Polymyositis (PM) and dermatomyositis (DM) are idiopathic inflammatory myopathies with a presumed autoimmune pathogenesis (1, 2). The generally accepted first-choice therapy is high-dose prednisone (2). When this treatment fails because of insufficient improvement, frequent relapses during tapering of the dosage, or unacceptable side effects, a second-line therapy is started. Many different treatments are being used for second-line treatment in PM and DM, and there is no consensus on which is the best choice (2).
Efficacy of therapies can only be assessed appropriately with randomized controlled trials (RCTs). When RCTs are not available, other types of controlled studies can be useful, although they provide a lower level of evidence because of their potential for bias. If there are not a sufficient number of controlled studies on which to base a treatment decision, as is often the case in rare diseases, the clinician's decision must be based on reports of uncontrolled observations and the opinion of experts (3).
Criteria for good-quality RCTs have been well established (4) and are applied widely. Criteria for a clear and adequate description of single cases and case series have also been formulated (5), but the methodologic quality of this type of evidence in published reports has never been evaluated (6). This hampers the appreciation of their validity and practical value. Controlled studies in PM and DM are extremely rare, but in contrast, there is a large body of reported single cases and uncontrolled case series. We undertook a study to assess the quality of reported descriptions of uncontrolled observations on the second-line treatment of PM and DM.
Single case reports and case series reports of second-line therapy in adult DM, PM, or myositis associated with a connective tissue disease or malignancy were found by searching Medline and Embase for articles published in English, French, or, German between 1966 and the end of 2001 using the following search terms: “dermatomyositis,” “polymyositis,” “inflammatory myopathy,” “treatment,” and “therapy.” We hand-searched all reference lists of identified publications and of relevant review articles for additional publications. Excluded were controlled studies, abstracts, and publications on inclusion body myositis, juvenile DM, and aspects of myositis other than weakness (e.g., pulmonary involvement).
Two investigators systematically and independently examined all eligible reports. A third investigator helped resolve differences through discussion. All articles were reviewed for a clear and adequate case description, which would allow a clinician to recognize his or her own patient in the patient described in the study, copy the described treatment, and get a fair impression of the treatment effects (5). On the basis of these requirements, we arbitrarily predesigned the following 10 criteria with which we would assess the reports: 1) sex and age; 2) credible diagnosis; 3) clear description of disease duration; 4) clear description of previous therapies; 5) clear description of severity of the disease at initiation of second-line treatment; 6) reason for failure of previous treatments; 7) dosage, mode of administration, and duration of the second-line therapy; 8) clear description of effect of the therapy on muscle strength or function; 9) description of side effects; and 10) followup at least 6 months (because of the chronicity of these diseases).
For a credible diagnosis of DM, reference to the criteria of Bohan and Peters sufficed (1), but a mere reference to these diagnostic criteria was accepted for PM only if it was clear from the text that the symptoms had evolved over weeks to months. This criterion was chosen as a feasible attempt to rule out inclusion body myositis and muscular dystrophies in the studied publications (obviously, nowadays specific investigations of muscle biopsy material are considered to be required for an accurate diagnosis of PM). By applying this criterion, we may have judged the diagnosis as not credible in some patients with slowly evolving PM.
Severity of disease before and after second-line treatment was preferably described in terms of MRC scores, dynamometry, or accepted scales of disability or handicap, but we also accepted any ad hoc scales and mere descriptions of functional abilities as long as they gave a fair impression of muscle strength or function. A clear description of the reason for failure of previous treatments was considered mandatory because the therapeutic prospects are probably quite different for a patient who had improved on prednisone, but did not tolerate it, than for a patient who did not benefit from previous therapies.
We analyzed articles published before and after 1985 (when evidence-based medicine became into general use) separately. We made a distinction between single case reports (in which the patient or patients are described individually) and case series reports (in which patients are described as a group).
We identified 148 publications, of which 92 were eligible for the study (references available from the authors). These 92 publications described a total of 915 patients (median 2 per publication; 75th percentile = 7). Ninety-two patients were described more than once in different articles. There were 74 single case reports and 18 case series reports. Most (77%) of the studies were retrospective, 14% were prospective, and the design was equivocal in 9%.
The number of fulfilled criteria by percentage of patients studied is shown in Figure 1. All 10 criteria were fulfilled in 9 publications (10%), describing 2.6% of all patients. All 9 articles were single-case reports (references available from the authors). The number of publications fulfilling each criterion is shown in Table 1. Four criteria (previous treatment, reason for initiating second-line treatment, and disease severity before and after treatment) were met in <50% of described patients. For example, treatment results were often indicated using such phrases as “remarkable improvement of strength,” “better than ever,” “doing well,” and “muscle condition satisfactory.” Of the 10 criteria, 9 were fulfilled in single case reports more often than in case series reports. Reports published after 1985 were not of better quality than those published earlier (Figure 1).
|Criteria||Single case reports† n = 74||Case series reports† n = 18||Patients n = 915|
|1. Sex and age||73 (99)||15 (83)||692 (76)|
|2. Credible diagnosis||58 (78)||13 (72)||473 (52)|
|3. Duration of disease prior to 2nd-line treatment||68 (92)||10 (56)||598 (65)|
|4. Previous treatment||36 (49)||4 (22)||311 (34)|
|5. Disease severity prior to 2nd-line treatment||43 (58)||3 (17)||265 (29)|
|6. Reason for starting 2nd-line treatment||66 (89)||7 (39)||357 (39)|
|7. Dose and scheme 2nd-line treatment||74 (100)||15 (83)||716 (78)|
|8. Disease severity after 2nd treatment||51 (69)||6 (33)||434 (47)|
|9. Side effects||51 (69)||15 (83)||738 (81)|
|10. Follow up at least 6 months||56 (76)||10 (56)||593 (65)|
In conclusion, we found the methodology of patient descriptions (the evidence) unsatisfactory in single case reports and case series reports of second-line treatments in PM and DM. Therefore, the added value of these reports for making treatment decisions in clinical practice, or for identifying new treatments of interest, is dubious. Also, any attempts at doing a systematic review of treatment results reported in these articles are unrealistic (6). It is noted that readers have a better chance of finding relevant and complete information in single case reports than in reports of large series of patients (Table 1). Our results further show that introduction of the principles of evidence-based medicine in recent years has not lead to more adequate data presentation in these types of studies. We conclude that reports of uncontrolled observations can improve considerably if criteria for good quality are taken into account.