The coronavirus disease 2019 (COVID-19) is an infectious disease caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). COVID-19 was declared a pandemic by the World Health Organization (WHO) on March 11, 2020 [1]. One of the unfortunate consequences of the COVID-19 pandemic was the increase in rates of mental health conditions such as depression, anxiety, and psychological distress [2-4]. Moreover, patients with preexisting depressive and anxiety disorders experienced worsening of their psychiatric conditions during the COVID-19 pandemic [5-7].

Given that first-line pharmacological treatments for both depression and anxiety are antidepressants, particularly selective serotonin reuptake inhibitors (SSRIs) and serotonin-norepinephrine reuptake inhibitors (SNRIs) [8,9], it would be reasonable to assume that antidepressants would have been prescribed and dispensed at an amplified rate after the onset of the COVID-19 pandemic. However, various population-based studies in the US and Canada that have examined dispensing patterns of antidepressants (based on fills/tablets dispensed, unique patients, or new prescriptions) have shown stable or even reduced rates relative to expected trends in the initial months following the onset of the COVID-19 pandemic [10-14], although there was an initial increase in rates in March 2020 [10,15].

While inferences regarding antidepressant adherence during the COVID-19 pandemic can be made from some of the aforementioned studies, explicit measures of antidepressant adherence were not included.

Nonadherence to antidepressants was already a familiar problem in both psychiatric and primary care populations well before the onset of the COVID-19 pandemic [16]. Unfortunately, the COVID-19 pandemic introduced additional factors that heightened the risk for medication nonadherence, including medication shortages, increased unemployment with the resulting loss of employer-based health insurance, fewer health care visits, decreased access to pharmacies, and psychological distress [17]. A survey that was conducted in the US found self-reported difficulties with medication adherence during the pandemic in those with psychiatric disorders, with 46% of patients admitting to forgetting or choosing not to take medications and 19% of patients claiming that they encountered problems in obtaining medications [18]. Two population-based studies have examined antidepressant adherence during the pandemic. Clement et al. [19] reported that patients taking escitalopram – the representative antidepressant in the study – were more likely to discontinue its use post-COVID compared to pre-COVID. Froese et al. [20] found that patients’ antidepressant adherence varied over time in 2020, but there was also a decrease in antidepressant discontinuation in previously adherent patients during 2020 compared with 2019.

There is a dearth of information regarding the effect of the COVID-19 pandemic on antidepressant adherence measures. The current study was conducted to characterize antidepressant adherence during the COVID-19 pandemic at the community pharmacy level.

Study Design

The study received approval from Samford University’s Institutional Review Board (EXMT-P-22-SUM-11). It was a retrospective review that was conducted at an independent community pharmacy that did not alter hours of operation during the COVID-19 pandemic. Dispensing data were extracted from the pharmacy’s Rx30 system software. The two time periods of the study were 6 months before (pre-pandemic) and 6 months after (post-pandemic) the index date of March 11, 2020. This index date was applicable to all patients in the study because it was the date that the WHO declared COVID-19 as a pandemic.

Study Population

Inclusion criteria included patients >18 years old who received therapeutic doses of an antidepressant medication with a minimum of two fills in the pre-pandemic period. Patients could be receiving any of the following antidepressants: bupropion, citalopram, desvenlafaxine, duloxetine, escitalopram, fluoxetine, fluvoxamine, levomilnacipran, mirtazapine, paroxetine, sertraline, venlafaxine, vilazodone, and vortioxetine. Patients receiving tricyclic antidepressants or trazodone as sole agents were excluded from the study because those drugs are frequently used for insomnia instead of depression and anxiety. Patients had to receive at least minimum daily dosages of antidepressants as specified in approved labeling. The requirement for patients to have a minimum of two fills in the pre-pandemic period was to ensure continuing use of the antidepressant as well as dedicated use of the particular community pharmacy to fill prescriptions. A total of 418 patients met inclusion criteria. A sample size calculation revealed the need for 201 patients using conservative assumptions to achieve a power of 80% and level of significance of 5%. Thus, 201 patients were randomly selected via number generator from 418 eligible patients.

Data Collection

Data collected from each patient’s chart included age, sex, insurance status, presence of maintenance medication for medical conditions, presence of other maintenance psychiatric medications, antidepressant(s) received and dates of fills. Regarding insurance status, coverage included commercial insurance, Medicaid, or Medicare; otherwise, patients were considered cash payers.

Outcomes

The primary outcomes were the changes in antidepressant adherence rates and percentage of patients with acceptable antidepressant adherence rates before and after pandemic onset. The secondary outcomes were the changes in antidepressant adherence rates and percentage of patients with acceptable antidepressant adherence rates before and after pandemic onset based on patients’ age, sex, insurance coverage, receiving maintenance medications for medical conditions, and receiving other maintenance psychiatric medications.

Statistical Analyses

Medication possession ratio (MPR) was used to measure adherence to antidepressant medications. The MPR was calculated for pre-pandemic and post-pandemic periods for all patients according to the following equation – (sum of days’ supply for all fills in period/number of days in period)×100% [21]. An acceptable adherence rate was considered 80% [21]. The mean MPR and percentage of patients with MPR ≥80% were calculated and compared statistically between the two time periods using a paired sample t-test and Fisher’s exact test, respectively. Multiple regression analyses were conducted to determine the possible effects of age, sex, insurance coverage, receiving maintenance medications for medical conditions, or receiving other maintenance psychiatric medications on the primary outcomes. A linear regression analysis was conducted using post-pandemic MPR – pre-pandemic MPR as the dependent variable. The regression is reasonable as the collinearity among the variables is low (VIF <5 for all variables). A logistic regression analysis was also conducted, which treated MPR as a binary variable of <80% or ≥80%. All analyses were conducted using SPSS software version 29.0.

Demographic Information

Of the 201 patients included in the study, 116 (57.7%) were 18–64 years old, 150 (74.6%) were females, 181 (90.0%) had insurance coverage, 164 (81.6%) received maintenance medications for medical conditions, and 74 (36.8%) received other maintenance psychiatric medications. Antidepressant medications that were prescribed to the patients in the study included sertraline (n=33), duloxetine (n=31), escitalopram (n=29), fluoxetine (n=28), venlafaxine (n=26), citalopram (n=22), bupropion (n=15), paroxetine (n=12), desvenlafaxine (n=2), vortioxetine (n=1), duloxetine/sertraline (n=1), and sertraline/vortioxetine (n=1).

Primary Outcomes

Primary outcomes are presented in Table 1. There was a 15.8 percentage point decrease in mean MPR from pre-pandemic to post-pandemic periods, and this decrease was statistically significant (p<0.001). There was a 14.5 percentage point decrease in the percentage of patients with MPR at least 80% from pre-pandemic to post-pandemic periods, and this decrease was statistically significant (p=0.005). Of the 201 patients, 142 (70.6%) experienced a decrease in MPR in the post-pandemic period compared to the pre-pandemic period. Thirty-two (15.9%) patients had no antidepressant fills in the post-pandemic period; their mean MPR in the pre-pandemic period was 54.7%, and only seven (21.9%) of those patients had a MPR ≥80% in the pre-pandemic period.

Table 1 . Primary outcomes (N=201).

Variable	Pre-pandemic	Post-pandemic	p-value
Mean MPR (%)	80.4	64.6	<0.001
Patients with MPR ≥80% (%)	65.2	50.7	0.005

Secondary Outcomes

Secondary outcomes are presented in Table 2. Mean MPR and percentage of patients with MPR at least 80% decreased from pre-pandemic to post-pandemic periods across all patient subgroups with no statistically significant effect of patient subgroups on the primary outcome variables.

Table 2 . Secondary outcomes (N=201).

Variable	Mean MPR (%)			Patients with MPR ≥80% (%)
	Pre-pandemic	Post-pandemic	p-value*	Pre-pandemic	Post-pandemic	p-value†
Age			0.269			0.273
>65	84.0	66.5		70.6	47.1
18–64	77.8	63.2		61.2	44.8
Sex			0.911			0.462
Female	79.9	63.9		64.0	49.3
Male	81.9	66.6		66.7	54.9
Insurance coverage			0.209			0.879
Yes	81.5	64.8		66.3	50.8
No	70.5	62.0		50.0	50.0
Maintenance medical medications			0.053			0.850
Yes	82.7	68.0		70.4	54.9
No	70.7	49.8		44.7	34.2
Other maintenance psychiatric medications			0.097			0.392
Yes	80.3	60.2		63.5	44.6
No	80.5	67.1		68.5	54.3

*Regression analysis of mean MPR vs. variable..

^†Regression analysis of patients with MPR ≥80% vs. variable..

This study found a significant reduction in adherence rates as well as satisfactory adherence rates to antidepressant therapy in patients at a community pharmacy in the six months following the COVID-19 pandemic relative to the six months prior to the pandemic. We are unaware of previous studies that have examined this issue on the community pharmacy level. Numerous studies have investigated the potential impact of COVID-19 on adherence to various medications, and the results have been variable. However, our findings of decreased antidepressant adherence during the COVID-19 pandemic are consistent with those from studies that have examined adherence to antiretroviral therapy, asthma controller medications, disease-modifying antirheumatic drugs, antiseizure medications, antipsychotics, antiglaucoma medications, and antihypertensive medications [22-28].

This study also found that the primary outcomes were unaffected by patients’ age, sex, insurance coverage, receiving maintenance medications for medical conditions, or receiving other maintenance psychiatric medications. We were particularly interested in two of these variables – insurance coverage and receiving maintenance medications for medical conditions. Unaffordability of medications was an obvious problem for many patients during the COVID-19 pandemic because of loss of employment and loss of insurance coverage; however, it turned out that 90% of our patients had coverage through commercial insurance, Medicaid, or Medicare. We had hypothesized that patients requiring maintenance medications for medical conditions would be more likely to overcome medication adherence barriers out of fear of worsened physical health, consequently positively affecting antidepressant adherence as well. Interestingly, this was the only independent variable in the study that approached statistical significance (p=0.053). The power analysis that was conducted to determine an adequate sample size in the study was based on primary outcomes only. Therefore, it is possible that a Type II error occurred, and a larger sample size would have detected a statistical difference in the case of this secondary outcome.

Approximately 1 out of every 6 patients failed to fill an antidepressant prescription during the post-pandemic period. Clement et al. [19] found that patients’ likelihood of discontinuing antidepressant therapy after the spread of COVID-19 was statistically significantly greater than during the pre-pandemic period. The patients in our study who had no antidepressant fills in the post-pandemic period also had a much lower mean MPR and percentage of patients with MPR ≥80% relative to the entire sample, suggesting that those patients who already have preexisting difficulties with antidepressant adherence are even more susceptible to discontinuing antidepressant therapy altogether in the face of mounting challenges associated with health care crises.

The most common antidepressant in this study was sertraline. Because this particular antidepressant had a well-known supply shortage during the COVID-19 pandemic [29], one possible explanation for the decreased adherence rates to antidepressants seen in our study was lack of access to sertraline. However, a post-hoc analysis revealed that adherence measures during the post-pandemic period for sertraline-treated patients were very similar to those of the entire sample. Specifically, the mean MPR in the post-pandemic period was 61.4% for sertraline-treated patients vs. 64.6% for all patients, and acceptable adherence (i.e., MPR ≥80%) in the post-pandemic period was 51.5% for sertraline-treated patients vs. 50.7% for all patients.

We assessed antidepressant adherence in the post- pandemic period as a single 6-month time frame (i.e., mid-March through mid-September 2020). Uthayakumar et al. [14] found that rates of antidepressant tablet dispensing decreased considerably from March to April and took until August 2020 to restabilize. Froese et al. [20] found that the odds of antidepressant adherence were lower in April–June 2020 compared with the previous quarter but were higher in both July–September and October–December 2020 compared with the quarter prior and compared with those quarters in 2019. Thus, there is some evidence that antidepressant adherence was dynamic during the months following onset of the pandemic, which is reasonable since factors that underlie medication adherence were evolving as well. It is possible that we would have found varying rates of antidepressant adherence over time in this study if we had examined different intervals within the 6-month time frame.

Community pharmacists are very accessible health care providers, and studies have shown that they are effective at improving patient adherence to antidepressants [30,31]. But our findings point to the need for community pharmacists to further identify opportunities to improve antidepressant adherence during health care crises. Such measures might include things like reviewing antidepressant adherence in real time and contacting patients when there are concerns, encouraging prescribers to utilize 90-days fills, aiding in prescription renewals when patients miss appointments with clinicians, and providing assistance with prescription pick-up or delivery.

There were certain limitations of our study, chiefly due to the fact that we relied solely on data available through the community pharmacy’s dispensing software. First, there was a reliance on prescription fills as a proxy for actual medication use. Second, it was not possible to definitively establish diagnoses that prompted antidepressant use. Third, it was not possible to verify exact causes of antidepressant nonadherence. Lastly, there was an assumption of static conditions during the study time frame (e.g., patients continued to use the same community pharmacy instead of changing pharmacies or changing to a mail-order service, and antidepressant treatment was not discontinued by the health care provider). Another limitation of our study was that it was conducted at a single community pharmacy site, so results may not be generalizable to other community pharmacy settings. Finally, we were unable to discern the impact of possible confounders (e.g., whether patients were diagnosed with COVID-19, whether there was a supply shortage of certain medications).

Adherence to antidepressants significantly decreased after onset of the COVID-19 pandemic for patients at a community pharmacy. These findings suggest the need for community pharmacists to identify opportunities to improve antidepressant adherence during future health care crises.

None.

None.

No potential conflict of interest relevant to this article was reported.