PROMIS Pain-Related Measures: An Overview
The National Institutes of Health (NIH) implemented the Patient Reported Outcomes Measurement Information System (PROMIS) more than 10 years ago. NIH created PROMIS to develop and evaluate measures to target important health outcomes across various chronic diseases.1,2 The PROMIS project combines systematic state-of-the-science instrument development with modern measurement theory to develop health outcome assessments that provide flexibility, precision, and validity for clinical trials and clinical practice settings. PROMIS measures are based on “item banks,” sets of items that represent symptom or functional domains.
PROMIS initially focused on item banks to assess physical functioning, fatigue, social activities, depression, anxiety, and other domains, including pain-related domains. The pain-related domains include item banks for pain interference and pain behavior,3,4 and measures of pain intensity and pain quality.5,6 The PROMIS pain-related item banks and measures were developed based on systematic qualitative and psychometric research methods.3,4,7,8 Figure 1 provides a summary of the PROMIS conceptual framework.
Item-response parameter estimates enable the development of static short-forms and computerized adaptive tests (CATs).8 In addition, researchers can review the content of these item banks and relevant item location and threshold parameters, and use this information in the intelligent design of study-specific or disease-targeted measures that can be used to assess outcomes. Each CAT is unique, in that it represents an assessment approach that tailors the measurement to individual patients, allowing for very precise assessment with few items (4 to 6 items for many applications). Items are selected for individual patients based on their answers to previous items. Thus, individuals receive different sets of items depending on their locations along the health domain continuum. For example, when assessing physical functioning, if a patient responds that she cannot walk up one flight of stairs, then she is not asked questions about walking up two flights of stairs or about the performance of more strenuous activities.
This report provides an overview for clinicians of the PROMIS pain-related item banks and measures, and briefly summarizes some of the psychometric results for these measures.
Clearly, any assessment package for pain-related outcomes needs to include measures of pain intensity or severity. The PROMIS measure includes the typical 0-to-10 numeric rating scale (NRS) for pain intensity. The pain intensity NRS can be used as measure of pain severity, or can be used as part of the PROMIS Global Scale.9 In addition, there is a small set of pain severity items that cover average pain (1 item), worst pain (1 item), and pain right now
(1 item).10 These 3 items cover a broad range of the pain intensity continuum that is comparable to the pain intensity NRS (Figure 2).
Pain interference refers to the degree to which pain limits or interferes with an individual’s physical, mental, and social activities. PROMIS network investigators developed a pain interference item bank based on reviews of the pain literature and existing published measures, clinical expert input, and qualitative research with patients who reported having chronic pain conditions.4 The resulting item bank contains 41 items covering the negative impact of pain on functioning. Example questions are summarized in Table 1. A 7-day recall is used, and 3 sets of response options are included (never to always; not at all to very much; never to every few hours). The PROMIS investigators have developed 4-item, 6-item and 8-item short-form pain interference scales.
Amtmann et al showed that the PROMIS pain interference item bank provides precise measurement across a broad range of the pain interference continuum.4 Scores are correlated with those of other pain-related measures, and vary significantly by pain intensity levels, number of chronic conditions, disabling conditions, and general health status. For example, pain interference scores increased with lower levels of reported health status (P<0.0001; Figure 3). Pain interference scores also varied significantly by pain intensity categories (P<0.0001; Figure 4).
The PROMIS pain interference item bank can be used to generate various static scales (short-forms), as well as for computerized adaptive tests. Research demonstrates that with CAT, 4 to 6 items are sufficient to assess pain interference, given the tailoring of the specific items to the functioning of the individual respondent. In addition, the PROMIS pain interference short-form and CAT scores have comparable responsiveness compared with the Brief Pain Inventory interference scale in patients with low back pain.11
Pain behaviors communicate to others that a person is experiencing pain.12,13 Pain behaviors can include verbal complaints of pain and suffering, non-language sounds, facial expressions, body posturing and gesturing, and limitations in activities. There is growing recognition that pain behavior is a key outcome of persons suffering from chronic pain.14 Direct observation provides the most objective approach to assessing pain behavior, however, self-report provides an alternative method.12 The PROMIS investigators developed a comprehensive pain behavior item bank based on self-reports that can reliably assesses the full array of pain behaviors for clinical studies.
The original PROMIS pain behavior item bank consists of 39 items covering movement (eg, move slowly, stiffness), affect (eg, irritable, angry), social interactions (eg, ask for help, withdraw), and facial/verbal expressions (eg, groan, grimace) (see Table 2, page 36). The initial pain behavior item bank was developed based on the published literature, review of existing observer-rated and self-report pain behavior instruments, clinician review, and qualitative research with patients experiencing various kinds of pain. The items cover a wide range of pain behaviors. The response scales focus on how frequently patients engage in each pain behavior using a 6-point Likert-type scale, ranging from 1 (had no pain) to 6 (always), with responses reflecting the 7-day recall period. The PROMIS investigators also have developed a 7-item short-form pain behavior scale.
Empirical evidence has supported the validity of PROMIS pain behavior scores.3 For example, pain behavior scores vary significantly by level of pain intensity, with increasing higher scores associated with increased pain intensity (see Figure 5, page 36).