According to the selection criteria (i.e., minimum volume of adalimumab dispensation from hospital prescriptions), data from the 40 hospitals in the experimental case and from 91 out of the 211 hospitals in the general case were examined from the original dataset. The 120 general case hospitals were removed from the analysis because they were associated with less than 30 boxes of adalimumab dispensed each month.

As shown in Table 1, both groups are geographically well distributed, with the exception of Corse. The experimental group contains far more public teaching hospitals than the general case group. These hospitals are also larger, as illustrated by a higher level of MSO activity, and the majority (70%) of them treat more than 40,400 patients per year (T4 and T5 levels of MSO activity ranges). On the other hand, most of the general case hospitals (55%) treat between 15,000 and 40,000 patients per year (T3 level of MSO activity ranges). The medical activity level appears to be quite similar, with most hospitals having over 20,000 patients yearly (M4 level of activity) in each group. Prior to biosimilar launch, the monthly mean number of adalimumab boxes dispensed from each hospital prescription and the number of hospitals that had previous experience with the experiment (ETA+) were the main differences between each group and were higher in favor of the experimental case.

Figure 3 highlights that hospitals in the experimental case had a significantly higher monthly volume of adalimumab prescriptions before the first biosimilar launch (October 2018). Thereafter, the volume of adalimumab dispensations from hospital prescriptions increased over months: from November 2017 to September 2018, 23,200 boxes per month on average were dispensed, and from October 2018 to October 2020, 29,200 boxes per month were dispensed. The distribution remained consistent across groups during the study period, with an average of 62.6% of adalimumab boxes dispensed from hospitals’ prescriptions in the experimental case (minimum 61.3%; maximum 64.2%) (Fig. 3).

Adalimumab market share (monthly global volume of boxes) from November 2017 to October 2020 by hospital group of incentives. The total volume of adalimumab boxes dispensed by community pharmacies is given for the months from November 2017 to October 2020. Each month, the total quantity of adalimumab dispensed comes from prescriptions from hospitals in the experimental case (black bars) and from hospitals in the general case (white bars), which was consistent across months. The black line highlights the adalimumab boxes ratio dispensed from experimental case hospitals prescriptions

Since adalimumab’s first biosimilar launch, the volume of adalimumab biosimilars has increased quickly, from 280 boxes in November 2018 to more than 10,000 boxes dispensed in October 2020. Between October 2018 and October 2020, the volume of adalimumab biosimilar dispensed from hospital prescriptions increased by 64% in the experimental case and 72% in the general case. Thus, the quantity index increased by +2.1% between 2018 and 2019 and remain constant at −0.2% in 2020. In contrast, the price index decreased by −13.5% in 2019 compared with 2018, and by −1% in 2020 versus 2019. Globally, hospitals in the experimental case accounted for 66% of total of adalimumab biosimilar volume (minimum 62.9%; maximum 87.7%).

The biosimilar rate increased in both groups, but hospitals in the experimental case maintained a higher biosimilar penetration than the general case group (Fig. 4). An upward trend was observed in both cases, but their respective growth rates resulted in a significantly lower biosimilar uptake (p = 0.029) on average throughout the period for general case hospitals (16.7% vs 23.2%) (Table 1). This difference widened over time, reaching mean biosimilar rates of 29.9% and 38.6% in October 2020, respectively. The boxplot also reveals wide variation among hospitals in each group (Fig. 4).

Monthly distribution of the rate of adalimumab biosimilars (BS) by incentive group. Dark boxplots illustrate the distribution of the adalimumab BS rate for hospitals in the experimental case by month from November 2018 to October 2020. Gray boxplots are the same for hospitals in the general case. Their concomitant study allowed us to observe that the mean BS rates by group both increased since the beginning of 2019. The boxplot shows that the difference in the BS rate is more concentrated in the experimental case

Finally, hospitals in the experimental case were mainly larger public teaching hospitals that induced a higher volume of adalimumab dispensations in community pharmacies and more rapid biosimilar uptake in these dispensations.

Four subgroups according to the type of incentive (general [ADA−] or experimental [ADA+]) and previous experience with the etanercept biosimilar experiment (“no” [ETA−] or “yes” [ETA+]) were developed. There were 23 hospitals in the experimental case (ADA+) with previous experience with the incentive (ADA+/ETA+ subgroup) and 18 hospitals with no experience (ADA+/ETA−). Among hospitals in the general case (ADA−), ten were in the subgroup with previous experience with this incentive (ADA−/ETA+), and 81 hospitals had never had any experience with the incentive (ADA−/ETA−).

Three of the four subgroups (ADA+/ETA+, ADA+/ETA−, and ADA−/ETA+) were geographically well distributed. They were also similar in terms of establishment type because most of them are public nonteaching hospitals and at least one-third are public teaching hospitals. The activity levels of the three subgroups were comparable and concentrated between 15,000 and 90,000 or more patients treated annually (T3–T5 MSO activity levels). The same was true for the medicine activity (M3–M5 activity levels), with the majority of hospitals (60%) treating 20,000–55,000 patients per year (Table 2). Otherwise, the hospitals in the ADA−/ETA− subgroup were different from one another, with the greatest number of public nonteaching hospitals (82%) and the lowest rate of public teaching hospitals (6%). These hospitals are consequently smaller, with 14% with T0–T2 MSO activity levels and a majority (57%) with T3 MSO activity levels. Moreover, no hospital had an M5 level of medicine activity in this subgroup (Table 2).

Regardless of the adalimumab experimental case subgroup (ADA+/ETA+ and ADA+/ETA−), adalimumab biosimilar uptake in community pharmacy dispensation was identical (mean biosimilar rate of 23%). This uptake has increased in parallel since the first biosimilar launch in October 2018 (Table 2, Fig. 4). ADA+/ETA− hospitals had a higher adalimumab biosimilar uptake than ADA+/ETA+ hospitals in October 2020, but this difference was not significant (40.4% vs 37.3%, p = 0.875).

In contrast, the ADA−/ETA− subgroup had the lowest monthly mean number of adalimumab boxes dispensed from the hospitals’ prescriptions (73 boxes on average vs 262–504 boxes on average for the other subgroups). This group also had a lower rate of biosimilar use (16% on average for the period studied) (Table 2).

There was a difference between general case subgroups in the ADA− group contrary to the experimental case subgroups. However, this difference between ADA−/ETA+ hospitals (mean biosimilar rate of 21%) and ADA−/ETA− hospitals (mean biosimilar rate of 16%) was not statistically significant (p = 0.086).

Figure 5 highlights a comparable trend in the progression of the monthly mean biosimilar rate of each subgroup, with the exception of ADA−/ETA−. They all constantly increased from October 2018 to February 2020 before a downturn. The monthly mean rate of biosimilar growth in the ADA−/ETA− subgroup appeared to be more consistent and slower than that in the other subgroups.

Monthly mean biosimilar rate (mBSr) by incentive subgroup: general (ADA−) or experimental (ADA+) case and prior (ETA+) or no prior (ETA−) experience with the experimental incentive. The mBSr of the two subgroups ADA+/ETA− and ADA+/ETA+ is nearly identical across months. The ADA−/ETA+ subgroup appears to follow the same pattern as the previous two subgroups but with a lower mBSr. The ADA−/ETA− subgroup mBSr appears to be evolving at a much slower rate than the other three groups

Graphically, we also observed a gap between the three previous subgroups who had experienced the experimental incentive at least once (ADA+/ETA−, ADA+/ETA+, ADA−/ETA+) and the ADA−/ETA− subgroup since the launch of the experimental case in April 2019. However, no significant difference was observed to confirm the greater acceleration (in March 2020, mBSr [SD] 24.1% [14.1%] for the ADA−/ETA− subgroup vs 31.9% [12.4%] for the ADA−/ETA+ subgroup). The ADA+/ETA− and ADA+/ETA+ curves nearly overlap each other, thus a previous experience with ETA seems to not have a significant impact on adalimumab experimental case biosimilar uptake. Otherwise, the curves observations and the inflection points found (see Fig. 6 to Fig. 9 in the Appendix [see the electronic supplementary material]) show a consistent increase of adalimumab biosimilar uptake for all subgroups from the start of the experiment (decree published in April 2019) until a plateau around March 2020 that could indicate a ceiling effect. However, this was also the first coronavirus 2019 (COVID-19) lockdown, between March and May 2020, in France. This event should have paused everything as we observed a clear slowdown of biosimilar uptake in all cases. The fact that the curves did not revert to their pre-March 2020 slope suggests that the probable influence of the COVID-19 crisis was still active in October and November 2020, when a new lockdown was implemented.

Finally, hospitals not under the experimental incentives for adalimumab (ADA−) seemed to be positively influenced by their previous experience with the experimental incentive for etanercept biosimilars that induced a better adalimumab biosimilar uptake. However, this influence appears to be minimal or nonexistent for the hospitals in the experimental case.

Using adalimumab biosimilars will have saved €8.9 million in the experimental case hospitals and €7.5 million in the general case between January 2019 and October 2020, based on hospital prescriptions of our sample (Table 3). The savings estimates for the NHI were €1.4 million higher in the experimental case, but the mean saved by box was lower for the experimental case than in the general case when compared to the volume of adalimumab dispensed in community pharmacies (Table 3). Otherwise, the simulation with general case hospitals having the same monthly mean biosimilar rate as experimental case hospitals resulted in a €8.5 million savings, compared to €7.5 million with its proper mean biosimilar rate.

The savings calculated are probably underestimated as we only had access to a sample of community pharmacies, and therefore only a sample of hospitals in the general case, and prescription renewals, which form part of the remuneration calculation, could not be extracted.