Utilization of brief pain inventory as an assessment tool for pain in patients with cancer: a focused review.

The Pain Research Group of the world health organization (WHO) Collaborating Centre for Symptom Evaluation in Cancer Care had developed the Brief Pain Inventory (BPI), a pain assessment tool for use with cancer patients. The BPI measures both the intensity of pain (sensory dimension) and interference of pain in the patient's life (reactive dimension). The objective of this review paper was to provide a detailed update of existing evidence on applicability of BPI in evaluation of patients with cancer pain. The BPI demonstrated good construct and concurrent validity. It was translated and validated into many languages - Brazilian, Chinese, Greek, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Russian, Spanish, Taiwanese and Thai. The BPI was validated in patient populations such as bone metastases, breast cancer and postoperative cancer patients. The BPI can be used both as a quantitative or a qualitative measure for statistical analysis. The BPI was a powerful tool and, having demonstrated both reliability and validity across cultures and languages, was being adopted in many countries for clinical pain assessment, epidemiological studies, and in studies on the effectiveness of pain treatment. Future studies are warranted on its responsiveness and cross-cultural adaptation into other cancer pain syndromes and into other Indian languages.

INTRODUCTION

Pain is among the most common and distressing symptoms encountered by patients with advanced cancer and other terminal illnesses. Holen et al.[1] in their systematic review identified 64 pain assessment tools for patients with advanced cancer who were receiving palliative care and they found that common dimensions assessed were intensity, temporal pattern, treatment and exacerbating/relieving factors, location, and interference with health-related quality of life. They also found that many tools included dimensions and items of limited relevance for patients with advanced cancer. This might have reduced compliance and threatens the validity of the pain assessment. The authors recommended that new tools should reflect the clinical relevance of different dimensions and be user-friendly.

The first step in understanding cancer pain in diagnosis and treatment is classification. Knudsen et al.[2] found 92 papers in their systematic review, which reported six standardized classification systems (three of them systematically developed and partially validated) that included both pain characteristics and patient characteristics. Recently, a proposed mechanism-based classification for cancer pain was put forth by Kumar,[3] which would direct clinical decision-making in palliative care management of cancer pain. Haugen et al.,[4] on the contrary, found that there was no widely accepted definition, classification system or well-validated assessment tool for cancer-related breakthrough pain among 51 articles in another systematic review.

Whilst these inadequacies exist in understanding cancer pain, the second step in evaluation of patients with cancer pain to assess and quantify the pain by means of objective scoring methods which may either be patient- or caregiver-reported and/or clinician-rated is still rapidly growing in emphasis. Considerable attention toward pain control in patients with advanced cancer warrants better clinical assessment tools and evaluation of their applicability in palliative care.[5]

Various factors influence pain measurement in palliative care clinical practice, such as patient or caregiver compliance,[6] content of the assessment tool[1] and its relationship to physical functioning and quality of life.[7] Hjermstad et al.,[8] as part of The European Palliative Care Research Collaborative (EPCRC), aimed to develop an international consensus-based computerized pain assessment tool and in their systematic review identified 11 different pain assessment tools of which 9 were multidimensional, 7 were for pain intensity, and 5 were on pain management.

The need for efficient clinical pain assessment and measurement tool in palliative care for patients with cancer witnessed the effort of an Expert Working Group which was convened under the auspices of the Steering Committee of the Research Network of the European Association of Palliative Care[9] to review the status of the use of pain measurement tools (PMTs) in palliative care research conducted in a multilingual–multicenter setting. Based on a literature review and the experts’ opinion, the authors opined that among the multidimensional questionnaires designed to assess pain, the McGill Pain Questionnaire and Brief Pain Inventory (BPI) are valid in many multilingual versions.[10] The BPI in addition had the ability to assess the effect of cancer pain on the patient.[11]

In order to address the significant public health problem of poorly controlled cancer pain throughout the world, the Pain Research Group of the world health organization (WHO) Collaborating Centre for Symptom Evaluation in Cancer Care had developed the BPI, a pain assessment tool for use with cancer patients. The BPI measures both the intensity of pain (sensory dimension) and interference of pain in the patient's life (reactive dimension). It also queries the patient about pain relief, pain quality, and patient perception of the cause of pain.[12]

The objective of this review paper was to provide a detailed update of existing evidence on applicability of BPI in evaluation of patients with cancer pain.

APPLICABILITY OF A MEASUREMENT TOOL

The applicability of a measurement tool depends upon the properties of the tool that determine its application as an outcome measure. Conventionally such properties are termed as measurement properties, whilst some authors mention them as psychometric properties or clinimetric properties of the tool. Psychometric properties involve construction and validation of measurement instruments, whereas clinimetric properties involve the clinical application into screening, diagnosis and prognosis.[13] The applicability is thus determined by two factors: (1) how accurately it measures what it was supposed to measure and (2) how accurately the measure reflects differences due to time, repeated testing, situation, condition, intervention and testers. The first property is termed as validity and the second is reliability (component–component inter-relationship, repeated testing, between-testers, and between-trials) and responsiveness (over time, upon intervention).[14] The measurement property of validity is the construct validity [Figure 1] which is further subcategorized into translation validity (measured as either face validity or content validity) and criterion-related validity (measured as either of its four subtypes: predictive validity, concurrent validity, convergent validity, and discriminant validity).[15]

Figure 1

Schematic flow diagram showing types and subtypes of validity as a measurement property

VALIDITY OF A MEASUREMENT TOOL

Construct validity is the approximate truth of the conclusion that an operationalization accurately reflects its construct. In translation validity, the focus is on whether the operationalization is a good reflection of the construct. This approach is definitional in nature – it assumes to have a good detailed definition of the construct and that we can check the operationalization against it. In criterion-related validity, we examine whether the operationalization behaves the way it should given the theory of the construct. This is a more relational approach to construct validity. It assumes that our operationalization should function in predictable ways in relation to other operationalizations based upon our theory of the construct.[16]

In face validity, we look at the operationalization and see whether “on its face” it seems like a good translation of the construct. In content validity, we essentially check the operationalization against the relevant content domain for the construct.[17]

In predictive validity, we assess the operationalization's ability to predict something it should theoretically be able to predict. In concurrent validity, we assess the operationalization's ability to distinguish between groups that it should theoretically be able to distinguish between. In convergent validity, we examine the degree to which the operationalization is similar to (converges on) other operationalizations that it theoretically should be similar to. In discriminant validity, we examine the degree to which the operationalization is not similar to (diverges from) other operationalizations that it theoretically should not be similar to.[18] The classification of validity is shown in Figure 1.

RELIABILITY OF A MEASUREMENT TOOL

The subtypes of reliability are inter-rater reliability, test–retest reliability, parallel-forms reliability and internal consistency. The classification of reliability is shown in Figure 2.

Figure 2

Schematic flow diagram showing types and subtypes of reliability as a measurement property

Inter-rater or inter-observer reliability is used to assess the degree to which different raters/observers give consistent estimates of the same phenomenon. Test–retest reliability is used to assess the consistency of a measure from one time to another. Parallel-forms reliability is used to assess the consistency of the results of two tests constructed in the same way from the same content domain. Internal consistency reliability is used to assess the consistency of results across items within a test. It has four subtypes: average inter-item correlation (the average inter-item correlation uses all of the items on our instrument that are designed to measure the same construct), average item-total correlation (this approach also uses the inter-item correlations; in addition, we compute a total score for the items and use that as another variable in the analysis), split-half reliability (random division of all items that purport to measure the same construct into two sets and then administer the entire instrument to a sample of people and calculate the total score for each randomly divided half; the split-half reliability estimate is simply the correlation between these two total scores) and Cronbach's alpha (computation of one split-half reliability and then random division of the items into another set of split halves and recomputation, until computation of all possible split-half estimates of reliability is done; Cronbach's alpha is mathematically equivalent to the average of all possible split-half estimates).[15]

Often, a measurement tool which has a good validity need not be reliable; and a reliable tool need not necessarily be valid.[19] The responsiveness of a measure is an extension of reliability where measurement error and minimum clinically important difference (MCID) are defined. Responsiveness is an indicator of how much change in the measurement score is the real change. The lower limit of this measure gives the measurement error and the upper limit gives the MCID. The MCID is useful for sample size estimation in intervention studies and for interpretation of treatment effect size upon completion of the study.[20]

BPI: MEASUREMENT PROPERTIES

Construct validity

Atkinson et al.[21] studied the construct validity of BPI using a confirmatory factor analysis for the two factors (pain intensity and pain interference) based on a range of demographic variables (disease, age, ethnicity groups) in patients with HIV/AIDS and cancer. A three-factor model (i.e., pain intensity, activity interference, and affective interference) was then suggested for use in clinical research as against the existing two-factor model of BPI.

Concurrent validity

As part of a longitudinal prospective study, Philip et al.[22] validated the modified Edmonton Symptom Assessment System (ESAS) with the Rotterdam Symptom Checklist and the BPI, the two instruments widely used in patients receiving palliative therapy for cancer. The authors concluded that the modified ESAS is a valid, self-administered instrument to assess the symptoms for patients from differing palliative care settings. The authors used BPI as a criterion gold standard assessment tool.

CROSS-CULTURAL ADAPTATION OF THE BPI

The cross-cultural adaptation of a health status self-administered questionnaire for use in a new country, culture, and/or language necessitates use of a unique method to reach equivalence between the original source and target versions of the questionnaire. It is now recognized that if measures are to be used across cultures, the items must not only be translated well linguistically, but also must be adapted culturally to maintain the content validity of the instrument at a conceptual level across different cultures.[23] Also, there exist standardized methodological approaches to validate a questionnaire for pain assessment/measurement in the process of cross-cultural adaptation.[24] The BPI was translated into 14 languages – Brazilian,[25] Chinese,[26] Greek,[27] Hindi,[28] Italian,[29] Japanese,[30] Korean,[31] Malay,[32] Norwegian,[33] Polish,[34] Russian,[35] Spanish,[36] Taiwanese[37] and Thai[38] around the world. Comparison of these studies is provided in Table 1.

Table 1

Comparison of studies on cross-cultural adaptation of the brief pain inventory in patients with cancer pain

BPI: USE IN SPECIFIC PATIENT POPULATIONS

Bone metastases

Wu et al.[39] validated the psychometric properties of the BPI and its Pain and Interference subscales in patients with clinically significant metastatic bone pain requiring palliative radiotherapy and examined the differences in BPI subscales among predefined subgroups of 258 bone metastases patients. High internal consistency of the BPI subscales of Pain, Activity interference, and Affect interference was demonstrated by Cronbach's alpha between 0.81 and 0.89. Removing sleep interference improved model fit in confirmatory factor analysis. The BPI revealed an alarming pattern in patients with lower body metastases, who reported substantial interference of activity even though pain levels were mild or moderate. BPI was thus able to indicate such patients who might require prompt clinical attention to better meet their needs.

Harris et al.[40] studied BPI to evaluate the patient response to radiotherapy and evaluated 199 patients referred to the Rapid Response Radiotherapy Program for palliative radiotherapy of symptomatic bone metastases. Patients rated the intensity and functional interference of their pain at the irradiated sites according to the BPI before and 2 months after radiotherapy. All pain intensity and interference scores for evaluable patients were significantly lower at 2 months. Response rates differed depending on the definition of pain intensity. An overall response rate was observed in 66, 58, and 54% of patients for worst, average, and current pain, respectively. Worst pain showed the best correlation with functional interference. Responders reported significantly larger decreases in functional interference scores at follow-up in general activity, normal work, enjoyment of life and average functional interference.

Canine bone cancer

Brown et al.[41] validated the Canine Brief Pain Inventory (CBPI), which was based on the human BPI, in a canine model of spontaneous bone cancer. One hundred owners of dogs with bone cancer self-administered the CBPI on three occasions to test the reliability, validity, and responsiveness of the measure. Factor analysis, internal consistency, convergent validity, and an extreme group validation assessment were completed using the responses from the first administration of the CBPI. Test–retest reliability was evaluated using two administrations of the instrument, 1 week apart. Responsiveness was tested by comparing responses 3 weeks apart. The “severity” and “interference” factors hypothesized based on the BPI were demonstrated in the CBPI in dogs with bone cancer. Internal consistency was high (Cronbach's alpha, 0.95 and 0.93), as was test–retest reliability (kappa, 0.73 and 0.65). Convergent validity was demonstrated with respect to quality of life (r = 0.49 and 0.63). Extreme group validation against normal dogs showed significantly higher factor scores. This innovative approach to preclinical outcome development, validating a preclinical outcome measure that directly corresponds to an outcome measure routinely used in clinical research, applied to a readily available animal model of spontaneous disease could transform the predictive ability of preclinical pain studies.

Breast cancer

Among patients with metastatic breast cancer, Castel et al.[42] assessed the impact of baseline clinical and demographic risk factors on patients reaching different pain severity and interference scores. Pain was measured by the BPI severity and interference with daily living 0–10 subscales. The authors’ findings that non-Caucasian race and restricted performance status were associated with greater pain hazards over time confirm previous cross-sectional findings that these characteristics are pain risk factors. Because they found that the most influential demographic and clinical baseline factors had predictive value for worsening outcomes as early as cut-point 5, they recommended that pain management strategies use cut-points informed by risk factors for worsening outcomes as cues for earlier intervention, thus delaying or preventing worst pain among women with metastatic disease who were at greatest risk.

Surgical patients with cancer

Tittle et al.[43] examined the psychometric characteristics of the BPI for surgical patients with cancer and compared the validity and reliability results between 159 surgical and 229 medical patients with cancer. The BPI was administered to the patients once and a pain visual analog scale (VAS) was administered to the patients three times. The VAS was correlated with individual items of the BPI and with the Pain Interference Subscale of the BPI; correlations were conducted separately for medical and surgical patients as a study of validity. Correlations between the Pain Interference Subscale and the other items on the BPI were similar for both the groups. Correlations between the VAS and the Pain Interference Subscale of the BPI were equally high for the medical and surgical oncology groups. Reliability evaluated by the coefficient alpha was very high for the medical and surgical oncology groups. The authors found the BPI to be equally valid and reliable for medical and surgical male, Caucasian patients with cancer.

USE OF BPI FOR DIAGNOSIS

Comparison of cancer and non-cancer patients

Holen et al.[44] aimed to explore how patients respond to pain interference items by comparing responses from patients who had cancer with patients who had non-cancer chronic pain (NCCP), and to explore how different levels of health-related quality of life affect upon pain's interference with functions. The authors studied 300 patients with cancer and 286 patients with NCCP who were asked to complete the BPI and the European Organization for Research and Treatment of Cancer's Quality of Life Questionnaire (EORTC QLQ-C30). The pain interference items were indexed into total interference, interference with physical functions, and interference with psychologic functions. age, sex, and all EORTC QLQ-C30 scales. The authors found that cancer patients reported higher values of physical interference than NCCP patients with the same level of pain intensity. NCCP patients reported higher values of psychologic interference than cancer patients. The study results indicated that patients were unable to report isolated pain's interference using the BPI. When reporting pain's interference with physical functioning, the level of physical functioning was more important than the level of pain.

Influence of pain on function

Stenseth et al.[45] investigated if pain interference items are influenced by factors other than pain. In the original version of the BPI, the patients were asked how, during the last 24 hours, pain had interfered with functions. In the revised BPI, this question was changed to how, during the last 24 hours, these functions were affected in general. Forty-eight of the 55 included patients completed both assessments. The BPI pain intensities scores and the health-related quality of life scores were similar on the two study days. Except for mood, this study observed no significant distinctions between the patients’ BPI interference items scores in the original (pain influence on function) and the revised BPI (function in general). Seventeen patients reported higher influence from pain on functions than the total influence on function from all causes. The authors observed similar scores in the original BPI interference scores (pain influence on function) compared with the revised BPI interference scores (decreased function in general). Their finding might imply that the BPI interference scale measures were partly responded to as more of a global interference measure.

DISCUSSION

This review is the first of its kind on an assessment tools such as BPI used for patients with cancer pain. Reviews on BPI and its measurement properties on other patient populations could not be found, and albeit its common use, the properties had not been summarized in a practically scientific manner. Other clinical tools commonly reported earlier were Edmonton Symptom Assessment,[22] Rotterdam Symptom Checklist,[36] Pain Management Index,[262735] Wisconsin Brief Pain Questionnaire[46] and McGill Pain Questionnaire.[25] The other measurement scales used for pain assessment in patients with cancer were VAS, a numerical rating scale from 0 to 10 (NRS), a verbal rating scale (VRS), the Italian Pain Questionnaire (Italian version of the McGill Pain Questionnaire), and the Integrated Pain Score (IPS).[47] Indirect activity-related pain measurements such as finger dynamometer were also used by other authors.[48] Of all the widely used variety of pain measures and scales, the numerical rating scales were shown to be more valid, reliable and responsive across settings and disease states in cancer pain.[49] The advantage of NRS is that it can also be administered verbally in addition to visual method.[50] Since BPI incorporates NRS for all items of both its subscales – pain severity and pain interference – this makes the BPI the best clinical measurement tool for pain in cancer patients.

From this review, information on the responsiveness or minimum clinically important change (MCIC) of BPI was not available in patient population with cancer pain. Future studies could address this issue by using patient-rated global impression of change as a criterion measure to derive the MCIC.[51] MCIC is useful for sample size estimation for prospectively planned, well-designed, randomized clinical trials.[52] Also, surprisingly though BPI was used as an outcome measure in a plethora of published clinical trials across all types of cancer, cross-cultural adaptation into disease-specific considerations was missing in the literature. Cross-cultural adaptation into many of the Indian languages[28] is obviously the need of the hour, taking into consideration language, time, setting and statistical methods.[53]

There was an increasing prevalence of cancer pain[54] in six major types of cancer: 70% in head and neck cancer, 59% in gastrointestinal cancer, 55% in lung cancer, 54% in breast cancer, 52% in urogenital cancer and 60% in gynecological cancer, each of which necessitates adequate pain measurement.[55] Despite the huge prevalence rates for cancer pain, prevalence rates for under-treatment are also very high,[56] and to add on to this, reporting rates of cancer pain among palliative care journals are very low.[57] When such an evidence to practice gap is witnessed, it is the responsibility of the clinicians and researchers to establish better and adequate evidence to inform current practice.[58] Each component of the measurement process (i.e., choice of an instrument to measure pain, timing and frequency of measurement, measurement of symptoms accompanying pain or its treatment, and measurement of functional status) is important in developing an accurate and comprehensive assessment of cancer pain.[59]

Multicenter outcome research using BPI will provide palliative care clinicians with answers toward applicability of BPI as an outcome measure for evaluating treatment effectiveness in randomized controlled trials.[60] The need for international collaboration in symptom assessment[61] and evaluation of physical functioning[62] in patients with cancer, however, cannot be overemphasized. Measurement tools are important for assessing not only pain but also analgesia in cancer pain.[63] Such measurements are to be integrated with the findings of comprehensive physical examination[64] in order to enable a wholistic assessment of a person with cancer pain.[65]

CONCLUSION

This paper describes the development of the BPI and the various applications to which the BPI is suited. The BPI can be used as both a quantitative and a qualitative measure for statistical analysis in research. The BPI is a powerful tool and, having demonstrated both reliability and validity across cultures and languages, is being adopted in many countries for clinical pain assessment, epidemiological studies, and in studies of the effectiveness of pain treatment in patients with cancer. Further studies on validation into Indian languages are essential for improving its applicability in Indian palliative care settings.