Breakthrough cancer pain should be properly assessed for better-personalized treatment plan. The Breakthrough Pain Assessment Tool is a 14-item tool validated in English developed for this purpose; no French version is currently available and validated. This study aimed to translate it in French and assess the psychometric properties of a French version of the Breakthrough Pain Assessment Tool (BAT-FR).
First, translation and cross-cultural adaptation of the 14 items (9 ordinal and 5 nominal) of the original BAT tool in French language was made. Second, assessments of validity (convergent, divergent and discriminant validity), factorial structure (exploratory factor analysis) and test-retest reliability of the 9 ordinal items were done with data of 130 adult cancer patients suffering from breakthrough pain in a hospital-academic palliative care center. Test-retest reliability and responsiveness of total and dimension scores derived from these 9 items were also assessed. Acceptability of the 14 items was also assessed on the 130 patients.
The 14 items had good content and face validity. Convergent and divergent validity, discriminant validity and test-retest reliability of the ordinal items were acceptable. Test-retest reliability and responsiveness of total and dimensions derived from ordinal items were also acceptable. The factorial structure of the ordinal items had two dimensions similar to the original version: “1—pain severity and impact” and “2—pain duration and medication”. Items 2 and 8 had a low contribution to the dimension 1 they were assigned and item 14 clearly changed of dimension compared with the original tool. The acceptability of the 14 items was good.
Citation: Perceau-Chambard E, Roche S, Tricou C, Mercier C, Barbaret C, Davies A, et al. (2023) Validation of a French version of the Breakthrough Pain Assessment Tool in cancer patients: Factorial structure, reliability and responsiveness. PLoS ONE 18(7): e0286947. https://doi.org/10.1371/journal.pone.0286947
Editor: Iftikhar Ahmed Khan, COMSATS University Islamabad, PAKISTAN
Received: September 27, 2022; Accepted: April 28, 2023; Published: July 10, 2023
Copyright: © 2023 Perceau-Chambard et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The supporting data are available in the Supporting information.
Funding: The Hospices Civils de Lyon "Young Researchers Grant" financially supported this work.
Competing interests: Andrew Davies and Katherin Webber developed the original English version of the tool.
Cancer pain is one of the most prevalent cancer related symptom [1, 2], resulting in disturbing patient’s quality of life and functioning, and increasing the burden of symptoms belonging to the same cluster, such as fatigue, insomnia or mood disturbance [3–5]. Cancer pain can be characterized by the different types of pain (somatic nociceptive pain, visceral nociceptive pain, or neuropathic pain) , but also by its manifestation as persistent, transient, or described as a breakthrough cancer pain. Breakthrough cancer pain is a transient exacerbation of pain that occurs despite relatively stable controlled background pain which might be triggered by a predictable specific event or an unpredictable one . It affects 40 to 80% of cancer patients , depending on cancer site, stage, treatments and intercurrent diseases. It has various characteristics, which are different between individuals and vary over time for a same individual. However, common features includes a quick onset up to three minutes, short duration up to thirty minutes and a moderate to severe intensity [9, 10].
A prerequisite for improving patients’ quality of life through effective pain management is proper pain assessment , it will guide the personalized pain medication plan . Considering its brief and sudden characteristics, breakthrough cancer pain is not properly assessed using generic pain assessment tools. This might prevent clinicians to adequately identify it, and chose specific medications such as intranasal fentanyl sprays or sublingual fentanyl pills [6, 12].
In the purpose of better assessing breakthrough pain, Webber et al. have developed and validated the Breakthrough Pain Assessment Tool (BAT) . This tool is a fourteen item tool of which nine are related to pain and five to its treatments , which is now widely recognized as a valid and reliable tool for use in clinical settings. It has been translated in various languages [14, 15], but not into French, preventing French languages countries to use it while French is the fifth most spoken language worldwide.
Our study aimed to translate the BAT into French and assess the psychometric properties (validity, reliability and responsiveness) of a French version of the Breakthrough Pain Assessment Tool (BAT-FR).
Material and methods
The Breakthrough Pain Assessment Tool (BAT)
The BAT is a 14 question tool designed to clinically assess breakthrough pain in cancer patients . The tool includes nine questions related to pain and five related to its management. The first question uses a body shape to locate the painful areas. The next questions uses free text (four questions), 10-points rating scales (six questions), and categorical scales (three questions).
Translation and adaptation of the original tool
We used a process of translation-adaptation adapted from the one proposed by the World Health Organization . First, a French native speaker made a forward translation laying on keeping the units of meaning (rather than an exact literal translation) from the original English version to a French one.
Second, a review board of pain experts reviewed the French version for identification of discrepancies or inadequate translations. Third, an English native speaker backward translated the French version into an English one. It was then, reviewed again by the review board to assess its equivalence with the original BAT. There was no need to modify the previous French version.
Finally, ten cancer patients suffering from pain were asked to assess the readability, and understandability of the tool. Slight changes have been made after this step. The BAT-FR is provided as S1 File.
This entire process with experts and a small sample of patients gave evidence for good content and face validity. The overall process was co-supervised by the original developers of the instrument who are co-authors of this article.
Study settings and participants
Data collection ran from March 2015 to September 2019 in the Lyon Sud University Hospital Palliative Care Center.
Participants and setting.
The study was conducted in the Lyon’s University Hospital—France. The inclusion settings were a University hospital palliative care center providing palliative care services for more than 1000 patients yearly.
All patients visiting the palliative care center were consecutively screened for inclusion. Inclusion criteria were: being over 18 years old, having cancer pain, having a regular opioid treatment for a background cancer-related pain for at least one week, and having been diagnosed from breakthrough cancer-pain by a specialist cancer pain physician. Inclusion criteria also included being a fluent French speaker, able to understand written French, to complete the questionnaire and to sign informed consent. Non-inclusion criteria were already known cognitive impairment (when listed in the medical history of the patient), and overwhelming fatigue.
1268 patients were screened for inclusion.
Exclusion criteria were the absence of pain (597), refusing to take part (20), agony (207), tiredness (13), inability to consent (235), trouble in understanding French (14), having already participated (23). Seven were excluded for an unknown reason. It resulted in one-hundred and thirty one included patients who received a questionnaire.
Procedure and data collection.
Participants had to fill the study questionnaire on inclusion, 24h after and 7 days after (noted as inclusion, visit 2 and 3 in the following). The test conditions were the same at each measurement. The questionnaire had two parts, the first recorded socio-demographic data (gender, age, cancer type and localizations, metastatic status, and Performance Status rated using the five-point Eastern Cooperative Oncology Group (ECOG) tool , the second was made of the BAT-FR (described above) and a validated French version of the Brief Pain Inventory (BPI) [18, 19].
The BPI is an instrument designed to assess the severity of pain and the functional impairment due to pain. The instrument includes one categorical scale question, one body shape for the localization of pain, four 10-points Likert-scales for rating pain intensity at various times, one free-text question and seven numerical scales to assess the impact of pain on the patient’s daily life.
As for the English version, the nine ordinal items were used to obtain scores and the five nominal items gave additional information and help to interpret the scores. To determine the sample size for factor analysis, we used the rule of 5 x number of parameters . This rule gave 130 patients because the model of the confirmatory factor analysis on the nine ordinal had 26 parameters (with the same structure as the English structure and items with 11 response categories considered as continuous). Moreover, this sample size was greater than the number of parameters for the exploratory factor analysis (up to 36 for four dimensions). According to COSMIN guidelines, a sample size of 130 was considered adequate to assess the other properties of the questionnaire .
Demographic and clinical characteristics of the sample.
Continuous variables were described using the median, the quartiles and the range values (and mean±SD for age and delay). Categorical variables were described using the number and percentage of patients in each category.
Acceptability of the BAT-FR.
Acceptability was assessed using the percentage of missing data for each of the 14 items at each assessment. A cut-off of 5% of missing data was chosen for the definition of non-acceptability .
Factorial structure of the BAT-FR.
As for the original English version, factor analysis was on the nine ordinal items: n°2, 5, 6, 7, 8, 9, 11, 12 and 14 (see Weber 2014 , Table 3). The five other items were nominal (free text or no order in the response categories) and thus could not be subject to a factor analysis; they gave additional information for the clinicians to interpret the ordinal items.
To assess if factor analysis is relevant, a test similar to the Bartlett’s sphericity test was used: the Chi-Square Test of Model Fit for the Baseline Model of uncorrelated dependent variables, available in Mplus for model fit information (p-value ≤ 0.05 indicates that a factor analysis may be useful with the data) .
To assess if the structure of the BAT-FR is the same as that of the English version , a confirmatory factor analysis (CFA) was conducted on the scores of the nine ordinal items on inclusion. It showed that the model of the English version (two dimensions and each of the nine ordinal items reflecting only the dimension it was supposed to measure) did not fit to data of BAT-FR (results not shown).
Therefore, an exploratory factor analysis (EFA) has been done on the nine ordinal items to produce hypotheses on (i) the number of dimensions and (ii) the structure of the items for the retained dimensions . As some items are ordinal, the WLSMV (Weighted Least Squares Mean and Variance adjusted) estimator was used to estimate the parameters of the EFA models from the correlation matrix (polychoric, polyserial or Pearson correlation). Several EFA models with different number of dimensions (from one to four) have been compared using global fit indexes. The number of dimensions retained was the smallest one with indexes indicating a good fit (parsimony criterion). The global fit indexes used to assess the fit of the EFA models were: Comparative Fit Index (CFI) and Tucker-Lewis Index (TLI) with good fit if >0.95 , Root Mean Square Error of Approximation (RMSEA) and its 90% confidence interval (CI) with close fit if <0.05 and fair fit in the 0.05 to 0.08 range , Standardized Root Mean square Residual (SRMR) with good fit if <0.08 . To make the dimensions interpretable and delineate the structure, oblique rotations CF-EQUAMAX and CF-FACPARSIM were used in the EFA models for the development of a new structure . After estimation of the number of dimensions, the criteria for generating a structure were the following: (i) a factor loading value greater than 0.4 was retained to assign an item to a dimension , and (ii) an item which reflects more than one dimension was assigned to the dimension for which its factor loading was the highest.
Discriminant validity of the nine BAT-FR ordinal items between groups determined by ECOG scores (0, 1 or 2 vs 3 or 4), global impression of pain control, and changes in treatments’ management at inclusion, was assessed by Kruskal-Wallis tests. Differences of items scores between groups of ECOG scores and more differences for breakthrough pain compared to background pain were supposed to show a good discriminant validity .
Convergent and divergent validity.
Convergent validity between BAT-FR and BPI scores was assessed by Spearman correlations between the nine BAT-FR ordinal items scores and BPI items and dimensions scores at inclusion. The higher the number of medium or large correlations, the better the convergent validity is. Correlation coefficient values of > 0.1 were considered as a small correlation, >0.3 a medium correlation and >0.5 a large correlation (in absolute value) . For analgesic treatments at inclusion, convergent validity was assessed by polychoric correlations between BAT-FR ordinal items scores and oral administration of transmucosal fentanyl and divergent validity was assessed by Spearman correlations between BAT-FR ordinal items scores and background medication dosage. These two validities were established if correlations with breakthrough analgesia were higher than correlations with background analgesia .
The test-retest reliability of each of the nine ordinal item of the BAT-FR was estimated with data at inclusion and visit 2 using weighted Kappa coefficient (κw). Fleiss-Cohen weighting scheme was used. Results’ interpretations were: κw ≥0.81 indicated almost perfect agreement, 0.61≤ κw ≤0.80 substantial agreement, 0.41≤ κw ≤0.60 moderate agreement, 0.21≤ κw ≤0.40 fair agreement, and κw ≤0.20 slight agreement . The test-retest reliability of the two dimensions (as specified in the English BAT) and the total score of the BAT-FR were estimated using Intraclass Correlation Coefficient (ICC). The results on ICC were interpreted with the previous thresholds as weighted kappa and ICC are equivalent .
The responsiveness of the total score of the BAT-FR was estimated using correlation between (a) change of total score of the BAT-FR between visits 1 and 3 and (b) the assessment at visit 3 of Breakthrough Pain (BP) compared with the last week in an ordinal response . Polyserial correlation was used, a correlation greater than 0.5 was considered large . The responsiveness of the two dimensions (as specified in the English BAT) were assessed with the same method (total score of each dimension instead of total score of BAT-FR).
All tests were 2-tailed, and p < 0.05 was considered for statistical significance.
The CFA and EFA were carried out using Mplus, version 8.4 . All other analyses used SAS software, version 9.4.
Oral information was delivered to each patient screened for inclusion. If the patient expressed agreement for inclusion, then, the inclusion criteria were verified. If all criteria were met, the patient was given a full oral and written information before signing an informed consent form.
The Committee For Persons’ Protection Sud-Est IV was seized on August, 25th 2014. It reviewed and approved the protocol on the 10th September 2014 (Reference number: L14-155).
Participants (Table 1)
One hundred and thirty one participants were included.
The participants were 62 ± 13 years old and diagnosed from cancer for 3.6 ± 3.7 years. Most suffered from gastro-intestinal or urological cancers, which were mainly on a metastatic stage. The etiopathogeny of the breakthrough cancer pain was cancer-related for nine in ten participants and had mixed characteristics for two thirds of participants and only nociceptive for around a third of participants. Neuropathic breakthrough cancer pain was minority.
All items had less than 5% of missing items suggesting a good acceptability, except at visit 3.
The model with three dimensions had a good fit but unreliable estimates (for both rotations, factor loading >1 and negative variance; bad condition number (for CF-EQUAMAX only) . The two-dimension model was retained because it had a fair fit with reliable parameter estimates (all factor loading <1 and no negative variance, good condition number), and a clear and stable structure whatever the oblique rotation (cf. S4 File for goodness-of-fit and model assumptions of the EFA models). The solution with the oblique rotation CF-FACPARSIM is presented in Table 2 (cf. S3 File for CF-EQUAMAX). The first dimension included five items (2, 6, 7, 8 and 9) and the second dimension included four items (5, 11, 12 and 14). Nevertheless items 2 and 8 had a low contribution to the first dimension and might be excluded (factor loading <0.40) and item 14 was not on the same dimension than for English and Dutch versions (with items 2, 6, 7, 8 and 9). The correlation between the two dimensions was low for the two rotations (r = 0.16 for both).
In the following parts of the article, the two dimensions are, unless otherwise stated, the dimensions based on the structure of the English version.
Discriminant validity (Table 3)
There was more differences for breakthrough pain than for background pain for items 8, 11 and 12 in control of pain (p-values <0.05) and for items 8 and 12 in changes in management of treatments (p-values <0.05). However, there was more differences for background pain than for breakthrough pain for item 2 in control of pain and changes in management of treatments, and for item 7 in changes in management of treatments (p-values <0.05). There were no significant differences of items scores between ECOG groups (all p-values were > 0.34 for the nine items).
Convergent and divergent validity
Correlations between the BAT-FR and BPI were small to medium (Table 4). Regarding the correlation between the BAT-FR and analgesic treatments (Table 5), the correlation with taking breakthrough analgesia treatment was greater than the correlation with background medication dosage for six items (5, 6, 7, 9, 11 and 14). The difference between coefficients was lower than 0.1 for half items and all correlations were small.
Test-retest reliability (Table 6)
All coefficients (weighted kappa and ICC) ranged between 0.31 and 0.66. The reliability was substantial for three items (5, 6 and 8), dimension “breakthrough pain severity and impact”, and BAT-FR total score. It was moderate for four items (2, 7, 9 and 12), and dimension “breakthrough pain duration and medication efficacy”. Reliability was fair for items 11 and 14.
The assessments of breakthrough pain by patient were available for 102 patients. The polyserial correlations of assessment by patient with change in total score and with change in English-like dimension “breakthrough pain severity and impact” were large (r = 0.59 on n = 89 and n = 95 patients respectively). The polyserial correlation with change in English-like dimension “breakthrough pain duration and medication efficacy” was medium (r = 0.31 on n = 91 patients).
This study aimed to assess the psychometric properties of a translation of the BAT in French language and in particular its validity. By comparing with the results of the original BAT , items 2 (frequency) and 8 (distress) should be excluded from the items of the BAT-FR, and item 14 (side effects) would change dimension. However, the structure for the BAT-FR has two dimensions like the original one; with for each dimension at least 3 items in common with the corresponding dimension of the original BAT. We therefore consider the dimensions of the BAT-FR as close to those of the original version. However, the values of the factor loadings are mostly lower for the BAT-FR compared to the English BAT. Note that, for the Dutch version of the BAT , the structure is identical to the English BAT but with factor loadings of generally lower values as well, therefore the results of the exploratory factor analysis on the data of the BAT-FR are not in contradiction with the results of the English BAT.
It is therefore necessary to decide on keeping, and the dimensions to which to assign items 2 (frequency), 8 (distress) and 14 (side effects) in the BAT-FR. This must be done taking into account the clinical relevance, the meaning of the two dimensions for BAT-FR and for English BAT and also the results of the other aspects of the validation: convergent validity with BPI (poor for the item 2, moderate for items 8 and 14), convergent and divergent validities with pain treatments (poor for items 2, 8 and 14), discriminant validity (poor for items 2 and 14, moderate for item 8), reliability (moderate for item 2, substantial for item 8, fair for item 14).
First, decision must be made on items 2 (frequency) and 8 (distress). Both items measures key concepts in pain assessment that are clinically highly relevant for clinicians. For item 2, it can reflect the equilibration of the background pain treatment and guide tailored treatment adaptation [34, 35]. While item 8 reflects the global impact of pain on patients. Inadequately controlled pain leads to anxiety, depression and sleep disorders . Assessing the distress associated with the experience of pain might trigger non-pharmacological interventions for managing it, such as psychosocial support, resulting in better quality of life. We therefore, would suggest keeping these items into the BAT-FR by assigning them the dimension on which the factor loading is the greatest. The “pain severity and impact” English BAT dimension would then be chosen making the structure of the BAT-FR even closer to that of the original version.
Second, for item 14 (side effects), the values of the factor loadings clearly indicate a change in dimension (from the English BAT dimension “pain severity and impact” to “pain duration and medication”). Both dimensions are relevant for this item. In the original English version, it might be argued that medication side effects are part of the patient’s burden related to pain and contributes to its impact which explains that both contributes to the same dimension . However, the treatment side effects are key information to make a choice on whether or not pursing the medication. It has a strong impact on medication adherence [38, 39], and it might lead to treatment discontinuation or adaptation . Therefore, we would argue that item 14 is clinically more relevant to be included in dimension “pain duration and medication”, despite the fact that it differs for the original version.
The differences observed between the two tools might be related to the sample involved. When compared to the sample used by Webber et al. for the initial instrument validation, both studies had mainly gastro-intestinal cancers and urological cancers, suggesting that these cancers might be more prone to trigger breakthrough cancer pain. However, in our study, the majority of included patients had impaired Performance Status when the majority of patients included in Webber et al. had 1 to 2 ECOG scores. Our population was also mainly concerned by mixed cancer pain when there were a minority compared to nociceptive pain in Webber et al. study. These differences in our population might influence the experience of pain [41, 42], resulting in the observed differences.
The BAT-FR had a close structure with the original version except for the item 14 related to side effects. All properties (validity, reliability and responsiveness) have moderate quality, but are relatively consistent with those of the original version. The BAT-FR needs assessment on another sample to confirm the structure and its properties.
S1 File. French version of the Breakthrough Pain Assessment Tool (OFEA).
S2 File. Indexes of fit for exploratory factorial analysis models with one to four dimensions estimated with data of n = 130 patients.
CFI: comparative fit index; TLI: Tucker-Lewis Index; RMSEA: root mean square error of approximation; CI: Confidence interval; SRMR: Standardized Root Mean Square Residual.
S3 File. Items grouping after CF-EQUAMAX rotation: Estimate of the factor loadings associated to each of the two dimensions of the French BAT estimated with data of n = 130 patients.
S4 File. Table reporting the goodness-of-fit and model assumptions of the EFA models.
The condition number is an index of good quality of numerical results (ratio of smallest to largest eigenvalue for the Information Matrix, good if >10−6) e.g. no problem as multicolinearity of items.
We acknowledge all participants who agreed to take part to the study.
The authors sorely Mrs Dr. Catherine Mercier, PhD, who passed away and has extensively, contributed to this work. Beyond this study, Catherine Mercier has greatly contributed to improve the quality of biostatistics in research to always purse the goal of a better and trustworthy research in health.
- 1. Hearn J, Higginson IJ. Cancer pain epidemiology: a systematic review. Cancer Pain: Assessment and Management. Cambridge University Press; 2003.
- 2. van den Beuken-van Everdingen MHJ, Hochstenbach LMJ, Joosten EAJ, Tjan-Heijnen VCG, Janssen DJA. Update on Prevalence of Pain in Patients With Cancer: Systematic Review and Meta-Analysis. Journal of Pain and Symptom Management. 2016;51: 1070–1090.e9. pmid:27112310
- 3. Cheng KKF, Lee DTF. Effects of pain, fatigue, insomnia, and mood disturbance on functional status and quality of life of elderly patients with cancer. Critical Reviews in Oncology/Hematology. 2011;78: 127–137. pmid:20403706
- 4. Rodriguez C, Ji M, Wang H-L, Padhya T, McMillan SC. Cancer Pain and Quality of Life: Journal of Hospice & Palliative Nursing. 2019;21: 116–123. pmid:30829932
- 5. Kroenke K, Theobald D, Wu J, Loza JK, Carpenter JS, Tu W. The Association of Depression and Pain with Health-Related Quality of Life, Disability, and Health Care Use in Cancer Patients. Journal of Pain and Symptom Management. 2010;40: 327–341. pmid:20580201
- 6. Hui D, Bruera E. A Personalized Approach to Assessing and Managing Pain in Patients With Cancer. JCO. 2014;32: 1640–1646. pmid:24799495
- 7. Davies AN, Dickman A, Reid C, Stevens A-M, Zeppetella G. The management of cancer-related breakthrough pain: Recommendations of a task group of the Science Committee of the Association for Palliative Medicine of Great Britain and Ireland. European Journal of Pain. 2009;13: 331–338. pmid:18707904
- 8. Mercadante S, Radbruch L, Caraceni A, Cherny N, Kaasa S, Nauck F, et al. Episodic (breakthrough) pain: Consensus conference of an expert working group of the European Association for Palliative Care. Cancer. 2002;94: 832–839. pmid:11857319
- 9. Portenoy RK, Payne D, Jacobsen P. Breakthrough pain: characteristics and impact in patients with cancer pain: Pain. 1999;81: 129–134. pmid:10353500
- 10. Davies A, Buchanan A, Zeppetella G, Porta-Sales J, Likar R, Weismayr W, et al. Breakthrough Cancer Pain: An Observational Study of 1000 European Oncology Patients. Journal of Pain and Symptom Management. 2013;46: 619–628. pmid:23523361
- 11. Wells N, Pasero C, McCaffery M. Improving the Quality of Care Through Pain Assessment and Management. In: Hughes RG, editor. Patient Safety and Quality: An Evidence-Based Handbook for Nurses. Rockville (MD): Agency for Healthcare Research and Quality (US); 2008. http://www.ncbi.nlm.nih.gov/books/NBK2658/
- 12. Hagen NA, Biondo P, Stiles C. Assessment and management of breakthrough pain in cancer patients: Current approaches and emerging research. Current Science Inc. 2008;12: 241–248. pmid:18625100
- 13. Webber K, Davies AN, Zeppetella G, Cowie MR. Development and Validation of the Breakthrough Pain Assessment Tool (BAT) in Cancer Patients. Journal of Pain and Symptom Management. 2014;48: 619–631. pmid:24766740
- 14. Oldenmenger WH, Lucas A, van der Werff GFH, Webber K, Visser D, van der Velden AWG, et al. Validation of the Dutch Version of the Breakthrough Pain Assessment Tool in Patients With Cancer. Journal of Pain and Symptom Management. 2020;59: 709–716.e2. pmid:31877346
- 15. Shin J, Cho SJ, Lee J, Choi YS. Validation of the Korean Version of the Breakthrough Pain Assessment Tool in Cancer Patients. Journal of Pain and Symptom Management. 2017;54: 361–367. pmid:28711749
- 16. WHO | Process of translation and adaptation of instruments. In: WHO [Internet]. World Health Organization; [cited 9 Feb 2021]. https://www.who.int/substance_abuse/research_tools/translation/en/
- 17. Oken MM, Creech RH, Tormey DC, Horton J, Davis TE, McFadden ET, et al. Toxicity and response criteria of the Eastern Cooperative Oncology Group. Am J Clin Oncol. 1982;5: 649–655. pmid:7165009
- 18. Cleeland CS, Ryan KM. Pain assessment: global use of the Brief Pain Inventory. Ann Acad Med Singap. 1994;23: 129–138. pmid:8080219
- 19. Larue F, Colleau SM, Brasseur L, Cleeland CS. Multicentre study of cancer pain and its treatment in France. BMJ. 1995;310: 1034–1037. pmid:7728056
- 20. Bentler PM, Chou C-P. Practical Issues in Structural Modeling. Sociological Methods & Research. 1987;16: 78–117.
- 21. Mokkink LB, Terwee CB, Patrick DL, Alonso J, Stratford PW, Knol DL, et al. The COSMIN checklist for assessing the methodological quality of studies on measurement properties of health status measurement instruments: an international Delphi study. Qual Life Res. 2010;19: 539–549. pmid:20169472
- 22. Muthén LK, Muthén BO. Mplus User’s Guide. Eighth Edition. Los Angeles; 2017.
- 23. Asparouhov T, Muthén B. Exploratory Structural Equation Modeling. Structural Equation Modeling: A Multidisciplinary Journal. 2009;16: 397–438.
- 24. Hu L, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal. 1999;6: 1–55.
- 25. Browne MW, Cudeck R. Alternative Ways of Assessing Model Fit. Sociological Methods & Research. 1992;21: 230–258.
- 26. Asparouhov T, Bengt O. M. SRMR in Mplus. 2018.
- 27. Schmitt TA. Current Methodological Considerations in Exploratory and Confirmatory Factor Analysis. Journal of Psychoeducational Assessment. 2011;29: 304–321.
- 28. Ford JK, MacCALLUM RC, Tait M. The application of exploratory factor analysis in applied psychology: a critical review and analysis. Personnel Psychology. 1986;39: 291–314.
- 29. de Vet HCW, editor. Measurement in medicine: a practical guide. Cambridge; New York: Cambridge University Press; 2011.
- 30. Streiner DL, Norman GR, Cairney J. Health Measurement Scales. Oxford University Press; 2015.
- 31. Landis JR, Koch GG. The Measurement of Observer Agreement for Categorical Data. Biometrics. 1977;33: 159. pmid:843571
- 32. Fleiss JL, Cohen J. The Equivalence of Weighted Kappa and the Intraclass Correlation Coefficient as Measures of Reliability. Educational and Psychological Measurement. 1973;33: 613–619.
- 33. Cohen J. Statistical Power Analysis. Curr Dir Psychol Sci. 1992;1: 98–101.
- 34. Bruera ED, Portenoy RK, Bruera MD. Cancer Pain: Assessment and Management. Cambridge University Press; 2003.
- 35. Pasero C, McCaffery M. Pain assessment and pharmacologic management. 2011. https://www.clinicalkey.com/nursing/dura/browse/bookChapter/3-s2.0-C20090386638
- 36. Woo AK. Depression and Anxiety in Pain. Reviews in Pain. 2010;4: 8–12. pmid:26527193
- 37. Martel MO, Finan PH, Dolman AJ, Subramanian S, Edwards RR, Wasan AD, et al. Self-reports of medication side effects and pain-related activity interference in patients with chronic pain: a longitudinal cohort study. Pain. 2015;156: 1092–1100. pmid:25782367
- 38. Anastassopoulos KP, MS, Chow W, MPH, PharmD, Ackerman SJ, MSE, PhD, Tapia C, MHS, et al. Oxycodone-related side effects: Impact on degree of bother, adherence, pain relief, satisfaction, and quality of life. JOM. 2011;7: 203–215. pmid:21823551
- 39. Timmerman L, Stronks DL, Groeneweg JG, Huygen FJ. Prevalence and determinants of medication non-adherence in chronic pain patients: a systematic review. Acta Anaesthesiol Scand. 2016;60: 416–431. pmid:26860919
- 40. World Health Organization. WHO guidelines for the pharmacological and radiotherapeutic management of cancer pain in adults and adolescents. 2018. http://www.ncbi.nlm.nih.gov/books/NBK537492/
- 41. Zhao F, Chang VT, Cleeland C, Cleary JF, Mitchell EP, Wagner LI, et al. Determinants of Pain Severity Changes in Ambulatory Patients With Cancer: An Analysis From Eastern Cooperative Oncology Group Trial E2Z02. JCO. 2014;32: 312–319. pmid:24366929
- 42. Daniel HC, Narewska J, Serpell M, Hoggart B, Johnson R, Rice ASC. Comparison of psychological and physical function in neuropathic pain and nociceptive pain: Implications for cognitive behavioral pain management programs. European Journal of Pain. 2008;12: 731–741. pmid:18164225