This study shows that 44%–47% of the patients with sciatica who were referred for secondary care had a non-successful outcome at 1
year and 39%–42% at 2
years. For the multivariable models, a high score for comorbid subjective health complaints was the only variable that predicted non-success at both 1 and 2
years. This finding was true for both for the main and the secondary outcome. For the main outcome, males, smokers, patients with higher scores for low back pain and patients who had not undergone surgery had an independent association with non-success at 1
year, but not at 2
years of follow-up. A long duration of back pain and sciatica symptoms and a high level of kinesiophobia were associated with non-success at 2
years. No sciatica-specific clinical findings were associated with non-success, except for a weak association with abnormal tendon reflex at 1
year. For the secondary outcome, muscular weakness at the clinical examination, higher scores for low back pain, and no surgical treatment were associated with non-success after both 1 and 2
years of follow-up. Smoking was also associated with non-success after 2
years of follow-up.
The main strength of this study is the large sample size, the high response rate and the use of imaging to confirm the diagnosis of disc herniation. We used the most precise outcome measures for the current cohort, which in a previous study showed the highest sensitivity and specificity to discriminate between successful outcome or not for sciatica patients [29
]. A broad range of potential prognostic variables including several clinical findings, psychological variables and comorbid subjective health complaints were investigated.
A limitation to the internal validity of the study was an incomplete recording of patients who according to the inclusion and exclusion criteria were eligible but for some reason either were not invited or declined to participate. Non-response to the follow-up questionnaires may also have biased the results. Other potential prognostic factors were not investigated in the current study, for example details of imaging findings [20
] and the phenomenon of centralization [35
] may become important. Lastly, a similar model should be tested in another sciatica sample of sciatica patients with a follow-up period exceeding 1
The Nagelkerke R2
values of 20%–26% is consistent with or lower than those of other studies of sciatica cohorts, and indicates that only a small proportion of the variance in the outcome was explained by the included variables [14
The choice of the main dependent variable, MSBQ ≥ 5, was based on the results of a validation of outcome measurements in the current sciatica cohort [29
]. MSBQ with a cut-off of 4.5 had the highest sensitivity and specificity when the global change score was used as an external criterion. The cut-off value for the secondary dependent variable, SBI ≥ 7, was based on the same validation. There is no gold standard for the definition of non-success in patients with sciatica and disc herniation. In 2011, Kamper et al. [37
] presented a systematic review of 82 studies of low back pain, including 14 studies of sciatica. They concluded that there is a great variation in how recovery is measured. In two of the studies of sciatica, a composite measure was used that included many different scales of pain and function [20
]. Eight of the reports did not describe the details of the definition of a good outcome for sciatica. The lack of a gold standard has contributed to the fact that practically no-one has used the same definition or measures of recovery or a successful outcome for sciatica.
The surgery variable was the only variable not recorded at baseline, but only during the follow-up period. This variable is complex because it contains both the decision regarding surgery and the fact that surgery was performed, hence, it was not possible to adjust the variable for symptoms and signs at the time it was decided to operate the patient. This might influence the interpretation of the results of the surgical treatment. However, 81% of the patients who were treated surgically were operated on during the first 3
months of the follow-up period. Consideration might also be given to the fact that the conservative treatments offered could also modify the effects of some of the prognostic factors.
The long-term prognosis in the present study was in accordance with the findings of other studies, but the prognosis for the surgically treated patients in the current study was poorer than that of other studies [9
]. Additionally, the differences between surgically treated and non-surgically treated patients were smaller than in comparable studies [10
Comorbid subjective health complaints was the only variable associated with non-success at both 1 and 2
years. This variable was associated with both the main and secondary outcome. One possible explanation is that patients with a lower threshold for reporting bodily discomfort report more complaints and have a poorer prognosis. In two Norwegian studies, high scores were associated with reduced function and more complaints in patients with nonspecific low back pain [38
] and whiplash [39
]. We have previously reported that the patients in the current cohort at baseline reported more comorbid subjective health complaints than the general population and that the number of subjective health complaints nearly doubled in those with persisting sciatica at the 1
year follow-up [40
]. In the final model, the significance of emotional distress was not maintained. This is contrary to the results of other studies of sciatica, where emotional distress has been found to be related to pain and disability [17
]. Emotional distress is also reported to be an important prognostic factor for nonspecific low back pain [42
]. However, none of those studies tested the influence of comorbidity.
In the present study, in terms of the primary outcome measure, females had better outcomes than males at the 1-year follow-up. This contrasts with the results of the study of Peul [15
], in which female sex was a strong predictor for an unsatisfactory outcome at 1
year for patients with sciatica and disc herniation. One possible explanation for the divergent results is that our data were adjusted for subjective health complaints, emotional distress and kinesiophobia. The exclusion criteria in the study of Peul were duration of sciatica symptoms of more than 12
weeks, similar complaints during the previous year, or severe comorbidity. Therefore, our study is probably more representative of the majority of patients with sciatica and disc herniation who are referred for secondary care.
The poor prognosis among smokers is in agreement with the results of some studies on surgically treated patients [19
], but conflicting results have also been published [14
Kinesiophobia was an independent prognostic variable for non-success at 2
years. The fear-of-movement/(re)injury model states that the reaction to pain may consist of confrontation in patients with a non-catastrophizing behaviour, and of avoidance in patients with a high catastrophizing behaviour. Pain-related fear may lead to prolonged chronic pain and disability [45
] and was prognostic for non-success in patients with nonspecific low back pain after 6
months in two Dutch studies, one of which was a population-based survey [46
] and one of which included army workers [47
]. In a cross-sectional study of a Swedish population with specific low back pain (defined as disc herniation, isthmic spondylolisthesis or spinal stenosis) attending an orthopaedic clinic, high scores for kinesiophobia were associated with high disability scores [48
]. Another cross-sectional Swedish study on patients treated with surgery for disc herniation found that half of the patients suffered from kinesiophobia 10–34
months after the operation and that patients with kinesiophobia were more affected in several other variables as pain, disability and symptoms of depression [49
]. Contrary to these findings, fear of movement and pain catastrophizing were not associated with recovery among patients with residual complaints at 3 and 12
months following lumbar disc surgery [50
Patients who had back pain and sciatica of longer duration at inclusion were about twice as likely to report non-success at 2
years. This is consistent with some studies on surgically treated patients [19
], but inconsistent with studies of conservatively treated patients [20
]. None of these studies had a follow-up period exceeding 1
year, and therefore they cannot be properly compared with the results of the current study.
The final models for the primary outcome showed that patients who were not treated surgically were nearly three times more likely to report non-success at 1
year, but no significant association was identified between surgical treatment and outcome at 2
years. Most operations were performed during the first 3
months of follow-up. The benefits of surgical treatment decreased with time, which is similar to the results of other studies [3
]. However, when SBI was used as the dependent variable, there was an association between non-success and no surgical treatment after both 1 and 2
years. The SBI is a variable that describes the radicular symptoms of sciatica as pain, sensory symptoms, and paresis. Decision regarding surgical treatment might depend more on the specific sciatic symptoms described in the SBI than on the symptoms related to function in the MSBQ. The interpretation of the interaction effect between smoking and surgery might be that the association between (non-)surgical treatment and (non-)success differs for smokers and non-smokers.
Different prognostic factors were identified for the 1- and the 2-year observations. Comorbidity, kinesiophobia, and duration of symptoms at baseline were associated with non-success at the 2-year follow-up and may indicate that psychosocial factors are more important for the long-term prognosis than sciatica specific symptoms and disability. Fear avoidance and comorbidity are not routinely assessed in consultations for sciatica. Factors and treatments that improve the short-term outcome are important for the patient and for society, but it may be even more important to identify the factors that predict long-term outcomes at an early stage in order to help the patient to solve their problems.
This study identified prognostic factors associated with non-success in sciatica patients. Predictor studies are provided to make estimates of probability and are a supplement for clinicians in their work with the patients [30
]. The current results suggest that the prognosis for sciatica patients referred to secondary care is not as good as previously reported and is only slightly better after surgery, and that comorbidity and kinesiophobia should be assessed in patients with sciatica, including surgical candidates.