Affective disorders, in particular depression and anxiety, are common non-motor disorders in Parkinson’s disease (PD). The burden of depression and anxiety has been studied in several clinical and research setting studies, and the reported prevalence ranges from 3% to 89% for depression¹ and 7% to 55% for anxiety.² Additionally, patients commonly present with both disorders.³ These disorders have negative impacts on disease outcomes; for example, depression in PD is associated with worse disability and health-related quality of life.^4–6

Persons with PD are typically above age 60,⁷ at risk for cognitive impairment,^8,9 and have dysfunctional dopaminergic^10,11 and cholinergic systems,^11–13 which makes treatment selection among those with an affective disorder challenging. Several U.S. Food & Drug Administration (FDA) medications indicated for depression and anxiety have anticholinergic properties, dopamine receptor-blocking properties, or gamma-aminobutyric acid (GABA) effects and are discouraged from use in older adults, individuals with cognitive impairment, or individuals with PD due to adverse event concerns.^14,15 The American Geriatrics Society Beers Criteria identifies these discouraged drugs as potentially inappropriate medications (PIMs).¹⁴ Clinicians, researchers, and regulators can use the Beers Criteria to identify gaps in care quality and quantify PIM-associated negative health events.¹⁴ Furthermore, based on evidence-based medicine reviews by the Movement Disorders Society and American Academy of Neurology, few of these medications have demonstrated efficacy and safety specifically in persons with PD.^16–19

Prior studies have quantified the frequency and type of affective disorder treatments PD patients receive.^20–30 Many have been conducted in smaller samples,^{20,22–24,29,30} with the largest study sample derived from a specialty center-based PD cohort.²⁶ However, many persons diagnosed with PD in the United States do not receive care from a neurologist or movement disorders specialist.³¹ Prior studies have used self-report^21,24,29 or clinician documentation³⁰ to collect information about medication use across all health care settings. No prior study has used national data to determine the frequency and type of treatment persons with PD receive for affective disorders in the United States, nor used standardized criteria to evaluate the safety of medications used for affective disorders in older adults with PD. To address these gaps in knowledge, and provide a foundation for interventions with the potential to improve the clinical management of PD, the aims of this study were to 1) examine the proportion of older adults with PD treated for affective disorders, 2) determine the proportion treated receiving a PIM for an affective disorder, and 3) determine whether patient or clinician characteristics were associated with receipt of a PIM for an affective disorder.

A waiver of consent was obtained for this study by the University of Pennsylvania Human Research Protections Office and Centers for Medicare and Medicaid Services (CMS).

There were 56,867,603 Medicare beneficiaries in 2014. Of those, 10,364,579 met coverage criteria and were alive at the end of 2014. Restricting to those with PD, a minimum of one prescription fill, and no nursing home/hospice care, resulted in a sample of 104,264 beneficiaries. The exclusion of persons with other neurodegenerative diseases and mental health conditions resulted in a final analytic sample of 84,323 beneficiaries.

By the end of 2014, 23.6% (n=19,866) and 16.0% (n=13,485) of beneficiaries with PD had a new or prevalent diagnosis of depression or anxiety. Examining mutually exclusive groups, 15.1% (n=12,703) had a diagnosis of depression only, 7.5% (6,322) had a diagnosis of anxiety only, and 8.5% (n=7,163) had comorbid depression and anxiety. Significant (p<0.05) and meaningful (standardized difference ≥0.1) differences were noted for several characteristics (Table 1). Beneficiaries with an affective disorder diagnosis were more likely to be female, diagnosed with a greater number of chronic conditions, diagnosed with non-AD dementia, and prescribed affective disorder treatment (Table 1). Of note, 20.8% (n=12,090) and 21.1% (n=12,268) of individuals without a formal affective disorder diagnosis received a medication indicated for the treatment of depression or anxiety, respectively.

Of those with depression only, 80.7% (n=10,253) of beneficiaries filled a prescription indicated for the treatment of depression. The five most common prescriptions to treat those with depression only were sertraline (16.0% of all prescriptions), citalopram (15.4%), escitalopram oxalate (12.8%), duloxetine (9.1%), and mirtazapine (8.2%). On-label anxiety treatment was prescribed to 62.9% (n=6,322) of beneficiaries with anxiety only. Alprazolam (18.8% of all prescriptions), clonazepam (17.1%), lorazepam (14.0%), sertraline (12.0%), and escitalopram oxalate (9.6%) were the most common prescriptions for individuals diagnosed with anxiety only. In total, 93.1% (n=6,670) of beneficiaries with comorbid depression and anxiety received at least one affective disorder prescription. Beneficiaries with comorbid depression and anxiety most commonly received prescriptions for clonazepam (10.5% of all prescriptions), sertraline (9.6%), alprazolam (9.6%), escitalopram oxalate (8.4%), and citalopram hydrobromide (8.3%).

Table 2 displays characteristics associated with treatment for each affective disorder group. Treatment prevalence differed by sex; those with depression only undergoing treatment were younger, and those with comorbid depression and anxiety undergoing treatment were more likely to be white (χ² test p-values <0.05 and standardized differences ≥0.1).

Among those treated, there were no characteristics that substantially differed between those treated with a PIM versus not treated with a PIM in any of the affective disorder groupings, except that those treated for comorbid depression and anxiety were younger (X²_(df=2)=19.78, p<0.0001, standardized difference=0.1095) and more likely to be female (X²_(df=1)=27.06, p<0.0001, standardized difference=−0.132) (Table 3). PIM receipt was least common among persons with depression only (12.8%); on-label PIM use was much higher for persons with anxiety only (75.9%) and with comorbid depression and anxiety (63.8%). Pooling all groups, clonazepam accounted for 24.2% of all PIM prescriptions, followed by alprazolam (23.8%), lorazepam (18.4%), paroxetine (17.6%), and amitriptyline hydrochloride (4.2%).

In the fully adjusted models of prescription-level data, female sex was significantly associated with PIM prescriptions among those with depression only (aOR=1.18, 95% CI: 1.04, 1.34; X²_(df=1)=6.87, p=0.0088) and with both depression and anxiety (aOR=1.15, 95% CI: 1.07, 1.24; X²_(df=1)=13.15, p=0.0003) (Table 4). Beneficiaries ≥85 years of age with comorbid depression and anxiety had a lower odds of PIM than those 65-74 (aOR=0.85, 95% CI: 0.76, 0.94; X²_(df=1)=9.33, p=0.0023). Among those with anxiety only, each additional comorbidity was associated with an increased odds of PIM receipt (aOR=1.03, 95% CI: 1.00, 1.05; X²_(df=1)=4.07, p=0.0436). Dementia was associated with lower odds of receiving PIMs in both the anxiety only group (aOR=0.78, 95% CI: 0.68, 0.90; X²_(df=1)=11.39, p=0.0007) and the comorbid group (aOR=0.89, 95% CI: 0.83, 0.97; X²_(df=1)=7.78, p=0.0053). In general, psychiatrists and geriatricians were less likely to prescribe PIMs than other clinicians, particularly for anxiety (aOR=0.42, 95% CI: 0.22, 0.78; X²_(df=1)=7.51, p=0.0061). Neurologists were more likely to prescribe PIMs for PD patients with comorbid disorders (aOR=1.21, 95% CI: 1.07, 1.38; X²_(df=1)=8.67, p=0.0032).

In the sensitivity analysis, the number of treated individuals dropped in all affective disorder groups (74.1% [n=9,410] for depression alone, 55.8% [n=3,525] for anxiety only, and 88.9% [n=6,368] for comorbid depression and anxiety). The proportion treated with PIMs also slightly declined (11.9% [n=1,121] for depression only, 73.2% [n=2,581] for anxiety only, and 58.5% [n=3,725] in the comorbid group). The GEE sensitivity analysis (see Table, Supplemental Digital Content 2, which details the model results) found the same factors significantly associated with PIMs; however, there were two additional significant findings: age (for anxiety alone) and residence (for comorbid depression and anxiety).

We studied the patterns of diagnosis and treatment of affective disorders among older adults with PD in the United States. Affective disorders are frequently diagnosed in this sample; annual treatment prevalence was high for persons diagnosed with depression with or without comorbid anxiety, but modest for anxiety alone. Prescription management of affective disorders with PIMs was somewhat common, and varied by diagnosis, patient demographic characteristics, comorbid disease, and prescribing physician specialty.

In this nationally representative sample of older adults, 23.6% and 16.0% of individuals diagnosed with PD met criteria for a depressive disorder and an anxiety disorder. The depression estimates we found in this study are consistent with estimates from two meta-analyses conducted in PD, which found prevalence estimates of 17%¹ and 22.9%⁴³ for major depression and 22%¹ and 31,3%⁴³ for minor depression. The prevalence of anxiety disorders in our study (16.0%) was lower than that found in a prior meta-analysis (31%).² However, that meta-analysis defined anxiety disorders using specialty diagnostic criteria used primarily by psychiatrists (i.e., the Diagnostic and Statistical Manual) and often included studies based in specialty care settings.² We found treatment with prescription medication to be particularly high for depression, with 80.7% of those with depression only receiving prescription treatment, and 93.1% of those with depression and comorbid anxiety receiving treatment. The majority of individuals with anxiety only were also treated (62.9%), but this proportion was lower than that for depression. These treatment rates are higher than previously reported.^20–30 Our claims-based method of identifying affective disorders diagnoses and treatment may have resulted in restricting our sample to individuals with clinically definite or more severe disease, which would represent groups more likely to be treated. Additionally, most prior studies did not examine the use of all FDA on-label treatments.

The presence of an anxiety diagnosis substantially increased the likelihood of PIM receipt, with 75.9% of treated individuals with anxiety alone receiving a PIM and 63.8% of treated individuals with comorbid depression and anxiety receiving a PIM. This PIM use compares to only 12.8% in those with depression alone. This higher prevalence among those with anxiety was driven by benzodiazepine use. Clinical practice guidelines uniformly recommend against using benzodiazepines as first-line treatment for anxiety disorders in older adults ⁴⁴; however, benzodiazepines were by far the commonly prescribed PIM (66.4%) in our PD sample. The risks of benzodiazepine use in older adults are well documented and include reduced mobility, decreased ability to drive, cognitive impairment, and increased risk of falls.^45–47 These risks and the associated negative health outcomes may be even greater in persons with underlying PD. Several factors may explain why PD patients are still prescribed benzodiazepines. PD patients may have persistent or resistant anxiety disorders. Competing demands for physician attention and limited time or access to alternative treatments for anxiety, such as cognitive-behavioral therapy, may also result in higher benzodiazepine prescribing. Persons with PD often see multiple clinicians for their PD and non-PD related symptoms/conditions. They may receive benzodiazepines from clinicians without knowledge, experience, or resources to monitor for the harms of or taper these medications. Finally, benzodiazepines are viewed as more useful than alternatives for the treatment of PD-related dystonia and sleep disorders, particularly REM behavior disorder. Our sensitivity analysis suggested that while some benzodiazepine use was short-term, committed PIM use still occurred in at least 50% of individuals with anxiety or anxiety and depression. Future studies should examine longitudinal trends in affective disorder treatment in PD, including the impact of prescription drug monitoring programs on benzodiazepine prescriptions.

Paroxetine, a selective serotonin reuptake inhibitor with high anticholinergic properties, was the most frequently prescribed non-benzodiazepine PIM. Although drugs with high anticholinergic properties are discouraged from use in older adults,¹⁴ the negative impacts of these medications specifically in individuals with PD, who have cholinergic dysfunction at baseline, are just beginning to be studied.^48–50 There are multiple possible explanations for the observed level of paroxetine use. A multi-center randomized controlled trial found paroxetine to be more effective than placebo, and similarly effective to venlafaxine, for depression treatment in PD,⁵¹ although anticholinergic adverse effects were preferentially reported in the paroxetine-treated group. Anxiety is a common non-motor fluctuation symptom, and a manifestation of early cognitive impairment in PD; paroxetine has a clinical indication for mixed anxiety/depression. In spite of these potential benefits, the consistent designation of paroxetine as a Beers Criteria PIM for all older adults, combined with the possibility that high anticholinergic medications may interfere with acetylcholinesterase inhibitors for the treatment of cognitive impairment highlights the need for additional drug-disease interaction and comparative safety/effectiveness research in PD.

Affective disorders were more common in those diagnosed with dementia. There may be a bidirectional relationship between cognitive impairment and depression in PD.¹⁵ We did also find, in adjusted models, that those with dementia had lower odds of PIM prescriptions in both the anxiety only and comorbid group. The general, decreased likelihood of PIM prescriptions among those with dementia may reflect good adherence to the recommendation against the use of anticholinergics among those with cognitive impairment.¹⁴

In our study, females were more likely to be diagnosed with an affective disorder. Consistent with our findings, females with PD are known to be more likely to have an affective disorder.⁵² Despite a prior study suggesting that males with PD have more negative views about treatment for affective disorders,²¹ we found, per standardized differences, inconsistent differences in prescription treatment by sex. Females with depression only, as well as with comorbid depression and anxiety, after adjustment, were slightly more likely to receive PIM prescriptions.

At least one other study has found that females with PD are at greater risk for receiving inappropriate prescriptions.⁵³ Given the multitude of treatment options, further research is needed to understand why females are not being prescribed more benign treatment options.

A prior study found that geriatricians are less likely to prescribe PIMs.⁵⁴ Consequently, our finding that geriatricians less frequently prescribe PIMs for the treatment of depression only and anxiety only is not surprising. Psychiatrists were significantly less likely to prescribe PIMs for those with comorbid affective disorders. No benefit of neurologist prescribing was found; in fact, neurologists were more likely to prescribe PIMs (primarily benzodiazepines) among the anxiety alone group. Again, further study is needed to disentangle benzodiazepine use for anxiety alone (which is potentially inappropriate) versus benzodiazepine use in a patient with multiple motor and non-motor indications (i.e., dystonia, sleep disorders). Studies that quantify the potential harms from short- and long-term benzodiazepine use in persons with PD are needed to support shifts in prescribing practices and modification of PD clinical treatment guidelines.

This study had several strengths, including a large sample that allowed us to produce population-representative estimates of PD care, which does not center around tertiary care centers. However, we were limited in our ability to assess non-medication treatment modalities, such as cognitive-behavioral therapy, in our study. Many of the medications we studied have other off-label and on-label indications, although we attempted to account for this as in the case of antipsychotics and selegiline. PD medications, including dopamine agonists and rasagiline, may have symptomatic benefit for affective disorders in PD.¹⁷ Consequently, we could be over- or under-estimating the treatment prevalence in our study. We also did not examine prescribing patterns among individuals who had depressive or anxiety symptoms or received affective disorder medications, but did not meet study ICD-9 CM diagnostic criteria. Thus, we may have underestimated the total burden of affective disorders in our sample. We also were unable to examine clinical factors that may influence prescribing patterns, such as affective symptoms or severity, motor and cognitive function, previous history of medication response, and patient preference. Although we found several factors to be significantly associated with PIM prescriptions, most associations were reasonably modest; thus the clinical and population-level significance of these initial data remains to be determined. Associations were strongest for prescriber specialty, and future studies will determine whether these associations can be detected in a clinical setting.

Despite these limitations, we present national estimates of affective disorder treatment in PD along with the prevalence and risk factors for PIM receipt. These data add to the growing evidence basis for future updates of clinical guidelines that will approach treatment of PD patients from a drug-disease safety perspective.