The main objective of treatment for rheumatoid arthritis (RA) is to control the inflammatory activity and prevent structural damage. It should be begun with a disease-modifying antirheumatic drug (DMARD) as soon as possible.1 Patients in whom treatment with a synthetic DMARD (sDMARD) has failed to achieve the therapeutic objective should be considered candidates to receive biological therapy.1,2 Nevertheless, it has been observed that between 17% and 35% of RA patients do not respond adequately to the first biological therapy.3

The choice of a biological DMARD (bDMARD) is generally based on clinical needs or the preference of the patient, the experience of the prescribing physician and the cost and availability of the drug. It is not usual to pose individualized therapy, although it would be desirable, in order to choose the best agent for each patient. In other diseases, there have been notable advances in individualized therapy through the use of biological markers that predict the response to certain drugs. In RA, this is still not possible.

For these reasons, the purpose of the ACORDAR project is to identify and assess the attributes that would help to decide on the most suitable bDMARD for each case.

The project was carried out between the months of February and September 2015.

To respond to the research question “what are the relevant attributes of bDMARD therapy that make it suitable in the first lines of treatment of RA?”, we started with a strategy involving a systematic literature search, following the PICO method, where the population consisted of adults with RA, the intervention was treatment with a bDMARD, comparisons were carried out and the outcome was the general attributes of these agents, their pharmacology, efficacy, safety, administration and cost.

As primary sources, we reviewed the journal Reumatología Clínica, as well as the web pages of the Spanish Society of Rheumatology and of other prestigious entities (the European Medicines Agency, editorials, etc.), PubMed, EMBASE and the Cochrane Library Plus. We employed combinations of MeSH terms: population, “rheumatoid arthritis”; intervention, “biologic” or “biologic + DMARD” in first lines; outcome, “characteristic”, “efficacy”, “safety”, “side effects”, “pharmacology”, “adherence”, “route-administration” and “cost”.

The terms were looked for in both Castilian and English, starting in 1990 to coincide with the development of the bDMARD. We reviewed the abstracts and/or articles of a total of 548 publications of the 4510 found in the different databases investigated. This was carried out by a single person who is an expert in the matter. This collaborator identified attributes of these agents that, aside from their other values, could help to evaluate or compare the use of biological drugs as applied in RA. In all, 109 attributes were proposed to the committee, of which ultimately 77 were selected, grouped and drawn up: 7 on general aspects, 5 related to pharmacological aspects, 18 relative to efficacy, 31 to safety, 6 to their administration and 10 to cost (Fig. 1).1,4–27

Fig. 1.

Flow chart showing the results of the literature review. EMA, European Medicines Agency; SER, Spanish Society of Rheumatology.

(0.18MB).

In the months of May, June and July, 12 meetings were held in various localities in Spain. To each of them, we invited 7 rheumatologists who were experts in this matter, and explained the project to 83 rheumatologists from all parts of Spain. After each meeting, we distributed the first Delphi findings, in which each attribute was assigned the level of importance it had been designated using a numerical Likert scale with scores ranging from 1 to 9. Each segment had the following correlations: from 1 to 3, minor importance; from 4 to 6, intermediate importance; and from 7 to 9, maximum importance (questionnaire included in the supplementary material). The Delphi process was completed with a second round that was sent in the month of July, when we had received responses from all of the participants.

To determine the consistency of the level of agreement, we simultaneously assessed 2 criteria (the standard deviation [SD], interquartile range [IQR] and coefficient of variation [CV]):

• –

  Criterion based on the median and dispersion, assigning the corresponding importance to the median, if the IQR≤1.00.

• –

  Criterion based on simultaneously establishing the following values: mean and median in the same segment, in addition to the SD and IQR≤1, and the CV≤0.25.

A result was considered to be in agreement, if less than a third of the responses remained outside the segment of importance attributed by the median.

Finally, we distributed the list of attributes with the resulting level of importance together with the criteria for consistency and agreement, for their approval by the participants. We were thus able to obtain the percentage of those who ultimately supported each attribute.

The analysis of the segmentation of the 77 selected attributes was done using the statistical tools provided by Microsoft Excel 2012: mean and SD; median, quartiles and IQR; mode, minimum and maximum values; as well as the CV. The Spearman correlation coefficient was applied to assess the correlation between the 2 rounds and Cronbach's alpha to quantify the internal consistency of the questionnaire.

The 83 participating rheumatologists completed the 2 Delphi rounds, and those who approved had a mean experience in the management of RA of 19.07 years (SD±7.5), and had been prescribing bDMARD therapy for an average of 13.19 years (SD±4.38). In all, 8.43% of the participants were department heads, 10.84% were section heads and the remainder (80.72%) were staff members.

Table 1 summarizes the levels of importance and agreement with each of the attributes included in the study in accordance with the rheumatologists. Only 2 attributes related to the administration of the drug (“a loading dose is not necessary” and “does not contain latex in the vial”) were considered to be of intermediate importance. In the rest, the level of importance was found to be high. The level of agreement with all of the attributes was over 73%, with the exception of those mentioned above, in which the degree of agreement was lower (48.19% and 59.04%, respectively).

The Spearman correlation coefficients between the 2 rounds grouped according to subject and Cronbach's alpha are shown in Table 2.

Fifteen attributes had an agreement of 100% and were considered to be those with the greatest importance (Table 3).

The purpose of this article is to make an effort in support of personalized medicine to define the ideal drug for RA patients who need to receive a bDMARD after the failure of a first bDMARD or of a sDMARD. In the absence of biological markers that can define the most suitable drug for a patient, studies of this type may be helpful in the attempt to achieve greater efficacy in the use of agents with a high impact.

The participants in this project were rheumatologists from different Spanish autonomic communities that accounted for 91.45% of the total population. The number of participating experts was around 10% of the rheumatologists practicing in Spanish hospitals. There was a heterogeneity of 50% and, at the 95% confidence level, the margin of sampling error would be ±10.19%.

The Delphi process was chosen as it is a widely accepted scientific method that facilitates reaching consensus.28 It also enables the participation of a large number of experts through the utilization of telematics.

The Spearman correlation is acceptable and the P value is very small. Thus, it can be considered that there was agreement between the 2 Delphi rounds. On the other hand, the Cronbach alphas were always more than 0.7, demonstrating the internal consistency of the questionnaire.

As far as we know, this is the first endeavor to attempt to establish the ideal attributes to look for in choosing a bDMARD. From the results obtained, we can observe that it appears that, for the group of specialists, it was important that there be an approved indication for first-line treatment for RA, that there be a varied range of mechanisms of action and therapeutic targets, and that the drug did not generate neutralizing antibodies or need an increase in the dose to achieve efficacy. The availability of efficacy data on certain immunological subgroups of patients such as, for example, the presence or absence of anti-cyclic citrullinated peptide antibodies, is also widely supported in the study. Aspects like safety are also highly valued by the group of specialists.

In questions like the administration of the drug is where we find a wider dispersion in the responses. The need for a loading dose of the agent was, specifically, the aspect that was considered the least important in the study.

The strength of the study lies, in our opinion, in the originality of attempting to define, by means of a Delphi study, the ideal characteristics for choosing a bDMARD, which are those displayed in Table 3. Studies should be conducted specifically on each of these attributes to demonstrate that its presence is associated with a greater efficacy of each drug. However, carrying out clinical studies of this type would be costly and difficult, and it does not seem feasible that they come to be performed. Therefore, this type of Delphi study can contribute to helping clinicians in the choice of the best agent for each patient.

In conclusion, we can point out that:

• –

  There was a high level of agreement on the attributes selected from the literature.

• –

  The majority of the experts considered that 75 of the 77 aspects (97.40%) were of high importance, whereas none of them were assigned a low level of importance.

• –

  There are 15 attributes ratified by 100% of the specialists, which could define the ideal bDMARD after the failure of a sDMARD or another bDMARD. All of them refer to general aspects, efficacy and safety.

This project was financed by Bristol-Myers Squibb, with the support of Scientia Salus, who provided the technical secretariat.

The authors declare they have no conflicts of interest.