多的局限，MLSS还是提供了目前最好的长期随访资料，对腰椎间盘突出症的手术与非手术治疗进行对比。