Challenges related to study design

Neither randomised clinical trials (RCTs) nor registries describe usual clinical care. Most of the medical literature concerning quantitative measures of RA is based on RCTs for which clinical data are being extensively collected. RCTs are regarded as providing the strongest evidence on the efficacy of an intervention and they are regarded as the highest level of evidence-based medicine . However, patients in trials are highly selected, and therefore RCTs do not profile usual clinical care, despite being increasingly conducted in various countries on all continents. Consequently, patients in RCTs may differ substantially from patients seen in standard care . Also RA registers are not generalizable to all patients receiving usual care. Certainly, clinical registries monitor patients outside of clinical trials and better reflect health status of patients treated in clinical care . However, they generally include only selected patients, e.g., patients with early disease or patients who receive certain therapies. Moreover, there is evidence that not all patients eligible for registers are invited and that not all patients invited will agree to participate. For example, illiteracy is a frequent reason to exclude patients from registries. There even may be bias in the centres or rheumatologists that are invited or that agree to participate in such registries. Therefore, these registers do not provide a generalizable picture of patients who receive usual clinical care.

As a consequence, to be really informed about the health status or about benefits and harms of treatments in all patients, information on unselected patients should be available as provided in clinical care. This is the only approach to generate data that can provide insight into which subgroups of patients are at risk for low health-care quality or which countries are at risk of not being able to provide recommended care.

Challenges related to choice and source of measures

Outside of RCTs and registers, most usual clinical rheumatology care continues to be conducted according to physicians’ impressions rather than to quantitative measures. Two main types of quantitative measures for RA outcome assessment could be distinguished: laboratory tests and medical records.

Laboratory tests are usually the only quantitative measures collected in all patients. Outcome measures of many chronic diseases can be assessed effectively from a usual medical record through objective measurements or laboratory tests as is the case of diabetes (glucose and haemoglobin A1C), hyperlipidaemia (cholesterol), hypertension (blood pressure) and osteoporosis (bone density). These data can be used to develop possible goals to improve quality and document the extent to which these goals might or might not be met. Laboratory tests are available in medical records and are the traditional source of the disease-related information in many diseases. However, they often are alone not so informative about patient status in RA, due to their lack of sensitivity, specificity or prognostic value. This is the case, for example, of rheumatoid factor, anti-citrullinated antibodies, C reactive protein or erythrocyte sedimentation rate (ESR), which help the physician when taking decisions, but do not alone guide clinical practice. Overall, laboratory tests in rheumatology are as yet frequently uninformative and contribute little to documentation of long-term outcomes.

On the other hand, the assessment of outcomes through medical records has its own challenges. No single ‘gold standard’ measure (such as blood pressure or serum cholesterol) is available in RA for diagnosis, assessment and monitoring in all patients in clinical trials, clinical research and clinical care. Therefore, pooled indices of multiple measures such as the American College of Rheumatology (ACR) Core Data Set , disease activity score (DAS) , disease activity score with 28-joint assessment (DAS28) and clinical disease activity index (CDAI) have been developed. Calculation of these indices requires values of individual components to be available and also extra work from a health professional, because automatic calculations are not always built as part of general medical record systems. Although frequent monitoring of disease activity guiding adjustments of therapy is recommended, formal assessments are not always performed . All of these disease activity indices require a formal quantitative joint count, which is not performed in most regular care of RA patients , although rheumatologists generally perform a careful qualitative joint evaluation. Tender and swollen joint counts are tedious to perform, have poor reliability and may indicate improvement in patient status while progressive damage is seen . Quantitative joint counts generally are not available in most medical records.

Patient questionnaire scores for physical function, pain, global status and fatigue are increasingly recognised to be of considerable value for assessment and monitoring of patients with RA. Questionnaire scores for physical function provide the most significant prognostic clinical measures for severe outcomes of RA, such as work disability, costs and premature mortality. Indices of only patient data distinguish active from control treatments in clinical trials as effectively as indices that include joint counts, such as the DAS . Notwithstanding the value of questionnaires to assess, monitor and predict patient status in RA at the individual or group level, most regular care does not include patient questionnaires . Therefore, while questionnaires may be the most informative measures to assess quality of care , such data are available in only a handful of usual medical records. When the aim of the data collection is to compare data between institutions and between countries, it is essential that rheumatologists agree upon a core set of outcome measures. The Outcome Measures Rheumatology Clinical trials (OMERACTs) international group has made an effort to propose minimal set of domains and candidate measures for data collection longitudinal studies which could serve as a starting point .

Electronic systems for data acquisition and data sharing

Even if quantitative data concerning laboratory tests, joint counts, clinical status measures, radiographic scores, joint replacements, patient self-reported functional status and other patient reported outcomes, work status, co-morbidities and other clinical status and outcome measures were available in the medical records, retrieving data to an analysable format is generally an impossible task due to the work load required. Therefore, a supplementary programme or data collection system is needed, to facilitate quantitative monitoring of patients in a rheumatology clinic, to analyse the collected data. In a multicentre and multinational effort, such systems should be compatible with a large number of different information and communication technologies (ICTs) systems.

Finally, even if quantitative measures were available and analysable, reporting of observations of single clinics is quite difficult during the era when results of clinical trials dominate medical literature.