Introduction
Spinal muscular atrophy (SMA) is an autosomal recessive neuromuscular disease caused by a homozygous deletion or mutation of the survival motor neuron 1 (
SMN1) gene on chromosome 5q13, leading to progressive muscle weakness and atrophy [
1‐
3]. It is one of the most frequent monogenic neurodegenerative diseases with an estimated incidence ranging from 1/6000 to 1/10,000 newborns [
1,
4].
In the last decade, therapeutic management of SMA has changed with the discovery and approval of antisense oligonucleotides, gene therapy, and small molecules achieving key improvements in motor and respiratory outcomes and increased life expectancy [
5‐
7]. Current therapies are intended to slow or stop the progression of the disease by either modifying the splicing of the
SMN2 gene, replacing the
SMN1 gene, or upregulating muscle growth [
5‐
7]. Earlier treatment after symptom onset enables a greater response than treatment after a longer disease duration according to clinical trial results [
5]. The survival of treated patients with SMA has generated new clinical phenotypes and long-term outcomes are unknown [
5,
8,
9]. In addition, we have no clinical trials that directly compare the efficacy and safety of different treatments [
5]. In this context, treatment selection is becoming more complicated and requires considering different factors, including patient age, disease duration, comorbidities, route of administration, and patient preferences [
5,
8‐
10].
Uncertainty is one of the most important contributing factors affecting decisions in medical care [
11]. Many decisions are made with limited information from clinical trials or observational studies that may not apply to particular patients. Decisions based on erroneous assessments may result in incorrect patient and family expectations, and potentially suboptimal advice, treatment, and prognosis [
11]. Despite these advances in the treatment landscape of SMA, there is limited information regarding preferences of pediatric neurologists that lead to specific therapeutic choices under uncertainty. Together, those were the concepts for design of DECISIONS-SMA study (Therapeutic Decision-Making under uncertainty in the management of SMA). DECISIONS-SMA assessed therapeutic expectations, preferences, and choices of pediatric neurologists when facing simulated case scenarios with different SMA subtypes and clinical status.
Methods
Study Design and Participants
DECISIONS-SMA is a non-interventional, cross-sectional, web-based pilot study in collaboration with the Spanish Society of Pediatric Neurology (SENEP). The selection criteria included (i) pediatric neurologists (with or without specialization in neuromuscular disorders) and (ii) active practice either in an academic or non-academic setting. Participants were recruited by receiving an invitation by SENEP from June 3 to November 2, 2021. The study was approved by the ethics committee of Hospital Clínico San Carlos, Madrid, Spain (reference 21/313-E), and performed in accordance with the 1964 Helsinki Declaration and its later amendments. Participants provided written informed consent. Further details of the study protocol were described in a previous publication [
12].
Study Objectives
The primary objective of DECISIONS-SMA was to assess preferences and expectations of pediatric neurologists regarding treatment choices for SMA. We also assessed treatment initiation and escalation when warranted by contemporary recommendations (see definitions in the next section). Participants were exposed to 11 simulated case scenarios or case vignettes (Supplementary Material; answers in bold were considered suboptimal treatment decisions). Case scenarios were designed by a research team led by GS and IM and based on the most common situations experienced by pediatric neurologists in clinical practice and after reviewing SMA clinical trials and patients’ and caregivers’ preferences from the literature [
5,
10,
13‐
16].
Outcome Measures and Definitions
The primary outcome of interest was therapeutic inertia (TI), defined as the absence of treatment initiation or intensification when treatment goals are unmet [
17]. As in our previous research, we created a TI score to represent the number of case scenarios where treatment initiation or escalation was warranted over the 11 presented case scenarios [
18,
19]. This score may range from 0 to 11, where higher values represent a higher degree of TI. Participants with a TI score ≥ 1 (i.e., therapeutic inertia in at least one case scenario) were considered to calculate the prevalence of therapeutic inertia. An appropriate treatment switch was defined by the number of case scenarios where the initial treatment was changed given the clinical evidence provided of disease progression (i.e., a decrease in baseline scale score greater than the scale’s minimal clinically important difference) over the total of five case scenarios (nos. 2, 3, 7, 8, and 9; Supplementary Material) assessing this strategy according to contemporary treatment recommendations [
9,
20‐
24].
Given the complexity of analyzing treatment effects observed in randomized clinical trials in SMA, we also assessed participants’ expectation of treatment benefit using four simulated case scenarios (e.g., a 5-month-old patient with SMA type 1, a 1-year-old patient with SMA type 2, a 16-year-old patient with advanced SMA type 2 and delayed diagnosis, and a 15-year-old stable patient with SMA type 2 diagnosed at 3 years of age; these correspond to cases 1, 6, 10, and 11, respectively—Supplementary Material). Participants were asked: “On a scale from 1% to 100%, what are your expectations of improvement in 2 years for this patient with any of the treatments currently available?” We reported participants’ expectation of improvement for each case scenario and a global metric by combining all four cases.
Clinical stabilization is a success in the context of a progressive disease like SMA. However, when creating simulated case scenarios we wanted to pose the questions in a broader and more open way by including improvement. According to behavioral economics, participants made decisions on the basis of their perception of benefits (instead of clinically defined or proven motor or respiratory metrics).
We applied concepts from behavioral economics that were previously associated with suboptimal therapeutic decisions or TI [
18,
19,
25]. Physicians’ tolerance to uncertainty was assessed using the standardized physician’s reaction to an uncertainty test [
26,
27]. Participants rated their level of agreement with each question from 0 (strongly disagree) to 5 (strongly agree), and a total score was calculated [
27]. Higher values indicate lower tolerance to uncertainty [
18]. In the present study, a score of 12 or higher indicated low tolerance to uncertainty. Ambiguity aversion is defined as dislike for events with unknown probability over events with known probability. As in our previous studies, participants were asked to choose between a visual option represented by bars with known 50/50 probability of winning €400 (blue bar) or €0 (red bar) and an option with an unknown probability of the same outcomes in one of the following degrees of uncertainty representing a 10%, 30%, 50%, 70%, and 90% of probability of winning, illustrated by a gray area covering in the bar—Supplementary Material, concepts from behavioral economics) [
18]. There was no cutoff point. The degree of aversion to ambiguity was defined as the proportion of times participants chose the 50/50 option over the ambiguous option combining all five uncertainty options.
Statistical Analysis
We used descriptive statistics to report frequency distributions of qualitative variables, measures of central tendency, and dispersion of quantitative variables using non-parametric tests, and 95% confidence intervals. Wilcoxon’s sign test was used to compare participant’s expectations with treatment. Factors associated with TI, treatment initiation, and intensification (switches) were analyzed using linear regression analysis with backward selection. We included the following explanatory variables: age, gender, specialization in neuromuscular disorders, years of experience as a pediatric neurologist and also seeing patients with SMA, number of patients seen per week, practice setting (academic vs. non-academic), proportion of time devoted to clinical care, co-author of a peer-reviewed publication within the last 3 years (yes/no), aversion to ambiguity, physicians’ reaction to uncertainty.
We also assessed participants’ expectations of improvement with treatment for different SMA scenarios (only for cases 1, 6, 10, and 11). Participants could select the expectation of improvement with treatment ranging from 0 to 100%. Results are presented as mean percentage of expected improvement (and standard deviation, SD), and illustrated by box plots. All tests were two-tailed, and p values less than 0.05 were considered significant. Unavailable data was described as missing, without any imputation/allocation. The statistical analysis will be performed using Stata Statistical Software 17.0 (StataCorp., College Station, TX, USA) and considering a significant level of 0.05.
Discussion
Recent discoveries targeting different disease mechanisms have altered the treatment landscape and management of SMA [
5,
6]. Despite these advances, we have limited information on how pediatric neurologists make treatment selection [
10,
15]. Diagnosis and follow-up of patients with SMA in Spain can be carried out in all hospital-based neuromuscular units in the country. However, the administration of SMA agents is centralized in referral units.
DECISIONS-SMA specifically assessed participants’ therapeutic expectations and preferences, as well as their choices regarding treatment initiation, intensification (e.g., switches), and TI. We found a high prevalence of TI among all participants in at least one simulated case scenario, affecting nearly 4 out of 10 (38.2%) therapeutic decisions. On average, pediatric neurologists had a modest (43%) overall expectation for improvement, which was significantly higher for SMA with a recent diagnosis compared to those with a longer disease course (59% vs. 20.5%). Lower aversion to ambiguity and lower expectation of treatment response were the two factors associated with higher TI after adjustment. Similar results were observed for treatment switches, whereas older age, lower years of experience as pediatric neurologist, lower aversion to ambiguity, and lower expectation with treatment were associated with inertia in treatment initiation.
The therapeutic landscape of SMA has become more complex with the approval of new therapies unveiling physicians’ inherent uncertainties when starting a new agent [
5,
10,
24]. Most studies have been evaluating therapeutic decision-making in SMA from the perspective of patients and their caregivers and family members [
13‐
16,
28‐
31]. Briefly, these studies highlighted the importance of maintaining functional status and the need to develop sensitive scales able to detect small changes in a patient-centered spectrum of dimensions beyond motor function. McGraw et al. conducted a qualitative study among patients, parents, and clinicians in the USA to understand what they considered as a meaningful change in SMA type 2 and 3 [
15]. Clinicians cited the importance of small motor improvements, usually undetected by the scales used in clinical trials and clinical practice, but which have an impact on activities of daily living and autonomy of patients and families. However, the perspective of the treating clinician has been little studied, especially in their behavioral aspects that could influence decision-making. The relationship between TI and aversion to ambiguity is intriguing. Participants who were more prone to choose options with unknown probability were more likely to select treatment options that were not based on recommended guidelines (higher TI scores). This finding may reflect that those participants may feel more comfortable in dealing with ambiguity resulting in making choices outside best practice recommendations. Previous studies showed that the preference for ambiguity was attributed to overconfidence, which could be a potential explanation for our findings [
32,
33].
Age, baseline functional status, SMA type, and the number of copies of
SMN2 are the most important factors predicting response to disease-modifying therapies [
6,
34‐
36]. There is convincing evidence that early initiation of treatment, ideally in the presymptomatic stage of the disease, is associated with markedly better clinical outcomes compared to delayed treatment initiation [
5]. A recent consensus panel including family members of patients with SMA also supports these recommendations [
37]. Our study findings are clearly aligned with both consensus statements as reflected by the results on treatment expectations for different SMA scenarios. Furthermore, families of patients with SMA also highlighted the risk that decisions be influenced by subjective and individual preferences of pediatric neurologists rather than being primarily based on evidence [
37]. This is a well-known concept in the medical literature [
38,
39]. We opted to have a better understanding of participants’ preferences with treatment versus no treatment and the treatment modality in order to avoid any bias related to a specific agent. By providing pediatric neurologists’ preferences and treatment choices, our study explicitly illustrates some real concerns raised in the family member-based consensus statement [
37]. For example, TI was observed in nearly all participants affecting on average over one-third of therapeutic decisions.
Our study has some limitations that deserve mention. First, our sample size is relatively small, which may affect outcome estimates, as well as the association of specific variables with the outcomes of interest (type II error due to low power to detect differences). In addition, observations could be clustered per neurologist and possibly also per simulated case. However, it was not possible to perform a mixed-effects model analysis of all original observations with a random intercept per neurologist because of the small sample size. Second, although other multicenter studies showed that TI is a global phenomenon, our results should be seen as a pilot study and are not necessarily generalizable to therapeutic decisions in SMA in other countries. Third, we have not covered the whole case-mix of SMA to avoid overwhelming participants with additional layers of uncertainty. Instead, therapeutic decisions were assessed by simulated case scenarios as representative of the most common situations faced by pediatric neurologists in routine clinical practice. Fourth, although optimal therapeutic choices for case scenarios were based on the recommended guidelines, some participants may have selected options prioritizing the principle of safety over treatment efficacy (primum non nocere) in a context of uncertainty and little comparative drug evidence. Finally, we cannot rule out the possibility of residual confounders given the limitations in the adjustment as a result of the small sample size. Despite these limitations, our study provides additional perspective and answers regarding the existing gaps for the management of patients with SMA. The application of concepts from behavioral economics under uncertainty revealed some unconscious biases (e.g., status quo) that lead to lack of treatment initiation, escalation (treatment switches), and overall TI. We expect that lessons learned from our results would allow the development of educational interventions and health policy strategies that ultimately improve the well-being and outcomes of patients with SMA and their families. Further research would be desirable to confirm the study findings and explore their generalizability to other countries with different backgrounds and healthcare systems.