Protocol. This trial protocol has been provided by the authors to give readers additional information about their work.

Transcription

1 Protocol This trial protocol has been provided by the authors to give readers additional information about their work. Protocol for: el Barzouhi A, Vleggeert-Lankamp CLAM, Lycklama à Nijeholt GJ, et al. Magnetic resonance imaging in follow-up assessment of sciatica. N Engl J Med 2013;368: DOI: /NEJMoa

2 This supplement contains the following items: 1. Original research protocol of the randomized controlled clinical trial (Page 2 to 22) with the MRI findings at follow-up as an outcome of the clinical trial (bottom page 9) 2. Original research protocols with regard to the MRI diagnostic part of the Clinical Trial (Page 23 to 68). 3. Final protocol regarding the 1-year MRI findings including the case record forms used to evaluate the baseline and one year MRI (Page 69 to 84) 4. Summary of changes MRI protocol (Page 85 to 86) 5. Statistical Analysis Plan (Page 87 to 88) 6. Final Statistical Analysis Plan (Page 89) 7. Summary of changes Statistical Analysis Plan (Page 90) 1

3 Original research protocol of the randomized controlled clinical trial February

4 Objective(s) of the proposed study The lumbosacral radicular syndrome (LSRS or LRS; also called sciatica) is characterized by radiating pain in an area of the leg served by one or more lumbar and sacral spinal nerve roots. The pain may be combined with other root -tension associated phenomena, such as sensory loss and motor deficit. As a rule LSRS is the result of an herniated disc, but there are also other causes, such as stenosis of the spinal and/or root canal, infections, diabetes, and tumours. This study will focus only on lumbar disc herniations at the levels between the fourth and fifth lumbar vertebrae and between the fifth lumbar and first sacral vertebrae, since they are the most common cause of LSRS in our population. It is estimated that there are between 60,000 and 75,000 new cases of LSRS in the Netherlands each year (1). The presumed direct medical costs of treatment of LSRS are 291 million guilders (2). Most of these costs are attributable to in-hospital treatment; only a small portion are due to general practictioners or physiotherapists (7 million guilders). In that study, performed in 1988, 11,424 patients were operated on for a lumbar disc herniation (2). In ,323 patients underwent surgery and another 23,438 patients were admitted to hospital for at least two days in the Netherlands (3). The total number of operations performed for LSRS each year has not changed recently. The combined direct and indirect costs are estimated to be 2.6 billion guilders per year (4). The natural history of LSRS is in general favourable. In percent of patients, the leg pain decreased or disappeared 6-12 weeks after the first complaints (5,6,7). Most of these patients no longer experienced problems at work or in their private lives after three months. Furthermore, a large percentage of those who continue to experience problems improve considerably after another three months or even after one year. At present it is not possible to identify in an early stage of their disease patients who are likely to suffer significant ongoing problems by means of symptoms, neurological deficit, root tension signs, or diagnostic imaging. For this reason it is not helpful to perform diagnostic imaging (CT or MRI) in an early stage of the disease, unless an other cause than disc herniation is suspected. After the indication for surgery diagnostic imaging is performed in order to define the exact level of disc pathology and the precise anatomic relationship of the disc herniation with respect to the nerve root. Whether particular demographic findings, symptoms, physical signs and/or MRI findings are of prognostic value has not been investigated as yet. It would be of great value if one were able to define early in the course of the disease those patients who will not have this favourable natural course without surgery. This refers to the second research question in the next paragraph. In spite of the known favourable natural course the surgical rate in the Netherlands is quite high. (8). We perform six times as many lumbar discectomies compared to Scotland, four times the number in England and two times the number in Sweden. The United States is the only country in the world where more operations are performed for the indication LSRS. There are no substantial differences in the incidence of this disease in the countries mentioned to explain the different surgical rates. There is no indication (4) that the surgical rate has changed under influence of the consensus reports (9, 10, and 11). In fact change was not likely to occur because the published guidelines were representative for daily practice and normal care before 1996 in the Netherlands. With regard to the indications for and timing of surgery no evidence in the literature is available to contradict these guidelines. These guidelines were produced after agreement between all medical (sub-) disciplines involved in the care for patients with LSRS. Our high surgical rate, as contradictory as it may seem, may reflect good clinical practice. Because of the observation that most people recover from their complaints in the first 6 weeks this period of 6-8 weeks of persistent radicular leg pain is considered a good indication for surgery in the Netherlands. Sometimes patients are operated on within the first six weeks because of severe neurological deficit and/or intolerable pain, with the aim of immediate pain 3

5 relief and conservation or restoration of normal day-to-day activities. The timing of 6-8 weeks of persistent radicular pain before elective surgery is offered is not a case of evidence-based medicine but is a reflection of normal practice. Although there is consensus that surgery is only offered in the case of persistent pain, the timing of this treatment seems to depend on local production capacity and patient and doctor preferences rather than evidence-based practice. This lack of evidence for the timing of surgery after the 6-week period explains the large variations. Knowledge about the problems associated with surgery, such as surgical failure, recidivism and side-effects, is limited. This is one of the reasons that in some regions surgery is only carried out after a period of 3-6 months of LSRS. (12). The main goal of this comparative study is to investigate whether a period of 6-8 weeks of persistent pain is justified as a solid indication for surgery and superior to a longer wait-and-see policy for this favourable self- limiting disease. This study to investigate this scientific gap in our understanding of the effectiveness of surgery for LSRS is in line with a recommendation by the Dutch Health Council in 1999 to the Minister of Health (12) and the Cochrane Review (13, 14). The results of this study will lead to a more rational use of the existing guidelines if the hypothesis is rejected. If the latter is accepted and prolongation of the conservative treatment policy is more cost-effective than surgery after 6 weeks, the existing guidelines for the timing of surgery have to be altered. The results, one way or the other, will provide a solution for this policy problem. Define research questions of the proposed study 1. Is a policy of prolonged conservative care sometimes followed by late surgery more (cost-)effective than conventional surgical intervention after six to twelve weeks for a persisting lumbosacral radicular syndrome? 2. Is it possible to define subgroups of patients who will benefit substantially from one of the proposed treatment strategies? Topic of separate HTA Methodology proposal: Validation of health utility measurements with emphasis on evaluation of temporary health states. Background and state of the art Since the first publication on lumbar disc surgery by Mixter and Barr (15) in 1934 many studies have demonstrated the success of surgery for the treatment of LSRS. It is a pity that only a few prospective studies of the difference between surgical and conservative care are available (5,6,16,17,18,19,20). The published treatment results vary as much as the frequency of reported complications and the recurrence rate. The only study which compared surgery with conservative care directly in a randomized clinical trial (RCT) was performed by Weber and published in (5,6). They found good results for surgery after one year of follow-up. At four years and ten years the results of surgical and conservative care no longer differed. Although it is the only published RCT comparing surgical and conservative care, this study has a lot of methodological flaws in design and outcome measures (21). One of the greatest shortcomings is the exclusion of patients with an indication for surgery because of intolerable pain. Therefore it is not possible to extrapolate the results of this RCT to the treatment policy of today. Since 1983 a few cohort studies have been published on non-surgical treatment of patients with at least six weeks of leg pain with good short-term results at one-year follow-up (23,20). These studies also suffer from methodological flaws. The only conclusion drawn from these reports and the study of Weber is that the policy of prolonged conservative care can be effective, 4

6 because of the favourable natural course of LSRS. Epidemiological and clinical studies have shown that most lumbar disc protrusions resolve naturally with the passage of time and conservative management without surgery (13, 14). Another fact is that prolonged conservative care is safe and without complications if the patients remain active. For a review of the literature,see Part 5. Preliminary studies by applicants on the subject of this proposal Prof. Koes, who is in charge of the epidemiological aspects of this study, participated in the commission for the development of guidelines for General Practice (9) and the CBO consensus commission (10,11), both for the diagnosis and treatment of the lumbosacral radicular syndrome. Furthermore Prof. Koes, together with his former colleagues of the EMGO Institute, performed research studies on disorders of the vertebral column, especially the lumbar spine. Prof. Kievit is a member of the Central Committee of the Health Council of the Netherlands and deals with Medical Technology Assessment (MTA), which is the evaluation of the topics of the list of 126 in relation to cost-effectiveness. The treatment of LSRS was one of these topics. Prof. Koes and Dr. Peul, the Principal Investigator, were members of the MTA subcommittee, which was established to review and investigate the scientific evidence and gaps underlying the diagnosis and treatment of LSRS caused by a disc herniation (12). Prof.Thomeer wrote his PhD Thesis on the surgical aspects of the lumbar spine in achondroplasia and also conducted a prospective study on lumbar fusion surgery. This study recently resulted in a dissertation on the innervation of lumbar discs and the surgical treatment of low back pain. 5

7 Study design (motivate) To answer the main research question the investigators propose to conduct a multi-centre comparative randomized clinical trial with parallel group design. The follow-up will last one year. The results of a randomized trial are less difficult to implement because of the strength of evidence when compared with a cohort study. The multi-centre design is necessary to collect enough patients in one year. The cost-effectiveness results will be a trade-off between a quicker relief of leg pain in the surgery group versus the advantage of lower costs and avoiding the negative effects of surgery in the conservatively treated group. The difference in quality of life depends on the duration of persisting pain after randomization in the last group. To answer the first question the short-term results are important and will not be affected by the results at two years. For this NWO study a grant is requested for the first year of follow-up. The intended follow-up at two years, which is necessary for a surgical study, will be conducted at our own expense. To answer the second question explorative subgroup analysis will be conducted. For a description of these subgroups, see the paragraph Data analysis and presentation below. Selection criteria for study population / data sources / search strategy All patients (18-65 yr.) with a lumbosacral radicular syndrome with a duration of less than 12 weeks are eligible for this study. Because of the multi-centre design the patients in and around The Hague, Delft, Gouda, Haarlem, Heemstede- Aerdenhout, Hoofddorp, Alphen a/d Rijn, Leiderdorp and Leiden can be included in this trial if they meet the inclusion and exclusion criteria (see below). These patients will reflect a normal population treated in primary and secondary care. Comparison of baseline-variables will be made with the population of patients investigated in the parallel HTA-study. Those patients are a truly representative sample of usual care. Inclusion of patients will be started after a visit to the neurological outpatient clinics. Randomization will start after 6 weeks persistent pain in the dermatome of the leg served by the L4, L5 or S1 root. The general practitioners in these areas will be informed about this study and receive information about developments and the results of the trial. The general practitioners will be asked to refer patients within the first 5 weeks of the start of leg pain. Communication with the large number of GP s and physiotherapists in the regions of this study is satisfactory. Coordination of information for the former will be facilitated by the General Practitioner Department of Leiden University. The physiotherapists will be informed with the help of the Department of Physiotherapy of the LUMC. Inclusion criteria; age yr., persistent radicular pain in the L4, L5 or S1 dermatome with or without mild neurological deficit, intensity of the pain in the leg more than 40 mm measured on the 100 mm horizontal VAS, leg pain of 6-12 weeks duration, evidence of a unilateral disc herniation confirmed on MRI, informed consent. Exclusion criteria; cauda equina syndrome or severe paresis (MRC<3), complaints of a lumbosacral radicular syndrome in the same dermatome within the past 12 months, low back surgery in the past, significant central canal stenosis, degenerative or lytic spondylolisthesis, pregnancy, severe life-threatening or psychiatric illness, insufficient knowledge of Dutch language, planned removal to another country in the year after randomization, contra-indications for MRI. Description of the intervention(s) (including follow-up) / selection of information / data extraction The GP s will receive detailed information about the study and will be asked to refer their patients after 4-5 weeks of persisting pain. Patients who do not want to be referred or come to the GP for their first visit >12 weeks after the start of the leg pain are not eligible. 6

8 During the first visit to the neurological outpatient clinic the patient s history will be taken and the neurologist will perform a standardized neurological examination. During this visit the neurologist will inform the patient that he/she has a lumbosacral radicular syndrome and the doubt regarding the timing of surgery for this condition. The study will be explained to the patiënt and in case of a positive reaction an appointment is made to meet the research nurse as soon as possible after a scheduled MRI scan. After this visit the neurologist will write a letter to the GP and neurosurgeon to inform them about the clinical status. He will see the patient for the second time 26 weeks after randomization. Preferably the MRI scan will be performed before the visit to the research nurse. This nurse will give all information needed to understand the trial and will ask the patient if he/she agrees to be randomized. The research nurse, informed by the radiologist, will only ask for informed consent if the MRI confirms the presence of unilateral disc herniation and the patient is eligible according to the inclusion and exclusion criteria (checked by the neurosurgeon). The patient will not be aware of detailed MRI data. The MRI data will be registered by the radiologist and neurosurgeon using a standardized Case Record Form (CRF). The patient will decide at home and will be asked to inform the outpatient clinic within 72 hours. A new appointment will be made with the research nurse after a positive decision to participate. Non-participating patients will be registered at the data coordination centre and given an appointment with the neurologist. The research nurse will receive the written informed consent form, will again perform a standard neurological examination and will fill out the form needed for evaluation of outcome measures and baseline variables. Preferences of patients and doctors will be registered. This will be done by using a 5-point scale varying from strong preference for surgery to strong preference for conservative treatment. After registration of the baseline variables, the research nurse will open the envelope in front of the patient. Randomization will take place in permutation blocks of 4 per hospital (treatment coordination centre). After randomization two groups of patients will exist. Group A; the surgically treated patients and group B; the conservatively managed patients. Surgical treatment (A) will be performed in the conventional manner with loupe magnification. The investigators prefer the standard surgical approach because the other (minimally invasive) surgical approaches have limited indications, are not more cost-effective, and have a long learning curve. The 10 participating surgeons are certified neurosurgeons with experience in the standard approach with loupe magnification. Surgery will take place as soon as possible with a maximum of two weeks after randomization. Hospital admission will be 4-6 days, including the day of surgery. During the post-operative period the patients will be mobilised with the help of a physiotherapist. At home the patients will visit their own physiotherapist who will guide them and provide exercises. The frequency will be 2 times a week for 6 weeks. After their last visit to the outpatient clinic, 6-8 weeks post-operatively, patients with complaints of persisting leg pain or aggravated back pain will need some extra physiotherapy sessions, prescribed by the neurosurgeon. If possible most patients will gradually start back to work. Conservative management (B) will be a longer wait-and-see policy conducted by the general practitioner (GP). The GP will provide information about the favourable prognosis of LSRS. The treatment of LSRS is aimed primarily at pain relief and maintenance/restoration of normal dayto-day activities. In view of the fact that, despite the frequently worrying nature of the condition, the prognosis for LSRS is generally quite good, great importance is attached to the provision of information. Unfortunately, the effect of giving information and counselling has not been studied specifically among LSRS patients. However, various studies have evaluated the effect of such support for people suffering from other pain syndromes (24). Inferences can reasonably be made from the findings of these studies. Hence, it may be assumed that adequate and 7

9 unambiguous information about what is wrong (the nature of the condition) and what the patient can expect (the prognosis), together with trustworthy counselling can reduce the anxiety and uncertainty felt by the patients and thus ease the pain (12). The GP s will encourage the patients to continue with normal day-to-day activities in so far as possible. If necessary analgesic medication can be prescribed according to the guidelines of the NHG. The GP will advise the patients to stay active and if possible return to work and/or their leisure activities. Graded activity counselling is not based on the pain experienced and therefore is not hampered by this factor. Instead it is graded according to a time schedule. This time schedule is made beforehand together with the patient (25, 26). Information about this graded activity will also be provided in written form. After the first consultation the GP will make a follow-up schedule. During the next visit the patient and doctor will look at the changes since the first visit to determine whether there is any improvement in the ability to perform normal activities. The doctor will check the efficacy of the prescribed pain medication and may adjust the dose or sort of analgesics according to the NHG guidelines. In these guidelines paracetamol is the first choice. If not effective, NSAID s (ibuprofen, diclofenac or naproxen) are to be prescribed. Only in the event of severe disabling pain can morphine be given for a restricted period of time. By preference all analgesics should be taken at fixed times of the day rather than on an if necessary basis. If the GP and the patient conclude that there is considerable kinesiophobia because of the fear that the radicular or low back pain will increase, the help of a physiotherapist can be recommended. Guided by the GP (and physiotherapist) the patient will upgrade his or her activities according to the agreed time schedule. The guide will be time, not the intensity of the pain. The GP will be free in her/his choice of prescription of medication and referral to physiotherapists. The research nurse will register the conservative management strategy after communication with the responsible GP. In case of progressive neurological deficit or worsening intolerable pain the GP can refer the patient back to the research nurse or neurosurgeon. If, six months after randomization, the patient has still not improved or suffers from intermittent LSRS, surgical treatment will be offered. Some patients will ask for surgery earlier because of worsening leg pain. In these cross-over cases and in the case of a progressive neurological deficit, surgery will be performed in consultation with the patient. The research nurses will inform the GP s about the result of randomization. Follow-up examinations by the research nurse will take place 4, 8, 12, 26, 38 and 52 weeks after randomization and the patients will keep a diary. The patients will have an appointment with his/her neurologist 26 weeks after randomization. The neurologist will not be aware of the MRI data and the treatment strategy received. He/she will perform a standard neurological examination and ask some standardized questions, without running the risk that the patient discloses the perceived management strategy. The patient will receive instructions to keep the neurologist unaware of the treatment. The neurologist fills out the PROLO scale and the same 7-point Likert scale. It is only possible to keep the MRI data from the neurologist, GP and patients. For purposes of organisation it is not possible to blind the research nurses and the Principal Investigator. Data collection will take place per treatment coordination centre by the research nurse and will be recorded and analysed at the Data Coordination Centre in Leiden. Primary and secondary outcome parameters (theoretical and operational) The primary and secondary outcome measures will be registered at baseline and during followup of 4, 8, 12, 26, 38, 52 weeks after assignment of the treatment. In the LSRS the most common complaints are pain and disability to perform normal daily activities. We will use 8

10 validated outcome parameters, which will be assessed by means of questionnaires. The blind neurologist will see the patients 26 weeks post-randomization to register the Perceived Recovery and PROLO scale. Primary outcome measure: Roland Disability Questionnaire for Sciatica. This illness-specific 23-item functional assessment questionnaire is frequently used for low back pain and sciatica (38,39). Scores range from 0 to 23, reflecting a simple unweighted sums of items endorsed by the respondent. Patients with high scores at baseline do have a severe disabling LSRS. To define recovery a difference of at least 11 points from baseline has to be seen (38, 17) The Roland Questionnaire for Sciatica has a documented high level of internal consistency, construct validity, and responsiveness (38, 39). Secondary outcome measures: Perceived recovery. This is a seven-point Likert scale measuring the perceived recovery, varying from completely recovered to worse than ever. This outcome scale has been used in previous studies and appears to be valid and responsive to change (27). VAS pain in the leg. This parameter will measure the experienced intensity of pain in the leg during the week before visiting the research nurse. Pain will be assessed on a horizontal 100 mm scale varying from 0 mm, no pain in the leg, to 100 mm, the worst pain ever. Patients do not see the results of earlier assessments and will score the pain experienced at the visit. (28,29,30,31,32). Short-Form 36 (SF-36). This is a generic health status questionnaire, which can easily be filled out at home. The questionnaire has 8 topics; 1) physical functioning, 2) physical restrictions, 3) emotional restrictions, 4) social functioning, 5) somatic pain, 6) general mental health, 7) vitality, 8) general health experience. The answers to the questions result in a score varying from 0 unhealthy to 100 optimal health. This questionnaire has been used frequently and was validated in studies on low back pathology and surgery. (33,34,35,36,37). Sciatica Frequency and Bothersome Index. This is a scale from 0 to 6 which can assess the frequency (0=not at all to 6=always) and bothersomeness (0=not bothersome to 6=extreme bothersome) of back and leg symptoms. The sum of the results of four symptom questions yields both indexes, ranging from 0 to 24: leg pain; numbness and/or tingling in the leg; weakness in the leg or foot; pain in the back or leg while sitting. (17). PROLO-scale. This scale measures the evaluation of the research nurse and neurologist of the functional-economic status of the patients. This parameter has been used in studies on the difference in functional outcome between different techniques of lumbar spine fusion. (40) VAS pain in the back. This parameter measures the intensity of the pain in the back experienced during the week before visiting the research nurse. Assessment will be based on a horizontal 100 mm scale varying from 0 mm, no pain in the back, to 100 mm, the worst pain ever. Patients do not see the results of earlier assessments and will score their pain during the visit. This parameter is included because a lot of patients with LSRS also have back pain in varying intensities, which can become worse after surgery. Incidence of (re-)surgery. One of the goals of the policy for group B is to avoid surgery while achieving at least the same effects. The disc surgical rate is therefore an indication of the success or failure of this policy. The incidence of re-operation at the same disc level in group A will be an indication of the failure rate for surgery. MRI findings. The results of the differences between the baseline MRI and the MRI s made 26 and 52 weeks after randomization are important secondary outcome measures. The difference in size of the disc herniation (in mm), nerve root compression, and amount of scar tissue will be registered. Failures of surgery can be recognized by inadequate disc removal or decompression of the nerve.the data will be gathered, using a standardized CRF which will be filled out by the local radiologist, neurosurgeon and neuroradiologist. 9

11 EuroQol and VAS-Health-Status. A cost-utility analysis will be performed using QALY s based on the EuroQol questionnaire which has been validated in many studies and is easy to fill out (41,42, 51). The EuroQol will be measured twice a week during the first four weeks an at all follow-up moments. In this questionnaire a VAS Health will be measured (0 dead to 1 complete healthy ). This outcome measure has not yet been validated. The perceived health VAS can be transformed into utilities with the formula U=1-(1-VAS). Sample size calculation (motivate assumptions) and feasibility of recruitment This randomized clinical trial will investigate the difference between surgery after 6-12 weeks of persisting LSRS and a longer wait-and-see policy in a parallel group design. The result of this study is based on the short-term success of surgical intervention and will be a trade-off between a quicker relief of leg pain versus an advantage in cost-effectiveness for conservatively managed patients. The sample size is calculated on the basis of the Roland Disability Questionnaire for Sciatica during the 12 months follow-up period. The numbers used for this sample-size are drawn from the Maine Lumbar Spine Study 1 year and recently published 5 year results (17, 53). The difference in the Roland score between the surgical- and non-surgical group in this study did not change between 3 and 12 months followup as shown in their study (17; fig. 2 A) and can be averaged over the first year. Surgical treatment is justified when the post treatment change is at least 4 points more when compared to the conservative treatment arm (38) and constant over time. A minimal change from baseline of 11 points (incomplete recovery) is required (38). Considering this constant difference and a mean standard deviation=10 (53; table 2) 140 patients per treatment arm are needed to reach a power of 90 % (=0.05). A total of 300 (150 per arm) patients will be enrolled in the study, including 8 % loss to follow-up after 1 year (17). In the 8 participating regional hospitals at least 100 patients every year are operated on for a lumbar disc herniation. In the Westeinde, Leyenburg, Groene Hart, Reinier de Graaf, Rijnland and Spaarne hospitals at least 200 patients each year undergo lumbar disc surgery because of LSRS. The investigators have excluded the LUMC. For this study patients will also be recruited from this University Hospital, but because of the academic status only a minority of the patients will be operated on in the LUMC. In this study the LUMC will function as the data coordination centre. The total number of patients each year in all participating hospitals exceeds Because of this study GP s will refer patients sooner than they usually do, which will result in an increase in the total number of referred patients when compared with the normal situation. The favourable natural course in the period following the consensus period of 6-8 weeks is one of the reasons why experienced GP s and neurologists delay the indication for surgery. The preference of the doctors and patients is an important factor. Patients who nowadays would not be referred because of spontaneous recovery in the first months will be potential candidates for recruitment in this study, which explains the expected rise in the total number of patients with LSRS at the outpatient clinic. Even without taking this fact into consideration it is feasible to enroll the total number of 300 patients in one year. 10

12 Data analysis and presentation / synthesis The answer to the main research question will be the results of the data analysis based on the Intention to Treat principle. The efficiency of the intervention will be assessed using mixed model analysis of variance. In addition to an analysis of the difference in recovery between the two groups (as explained under the paragraph Sample Size Calculation) analyses of the difference in time to recovery will be carried out. Due to lack of data in the literature we could not base our sample size calculations on these differences. To answer the second research question explorative subgroup analysis will be performed. This analysis is not intended as a thorough study to test the hypotheses described below. To test these we need a larger sample of patients that are not needed for the main research question. This is a longitudinal study with repeated measurement analysis of variance to assess the relationship between prognostic variables and recovery of patients in both treatment arms. For the complete analysis demographic, anamnestic, physical, and radiological findings will be used. Subgroups will be composed of the following selected variables (7,17,43,44): age; level of education; a long history of low back pain; intensity of leg pain; Lasegue s sign; ratio leg pain/back pain intensity; provocation of leg pain while sitting; the size of the disc herniation measured by MRI. To complete this analysis the investigator will register the preference for group A or B of both patients and doctors. This will be done by using a 5-point scale varying from strong prefence for surgery to strong preference for conservative management. Anxiety for chronic pain is an important factor in the treatment preference.the registration of anxiety (McGill Pain Questionnaire) and patient and doctor preference will take place just before randomization is performed and during follow-up. The following subgroups and hypotheses will be investigated in this trial: Highly educated young patients (18-40 yr.) with an intense LSRS (VAS > 70 mm) and without a history of back pain fare better with a conservative management strategy. Circumferential gadolinium enhancement of a large disc sequester (MRI) is a predictor of fast recovery without surgery. Patients with selective gadolinium enhancement of a unilateral nerve root (MRI), as a sign of root compression, fare better with a surgical treatment strategy. Older patients (40-65 yr.) with an intermediate intensity of leg pain (VAS mm), a small contained disc herniation and some lateral recess stenosis (MRI) profit from surgical treatment only. Patients with increased leg pain while sitting and paraesthesia/ dysaesthesia in the leg dermatome fare better with a surgical treatment strategy. Patients with LSRS, caused by a large sequestered disc fragment (MRI), fare better with a conservative management strategy. Patients with a positive Lasegue s sign of less than 30 degrees and without back pain fare better with a surgical treatment strategy. Median disc herniations (MRI) fare better with a conservative management strategy. Patients with a relatively high interbody space (MRI) at the level of the disc herniation fare better with a conservative management strategy. 11

13 Economic evaluation: general considerations Conservative treatment may decrease costs compared to surgery but possibly at the expense of delayed effectiveness. In an incremental cost-effectiveness analysis, societal costs during the first year will be compared to the primary outcome measure (Roland Disability Questionnaire for Sciatica, averaged over the first year), Quality of Life (SF-36, during the first year) and perceived recovery (7-points Likert scale). Cost-effectiveness analyses with these effectiveness measures have been conducted before, allowing comparison with other spine interventions. For comparison with a wider range of interventions an incremental cost-utility analysis will be conducted, comparing the same societal costs during the first year to the societal utility obtained from the descriptive classification system of the EuroQol (QALY during the first year). An attempt will be made to model differences in effectiveness and costs beyond the time horizon of the study itself, by extrapolating the rates of surgery, physiotherapy and absenteeism. Here, time horizons of 2 and 5 years will be used with discounting of both costs and effectiveness by 4%. Sensitivity analyses will be performed of the costs of hospital admission for surgical treatment, the costs of physiotherapy, percentage employment and the utility measure (EuroQol versus transformed VAS). Economic evaluation: cost analysis The societal costs during the first year will be estimated in accordance with the recent pharmacoeconomic guideline (45). The costs of hospital admission and surgery will be based on an integral top-down cost analysis in the Westeinde, Spaarne and Rijnland hospitals (aggregated according to the total number of patients per department). For this institutional analysis, the fixed costs per admission and the variable costs per admission day will be estimated. From these fixed and variable costs, the individual costs of hospital admission and surgery for all patients can be estimated from the duration of the hospitalisation. In the study an MRI is performed in all cases. The costs of this MRI will only be calculated for patients undergoing surgery, because in the normal situation MRI would only be performed when a surgical indication exists. Other health care needs will be registered by the patients in a diary (including physiotherapy, visits to GP s and specialists, nursing care and medication). Each diary covers a period of 3 months and will be discussed with the patient during the follow-up visits to the research nurse. The volume of health care will be assessed using standard prices (46). In the diary the patient will also register direct non-medical costs (including time costs, travel expenses and domestic help). To estimate productivity costs the patients will also report absenteeism in the diary. At the follow-up visits, the research nurse will register the work situation, work efficiency and bruto fare. Absenteeism will be assessed according to the frictioncost method. Economic evaluation: patient outcome analysis The result of this study will be a trade off between the disadvantages of surgery (hospitalisation, reduced quality of life and costs) versus the possible advantage of earlier relief of pain. For that reason recovery, measured as a difference in score when compared to baseline (Roland Disability Questionnaire for Sciatica), is the clinically most relevant patient outcome. Quality of Life (SF-36) and perceived recovery are important to compare the reduced quality of life from surgery to the possibly prolonged pain from conservative therapy and also to be able to compare cost-effectiveness with that of other spine interventions. The EuroQol is important to obtain cost-effectiveness ratio's that can be compared with those of a wide range of other interventions. Utilities are obtained from the descriptive classification system of the EuroQol, using the model described by Dolan (41,51). The EuroQol is used because it provides societal 12

14 utilities. Furthermore, the patients can fill out the EuroQol at home without the help of the research nurse, and filling it out under changing conditions poses no difficulties. To allow for a utility analysis from the patient perspective, the VAS of the EuroQoL will be used, since this is more feasible than the TTO for temporary states. VAS scores will be transformed into TTO scores based on the formula proposed in the literature (50). Outcome measures are evaluated at baseline and at 4, 8, 12, 26, 39 and 52 weeks. 13

15 Optional parallel study HTA-methodology: outcome measurement methodology Describe the methodological problem that is underlying the research proposal: In the Sciatica trial temporary states need to be evaluated. To evaluate such temporary states using the TTO (the generally accepted utility instrument (47)) a chained procedure must be followed. This has rarely been done, given the complicated nature of the procedure. It has been shown to be feasible in oncology (49), but a careful selection of the anchor state is required. For CEAs, the use of a health state classification system is recommended. In Europe, the EuroQol is generally used. This will also be the case in the Sciatica trial. In the EuroQoL, the descriptive score obtained from patients is transformed into a TTO utility for the general public by means of a formula which has been published in the literature (41). This formula (or tariff ) has been developed using a time period of 10 years. The members of the general public whose data have been used to construct the formula, were instructed that each health state (defined by the five dimensions of the EuroQol) was to be regarded as lasting for 10 years, followed by death. However, it has been shown that the duration of a state has an impact on the valuation of that state (in healthy subjects the value is negatively correlated with the duration). Therefore, Dolan (51) has proposed an adapted tariff for all EuroQol states, based on different durations (1 month and 1 year). This tariff is limited in several ways. The state to follow these two durations was specified to the respondents as unknown (and not as dead, as was the case in the above mentioned study (41) that used the 10-year period). Moreover, the author stresses that his tariff only applies for the VAS, and more research is needed on choice-based methods, such as the Standard Gamble and the TTO. To our knowledge, CEAs that use the EuroQol never discuss this issue: it is assumed that the utilities based on a 10-year time period apply to the health states under analysis, even if these states are either chronic states with a duration that far exceeds 10 years, e.g. as in the case of the life expectancy of young subjects, or temporary states such as in the case of LSRS. We therefore believe that a more in-depth investigation is needed than was done by Dolan (51). We will assess these duration effects, and determine whether the results are similar when the 10 year period is framed as temporary, not chronic, i.e. not followed by death (a validation of the proposed adapted tariff). Further we wish to study whether the duration effects in the TTO are similar to those in the VAS. The Sciatica trial provides a unique opportunity to study these issues, due to the suitability of the clinical subject (the health states) and the combination of expertise in the Departments of Neurosurgery and Medical Decision Making. Objectives: First, we wish to study these duration effect using the TTO, in which durations are more salient than in the VAS. Second, we want to study these effects not only in healthy subjects (as in Dolan s study) but also in patients, who will be subject to adaptation. As a result the effects for the latter are expected to be different from perhaps even contrary to- those for the general public. The results of this study will lead to a more evidence-based choice of measurement methods in future studies assessing temporary states and will provide guidance in the use of the EuroQol in such studies. Research questions: 1. Is the utility of a 6-month temporary state different from the utility of that state when it lasts 10 years, the latter still considered as a temporary state, followed by an unknown health state (as in the adapted tariff (51)). 2. Is the utility of a 6-month temporary state similar to the utility of that state when it lasts 10 years, but now considered as a chronic state, followed by death (as was the case in the original tariff (41)). 3. Are duration effects, if found, different for VAS than for TTO? 4. Are duration effects, if found, different for patients and healthy subjects? Study design/procedures: A cross-sectional preference study will be appended to the clinical study. Subjects will be LSRS patients and healthy subjects. Patients will be interviewed individually, healthy subjects will be interviewed in a classroom setting, which according to Stolk et al. (52) is feasible. 14

16 Two representative states (mild, and severe LSRS) will be obtained from the descriptive classification system of the EuroQol, based on a pilot study in this patient population. The valuation is limited to two states for reason of respondent burden. The valuation of more states is not deemed feasible, given the complexity of the TTO for temporary states. The states will be labeled as LSRS states, and described in scenarios using the five dimensions of the EuroQoL descriptive system. Temporary states will be followed by an unknown health state, chronic states by death. Both states (mild and severe LSRS) will thus be presented in three durations: 6-month temporary, 10-year temporary, and 10-year chronic. The order in which the mild and severe states are presented to the subjects will be at random, as will be the order of the (one) chronic vs. (two) temporary states, and the order of the 6-month and 10-year periods (the two temporary states will always follow one another, since the chained method is complicated and thus needs explanation only once). The subjects will value the first selected state by the VAS, since this has been shown to provide a good introduction to the more complicated TTO. Next they will proceed with the TTO and the VAS for that same state. This procedure is repeated for the remaining five states. Patients: Patients will be accrued from the hospitals of the trial but will not participate in the trial, in order to decrease the burden on the participating patients. This is possible since the pool of potential patients is much larger than that needed for the trial. Healthy subjects will be recruited through regional community registries. Based on data from Jansen et al. (48) a sample of 125 patients and 125 healthy subjects would be needed to enable a 0.05 difference in valuations between the two groups to be detected (alpha=0.05, 1-beta=0.80, s=0.14, s D =0.14). The sample size needed for the paired analyses differences between durations- is 125. The sample sizes are based on the between group differences, since the largest sample is needed for this comparison. This leaves us more than 80% power for the paired analyses (durations and methods). Since methods and durations are correlated, our calculations will be conservative. We did not wish to speculate on the exact value of the correlations, and therefore did not adjust our estimates downward. Our main interest is the duration effect (differences between 6-month and 10- year temporary states; and differences between chronic and temporary versions of the 10-year state). These effects are all assessed using within-subjects analyses. Analysis: Differences between methods and between durations will be tested by paired tests/manovas (if necessary, incorporating a factor for between-group differences), differences between groups (patients, healthy subjects) will be tested by unpaired tests/anovas, Since we base our power calculations on the between group effects (see under 1), but our main interest is the within group effects (which needs smaller numbers), we believe that we will we be able to answer the research questions with the proposed research. Time schedule Clinical Study Month 1-6: During the first 6 months preparations will be made. The 4 research nurses will be recruited and installed in the different hospitals. Each research nurse will coordinate the intake and follow-up in two hospitals in close contact with the neurologist-coordinator and the neurosurgeon of the participating hospital. The Principal Investigator will also introduce them to the different departments of the hospital, such as the operating room, the inpatient and outpatient clinics, radiology and physiotherapy. During the same period the project manager will install the software for the database at the LUMC in conjunction with the Department of Statistics. The data forms, patient information forms, outcome measurement forms and questionnaires will be developed and printed. The Principal Investigator will again inform all participating neurologists and GP s. The Principal Investigator, together with the GP coordinator of the university, will contact the latter group. The physiotherapists in the different regions will be informed by mail about the trial and asked to register their treatment in the diary of the participating patients, if any. After the first three months the data coordination centre will start functioning at the LUMC and the treatment 15

17 coordination centres in the participating hospitals. Month 7-18: During this twelve-month period, intake, MRI, randomization and treatment will take place. The Principal Investigator together with the project manager and GP coordinator will be responsible for coordination during this period and the overlapping follow-up. The econometrist will measure costs in three regional hospitals. Month 8-30: The follow-up period with data collection according to schedule will last twelve months post-randomization. Patients who were included early in the study will have their followup visits at 78 and 104 weeks post-randomization during the NWO study period. This is necessary because of the international agreement on a minimal two-year follow-up period for a surgical spine study. No grant is asked for these visits, which will be scheduled until 24 months (month 42) after the last patient is enrolled (month 18). The research nurse s activities will decrease after the intake and treatment period. Two of the research nurses will stop after month 18 and the other 2 after month 30. Primary data analysis can be started and a synthesis performed, based on baseline variables and extracted short-term follow-up parameters. Month 31-36: During the last six months final data analysis and synthesis can be performed to answer the main research question. The results on cost-effectiveness can be reported to NWO and the different national scientific medical societies. If necessary, depending on the results an implementation policy and schedule can be discussed with CBO and the Dutch Health Council. Presentation of the final (cost-) effectiveness results at an international forum cannot take place until after the final follow-up visit of the last patient (month 42). Research focus and selected publications (max 5) of the (co)- investigators -Poppel MNM van, Koes BW, Ploeg GE van der, Smid T, Bouter LM. A randomized controlled trial of lumbar supports and education for the prevention of low back pain in industry. JAMA 1998; 279: Koes BW, Bouter LM, Mameren H van, et al. Randomised clinical trial of manual therapy and physiotherapy for persistent back and neck complaint: results of one year follow-up. Br.Med. J 1992;304: MTA-committee including Koes BW, Peul WC, Kievit J. Health Council of the Netherlands: Management of the lumbosacral radicular syndrome (sciatica). No. 1999/18 -Coppes MH, promotores Thomeer RTWM, and Mooij J.J. Lumbar Spondylodesis Revisited. A prospective surgical study. Dissertation Peul WC, Voormolen JHC. Hoe groot is de kans op een recidief na een hernia-nuclei-pulposi operatie en wat is de beste manier om dit te voorkomen? Neurologen Vademecum 1999/3. References 1)-Van de Velden J, de Bakker DH. Basisrapport: Morbiditeit in de huisartsenpraktijk. Utrecht Nivel )-Van den Bosch JH, Kardaun JWPF. Ziekten van het zenuwstelsel in Nederland. Den Haag: SDU/uitgeverij )-SIG Zorginformatie, Utrecht Gegevens uit de landelijke registratie van Nederlandse Ziekenhuizen. 4)-van Tulder, Koes BW, Bouter LM. A cost-of-illness study of back pain in the netherland. Pain 1995;62: )-Weber H.Lumbar Disc Herniation. A controlled prospective study with ten years of observation. Spine 1983; 8: )-Weber H.Lumbar Disc Herniation. A prospective study of prognostic factors including a controlled trial. J. of Oslo City Hospital. 1978;28: )-Vroomen PCAJ. The diagnosis and conservative treatment of Sciatica. Thesis Maastricht )-Cherkin DC, Deyo RA. An international comparison of back surgery rates. Spine 1994;19: )-Smeele IJM, van den Hooghen JMM, Mens JMA. e.a. NHG-standaard Lumbosacraal Radiculair Syndroom. Huisarts Wet 1996; 39: )-Stam J. Consensus over diagnostiek en behandeling van het lumbosacraal radiculair syndroom. Ned. Tijdschr. Geneesk. 1996; 140; )-CBO Consensus Het lumbosacraal radiculair syndroom )-Gezondheidsraad 1999/18. Diagnostiek en behandeling van het lumbosacraal radiculair syndroom. 13)-Gibson JNA, Grant IC, Waddell G. The Cochrane Review of Surgery for Lumbar Disc Prolapse and Degenerative Lumbar Spondylosis. Spine 1999; 24 : )-Gibson JNA, Grant IC, Waddell G. Surgery for lumbar disc prolapse (Cochrane Review). In: Cochrane Library, Issue 4, 2000 Oxford: Update Software 16

18 15)-Mixter WJ, Barr J. Rupture of the intervertebral disc with involvement of the spinal canal. N.Engl. J.Med. 1934;211: )-Spanfort EV. Lumbar disc herniation: a computer aided analysis of 2504 operations. Acra Orthopaedica Scandinavica 1972;suppl: )-Atlas SJ, Deyo RA. The Maine Lumbar Spine Study, part II. 1-Year Outcomes of Surgical and Nonsurgical Management of Sciatica. Spine 1996; 15: )-Hakelius A. Prognosis in Sciatica: A clinical follow-up of surgical and nonsurgical treatment. Acta Orthop Scand. 1970; Suppl 129: )-Malter A.D., Deyo RA. 5-Year Reoperation Rates After Different Types of Lumbar Spine Surgery. Spine 1998;23: )-Saal JA. Natural history and non-operative treatmenet of lumbar disc herniation. Spine 1996; Suppl. 24 S: 2S-9S. 21)-Bessette L, Liang MH. Classics in Spine. Surgery literature revisited. Spine 1996;21: )-Malter A.D., Deyo R.A.. Cost-Effectiveness of lumbar discectomy for the treatment of herniated intervertebral disc. Spine 1996;21: )-Saal JA. Saal JS The natural history of lumbar intervertebral disc extrusions treated non-operatively.spine 1989;15: )-Gatchel, Turk DC,. Psychological approaches to painmanagement: a practitioners handbook. The Guilford Press, )-Main CJ, Wood PLR, Hollis S. The distress and risk assesment method. Spine 1992; 17: )-Fordyce WE. Behavioral methods for chronic pain and illness. St. Louis: CV Mosby Company, )-Deyo RA. Outcome measures for low back pain research. A proposal for standardized use. Spine 1998:23; )-Scott J., Huskisson EC. Graphic representation of pain. Pain 1976;2: )-Huskisson. Measurement of pain. Lancet ii (1974) )-Joyce CRB, Zutshi DW. Comparison of fixed interval and Visual Analogue Scales for rating Chronic Pain. Eur. J. Clin. Pharmacol. 1975;8: )-Melzack R, Katz J. Pain measurement in persons in pain. Textbook of Pain. 32)-Collins SL, Moore A. The visual analogue pain intensity scale: what is moderate pain in millimetres? Pain 1997;72: )-Ware JE, Sherbourne C.The MOS 36-item short-form survey(sf 36): Conceptual framework and item selection. Med.Care 1992;30: )-van der Zee K, Sanderman R. De psychometrische kwaliteiten van de MOS 36-item Short Form Health Survey (SF-36) in een Nederlandse populatie. T. Soc. Gezondheidsz. 1993; 71: )-Brazier JE, Harper R. Validating the SF-36 health survey questionnaire. BMJ 1992;305: )-Aaronson NK, Acquadro C. International quality of life assesment (IQOLA) project. Q. of Life Research.1992; 1: )-Stansfeld SA, Roberts R. Assessing the validity of the SF-36 General Health Survey. Q. of Life Research 1997;6: )-Patrick DL, Deyo RA. Assessing health related quality of life in patients with sciatica. Spine 1995;20: )-Gommans I, Koes BW. Validity and responsiveness of the dutch adaptation of the Roland Disability Questionnaire. Low Back Pain. Editors: Tulder MW, Koes BW, Bouter LM. EMGO 1996; )-Prolo DJ, Oklund SA. Toward uniformity in evaluating results of lumbar spine operations. Spine 1986; 11: )-Dolan PP. Modeling valuations for EuroQol health states. Med. Care 1997; 35: )-EuroQol Group: A new facility for the measurement of health-related quality of life. Health Policy 1990;16: )-Matsubara Y. Serial changes on MRI in lumbar disc herniations treated conservatively. Neuroradiology 1995; 37: )-Bozzao A,Galluci M.MRI imaging assessment of natural history in patiënts treated without surgery. Radiology 1992;185: )-College voor zorgverzekeringen. Richtlijnen voor farmaco-economisch onderzoek. CvZ, afdeling FO/G&s, )-Oostenbrink JB, Koopmanschap MA, Rutten FFH. Handleiding voor kostenonderzoek, methoden en richtlijnprijzen voor economische evaluaties in de gezondheidszorg. CvZ )-Stiggelbout AM, De Haes JCJM. Patient preference for cancer therapy: an overview of measurement approaches. J Clin Oncol (in press) 48)-Jansen SJT, Stiggelbout AM, Wakker PP, Nooij MA, Noordijk EM, Kievit J. Unstable preferences: a shift in valuation or an effect of the elicitation procedure? Med Decis Making 2000;20: )-Jansen SJT, Stiggelbout AM, Wakker PP, Vliet Vlieland TPM, Leer JWH, Nooy MA, Kievit J. Patients utilities for cancer treatments: a study of the chained procedure for the standard gamble and time tradeoff. Med Decis Making 1998; 18: )-Stiggelbout AM, Eijkemans MJC, Kiebert GM, Kievit J, Leer JWH, De Haes JCJM. The "utility" of the Visual Analog Scale in medical decision making and technology assessment: is it an alternative to the Time Trade-Off? Int J Technol Assessm Health Care 1996;12: )- Dolan P. Modelling valuations for health states: the effect of duration. Health Policy 1996;38: )- Stolk EA, Busschbach JJ, Caffa M, Meuleman EJ, Rutten FF. Cost utility analysis of sildenafil compared with papaverinephentolamine injections. BMJ 2000; 29; 320: )- Atlas SJ, Keller RB, Deyo RA. Surgical and non-surgical Management of Sciatica Secondary to a Lumbar Disc Herniation. Five-Year Outcomes from the Maine Lumbar Spine Study. Spine ; 26(10): Disease / condition, subject of this proposal The lumbosacral radicular syndrome caused by a lumbar disc herniation. (Sub-) group of patients, targeted in this proposal Patients (18-65 yrs) with lumbosacral root irritation persisting 6-12 weeks and causing dermatomic pain in the leg with a minimum score of 40 mm on the 100 mm horizontal VAS for pain intensity. The MRI scan must show unilateral extrusion of a disc between the fourth and 17

19 fifth lumbar vertebrae or between the fifth lumbar and first sacral vertebrae. Patients presenting with a cauda equina syndrome, a recently acquired severe paresis (MRC 2 or less) of a muscle group, lumbosacral radicular syndrome in the past year, a history of low back surgery, congenital stenosis, spondylolisthesis, pregnancy, or other serious illnesses, as well as those who are not insured, suffer from claustrophobia or have metal implants will be excluded from the study. Usual care in the Netherlands for the (sub-) group of patients involved There is consensus between the various medical disciplines involved that, after a period of 6-8 weeks of persistant pain in the leg, surgical discectomy can be offered to these patients (1). This time period was probably chosen because several studies have demonstrated that 70 to 80 percent of patients experience partial or total relief of complaints within this time period. After 8 weeks of complaints it is unknown which percentage of patients will recover without intervention and when. Due to the lack of scientific data general practitioners and specialists of other medical disciplines prefer to delay the decision on surgical treatment, in the hope that the natural history will be favourable. Meanwhile physiotherapy is frequently prescribed. The opposite, however, occurs as well. The patients themselves may demand surgery after a relatively short period of complaints. Some requests are granted, others are not. This wide variation in the timing of referral for disc surgery partly depends on such factors as the severity of the symptoms, the ability of these patients to cope with the radicular pain during their daily activities or social intercourse and fear that the pain will become chronic. The fact that the period of 6-8 weeks of pain is based on consensus rather than scientific evidence is probably the main reason for the wide variation in the timing of disc surgery in daily practice. Medical (sub-) disciplines involved in the usual care and their participation in this proposal General Practitioners will refer the patients and will be entirely responsible for the treatment of those patients randomized to the conservative treatment arm. Neurologists together with a research nurse will do most of the intake, including physical examination and obtaining informed consent, and will perform the follow-up evaluations. Neuroradiologists will take the MRI scans and conventional x-rays of the spine before randomization and during the trial. Neurosurgeons will perform the surgery. One of the neurosurgeons is the Principal Investigator. Physiotherapists will guide the surgically treated patients as well as those conservatively treated patients referred by the general practitioner. Anaesthesiologists and office, surgical and clinical nurses will be involved in the treatment of the patients. Describe the policy problem that is underlying this research proposal In the Netherlands every year more than 70,000 patients suffer from a newly developed lumbosacral radicular syndrome. Combined direct and indirect costs are estimated at 2.6 billion guilders per year (2). At least 11,000 patients (3) per year undergo surgery for lumbar disc herniation despite the ultimately favourable natural course. The surgical rate in our country is high when compared to the rest of the world, e.g. 4 times higher than in the United Kingdom (4). Only in the United States are more disc operations performed. Two reasons for the high surgical rate in our country are the timing of surgery and patient/doctor preferences. There is no evidence-based data to justify any particular time of surgery, which may explain the wide variation in treatment in our country. It is not known whether our high rate of disc surgery is cost-effective or not, compared to other countries. The period of 6 to 8 weeks of conservative treatment before surgery is offered seems to be arbitrarily chosen. Between 70 and 80 percent of patients have a favourable course in the first 6-8 weeks. It is unknown how many patients no longer have leg complaints in the weeks or 18

20 months after this period. In 1999 the Gezondheidsraad (Dutch Health Council) MTA subcommittee, which investigated the efficacy of the diagnostics and treatment of the lumbosacral radicular syndrome caused by disc herniation, recommended to the Minister of Health that a study on the indications for and timing of surgery and the complications of this intervention should be conducted. It was also concluded that more insight is needed into patient and doctor preferences with respect to their treatment (3). Indicate how the results of the proposed study may contribute to the resolution of this policy problem More insight is needed into the potential short-term effects after early surgery, as advised in the Dutch Consensus, compared to a longer wait-and-see period. In particular the effects on the return to work or resumption of previous daily activities as well as the complications of both strategies have not yet been investigated. Our proposed study will supply evidence that will show whether early surgery is more (cost-)effective than prolonged conservative care followed by late surgery when indicated. The high surgical rate and the number of MRI scans performed will decrease if it can be shown that a longer wait-and-see period results in recovery of the majority of patients. The latter are often able to continue working or, when on sick leave, to go back to work earlier than those who undergo surgery with the risks of complications. This scenario results in a gain in health effects and will save money currently used for in-hospital treatment diagnostics, consultation of medical specialists, and surgery. Fewer productive hours will be lost since a lot of patients operated upon will not resume their work until 2 months after surgery. The Dutch Consensus advises that the patient should start working 1 month postoperatively for light and 2 months for heavy physically demanding jobs (1). An unknown number of patients continues working in spite of complaints while the decision for surgery has been made and they will have to suspend their activities for 2 months because of the intervention. They would not need to stop their activities in the conservative treatment arm. On the other hand if the results are favourable for the surgical group, this will be evidencebased confirmation of the advice of the Consensus Committee, proving that the surgical policy for this condition in the Netherlands is adequate. Prevalence of the disease /condition of this subgroup in the Netherlands It is difficult to estimate accurately the prevalence of this self-limiting condition. There is a lack of reliable data partially due to the difference in definition between studies. Since 75,000 patients develop a lumbosacral radicular syndrome each year and % have a favourable natural course in the first 6 8 weeks, we may roughly estimate a prevalence of between 15,000-22,500 people in the Netherlands. The prevalence may be higher if we take into account the fact that an unknown number of patients has a persisting radicular syndrome, which developed the previous years and continues to cause complaints. Incidence of the disease /condition of this subgroup in the Netherlands The incidence of the disease in the Netherlands is estimated to be 5 per 1000 persons a year (5), which results in 75,000 new cases of lumbosacral radicular syndrome each year. More than 11,000 (3) of these patients are operated on annually for this condition. The incidence of patients with persisting pain is estimated to be 19,000. Between % of 75,000 patients recover within the first 6-8 weeks. 19

21 Estimate the potential effects on health from the intervention(s) that will be evaluated in this proposal compared to usual care In the short term we expect immediate relief of leg pain in the surgical group, which probably results in a better quality of life when compared to the conservatively treated patients. We are not able to estimate the effects on returning to work and side-effects. The purpose of the present study is to identify the magnitude of the effects on these outcome measures. The longterm effects are also difficult to estimate because of insufficient data. One randomized study (6) concluded that in the long run (more than 4 years post-randomization) there was no difference in effects between the two treatment arms. This is the only randomized intervention trial that compares surgery with conservative treatment and it was not properly designed methodologically. Moreover it is difficult to apply these results in daily practice, especially in relation to cost-effectiveness. Withholding disc surgery after randomization prolongs pain and eventually postpones the intervention for 6 months or less in some cases. Hypothetically most patients in the conservatively treated group will have a favourable natural course of their leg pain between the allocation of treatment, 6-12 weeks after the start of their complaints, and 3 months postrandomization. This percentage will be even higher at the six-month post-randomization evaluation. These conservatively treated patients continue to work or, when not working because of their complaints, may start with their previous activities earlier when relieved of the pain (e.g. most patients start back to work 8 weeks after the surgical intervention). They would not have the potentially negative effects of surgery, including back pain and/or stiffness and complications. Estimate the potential effects on costs from the intervention(s) that will be evaluated in this proposal compared to usual care The direct medical costs in the surgical treatment arm are estimated to be 26,500 guilders per patient (7). For the conservatively treated patients these medical costs will hypothetically be much lower since only a small portion of these patients will eventually undergo surgery. The medical costs of physiotherapy are estimated to be 500 guilders. For a 40-year-old employee, the productivity costs are estimated to be 7,000 guilders per month of absenteeism. These estimates render the following a-priori estimated societal costs: Group A (Early surgery):37,000 guilders, i.e. 26,500 medical costs plus 10,500 due to 1-2 months of absenteeism. Group B1 (Conservative treatment only): 4,000 guilders, i.e. 500 medical costs plus 3,500 due to 0-1 month of absenteeism. Group B2 (Conservative treatment followed by surgery): at least 41,000 guilders, i.e. 27,000 medical costs (physiotherapy followed by surgery) plus 14,000 due to at least 2 months of absenteeism. The cost-effectiveness of conservative treatment depends critically on the proportion of conservatively treated patients in groups B1 and B2. Assuming an unfavourable scenario with 50% of patients requiring surgery, the estimated costs for group B would be 22,500 guilders per patiënt (50% 4, % 41,000). Compared to group A this implies a saving of 14,500 guilders. Translating these savings per patient to the total group of 11,000 patients undergoing surgery annually, the total cost savings per year would be 160 million guilders. If 100% and 0% reguire surgery, the estimate would range from 44 million guilders additional costs to 360 million guilders cost savings. The break-even point is at 89% undergoing surgery. 20

22 The result of this study will be a trade-off between a quicker relief of leg pain in the surgical group versus an advantage in cost-effectiveness for the conservatively treated patients. Everything depends on the proportion of conservatively treated patients who are relieved of leg pain without surgery and, no less important, when is the pain gone. If any obstacles regarding the implementation of the results of the proposed study can be foreseen, please describe In this context hypothetically there are two possible outcome scenarios with different effects on implementation. 1) If the results of this study show that surgery between 6 to 12 weeks is more (cost-) effective than conservative treatment we do not expect many problems with the implementation of these results. 2) If the results show that the opposite is true, the existing guidelines for this condition need to be altered. The preferences of patients and doctors are difficult to predict and can become an obstacle to implementing the results in daily practice. These preferences will be assessed before randomization and in the follow-up period, to obtain an indication of the strength of these preferences and to evaluate the common assertion that patients and doctors are unwilling to wait. MRI s during follow-up will show the anatomical results of surgery. An adequate root decompression will convince the critical surgeons and neurologists. The aim of this study was defined by the Gezondheidsraad 1999 (2) and is supported by the Dutch Society of Neurosurgeons (NVVN). The principal investigator is a member of the organizing directory of the spine section of this NVVN and the NOV and is active in almost every Dutch committee concerning guidelines with regard to spinal surgery. The results will also be incorporated in future updates of the NHG clinical guideline for Lumbosacral Radicular Syndrome, of which professor Koes is co-author. Both of these facts will help the implementation of results if it means that a longer wait-and-see period, e.g. 3-6 months, is recommended. Are there any studies performed or underway similar to the present study proposal or related to the policy problem of the present study proposal (nationally or internationally)? No. Many studies have been carried out to investigate the diagnosis and prognosis of the lumbosacral radicular syndrome, especially directly after first complaints. One on-going study (Verhagen, Rotterdam) will investigate the acute phase (< 4 weeks) of the lumbosacral radicular syndrome and the appropriate conservative treatment. Internationally there are a few trials in progress, which compare new surgical treatment modalities with conventional surgery. To our knowledge no study is underway which investigates the cost-effectiveness of this intervention and the efficiency of timing of surgery compared to prolonged conservative therapy. References; 1)-CBO Consensus. Het lumbosacraal radiculair syndroom )-Gezondheidsraad 1999/18. Diagnostiek en behandeling van het lumbosacraal radiculair syndroom. 3)-SIG Zorginformatie, Utrecht Gegevens uit de landelijke registratie van Nederlandse Ziekenhuizen. 4)-Cherkin DC, Deyo RA. An international comparison of back surgery rates. Spine 1994; 19: )-Nivel Utrecht Van der Velden J. Basisrapport: Morbiditeit in de huisartsenpraktijk. 6)-Weber H. Lumbar disc herniation: a controlled, prospective study with ten years observation. Spine 83; 8:

23 7)-Koopmanschap 1991: Kosten van ziekten in Nederland. Survey of requested funding NZI-nrs. Year 1 Year 2 Year 3 Personnel 413, 422, Equipment 4814, 485, 466 Procedures COTG-nrs Medical equipment 461, 462 Total Costs Contribution Balance Motivation regarding the personnel requested Methodological Superviser ( 0.05 FTE 3 yr.) is responsible for supervision of the Principal Investigator and the trial. The Principal Investigator (0.5 FTE 3 yr.) is responsible for the design and execution of the project. This includes coordination in each treatment centre and communication with all neurologists, neurosurgeons, general practitioners, physiotherapists and different hospital departments. He will supervise the project manager and the 4 research nurses to obtain high quality data collection and processing. Finally he will prepare the Final Report to NWO. The Project Manager (M.Sc.1.0 FTE 2 yr. and 0.5 FTE 1 yr.) is responsible for data management (0,4 fte) at the Data Coordination Centre in the LUMC. (S)he will assist the Principal Investigator in the management of mailings, communication and solving organisational problems in the different treatment centres (0,6 fte). During absenteeism or surgery by the Principal Investigator (s)he will be responsible for the project. Local Project Coordinator (0.1 FTE 2 yr and 0.05 FTE last year per hospital): In each hospital a neurologist/neurosurgeon will be responsible for local coordination of the trial and communication with the research nurses. The General Practitioner Coordinator (2 yrs 0.1 FTE) will be responsible for mailings to all GP s in the regions around the hospitals. The GP s will treat the patients of group B and can contact the GP coordinator, who will guarantee the quality of care in general practice. Research nurses (4 * 1.0 FTE 2 yr. and 2*1.0 FTE 1 yr.): 300 patients will participate in the study in 9 different hospitals. The first visit will last 30 minutes, the second visit 60 minutes (obtaining informed consent, conducting the neurological examination, filling out of the questionnaires with the patients performing randomization, and planning the operation for those in group A). The 6 follow-up will last 30 minutes. During the study period the research nurses will be in contact with each patient for at least 4.5 hours. This is 1350 hours in total, assuming all 300 give informed consent, or 315 hours per research nurse (75 patients). In addition to the intake and follow-up visits the data form must be filled out properly and communication maintained between research nurse and Principal Investigator, neurologists, neurosurgeons, GP s, office, surgical and clinical nurses, secretaries. The first two years the Biostatistician (0.1 FTE 2 yr., 0.2 FTE 1 yr.) will supervise data collection together with the Project Manager and Principal Investigator. During the second half of the second year the short-term results can be analysed; in the last year complete data analysis and processing will be performed. Econometrist (0.1 FTE): He will conduct the economic analysis and perform a cost estimation in three hospitals. More than 300 MRI s will be taken before randomization, but they would also have been performed under normal circumstances. The 600 MRI s done after randomisation are needed for this study as is the extra work for the (Neuro-)Radiologists (0.3 FTE 3 yr.) in each hospital. ( Separate HTA Question) (For Requested Support see page 9 of Excel) A Post-Doc (1.0 FTE 1.5 yr.) is needed to conduct the parallel study to answer the HTA question. Two extra Research nurses (2 * 0.6 FTE 0.5 yr.) are needed for the HTA Methodology study. 22

24 Diagnosis of Disc Related Sciatica 1. Introduction The clinical diagnosis of disc related sciatica is generally supposed to be straightforward. The indication for surgery is based on clinical signs and symptoms, and duration of persisting pain. As a rule a wait-and-see policy of at least 6 to 8 weeks is maintained. Thereupon a MRI or CT-scan is performed in order to confirm the level and side of root compression anticipating the surgical intervention. Recently doubt has been raised around the sensitivity, specificity and predictive value of radicular root compression signs on the presence of a lumbar disc herniation (LDH). 1 Moreover some studies attributed possible prognostic value to the MRI findings helpful in decision making between surgical or non-surgical strategy in individual patients. 2-4 Vroomen et. al developed a model predicting future surgical intervention in sciatica patients directly derived from primary care. 5 However, this model is has not been tested in other populations and extrapolation to a secondary care population has not been examined yet. The present study, which is linked to the Sciatica Trial, aims at answering the research questions described below. To simplify the protocol the diagnostic questions have been split up into in four different chapters and two separate papers: the first concerning the predictive value of history and neurological signs on the presence of a LDH and root compression (protocol 1 and this document), the second discussing the value of neurological signs predicting the level of LDH and site of root compression (protocol 2), the third on the predictive value of the Maastricht formula (protocol 3), the fourth comparing MRI aspects at baseline and at 1 year follow-up in correlation to neurological signs (protocol 4), an intra- and interobserver study (described by Donald Bouman) a more in depth qualitative and quantitative analysis of MRI aspects and their correlation with low back pain and sciatica (described by Barry Schenk). 23

25 A study looking at the prognostic value of MRI findings will be conducted separately. For the diagnostic protocols data drawn from the sciatica trial are used. The details of this randomized trial and the recently published 1-year results are described elsewhere

26 2. The Sciatica Trial, design and patient population The Sciatica Trial is a multicenter, prospective, randomized trial, investigating the timing of surgery in protruded disc related lumbosacral radicular syndrome (LSRS). Two treatment strategies were compared in a population of patients with persisting sciatica: Surgery performed following the current guidelines after a 6-8 weeks period of failed conservative treatment Versus Prolonged conservative care by the General Practitioner with eventually delayed surgical treatment. All consecutive patients with a disabling radicular syndrome of at least six weeks duration, recruited from general practice (GP) and subsequently neurological practices in 9 hospital regions, between January 2003 and February 2005 have been analysed. All patients were asked to participate in a randomized controlled trial comparing early surgery with a conservative wait-and-see strategy. Since these patients were referred by GP s, a radiological diagnosis had not been made yet. After neurological secondary care confirmation of the diagnosis LSRS with high suspicion of LDH and surgical indication the patients were enrolled in the study. The Research Nurses were trained to interview the patients in a standardized way. The same applied to the neurological examination. The variables used for the diagnostic studies were restricted to those gathered by the neurologist. The neurologist concluded which nerve root(s) were suspected of nerve root compression. As the MRI was not performed yet the independent research nurses and neurologist were not aware of the presence or absence of LDH. After receiving informed consent the patients underwent a study protocol MRI. As stated above a considerable proportion of these eligible patients did not harbour a disc herniation with root compression on MRI, and were not randomized. The main reason for MRI was to prove root compression by a disc herniation, before definite randomization was allowed. A priori statements included that a clear-cut sciatica patient should have root compression. All patients undergoing MRI scanning in the Sciatica Trial protocol were analysed. These patients did have a surgical indication as stated by their general practitioner, neurologist, research nurse and neurosurgeon, as described in the Dutch Guidelines. 25

27 The MRI data were registered by the attending radiologist and neurosurgeon using a standardized Case Record Form (CRF). Patients were not informed on the results of the MRIscan. Inclusion criteria Patient age years. Persistent radicular pain in the L4, L5 or S1 dermatome with or without mild neurological deficit. Intensity of leg pain more than 40 mm measured on the 100 mm horizontal VAS. Leg pain of 6-12 weeks duration. Evidence of a unilateral disc herniation confirmed on MRI. Informed consent. Exclusion criteria Cauda equina syndrome or severe paresis (MRC<3). Complaints of a lumbosacral radicular syndrome in the same dermatome within the past 12 months. Low back surgery in the past. Significant central canal stenosis, degenerative or lytic spondylolisthesis. Pregnancy. Severe life-threatening or psychiatric illness. Insufficient knowledge of Dutch language. Plans to move to another country in the year after randomization. Contra-indications for MRI. Table 1. In- and exclusion criteria Sciatica Trial After registration of the baseline variables, patients in the Sciatica Trial were randomized into two groups, receiving either lumbar disc surgery or conservative management. Lumbar disc surgery (group B) was performed in the conventional manner with loupe magnification. All participating surgeons were certified neurosurgeons with experience in the standard approach with loupe magnification. Surgery took place as soon as possible with a maximum of two weeks after randomization. Hospital admission was 4-6 days, including the day of surgery. During the post-operative period the patients were mobilized with the aid of a 26

28 physiotherapist. At home patients visited their own physiotherapist who guided them and provided exercises 2 times a week for 6 weeks. If, at their last visit to the outpatient clinic 6-8 weeks post-operatively, patients still complained of persisting leg pain or aggravated back pain, additional physiotherapy sessions were prescribed by the neurosurgeon. Conservative management consisted of a wait-and-see policy conducted by the general practitioner (GP). The GP provided information on the favorable prognosis of LSRS. The treatment of LSRS was aimed primarily at pain relief and maintenance/restoration of normal day-to-day activities. The GP s encouraged patients to continue their normal day-to-day activities as far as possible. Analgesic medication was prescribed if necessary. The doctor periodically checked the efficacy of the prescribed pain medication and if necessary adjusted the dose or sort of analgesics. Paracetamol was the first choice, if not effective, NSAID s (ibuprofen, diclofenac or naproxen) were prescribed. Only in case of severe disabling pain, morphine was given for a restricted period of time. By preference all analgesics should be taken at fixed times of the day rather than on an if necessary basis. If the GP and the patient concluded that there was considerable kinesiophobia (fear that movement will increase pain), the aid of a physiotherapist can be recommended. The GP was free in her/his choice of prescription of medication and referral to physiotherapists. The research nurse registered the conservative management strategy after communication with the responsible GP. In case of progressive neurological deficit or worsening intolerable pain the GP could refer the patient back to the research nurse or neurosurgeon. If, six months after randomization, the patient had still not improved or suffers from intermittent LSRS, surgical treatment was offered. Patients might ask for earlier surgery because of worsening leg pain. In these crossover cases and in the case of a progressive neurological deficit, surgery was performed in consultation with the patient. Research nurses, who were thoroughly trained in neurological examination, conducted follow-up examinations in both groups at 4, 8, 26, and 52 weeks after randomization. 2.1 Measured variables and data acquisition in the Sciatica Trial In the Sciatica Trial, various clinical and demographic parameters were recorded for all patients, as shown in Table 2. Demographic variables include age (18-45 vs ), sex, Quetelet index, and heavy labour. History variables include low back pain, duration of pain, dermatomal pattern of pain, coldness in the area of pain, paresthesias, anamnestic paresis; aggravation by walking, coughing-sneezing-pressing and disability to work. Affective variables were measured using the McGill Questionnaire. 27

29 Physical examination includes root compression signs such as the Lasegue manoeuvre, finger-ground-distance, Kemp manoeuvre, scoliosis, reflex difference, sensory deficit and motor deficit. Other signs were analgetic posture or walking pattern. Physical examination and clinical interview Weight Length Well-leg lifting Restricted lumbar motion Painful lumbar motion Crossover pain Kemp's sign Lasegue's sign (Straight Leg Raising test) Crossed Lasegue's sign Bragard's sign Trendelenburg's sign Knee tendon reflex Achilles tendon reflex Muscle strength, MRC graded (M. Tibialis anterior, M. Peronei, M. Extensor hallucis longus, M. Triceps surae) Sensory exam Demographic parameters Age Sex Occupation Marital status Questionnaires Roland Disability Questionnaire for Sciatica. Perceived recovery (7-point Likert scale) VAS for leg pain McGill Questionnaire for Pain Fingers-to-ground distance Sciatica history Time since first onset of leg pain Localization of leg pain Nature and frequency of leg pain Previous episodes of leg pain Sensory deficits Muscle weakness Signs of cauda equina syndrome Low back pain history Time since first onset of leg pain Time since first onset of low back pain Disability / time off work Traumatic onset of complaints Factors provoking leg pain Pain medication Complications Highest level of education Children Recreational activities Short-Form 36 Sciatica Frequency and Bothersome Index. VAS for low back pain 28

30 Table 2. Clinical parameters, demographic parameters and questionnaires collected at baseline and follow-up in the Sciatica Trial 2.2 MR-imaging in the Sciatica Trial All patients in the Sciatica Trial underwent MR-imaging at the time of the initial diagnosis of sciatica and after one year of follow-up. MR-imaging was performed using a standardized protocol (Table 3) and the radiologist and surgeon completed CRF s independently. (Separate interobserver study) 1 Sagitaal T1SE 2 Sagitaal T2TSE 3 Sagitaal T1SE na Gado 4 Transversaal T1-TSE 5 Transversaal T2-TSE 6 Transversaal T2-FFE 7 Transversaal T1-TSE na gado 8 MR caudografie Table 3. MRI protocol 29

31 3. Protocol 1: Predictive value of history and neurological signs on the presence of a LDH and root compression 3.1 Background As stated in the outline, doubt has been raised about the predictive value (PV) of radicular symptoms and signs on the presence of a lumbar disc herniation (LDH) and nerve root compression. Especially the diagnostic part of the bed rest study by Vroomen showed less disc herniations than expected in patients with clear-cut sciatica of short duration. 9 Other studies showed disc protrusions and extrusions in asymptomatic people The old rule in textbooks and guidelines has to be evaluated. If the sensitivity, specificity and predictive value is lower than expected the role of MRI will get another perspective in diagnosis, instead of confirming the clinical suspicion. This may affect the treatment in primary and secondary care. It is a fact that the proportion of good and excellent outcomes in daily care depends on a pre-treatment sufficient diagnosis. This study is designed to estimate the value of clinical signs and symptoms on predicting the presence of a lumbar disc herniation and/or compression of nerve roots. 3.2 Research Questions Is it possible to predict the presence of a Lumbar Disc Herniation by history and neurological examination in patients with persisting sciatica? The presence of a disc herniation is not enough for a surgical indication. Outcome of intervention depends on the presence of root compression. A disc herniation without root compression yields less satisfactory results of surgery. A compromised nerve root however can also exist without a disc herniation (e.g. lateral recess stenosis or neuroforamen problems). This results in the second research question. Is it possible to predict the presence of nerve root compression by history and neurological examination in patients with persisting sciatica? Question one will result in a simple research answer, whereas answering question two results in a more clinical oriented conclusion. 30

32 Figure 1: Predictive value LDH (question 1) and root compression (question 2) 3.3 Methods For the scientific value of all diagnostic research questions four experienced spine experts will evaluate the images uniformly and independent of each other. A senior neurororadiologist (JW) and a senior neurosurgeon (RT) supervised the scientific observations. Two neuroradiologists (NA and PB) and two neurosurgeons ( and WP) filled out the CRF s of all MRI s and X-rays after double evaluation of the scans (One CRF blinded for the clinical question and the second CRF with the knowledge of the suspected nerve root compression). The Sciatica Research Team will gather all MRI studies including follow-up MRI s. The hard-copies and CD s are burned on hard-disc, which will be read using uniformly configured Apple Powerbook laptops equipped with OsiriX medical imaging software. All scans will be coded using the Unique Identification Code (UIC) of the ProMIse statistical software. Barry Schenk will coordinate the diagnostic study and thereby the gathering of all the data. The MRI sagittal and axial slides will be scored on the presence or absence of clinically relevant disc herniation root compression and other radiologically relevant signs such as lateral recess stenosis, neuroforamen stenosis, anatomical location of herniation, sequestration, modic changes, facet hypertrophy, lytic spondylolisthesis and other deformities. These variables will be used for the other protocols but scored at once to heighten the efficiency. For diagnostic purposes each scanned level will be investigated separately, and a 10 cm VAS scale will be used to assess the probability of presence or absence of these abnormalities. 31

33 During the examinations the radiologists and surgeons are unaware of the clinical suspected root. They will be blinded for the information from the neurologist and research nurse. As earlier studies demonstrated the multiplicity of radiological lumbar abnormalities without any clinical correlating pain syndrome, a re-evaluation of the scans was done with known clinical suspicion of root compression. The investigators will again evaluate all MRI s as if a patient has to be diagnosed before surgical treatment is undertaken. They will conclude the presence or absence of clinical and surgical relevant abnormalities. For this investigation clinical information is essential and will be attached to the second CRF after filling out the blinded CRF. If as expected, disagreement exists between the investigators they will try to reach consensus by discussing the patient. In case of no definite agreement the two senior investigators (RT and JW) will be contacted by the coordinator Schenk and a definite conclusion will be made. The conclusion will be used for the second analysis of this study, after a blind evaluation has been performed. The first analysis (blind observation) will focus on the scientific value of the correlation between symptomatology and radiological abnormalities. The second analysis will focus on clinical significance and search for treatable pathologies. Univariate analysis will be performed using all variables of interest for both items (disc herniation and root compression) separately and combined. Logistic regression will be performed for both items (disc herniation and root compression) separately and combined. The authors are aware of the fact that root compression is rare in case of an absent disc herniation. In case of lateral recess stenosis however it is possible to diagnose nerve root compression sec. 32

34 4. Protocol 2: Level and site of lumbosacral root compression in sciatica 4.1 Background The clinical diagnosis of disc related sciatica is generally supposed to be easy. This is one of the main reasons why MRI-scanning is restricted to cases of clear-cut radicular syndrome, in which surgery is considered. Imaging of the site of root compression enables the surgeon to perform the discotomy efficiently and safely. Recently however doubt has been raised about the height of correlation between clinical signs and symptoms and MRI findings. 1 Firstly classical neurological signs are not invariably associated with the presumed level of disc herniation. Secondly patients with non-correlating symptoms but with root-tension signs may very well experience good relief of function and pain after surgery. These exceptions conflict with our medical beliefs and need data towards supported evidence. Kortelainen was the first to raise doubt about the strength of the predictive value of the classical radicular syndrome caused by a lumbar disc herniation (LDH). 13 If indeed the sensitivity and specificity are not as high as medical doctors use to think, and also the result of surgery does not merely depend on the classical syndrome more people could benefit from surgery. On the other hand if the classical LSRS appears the crucial indicator to predict the site of disc herniations and thereby the result of surgery, only patients with a fitting radicular syndrome should be offered surgery which will lower the surgery rates. This protocol will investigate the value of the classical lumbosacral radicular syndrome, in correlation with the level and site of the disc herniation as visualised by MRI. 4.2 Research Questions Do classical root related symptoms and signs fit to the clinically expected level of root compression in patients with sciatica? The nerve root compressed will be analysed in relation to some of the medical history and neurological findings. E.g. does a patient with pain in the S1 dermatome and/or hypaesthesia, lowered ankle jerk, and a motor deficit of the gastrocnemicus have disc herniation at Disc Level L5-S1 and root compression S1? In other words is the level of disc herniation predictable or does this old adagio of standard neurological and neurosurgical textbooks have no value? 33

35 Does the classical compound lumbosacral radicular syndrome L4 or L5 or S1 predict the expected compressed nerve root when compared with patients with less well defined sciatica? In addition to the answers to the first question, patients with a combination of signs correlating to one root will be compared to patients with a less clear cut radicular syndrome. 4.3 Methods In order to answer the above-mentioned questions, data of the Sciatica Trial are used. As described above this multicenter randomized trial investigated the timing of surgery in disc related LSRS. All patients undergoing MRI scanning in the Sciatica Trial were analysed. The scans were evaluated in usual care but written down on a standardized Case Record Form (CRF). This procedure was used for inclusion of patients and for a separate inter-observer study. The neurologist concluded the CRF with the predicted compressed nerve root. All variables of interest for this study are related to the classical root compression signs and symptoms. Pain in calf, heel, sole foot and lateral part foot (S1) Pain lateral lower leg, forefoot and great toe (L5) Pain anterior tibia (L4) Paresis M. Gastrocnemicus (S1) Paresis M. Peroneai (S1, L5) Paresis M. Ext. Hallucis Longus (L5) Paresis M. Tibialis Anterior (L5) Achilles tendon reflex deficit (S1) Knee tendon reflex deficit (L4) Sensory deficit area calf, heel, sole foot and lateral part foot (S1) Sensory deficit area lateral leg under knee, forefoot and great toe (L5) Sensory deficit area medial leg under knee without foot (L4) 34

36 For answering research question 2 at least three or more combinations (a, b or c) of the following signs are needed, without a sign of another root to define a one root syndrome. a. S1 pain, Lasegue positive < 60 degrees, sensory deficit in S1 dermatome, m.gastrocnemnius deficit, m.peronaei deficit and Achilles tendon difference b. L5 pain, Lasegue positive <60 degrees, sensory deficit in L5 dermatome, m. extensor hallucis longus, m. tib. ant. and or, m. ext.hall. longus paresis, positive Trendelenburg sign c. L4 pain, inverted Lasegue, sensory deficit L4, m. quadriceps deficit, knee tendon difference? MR imaging was performed using a standardized protocol (Table 2) and the radiologist and surgeon completed CRF s independently. (Separate interobserver study, protocol 4) 1 Sagitaal T1SE 2 Sagitaal T2TSE 3 Sagitaal T1SE na Gado 4 Transversaal T1-TSE 5 Transversaal T2-TSE 6 Transversaal T2-FFE 7 Transversaal T1-TSE na gado 8 MR caudografie Table 2. MRI Protocol The Sciatica Research Team gathered all MRI studies including follow-up MRI s. The hardcopies and CD s are burned on hard-disc, which will be read using uniformly configured Apple Powerbook laptops equipped with OsiriX medical imaging software. All scans were coded using the Unique Identification Code (UIC) of ProMIse statistical software. Barry Schenk will coordinate the diagnostic study and thereby the gathering of all the data. 35

37 For the scientific value of all diagnostic research questions four experienced spine experts evaluated the images uniformly and independent of each other. A senior neurororadiologist (JW) and a senior neurosurgeon (RT) supervised the scientific observations. Two neuroradiologists (NA and PB) and two neurosurgeons ( and WP) filled out the CRF s of all MRI s and X-rays after double evaluation of the scans (One CRF blinded for the clinical question and the second CRF with the knowledge of the suspected nerve root compression). The MRI sagittal and axial slides will be scored on the presence or absence of Disc herniation Root compression The root number Other radiological relevant signs such as lateral recess stenosis, neuroforamen stenosis, anatomical location of herniation, sequestration, modic changes, facet hypertrophy, lytic spondylolisthesis and other deformities will be scored. These variables will be used for the other protocols but scored at once to heighten the efficiency. For diagnostic purposes each scanned level will be investigated separately, and a 10 cm VAS scale will be used to assess the probability of presence or absence of these abnormalities. During the examinations the radiologists and surgeons are unaware of the clinical suspected root. They will be blinded for the information from the neurologist and research nurse. In case of no definite agreement the two senior investigators (RT and JW) will be contacted by the coordinator Schenk and a definite conclusion will be made. The conclusion will be used for the second analysis of this study, after a blind evaluation has been performed. Univariate analysis will be performed using all variables of interest for the specific roots L4, L5 and S1 (Question 1) followed by a per root (L4, L5 and S1, (present or absent)) logistic regression analysis (Question 2). Logistic regression analysis will be performed for both disc herniation and root compression, separately and combined. The authors are aware of the fact that root compression is rare in case of an absent disc herniation. In case of lateral recess stenosis however it is possible to diagnose nerve root compression sec. 36

38 5. Protocol 3: Predictability of persistent disc related root compression in sciatica patients 5.1 Background The clinical diagnosis of disc related sciatica is generally supposed to be straightforward. This is one of the arguments to perform MRI or CT-scanning in case of a clear-cut radicular syndrome, only to confirm the level and side of root compression to facilitate the surgeon to perform the discotomy safely. Recently doubt has been raised around the presence of a lumbar disc herniation (LDH) in patients with signs of root compression. 1 Although Vroomen et. al tried to approach the truth by multiple regression analysis, their study population is not to be compared to secondary care. 5 In the study performed by Vroomen et al. 183 patients with nerve root compression were prospectively sampled from a primary care population. 2 Models were reconstructed by logistic regression: 1. Model A predicting future surgery after first consultation 2. Model B predicting future surgery after second consultation 2 weeks later The probability of surgical intervention in the next 6 months after first consultation (model A) was held by the following paradigm: Probability (X) = 1/ (1 + e -y ) Y = α + ln(4)*x1 + ln(0.22)*x2 + ln(4.3)*x3 + ln(2.3)*x4 + ln(1.033)*x5 + ln(3.2)*x6 α = intercept and X1 through X6 are the six variables in the order of the table below. 37

39 Baseline predictors of surgery Crude odds ratio 95% CI Adjusted odds ratio 95% CI History/questionnaire: Mentally demanding job Sudden onset More pain on coughing/sneezing/straining Difficulty putting on socks/stockings VAS of pain intensity in the leg Physical examination: Reversed straight leg raising Exponential value of the intercept 3.6* * *10-2 After second consultation two weeks later (model B) the probability of surgery in the 6 months was held by the following paradigm: Probability (X) = 1/(1 + e -y ) with Y = α + ln(5.8)*x1 + ln(1.2)*x2 + ln(0.17)*x3 + ln(7.3)*x4 + ln(10.1)*x5 + ln(4.1)*x6 + ln(4.4)*x7 + ln(0.2)*x8 + ln(24.3)*x9 + ln(1.058)*x10 + ln(1.35)*x11 + ln(4.2)*x12 X1 through X12 are the variables listed in the table below; α is the intercept with the value as stated in the table below. 38

40 Baseline predictors of surgery Crude odds ratio 95% CI Adjusted odds ratio 95% CI History/questionnaire: Duration of disease at presentation: days More than 30 days Previous episodes of pain in the leg Previous episodes of low back pain Positive family history for: Pain in the back Pain in the leg Mentally demanding job Paraesthesias Pain worsening on sitting after weeks Change over 2 weeks in VAS of leg pain PRI-A of McGill pain questionnaire Physical examination: Crossed straight leg raising Exponential of the intercept 2.3*10-3 (5.3* *10-2 ) Model A was designed to have a specificity exceeding 95%, limiting the rate of MRI studies performed in patients eventually not undergoing surgery, whereas model B was designed to reliably expedite surgery and was required to have a specificity of nearly 100%. For model B, the probability cut off with a 100% specificity was The value of both formulas will be tested on our population. Whereas Vroomen s study concentrated on sciatica of less than six weeks his formula is supposed to have the same predictive value. The only retesting has been done on his own population; whereas the value may be greater if tested in another population. 5.2 Research questions Can the presence of disc related non-recovering sciatica, indicating surgical decompression in the Sciatica trial population be predicted by the logistic regression formula of the University of Maastricht? 39

41 By applying the paradigms for Model A and B the probability of surgical intervention within 6 months will be calculated for each individual patient in the conservative treatment group. This will be compared with the actual number of patients in the conservative treatment group undergoing surgery within 6 months. The nullhypothesis tested is that Vroomen s model does apply to patients in the sciatica trial. Is it possible to create a model, based on the data collected in the sciatica trial, predicting surgical intervention in sciatica patients? A similar model as in Vroomen s study will be created based on the data collected in the sciatica trial, by backward stepwise logistic regression analysis. 3 Models will be constructed using baseline variables, predicting surgical intervention within the next 3, 6 and 12 months, respectively. 5.3 Methods To investigate the above mentioned questions, all data of the Sciatica Trial were used. As described above this multicenter randomized trial investigated the timing of surgery in protruded disc related lumbosacral radicular syndrome (LSRS). Two treatment strategies were compared in a population of patients with persisting sciatica: Surgery performed following the current guidelines after a 6-8 weeks period of failed conservative treatment versus prolonged conservative care by the General Practitioner with eventually delayed surgical treatment. To answer the above stated research questions the data of the conservative treatment group will be analysed. Both paradigms of Vroomen s study will be applied to calculate the probability of surgical intervention within 6 months (correction for duration of complaints????) for each individual patient. The probability cut-off will be maintained at Some patients will be below the cut off and some above. These numbers will be compared with the numbers actually operated in a 2x2 table. The specificity and sensitivity for both models will be calculated and compared to Vroomen s study. Backward stepwise logistic regression analysis with the likelihood ratio test on SPSS software will be used to construct 3 models predicting surgical intervention in the next 3, 6 and 12 months, respectively, based on all variables in the Sciatica trial collected at baseline. At any given probability cut off, some patients will be below the cut off and some above. These numbers will be compared with the numbers actually operated in a 2x2 table. In this way, 40

42 sensitivity and specificity values will be calculated for any probability cut off value. The (1- specificity) versus sensitivity values for the various probability cut offs of both models will be depicted in a receiver operating characteristic. 14, 15 The area under the curve will be estimated non-parametrically. This area reflects the overall predictive properties of each model. The probability at which the specificity is 100% will be determined. 41

43 6. Protocol 4: MRI aspects at baseline and at 1 year follow-up in correlation to neurological signs 6.1 Background Many studies examining the correlation between MRI aspects and clinical findings focus on baseline findings predicting outcome dependant of treatment modality, without further MRI follow-up. In the sciatica trial, where the recently published 1-year outcomes were similar for patients assigned to early surgery and those assigned to conservative treatment with eventual surgery if needed, follow-up MR imaging was performed 1 year after baseline. 6 The availability of these additional data offers the opportunity to examine the evolution of MRI aspects in relation to clinical findings for both treatment groups. This study will mainly focus on the extent in which the evolution of MRI aspects are reflected in changes in clinical findings changes. In addition, the correlation between MRI aspects at 1-year follow-up and treatment outcome will be examined. 6.2 Research questions Is the evolution of clinical/neurological symptoms related to a change in the rate of LDH/nerve root compression? The assumption is that for example a decrease in clinical symptoms is related to a diminished or absent LDH/nerve root compression. However, could a significant decrease in clinical symptoms be possible even in the presence of a prolonged existing nerve root compression at MRI follow-up? Is outcome related to MRI characteristics at 1 year follow-up? In patients that underwent surgery the MRI after 1 year will also be used to verify the level of surgery and to examine the formation of epidural scar tissue with possible nerve root compression. 6.3 Methods For the scientific value of all diagnostic research questions four experienced spine experts will evaluate the images uniformly and independent of each other. A senior neurororadiologist (JW) and a senior neurosurgeon (RT) supervised the scientific observations. Two neuroradiologists (NA and PB) and two neurosurgeons ( and WP) filled out the CRF s of all MRI s and X-rays after double evaluation of the scans (One CRF blinded 42

44 for the clinical question and the second CRF with the knowledge of the suspected nerve root compression). The Sciatica Research Team will gather all MRI studies including follow-up MRI s. The hard-copies and CD s are burned on hard-disc, which will be read using uniformly configured Apple Powerbook laptops equipped with OsiriX medical imaging software. All scans will be coded using the Unique Identification Code (UIC) of the ProMIse statistical software. Barry Schenk will coordinate the diagnostic study and thereby the gathering of all the data. The MRI sagittal and axial slides will be scored on the presence or absence of clinically relevant disc herniation, root compression, epidural scar tissue formation and other radiologically relevant signs such as lateral recess stenosis, neuroforamen stenosis, anatomical location of herniation, sequestration, modic changes, facet hypertrophy, lytic spondylolisthesis and other deformities. These variables will be used for the other protocols but scored at once to heighten the efficiency. The level of surgery will be appointed based on post-operative radiological effects. For diagnostic purposes each scanned level will be investigated separately, and a 10 cm VAS scale will be used to assess the probability of presence or absence of these abnormalities. During the examinations the radiologists and surgeons are unaware of the clinical suspected root and will be blinded for the information from the neurologist and research nurse. As earlier studies demonstrated the multiplicity of radiological lumbar abnormalities without any clinical correlating pain syndrome, a re-evaluation of the scans was done with known clinical suspicion of root compression. The investigators will again evaluate all MRI s as if a patient has to be diagnosed before surgical treatment is undertaken. They will conclude the presence or absence of clinical and surgical relevant abnormalities. For this investigation clinical information is essential and will be attached to the second CRF after filling out the blinded CRF. If as expected, disagreement exists between the investigators they will try to reach consensus by discussing the patient. In case of no definite agreement the two senior investigators (RT and JW) will be contacted by the coordinator Schenk and a definite conclusion will be made. The conclusion will be used for the second analysis of this study, after a blind evaluation has been performed. The first analysis (blind observation) will focus on the scientific value of the correlation between symptomatology and radiological abnormalities. The second analysis will focus on clinical significance and search for treatable pathologies. A successful outcome is defined as an improvement of at least 11 points on the one-yearfollow-up-roland Disability Scale for Sciatica compared to baseline. Treatment failure is defined as worsening or less than 11 points improvement on the Roland Disability Scale for 43

45 Sciatica compared to baseline. Patients requiring additional lumbar disc surgery will also be considered failures. Univariate analysis will be performed using all variables of interest for both items (disc herniation and root compression) separately and combined. Logistic regression will be performed for both items (disc herniation and root compression) separately and combined. 44

46 7. References 1. Vroomen PC, de Krom MC, Wilmink JT, Kester AD, Knottnerus JA. Diagnostic value of history and physical examination in patients suspected of lumbosacral nerve root compression. Journal of neurology, neurosurgery, and psychiatry 2002;72(5): Choi SJ, Song JS, Kim C, et al. The use of magnetic resonance imaging to predict the clinical outcome of non-surgical treatment for lumbar intervertebral disc herniation. Korean J Radiol 2007;8(2): Jensen TS, Albert HB, Sorensen JS, Manniche C, Leboeuf-Yde C. Magnetic resonance imaging findings as predictors of clinical outcome in patients with sciatica receiving active conservative treatment. Journal of manipulative and physiological therapeutics 2007;30(2): Carragee EJ, Kim DH. A prospective analysis of magnetic resonance imaging findings in patients with sciatica and lumbar disc herniation. Correlation of outcomes with disc fragment and canal morphology. Spine 1997;22(14): Vroomen PC, de Krom MC, Knottnerus JA. When does the patient with a disc herniation undergo lumbosacral discectomy? Journal of neurology, neurosurgery, and psychiatry 2000;68(1): Peul WC, van Houwelingen HC, van den Hout WB, et al. Surgery versus prolonged conservative treatment for sciatica. The New England journal of medicine 2007;356(22): Peul WC, van Houwelingen HC, van der Hout WB, et al. Prolonged conservative treatment or 'early' surgery in sciatica caused by a lumbar disc herniation: rationale and design of a randomized trial [ISRCT ]. BMC musculoskeletal disorders 2005;6:8. 8. The Sciatica Trial. (Accessed at 9. Vroomen PC, de Krom MC, Wilmink JT. Pathoanatomy of clinical findings in patients with sciatica: a magnetic resonance imaging study. Journal of neurosurgery 2000;92(2 Suppl): Boden SD, Davis DO, Dina TS, Patronas NJ, Wiesel SW. Abnormal magneticresonance scans of the lumbar spine in asymptomatic subjects. A prospective investigation. The Journal of bone and joint surgery 1990;72(3): Brant-Zawadzki MN, Jensen MC, Obuchowski N, Ross JS, Modic MT. Interobserver and intraobserver variability in interpretation of lumbar disc abnormalities. A comparison of two nomenclatures. Spine 1995;20(11): ; discussion Jensen MC, Brant-Zawadzki MN, Obuchowski N, Modic MT, Malkasian D, Ross JS. Magnetic resonance imaging of the lumbar spine in people without back pain. The New England journal of medicine 1994;331(2): Kortelainen P, Puranen J, Koivisto E, Lahde S. Symptoms and signs of sciatica and their relation to the localization of the lumbar disc herniation. Spine 1985;10(1): Hanley JA. Receiver operating characteristic (ROC) methodology: the state of the art. Critical reviews in diagnostic imaging 1989;29(3): Metz CE. Basic principles of ROC analysis. Seminars in nuclear medicine 1978;8(4):

47 MRI-findings in lumbar disc herniation 1. Introduction The relationship between lumbar disc herniation and sciatica has been well known ever since Mixter and Barr's first published their paper on lumbar disc lesions in Their mechanical concept of root compression by a herniated portion of the disc offers a satisfying explanation for most symptomatic patients. However, this concept cannot explain the high rate of asymptomatic disc herniations that can be found in patients without sciatica or low back pain, ranging from 28% 2 to 76% 3.With the introduction of sophisticated imaging modalities such as CT- or MR-imaging, many investigators have been trying to increase the knowledge of the pathophysiologic mechanisms at work The increasing accuracy of MR-imaging has made it possible to visualize in detail the various anatomical aspects of the intervertebral disc, nerve root and surrounding structures. 6;8;10;16-22 MR-findings such as the presence of disc extrusion or severe nerve root compression have been reported to be strongly associated with sciatica 23, but it remains unclear to what extent the morphological variations seen on MR-imaging are associated with low back pain as well, rather than being a representation of idiosyncratic differences between individuals. 3;23-28 The establishment of correlations between certain lumbar disc- and nerve root characteristics and the presence of sciatica and/or low back pain may not only help improve our understanding of its etiology, but it may also provide anchor points for new therapeutic approaches or fine-tuning of existing therapeutic strategies. This protocol describes a diagnostic study that aims to assess both diagnostic- and prognostic aspects of Magnetic Resonance Imaging in patients with sciatica with or without low back pain in the presence of lumbar disc herniation. The first chapter of this protocol provides background information concerning the Sciatica-trial, which is the source of the entire study population. The next chapter describes the scanning protocol used in obtaining all MR-images, together with a list of all MRI-parameters that will be assessed. The remaining chapters discuss in more detail the various research questions and their background. 46

48 2. The Sciatica Trial, design and patient population The Sciatica Trial is a multicenter, randomized cost-effectiveness trial concerning the timing of surgery for sciatica caused by a lumbar disc herniation. Patients (18-65 yr.) with a lumbosacral radicular syndrome with duration of less than 12 weeks are eligible for this trial. Patients are included after a visit to the neurological outpatient clinics. During the first visit to the neurological outpatient clinic the patient s history is taken and a neurologist performs a standardized neurological examination. All patients eligible for inclusion are seen by a research nurse, who provides extensive information on the trial and the consequentions of participation. After a positive decision to participate, patients are again seen by a research nurse, who performs a standard neurological examination and fills out the forms needed for evaluation of outcome measures and baseline variables. An MRI-scan and lumbar spine X- ray are scheduled. A neurosurgeon checks if the MRI confirms the presence of unilateral disc herniation and the patient is eligible according to the inclusion and exclusion criteria described in table 1. Inclusion criteria Patient age years. Persistent radicular pain in the L4, L5 or S1 dermatome with or without mild neurological deficit. Intensity of leg pain more than 40 mm measured on the 100 mm horizontal VAS. Leg pain of 6-12 weeks duration. Evidence of a unilateral disc herniation confirmed on MRI. Informed consent. Exclusion criteria Cauda equina syndrome or severe paresis (MRC<3). Complaints of a lumbosacral radicular syndrome in the same dermatome within the past 12 months. Low back surgery in the past. Significant central canal stenosis, degenerative or lytic spondylolisthesis. Pregnancy. Severe life-threatening or psychiatric illness. Insufficient knowledge of Dutch language. Plans to move to another country in the year after randomization. Contra-indications for MRI. Table 1. In- and exclusion criteria Sciatica Trial The MRI data is registered by the attending radiologist and neurosurgeon using a standardized Case Record Form (CRF). Patients are not informed on the results of the MRI-scan. 47

49 After registration of the baseline variables, patients in the Sciatica Trial are randomized into two groups, receiving either lumbar disc surgery or conservative management. Conservative management consists of a wait-and-see policy conducted by the general practitioner (GP). The GP provides information on the favorable prognosis of LSRS. The treatment of LSRS is aimed primarily at pain relief and maintenance/restoration of normal day-to-day activities. The GP s encourage patients to continue their normal day-to-day activities as far as possible. Analgesic medication is prescribed if necessary. The doctor periodically checks the efficacy of the prescribed pain medication and may adjust the dose or sort of analgesics. Paracetamol is the first choice, if not effective, NSAID s (ibuprofen, diclofenac or naproxen) are prescribed. Only in case of severe disabling pain, morphine is given for a restricted period of time. By preference all analgesics should be taken at fixed times of the day rather than on an if necessary basis. If the GP and the patient conclude that there is considerable kinesiophobia (fear that movement will increase pain), the aid of a physiotherapist can be recommended. The GP is free in her/his choice of prescription of medication and referral to physiotherapists. The research nurse will register the conservative management strategy after communication with the responsible GP. In case of progressive neurological deficit or worsening intolerable pain the GP can refer the patient back to the research nurse or neurosurgeon. If, six months after randomization, the patient has still not improved or suffers from intermittent LSRS, surgical treatment is offered. Patients might ask for earlier surgery because of worsening leg pain. In these crossover cases and in the case of a progressive neurological deficit, surgery is performed in consultation with the patient. Lumbar disc surgery (group B) is performed in the conventional manner with loupe magnification. All participating surgeons are certified neurosurgeons with experience in the standard approach with loupe magnification. Surgery takes place as soon as possible with a maximum of two weeks after randomization. Hospital admission is 4-6 days, including the day of surgery. During the post-operative period the patients are mobilized with the aid of a physiotherapist. At home patients visit their own physiotherapist who guides them and provides exercises 2 times a week for 6 weeks. If, at their last visit to the outpatient clinic 6-8 weeks post-operatively, patients still complain of persisting leg pain or aggravated back pain, additional physiotherapy sessions are prescribed by the neurosurgeon. Research nurses, who are thoroughly trained in neurological examination, will conduct follow-up examinations in both groups at 4, 8, 26, and 52 weeks after randomization. 2.1 Measured variables and data acquisition in the Sciatica Trial In the Sciatica Trial, various clinical and demographic parameters are recorded for all patients, as shown in Table 2. Physical examination and clinical interview Weight Length Well-leg lifting Restricted lumbar motion Painful lumbar motion Crossover pain Kemp's sign Lasegue's sign (Straight Leg Raising test) Crossed Lasegue's sign Bragard's sign Trendelenburg's sign Knee tendon reflex Achilles tendon reflex Muscle strength, MRC graded (M. Tibialis anterior, M. Peronei, M. Extensor hallucis longus, M. Triceps surae) Sensory exam Demographic parameters Fingers-to-ground distance Sciatica history Time since first onset of leg pain Localization of leg pain Nature and frequency of leg pain Previous episodes of leg pain Sensory deficits Muscle weakness Signs of cauda equina syndrome Low back pain history Time since first onset of leg pain Time since first onset of low back pain Disability / time off work Traumatic onset of complaints Factors provoking leg pain Pain medication Complications 48

50 Age Sex Occupation Marital status Highest level of education Children Recreational activities Questionnaires Roland Disability Questionnaire for Sciatica. Perceived recovery (7-point Likert scale) VAS for leg pain McGill Questionnaire for Pain Short-Form 36 Sciatica Frequency and Bothersome Index. VAS for low back pain Table 2. Clinical parameters, demographic parameters and questionnaires collected at baseline and follow-up in the Sciatica Trial 2.2 MR-imaging in the Sciatica Trial All patients in the Sciatica Trial undergo MR-imaging at the time of the initial diagnosis of sciatica and after one year of follow-up. The MRI is performed uniformly in all hospitals using the following scanning protocol: 1. Sagittal T1 spin echo (4 mm coupes; interslice gap 0,4 mm) 2. Sagittal T2 turbo spin echo (4 mm coupes; interslice gap 0,4 mm) 3. Transversal T1 spin echo (4 mm coupes; interslice gap 0,4 mm) 4. Transversal T2 TSE Step 1 and 3 are repeated after intravenous injection of Gadolineum. 3. MRI-findings in lumbar disc herniation - materials and methods 3.1 Patient population 49

51 As previously described, the patient population (and their MR-images) for this MR-imaging study will be derived from the Sciatica Trial. Not all patients eligible for inclusion were randomised, due to various reasons independent of in- or exclusion criteria. However, a complete set of baseline variables, including an MRI-scan, was collected for this group of patients (n=...) during the intake period. These patients were not randomized and therefore did not undergo the full follow-up regime. However, they fit all inclusion criteria and can therefore, at baseline, be considered equal to patients from group 1 and 2. The total patient population can be divided into 4 main subgroups: 1. Randomized patients with low back pain and sciatica (n=69) 2. Randomized patients with sciatica, but without low back pain (n=214) 3. Non-randomized patients that fit all inclusion criteria, with low back pain and sciatica (n=40) 4. Non-randomized patients that fit all inclusion criteria, with sciatica, but without low back pain (n=71) After randomization the first two subgroups can be divided into four new subgroups: 1A. Patients with low back pain and sciatica, randomised to lumbar disc surgery (n=36) 1B. Patients with low back pain and sciatica, randomised to conservative treatment (n=33) 2A. Patients with sciatica, but without low back pain, randomised to lumbar disc surgery (n=105) 2B. Patients with sciatica, but without low back pain, randomised to conservative management (n=109) Low back pain is considered present when the patient has a score of at least 20 mm on the 100 mm horizontal VAS for low back pain. Depending on the research question under assessment, a selection of patients from one or more of these subgroups will be included for analysis. 3.2 Assessment of MRI-parameters All MR-images will be collected and subsequently assessed without additional clinical information by four spine experts (two neuroradiologists (PB and GL) and two neurosurgeons (WP and x), none of whom have been involved in the initial radiological assessment during the diagnostic phase. Morphometric parameters will be assessed by PB and BS. In case two scores have different opinions during their assessment of MRI parameters, they will first attempt to reach agreement. A senior neuroradiologist (JW) will act as supervisor and will make the final judgement in case of remaining disagreement between scorers. Inter-observer variability will be assessed in a separate study by DB. All data are registered using standardized Case Record Forms (CRF) (appendix I and II). The MRI-parameters under investigation can be divided into qualitative- and morphometric parameters. The qualitative parameters present on the first CRF (appendix 1) will be assessed by all scorers. The morphometric parameters present on the second CRF (appendix II) will be assessed only by PB and BS. Hardcopy MR-images are digitized prior to assessment. All digital MR-images are collected and labeled in accordance with each patient's unique trial number, and stored together on external hard disk drives. Assessment of all MRI parameters will take place using uniformly configured Apple Powerbook laptops equipped with OsiriX medical imaging software. These laptops will be shared between scorers MRI-parameters assessed by all scorers Location of disc herniation 50

52 The location of each lumbar disc herniation is assessed on both sagittal and axial images and classified in concordance with the criteria of the ASNR task force (central, subarticular, intraforaminal, extraforaminal). In addition, central disc herniations are divided into central-median and centralparamedian herniations. Properties of disc herniation The properties of each disc herniation are described in concordance with the criteria of the ASNR task force (focal/broad based, protrusion/extrusion, contained/non-contained, migration, sequestration). Compression of nerve root and thecal sac Compression of nerve root and thecal sac in the spinal canal are assessed on axial T2 images and graded according to the following criteria (adapted from Wilmink 29 ): 1. Decrease in area of the dural sac (AD) is measured and rated on a four-point scale. 2. Flattening of the ventrolateral angle of the dural sac (AN1) or the emerging root sheath (AN2), rated on a five-point scale. 3. Disappearance of epidural fat adjacent to the dural sac (FA1) or surrounding the nerve root sheath (FA2) 4. Nerve root swelling distal to the site of compression (NC), rated on a three-point scale. AD: At the site of compression the area of visible dural sac is measured. Using area measurements obtained at adjacent levels, the average reduction percentage is calculated. Score 1 for reduction by 0-19%; 2 for reduction by 20-29%; 3 for reduction by 39-40%; 4 for reduction by 40% or more AN1 and AN2: In case of flattening of the ventrolateral angle of the dural sac (AN1), score 1 for a rounded ventrolateral aspect of the dural sac; score 3 when ventrolateral angle of dural sac is no longer rounded, but adjacent dural surfaces forming angle are flat or outwardly convex; score 5 when ventrolateral angle is sharp and adjacent dural surfaces are inwardly convex. Scores 2 and 4 are awarded in transitional cases. In case of AN2 score 1 when root sheath is not deformed, 3 when flattened, 5 when obliterated. Score 2 or 4 in transitional cases. FA1 and FA2: Fat is normally present at level of disk and intervertebral foramen, usually absent at level of lateral recess. Score 1 when fat is present, 3 when fat is unilaterally diminished or bilaterally absent at levels where fat should normally be present. Score 2 in transitional cases. NC: Score 1 when root in question is equal or smaller in size than its contralateral counterpart, score 3 when larger. Score 2 in transitional stages. Do not score in case of bilateral lesions, as symmetrical roots could in that case be due to bilateral swelling. Gadolineum enhancement Gadolineum enhancement of the intervertebral disc, nerve root and surrounding tissues is assessed on axial T1-images for each level (and side, if applicable). 51