Prescription Drug Utilization among Patients with Essential Tremor: A Cross‐Sectional Study of More Than 36,000 Patients

Abstract

Background

Essential tremor (ET) is a chronic, progressive neurological disease that affects an estimated 7 million individuals in the United States (ie, 2.2% of the entire U.S. population). Despite its high prevalence, there are a few published studies on patterns of prescription medication use among patients.

Objective

The aim was to examine prescription drug medication use among ET patients.

Methods

This is a cross‐sectional study of ET patients, age ≥40, with at least 1 prescription medication fill using the Optum's de‐identified Clinformatics Data Mart Database from 2018 through 2019. We examined patterns of fills of key agents used to treat ET.

Results

The final sample comprised 36,839 ET patients in the United States; 89% had at least 1 prescription drug claim over a 2‐year period, indicating that 9 of 10 ET patients take a medication to treat their disease. For each of the 3 most frequently prescribed medications, only a modest fraction (1/5 to 1/4) of patients were taking that medication. Adherence to these agents was 52% to 61%. A high percentage of patients had fills for more than 1 of the main agents we studied.

Conclusion

These data illustrate a need for medication in the ET population. There is only 1 FDA‐approved medication to treat ET, propranolol, and less than 25% of ET patients used this drug during our study period. At the same time, no single agent was utilized by more than one quarter of ET patients, adherence was low, and use of multiple agents was common. For such a common disease, the pharmacotherapeutic landscape is impoverished.

Essential tremor (ET) is a chronic, progressive neurological disease that affects an estimated 7 million individuals in the United States (ie, approximately 2.2% of the entire U.S. population). ¹ ET is the most common cause of tremor in humans; in some studies it is 20 times more prevalent than Parkinson's disease (PD). ² Incidence data ³ suggest that approximately 1000 new ET cases arise in the United States every day.

The central clinical feature of ET is an 8‐ to 12‐Hz kinetic tremor (ie, tremor during volitional movement). This tremor is usually progressive, ⁴ , ⁵ , ⁶ producing disabilities with basic daily activities such as eating and writing, ⁷ , ⁸ , ⁹ and it has been estimated that 15% to 25% are sufficiently motorically disabled by the high‐amplitude shaking that they cannot continue to work. ⁴ , ⁵ , ⁶ , ¹⁰

Evidence‐based reviews indicate that there are 2 or 3 frontline medications, ¹¹ , ¹² although their efficacy is limited, with an estimated 30% to 50% of patients not responding to these medications. ¹¹ Further, the side effect profile of these medications is often intolerable. ¹¹ , ¹³ In a survey of 223 ET patients using medication, although nearly 71% had taken primidone or propranolol, the first‐line agents, 56% of those individuals discontinued use. ¹⁴ A more recent study of 144 ET patients who were followed prospectively reported that even among elders with chronic, long‐standing ET, there is considerable ongoing medication adjustment, underscoring the need to improve the medication situation for ET patients. ¹⁵

Although these studies provide a valuable window with which to view medication usage in ET, they include relatively small samples of patients from a single clinic. Studies using secondary, administrative data such as medical claims—the bills providers submit to patient's payers (eg, insurance providers)—provide large, often national samples and real‐world evidence on utilization. ¹⁶ , ¹⁷ , ¹⁸ A recent study using medical claims data (including pharmacy claims) on 128,263 continuously insured patients with at least 1 ET claim per year for at least 3 years after the first ET diagnosis found that two‐thirds of patients had at least 1 fill. ¹⁹ However, this inclusion criterion meant only patients with at least 1 record of an encounter with an ET diagnosis each year for 3 years in a row were included. Apart from this study, surprisingly little has been written about patterns of prescription medication use among patients in the United States with this common disorder, including the extent to which the different medications are used, their use in conjunction with other medications, and the typical duration of use. To address these fundamental questions, we leveraged 2 years of rich de‐identified medical claims data from Clinformatics Data Mart Database (CDM), ²⁰ which were derived from a database of claims for members of large commercial and Medicare Advantage health plans in the United States.

Patients and Methods

Study Design

This is a cross‐sectional secondary data analysis of adults aged ≥40 years with at least 1 medical claim extracted from CDM from 2018 through 2019. ²¹ , ²² , ²³ , ²⁴ The inclusion criteria for the analytic sample were (1) age ≥40 years, (2) continuous enrollment in the health insurance plan for the full calendar year, and (3) at least 1 claim with a diagnosis of ET (G25.0). Unlike the earlier study, ¹⁹ we did not require that patients have an ET claim both years of our data. We excluded any patients with related neurological conditions (G20 [PD], G21 [secondary parkinsonism], G23 [other degenerative diseases of basal ganglia], G24 [dystonia], G25 [other extrapyramidal and movement disorders], R49.0 [dysphonia], and F44.4 [hysterical tremor]), which are sometimes conflated with ET. This resulted in a sample size of 41,200 patients. We then restricted the sample to patients with at least 1 prescription drug claim over the 2‐year period, yielding a final sample size of 36,839.

Measures

Prescription medication claims were employed to identify the use of any of the following agents that can be used in the treatment of ET: (1) propranolol, (2) primidone, (3) topiramate, (4) metoprolol, (5) atenolol, (6) gabapentin, (7) alprazolam, and (8) any benzodiazepines. Propranolol and primidone are first‐line agents, and some reviews include topiramate as well. ¹¹ , ¹² Apart from propranolol, metoprolol and atenolol are the commonly used β‐blockers, and gabapentin, primidone, topiramate, alprazolam, and benzodiazepine (American Hospital Formulary Service® class = 28:24.08) are the commonly used agents with γ‐aminobutyric acid (GABA)‐ergic properties. For each agent prescribed, standard measures of drug utilization were calculated, including the number of fills for each agent, percentage with only 1 fill of a given agent, the mean number of months with a fill (of 24), and the mean proportion of days covered (PDC) starting from the date of the first fill to the end of the study period. ²⁵ The PDC represents the number of days' supply from all fills for a given agent during the study period divided by the number of days from the initial fill to the end of the study period. A PDC of greater than 80% is typically considered “adequate adherence,” and the fraction of patients meeting this criterion was calculated.

In multivariate analyses, we created the following collapsed measures of prescription fills for brevity: indicator for any GABA‐ergic medications (gabapentin, topiramate, primidone, and any benzodiazepine) and indicator for either β‐blocker medication (metoprolol or atenolol). In those models, we included the following covariates: patient age, sex, race/ethnicity, Elixhauser comorbidity indicators, total number of comorbid conditions, and year fixed effects. The Elixhauser conditions include 31 conditions, which have been used to prepare a well‐validated index for risk adjustment and mortality prediction. ²⁶ , ²⁷ Because our sample is large, we used indicators for each of the 31 conditions as covariates in our regressions.

Statistical Analyses

We first calculated the descriptive statistics of patient attributes of the full sample of ET patients with any fills as well as the subset samples of patients using each of the aforementioned agents. Second, we examined the predictors of monthly use of each agent (ie, at least 1 fill) using logistic regression and adjusting for the aforementioned covariates. We presented adjusted odds ratios (aOR) post‐estimation. Stata SE, version 16, was used.

Results

Table 1 presents the sample descriptive statistics for the full analytic sample and for each subset of patients with at least 1 fill of each of the agents that can be prescribed for ET. Propranolol was the most frequently prescribed agent, with 27% of patients with a prescription medication claim having at least 1 fill. Gabapentin and primidone were the next 2 most common at 22% and 18%, respectively. Patients with a topiramate fill were aged 68.3 years, on average, lower than the full sample mean of 72.2 years.

TABLE 1.

Sample descriptive statistics

	Full sample	At least 1 fill between 2018 and 2019 of				Atenolol	Gabapentin	Alprazolam	Any benzodiazepines
		Propranolol	Primidone	Topiramate	Metoprolol
n	36,839	10,077	6754	1745	7178	1744	7928	3424	7490
Percentage of all ET patients	89	23	16	4	16	4	18	8	17
Percentage of ET patients with 1+ RX		27	18	4	16	4	22	8	17
Age—mean (SD)	72.17 (10.07)	70.61 (10.40)	73.43 (9.35)	68.28 (11.28)	74.33 (9.55)	73.46 (9.80)	71.73 (10.19)	70.54 (10.75)	71.09 (10.59)
Male (%)	45
White (%)	81	85	84	86	83	84	88	83	83%
Black (%)	7	7	9	9	9	8	8	7	10%
Hispanic (%)	7	7	7	7	8	8	7	10	9%
Asian (%)	2	2	2	1	2	2	0	1	2%
Payer = commercial (%)	18	22	13	20	13	15	11	20	14%
Payer = Medicare (%)	85	84	94	87	93	91	97	87	94%

Abbreviation: ET, essential tremor; RX, prescription.

Figure 1 shows the patterns of use among ET patients with at least 1 fill of each of these medications. On average, patients had ~7 to 8 months with a fill of each of the agents (panel A). Among patients with at least 1 fill of each agent, between 9% and 29% had no other fills of the agent (panel A, shown in orange). For example, 9% of patients with at least 1 fill of atenolol had only 1 fill of atenolol, and for any benzodiazepines, 29% of patients with at least 1 fill had only 1 fill. Panel B shows the mean PDC, which ranged from 36% to 80% (orange circles). The PDC was highest, on average, among those taking β‐blockers (77% and 80% of days covered), whereas those taking benzodiazepines were less likely to be taking those medications during the full period from the first fill date to the end of the study. The fraction of patients taking a medication who had more than 80% of their “days covered” (from the first fill to the end of the study) ranged from 20% for patients with any benzodiazepine fills to 67% of patients with any atenolol fills.

FIG. 1.

Prescription medication patterns of use among essential tremor (ET) patients. (A) Mean number of months with a fill among users and percentage with only 1 fill. (B) Mean proportion (%) of days covered. Sample sizes are presented in Table 1.

Table 2 presents the extent to which patients have fills for more than 1 of the main agents we study. Conditional on having at least 1 fill of the agent shown in the column headings, the rows represent the fraction of patients with at least 1 fill of each of the other agents. For example, among those with at least 1 fill of propranolol (first column), 28% also have at least 1 fill of primidone.

TABLE 2.

Percentage of users with fills of other prescriptions

	Primidone	Propranolol	Topiramate	Alprazolam	Metoprolol	Atenolol	Sotalol	Gabapentin	Any benzodiazepines
Primidone		0.28160664	0.08062843	0.01	0.22839781	0.05380169	0.00889284	0.2854602	0.22528531
Propranolol	0.18871673		0.05552245	0.01	0.07945967	0.02274533	0.00397298	0.21017084	0.20709178
Topiramate	0.31174785	0.32034384			0.2034384	0.05157593	0.00744986	0.34727794	0.27048711
Alprazolam	0.01	0.01	0.01		0.01	0.01	0.01	0.01	0.01
Metoprolol	0.21492329	0.11157601	0.04951185	0.01		0.02329149	0.01199442	0.27029289	0.23863319
Atenolol	0.20826162	0.13138267	0.05163511	0.01	0.09581182		0.00458979	0.2444062	0.26333907
Sotalol	0.25531915	0.17021277	0.05531915	0.01	0.36595745	0.03404255		0.27234043	0.25531915
Gabapentin	0.24315112	0.26713799	0.07650549	0.01	0.24466608	0.05378109	0.00807979		0.29642722
Any benzodiazepines	0.20345335	0.27907911	0.06317762	0.01	0.22901887	0.06143756	0.00803105	0.31428189

Table 3 presents the aORs for each of the predictors of medication use among ET patients using the collapsed medication categories (propranolol, primidone), any GABA‐ergic agent, any β‐blockers (metoprolol or atenolol), and any benzodiazepines. Predictors assessed included sociodemographic characteristics and comorbidities. aORs >1 indicate the predictor has a positive association with use of the agent. After adjusting for other covariates (as presented in Table 3), the odds of at least 1 fill of primidone or any β‐blocker increased as age increased. Black ET patients were more likely to have at least 1 fill of primidone or any GABA‐ergic agent and less likely to have any fills of propranolol relative to White ET patients. Asian patients with ET were more likely to have 1 of the β β‐blocker agent (metoprolol and atenolol) fills relative to White patients. Hispanic patients were more likely to have 1 or more fills of any benzodiazepine relative to White patients. There were also significant differences in sex, with men less likely to have a single fill of propranolol, primidone, any GABA‐ergic agent, or any β‐blockers and more likely to have any benzodiazepine fills.

TABLE 3.

Predictors of prescription medication use among ET patients, adjusted odds ratios reported (95% confidence intervals)

	Propranolol	Primidone	Any GABA‐ergic	Any β‐blockers	Any benzodiazepines
Age	1.00	1.02***	0.99*	1.01***	1.00
	(1)	(1.01, 1.01)	(0.99, 0.99)	(1.01, 1.01)	(0.99, 1)
Race (reference = White)
Asian	0.92	0.99	0.92	1.29*	0.86
	(0.76, 1.1)	(0.79, 1.23)	(0.71, 1.2)	(1.06, 1.56)	(0.63, 1.17)
Black	0.90*	1.16**	1.22**	1.00	1.01
	(0.81, 0.98)	(1.04, 1.29)	(1.07, 1.39)	(0.9, 1.1)	(0.87, 1.16)
Hispanic	0.93	0.90	1.12	0.92	1.30***
	(0.84, 1.02)	(0.8, 1)	(0.97, 1.28)	(0.83, 1.02)	(1.13, 1.5)
Missing	1.09	1.04	0.98	0.97	0.94
	(0.96, 1.23)	(0.9, 1.2)	(0.82, 1.16)	(0.84, 1.11)	(0.77, 1.14)
Male (reference = female)	0.95*	0.63***	0.93*	0.96	1.42***
	(0.9, 0.99)	(0.59, 0.66)	(0.86, 1)	(0.9, 1.01)	(1.3, 1.53)
Insurance (reference: commercial and Medicare)
Medicare only	0.86***	1.48***	1.52***	0.90*	0.73***
	(0.8, 0.93)	(1.33, 1.63)	(1.35, 1.71)	(0.82, 0.99)	(0.65, 0.82)
Commercial only	1.03	1.11	1.53	1.14	1.22
	(0.68, 1.57)	(0.64, 1.92)	(0.83, 2.81)	(0.71, 1.82)	(0.66, 2.24)
Total number of Elixhauser conditions	0.90*	0.91*	0.98	1.23***	0.91
	(0.83, 0.97)	(0.83, 0.99)	(0.87, 1.1)	(1.13, 1.33)	(0.8, 1.02)

Each logistic regression (column) also includes an indicator for each of the 31 Elixhauser comorbidities; see Figure 2 for those adjusted odds ratios. “Any GABA‐ergic” includes any fills of primidone, topiramate, gabapentin, or any benzodiazepines. “Any β‐blockers” includes either metoprolol or atenolol.

^* P < 0.05,

^** P < 0.01,

^*** P < 0.001.

Abbreviations: ET, essential tremor; GABA, γ‐aminobutyric acid.

There were no significant differences between ET patients covered by only Medicare and those covered by only commercial insurance; the confidence intervals of the aORs overlap in all cases (Table 3). However, those covered only by Medicare were less likely to have any fills of propranolol, any β‐blockers, or any benzodiazepines and more likely to have any fills of primidone and any‐GABA‐ergic agent relative to those covered by both Medicare and commercial insurance (Table 3).

Figure 2 shows the aORs by comorbidities. Patterns of use varied by the presence of comorbidities. Panel A shows the aORs of any use of the 2 first‐line agents used in treating ET where the presence of a comorbidity is significantly associated with the odds of use of the agent. Several comorbidities increased the odds of a patient's use of primidone, including obesity, diabetes (complicated or uncomplicated), chronic pulmonary diseases, other neurological disorders, peripheral vascular disorders, and pulmonary circulation disorders. Those with psychoses and substance use disorder were less likely to have a primidone fill. For propranolol, only 3 comorbidities increased the likelihood of use: alcohol use disorder, coagulopathy, and metastatic cancer. On the contrary, several comorbidities were associated with lower odds of use of propranolol: depression, substance use disorder, weight loss, chronic pulmonary disease, other neurological disorders, cardiac arrhythmias, and congestive heart failure. The prescription patterns for propranolol, in part, may stem from a desire to avoid known medication side effects such as reactive airway disease in patients with chronic pulmonary disease. Panel B shows the aORs for the most common second‐line agents for treating ET. Not surprisingly, ET patients who had any of the mental health comorbidities were more likely to have any benzodiazepine fill. In addition, those with cancer (solid tumor or metastatic) or liver disease were more likely to have a benzodiazepine fill. There were numerous comorbidities that decreased the odds of any fill for a β‐blocker, whereas any ET patient with hypertension or cardiac arrhythmias were more likely to have a fill for a β‐blocker. Those patients with hypertension, cardiac arrhythmias, or valvular disease were less likely to have a fill of any GABA‐ergic agent.

FIG. 2.

Association between comorbidities and prescription medication use among essential tremor (ET) patients, adjusted odds ratios (aOR). (A) First‐line agents. (B) Second‐line agents. Full set of covariates included in logistic regressions is presented in Table 3. Only aORs that are statistically significant at P < 0.05 are shown.

Discussion

To understand in depth medication use patterns among ET patients, we utilized a sample of medical claims data from CDM of more than 36,000 ET patients. We observed that the vast majority of ET patients, 89%, had at least 1 prescription drug claim over a 2‐year period, indicating that nearly 9 of 10 ET patients take a medication to treat their disease. This is remarkable given the broadly articulated concerns in the literature about both the efficacy and side effect profiles of current pharmacotherapeutic options for ET. ⁶ , ¹² , ¹⁴ It is important to note, however, that this high percentage of usage does not indicate that patients are necessarily satisfied with their treatment or that the treatments are efficacious. Our results suggest greater use of prescription medications than another recent study that focused on patients who had at least 1 claim per year with an ET diagnosis. Without this sample inclusion criteria, we might expect that our sample includes more patients with milder ET (eg, patient shows up for an unrelated medical visit, and provider does not flag ET diagnosis).

Propranolol was the most frequently prescribed agent: 27% of ET patients with a prescription medication claim had at least 1 fill of this medication. Gabapentin and primidone were the next 2 most common at 22% and 18%, respectively. However, both propranolol and gabapentin are commonly used to treat other medical conditions, which could have resulted in higher proportions. One remarkable feature of these data is that for each of these 3 agents, only a fraction (approximately one quarter to one fifth) of patients were taking that medication.

For the majority, although not all ET medications, most patients with at least 1 fill had more than 1 fill of that agent (Fig. 1A). For propranolol, primidone, and topiramate, which are often cited as first‐line agents, 12%, 18%, and 20%, respectively, had only 1 fill of that agent, suggesting a rapid trial followed by discontinuation, or alternatively, a filled prescription but a decision on the part of the patient not to take the medication. For primidone, with an acute reaction in many patients, and topiramate, with side effects (eg, paresthesia) in a high percentage of patients, the proportion of patients with only 1 fill (18% and 20%, respectively) was approximately double that of primidone and on the order of 1 in every 5 patients. One caveat is that propranolol is often prescribed on an as‐needed schedule, rather than daily, likely resulting in greater tolerance.

Adherence to medications for these 3 first‐line agents ranged from 52% to 61% (Fig. 1B), indicating that in more than one half of patients, the PDC was less than 80%. Adherence was numerically higher for those taking β‐blockers and lower for those taking topiramate, gabapentin, or any of the benzodiazepines (Fig. 1B).

Studies have demonstrated that adherence rates are typically higher in patients with acute rather than chronic conditions and that patients with chronic illnesses take only approximately 50% of medications prescribed for these conditions. ²⁸ A study of 151 ET cases enrolled in a clinical‐epidemiological study showed that one‐third of cases reported that they sometimes forgot to take their medication, and one‐fifth reported missed doses within the past week. ²⁹ Nearly 1 in 4 reported that there were whole days in the past 2 weeks in which they had not taken their medication. ²⁹

We also observed that a high percentage of ET patients have fills for more than 1 of the main agents we studied (Table 2). For propranolol, a median of 28% of users had fills of each of the other prescriptions we studied; for primidone, the median was 21% (Table 2).

We found several patient‐level factors that were associated with the use of certain agents, including age, race, sex, insurance, and number of Elixhauser conditions. For example, ET patients who were female; those not covered by Medicare; or those with an alcohol use disorder diagnosis, coagulopathy, or metastatic cancer were more likely to use propranolol. ET patients who were older, Black, and female; were covered by Medicare; and had a diagnosis of obesity, diabetes, chronic pulmonary disease, other neurological disorders, peripheral vascular disorders, and pulmonary circulation disorders were more likely to use primidone compared to their counterparts. Although these findings are of interest, in many instances, the underlying mechanisms for the association are not totally clear; additional studies are needed to corroborate these results. Furthermore, the explanation for associations between patient‐level factors and classes of agents is likely to be even more complex and difficult to disentangle. A consistent finding across agents was that the greater the number of comorbidities, the lower the odds of medication use (Table 3). Cardiac arrhythmias, in particular, were associated with a reduced usage of propranolol, primidone, and any GABA‐ergic agents (Fig. 2).

Our results of the key agents (propranolol and primidone) used to treat ET were similar to those found in the study of Vetterick et al., ¹⁹ except that more patients in our sample had at least 1 prescription fill. Our study sample inclusion criteria and definition of ET differed from those of the earlier study. We examined medication adherence as measured by the PDC based on the index date (first fill observed in claims), whereas the earlier study ¹⁹ examined discontinuation rates (typically defined by whether patients have a medication gap of ≥60 days). We found similar adherence rates, however, with about 60% of those taking propranolol or primidone having 80% or more of days covered. Additional studies such as these 2 foundational studies will provide the most comprehensive pharmacological data on this highly prevalent disease.

This study was not without certain limitations. First, although our sample was derived from a large commercial claims database, we were unable to draw conclusions about milder ET patients who do not seek care or medication; inclusion of such patients would likely have yielded lower estimates of medication use in this disease. Similarly, patterns and correlates of medication use may be different for ET patients without health insurance. Second, we were not able to generalize to ET patients in other countries. Comparisons across countries are problematic given the significant differences in health care systems, utilization, and costs. ³⁰ , ³¹ However, a recent study from the United Kingdom and France after incident ET cases reports similar rates of use of propranolol (about 20% within 6 months of diagnosis) but much lower rates of primidone (<5%). ³² Third, our measure of adherence is based on the first fill of agents as observed in our data, which may be measured with error. Fourth, our analysis of predictors of use should not be interpreted as causal. Finally, although we adjust for a set of 31 comorbidities, the diagnosis for which each medication was prescribed was not determined from these data. Thus, for agents that have multiple clinical indications, we were unable to attribute their use entirely to ET. For example, it is certainly likely that some proportion of the β‐blocker use in our ET cases could have been for other medical conditions, particularly hypertension. With that being said, with the advent of newer classes of antihypertensive agents, the use of β‐blockers has declined and, in some recent guidelines, is no longer included as a first‐line agent. ³³ , ³⁴ Nevertheless, the caveat should be added to the interpretation of our findings. With all this being noted, our study leverages a large administrative dataset that allows us to document patterns of prescription medication and examine predictors of the use of different agents.

In summary, these data illustrate a high need for medication in the ET population. At the same time, no single agent was utilized by more than one quarter of ET patients, most prescriptions were filled only once, adherence was low, and use of multiple agents was common. For such a common disease, the pharmacotherapeutic landscape is impoverished.

Author Roles

(1) Research project: A. Conception, B. Organization, C. Execution; (2) Statistical analysis: A. Design, B. Execution, C. Review and critique; (3) Manuscript: A. Writing of the first draft, B. Review and critique.

K.A.K.: 1B, 1C, 2A, 2B, 3A, 3B

E.D.L.: 1A, 1B, 1C, 2C, 3A, 3B

Disclosures

Ethical Compliance Statement: The University of Texas Southwestern Medical Center internal review board (IRB) deemed this secondary, de‐identified study as exempt (STU‐2022‐0613). As the data are de‐identified, informed consent was not possible. We confirm that we have read the journal's position on issues involved in ethical publication and affirm that this work is consistent with those guidelines.

Funding Sources and Conflict of Interest: “No specific funding was received for this work” and/or “The authors declare that there are no conflicts of interest relevant to this work.”

Financial Disclosures for the Previous 12 Months: Document all funding sources, not related to the current research in the article, from each author. The authors declare that there are no additional disclosures to report.

Data Availability Statement

The data that support the findings of this study are available from Optum. Restrictions apply to the availability of these data, which were used under license for this study.