Table 2. Patient pool and changes in discontinuation.

Class	Average Number of Active Patientsa	Extra Patients “Stocking Up” in March 2020b	Total	Estimated Discontinuations Resulting from Covid-19d	Avg Change in New Monthly Patientse
Drug			Discontinuations Post-Covid (March-Aug 2020)c
Addiction	520,407	27,145	179,835	-20,531	-7,933
Buprenorphine/
Naloxone
Immunosuppression	200,773	20,161	27,893	-1,922	-1,450f
Tacrolimus
Hormonal Contraceptive	240,740	6757	26,959	8,681	-2,771
Norgestrel-Ethinyl Estradiol
ADHD (Stimulant)	358,472	7,980	148,426	54,777	-15,335g
Dexmethylphenidate HCL
SSRI	4,511,608	279,061	897,621	164,867	-60,011
Escitalopram Oxalate
Antipsychotic	82,658	5,734	30,440	8042	91
Haloperidol

This table represents the potential scope of the impact of Covid-19 on medication adherence. These numbers represent only patients on these six drugs; the impacts are obviously magnified across all drugs and therapies. As seen in this table and the other figures, the exact impact of Covid-19 on a particular patient is uncertain: some patients stocked up more than normal, while others were more likely to discontinue use. The effects are heterogeneous across therapies as well.

^aThis is the number of patients demonstrating a pattern of actively filling prescriptions (beyond a single month or a trial of the drug). It is calculated by averaging the number of active patients from September 2019-February 2020 (pre-Covid months).

^bThis is the number of extra patients who “stocked up” on their medication in March, 2020. It is calculated by comparing the number of patients who filled prescriptions for 60+ DOS in March, 2020 against the average number of patients filling prescriptions for 60+ DOS in September 2019-February 2020.

^cThis represents the total number of patients discontinuing the medication from March 2020-August 2020. Some level of discontinuations is expected (e.g. changing to a different therapy).

^dThis is calculated by multiplying the linear probability model coefficients presented in Table 1 by the total number of active patient-months post-COVID; conceptually, it is the sum of exposing the active patients to the “change in likelihood of discontinuing.” Even small increases in the likelihood of discontinuing (e.g. a fraction of a percentage point) implies that thousands of additional patients will discontinue use. Note that in each case, it is only a fraction of the total discontinuations that are potentially attributable to Covid; however, it is a sizeable fraction. This leads us to believe that our results are plausible; even if estimates are off by an order of magnitude or more, tens of millions of patients are impacted across all drug categories.

^eThis is calculated by comparing the average number of new patients per month starting therapy in March 2020 –August 2020 against September 2019 –February 2020. Any patient with an approved claim is included in this analysis.

^fThe total number of transplants performed will be the primary driver of this number; access to immunosuppression is normally not a primary consideration. There will be some assessment of a patient’s access/coverage at the time of listing for transplant, but other factors (especially finding a match) are more significant. Live donor transplants were essentially halted in March 2020.

^gSee S3 Fig for additional discussion on seasonality.