Original Research Article Clinical Course and Prognostic Factors in Acute Neck Pain: An Inception Cohort Study in General Practice

Transcription

1 PAIN MEDICINE Volume 9 Number SPINE SECTION Original Research Article Clinical Course and Prognostic Factors in Acute Neck Pain: An Inception Cohort Study in General Practice Cees J. Vos, MD, PhD,* Arianne P. Verhagen, PhD,* Jan Passchier, PhD, and Bart W. Koes, PhD* Departments of *General Practice and Medical Psychology, Erasmus Medical Center, University Medical Center Rotterdam, Rotterdam, The Netherlands ABSTRACT Objective. To describe the natural course of patients with acute neck pain presenting in general practice and to identify prognostic factors for recovery and sick leave. Design. We conducted a prospective cohort study with a 1-year follow-up in general practice. Questionnaires were collected at baseline and after 6, 12, 26, and 52 weeks. Days of sick leave were dichotomized into two groups: below and above 7 days of sick leave. Logistic regression was used to identify prognostic factors for recovery and sick leave. Patients. Consecutive patients with nonspecific neck pain lasting no longer than 6 weeks were invited to participate. Results. One hundred eighty-seven patients were included and 138 (74%) provided follow-up data. After 1 year, 76% of the patients stated to be fully recovered or much improved, although 47% reported to have ongoing neck pain. Almost half of the patients on sick leave at baseline returned to work within 7 days. Multivariate analysis showed that the highest association with recovery was the advice of the general practitioner (GP) to wait and see (odds ratio [OR] 6.7, 95% confidence interval [CI] ). For sick leave, referral by the GP, for physical therapy or to a medical specialist, showed the highest association (OR 2.8, 95% CI ). Conclusion. Acute neck pain had a good prognosis for the majority of patients, but still a relatively high proportion of patients reported neck pain after 1-year follow-up. The advice given by the GP to wait and see was associated with recovery, and referral was associated with prolonged sick leave. Key Words. Neck Pain; Clinical Course; Prognosis; General Practice; Follow-Up Introduction Neck pain is a common condition in the general population affecting many people at some point in life. Point prevalences in the general Reprint requests to: Cees J. Vos, MD, PhD, Department of General Practice, Erasmus Medical Center, University of Rotterdam, PO Box 1738, 3000 DR Rotterdam, The Netherlands. Tel: ; Fax: ; c.vos@erasmusmc.nl. population range from 10% to 15% [1]. In the literature, 12-month-period prevalences as high as 40% have been reported [1]. In most studies, prevalence figures for women are higher than for men [2 4]. A higher prevalence in women occurs both in the general population and in selected populations [5]. For both men and women, the prevalence of neck pain increases with age until the age of 50, after which it no longer increases [6,7]. American Academy of Pain Medicine /08/$15.00/ doi: /j x

2 Clinical Course of Acute Neck Pain 573 Neck pain is not life-threatening, but it affects activities of daily living and can, therefore, have a major impact on the quality of life [1]. Apart from personal suffering, neck pain often necessitates sick leave and is, therefore, responsible for substantial costs to society [8]. Neck pain is one of the major musculoskeletal complaints for which health care is sought, and contributes substantially to the workload in general practice. In The Netherlands, neck pain accounts for up to 1% of consultations with general practitioners (GPs) [9]. In general, information is lacking on the clinical course of acute neck pain in primary care. In their systematic review, Borghouts et al. found 21 studies on the clinical course of acute and chronic neck pain [10]. Twelve studies concerned a secondary setting; eight studies were from an occupational setting, but only one was carried out in a primary care setting [10]. They reported a lack of information on the course of acute neck pain in primary care. Prevailing views differ. Some regard neck pain as a self-limiting condition, from which recovery occurs in most cases without any medical treatment [1]. On the other hand, according to population surveys, 5% of individuals report disabling neck pain [11,12]. A number of studies have explored prognostic factors for neck pain [10,13,14]. Most frequently reported prognostic factors are age, gender, pain severity, a history of neck pain, concomitant low back pain, duration of pain, occupation, previous trauma, and degenerative changes on X-ray [1,13,14]. Identifying such factors serves to distinguish patients with a good prognosis from those with a less favorable prognosis, so that health care attention can be focused on those with an expected slower recovery [15]. Relationships between chronic neck pain, on the one hand, and physical, occupational, and psychological risk factors, on the other, have been investigated in many studies. Some have focused on certain occupations, such as office work, assembly line work, and industrial work including heavy labor [1,3,16]. These and other studies show that work-related, as well as non-work-related, psychosocial risk factors are important determinants for chronicity [9]. It is disputed whether a motor vehicle accident is a relevant factor for the development of chronic neck pain. Some studies found that the prevalence of chronic neck pain is associated with a history of injury and state that it is a distinct and separate risk factor apart from other predictors of chronic neck pain [17,18]. The primary aim of our study was to describe the clinical course of patients with acute neck pain in general practice in a 1-year follow-up study. The secondary aim was to identify prognostic factors for self-perceived recovery and prolonged sick leave. Methods Study Population GPs working in the city of Rotterdam and its suburban region were invited to participate in this study. The study design was a prospective cohort study with a follow-up period of 1 year. A priori, we aimed to include 200 patients with acute neck pain. A generally accepted, time-based classification of neck pain is threefold: acute (0 6 weeks), subacute (6 12 weeks), and chronic (>3 months) [19]. Accordingly, we invited to participate in the study patients who had neck pain lasting no longer than 6 weeks. This could be neck pain for the first time or recurrent neck pain after a period free of pain of at least 3 months. Additional inclusion criteria were age above 18 years and sufficient knowledge of the Dutch language to be able to complete written questionnaires. Excluded were patients with specific causes of neck pain (e.g., known vascular or neurological disorders, neoplasms, rheumatic conditions, or referred pain from internal organs). A patient could be included only once during the follow-up period. After oral consent, the GP handed over an envelope containing the baseline questionnaire, a patient information form concerning the content of the study, an informed consent form, and a prepaid return envelope. Only after having returned a completed baseline questionnaire as well as a written informed consent were patients included in the study. Approval for this study was obtained from the Ethical Committee of the Erasmus University Medical Center. Nonresponders Nonresponders were defined as patients who were approached by their GP to participate but did not enrol. At baseline, the GPs completed a short form to describe all patients whom they asked to participate in the study. This covered date of birth, gender, reported cause of neck pain, outcome of physical examination, diagnosis, and proposed diagnostic and treatment modalities. We asked the

3 574 GP to submit the short forms immediately after the visit. Explanatory Variables The baseline questionnaire contained questions on age, gender, employment status, previous history of neck pain and previous treatments (if any), duration and self-reported cause of current neck complaints, previous and concomitant headache, smoking habits, and sudden onset of the neck pain. The patients were asked if they were on sick leave because of neck pain. They were also asked which treatments, advice, and referrals were given by the GP. The patients scored their average severity of neck pain on an 11-point, numerical pain rating scale (NRS), ranging from 0 (no pain) to 10 (unbearable pain). The NRS validity and reliability were well established [20]. Patients completed the Neck Disability Index (NDI) as well as the Acute Low Back Pain Screening Questionnaire (ALBPSQ). The NDI is a 10-item disability questionnaire containing questions on three different domains: pain intensity (neck pain, headache), work-related activities (work, lifting, and concentration), and non-work-related activities (personal care, reading, driving, sleeping, and recreation) [21]. Patients chose one out of six answer categories for each item describing the degree of disability from 0 (no activity limitation) to 5 (major activity limitation). After summing all items, the total score ranged between 0 and 50. The ALBPSQ is a biopsychosocial screening instrument [22]. The questionnaire is composed of 20 items divided in five domains. The questions deal with background, physical functioning, fearavoidance beliefs, work, the experience of pain and reactions to pain, and miscellaneous. Patients scored their answers on an 11-point rating scale, generating a total score between 0 and 200 points. The patients received a follow-up questionnaire 6, 12, 26, and 52 weeks after enrollment. The patients were asked if they still experienced neck pain or had a recurrence, and to rate their current pain level on the NRS. The patients judged for themselves whether their current pain should be classified as ongoing or recurrent. We asked if the patient was still on sick leave due to neck pain and, for those who resumed work, how many days they had been on sick leave. The patients also completed the NDI and ALBPSQ. When a successive questionnaire was not returned within 2 weeks, the patient received a written reminder, followed by an additional telephone call 2 weeks later. Vos et al. Outcome Measures Recovery and sick-leave days were chosen as outcome variables. Both outcome measures have been used in methodologically similar studies [13,14,18,22]. The follow-up questionnaires contained questions about the total days of sick leave. The patients rated their perceived recovery on a separate seven-point ordinal scale. The scale ranged from 1 (complete recovery) over 4 (no change) to 7 (my complaints are worse than ever). The perceived recovery scale has been frequently used in outcome studies and has been validated [23,24]. Statistical Analysis Frequencies, means, and standard deviation (SD) were determined for the explanatory variables, the outcome variable, and for each item and total scores on the NDI and ALBPSQ. Descriptive statistics were used to calculate the frequencies and means of the outcome variables: sick leave and recovery. Differences between responders and nonresponders were assessed with a Student t-test for independent samples. Nonresponders were defined as patients who were approached by their GP but decided not to participate. Means of the NRS scores were presented graphically. Frequencies on the perceived recovery scale and for days on sick leave, in every follow-up questionnaire, were calculated and graphically presented. Patients were classified as recovered if they scored 1 ( I am completely recovered ) or 2 ( I am much improved ) on the perceived recovery scale. Other scores were considered as not recovered. The sick-leave score was the total of the selfreported days absent from work due to neck pain, during the follow-up year, whether it was continuous or recurrent. After data collection, we dichotomized days of sick leave into two groups: less than 7 days of reported sick leave and otherwise 7 days or more. We agreed that 7 days of sick leave was appropriate as a criterion as this was the median time-off for those taking sick leave. Univariate and multivariate logistic regression analysis was performed with recovery at 12 months and sick leave as the outcome measures. Explanatory variables were checked for possible interdependent correlations by means of a correlation matrix. Variables with a correlation factor above 0.90 were removed. After univariate analysis, variables with a P < 0.1 were entered in a final multivariate regression model by means of the

4 Clinical Course of Acute Neck Pain 575 backward Wald method. Odds ratios (ORs), 95% confidence interval (CI), and beta values were calculated [25]. Negative or positive beta values refer to a negative or positive relation between an individual variable and outcome. The proportion of correctly classified patients with the final model was calculated. We corrected for age and gender by including them in the final analysis. Nagelkerke R square represents the explanatory variance of the model. Whether nonresponse due to dropouts during the follow-up year was selective and caused bias was evaluated separately. We imputed the last known data on recovery of every dropout in the final follow-up outcome data. Imputing in this way is known as the last measurement carried forward procedure. Univariate and multivariate logistic regression analysis were performed with available cases and also with imputed data. Imputing will generally result in more conservative results. A higher number of patients will result in less wide CIs in the final model. All statistical analyses were carried out using the SPSS version 10.0 (SPSS Inc., Chicago, IL) for Windows program. Results Twenty-nine GPs enrolled patients during the recruitment period from March 2001 until August Two hundred forty-nine patients with acute neck pain were asked by their GP to join the study and were given the information packet. One hundred ninety patients (76%) responded and returned the baseline questionnaire and the signed informed consent form. Three patients did not meet the inclusion criteria and were excluded (two patients had chronic neck pain and one patient was too young). Finally, 187 patients formed our inception cohort. Patient characteristics at baseline are presented in Table 1. The patients were predominantly younger females. Most patients had experienced neck pain episodes before (63%) and had received previous treatments for this complaint. Mean duration of neck pain at baseline was 16 days (SD 13.1). Pain at multiple sites was common (81% had pain at one or more additional sites). Motor vehicle accidents were the self-reported cause of neck pain in a substantial proportion of cases (23%). There were significantly more male (51% vs 36%, P < 0.05) nonresponders (N = 59). Although nonresponders were on average younger (36.8 vs 40.0 years), age as well as the other variables Table 1 Patient characteristics of the study population at baseline (N = 187) N (percentages) Mean age Gender Female 119 (64) Male 68 (36) Employed 148 (79) Had previous episodes of acute 118 (63) neck pain Underwent previous treatment for 74 (40) neck pain Duration of acute neck pain shorter 79 (42) than 2 weeks Pain radiating to back 10 (5) Neck pain accompanied by 117 (62) headache Sudden onset of neck pain 88 (47) On sick leave due to neck pain 53 (28) Onset of neck pain Spontaneously/unknown 70 (38) Due to a motor vehicle accident 42 (23) Noticed after waking up 32 (17) After a fall or hitting the head 13 (7) Sudden onset 12 (6) Stress 10 (5) Work related 8 (4) SD = standard deviation. that were taken into account did not differ significantly. At the 1-year follow-up, 138 patients (74%) participated. Almost half of the initial cohorts (47%) still experienced neck pain, and 5.6% of the patients reported a recurrence. The scores on the perceived recovery scale during the follow-up year are presented in Figure 1. Twenty-six percent of the patients stated at 6 weeks after baseline that they regarded themselves to be completely recovered. During the rest of the follow-up year, an additional 20% reported themselves to be recovered. The scores regarding sick leave during the follow-up year are presented in Figure 2. At baseline, almost one-third (N = 52) of the employed patients (N = 148) reported to be on sick leave. Almost half of the patients on sick leave returned to work within 7 days. At baseline, sick leave in the male patients was somewhat higher than in the female patients (38% vs 33% of employed patients). After 1 year, none of the men and only four women (8%) were still absent from work. The duration of the self-reported periods of sick leave was up to 1 week for 37% of patients, between 1 week and 1 month for 22%, between 1 and 3 months for 20%, and above 3 months for 21% of patients. Mean scores on the NRS are presented in Figure 3. At baseline, the mean score on the NRS SD

5 576 Vos et al. % 50 primary outcome: recovery recovered much improved baseline 6 weeks 12 weeks 26 weeks 52 weeks 35 primary outcome: sick leave 28 baseline 6 weeks 12 weeks 26 weeks 52 weeks n=52 n=26 n=14 n=7 n=4 was 6.4 (SD 2.0); scores for pain were not significantly different between those patients who eventually recovered and those who did not. After 1 year, the mean scores dropped for non-recovered patients to 5.3 (SD 1.6) and for recovered patients to 2.9 (SD 1.8). Some of those who rated themselves as much improved still reported to experience pain, although of much lower levels. After 1-year follow-up, only five patients scored a 4 or higher on the NRS. The mean total score on the NDI at baseline was 14.4 (SD 6.5), and at the end of follow-up was 4.5 (SD 4.3) and 14.3 (SD 6.5) for recovered and Pain numeric rating scale (NRS) baseline 6 weeks 12 weeks 26 weeks 52 weeks n=187 n=155 n=141 n=132 n= % Figure 1 Percentages of patients stating at successive measurement points to be recovered or much improved. Figure 2 Percentages and number of employed patients who reported to be on sick leave during the follow-up year. non-recovered patients, respectively. At baseline, scores for disability were somewhat different (12.2 vs 16.1) but not significantly between those patients who eventually recovered and those who did not. For the ALBPSQ, the baseline score was 71.3 (SD 32.2) and after 1 year for recovered and non-recovered patients, 49.8 (SD 28.2) and 87.9 (SD 25.4), respectively, but not significantly between those patients who eventually recovered and those who did not. Univariate regression analysis revealed 12 items that were significantly correlated with recovery after 12 months, and after multivariate non-recovered much improved recovered Figure 3 Mean score on the numerical pain rating scale for recovered, much improved, and non-recovered patients during the follow-up year.

6 Clinical Course of Acute Neck Pain 577 Table 2 Prognostic factors significantly associated with recovery after 1 year by univariate (P < 0.1) and multivariate (P < 0.05) logistic regression analyses (N = 138) Univariate analysis Item Beta/OR (90% CI) Beta/OR (95% CI) Female gender ( ) ( ) Had neck pain before ( ) Treated by physiotherapist before ( ) Treated by manual therapist before ( ) Severe initial pain ( ) Accompanying headache ( ) Pain in the upper part of the neck ( ) ( ) Pain radiating to the back ( ) ( ) Duration of complaints >2 weeks ( ) ( ) GP advised to wait and see ( ) ( ) GP advised to improve posture ( ) GP prescribed medication ( ) GP instructed in physical exercises ( ) Total score on the NDI ( ) Total score on the ALBPSQ ( ) analysis, five items remained (see Table 2). The explanatory variance of the model was 38%. The final model correctly classified 83% of patients. The highest OR for recovery was the advice given by the GP at baseline to wait and see for an expected favorable natural course (OR 6.7). In the final model, female gender, radiating pain to the upper part of the neck, radiating pain to the back, and duration of complaints longer than 2 weeks at baseline all had a negative association with outcome, meaning that their presence diminished the chance of recovery. After 1 year, data on sick leave were available for 109 employed patients (74%). Ten items were significantly associated with prolonged sick leave in Multivariate analysis OR = odds ratio; CI = confidence interval; GP = general practitioner; NDI = Neck Disability Index; ALBPSQ = Acute Low Back Pain Screening Questionnaire. the univariate regression analysis. After multivariate analysis, five items were significantly related to prolonged sick leave (Table 3). The explanatory variance of the model was 38%. The final model correctly classified 79% of patients with prolonged sick leave. For sick leave, referral by the GP had the highest positive association (OR 2.8) followed by the GP made a follow-up appointment (OR 1.7). Imputing data of 30 patients (16%) in a last measurement carried forward procedure did not reveal significant differences in the outcome recovery as well as in the outcome sick leave. Therefore, we only present data on available cases and none after imputation of data. Table 3 Prognostic factors significantly associated with sick leave by univariate (P < 0.1) and multivariate (P < 0.05) logistic regression analysis (N = 109) Univariate analysis Multivariate analysis Item Beta/OR (90% CI) Beta/OR (95% CI) Had neck pain before ( ) ( ) Had previous headache ( ) Had physiotherapy for neck pain before ( ) Caused by a motor vehicle accident ( ) Severe initial pain ( ) GP made a follow-up appointment ( ) ( ) GP advised to improve posture ( ) GP prescribed medication ( ) GP instructed in physical exercises ( ) GP advised to stop working ( ) Referral by GP for treatment ( ) ( ) Patients did exercises for the neck ( ) ( ) Total score on NDI ( ) ( ) Total score on ALBPSQ ( ) OR = odds ratio; CI = confidence interval; GP = general practitioner; NDI = Neck Disability Index; ALBPSQ = Acute Low Back Pain Screening Questionnaire.

7 578 Discussion This study has some limitations. The dropout number during follow-up was acceptable and we believe that it did not influence the results we presented, although it is possible that the dropouts were more severe cases that did not resolve or whose pain recurred, making the proportion of improvement look better than it actually was. On the other hand, if the dropouts were patients who recovered completely and did not bother to participate anymore, the proportion of recovery may have been underestimated. We have surveyed the dropouts for recovery as best as we had available data but found no indication for either one of the posed biases. The study size was moderate, resulting in a restriction of the number of prognostic variables that could be used in the final analysis. Another limitation of our study is that the actual cause of neck pain was not determined. It could be that different prognostic factors are associated with different causes of neck pain. However, in the majority of patients, the cause of neck pain cannot be ascertained by history, physical examination, or even by conventional medical imaging. In primary care, it is not practical to determine the cause of acute neck pain nor is it generally useful to do so. Investigations such as discography and diagnostic blocks may be appropriate for patients with persisting disabling pain, but are inappropriate for patients with acute neck pain who might still recover. Therefore, it is unlikely that the influence on recovery of the actual cause of acute neck pain will ever be determined. Previously published data on the natural course of acute neck pain provide an incomplete picture [1,10,26]. Although almost half of the patients recover or are much improved within the first week, the subsequent of the others is unknown. Some studies report that the percentage of people in whom neck pain becomes chronic, with at least mild to moderate symptoms, is about 10% [9,17], but more recent studies report higher percentages of patients still having neck pain after one or more years of follow-up [26]. In our study, 47% of patients (38% of men and 51% of women) still experienced (some) neck pain after 1 year. The 1-year prevalence of neck pain in literature ranges from 18% to 66% [12,18,27,28]. The recurrence rate we found was rather low. We defined neck pain as a new episode after a 3-month period free of neck pain. This period is rather arbitrary given the high percentage of patients that reported to have experienced one or Vos et al. more episodes of neck pain before. So we have to be cautious to draw a conclusion on recurrence rates. Neck pain is currently thought to run a recurrent and intermittent course rather than a picture of relentless or continuous symptoms [18]. There was a decrease in reported sick leave indicating that the majority of patients was only moderately affected by their neck pain. In general, the number of patients with musculoskeletal complaints is three to four times higher than the number of patients on sick leave [15,29]. Our study is no exception to that rule. In occupational settings, the same proportion is found in almost all regional musculoskeletal pain sites. This finding suggests that in a prospective cohort study in general practice as well as in occupational settings, patients continue their regular work while experiencing musculoskeletal pain. We identified several factors of prognostic value for recovery. In particular, the advice given by the GP to wait and see for an expected favorable course appeared to be a strong predictor. No other study in neck pain has reported this factor. However, the basis for this association cannot be derived from our data. An ideal interpretation would be that the predictive power of the GP s advice is due to the convincing strength of the advice itself, but the competing interpretation is that the GPs provided this advice to those patients with simpler or less severe complaints, who were destined to recover. Our data do not provide subgroups of sufficient size to test these interpretations. Being female turns out to be a negative prognostic factor and has been reported before [2 4]. Why females have a worse prognosis remains unexplained. Five factors correlated in the multivariate analysis with prolonged sick leave. Two of them concerned actions of the GP. Referral by the GP is the strongest predictor. The act of referral implies that GPs identified patients with more severe, or more complicated, problems, with the expectation that physical therapy or specialist care might improve outcome. Our data show that referral did not improve outcome, but they do not show why it failed. It may be that referral is intrinsically deleterious, in that it convinces patients that their condition is more serious than it really is, in which case more concerted explanation and reassurance by the GP is warranted, or it may be that treatment at all levels of specialty is simply ineffective for severe or complex neck pain.

8 Clinical Course of Acute Neck Pain 579 Another significant prognostic factor for sick leave was the GP s initiative to make a follow-up appointment. In general, the GP in The Netherlands does not make follow-up appointments at the end of the consultation. Normally, the consultation ends with a general remark like please return when your complaint does not resolve on its own within 14 days. Under these conditions, making an appointment implies that the GP is concerned that the patient may not recover in the usual manner. Thus, the patient s presenting features, rather than the act of making an appointment, may underlie the poor prognosis. Our data on presenting features are insufficiently detailed to explore what the underlying clinical cues might have been in this regard. Previous periods of neck pain turned out to be a significant prognostic factor with a slight negative correlation with prolonged sick leave meaning that it is associated with less sick leave. Other authors reported that previous complaints of the neck (when medical care was sought) were significant predictors for future sick leave due to these complaints [14,30,31]. The possibility that performing exercises for the neck, on the patients own initiative, has a positive influence on sick leave is an interesting finding. Perhaps it reassures the patients in the conviction that they can overcome the problem on their own. We, however, need to be cautious in drawing conclusions about effectiveness in an observational study. In order to compare outcomes, it is important to know what recovered means to patients and investigators. Patients probably consider themselves recovered when they are fully cured of their problem. In studies of treatment, much improved is often combined with recovered. Our data show that patients consider themselves much improved, not necessarily fully recovered, despite still having some pain. At issue is what operational definition of recovered is used. That definition is not necessarily the same as what the patients mean by it. This finding has implications for reviewing literature. If outcomes are to be compared, investigators should clearly define their operational criteria; otherwise, it hampers comparing results of different studies. Another finding in this study is that different prognostic factors appear to be relevant depending on what kind of outcome measure is chosen. Reporting the different prognostic factors in outcome research can only be carried out in direct conjunction with the used outcome measure. In that respect, we agree with the suggestion of Kjellman et al. that researchers should take into consideration that the prognostic factors that appear in an analysis are clearly associated with the outcome measure that is used [32]. Conclusions This study confirms the finding of others that acute neck pain does not have for all patients the favorable natural course it is generally thought to have. Although patients consider themselves recovered, still having neck pain is a common and seemingly accepted fact for many patients. Most patients reported that the pain intensity is on a substantial lower level and probably hardly of clinical importance. This emphasizes the need in future research for a clear definition of the often used outcome recovery. Different outcome measures are associated with different prognostic factors. In reviewing the literature, one should bear this in mind. The influence of the GP s advices and referrals as a predictive value for recovery and prolonged sick leave is a new finding. Further research to underline the relationship between the GP s actions and outcome is needed. References 1 Ariëns GAM, Borghouts JAJ, Koes BW. Neck pain. In: Crombie IK, ed. The Epidemiology of Pain. Seattle, WA: IASP Press; 1999: Frederiksson K, Alfredsson L, Köster M, et al. Risk factors for neck and upper limb disorders: Results from 24 years of follow-up. Occup Environ Med 1999;56: Skov T, Borg V, Ørhede E. Psychosocial and physical risk factors for musculoskeletal disorders of the neck, shoulders, and lower back in salespeople. Occup Environ Med 1996;53: Webb R, Brammah T, Lunt M, et al. Prevalence and predictors of intense, chronic and disabling neck and back pain in the UK general population. Spine 2003;23: Leclerc A, Niedhammer I, Landre M-F, et al. Oneyear predictive factors for various aspects of neck disorders. Spine 1999;24: Donk van der J, Schouten JSAG, Passchier J, et al. The associations of neck pain with radiological abnormalities of the cervical spine and personality traits in a general population. J Rheumatol 1991;18: Westerling D, Jonsson BG. Pain from the neckshoulder region and sick leave. Scand J Soc Med 1980;8:131 6.

9 580 8 Borghouts JAJ, Koes BW, Vondeling H, Bouter LM. Cost-of-illness of neck pain in the Netherlands in Pain 1999;80: Binder A. Neck pain. Clin Evid 2002;7: Borghouts JAJ, Koes BW, Bouter LM. The clinical course of non-specific neck pain: A systematic review. Pain 1998;77: Brattberg G, Thorsland M, Wikman A. The prevalence of pain in the general population. The results of a postal survey in a county of Sweden. Pain 1989;37: Côté P, Cassidy JD, Carroll L. The factors associated with neck pain and its related disability in the Saskatchewan population. Spine 2000;25: Bot SDM, van der Waal JM, Terwee CB, et al. Predictors of outcome in neck and shoulder symptoms. Spine 2005;30:E Hoving JL, de Vet HCW, Twisk JWR, et al. Prognostic factors for neck pain in general practice. Pain 2004;110: Burdorf A, Naaktgeboren B, Post W. Prognostic factors for musculoskeletal sickness absence and return to work among welders and metal workers. Occup Environ Med 1998;55: Eriksen W, Natvig B, Knardahl S, Bruusgaard D. Job satisfaction as predictors of neck pain: A 4-year prospective study. Occup Environ Med 1999;41: Mäkelä M, Heliovaara M, Sievers K, et al. Prevalence, determinants, and consequences of chronic neck pain in Finland. Am J Epidemiol 1991;134: Croft PR, Lewis M, Papageorgiou AC, et al. Risk factors for neck pain: A longitudinal study in the general population. Pain 2001;93: Merskey H, Bogduk N. Classification of Chronic Pain, 2nd edition. Seattle, WA: IASP Press; Miller MD, Ferris DG. Measurement of subjective phenomena in primary care research: The visual analogue scale. Fam Pract Res J 1993;13: Vernon HT, Mior S. The neck disability index: A study of reliability and validity. J Manipulative Physiol Ther 1991;14: Vos et al. 22 Linton SJ, Hallden K. Can we screen for problematic back pain? A screening questionnaire for predicting outcome in acute and subacute back pain. Clin J Pain 1998;14: Koes BW, Bouter LM, van Mameren H, et al. A randomized clinical trial of manual therapy and physiotherapy for persistent back and neck complaints: Subgroup analysis and relationship between outcome measures. J Manipulative Physiol Ther 1993;16: De Vet HC, Bouter LM, Bezemer PD, Beurskens AJ. Reproducibility and responsiveness of evaluative outcome measures. Theoretical considerations illustrated by an empirical example. Int J Technol Assess Heath Care 2001;17: Norusis MJ. SPSS 11.0 Guide to Data Analysis. New York, NY: Prentice Hall; Gureje O, Simon GE, Von Korff M. A crossnational study of the course of persistent pain in primary care. Pain 2001;92: Hill J, Lewis M, Papageorgiou AC, et al. Predicting persistent neck pain: A 1-year follow-up of a population cohort. Spine 2004;29: Picavet HSJ, Schouten JSAG. Musculoskeletal pain in the Netherlands: Prevalences, consequences and risk groups, the DMC3-study. Pain 2003;102: Croft PR, Macfarlane GJ, Papageorgiou AC, et al. Outcome of low back pain in general practice: A prospective study. BMJ 1998;316: Guez M, Hildingsson C, Nilsson M, Toolanen G. The prevalence of neck pain. A population based study from northern Sweden. Acta Orthop Scand 2002;73: Schiøttz-Christensen B, Nielsen GL, Hansen VK, et al. Long-term prognosis of acute low back pain in patients seen in general practice; a 1 year prospective follow-up study. Fam Pract 1999;16: Kjellman G, Skargren E, Öberg B. Prognostic factors for perceived pain and function at one year follow-up in primary care patients with neck pain. Disabil Rehabil 2002;24: