Secondary Logo

Journal Logo

Original Articles

The Relation Between Patients’ Beliefs About Pain Medication, Medication Adherence, and Treatment Outcome in Chronic Pain Patients

A Prospective Study

Timmerman, Leon PhD*; Stronks, Dirk L. PhD; Huygen, Frank J. PhD

Author Information
doi: 10.1097/AJP.0000000000000760
  • Free


Chronic pain has been reported by about 1 in 5 adults throughout the world.1 Pharmacotherapy is an important component of pain therapy. It is well recognized that pain perception and pain relief after analgesic therapy display large interindividual variability.2 Age, sex, ethnicity, and actual level of stress, mood, or diseases may modify individual pain perception and responses to drug treatment.2 Furthermore, pharmacogenetic differences may lead to pharmacokinetic and pharmacodynamic differences.3 Any of these factors may play a role in the (suboptimal) effect of analgesic treatment.

Another important reason for suboptimal treatment response might be that up to 40% of chronic pain patients do not use their medication as prescribed.4 In other chronic conditions, medication nonadherence has been described to lead to reduced clinical benefit, increased burden of side effects, medication wastage, and increased health care costs.5 Whether the same holds true for chronic pain is not clearly established. Although an association between medication adherence and treatment outcome has been shown, a causal relationship has not been demonstrated.6 However, chronic pain medications, for example, antidepressants, anticonvulsants, or long-acting opioids, are often prescribed to be used on a regular basis, and it seems plausible that this medication will work better if it is used accordingly. Furthermore, nonadherence to pain medication may result in unnecessary treatment changes and medication wastage as well, and overuse of analgesics may result in health care risks. Adherence to a medial regime is defined as the extent to which a person’s behavior corresponds with agreed recommendations from a health care provider.7 Nonadherent behavior may be intentional, unintentional, or both. Unintentional nonadherence occurs when a patient wants to adhere but is unable to because of lack of capacity or resources, such as low health literacy or financial constraints. Intentional nonadherence involves a decisional process not to follow the recommendations. Individual perceptions about disease severity or prescribed therapy can influence motivation to start or continue medication.8,9 Beliefs about medication are a well-investigated determinant of medication adherence in many chronic conditions.8,9 The individual balance between perceived necessity and concerns about the medication (the “Necessity-Concerns Framework”) may explain intentional nonadherence and provides a target for adherence-improving interventions.9

Perceived necessity and concerns toward pain medication have also been described to be associated with adherence in chronic noncancer pain patients.10,11 Using the 47-item “Pain Medication Attitudes Questionnaire (PMAQ),” perceived need was associated with analgesic overuse among chronic pain patients, and concerns over addiction and side effects were associated with underuse of pain medication in chronic pain patients.10 Recently, a 14-item version of the PMAQ showed similar results.12

Beliefs about medications for chronic disease have been described to be related to therapy outcome, for example, in diabetes, possibly by their effect on medication adherence.13 It is unknown whether this applies to chronic pain treatment outcomes as well.

Earlier cross-sectional studies reported a relationship between patients’ beliefs about pain medication and medication adherence.10,11,14 The aim of this study was to prospectively investigate the relationship between patients’ initial beliefs about pain medication and medication adherence patterns and treatment outcomes during follow-up.


This study was part of a randomized clinical trial, in which the effect of 3 different follow-up strategies on medication adherence and treatment outcome were compared. This single-center randomized controlled trial was performed in a pain treatment center of a large general hospital in The Netherlands after approval of the medical ethics committee. Newly referred patients with chronic noncancer pain >3 months with the following indications for structural pain medication were eligible for participation: central neuropathic pain, peripheral neuropathic pain, complex regional pain syndrome, and chronic low back pain. Inclusion was not restricted to the type of medication prescribed, as long as it was prescribed for structural use. Patients receiving medication on an “as needed” basis and patients below 18 years were excluded.

All consecutive patients who received a new analgesic prescription, or who received additional pain medication to an existing treatment regimen, were invited to participate. Participants had to be able to complete electronic questionnaires in the Dutch language. Participants were identified and informed by 1 of 3 doctors’ assistants according to the above-mentioned criteria, directly after the specialists’ consultation, and provided with written information. All participants provided written informed consent.


After receipt of informed consent, an e-mail was sent containing a link to the web-based baseline questionnaire. After 11 weeks, a follow-up questionnaire was sent.

Patients were randomized to 1 of the 3 study groups: (1) standard follow-up, which consisted of follow-up visits after 6 and 12 weeks; (2) intensified hospital-initiated follow-up, which consisted of standard follow-up and additional scheduled counselling by a specialized nurse after 3, 5, 8, and 10 weeks; or (3) patient-initiated follow-up, which consisted of standard follow-up and additional counselling at patients’ request.

Outcome Measures

Baseline characteristics collected were age, sex, level of education, mean and maximum pain intensity in the previous week (11-point Numeric Rating Scale), number of different medications, and duration of pain symptoms. Patient attitudes and concerns about pain medication were measured with a Dutch translation of the “PMAQ (permission for reuse by Elsevier),” a 47-item validated survey measuring attitudes and concerns toward pain medication with regard to 7 subscales (perceived need, mistrust in the prescribing doctor, and concerns over side effects, adverse scrutiny, withdrawal symptoms, addiction, and tolerance).10 The English questionnaire was translated forward by 2 persons, one of whom was not related to the study. Two other persons performed backward translation, one of whom was not related to the study and had English as his native language. Although a validated Dutch translation of the “Beliefs about Medication Questionnaire” is available, we decided to use the PMAQ because its questions and subscales were specifically designed for use in the chronic pain population.10,15

Medication adherence was measured 11 weeks after initiation of the new prescription. Adherence was measured by asking 2 questions. One question measured underuse of pain medication: “You received a prescription for pain medication from your doctor. How often do you, intentionally or unintentionally, miss or skip a dose?” The second question measured the overuse of pain medication: ‘How often do you take more medication than prescribed? Answers were given on a 6-point scale (0=never, 1=seldom, 2=once a month, 3=once a week, 4 more than once a week, not every day, 5=every day).

Moreover, after 11 weeks, patient satisfaction was measured using the module “Current pain medication” and “Satisfaction with current pain medication and care” of the Dutch translation of the Pain Treatment Satisfaction Scale (PTSS; permission for reuse by MAPI Institute, Lyon, France).16 The first module consists of 8 questions about the effect of pain medication on (1) physical health, (2) outlook on life, (3) daily activities, (4) leisure activities, (5) independency, (6) relationships, (7) mood, and (8) concentration. The latter module consists of 8 questions as regards satisfaction about (1) information received, (2) time devoted to patient, (3) care provided by nurses, (4) medication dosing frequency, (5) the amount of pain medication, (6) the time to the onset of medication effects, (7) the amount of pain relief, and (8) duration of pain relief provided by pain medication. The PTSS items were scored on a 5-point Likert scale: 1=very satisfied, 2=satisfied, 3=neither satisfied nor dissatisfied, 4=dissatisfied, and 5=very dissatisfied). Furthermore, the burden of side effects was registered (1=not bothered, 2=a little bothered, 3=moderately bothered, 4=quite bothered, and 5=extremely bothered).

Data Analysis

Underuse was defined as missing a dose every week up to every day. Overuse was defined as taking additional medication every week up to every day.

PMAQ items were rated on a 6-point numerical scale (0=never true to 5=always true). One missing value per subscale was accepted. Subscale scores were the mean score of valid item scores. Patient satisfaction subscale scores were the mean scores of PTSS item scores (both modules scored separately). For the regression analysis, side effects were dichotomized, considering scores of 3 (moderately bothered) and higher as “bothered by side effects.”

Statistical Analysis

Descriptive statistics were used to determine the frequencies of the demographic variables and PMAQ scores and to describe measures of central tendency and dispersion dependent on the shape of their distribution. The Shapiro-Wilk test was used to analyze whether or not parameters were normally distributed. Normally distributed data are presented as mean (SD), not normally distributed data as median (interquartile range). Differences in proportions between the experimental groups at baseline were tested using the Pearson χ2 test. Differences in continuous variables were evaluated using the independent-samples Kruskal-Wallis test if the parameter was not normally distributed and the 1-way analysis of variance if the parameter was normally distributed.

Binary logistic regression analysis was used to evaluate the contribution of study allocation, baseline characteristics and PMAQ subscores to the prediction of adherence and side effects 11 weeks after initiation of the newly prescribed medication. To prevent overfitting of the models, we performed univariate binary logistic regression analyses of PMAQ subscales. Only those parameters with a significance level of P-value ≤0.2 were entered into the final multivariate stepwise binary logistic regression analysis (method Backward Wald) with a probability out of P-value=0.1. To prevent multicollinearity, pairwise correlations between the parameters to be entered into the final model were calculated. Of those with a bivariate correlation of ≥0.7, only the parameter with the highest univariate significance level was entered into the final model. Linear regression was used to evaluate the contribution of study allocation and PMAQ subscores to the prediction of PTSS scores. The same method was used to prevent overfitting of the model. A stepwise (backward method) was used in the final model.

For all statistics, α was set at the traditional 0.05 level. All analyses were performed using IBM SPSS Statistics, version 24 (SPSS Inc., Chicago, IL).


From November 2014 up to November 2016, 133 of 139 eligible patients provided written informed consent and responded to the baseline questionnaire. Patient characteristics of nonconsenting patients and reasons for nonconsent were not documented. After 11 weeks, 99 patients (75%) completed the study. Baseline characteristics of the patients are presented in Table 1. Patients who did not complete the study were significantly younger than patients who did complete the study. All other baseline characteristics, including PMAQ scores, did not differ between the group that did and did not complete the study (Table 1). Prescribed medications are presented in Table 2. PMAQ scores are presented in Figure 1 and Table 3.

Number and Percentage of Prescribed Medications of Patients at Baseline and of Responding Patients After 11 Weeks
Boxplot of the subscales of the Pain Medication Attitudes Questionnaire scores (0=never true to 5=always true).
Median Scores of the Pain Medication Attitudes Questionnaire

Baseline characteristics and PMAQ scores did not differ between the randomization groups (data not shown). The study allocation (standard follow-up, intensive hospital-initiated follow-up, and patient-initiated follow-up) was not related to underuse of medication (P=0.74), overuse of medication (P=0.25), “satisfaction with the effect of pain medication” (P=0.92), “satisfaction with current pain medication and care” (P=0.27), and presence of side effects (P=0.31) in the regression analyses.

Attitudes and Concerns Towards Medication and Underuse

Underuse was reported by 13 (13.1%) patients. The univariate binary logistic regression analyses revealed that within the above-mentioned criteria for inclusion as covariates into the final multivariate analysis, maximum pain intensity (P=0.0143), concerns over addiction (P=0.033), concerns over side effects (P=0.004), and concerns over tolerance (P=0.135) contributed to the prediction of medication underuse. Entering these into the final multivariate analysis resulted in a significant positive contribution of only concerns over side effects to the prediction of medication underuse (odds ratio [OR]=3.12; 95% confidence interval [CI]: 1.45-6.60; P=0.003) (Table 4).

Results of the Multivariate Regression Analyses of Pain Medication Beliefs in Relation to Medication Adherence, Treatment Satisfaction and Side Effects

Attitudes and Concerns Towards Medication and Overuse

Overuse was reported by 8 (8.1%) patients. Mean pain intensity (P=0.024), maximum pain intensity (P=0.024), perceived need (P=0.006), concerns over withdrawal (P=0.171), and concerns over addiction (P=0.192) significantly contributed to the prediction of overuse in the univariate logistic regression. Mean pain intensity was not entered in the final analysis due to strong correlation with maximum pain intensity (r=0.71). In the final multivariate analysis, perceived need contributed positively (OR=15.50; 95% CI: 2.49-96.74; P=0.003), and concerns over addiction contributed negatively (OR=0.24; 95% CI: 0.06-0.98; P=0.047) to the prediction of medication overuse (Table 4).

Attitudes Towards Pain Medication and Satisfaction With Medication

The PTSS scores are presented in Table 5. Median “satisfaction with the effect of current medication” subscale score was 2.63 (1.5). Concerns over side effects (P=0.002) and tolerance (P<0.001) were significantly associated with “satisfaction with the effect of current medication” in the univariate analysis. Entering these items into the multivariate analysis resulted in a significant contribution of concerns over tolerance to the prediction of “satisfaction with the effect of current medication” (β=0.534; 95% CI: 0.284-0.784; P<0.001) (higher scores indicate lower satisfaction) (Table 4).

Pain Treatment Satisfaction Scores of the Individual Patients

Median “satisfaction with current pain medication and care” score was 2.25 (0.75). Concerns over tolerance (P=0.009), adverse scrutiny (P=0.108), side effects (P=0.017), mistrust (P=0.001), and withdrawal (P=0.179) were associated with “satisfaction with current pain medication and care” in the univariate analysis. Entering these items in the multivariate analysis resulted in a significant contribution of concerns over tolerance (β=0.211; 95% CI: 0.042-0.380; P=0.015) and mistrust (β=0.214; 95% CI: 0.036-0.392; P=0.019) to the prediction of “satisfaction with current medication and care” (higher scores indicate lower satisfaction) (Table 4).

Attitudes Towards Pain Medication and Burden of Side Effects

Presence of side effects were reported as follows: 28 (28.3%) patients were not bothered by side effects, 14 (14.1%) were a little bothered, 30 (30.3%) were moderately bothered, 21 (21.2%) were quite bothered, and 6 (6.1%) patients were extremely bothered by side effects. Maximum pain intensity (P=0.008), mean pain intensity (P=0.023), concerns over addiction (P=0.079), concerns over side effects (P=0.002), concerns over withdrawal (P=0.195), mistrust in the doctor (P=0.048), and concerns over tolerance (P=0.022) were associated with the burden of side effects after 11 weeks. Mean pain intensity was not entered in the final analysis due to strong correlation with maximum pain intensity (r=0.71). In the final multivariate analysis concerns over side effects contributed positively to the prediction of side effects after 11 weeks (OR=2.41; 95% CI: 1.36-4.29; P=0.003) (Table 4).


This prospective study confirms the results of earlier cross-sectional studies, in which associations were found between patient beliefs about prescribed medication and medication adherence.10 As previously discussed, underuse and overuse are 2 different entities that should be considered separately.11,17 This is confirmed by the fact that different beliefs about medication contribute to these separate behaviors. Concerns over side effects of pain medication were associated with underuse nonadherence. Overuse nonadherence was positively associated with the perceived need for pain medication and negatively associated with concerns over addiction. In the earlier study in which the PMAQ was used in secondary care, other PMAQ subscales were demonstrated to be associated with underuse (concerns over withdrawal, perceived need) and overuse (concerns over scrutiny) as well.10 Although the directions of the associations were similar, these factors did not reach significance in our study. A possible explanation might be the different operationalization of underuse and overuse. Instead of considering any deviation from the prescription as nonadherence, in our study patients were allowed to deviate from the prescription up to once a week to be considered adherent.

Concerns over pain medication tolerance and mistrust in the doctors’ decisions before initiation of new pain medication was associated with lower patient satisfaction scores. Many chronic pain patients may consider medication as a temporary symptomatic solution and may fear that, whether the therapy is helpful or not, a relapse might occur. Whereas positive expectations may result in an analgesic placebo response, negative expectations may result in analgesic nocebo response lowering the specific positive effects of analgesic therapy.18,19 Furthermore, divergent expectancies, leading to mistrust or doubts with regards to medical decisions, have been described earlier to suppress the effectiveness of pain therapy.18,19

The finding that patients’ concerns over side effects were associated with the actual experience of side effects after 11 weeks might be explained by earlier negative experiences with pain medication. Negative expectations due to prior complications or negative information have been described to result in actual adverse events by nocebo mechanisms as well.19 Furthermore, patients with earlier negative effects might be more susceptible to repeated side effects. Side effects may not only occur due to intolerance to a specific drug. Impaired drug metabolism or interactions with other medications are factors that might cause general susceptibility for medication side effects.

The nonadherence levels found in this study were lower than reported in earlier studies. First, this might be due to the measures of adherence we selected consisting of 1 question with regards to underuse and 1 with regard to overuse of medication. Although it was used anonymously, and patients did not have a reason to report different than their actual medication use, this self-report measure is susceptible to overestimation of adherence, because of social desirability and memory biases.20 Second, patients were allowed to deviate from the prescription up to once a week to be considered adherent. Other studies use a more strict definition of nonadherence as “any deviation of the prescribed therapy.” Up to now, there is no generally accepted definition of adherence and no general accepted subjective or objective operationalization of this concept. We chose this self-report measure because it is clinically applicable, and this definition because it seems unrealistic to expect patients to never, intentionally or unintentionally, omit or add a dose.

Findings of associations between patient beliefs about medication and medication adherence are clinically relevant and should support the design of future adherence-improving interventions. Taking account of patients’ necessity beliefs and concerns about pain medication when prescribing new analgesics could support adherence to these prescriptions. The first step in facilitating adherence is to take a “no-blame approach” and encourage an honest and open discussion to identify barriers to adherence.21 Individual beliefs and concerns should be addressed, leading to a shared decision as regards pharmacological therapy. This approach, which has had encouraging results, might lead to alternate (or no) treatment when specific barriers are too strong to overcome.22 The most challenging reality is that thorough assessment and discussion of patients’ views about medication costs significant time and is difficult to achieve during a busy schedule at a pain clinic. Given the finding that physicians generally take less than a minute to prescribe a new medication, there is a need for change.23 Questionnaires as the PMAQ, of which a shorter 14-item form has been introduced recently, might serve as a starting point for discussions about pain medication.12 Although standardized education might increase patient knowledge, tailored counseling is necessary to address specific concerns that are highly individual.24

The mean satisfaction scores presented to support the need for a change in the prescribing practice. Although patients are generally positive about the care provided by doctors and nurses, satisfaction scores about any aspect of prescribed pain medication are not so positive. Unrealistic expectations on the one hand and a common inability of medication to control chronic pain on the other may explain this unfavorable outcome.

Although the associations found are significant, the models account for only a small part of the variation found in the outcome parameters. Nevertheless, awareness of a relationship between patient views about medication on the one hand, and medication use and treatment results on the other, is clinically relevant. Chronic pain is a complex condition, and treatment results are often unsatisfactory. As a prescription of pain medication is very common, any small step in improving the use and the effects of this treatment should be considered.

A limitation of this study might have been the fact that it is a substudy within a randomized trial that investigated the effect of different follow-up strategies on medication adherence and pain treatment satisfaction. However, the study allocation did not contribute to any of the outcome parameters used in this study. Second, the follow-up period after the initiation of chronic pain therapy in this study was only 11 weeks, which may account for the relatively low nonadherence rates. It is likely that nonadherence rates would have been higher after, for example, 6 months. Third, we did not perform post hoc corrections for multiple testing. However, as the number of covariates in the analyses remained limited and most associations were highly significant, they were not likely to be based on coincidence. A final limitation is the dropout rate of 25%, for which the reasons were not recorded. The nonresponders were younger, which is an independent risk factor for nonadherence.4 Patients who do not adhere to a study protocol, which consists of the completion of 2 questionnaires, might have different medication adherence patterns as well.

In conclusion, this study prospectively confirms earlier cross-sectional reports about the association between attitudes towards prescribed pain medication and nonadherence patterns. Furthermore, attitudes and concerns towards pain medication were related to outcome parameters. To improve medication adherence and therapy outcome, patient beliefs about pain medication should be addressed.


1. Breivik H, Collett B, Ventafridda V, et al. Survey of chronic pain in Europe: prevalence, impact on daily life, and treatment. Eur J Pain. 2006;10:287–333.
2. Fillingim RB. Individual differences in pain responses. Curr Rheumatol Rep. 2005;7:342–347.
3. Svetlik S, Hronova K, Bakhouche H, et al. Pharmacogenetics of chronic pain and its treatment. Mediators Inflamm. 2013;2013:864319.
4. Timmerman L, Stronks DL, Groeneweg JG, et al. Prevalence and determinants of medication non-adherence in chronic pain patients: a systematic review. Acta Anaesthesiol Scand. 2016;60:416–431.
5. DiMatteo MR, Giordane MA, Lepper HS, et al. Patient adherence and medical treatment outcomes. Med Care. 2002;40:794–811.
6. Stern A, Sanchez-Magro I, Rull M. Chronic noncancer pain intensity is inversely related to analgesic adherence in pain clinics. J Med Econ. 2011;14:568–575.
7. Noncommunicable Diseases and Mental Health Cluster. Adherence to Long Term Therapies: Evidence for Action. Geneva, Switzerland: World Health Organization; 2003.
8. Morgan C, McBeth J, Cordingley L, et al. The influence of behavioural and psychological factors on medication adherence over time in rheumatoid arthritis patients: a study in the biologics era. Rheumatology (Oxford). 2015;54:1780–1791.
9. Horne R, Chapman SC, Parham R, et al. Understanding patients’ adherence-related beliefs about medicines prescribed for long-term conditions: a meta-analytic review of the Necessity-Concerns Framework. PLoS ONE. 2013;8:e80633.
10. McCracken LM, Hoskins J, Eccleston C. Concerns about medication and medication use in chronic pain. J Pain. 2006;7:726–734.
11. Rosser BA, McCracken LM, Velleman SC, et al. Concerns about medication and medication adherence in patients with chronic pain recruited from general practice. Pain. 2011;152:1201–1205.
12. Elander J, Said O, Maratos FA, et al. Development and validation of a short-form Pain Medication Attitudes Questionnaire (PMAQ-14). Pain. 2017;158:400–407.
13. Aikens JE, Piette JD. Diabetic patients’ medication underuse, illness outcomes, and beliefs about antihyperglycemic and antihypertensive treatments. Diabetes Care. 2009;32:19–24.
14. Lai YH, Keefe FJ, Sun WZ, et al. Relation between pain-specific beliefs and adherence to analgesic regimens in Taiwanese cancer patients: a preliminary study. J Pain Symptom Manage. 2002;24:415–423.
15. De Ridder D, Theunissen N. De rol van ziektepercepties in therapietrouw bij hypertensie. Gedrag en gezondheid [The role of illness perceptions and adherence in hypertension]. Behavior and Health (Dutch). 2003;31:237–245.
16. Evans CJ, Trudeau E, Mertzanis P, et al. Development and validation of the Pain Treatment Satisfaction Scale (PTSS): a patient satisfaction questionnaire for use in patients with chronic or acute pain. Pain. 2004;112:254–266.
17. Broekmans S, Dobbels F, Milisen K, et al. Medication adherence in patients with chronic non-malignant pain: is there a problem? Eur J Pain. 2009;13:115–123.
18. Klinger R, Blasini M, Schmitz J, et al. Nocebo effects in clinical studies: hints for pain therapy. Pain Rep. 2017;2:e586.
19. Colloca L, Finniss D. Nocebo effects, patient-clinician communication, and therapeutic outcomes. JAMA. 2012;307:567–568.
20. Jerant A, Dimatteo R, Arnsten J, et al. Self-report adherence measures in chronic illness. Retest reliability and predictive validity. Med Care. 2008;46:1134–1139.
21. Rosenbaum L. Beyond belief—how people feel about taking medications for heart disease. N Engl J Med. 2015;372:183–187.
22. Clifford S, Barber N, Elliott R, et al. Patient-centred advice is effective in improving adherence to medicines. Pharm World Sci. 2006;28:165–170.
23. Tarn DM, Paterniti DA, Kravitz RL, et al. How much time does it take to prescribe a new medication? Patient Educ Couns. 2008;72:311–319.
24. Timmerman L, Stellema R, Stronks DL, et al. Adherence to pharmacological pain therapy: the role of patients’ knowledge of pain medication. Pain Pract. 2014;14:701–708.

chronic pain; adherence; beliefs; medication

Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.