Applying reflective multicriteria decision analysis to understand the value of therapeutic alternatives in the management of gestational and peripartum anaemia in Spain

A systematic literature review (SLR) was conducted with two purposes, the first one was to define the specific subcriteria within the framework for the indication, using the EVIDEM domain-criteria framework, further subcriteria were identified according to the SLR identifying all the relevant characteristics to be considered in the decision-making process such as efficacy, safety, patient preferences, economic evaluations, political aspects, and others. The second objective of the SLR was to create the evidence matrix for each criterion and subcriterion. A medical database targeted search following the PICO approach (population, intervention, comparison, and outcomes) [29] was carried out, including a time horizon (2010-present) and type of studies (randomized clinical trials, systematic reviews, meta-analysis, observational studies) limitation; databases used were PubMed and Cochrane Library (Additional Figure 1, Additional Tables 1 and 2). Additional documents and evidences were searched in the grey literature, for example clinical practice guidelines, clinical trials [30], the European Medicines Agency [31], Agencia Española de Medicamentos y Productos Sanitarios [32], Anemia Working Group España [33], and Network for the Advancement of Patient Blood Management, Haemostasis and Thrombosis [34].

The EVIDEM framework was used to develop the evidence matrices, the framework includes the core model (risk–benefit, and modulators domains), that is further itemised in general and specific criteria (Fig. 1a), and the contextual criteria (Fig. 1b). Additional contextual criteria were included in the EVIDEM framework according to the report published by OSTEBA, the Agency of evaluation of Health Technologies of the Basque Country [35].

An expert panel composed of twelve people was created to collect insights from the different perspectives in the management of anaemia in gynaecology. There were clinical experts including the following profiles which are involved within the context of the gynaecological indication of interest gynaecology/obstetrics, midwifery, haematology, anaesthesiology, hospital pharmacy, regional decision makers, patient representatives, and patients. Due to national travelling restrictions, lockdowns, and public health situation all the exercises and meetings were carried out in a virtual approach developing online tools that fulfilled all the needs.

In a first basis all the experts received documents with the methodology of the project. The first phase of the project was selecting the specific subcriteria for the indication, that were validated as clinically accurate by the principal investigator of the project and to ensure their understandability, definitions were reviewed and validated by the patient involved. These criteria and subcriteria were subjected to Hierarchical Point Allocation method by the experts in a virtual workshop in order to obtain those aspects that were more relevant in the Spanish context when making a decision; this was performed by distributing 100 weighting points across every set of domains, criteria, and subcriteria; this rating method was preferred over the direct rating method also proposed by EVIDEM, as it forces the experts to weight each criterion/subcriterion while keeping in mind the rest of the criteria/subcriteria within the same level; this approach forces the prioritization of criteria/subcriteria and not allowing the participants to state all of them as very relevant or not relevant at all. After that, all weights were normalized. In the second phase of the project a description of the criteria/subcriteria based on the available evidence was presented to the participants and they reflected how the evidence on the criterion would impact their decision by using a direct rating scale method, in which they by assigned a score from + 5 (“Much in favour of option a”) to -5 (“Much in favour of option b”). For the contextual criteria, participants selected the type of the impact in their decisions of these criteria:”In favour of FS”, neutral, or”In favour of FCM”.

Two online tools were developed for the two exercises with the purpose of simplifying the participation and collection of data.

A final virtual workshop was carried out to share and reflect individual preferences on the value of compared alternatives, to discuss final results and to include any additional individual comments, this phase was of great importance to the project as it is the part in which the different participants’ perspectives are shared in order to reach an agreement and establish meeting points to improve patients’ management.

After the second exercise, direct rating scale, the mean scores were standardized in a scale from -1 to + 1 (Sx). Lastly, the value of intervention (Vi) of each criterion or subcriterion was calculated as the product of their normalised weight (from the first exercise, Wx) and the standardised score of the second exercise (Sx). These results were presented as mean ± standard deviation (SD) and represented in graphs. The final modulated relative benefit-risk balance (MRBRB) was calculated as the sum of the Vi of all subcriteria (x), in an additive linear model.

The contextual criteria were represented in a qualitative scale with the three options: positive, neutral, or negative impact.

Using hierarchical point allocation method to the core model, weights resulted in 62% (± 8%) risk–benefit criteria, and 0.38 (± 0.08) for the modulators (Fig. 2). Within risk–benefit domain, compared efficacy-effectiveness criterion received the greatest weight, 0.26 (± 0.10), with values ranging 0.06–0.48 across the different professional profile of the participants, corresponding subcriteria ranged between 0.03 (± 0.02) (Response duration) and 0.08 (± 0.06) (Haemoglobin increase). Compared safety/tolerability was weighted as 0.19 (± 0.06) of the risk–benefit, and the most important subcriterion was Non-serious and non-fatal adverse events (0.09 (± 0.05)), followed by Serious and non-fatal adverse events (0.05 (± 0.02)). The last criterion within the Risk–Benefit domain, Patient’s preferences, received a weight of 0.17 (± 0.11), being the most important Preferences on daily frequency of administration, 0.05 (± 0.05).

As additional contextual criteria were selected according to Spanish agency recommendations, these criteria were also subjected to hierarchical point allocation method to assess their appropriateness. Those criteria receiving higher weights were Feasibility (0.21 ± 0.12), General population priorities and access (0.20 ± 0.10), Ethical (0.20 ± 0.18), and Mandate and scope of the healthcare system (0.19 ± 0.09). The remaining criteria received very low weights and were excluded for being non-relevant: Common goals and specific interests (0.06 ± 0.06), Environmental impact (0.09 ± 0.07), and Political/historical/cultural context (0.04 ± 0.05).

Within de Risk–Benefit domain, all the profiles were aligned with valuations around 60% (minimum 50% for the healthcare decision-maker profile, and maximum 70% for the hospital pharmacists), in general the most important criterion was Compared Efficacy/Effectiveness (ranging 15%-36%), while anaesthesiologists and healthcare system decision-makers considered equally important Compared Safety/Tolerability (27% and 15% each, respectively). Additionally, for the healthcare system decision-makers and patients/patients’ representatives the most important criterion within the Risk–Benefit domain was Patient preferences (20% for the decision-makers, and 33.4% for the patients and patients’ representatives). Those giving more importance to Compared Safety/Tolerability were anaesthesiologists and hospital pharmacists (24%), while patients/patients representatives were valuating it the least (12.4%). Patients’ preferences were particularly important for patients/patients’ representatives, followed by haematologists (24%), those giving less importance to this criterion were anaesthesiologists (6%).

Modulators were as important as the Risk–Benefit domain for the decision-makers (50%-50%), but less important for the other profiles (30%-40%). The Type of benefit of the intervention was valuated as the most important criterion or as important as the other criteria by all the participants; preventive benefit was more important than therapeutic benefit for some profiles; anaesthesiology, haematology, decision-makers, and patients/patients’ representatives; while hospital pharmacy, and gynaecology/obstetrics considered the therapeutic benefit as more relevant; midwifery considered both criteria as equally relevant (10% each).

Regarding the economic consequences of the intervention, decision-makers were the group valuating the most this criterion (15%) and its subcriteria, followed by anaesthesiology (10%), and haematology (10%). These groups were also giving more importance to the Knowledge about intervention (20% decision-makers, and 10% anaesthesiology and haematology), while the rest of the participants did not consider this criterion as truly relevant (3–8.5%).

Participants chose between FCM or FS in relation to each criterion and subcriterion basing the decision on available evidence (Table 1). All participants were much in favour of FCM regarding Compared efficacy/effectiveness (3.52 ± 1.08) and most of its subcriteria, whilst Compared safety/Tolerability was slightly in favour of FCM (1.16 ± 2.62), but some participants chose FS; when considering Patients Preferences (1.95 ± 2.64); Daily frequency of administration (3.36 ± 2.06) and Duration of treatment (3.88 ± 2.60) seemed to be driving the decision towards FCM, while Preferences on administration location (0.48 ± 3.53) and administration procedure (-0.48 ± 3.13) were more neutral or in favour of FS.

When deciding the intervention considering the modulator criteria and subcriteria, all of them but one (Administration cost, healthcare system; -0.40 ± 3.52) resulted in favour of FCM, being the most relevant criteria Therapeutic benefit (3.67 ± 1.45) and Unmet needs (3.64 ± 1.60).

The global MRBRB favoured FCM (0.45 ± 0.1913) in a scale from -1 to + 1 (Fig. 3). Almost all the criteria favoured FCM as well, with values ranging between 0.183 ± 0.07 for Compared efficacy/effectiveness and 0.002 ± 0.01 for Time from treatment commercialization. Only one criterion favoured FS: Direct medical costs (-0.003 ± 0.03).

Regarding the qualitative, contextual criteria, participants scored their preference in terms of in favour of FCM, indifferent, or in favour of FS (without quantitative valuation). In general, evidence presented in most criteria resulted in a positive valuation of FCM (54.29%) (Fig. 4). However, some criteria raised some concern such as: General population priorities and access in which 30% of participants were in favour of FS and 10% were indifferent; and Healthcare system capacity to assume the technology cost, with 40% of participants in favour of FS and 10% not showing any preference.

From the different perspectives and taking into account the evidence, the decision-making process between FCM and FS was very diverse (Additional Table 3).

Six groups chose FCM regarding risk–benefit data, being the lowest gynaecology/obstetrics (0.11) and the highest anaesthesiology (0.48); while decision-makers chose FS (-0.15). All of them were in favour of FCM according to its Compared Efficacy/Effectiveness (0.07 for decision-makers; 0.12 for gynaecology/obstetrics; 0.17 for patients/patients’ representatives; 0.18 for midwifery; 0.24 for haematology; 0.25 for anaesthesiology, and 0.26 for hospital pharmacy); being the decision mainly driven by the Hb increase. Anaesthesiology was much in favour of FCM regarding Compared Safety/Tolerability (0.27), other profiles chose discretely FCM (midwifery, 0.08; haematology, 0.06; and patients/patients’ representatives, 0.04); hospital pharmacy did not choose any alternative (0.0); and gynaecology/obstetrics and decision-makers were in favour of FS, -0.05 and -0.08, respectively.

Patients’ preferences favoured the decision in favour of FCM by all the participants but decision-makers (-0.14 in favour of FS); with values ranging from 0.04 for hospital pharmacists and gynaecology/obstetrics, and 0.18 for haematology. The only subcriteria generating concerns among those participants in favour of FCM were Preferences on administration procedure.

As for Modulators domain, all participants were in favour of FCM but decision-makers, being especially relevant for this group the Knowledge about intervention (-0.10) and the economic consequences of the intervention (-0.09). The need for intervention was truly relevant for midwifery, anaesthesiology, and haematology; as well as the Type of benefit of the intervention that was also very relevant for these groups and patients/patients’ representatives. Economic consequences of the intervention made three of the seven groups chose FS (hospital pharmacy, gynaecology/obstetrics, and decision-makers, while the rest of the participants were slightly in favour of FCM (0.03–0.09). Regarding Knowledge about intervention, decision-makers were much in favour of FS (-0.10) while anaesthesiology were much in favour of FCM (0.10); the rest of the participants chose FCM with low values (0.01–0.06) or made no choice (0.00 gynaecology/obstetrics).