Secondary Logo

Journal Logo

Identifying the Best Treatment Among Common Nonsurgical Neck Pain Treatments: A Decision Analysis

van der Velde, Gabrielle, DC*†‡§; Hogg-Johnson, Sheilah, PhD; Bayoumi, Ahmed M., MD, MSc†∥**††; Cassidy, J David, PhD, DrMedSc†‡§; Côté, Pierre, DC, PhD*†‡§; Boyle, Eleanor, PhD‡§; Llewellyn-Thomas, Hilary, PhD‡‡; Chan, Stella, MSc*; Subrata, Peter, MSc*; Hoving, Jan Lucas, PhD§§; Hurwitz, Eric, DC, PhD¶¶; Bombardier, Claire, MD, MSc*†∥∥; Krahn, Murray, MD, MSc†∥∥***

doi: 10.1097/BRS.0b013e31816454f8
Supplementary Research Studies

Study Design. Decision analysis.

Objective. To identify the best treatment for nonspecific neck pain.

Summary of Background Data. In Canada and the United States, the most commonly prescribed neck pain treatments are nonsteroidal anti-inflammatory drugs (NSAIDs), exercise, and manual therapy. Deciding which treatment is best is difficult because of the trade-offs between beneficial and harmful effects, and because of the uncertainty of these effects.

Methods. (Quality-adjusted) life expectancy associated with standard NSAIDs, Cox-2 NSAIDs, exercise, mobilization, and manipulation were compared in a decision-analytic model. Estimates of the course of neck pain, background risk of adverse events in the general population, treatment effectiveness and risk, and patient-preferences were input into the model. Assuming equal effectiveness, we conducted a baseline analysis using risk of harm only. We assessed the stability of the baseline results by conducting a second analysis that incorporated effectiveness data from a high-quality randomized trial.

Results. There were no important differences across treatments. The difference between the highest and lowest ranked treatments predicted by the baseline model was 4.5 days of life expectancy and 3.4 quality-adjusted life-days. The difference between the highest and lowest ranked treatments predicted by the second model was 7.3 quality-adjusted life-days.

Conclusion. When the objective is to maximize life expectancy and quality-adjusted life expectancy, none of the treatments in our analysis were clearly superior.

A decision-analytic model was used to identify the best treatment among common nonsurgical neck pain treatments. With the objective to maximize (quality-adjusted) life expectancy, no treatment was clearly superior.

From the *Institute for Work & Health, Toronto, Canada; †Department of Health Policy, Management and Evaluation, University of Toronto, Toronto, Canada; ‡Division of Health Care and Outcomes Research, Toronto Western Research Institute, Toronto, Canada; §Centre of Research Expertise in Improved Disability Outcomes, University Health Network Rehabilitation Solutions, Toronto Western Hospital, Toronto, Canada; ¶Department of Public Health Sciences, University of Toronto; ∥Centre for Research on Inner City Health, Keenan Research Centre in the Li Ka Shing Knowledge Institute of St. Michael’s Hospital; **Department of Medicine, University of Toronto; ††Division of General Medicine, St. Michael’s Hospital, Toronto, Canada; ‡‡Department of Community and Family Medicine, Dartmouth Medical School, Dartmouth College, Hanover, NH; §§Coronel Institute of Occupational Health and Research Centre for Insurance Medicine, Academic Medical Centre, Universiteit van Amsterdam, Amsterdam, The Netherlands; ¶¶Department of Public Health Sciences, John A. Burns School of Medicine, University of Hawaii at Mᾱnoa, Honolulu, HI; ∥∥Toronto General Research Institute, Division of Clinical Decision Making and Health Care, Toronto, Canada; and ***Faculty of Pharmacy, University of Toronto, Toronto, Canada.

The manuscript submitted does not contain information about medical device(s)/drug(s).

Provincial funds were received in support of this work. No benefits in any form have been or will be received from a commercial party related directly or indirectly to the subject of this manuscript.

Supported by the Canadian Institutes of Health Research through a Fellowship Award (to G.v.d.V.), Canadian Institutes of Health Research through a New Investigator Award (to P.C.), Canada Research Chair in Knowledge Transfer for Musculoskeletal Care (to C.B.), and F. Norman Hughes Chair in Pharmacoeconomics (to M.K.).

The views expressed in this publication are the views of the authors and do not necessarily reflect the views of the Ontario Ministry of Health and Long-term Care.

Ethics Approval: University of Toronto Ethics Review Board Protocol Reference 18807.

Address correspondence and reprint requests to Gabrielle van der Velde, DC, Toronto Western Hospital, 399 Bathurst Street, Fell Pavilion 4-134, Toronto, ON, Canada M5T 2S8; E-mail:

In Canada and the United States, persons with neck pain generally visit primary care physicians or chiropractors, and are usually treated with nonsteroidal anti-inflammatory drugs (NSAIDs), exercise, or manual therapy.1–6 Physicians typically prescribe medication (most likely a NSAID) or refer to physical therapists for exercise,5,6 whereas most chiropractors typically administer manual therapy (most likely manipulation or mobilization).4,7

Deciding which treatment is best for neck pain is difficult. Each treatment carries trade-offs between potential beneficial and harmful effects. Furthermore, the probability of these effects is uncertain. For example, the evidence for the effectiveness of NSAIDs for relieving neck pain is scant.8,8a NSAIDs may also increase the risk of cardiovascular or gastrointestinal events, although the estimates of increased risk vary widely.9,10 Similarly, evidence about the effectiveness of manipulation is conflicting,8a and published estimates of the risk of stroke with neck manipulation vary substantially.11

Decision analysis is a method for identifying the best treatment based on (1) the outcome (benefits and harms) of a specific treatment and associated probabilities, and (2) patients’ preferences for these outcomes. This method typically synthesizes evidence from various sources.12 Our objective was to use decision analysis to identify the optimal treatment among commonly-used neck pain treatments using evidence from the Bone and Joint Decade 2000-2010 Task Force on Neck Pain and Its Associated Disorders (Neck Pain Task Force).

Back to Top | Article Outline


Decision analysis was used to quantify outcomes associated with 6 weeks of NSAID use, exercise, or manual treatments for nonspecific neck pain neck (i.e., neck pain that is not caused by an identifiable etiological lesion such as a fracture, infection, tumor, inflammatory arthritide, or myelopathy). We assumed the patient’s perspective and quantified beneficial and harmful outcomes in terms of life expectancy and quality-adjusted life expectancy.

Back to Top | Article Outline

Model Structure and Modeling

We built a decision-analytic Markov model that followed a simulated cohort of neck pain patients (TreeAge, Pro Suite 2007, Williamstown, MA). Five treatments were evaluated, including standard NSAIDs (sNSAIDs), selective Cox-2 inhibiting NSAIDs (Coxibs), exercise, mobilization, and manipulation (Figure 1). Model details are available online in a Supplementary Appendix (available online through Article Plus).

Figure 1

Figure 1

The model consisted of health states that represented the beneficial and harmful outcomes of these treatments, including Troublesome Neck Pain, No Troublesome Neck Pain, Myocardial Infarction, Post-Myocardial Infarction, Heart Failure, Minor Stroke Disability, Major Stroke Disability, Surgery for Upper Gastrointestinal Bleed, and Endoscopy for Upper Gastrointestinal Bleed.13 No Troublesome Neck Pain was defined as a Chronic Pain Score Grade of 0 (no neck pain) or I (pain of low intensity and few activity limitations).14–16 Troublesome Neck Pain was defined as grade II (pain of high intensity, but few activity limitations), III (pain associated with high levels of disability and moderate limitations in activities), or IV (pain with high levels of disability and several activity limitations). The model tracked patients over 13 four-week cycles (total 52 weeks). At baseline, all patients had Troublesome Neck Pain and could transition between health states at each cycle. A 1-year follow-up was modeled due to lack of evidence regarding long-term treatment effects. Projected years of remaining lifetime were modeled by assigning a mean life expectancy to each terminal health state at the end of the 1-year follow-up.13,17 Patients were not modeled to cross-over from one treatment to another, nor treated for recurrent episodes.

We modeled a baseline course of neck pain (without treatment) and then modified this course to reflect treatment effectiveness. We also modeled a background risk of adverse events (drawn from rates in the general population) that were modified with estimates of treatment-related risks of harmful events. Patients who received sNSAIDs or Coxibs were at an increased risk of stroke, myocardial infarction, heart failure, and serious upper gastrointestinal bleed (UGIB).9,18–21 Patients who received neck manipulation were assumed to have an increased risk of stroke.11

Life expectancy (measured in years) and quality-adjusted life expectancy (measured in quality-adjusted life years [QALYs]) were used to measure the impact of beneficial and harmful treatment outcomes. The QALY captures health-related gains or losses from reduced or increased morbidity (quality-of-life) and reduced or increased mortality (quantity-of-life) into a single measure, by assigning a quality-of-life weight to each time-cycle spent in a health state.13,17 The weight is based on a preference score, a quantitative expression of the desirability of a health state,17,22 expressed as a numerical anchored at 0, corresponding to Death (presumed to be the least desirable outcome) and 1, corresponding to Good Health (the most desirable outcome).13,17 QALYs were calculated by multiplying the weight associated with each health state by the duration of time spent in that state, then summing this over 1-year. The long-term impact of outcomes on quality-adjusted life expectancy over the projected years of remaining lifetime was estimated by multiplying the quality-of-life weight of each terminal health state by its mean life expectancy.

Back to Top | Article Outline

Patient Population

Our target population included persons with neck pain seeking care from a physician, a physical therapist, or a chiropractor. The baseline analysis considered a cohort of 45-year-old individuals from the general population with neck pain of ≥2 weeks duration. We chose these characteristics to mirror the age and duration distribution reported in previous studies.1,5,23-26,26a

Back to Top | Article Outline

Model Data

We obtained evidence about event probabilities and transition rates between health states from studies accepted by the Neck Pain Task Force that approximated the characteristics of our target population (Table 1). Where such data were not available, we searched the literature for data based on samples that matched our cohort as closely as possible (Table 1). Transition rates from Troublesome Neck Pain and No Troublesome Neck Pain were estimated from a population-based cohort (Table 1).27

Table 1

Table 1

Back to Top | Article Outline
Background Risk of Adverse Events in the General Population.

The population-based incidence of hospitalization for acute strokes in persons aged 45 years was obtained from data gathered for a Neck Pain Task Force cohort study (Table 1).27a The 1-year incidence of stroke after a myocardial infarction was obtained from a meta-analysis.39 The remaining adverse events were estimated by pooling data identified by systematic searches of Medline. Intercept-only random-effects regression models were used to pool incidence rates and incidence proportions. We calculated the following: (1) rates of hospitalization for acute myocardial infarction,28–33 and for serious UGIB53,54; (2) 28-day and 1-year mortality after hospitalization for the following index events: myocardial infarction,30–32,34–38 heart failure,43,44 serious UGIB,55–61 and stroke45–49; and (3) an estimate of the proportion of strokes that result in major disability50–52; serious UGIBs that are treated with surgery59,62–64; and acute myocardial infarctions that result in heart failure.34,36,40–42

Back to Top | Article Outline
Treatment Risk.

Pooled estimates of stroke, myocardial infarction, and serious UGIB associated with NSAIDs were identified by a systematic search of meta-analyses of harmful events associated with NSAIDs in Medline (Table 1). Wherever possible, we selected pooled estimates that are based on clinical trials comparing classes of NSAIDs to placebo.9,19–21 Where such estimates were not found, we selected estimates that fulfilled these criteria as closely as possible.

The Neck Pain Task Force provided 2 estimates of the risk of stroke after chiropractic visits.11,54a Both studies reported a positive association between chiropractic visits and posterior circulation stroke in persons aged <45 years and a negative association in persons aged ≥45 years. However, the analysis on patients who visited physicians conducted in one of these studies (which found a positive association between physician visits and stroke) suggests that patients with neck pain and headaches consult a health care provider during the prodromal phase of a stroke.54a We applied a risk estimate for all types of strokes associated with chiropractic visits for persons aged <45 years estimated from the Neck Pain Task Force data (Table 1). This estimate was used because we considered it unlikely that the risk of stroke changes discretely at the age of 45. Given the positive association between chiropractic and physician visits with stroke reported by Cassidy et al54a (2008), we performed a sensitivity analysis in which we assumed no excess risk of stroke with manipulation.

We assumed that (1) “chiropractic visit” was a reasonable proxy for “neck manipulation” based on a study that reported 80.8% of patients receive manipulation4; (2) the risk of harmful events with manipulation and NSAIDs ended 4 and 8 weeks, respectively, after treatment10,11; (3) mortality after harmful events was equivalent in treatment users and nonusers; (4) patients could experience only 1 type of harmful event (cardiovascular, cerebrovascular, gastrointestinal) because the probability of experiencing more than 1 harmful event in a 4-week cycle is exceedingly rare; and (5) there were no risks of harms with exercise and mobilization (Table 2).

Table 2

Table 2

Back to Top | Article Outline
Treatment Effectiveness.

Effectiveness estimates were derived from data provided by a high-quality randomized trial that compared the care provided by general practitioners, physical therapists, and manual therapists.67–69 These interventions were used as proxies for NSAID, exercise, and manual treatments, respectively. We assumed that (1) sNSAIDs and Coxibs have an equivalent effectiveness70–72; (2) manipulation and mobilization have an equivalent effectiveness8a,24,73; and (3) that any differences in effectiveness ends 52 weeks from the start of treatment (Table 2).67,68,74,75

Back to Top | Article Outline
Mortality and Life Expectancy.

At each cycle, patients were subject to age-specific mortality due to all causes based on Canadian Life Tables.65 Average health state-specific life expectancies for Major Stroke Disability, Minor Stroke Disability, Post-Myocardial Infarction, and Heart Failure were obtained from the Framingham Heart Study cohort (Table 1).66 We assumed that life expectancies for persons in the Troublesome Neck Pain and No Troublesome Neck Pain health states were equivalent to age-matched persons in the general population (Table 2).65

Back to Top | Article Outline
Quality-of-Life Weights.

Quality-of-life weights for the health states were obtained from a sample of 220 neck pain patients presenting to clinics in Ontario and California (Table 1). Quality-of-life estimates were elicited with the standard gamble using health state descriptions as stimuli.13 The descriptions were constructed in a standardized manner using a recognized health classification framework.76,77 Variances of the mean quality-of-life estimates were estimated by using a variance estimator method.78

Back to Top | Article Outline


The internal consistency of the model was evaluated according to the method proposed by Philips et al.12

Back to Top | Article Outline
Expected Value Analyses.

We calculated expected value by multiplying the proportion of the cohort in each health state by the corresponding quality-of-life weight in each cycle and then summing products over all cycles, yielding an estimate of quality-adjusted survival.22

Two expected value analyses were performed: a baseline analysis and a case example. For the baseline analysis, we assumed that all treatments had equivalent effectiveness because the Neck Pain Task Force found no clinically important differences in the effectiveness of NSAID, exercise, and manual treatments.8a,24,71–73,79–82 Treatments were therefore compared by their risks of harmful outcomes only. The case example compared treatments by their risks of harmful events and their effectiveness, to explore the extent to which plausible estimates of effectiveness might change the baseline analysis results.67

We estimated the average gain in quality-adjusted life expectancy for exercise, mobilization, and manipulation compared with NSAIDs (the referent group) by subtracting the expected value of a treatment when only harms were considered from the expected value of the treatment when both benefits and harms were considered. Sensitivity analyses determined whether the rank-order of the best treatment (treatment with the highest expected value) changed when variable estimates were varied over plausible ranges.83 We defined plausible range as the lower and upper 95% confidence limits.

Back to Top | Article Outline


External Validation

At the end of the 1-year follow-up, 50.6% of the cohort was in the Troublesome Neck Pain state when treatments were assumed to have equivalent effectiveness and have no excess risk of harms. This result is consistent with Hill et al84 and the Neck Pain Task Force.84a

Back to Top | Article Outline

Expected Value Analyses

Baseline Analysis.

The model predicted a small average loss of life expectancy due to the risk of harms associated with NSAIDs and manipulation (Table 3). The difference between treatments assumed to have no harms (exercise, mobilization) versus manipulation, sNSAIDs, and Coxibs was 0.6, 2.2, and 4.5 days of life expectancy, respectively. Note that a loss of life expectancy does not mean, for example, that a 45-year-old person with a life expectancy of 35 years will live 4.5 fewer days;85 a loss of life expectancy means that the probability of a shorter life span has increased. The model also predicted a small average loss of quality-adjusted life expectancy. The difference between the highest-ranked (exercise, mobilization) and lowest-ranked (Coxibs) treatments was 3.4 quality-adjusted life days.

Table 3

Table 3

Back to Top | Article Outline

Case Example Based on Randomized Control Trial Effectiveness Data.

The average difference in quality-adjusted life expectancy between the highest-ranked (mobilization) and lowest-ranked (Coxibs) treatment was 7.3 quality-adjusted life days (Table 3). Note that predicted life expectancies for the case example are identical to those of the baseline analysis because neck pain treatment effectiveness influences quality-of-life (morbidity) and not length of life (mortality). The model predicted small gains in quality-adjusted life expectancy due to effectiveness. Average gains for manipulation, mobilization, and exercise were 4.2, 3.9, and 1.1 quality-adjusted life days, respectively, compared to NSAIDs (referent group).

Back to Top | Article Outline

Sensitivity Analyses

Baseline Analysis.

sNSAIDs was predicted to be the highest-ranked treatment when its risk of myocardial infarction was <0.87 (for life expectancy) and <0.86 (for quality-adjusted life expectancy). However, differences in expected values between sNSAIDs and exercise/mobilization were small. As expected, when neck manipulation was assumed to have no excess risk of stroke, its rank-order was equivalent to that of the best treatments.

Back to Top | Article Outline

Case Example Based on Randomized Control Trial Effectiveness Data.

Exercise was predicted to be the best treatment when all mean hazard rate ratios for mobilization were minimally varied; however, differences in expected values between these treatments remained small. When manipulation was assumed to have no excess risk of stroke, its rank-order was equivalent to that of the highest-ranked treatment.

Back to Top | Article Outline


We compared the impact of harms associated with sNSAIDS, Coxibs, and manipulation compared with mobilization and exercise on the health of a simulated cohort of neck pain patients. The collective impact of harms with these treatments on life expectancy and quality-adjusted life expectancy was small. Our model also predicted small differences in loss of life expectancy85,86 and quality-adjusted life expectancy across treatments. Evaluating both benefits and harms also yielded minor differences across treatments in quality-adjusted life expectancy.

Our study has strengths. To our knowledge, nonsurgical neck pain treatments have not been previously compared in a decision-analytic Markov model. We synthesized high-quality evidence mainly provided by the Neck Pain Task Force. We quantified the impact of the benefits and harms associated with neck pain treatments on patients’ health on a common yardstick. We performed extensive sensitivity analyses.

Our study has limitations. First, certain background estimates of harmful events in the general population were based on samples that did not match our cohort because estimates from samples comparable to our cohort were not available from the literature. For example, the estimate of mortality after a stroke was based on samples that were older than our cohort. However, this limitation is unlikely to have led to a differential bias across treatments because the same background estimates were applied to all treatments. Furthermore, sensitivity analyses showed that our results were robust to variations in these estimates. Second, patients were not modeled to cross-over treatments. Therefore, the impact of serially-combined treatments was not evaluated. Third, we did not include renal complications associated with NSAIDs.87 This likely resulted in a moderate bias that favored NSAID treatments. Then again, the estimates of harmful events associated with NSAIDs tended to be based on samples of patients with serious health conditions, taking higher doses than we expect for our patient cohort. This likely resulted in a moderate bias against NSAIDs. Finally, we assumed an excess risk of stroke with neck manipulation, yet recent evidence suggests no excess risk.54a

Our study did not find overwhelming evidence to suggest that there is 1 treatment among the compared treatments that is, on average, clearly better. Because our results suggest a toss-up, other elements, such as patients’ preferences for treatment88–91 and their attitudes toward risk, likely play an important role in deciding the best treatment. More studies that consider these decisional elements are needed.

Back to Top | Article Outline


From the standpoint of health policy, and when the objective is to maximize life expectancy and quality-adjusted life expectancy, there is no single treatment among NSAID, exercise, and manual therapies that is clearly better for neck pain. Therefore, a practitioner’s decision about the best treatment to provide to an individual patient should be based on the patient’s informed treatment preferences and attitudes toward risk.

Back to Top | Article Outline

Key Points

  • Although a harmful event can have serious consequences for an individual patient, on an average, the overall impact of harms associated with 5 common nonsurgical neck pain treatments on patients’ health seems to be small.
  • When the objective is to maximize (quality-adjusted) life expectancy and only risks of harmful events are considered, differences across 5 common nonsurgical neck pain treatments are too small to establish the best treatment.
  • When the objective is to maximize (quality-adjusted) life expectancy, and both treatment-related harms and benefits are considered, no treatment among the 5 common nonsurgical neck pain treatments is clearly superior.

Appendix available online through Article Plus.

Back to Top | Article Outline


The authors are grateful to George Tomlinson (University Health Network, Toronto) and members of the Measurement Research Unit (Dorcas Beaton, Kim Cullen, Anusha Raj, Peter Smith, and Dwayne van Eerd) (Institute for Work & Health, Toronto), and the Members of the Scientific Secretariat of the Bone and Joint Decade 2000-2010 Task Force on Neck Pain and Its Associated Disorders (Eugene Carragee, Linda Carroll, Jaime Guzman, Scott Haldeman, Lena Holm, Margareta Nordin, and Paul Peloso) for the expertise they provided. They are also grateful to Rachel Couban, Emma Irvin, Sigmund Kaw, and Quenby Mahood (Institute for Work & Health, Toronto) for their help with the study. They gratefully acknowledge the financial and material support provided by the Bone and Joint Decade 2000-2010 Task Force On Neck Pain and Its Associated Disorders, Institute for Work & Health, Ontario Ministry of Health and Long-term Care, Ontario Chiropractic Association, and Canadian Memorial Chiropractic College.

Back to Top | Article Outline


1. Côté P, Cassidy JD, Carroll L. The treatment of neck and low back pain: who seeks care? Who goes where? Med Care 2001;39:956–67.
2. Hurwitz EL, Morgenstern H. The effect of comorbidity on care seeking for back problems in the United States. Ann Epidemiol 1999;9:262–70.
3. Lim KL, Jacobs P, Klarenbach S. A population-based analysis of healthcare utilization of persons with back disorders: results from the Canadian Community Health Survey 2000–2001. Spine 2006;31:212–18.
4. Hurwitz EL, Coulter ID, Adams AH, et al. Use of chiropractic services from 1985 through 1991 in the United States and Canada. Am J Public Health 1998;88:771–6.
5. Vos C, Verhagen A, Passchier J, et al. Management of acute neck pain in general practice: a prospective study. Br J Gen Prac 2007;57:23–8.
6. Riddle DL, Schappert SM. Volume and characteristics of inpatient and ambulatory medical care for neck pain in the United States: data from three national surveys. Spine 2007;32:132–41.
7. Christensen MG, Kerkoff D, Kollasch MW. Job Analysis of Chiropractic, 2000: A Project Report, Survey Analysis, and Summary of Chiropractic Practice Within the United States. Greeley: National Board of Chiropractic Examiners; 2000.
8. Binder AI. Cervical spondylosis and neck pain. BMJ 2007;334:527–31.
8a. Hurwitz EL, Carragee EJ, van der Velde G, et al. Treatment of neck pain: noninvasive interventions: Results of the Bone and Joint Decade 2000–2010 Task Force on Neck Pain and Its Associated Disorders. Spine 2008;33(Suppl):S123–S152.
9. Kearney PM, Baigent C, Godwin J, et al. Do selective cyclo-oxygenase-2 inhibitors and traditional non-steroidal anti-inflammatory drugs increase the risk of atherothrombosis? Meta-analysis of randomized trials. BMJ 2006;332:1302–8.
10. Hernandez-Diaz S, Rodriguez LA. Association between nonsteroidal anti-inflammatory drugs and upper gastrointestinal tract bleeding/perforation: an overview of epidemiologic studies published in the 1990s. Arch Intern Med 2000;160:2093–9.
11. Rothwell DM, Bondy SJ, Williams JI. Chiropractic manipulation and stroke: a population-based case-control study. Stroke 2001;32:1054–60.
12. Philips Z, Bojke L, Sculpher M, et al. Good practice guidelines for decision-analytic modelling in health technology assessment: a review and consolidation of quality assessment. Pharmacoeconomics 2006;24:355–71.
13. Drummond MF, O’Brien B, Stoddart GL, et al. Methods for the Economic Evaluation of Health Care Programmes. 3rd ed. Oxford: Oxford University Press; 2005.
14. Von Korff M, Ormel J, Keefe FJ, et al. Grading the severity of chronic pain. Pain 1992;50:133–49.
15. Von Korff M, Dworkin SF, Le Resche L. Graded chronic pain status: an epidemiologic evaluation. Pain 1990;40:279–91.
16. Von Korff M, Dworkin SF, Le Resche L, et al. An epidemiologic comparison of pain complaints. Pain 1988;32:173–83.
17. Gold MR, Siegel JE, Russell LB, et al. Cost-effectiveness in Health and Medicine. New York: Oxford University Press; 1996.
18. McGettigan P, Herbert RD. Cardiovascular risk and inhibition of cyclooxygenase. JAMA 2006;296.
19. Moore RA, Derry S, Makinson GT, et al. Tolerability and adverse events in clinical trials of celecoxib in osteoarthritis and rheumatoid arthritis: systematic review and meta-analysis of information from company clinical trial reports. Arthritis Res Ther 2005;7:R644–R665.
20. Ofman JJ, MacLean CH, Straus WL, et al. A metaanalysis of severe upper gastrointestinal complications of nonsteroidal antiinflammatory drugs. J Rheumatol 2002;29:804–12.
21. Salpeter SR, Gregor P, Ormiston TM, et al. Meta-analysis: cardiovascular events associated with nonsteroidal anti-inflammatory drugs. Am J Med 2006;119:552–9.
22. Tosteson ANA. Preference-based health outcome measures in low back pain. Spine 2000;25:3161–6.
23. Coulter ID, Hurwitz EL, Adams AH, et al. Patients using chiropractors in North America: who are they, and why are they in chiropractic care? Spine 2002;27:291–6.
24. Hurwitz EL, Morgenstern H, Harber P, et al. A randomized trial of chiropractic manipulation and mobilization for patients with neck pain: clinical outcomes from the UCLA neck-pain study. Am J Pub Health 2002;92:1634–41.
25. Hurwitz EL, Chiang LM. A comparative analysis of chiropractic and general practitioner patients in North America: findings from the joint Canada/United States Survey of Health, 2002–03. BMC Health Serv Res 2006;6:49.
26. Korthalsde BI, Hoving JL, van Tulder MW, et al. Cost effectiveness of physiotherapy, manual therapy, and general practitioner care for neck pain: economic evaluation alongside a randomized controlled trial. BMJ 2003;326:911.
26a. Hogg-Johnson S, van der Velde G, Carroll LJ, et al. The burden and determinants of neck pain in the general population: Results of the Bone and Joint decade 2000–2010 Task Force on Neck Pain and Its Associated Disorders. Spine 2008;33(Suppl):S39–S51.
27. Côté P, Cassidy JD, Carroll LJ, et al. The annual incidence and course of neck pain in the general population: a population-based cohort study. Pain 2004;112:267–73.
27a. Boyle E, Côté P, Grier A, et al. Examining vertebrobasilar artery stroke in two canadian provinces. Spine 2008;33(Suppl):S170–S175.
28. Vrbova L, Crighton EJ, Mamdani M, et al. Temporal analysis of acute myocardial infarction in Ontario, Canada. Can J Cardiol 2005;21:841–5.
29. Rosamond WD, Folsom AR, Chambless LE, et al. Coronary heart disease trends in four United States communities. The Atherosclerosis Risk in Communities (ARIC) study 1987–1996. Int J Epidemiol 2001;30:S17-S22.
30. McGovern PG, Jacobs DR Jr, Shahar E, et al. Trends in acute coronary heart disease mortality, morbidity, and medical care from 1985 through 1997: the Minnesota heart survey. Circulation 2001;104:19–24.
31. Bata IR, Gregor RD, Eastwood BJ, et al. Trends in the incidence of acute myocardial infarction between 1984 and 1993—The Halifax County MONICA Project. Can J Cardiol 2000;16:589–95.
32. Pilote L, Lavoie F, Ho V, et al. Changes in the treatment and outcomes of acute myocardial infarction in Quebec, 1988–1995. CMAJ 2000;163:31–6.
33. Rosamond WD, Chambless LE, Folsom AR, et al. Trends in the incidence of myocardial infarction and in mortality due to coronary heart disease, 1987 to 1994. N Engl J Med 1998;339:861–7.
34. Goldberg RJ, Cui J, Olendzki B, et al. Excess body weight, clinical profile, management practices, and hospital prognosis in men and women after acute myocardial infarction. Am Heart J 2006;151:1297–304.
35. Cox JL, Bata IR, Gregor RD, et al. Trends in event rate and case fatality of patients hospitalized with myocardial infarction between 1984 and 2001. Can J Physiol Pharmacol 2006;84:121–7.
36. Spencer FA, Meyer TE, Gore JM, et al. Heterogeneity in the management and outcomes of patients with acute myocardial infarction complicated by heart failure: the National Registry of Myocardial Infarction. Circulation 2002;105:2605–10.
37. Heidenreich PA, McClellan M. Trends in treatment and outcomes for acute myocardial infarction: 1975–1995. Am J Med 2001;110:165–74.
38. Goldberg RJ, Yarzebski J, Lessard D, et al. A two-decades (1975 to 1995) long experience in the incidence, in-hospital and long-term case-fatality rates of acute myocardial infarction: a community-wide perspective. J Am Coll Cardiol 1999;33:1533–9.
39. Witt BJ, Ballman KV, Brown RD Jr, et al. The incidence of stroke after myocardial infarction: a meta-analysis. Am J Med 2006;119:354e1–e9.
40. Hellermann JP, Goraya TY, Jacobsen SJ, et al. Incidence of heart failure after myocardial infarction: is it changing over time? Am J Epidemiol 2003;157:1101–7.
41. Spencer FA, Meyer TE, Goldberg RJ, et al. Twenty year trends (1975–1995) in the incidence, in-hospital and long-term death rates associated with heart failure complicating acute myocardial infarction: a community-wide perspective. J Am Coll Cardiol 1999;34:1378–87.
42. Ali AS, Rybicki BA, Alam M, et al. Clinical predictors of heart failure in patients with first acute myocardial infarction. Am Heart J 1999;138:1133–9.
43. Hellermann JP, Jacobsen SJ, Gersh BJ, et al. Heart failure after myocardial infarction: a review. Am J Med 2002;113:324–30.
44. Jong P, Vowinckel E, Liu PP, et al. Prognosis and determinants of survival in patients newly hospitalized for heart failure: a population-based study. Arch Intern Med 2002;162:1689–94.
45. Johansen HL, Wielgosz AT, Nguyen K, et al. Incidence, comorbidity, case fatality and readmission of hospitalized stroke patients in Canada. Can J Cardiol 2006;22:65–71.
46. Kleindorfer D, Broderick J, Khoury J, et al. The unchanging incidence and case-fatality of stroke in the 1990s: a population-based study. Stroke 2006;37:2473–8.
47. Field TS, Green TL, Roy K, et al. Trends in hospital admission for stroke in Calgary. Can J Neurol Sci 2004;31:387–93.
48. Tu JV, Gong Y. Trends in treatment and outcomes for acute stroke patients in Ontario, 1992–1998. Arch Intern Med 2003;163:293-7.
49. Kapral MK, Wang H, Mamdani M, et al. Effect of socioeconomic status on treatment and mortality after stroke. Stroke 2002;33:268–73.
50. Carandang R, Seshadri S, Beiser A, et al. Trends in incidence, lifetime risk, severity, and 30-day mortality of stroke over the past 50 years. JAMA 2006;296:2939–46.
51. Mayo NE, Hendlisz J, Goldberg MS, et al. Destinations of stroke patients discharged from the Montreal area acute-care hospitals. Stroke 1989;20:351–6.
52. Bonita R, Beaglehole R. Recovery of motor function after stroke. Stroke 1988;1497–500.
53. Mellemkjaer L, Blot WJ, Sorensen HT, et al. Upper gastrointestinal bleeding among users of NSAIDs: a population-based cohort study in Denmark. Br J Clin Pharmacol 2002;53:173–81.
54. Garcia Rodriguez LA, Walker AM, Perez GS. Nonsteroidal antiinflammatory drugs and gastrointestinal hospitalizations in Saskatchewan: a cohort study. Epidemiology 1992;3:337–42.
54a. Cassidy JD, Boyle E, Côté P, et al. Risk of vertebrobasilar stroke and chiropractic care: results of a population-based case control and case-crossover study. Spine 2008;33(Suppl):S176–S183.
55. Lassen A, Hallas J, Schaffalitzky de Muckadell OB. Complicated and uncomplicated peptic ulcers in a Danish county 1993–2002: a population-based cohort study. Am J Gastroenterol 2006;101:945–53.
56. Lanas A, Perez-Aisa MA, Feu F, et al. A nationwide study of mortality associated with hospital admission due to severe gastrointestinal events and those associated with nonsteroidal antiinflammatory drug use. Am J Gastroenterol 2005;100:1685–93.
57. Di Fiore F, Lecleire S, Merle V, et al. Changes in characteristics and outcome of acute upper gastrointestinal haemorrhage: a comparison of epidemiology and practices between 1996 and 2000 in a multicentre French study. Eur J Gastroenterol Hepatol 2005;17:641–7.
58. Garcia-Rodriguez LA, Ruigomez A, Hasselgren G, et al. Comparison of mortality from peptic ulcer bleed between patients with or without peptic ulcer antecedents. Epidemiology 1998;9:452–6.
59. Rockall TA. Management and outcome of patients undergoing surgery after acute upper gastrointestinal haemorrhage. Steering Group for the National Audit of Acute Upper Gastrointestinal Haemorrhage. J R Soc Med 1998;91:518–23.
60. Vreeburg EM, Snel P, de Bruijne JW, et al. Acute upper gastrointestinal bleeding in the Amsterdam area: incidence, diagnosis, and clinical outcome. Am J Gastroenterol 1997;92:236–43.
61. Rockall TA, Logan RF, Devlin HB, et al. Incidence of and mortality from acute upper gastrointestinal haemorrhage in the United Kingdom. Steering Committee and members of the National Audit of Acute Upper Gastrointestinal Haemorrhage. BMJ 1995;311:222–6.
62. Ramsoekh D, van Leerdam ME, Rauws EA, et al. Outcome of peptic ulcer bleeding, nonsteroidal anti-inflammatory drug use, and Helicobacter pylori infection. Clin Gastroenterol Hepatol 2005;3:859–64.
63. Rockall TA, Logan RF, Devlin HB, et al. Influencing the practice and outcome in acute upper gastrointestinal haemorrhage. Steering Committee of the National Audit of Acute Upper Gastrointestinal Haemorrhage. Gut 1997;41:606–11.
64. Sandel MH, Kolkman JJ, Kuipers EJ, et al. Nonvariceal upper gastrointestinal bleeding: differences in outcome for patients admitted to internal medicine and gastroenterological services. Am J Gastroenterol 2000;95:2357–62.
65. Statistics Canada. 2000-2002 Life Tables, Canada, Provinces and Territories, Ottawa, 2006. Statistics Canada Catalogue No. 84-537-XIE.
66. Peeters A, Mamun AA, Willekens F, et al. A cardiovascular life history. A life course analysis of the original Framingham Heart Study cohort. Eur Heart J 2002;23:458–66.
67. Hoving JLP, de Vet HCWP, Koes BWP, et al. Manual therapy, physical therapy, or continued care by the general practitioner for patients with neck pain: long-term results from a pragmatic randomized clinical trial. Clin J Pain 2006;22:370–7.
68. Hoving JL. Manual therapy, physical therapy, or continued care by a general practitioner for patients with neck pain. A randomized, controlled trial. Ann Intern Med 2002;136:713–22.
69. Allison PD. Survival Analysis Using the SAS System: A Practical Guide. Cary, NC: SAS Institute Inc.; 1995.
70. Bombardier C, Laine L, Reicin A, et al. Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis. VIGOR Study Group. N Engl J Med 2000;343:1520–8.
71. Nadarajah A, Abrahan L, Lau FL, et al. Efficacy and tolerability of celecoxib compared with diclofenac slow release in the treatment of acute ankle sprain in an Asian population. Singapore Med J 2006;47:534–42.
72. Singh G, Fort JG, Goldstein JL, et al. Celecoxib versus naproxen and diclofenac in osteoarthritis patients: SUCCESS-I Study. Am J Med 2006;119:255–66.
73. Jordan A, Bendix T, Nielsen H, et al. Intensive training, physiotherapy, or manipulation for patients with chronic neck pain. A prospective, single-blinded, randomized clinical trial. Spine 1998;23:311–18.
74. Koes BW, Bouter LM, van MH, et al. The effectiveness of manual therapy, physiotherapy, and treatment by the general practitioner for nonspecific back and neck complaints. A randomized clinical trial. Spine 1992;17:28–35.
75. Koes BW, Bouter LM, van MH, et al. Randomised clinical trial of manipulative therapy and physiotherapy for persistent back and neck complaints: results of one year follow up. BMJ 1992;304:601–5.
76. Froberg DG, Kane RL. Methodology for measuring health-state preferences. I. Measurement strategies. J Clin Epidemiol 1989;42:345–54.
77. World Health Organization. International Classification of Functioning, Disability and Health. World Health Organization; 2001.
78. Mach L, Dumais J, Robinson A. A Study of the Properties of a Bootstrap Variance Estimator Under Sampling Without Replacement. Arlington, VA: Federal Committee on Statistical Methodology (FCSM) Research Conference; 2005.
79. Kongsted A, Qerama E, Kasch H, et al. Neck collar, “act-as-usual” or active mobilization for whiplash injury? A randomized parallel-group trial. Spine 2007;32:618–26.
80. Horneij E, Hemborg B, Jensen I, et al. No significant differences between intervention programmes on neck, shoulder and low back pain: a prospective randomized study among home-care personnel. J Rehabil Med 2001;33:170–6.
81. Yamamoto M, Sugano T, Kashiwazaki S, et al. Double-blind comparison of piroxicam and indomethacin in the treatment of cervicobrachial syndrome and periarthritis scapulohumeralis (stiff shoulder). Eur J Rheumatol Inflamm 1983;6:266–73.
82. Zylbergold RS, Piper MC. Cervical spine disorders. A comparison of three types of traction. Spine 1985;10:867–71.
83. Felli JC, Hazen GB. Sensitivity analysis and the expected value of perfect information. Med Decis Making 1998;18:95–109.
84. Hill J, Lewis M, Papageorgiou AC, et al. Predicting persistent neck pain: a 1-year follow-up of a population cohort. Spine 2004;29:1648–54.
84a. Carroll LJ, Hogg-Johnson S, van der Velde G, et al. Course and prognostic factors for neck pain in the general population: Results of the Bone and Joint decade 2000–2010 Task Force on Neck Pain and Its Associated Disorders. Spine 2008;33(Suppl):S75–S82.
85. Naimark D, Naglie G, Detsky AS. The meaning of life expectancy: what is a clinically significant gain? J Gen Intern Med 1994;9:702–7.
86. Wright JC, Weinstein MC. Gains in life expectancy from medical interventions—standardizing data on outcomes. N Engl J Med 1998;339:380–6.
87. Zhang J, Ding EL, Song Y. Adverse effects of cyclooxygenase 2 inhibitors on renal and arrhythmia events: meta-analysis of randomized trials. JAMA 2006;296:1619–32.
88. Swan JS, Lawrence WF, Roy J. Process utility in breast biopsy. Med Decis Making 2006;26:347–59.
89. Swan JS, Sainfort F, Lawrence WF, et al. Process utility for imaging in cerebrovascular disease. Acad Radiol 2003;10:266–74.
90. Brouwer WB, van Exel NJ, van den BB, et al. Process utility from providing informal care: the benefit of caring. Health Policy 2005;74:85–99.
91. Donaldson C, Shackley P. Does “process utility” exist? A case study of willingness to pay for laparoscopic cholecystectomy. Soc Sci Med 1997;44:699–707.

decision analysis; exercise; quality-of-life; neck pain; nonsteroidal anti-inflammatory drug; manual therapy; effectiveness; treatment risk

© 2008 Lippincott Williams & Wilkins, Inc.