An increasing number of cancer patients live longer, and palliative care has become an important part of their treatment. Symptoms are often inadequately assessed and managed. A significant challenge in clinical trials is to control for the variability of the samples being studied. To overcome this problem, classification systems have been developed in order to characterise and stratify patients by grouping them according to major common characteristics. The lack of agreed methods for the assessment and classification of cancer pain has been clearly indicated in clinical trials and in clinical practice and may be one possible explanation for the inadequate treatment of cancer pain. This was the background to an international expert meeting arranged in September 2009 in Milan, Italy. The primary aims were to produce recommendations on how to assess and classify cancer pain and to recommend a strategy for the further development, validation and implementation of an international cancer pain classification and assessment system. The recommendations consisted of two basic working proposals, nine specific working proposals and seven recommendations for the further development of a cancer pain classification system. Examples of specific working proposals were to include pain intensity, pain mechanism, breakthrough pain and psychological distress as the core domains in this classification of cancer pain and to measure pain intensity with a 0–10 numerical rating scale with ‘no pain’ and ‘pain as bad as you can imagine’ as anchors. The proposed name for this international standard is Cancer Pain Assessment and Classification System (CPACS).
Statistics from Altmetric.com
The clinical relevance of symptom control in advanced cancer cannot be underestimated. Because an increasing number of patients with cancer are living longer, the palliative care of patients with advanced disease should be considered an integral part of overall disease management, and symptom control should start immediately upon diagnosis and continue through life prolonging therapy into palliative care. The improvement in pain control in many countries in recent decades may be due to the implementation of simple guidelines, new treatment strategies and the use of opioid analgesics. However, this improved clinical awareness has not been accompanied by a similar development in the scientific knowledge of cancer pain. Unfortunately, a consensus-based methodology that standardises symptom assessment and classification is still not routinely applied in clinical trials or clinical practice. Such methodology could help identify patients at risk for poor pain relief and improve the quality of clinical trials. The lack of homogeneous standard methods of assessment and classification for cancer pain in palliative cancer treatment is particularly evident in controlled clinical trials.1,–,6
A significant challenge in all areas of medical science, ranging from cell studies and animal studies to human studies, is to control for the variability of the samples studied. In laboratory studies, reproducible findings are expected across studies investigating the same cell lines, provided that the experimental conditions are controlled. If the environment of the cell is exposed to other than experimental factors, this may introduce bias to the end results.
Issues of replicability and bias are even more significant in human clinical studies, where numerous factors can influence study participants, thereby reducing the homogeneity of a study population. In cancer patients, major individual and group variability may occur due to the heterogeneity of the tumour and of the host. In clinical studies, standardisation of external factors (which is easier to accomplish in the laboratory) is not possible. To help address this issue, classification systems are used to characterise and/or stratify cohorts by grouping them according to major common characteristics that are judged or have proved to play a role in the outcomes. If the classification system is shown to be reliable and valid, its use can support comparison of results across studies, and may also facilitate a better understanding of diverging results.
Cancer pain classification
Classification originates from the grouping or categorisation of living organisms in biology.7 The International Classification of Diseases8 is an example of a widely used system for grouping or classifying different types of disease. Generally speaking, classification of patients can be applied as a prognostic tool to predict the course of a disease or the effect of an intervention, as a predictive tool (for example by selecting specific treatments based upon certain group characteristics) or in order to enhance research (for example by construction of a phenotype such as stage IV non-small cell lung cancer). For a new classification system to become useful it needs to have international acceptance, use conceptually clear language, be feasible and applicable for frequent use in clinical and practice research, and have sufficient psychometric properties (validity/reliability).6
For tumour classification, consensus has been achieved using the TNM system in clinical practice.9 10 More detailed classification is based on examination of the tumour histology and/or a variety of tumour biomarkers. During the last decade, a series of more advanced methodologies have been introduced to supplement clinical classification derived from, for example, techniques developed in molecular biology.11
These more advanced classification systems combine a series of variables and may ultimately improve the specificity of classification, thereby increasing the homogeneity of each class in the system. Better classifier specificity and sensitivity may improve patient treatment outcomes as well as the quality of clinical research. Further, enhanced specificity and sensitivity will facilitate the generalisability of research results through a better understanding of the characteristics of the population(s) under investigation.1 12,–,14 Thus, clinically relevant characteristics should be assessed through appropriate and agreed assessment methods in order to correctly classify patient populations. Such a classification system can then be evaluated in relation to the different clinical perspectives and treatment goals15 and will improve the possibility of selecting the most appropriate treatment for the individual patient based upon clinical guidelines, thereby contributing to evidence based practice. Ultimately, this should improve patient care, survival and symptom control, and reduce costs and the side effects of cancer treatments.
In general, the same challenges that make patient classification difficult at primary diagnosis may also apply to the classification of palliative cancer care. To our knowledge, however, there is no internationally agreed system for classifying a cancer palliative care population in general, or for classifying patients with cancer who experience pain, despite the common plea to develop such classification systems by researchers and scientific associations since the 1970s.16,–,19
Cancer pain assessment
Cancer pain is a complex phenomenon that has both qualitative and quantitative aspects. A standardised system for cancer pain assessment would enhance our ability to describe, qualify/quantify and evaluate pain and thereby monitor changes that occur after treatment.20 This would contribute to a reduction in systematic errors, thereby increasing the validity of the clinical observation and evaluation, which would improve the quality of symptom management and care. To achieve this objective, standard assessment protocols need to be developed. The facts that the development of new pain assessment tools seems to be a never-ending process, and many tool developers do not adhere to accepted methodologies with respect to validity and reliability, are major problems in the field.21 The diversity of measures and protocols in international clinical work and research also contributes to unnecessary variability.
Thorough cancer pain assessment is an essential part of a useful classification system, because it helps to identify the best indicators for the grouping of subjects into classes or categories of the same type, for example by correctly identifying patients with neuropathic pain as emphasised in the guidelines from the European Medicines Agency (EMA).22 23 The goal of a standardised assessment is to reduce variability within groups while increasing variability between groups. The classifiers and the nature of the class or group will depend on the intent and use of the assessment and classification system. For the assessment and classification of palliative cancer patients with pain, three potential uses are particularly relevant: (1) to describe or profile an individual or a group of individuals; (2) to classify patients (that is, to group patients based on preselected characteristics); and (3) to evaluate pain (that is, to assign a value or meaning to the results of assessment and classification). Thus, there are a number of different implications for clinical practice or for research derived from the unsolved matter of the standard assessment and classification of cancer pain.
The European Palliative Care Research Collaborative (EPCRC)24 was established in 2006 and funded by the European Commission 6th Framework Programme. One of the aims of the EPCRC is to develop an international classification system for cancer pain. An iterative approach to this task has been followed, consisting of several steps including systematic literature reviews, acquisition of expert opinions, patient input, empirical testing and validation, and international consensus processes, which has been previously described.24 25
As a part of the EPCRC, an international expert meeting was held in September 2009 in Milan, Italy. The primary aims of the meeting were to produce recommendations on how to assess and classify cancer pain based upon existing knowledge and to recommend a strategy for the implementation, further development and validation of international cancer pain classification and assessment systems. This paper describes the recommendations resulting from the expert consensus and proposes further steps for the implementation and evaluation of the recommendations proposed.
The conference was organised by the EPCRC24 as a joint project with several other institutions and organisations: PRC (the European Palliative Care Research Centre), NTNU (The Norwegian University of Science and Technology, Trondheim),26 MN (the Mario Negri Institute for Pharmacological Research, Milan),27 INT (the Fondazione IRCCS Istituto Nazionale dei Tumori, Milan)28 and EAPC RN (the European Association for Palliative Care Research Network).29 The steering committee (SK, GA, OC, AC) designed the conference format and the criteria used to select invited participants.
Participants were selected on the basis of their research and clinical expertise in cancer pain assessment and classification. They represented a wide range of disciplinary backgrounds, including oncology, neurology, epidemiology, psychology, biostatistics, palliative care, public health, anaesthesiology and nursing. The panel was also representative of international research groups addressing cancer pain assessment and classification such as the EPCRC,24 PROMIS (Patient-Reported Outcomes Measurement Information System),30 IMMPACT (Initiative on Methods, Measurement and Pain Assessment in Clinical Trials)31 and CPOR-SG (Cancer Pain Outcome Research Study Group)32 as well as members of scientific associations involved in oncology and pain such as the IASP (International Association for the Study of Pain),33 ASCO (American Society of Clinical Oncology),34 ESMO (European Society for Medical Oncology),35 AIOM (Associazione Italiana di Oncologia Medica),36 EORTC (European Organisation for the Research and Treatment of Cancer),37 EAPC (European Association for Palliative Care),38 SICP (Società Italiana Cura Palliative),39 and of international regulatory and health authorities: EMA (European Medicines Agency)40 and the WHO (World Health Organization).41
The Milan 2009 consensus meeting started with introductory plenary sessions on cancer pain assessment and classification, the role of health authorities and international collaboratives and the requirements of regulatory clinical trials in cancer pain. On the second day, two parallel workshops (one for pain assessment and one for pain classification) were held with the aim of defining core domains and items, and developing standardised procedures for the assessment and classification of pain in cancer patients for use in clinical trials and clinical practice. The results of the two workshops were then presented and discussed in a final plenary session.
The presentations in the plenary sessions were based upon a series of published systematic reviews5 20 21 42,–,44 and ongoing work based on data collected by the EPCRC and from other sources such as the IMMPACT and the PROMIS initiatives.31 45,–,50
The discussions in the workshops were based on a list of statements first drafted by the steering committee and then modified following discussions among the experts. For each session, two reporters summarised the discussions and then conveyed their summary to the writing committee coordinator. The present position paper was drafted by the writing committee, circulated to and edited by all participants, and unanimously acknowledged by all as the product of the expert consensus.
The workshop content
The systematic literature reviews on classification of pain in cancer patients5 identified six formal pain classification systems. Three were systematically developed and partly evaluated: the IASP classification,51 the Edmonton Classification System for Cancer Pain (ECS-CP)4 6 19 52,–,54 and the Cancer Pain Prognostic Scale (CPPS),55 but none of these have been widely applied in international studies or clinical settings. Two systematic reviews including one expert opinion survey on pain assessment21 56 identified more than 90 different assessment tools used for pain assessment. These reviews in combination with the expert survey identified the five top dimensions for clinical assessment of cancer pain: pain intensity, temporal pattern, treatment effect, pain quality and pain location.
Specific focus was placed on the following themes concerning pain assessment and classification.
▶ Pain outcomes as measured by available pain scales.
The experts' proposals
The working proposals by the expert group are presented under three main headings:
Basic working proposals (BWP)
Specific working proposals (SWP) that are divided into:
cancer pain classification
cancer pain assessment
Further development and follow-up (FD).
Basic working proposals (BWP)
The expert group proposed that an international consensus should be reached on how to classify and assess cancer pain. This does not preclude the use of other classification systems or measures. However, the recommended classification procedures and measures should be considered as minimum requirements that should be included whenever possible to allow for comparability across studies and settings.
The expert group proposed that similar assessment methods that are appropriate for the assessment of cancer pain should be used when possible, both in clinical practice and in clinical research.
Specific working proposals (SWP)
Core domains that should be included in a classification system for cancer pain are: (1) pain intensity, (2) pain mechanism, (3) breakthrough pain and (4) psychological distress. Other domains that were considered as ‘candidate domains’ for a classification system included: (1) pain-related cognitions, (2) addiction, (3) pain location, (4) other relevant symptoms such as sleep, depression and anxiety, and (5) genetic variations.
Domains that are most relevant as outcomes in clinical practice and in research are: (1) pain intensity, (2) pain relief and (3) temporal pattern of pain.
The preferred strategy for assessing cancer pain intensity is to use a 0–10 numerical rating scale with ‘no pain’ and ‘pain as bad as you can imagine’ as anchor words for the extreme values (figure 1).
Average pain intensity over a clinically relevant period of time is the preferred domain for the assessment of pain intensity, with the prompt question being: ‘How would you rate your average pain intensity?’. Two specific time periods are recommended, the selection of which should depend upon the purpose(s) of assessment: (1) average pain intensity during the last 24 hours and (2) average pain intensity during the last week. Worst pain and least pain are also clinically useful and may have a role in pain studies, but average pain intensity should always be measured.
‘Average pain intensity in the last 24 hours’ is recommended as the standard for the classification system for cancer pain (see SWP 1). Other time frames may also be appropriate but ‘average pain intensity in the last 24 hours’ should always be measured.
The recommended source of pain ratings is the patient. If self-assessment is not possible (due to cognitive dysfunction, drowsiness or other factors), proxy ratings may be conducted. If this option is selected, the specific methodology and evidence for the validity of the method chosen should be specified.
If change in pain over time is to be monitored, pain intensity is proposed as the primary outcome. In clinical research this can be accomplished by assessing change in pain intensity over prespecified time intervals of different lengths, depending on the specified outcome. The difference between initial and subsequent assessments should be evaluated.
The same procedure(s) should be used for repeated assessment of pain over time. The recommended formats are either (1) paper-and-pencil instruments or (2) patient interviews (either face-to face or via telephone, as appropriate). Computer-based technologies (Palm device, laptop, cell phone, direct web entries) reflecting the proposals in this publication should be developed and validated.
Further development and follow-up (FD)
For all published cancer pain studies, SWPs should be reported. The goal is that the proposed domains and items for pain classification and assessment be included in all cancer pain studies, according to the recommendations above. The group will work actively to gain wider consensus on the present BWPs and SWPs. The scientific community and journal editors will be approached in order to endorse SWPs to disseminate their use in the design of clinical trials.
Standardised and international consensus based methods for assessment of (1) breakthrough pain, (2) psychological distress and (3) pain mechanism (eg, neuropathic vs nociceptive pain) are urgently needed.
A reduction of ≥50% should be considered a ‘substantial decrease’ and ≥30% ‘a meaningful decrease’ in pain intensity. A reduction of 15–30% should be considered ‘a minor decrease’ in pain intensity. Table 1 provides recommendations on how to adapt this magnitude in pain reduction as a function of different baseline pain intensity for the individual patient.
Average pain ≤3 (on a 0–10 scale) is considered controlled or manageable pain, pain ≥4 but ≤7 is considered moderately controlled or clinical pain (in need of more control if possible), and pain ≥7 is considered inadequately controlled pain.
A standardised cross-culturally valid paper-and-pencil and computer-based 0–10 NRS that is in the public domain (ie, free to all users) (SWP 4) should be developed.
Updates and revisions of the structure of SWPs should be published, as needed.
An international panel based upon the Milan meeting will be established with affiliations to the EAPC RN,29 EFIC (European Federation of Chapters of the International Association for the Study of Pain),68 IASP,33 ESMO,35 ASCO,34 MASCC (Multinational Association of Supportive Care in Cancer),69 UICC9 and other relevant organisations as they are identified. The newly established PRC26 will take on the task of coordinating such activity.
This panel of experts agreed that an international consensus on how to classify and assess cancer pain is needed in order to allow for comparability between clinical and translational studies in cancer pain. These standards or minimum requirements do not necessarily preclude the use of other scales and measures but are proposed as a minimum set of standards that should always be applied.
To our knowledge, the success of the development and implementation of other systems like the TNM/UICC system,9 the WHO pain ladder70 and the DSM classification system for mental disorders71 into clinical practice and research was based, in part, on a combination of formal and informal consensus processes and panel meetings with the results of these being subsequently implemented into various scientific and clinical settings. When applied, these systems were found to be useful in clinical practice as they made a difference in the decision-making processes. They also represented the starting point for continuous validations and refinements.
According to the EPCRC algorithm for assessment and classification, a stepwise and iterative process is an effective approach to achieve standard procedures which facilitate comparisons between empirical studies and effective patient care.25 This or relatively similar approaches to tool development have been followed in the development of several health, symptom or quality of life instruments.6 30 37 53 At this stage, the process has included empirical data collection, literature reviews, expert consensus surveys and input from patient focus groups and surveys. The proposed name of this international standard for the assessment and classification of cancer pain is the Cancer Pain Assessment and Classification System (CPACS). This was pragmatically formulated based in part upon a published stepwise process recommended by the EU collaborations,24 influenced by the panel of experts who met in Milan and with representatives from various stakeholder organisations. This first version of an assessment and classification system – the CPACS – is hereby introduced to the public domain for discussions and input from other stakeholders within the area of palliative cancer care, pain and oncology.
According to the EPCRC strategy for the development of assessment and classification methodology,24 25 the final step in the process has not been completed, that is, testing of the proposed methodology in empirical studies. The lack of prospective large scale testing is a limitation. However, we believe that a system such as the CPACS has sufficient empirical basis to be immediately incorporated into clinical practice and research, because the items and procedures were selected from those that already have considerable empirical support. At present, then, the framework is developed and ready for testing.
It is likely that the CPACS may improve patient care by facilitating communication between clinicians, aid in the evaluation of treatment outcomes and compare the patient cohorts with samples from research reports. Moreover, as scientists and clinicians become experienced with the CPACS, its strengths and weaknesses will become more apparent, providing necessary feedback for its revision over time. More empirical data from the EPCRC-CSA, a recently finalised computer based data collection, described on the EPCRC website,24 and from studies employing the ECS-CP4 53 will hopefully add information about the content of several of the SWPs.
In order to standardise a cancer pain assessment and classification system, the indicators (specific measures) of each domain need to be agreed upon. In the present recommendation, pain intensity should be measured by a 0–10 NRS with standard endpoints (see SWP 4) regardless of setting or language. In order to ensure homogeneity across languages, the EAPC RN,29 in collaboration with the PRC,26 will provide validated translations for the recommended measures over time that will be accessible from a pain item bank on the internet. This will also be the case for the pain characteristics included in the ECS-CP (pain mechanism, incident pain, psychological distress, addictive behaviour and cognitive function). This system was proposed as a template of further development because of its long term and recently international development and validation.6 One limitation of the ECS-CP is that its primary focus is to predict the response to opioid treatment. However, four of the five domains of the ECS-CP have been verified to be of importance in a study using pain intensity as the outcome variable.72
For the classification component of the CPACS, four domains are now considered as core domains: pain intensity, pain mechanism (with or without neuropathic pain), incident pain (breakthrough pain) and psychological distress.73 A next step will be to evaluate the utility of these core domains in prospective international studies and as applied in clinical practice. Feedback should be sought regarding whether additional domains should be included, or if any of the proposed domains are inappropriate or unnecessary. Feedback is also needed to help evaluate the utility of the system applied in clinical practice. In the meantime, the CPACS proposals should initiate a discussion with the editors of the main pain, palliative care and cancer journals to recommend minimum criteria for cancer pain assessment and classification when publishing cancer pain clinical trials.2 3
In our opinion the most important accomplishment of the expert conference was the achievement of consensus regarding the need for an international cancer pain classification and assessment system. Such an approach has recently been supported in a commentary on the classification of cancer breakthrough pain.74 The SWPs should be considered as the initial version of such as system; we anticipate that revisions will be needed, based on clinical experience and ongoing and planned empirical studies, and will build upon work conducted by other groups such as a recent publication by a special interest group of the IASP on neuropathic pain.75 However, the potential need for revision should not limit the cancer clinical and research community from applying the present SWPs in their respective settings. Systematic documentation of clinical experience will in itself help to bring cancer pain into the era of evidence based practice through its inherent standardisation.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
The Consensus Panel on Cancer Pain Assessment and Classification 2009 Steering committee Stein Kaasa, Giovanni Apolone, Oscar Corli, Augusto Caraceni
Experts Henry McQuay, Mark P Jensen, John Farrar, Robin Fainsinger
International research networks Marianne J Hjermstad (European Palliative Care Research Collaborative, EPCRC), Pål Klepstad (EPCRC), Jon Håvard Loge (EPCRC), Nan E Rothrock (PROMIS, Patient-Reported Outcomes Measurement Information System)
Scientific organisations Florian Strasser (ESMO, European Society for Medical Oncology, Palliative and Supportive Care Group), Tarja Heiskanen (IASP, International Association for the Study of Pain), Massimo Costantini (SICP, Italian Association for Palliative Care), Vittorina Zagonel (AIOM, Italian Association of Medical Oncology Palliative Care Study Group), Mogens Groenvold (EORTC, European Organisation for Research and Treatment of Cancer), James Cleary (ASCO, American Society of Medical Oncology)
Regulatory health care authorities Cecilia Sepulveda (WHO), Catherine Deguines (Afssasp, Agence Française de Sécurité Sanitaire des Produits de Santé; appointed by EMA, European Medicines Agency)
Observers and rapporteurs Cinzia Brunelli (Italy), Alessandra Pigni (Italy), Franco De Conno (Italy), Anne Kari Knudsen (Norway), Furio Zucco (Italy), Silvia De Andrea (Italy), Mauro Montanari (Italy), Maria Teresa Greco (Italy).
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.