Switching medication products during the treatment of psychiatric illness

Abstract

Background. The common practice of switching between branded (reference) medications and their corresponding generic products, between generic products, or even from a generic product to a branded medication during the treatment of central nervous system (CNS) disorders may compromise efficacy and/or tolerability. Methods. We assessed the published literature from March 1, 2010 through June 30, 2017 via PubMed using the MeSH term ‘generics, drugs’ alone and in combination with class-specific terms (e.g., ‘anticonvulsants’, ‘mood stabilisers’), for studies detailing outcomes following product switches. Results. Although some studies comparing the initiation of reference versus generic drugs suggest equivalence between products, several studies detailing a switch between reference and generic products describe reductions in efficacy, reduced medication adherence and persistence, and increased overall health care resource utilization and costs associated with generic substitution. Conclusion. When product switches are considered, they should only proceed with the full knowledge of both patient and provider.

Keywords:

• Drugs
• generic
• therapeutic equivalency
• pharmacokinetics
• drug substitution

Introduction

Innovative drug development (development of drugs for new therapeutic targets and/or using a novel mechanism of action) is associated with very high costs and extended timelines (DiMasi, Grabowski, & Hansen, 2016; Nixon et al. 2017). This is particularly true for novel drugs targeting central nervous system (CNS) indications where the likelihood that a new chemical entity will proceed successfully through clinical trials to approval has been estimated at 8%, with many drugs failing in late-stage trials after significant investment (Brady & Insel, 2012; Riordan & Cutler, 2012).

Once patent or data protection for the innovator/branded (reference) product has expired, generic versions enter the market. Compared with the approval process for reference products, the introduction of a generic product entails substantially reduced investment. The generic product is required to be pharmaceutically equivalent (i.e., same active ingredients, dosage forms, route of administration, strength or concentration) and to demonstrate bioequivalence with the reference product, whereas the efficacy and safety of the reference product typically must be established in preclinical and clinical trials (Borgheini 2003; Seoane-Vazquez, Rodriguez-Monguio, & Hansen, 2016). Although generic versions generally must have the same active pharmaceutical ingredient (API) and the same dosage form as the reference product (although some generics have introduced different APIs, such as a different salt or additional dosage forms [e.g., clozapine 200-mg tablet form]), they may differ with respect to excipients and also may possess other alterations (such as the salt conjugated to the API) that potentially and critically affect drug stability, absorption, and toxicity (Borgheini 2003).

A key advantage that generic versions have over reference products is reduced cost to patients and payers (Borgheini 2003). In the United States, the cost reduction resulting from the initial generic entry of a given reference product can be limited, because the initial generic entry is afforded 6 months of exclusivity, but subsequent entries following that 6-month time-frame lead to far more substantial cost reductions (Druss, Marcus, Olfson, & Pincus, 2004). For example, the first generic version of fluoxetine (Prozac^®) was initially priced 12% below the reference product ($1.91 vs. $2.25 per dose), and the price remained steady for the 6-month exclusivity period. However, during the following 12 months, as new generic versions were introduced, the price for the generic dropped to $0.32 (vs. $2.40 for the reference product), and the percentage of Prozac prescriptions filled with generic fluoxetine rose to nearly 90% (Druss et al., 2004).

The operational definition of bioequivalence varies somewhat between the United States, Canada, and the European Union (Food and Drug Administration 2004; European Medicines Agency 2010; Health Canada 2012a). The US Food and Drug Administration (FDA) defines ‘bioequivalence’ as the absence of a significant difference between 2 or more products in the rate and extent of absorption of the API at the site of drug action when administered at the same (molar) dose under similar conditions (Food and Drug Administration 2003).The evaluation of bioequivalence rests on two key pharmacokinetic (PK) parameters: maximum plasma concentration (C_max) and area under the curve (AUC) of plasma drug concentration over time after administration, which describe the rate and extent of drug absorption (Meyer 2001; Food and Drug Administration 2004).

In the United States and Europe, a test product (in this context, a generic) and reference product (the ‘branded’ and/or innovator product) are considered bioequivalent if the 90% confidence intervals (CIs) of their geometric mean ratios (test/reference) for the area under the plasma concentration-time curve from time zero to the last measurable concentration (AUC_0-t) and C_max falls between 80% and 125% (Food and Drug Administration 2004; European Medicines Agency 2010). In Canada, a finding of bioequivalence requires satisfaction of the same criteria for AUC values, but for C_max, only the point estimate of the test product must fall between 80% and 125% of that for the reference product for most drugs (Health Canada 2012a); 90% CIs for the geometric mean C_max must fall between 80% and 125% only for critical dose drugs, which have a narrow therapeutic window. In general, if bioequivalence criteria are met, manufacturers of generic products are not required to conduct additional safety and efficacy studies to demonstrate clinical equivalence (Blier 2009; Health Canada 2012b; Food and Drug Administration 2017). Table 1 summarises bioequivalence requirements for the United States, Canada, and Europe (Galgatte, Jamdade, Aute, & Chaudhari, 2014).

Table 1. Bioequivalence study parameters in Canada, United States, and Europe.

Comparative assessment of study parameters^a
Parameter	Canada	United States	Europe
Study design	Nonreplicated, randomized, crossover	Nonreplicated, randomized, crossover	Nonreplicated, randomized, crossover
Fasting/fed	Fasting	Fasting and fed	Fasting
Volunteers	>80% power of acceptance criteria	Sufficient to achieve adequate power	>12 (min 80% power of acceptance criteria)
Study dose	Made by manufacturer	Made by manufacturer	Made by manufacturer
Test Reference	Canadian reference product	Reference listed drug in USA	European reference product
Sampling points	12–18 samples per subject/dose	12–18 samples (more collected at Tmax to be continued up to 3+ half-lives	At least 2 samples before expected Tmax, 3–4 terminal log-linear phase
Analytical method validation parameters	Accuracy	Accuracy	Accuracy
	LOQ	Calibration curve	Detection
	Precision	LLOQ	Limit of quantitation
	Recovery	Precision	Intermediate precision specificity
	Specificity	Reproducibility	Linearity range
	Stability	Selectivity	Precision
	Standard curves	Sensitivity	Repeatability
		Stability
Moieties to be measured in plasma	Active drug	Active drug	Active drug
	Metabolites if applicable	Metabolites if applicable	Metabolites if applicable
PK parameters	Cmax	Cmax	Cmax
	Tmax	Tmax	Tmax
	AUC0-t and AUC0-∞	AUC0-t and AUC0-∞	AUC0-t and AUC0-∞
	t1/2	t1/2	t1/2
	γz	γz	γz
Criteria for bioequivalence	90% CI 80%–125% for Cmax, AUC0-t, AUC0-∞	90% CI 80%–125% for Cmax, AUC0-t, AUC0-∞	90% CI 80%–125% for Cmax, AUC0-t, AUC0-∞ (for highly variable drugs 75%–133%)
GCP requirements	ICH GCP guidelines	ICH GCP guidelines	ICH GCP guidelines

Data from Nation & Sansom, 1994; Borgheini 2003; Galgatte, Jamdade, and Aute, & Chaudhari, 2013. Reprinted from Galgatte et al., 2014.

AUC_0-t: denotes area under the plasma concentration-time curve from time zero to the last measurable concentration; AUC_0-∞: denotes area under the concentration curve from zero to infinity; C_max: denotes maximum serum concentration; GCP: denotes good clinical practice; ICH: denotes International Council for Harmonisation; LOQ: denotes limit of quantitation; LLOQ: denotes lower limit of quantitation; T_max: denotes time to reach maximal concentration; t_1/2: denotes terminal elimination half-life; γz: denotes terminal or elimination rate constant.

^aAll follow International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH) Good Clinical Practice (GCP) guidelines, with some minor variation.

Bioequivalence criteria allow for variability in C_max and AUC ranging from −20% to +25% between generic and reference products (Borgheini 2003). However, additional variability is introduced throughout the lifecycle of both generic and reference products. Potential sources for this variability include, for instance, increasing batch sizes, addition or change in manufacturing sites, and the production of additional formulations. (Meyer 2001). Although changes to a product over its lifecycle must pass manufacturing and regulatory hurdles (e.g., new batches or formulations can be rejected if the variability is outside acceptable limits), the cumulative effect of these variations may potentially result in larger than expected differences in generic compared with reference products, potentially impacting therapeutic equivalence. Furthermore, the potential variability between different generic versions of a single reference product can be even greater than the variability between a generic and its reference product (Meyer 2001). Consider, for example, two generic products, one that yields plasma drug levels 20% greater than the reference drug, and a second that yields plasma levels 15% lower than the reference (both within the bioequivalence acceptance range). Both generic products demonstrate bioequivalence with the reference product. However, switching from the first of these generic products to the second without possible dose adjustments could result in a clinically significant reduction in plasma drug levels and a potential loss of efficacy. A switch from the second generic product to the first, in contrast, could increase plasma drug levels, possibly reducing tolerability. In this scenario, differences in PK and potential therapeutic effects could grow larger over the lifecycle of the drug.

Several reports have questioned whether bioequivalence of a generic product to the reference product translates to therapeutic equivalence, at least for some generic medications and in some patient populations (Yamada & Welty, 2011; Tothfalusi & Endrenyi, 2013; Unterecker et al. 2014; Hsu, Lee, & Wang, 2017). This scepticism may be due in part to certain characteristics of bioequivalence studies. First, bioequivalence studies are typically conducted in small groups of healthy volunteers (who largely have been young, male, and Caucasian adults with no chronic medical issues) and are frequently single-dose comparisons. Additionally, bioequivalence studies are not typically conducted in patients for which the medication is intended, unless safety issues are of concern (Food and Drug Administration 2003; Health Canada 2012b). Moreover, even if multiple bioequivalence studies were conducted on a specific generic drug, and only one established bioequivalence, it used to be the case that it was acceptable for the generic drug sponsor to submit the positive study only in support of approval, eschewing the others (Meyer 2001; Blier 2009).

Given both the limitations of bioequivalence studies and the potential for variation throughout the lifecycle of a drug product, it is not clear that generic products are in each case therapeutically equivalent to their reference drug. Indeed, the US FDA has investigated suspected differences in efficacy and safety between generic and reference products based on reports received in the Adverse Event Reporting System database (Seoane-Vazquez et al., 2016). After examining adverse event (AE) reports and bioequivalence data used for approval, the FDA may find no clinically significant differences between products, but in some cases reassessment of bioequivalence and/or withdrawal of a product has been required (e.g., a generic bupropion product (Woodcock, Khan, & Yu, 2012)).

Concerns about the therapeutic equivalence of reference versus generic products are illustrated by experience with antiepileptic drugs (AEDs). A number of clinical reports and studies have documented increases in breakthrough seizures or in overall seizure frequency, as well as substantial changes in PK parameters and serum drug levels, following a switch from a reference AED to a generic version (MacDonald 1987; Welty, Pickering, Hale, & Arazi, 1992; Makus & McCormick, 2007; Berg, Gross, Tomaszewski, Zingaro, & Haskins, 2008; Nielsen, Dahl, Tommerup, & Wolf, 2008). Such clinical deterioration has been documented across a broad range of AEDs; in a number of cases, optimal seizure control was restored following reversion to the reference product (MacDonald 1987; Welty et al., 1992; Makus & McCormick, 2007; Berg et al., 2008). These findings are relevant to the treatment of bipolar disorder as AEDs are used in the standard of care in such patients (Fountoulakis et al. 2017).

Given the regulatory stance that bioequivalence does generally translate to therapeutic equivalence, generic products may be freely substituted for the reference product. It is common, for example, for switches between products (reference and generic or generic and generic) to be made immediately following a hospital discharge because hospital and commercial pharmacies do not necessarily use the same products. Clinicians may be unaware that this is a concern and that such switches may contribute to a high rate of relapse and readmission in the weeks after discharge (with treatment noncompliance an important factor) (DelBello, Hanseman, Adler, Fleck, & Strakowski, 2007; Schennach et al. 2012). In Canada, pharmacists may substitute among interchangeable generic and reference products on the formularies when filling prescriptions (Canadian Generic Drug Sector Study, 2007). They are also free to switch between different generic versions of the same medication, which, as discussed earlier, can potentially differ even more than do the generic medications and their reference drugs. Within the 2012 US retail pharmacy sector, generic products represented 84% of all prescriptions filled, and were dispensed 95% of the time when a generic product was available (Bartholow 2013). In light of the high rates of substitution between generic and reference products, it is imperative to underscore that therapeutic equivalence is best established via clinical trials involving an appropriately large sample of the target population for which the drug is intended. Without this step, automatic unconstrained switching to a generic drug from a reference drug, between different generic drug products, or even from a generic product back to the reference version may increase the risk of clinical deterioration and/or tolerability issues (Yamada & Welty, 2011; Tothfalusi & Endrenyi, 2013; Unterecker et al., 2014; Hsu et al., 2017).

Automatic generic substitution is of particular concern in the treatment of mental health disorders. The World Mental Health Survey Initiative, conducted by the World Health Organization, has found that the lifetime prevalence estimate for any Diagnostic and Statistical Manual of Mental Disorders mental disorder assessed in the Composite International Diagnostic Interview ranges from 12.0% (in Nigeria) to 47.4% (in the United States), with an interquartile range (25–75%) across all surveys of 18.1–36.1% (Kessler et al. 2009). The prevalence and unpredictability of mental illness, combined with the inherent variabilities of generic substitutions, even given pharmacokinetic bioequivalence with the reference, suggests that unintended harm may result with such substitution (Borgheini 2003). In 2011, a report reviewed and addressed the consequences of switching from innovator/branded (reference) to generic psychotropic products, assessing the available literature published from January 1, 1974 through March 1, 2010 (Desmarais, Beauclair, & Margolese, 2011). This report provides an update to that review, based on literature published between March 1, 2010 and June 30, 2017.

Methods

A PubMed literature search was conducted for articles published between March 1, 2010 and June 30, 2017 to identify relevant sources using a similar search strategy to that described in the previous review (Desmarais et al., 2011).The search strategy involved combining the MeSH term ‘generic, drugs’ with terms including ‘anticonvulsants’, ‘mood stabilisers’, ‘lithium’, ‘antidepressants’, ‘antipsychotics’, ‘stimulants’, ‘anxiolytics’ and ‘benzodiazepines’. The bibliographies of relevant references were reviewed to identify additional articles. We considered articles published in English, French, or Spanish based on their inclusion of topics including clinical equivalence of generic and brand-name medications, generic substitution, or the effectiveness, tolerability, compliance, or economics observed with generics. Because of the volume of available literature on anticonvulsant drugs, we concentrated our search on those commonly used in psychiatry, including carbamazepine, valproate, lamotrigine, gabapentin, and topiramate. We excluded PK studies conducted with healthy volunteers unless they contained important data relevant to the clinical population; in contrast to the previous literature review, we did include surveys, as we were interested in evaluating stakeholder opinion across various markets (Desmarais et al., 2011).

Generic substitution in general

Patient perspectives

A review of articles evaluating patient adherence and attitudes toward generic substitution (Weissenfeld, Stock, Lungen, & Gerber, 2010) in the published literature between March 25, 1989 and March 25, 2009 across a range of medications, found that about two thirds of patients were generally accepting of generic substitution. Factors influencing acceptance included financial incentives and being informed about the switch by their physician or pharmacist (Weissenfeld et al., 2010).

A postal survey (Heikkila, Mantyselka, & Ahonen, 2011) conducted in Finland (N = 1844 respondents [62% response rate]) following the adoption of mandatory generic substitution showed that 70.9% believed that generic substitution was a good law reform, but a substantial minority (26.9%) remained unsure. Most (80.9%) believed that generic medications were effective and most (84.9%) thought generic substitution did not pose a risk to safety (Heikkila et al., 2011).

A survey of renal transplant patients (N = 163) in the United Kingdom (Al Ameri, Whittaker, Tucker, Yaqoob, & Johnston, 2011) found that 84% believed that generic medications were not equivalent or were equivalent only sometimes in comparison to the reference product. About two thirds were dissatisfied or uncertain about their satisfaction regarding generics; 55% reported noticeable differences in effectiveness or side effects between generics and their reference products, and of these, 76% acknowledged problems in adapting to these differences (Al Ameri et al., 2011).

An assessment of attitudes regarding generic substitution among Japanese patients (N = 1215) (Kobayashi, Karigome, Sakurada, Satoh, & Ueda, 2011) found that among the 18.4% of patients who had experienced a generic substitution, their acceptance of the substitution was facilitated by a recommendation from a physician or a pharmacist (cited by 48.6% and 33.1%, respectively). Experience with a previous substitution was associated with an increased willingness for generic substitution (odds ratio [OR] vs patients with no previous experience 2.93; 95% CI =1.93–4.44) (Kobayashi et al., 2011).

Face-to-face questionnaire-based interviews with patients (N = 101) conducted in Turkey (Toklu et al. 2012) found that only 24% of patients believed that generic products do not differ from the original innovator drug; 26% of patients were willing to accept a generic substitute if suggested by a prescriber, but only 10% would accept a generic substitution suggested by a pharmacist (Toklu et al., 2012).

A survey of Jordanian patients (N = 400; 80% response rate) with chronic medical conditions (El-Dahiyat & Kayyali, 2013) revealed that 78% of respondents reported that they ‘did not mind’ the substitution of a lower priced generic for a reference product . However, an equivalent number (78%) also agreed that the choice between a generic and reference product should belong to the patient (El-Dahiyat & Kayyali, 2013).

A systematic review (Colgan et al. 2015) of 52 studies addressing perceptions of generics among patients and providers found that 35.6% (95% CI =34.83–36.37) and 25.1% (24.23–26.01) of patients viewed generics as less effective or of inferior quality, respectively, compared with reference products. Although 34.0% (33.16–34.90) viewed generic substitution negatively, only 18.8% (17.76–19.80) and 18.0% (16.96–19.04), respectively, believed generics were more likely to cause AEs or were less safe than innovative drugs (Colgan et al., 2015).

An interview-based study of the general public in Poland (N = 1000) conducted in 2013 (Drozdowska & Hermanowski, 2016b) found that the strongest predictors of choosing generic over reference products were patient sensitivity to physician opinion (p < .001), generic brand (p < .001), and household income (p = .011) (Drozdowska & Hermanowski, 2016b).

Provider perspectives

In the Turkish study by Toklu and colleagues mentioned above (Toklu et al., 2012), 32% and 31% of prescribers and pharmacists, respectively, believe that generics do not differ from reference products. However, 82% of prescribers and 40% of pharmacists expressed uncertainty about the bioequivalence of generics and innovative drugs. A large majority of prescribers (92%) and pharmacists (82%) believe that cost is the most important factor driving generic substitution (Toklu et al., 2012).

In interview-based surveys of general practitioners (GPs; n = 34) and pharmacists (n = 44) conducted in Ireland (Dunne, Shannon, Dunne, & Cullen, 2014) before the June 2013 introduction of generic substitution and reference pricing, both groups (GPs, 91.2%; pharmacists, 97.7%) strongly believed that generics and reference products were of similar quality. GPs were slightly less supportive of generic equivalence; significantly more GPs than pharmacists believed they do not work as well as innovative drugs (11.8% vs. 2.3%; p = .03). In addition, 32/34 (94.1%) of GPs and 39/44 (88.6%) of pharmacists had received patient complaints about poorer efficacy and/or increased side effects with generics (Dunne et al., 2014).

A systematic review of 24 studies published between 2002 and 2012 reporting on physician and pharmacist attitudes and experiences with generics in multiple countries (Toverud, Hartmann, & Hakonsen, 2015) noted that pharmacists had a better understanding of bioequivalence compared with physicians. Respondents from countries with less mature health systems were generally less informed about bioequivalence, control routines, and manufacturer standards; were more likely to distrust quality and efficacy of generic products; and were more likely to consider patient socioeconomic status in prescribing decisions. Respondents from mature health systems trusted generic quality and prescribed them regardless of patient socioeconomic status. Both practitioner groups, regardless of country, expressed caution about generic substitution for certain drug types, including psychotropics and those with a narrow therapeutic window (Toverud et al., 2015).

In the Colgan et al. systematic review of 52 studies (Colgan et al., 2015), generics were regarded as less effective than their reference counterparts by 28.7% of physicians and by 23.6% of pharmacists; significantly more physicians than pharmacists believed generics were associated with more side effects (24.4% vs. 17.6%; p < .0001). Pharmacists were more likely than physicians to believe that generic quality was inferior to that of reference products (33.4% vs. 28.0%; p = .0006). A substantial fraction of both groups (physicians, 28.5%; pharmacists, 25.4%) believed generics were less safe than reference products. In general, more physicians than pharmacists held negative perceptions about generic substitution (24.1% vs. 11.0%; p < .0001) (Colgan et al., 2015).

A questionnaire-based survey of Nigerian physicians (N = 171) working in tertiary health care facilities (Fadare et al. 2016) found that generics were recognised as being equal to the reference medicines by only 44%. The mean (SD) knowledge score regarding generics was 5.3 (1.8) out of a possible 9; and 36.6%, 36.1%, and 27.2% of the respondents were rated as having poor, average, or good knowledge, respectively, about generics. Most respondents (71.7%) did not feel that generics were of lower quality than reference products. However, there was broad distrust of locally-produced generics, with 47.1% feeling they were of lower quality than those produced by multinational corporations, and only 15.7% feeling that local manufacturers were reputable suppliers; 82.7% expressed concern about therapeutic failures associated with locally-produced generics. Perhaps because of these concerns, 63.9% would not be comfortable with generic substitution by pharmacists (Fadare et al., 2016).

In a nationwide telephone survey of community pharmacists (N = 802) in Poland regarding attitudes toward generic medications (Drozdowska & Hermanowski, 2016a), 67% believed that the efficacy of generics was no worse than that of the reference products while 30% felt that generics may sometimes be less effective. About 41% stated they always informed customers regarding availability of a generic substitute, and 46% said they often did; length of experience was a factor, with inexperienced pharmacists (≤ 5 years’ experience) less likely to inform customers about generic alternatives than those with 11 years or more of experience (Drozdowska & Hermanowski, 2016a).

Straka and colleagues (Straka, Keohane, & Liu, 2017) conduced a systematic review that assessed the clinical and economic impact of switching from reference drugs to generics from peer-reviewed literature published from 2003 to 2013. Several studies documented negative consequences of generic substitution, including poorer adherence; a number also found that switching resulted in poorer clinical outcomes and switched back to the reference product. Although psychotropic medications (including AEDs) appeared to be the most commonly associated with negative consequences of switching, medications used in other clinical areas were also affected. In addition, some studies documented increased total health care costs associated with switching, in which lower drug costs were more than countered by increased physician and hospitalisation costs. The authors recommend a more complete consideration and accounting of the total cost impact of generic substitution (Straka et al., 2017).

Generic substitution of specific psychotropic agents

Medications used to treat seizures or bipolar disorder

In a prospective multicenter study conducted in Italy, Vari et al. (Vari et al. 2016) evaluated outcomes following a switch from reference levetiracetam to a generic version; about half of the included patients (n = 59) were taking levetiracetam monotherapy at baseline. Throughout the 6-month follow-up period, there was no evidence of increases in seizure frequency or in AEs compared with preswitch values; only 2 patients (3.4%) switched back to the reference version (Vari et al., 2016).

Using claims data from the PharMetrics Database, Labiner and others (Labiner et al. 2010) evaluated medical resource use among patients with epilepsy (N = 33,625) following generic substitution of five commonly used AEDs (carbamazepine, gabapentin, phenytoin, primidone, or zonisamide). During periods of generic AED use, there were increases in all-cause drug use (adjusted incidence rate ratio [IRR] = 1.13 [95% CI =1.13–1.14]), hospitalisations (IRR =1.31 [1.26–1.35]), and outpatient service utilisation (IRR =1.17 [1.17–1.18]), mean length of hospital stay (2.3 vs. 1.6 days; IRR =1.33 [1.32–1.35]), and risk of epilepsy-related injury (adjusted hazard ratio [HR] = 1.23 [1.07–1.41]), compared with periods of reference product use. Epilepsy-specific health care use demonstrated similar patterns (Labiner et al., 2010).

Erickson and colleagues (Erickson et al. 2011) conducted a retrospective matched-cohort study using a claims database to compare 6-month outcomes in patients (n = 2139) who switched to generic phenytoin, lamotrigine, or divalproex with patients (n = 2139) who remained on the reference product. Changes in AED use (discontinuations, dosage changes, or new AED initiation) were twice as likely for patients switched to generic phenytoin than those remaining on the reference product. The event rate ratio for all-cause hospitalisations and emergency department (ED) visits (comparing generic users with nonswitchers) was 0.96 (95% CI =0.80–1.16) for phenytoin, 0.97 (0.80–1.17) for lamotrigine, and 0.84 (0.66–1.06) for divalproex, indicating no greater incidence of those events with a switch to any of the three generic products (Erickson et al., 2011). However, patients switched to generic phenytoin had an IRR of 1.85 (95% CI =1.50–2.29) for changes in AED use. Indeed, the authors point out that, in the context of brand to generic substitution, changes in dose or the addition of an adjunctive AED may be more sensitive for detecting adverse outcomes than are number of ED visits and hospitalisations (Erickson et al., 2011).

The risk of seizure-related hospitalisation was evaluated (Polard, Nowak, Happe, Biraben, & Oger, 2015) in patients (N = 8379) who experienced a switch from a reference to generic product of 1 of 6 targeted drugs (carbamazepine, lamotrigine, levetiracetam, oxcarbazepine, topiramate, or valproic acid). In this case-crossover study using the French National Health Insurance Database, the risk of seizure-related hospitalisations was similar between those who were switched and those who remained on the reference product (OR 0.97 [95% CI =0.86–1.10]), suggesting therapeutic equivalence (Polard et al., 2015).

In a propensity-matched retrospective database study, Gagne et al. (Gagne et al. 2015) compared outcomes in epilepsy patients who initiated treatment with one of five reference AEDs (clonazepam, gabapentin, oxcarbazepine, phenytoin, or zonisamide; n = 1454) or with their generic counterparts (n = 18,306). Patients who initiated with a generic version demonstrated a lower rate of seizure-related hospitalisations and ED visits (HR vs reference products, 0.53 [95% CI =0.30–0.96]), in addition to longer treatment persistence before experiencing a gap (reference: mean [SD] = 124.2 [125.8] days; generic: 137.9 [148.6] days; p = .01) (Gagne et al., 2015).

Conclusions regarding product switches based on retrospective cohort analyses such as these are limited, however, in that plasma AED levels were not assessed. Inconsistencies in study findings therefore may be due in part to differences in reference and generic drug PK in the populations studied. In a series of PK modelling calculations, Tothfalusi and Endrenyi (Tothfalusi & Endrenyi, 2013) showed that two hypothetical generic carbamazepine products, differing from a reference version only with respect to their absorption rate constants (e.g., 0.94 h⁻¹ vs. 3.6 h⁻¹), would meet US bioequivalence standards with respect to C_max after a single administration in healthy volunteers (predicted geometric mean C_max ratios [95% CI]: 105% [100.9–109.1] and 110% [105.8–114.3], respectively). However, because carbamazepine is a strong inducer of hepatic cytochrome P450 enzymes, under steady-state (fully induced) conditions, C_max values for the generic product with a higher absorption rate constant are predicted to fall outside of bioequivalence range of 80–125% (predicted geometric mean C_max ratio [95% CI]: 128.9% [123.9–134.0]) whereas the generic product with a lower absorption rate constant would continue to meet bioequivalence criteria (110.1% [108.6–113.4]). The same would hold true in target populations already receiving carbamazepine (or other enzyme-inducing drug), possibly resulting in a lack of bioequivalence and necessitating dosage adjustments (Tothfalusi & Endrenyi, 2013).

Contin et al. (Contin, Alberghini, Candela, Benini, & Riva, 2016) assessed whether intrapatient PK variability (changes in plasma AED concentration) was greater in patients switched from one of three reference AEDs (lamotrigine, levetiracetam, or topiramate; n = 93) than in those who remained on the reference product (n = 400), using prospectively obtained and historical blood samples. The degree of intrapatient variability in drug level over time was high, with similar or smaller proportions of patients who switched demonstrating significant (>20%) variability (lamotrigine, 22%; levetiracetam, 44%; topiramate, 6%) compared with those who remained on reference products (lamotrigine, 33%; levetiracetam, 38%; topiramate, 41%). The authors concluded that AED plasma level changes after switching may not be related to the switch, but instead may reflect normal variation (Contin et al., 2016), although switch-related differences in plasma AED levels within individuals cannot be excluded.

Markoula et al. (Markoula et al. 2017) conducted a single-center prospective study in Greece to assess PK and clinical outcomes following a switch from reference to generic levetiracetam in patients with epilepsy (N = 12), comparing 4-week preswitch and postswitch periods. Mean [SD] baseline and 4-week postswitch PK parameters were similar (AUC: 288.4 [86.3] vs. 319.2 [104.7] mg/L·hr; C_max: 37.8 [10.4] vs. 41.6 [12.3] mg/L). Mean [SD] seizure frequency was similar between the preswitch and postswitch periods (1.2 [1.5] vs. 1.3 [1.6] per week), as was the incidence and severity of AEs (Markoula et al., 2017).

A systematic review conducted by Yamada and Welty (Yamada & Welty, 2011) that evaluated generic AED substitution outcomes from retrospective (n = 7) and prospective (n = 6) studies found inconsistencies. Retrospective studies generally demonstrated poorer outcomes with generics including increased medical resource utilization and higher switchback rates, but prospective studies (which possess higher quality evidence) generally suggested there should be no differences with respect to medical outcome or PK with generic substitution (with respect to outcomes and PK parameters). This finding however is with the caveat that 3 of these prospective studies were conducted with carbamazepine, bringing into question the applicability of these results in patients taking other anticonvulsants. The authors concluded that for most patients, evidence suggests generic switching is not associated with statistically significant differences in either clinical outcomes or PK parameters determining bioequivalence, but that some groups may be more susceptible to such switches. Importantly, switching between multiple generic AEDs may cause problems in some individuals leading to increased medical resource utilization (Yamada & Welty, 2011).

A second systematic review (Talati et al. 2012) of literature compared the efficacy of innovative and generic drugs either qualitatively (n = 71) or quantitatively (n = 18); most of the data concerned carbamazepine, phenytoin, or valproic acid. Although the strength of evidence was rated low for most publications in this review, it appears that initiation of anticonvulsant therapy with innovative versus generic drugs produced similar clinical results with respect to seizure frequency, discontinuations, and AEs. However, switching from a reference product to a generic drug may be associated with more hospitalisations and longer stays, although again, the quality of evidence was low (Talati et al., 2012).

Medications used to treat depression

In a retrospective case-control study based on the Ingenix Impact Database, Wu and coworkers (Wu et al. 2011) compared patients with major depressive disorder (MDD) who switched from a reference to a generic selective serotonin reuptake inhibitor (SSRI) with MDD patients who remained on the reference product (4449 matched patient pairs). During the 2003–2007 observation period, the risk (adjusted for baseline characteristics) of all-cause and MDD-related hospitalisations and hospitalisations/ED visits was significantly higher among patients who switched to a generic than those remaining on the reference product (ORs [95% CI] ranged from 1.149 [1.037–1.272] to1.948 [1.465–2.589]; all p ≤ .008). Total adjusted mean mental health-related and MDD-related medical costs were also significantly higher for switchers ($1154 vs. $935 and $733 vs. $511, respectively; both p ≤ .028), although antidepressant and SSRI costs were significantly lower ($384 vs. $500 and $266 vs. $371, respectively; both p < .001).The magnitude of increased medical costs was about twice that of reduced drug costs with generic switching (Wu et al., 2011).

Unterecker et al. (Unterecker et al., 2014) compared serum levels of active drug and metabolites in patients who received reference (n = 110) or generic (n = 68) extended-release (ER) venlafaxine. There were no significant between-group differences in the serum levels of venlafaxine (VEN), its metabolite O-desmethyl-VEN (ODV), or VEN + ODV; nor in the VEN/ODV ratio. The correlation coefficient for serum level versus dosage was also similar between groups (reference, 0.668; generic, 0.554). Females in the generic group, but not in the reference product group, demonstrated significantly higher dose-corrected VEN levels (p < 0.05); and smokers in the reference product group, but not the generic group, demonstrated significantly lower VEN and VEN + ODV levels (both p < 0.01). The authors concluded that despite the similarity between reference and generic serum levels, differences in specific subgroups emphasise the need to regularly monitor serum drug/metabolite levels (Unterecker et al., 2014). This interesting point may provide insight into why some patients are sensitive to the effects of switching to generic products while others tolerate switching well.

Using New Zealand’s national health and pharmacy databases, Lessing, Ashton, and Davis, (2015b) compared outcomes in patients who remained on reference venlafaxine (n = 12,407) with those in patients (n = 1624) who switched to generic venlafaxine between 2011 and 2013. There were no differences in health services utilisation (ED visits, hospitalisations, outpatient services) between nonswitchers and switchers, except for a larger drop in outpatient services visits by switchers (by 0.6 (95% CI = –1.2, –0.05) visits/person-year; p = .03), supporting the safety of switching from reference to generic venlafaxine (Lessing et al., 2015b).

Medications used to treat psychosis

In a study of Polish patients (N = 99) receiving treatment (amisulpride, olanzapine, quetiapine, risperidone) for chronic schizophrenia whose pharmacists had proposed a reference-to-generic switch within the previous 12 months, Murawiec et al. assessed emotional and behavioural responses to the switch proposal (Murawiec et al. 2015). Although initial responses to the proposed switch were predominantly negative (negative 42/99, neutral 36/99, positive 18/99, no response 3/99), more patients accepted (75/99, 76%) than rejected (24/99, 24%) the switch to the generic. Across all patients, 11/99 (11%) reported an unscheduled psychiatric visit, in part to discuss the proposed switch. Among those who switched, 21/75 (28%) reported negative consequences, and 5/75 (7%) reported a change in drug dosing; however, there were no discontinuations among switchers (Murawiec et al., 2015).

In a retrospective database study, Hsu et al. (Hsu et al., 2017) evaluated outcomes among patients with schizophrenia in Taiwan who used reference or generic risperidone (n = 404 and 145, respectively) and sulpiride (n = 334 and 991, respectively) products; patients who switched between reference and generic versions during the 2000 to 2012 study period were excluded. Reference risperidone users, compared with generic users, required a lower mean (SD) daily dose (2.14 [1.49] vs. 2.61 [1.53] mg, respectively; p < .001), and more generic users than reference users required augmentation therapy during the first 180 days of therapy (4.8% vs. 1.5%); but otherwise the two groups required similar rates of treatment intervention (discontinuations, augmentation, and psychiatric hospitalisations). Reference sulpiride users required a lower mean (SD) daily dose than generic sulpiride users (303 [244.3] vs. 341 [271.6] mg; p = .031) and had significantly lower rates of psychiatric hospitalisations (1.5% vs. 7.1%; adjusted HR 0.24, 95% CI =0.10–0.56). The authors suggest that the results cast doubt on the therapeutic equivalence of generic versus reference risperidone and sulpiride products (Hsu et al., 2017).

Switching patterns and clinical outcomes following the implementation of generic reference pricing for risperidone were evaluated in patients with schizophrenia in New Zealand who were receiving stable treatment with reference risperidone (Lessing, Ashton, & Davis, 2015a). In this retrospective database study, about 75% of reference risperidone users switched to one of the generic versions, and 1.5% switched back to the reference version; among those who switched, 50% remained with the same generic product, while the remainder made multiple between-product switches (40% made >2 switches, and 9% made >3 switches). Patterns of health care utilisation did not differ between switchers and nonswitchers during follow-up (Lessing et al., 2015a).

In a similar retrospective database study, Lessing et al. evaluated outcomes following the removal of reference olanzapine subsidies in New Zealand (Lessing, Ashton, & Davis, 2015c). Because the vast majority of reference olanzapine users (99.7%) switched to generic versions, the evaluations were within-subject (N = 5223), comparing outcomes for the 6 months preceding and following the switch. Most patients (86%) made >1 switch between generic products following the initial reference-to-generic switch. After switching to the generic, there was no significant increase in dosage or use of additional antipsychotic medicines, and health care utilisation remained the same before and after switching (Lessing et al., 2015c).

Italiano et al. (Italiano et al. 2015) assessed post-switch drug levels and clinical outcomes 4 weeks after switching from reference olanzapine to generic olanzapine in Italian patients with schizophrenia. Despite a significant reduction from baseline in measured serum olanzapine levels (from 27.7 ± 14.4 (reference) to 22.6 ± 12.3 ng/mL; p < .01), there was no evidence of clinical deterioration, as assessed by mean Positive and Negative Syndrome Scale scores using the Wilcoxon signed test (baseline 49.6 ± 8.3; endpoint 48.6 ± 9.5; p = .777). In addition, no patient demonstrated relapse or required dosage adjustment (Italiano et al., 2015). While indeed interesting, 4 weeks may not be a long enough time-frame to observe clinical deterioration.

In a prospective, randomized study in Ontario, Canada, Oluboka and colleagues (Oluboka, Stewart, Landry, & Adams, 2010) evaluated outcomes in patients with schizophrenia/schizoaffective disorder who were switched to a generic product after stabilisation on reference clozapine (n = 17) with those who remained on the reference product (n = 22). After 6 months, patients who switched demonstrated a significant improvement from baseline in mean Global Assessment Scale scores (t = –2.37, df = 8, p = .045), whereas those who remained on the reference product demonstrated significant improvement from baseline in mean 32-item Behaviour and Symptom Identification Scale scores (t = 2.49, df = 19, p = .022); no other significant pairwise comparisons were reported. These results suggest that there was no clinical deterioration following the switch from reference to generic clozapine (Oluboka et al., 2010). However, in an earlier, similarly designed study in 45 patients with schizophrenia, schizoaffective disorder, bipolar disorder with psychosis, or atypical psychosis with mood disorder, five patients relapsed while taking the generic product and nine worsened short of full relapse; two patients worsened while taking the reference product (0 relapse) (Kluznik, Walbek, Farnsworth, & Melstrom, 2001).

Medications used to treat attention deficit disorders

Two studies examined differences between two methylphenidate products, OROS (osmotic controlled release oral delivery system)-methylphenidate (Concerta^®) and generic methylphenidate ER, in patients treated for attention deficit hyperactivity disorder (ADHD) (Fallu, Dabouz, Furtado, Anand, & Katzman, 2016; Lally, Kral, & Boan, 2016). In a clinical case series analysis of 14 children and adolescents with ADHD, patients showed a significant improvement in Conners Parent Rating Scale Inattentive Scale scores after a switch from generic methylphenidate ER to the same dosage of the OROS product in a repeated measures analysis of mean difference scores (mean T-score reduction =23, p < .0001) (Lally et al., 2016). A switch from OROS methylphenidate to the generic product was not examined. Adult ADHD patients (N = 20) receiving ongoing treatment with OROS-methylphenidate who were switched to generic methylphenidate ER in a randomized, double-blind, crossover trial reported a significant decrease in satisfaction with treatment effectiveness in a mixed effects analysis of variance model using change from screening visit score (mean [SD] change from screening = –23.6 [36.8]; p < 0.0037), whereas patients who continued treatment with OROS-methylphenidate in the first study phase had no significant change in satisfaction scores (Fallu et al., 2016). Patients switched to the generic product also had a significant decrease in satisfaction with tolerability compared with scores for OROS-methylphenidate (mean [SD] change from screening = –33.7 [30.6]; p = 0.0001); those who stayed on OROS-methylphenidate did not. The differences observed in these studies may be related to different time-concentration profiles for the two products, and FDA revised guidance for bioequivalence studies for long-acting methylphenidate products after these studies were initiated to address this issue (Park-Wyllie et al. 2016).

Adherence

In the Gagne et al. (Gagne et al., 2015) retrospective database study described earlier, 93% of AED treatment initiators (18,306/19,760) started with a generic product. The mean time (SD) to the first treatment gap with the initial drug was significantly longer for initiators of generics than for initiators of reference products (137.9 [148.6] vs. 124.2 [125.8] days, respectively; p = .01) (Gagne et al., 2015). As noted above, plasma AED levels were not examined in the analysis.

In a retrospective study of patient electronic medical records between 2008 and 2012 in Spain, Sicras-Mainar et al. (Sicras-Mainar, Rejas-Gutierrez, & Navarro-Artieda, 2015) compared adherence and persistence for reference and generic products in patients who initiated treatment with gabapentin (N = 1369: reference, 400; generic, 969) or venlafaxine (n = 841: reference, 370, generic 471) for peripheral neuropathic pain and generalised anxiety disorder, respectively. Reference gabapentin and venlafaxine products demonstrated significantly longer mean (SD) persistence (time to final use of initial prescription) than generic counterparts (7.3 [3.0] vs. 6.3 [3.1] months, p < .001; 8.8 [2.9] vs. 8.1 [3.6] months, p = .002, respectively). Adherence (medication possession ratio [MPR: days’ supply/days on treatment]) was also significantly higher for reference gabapentin and venlafaxine vs generics (86.5% vs. 81.3%, p < .001; 82.9% vs. 79.0%, p = .045). Reference gabapentin produced a larger adjusted relative reduction in Numeric Rating Scale pain scores than generics (51% vs. 43%, p < .001), and reference venlafaxine was associated with a larger adjusted relative reduction in Hamilton Anxiety Scale score compared with generic products (62% vs. 48%, p < .001) (Sicras-Mainar et al., 2015). Both measures were analysed using a general linear model analysis of covariance with baseline score as a covariate.

In a retrospective study based on the Odense PharmacoEpidemiologic Database (Denmark), Rathe evaluated the impact of generic switching on treatment persistence among patients (N = 1386) who switched or did not switch from a reference AED or medication used to treat depression (Rathe 2015). Compared with patients who never switched, first-time switchers had a higher risk of non-persistence (HR 2.98; 95% CI =1.81–4.89); 35.7% of first-time switchers became non-persistent during the first year of follow-up. Compared with never switchers, the risk of non-persistence was similar in patients who had experienced a previous switch to a generic version of any drug, regardless of whether they switched or didn’t switch (HR 1.02 (0.72–1.43) and 0.98 (0.68–1.41), respectively) to a generic AED or a medication used to treat depression (Rathe 2015).

In a US claims database study designed to compare persistence of drugs used to specifically treat MDD, patients prescribed reference desvenlafaxine (Pristiq^®; n = 14,379) versus reference (n = 50,937) or generic (n = 208,198) SSRIs or serotonin-norepinephrine reuptake inhibitors (SNRIs), persistence rates were significantly higher at 6 and 12 months for reference desvenlafaxine (6 months: 60.4% [95% CI = 59.6–61.2]; 12 months: 42.5% [41.7–43.4]) compared with either generic (6 months: 55.2% [55.0–55.5]; 12 months: 38.6% [38.4–38.8]) or reference SSRIs/SNRIs (6 months: 51.9% [51.5–52.3]; 12 months: 31.6% [31.2–32.1]) (Solem et al. 2016). Median (95% CI) time to treatment discontinuation was longer for generic SSRIs/SNRIs (33.4 weeks [33.1–33.7]) compared with reference SSRIs/SNRIs (28.9 weeks [28.4–29.1]; reference desvenlafaxine: 40.7 weeks [39.3–42.0]). The study did not examine effects of a switch between generic and reference products (patients who switched medications were censored), and no generic desvenlafaxine was available at the time of the study (Solem et al., 2016).

Medication persistence was also examined in ADHD patients who were switched from OROS-methylphenidate to generic methylphenidate ER versus from the generic to the OROS product in a population-based retrospective cohort study (N = 21,940) (Park-Wyllie et al., 2016). When comparing methylphenidate ER products, 12-month persistence rates were significantly higher for patients switched to the reference product (76%) than for those switched to generic (56%; p <.0001).

Economic considerations

In a retrospective database study (N = 33,625), Helmers et al. (Helmers et al. 2010) used claims data from the PharMetrics Patient-Centric database to evaluate use of medical and pharmacy services during treatment with generic versus reference versions of five AEDs (carbamazepine, gabapentin, phenytoin, primidone, and zonisamide). Total adjusted annualised mean health care costs were 25.8% higher during periods of generic AED use compared with periods of reference AED use (between-period difference $3254; 95% CI = $2403–$4105; p < .05); which were primarily driven by higher medical service costs (difference $3186; 95% CI = $2359–$4012; p < .05), with pharmacy costs stable (difference $69; 95% CI = –$34–$171; p = NS) (Helmers et al., 2010).

In their retrospective cohort claims-based study (described above), Wu et al. (Wu et al., 2011) found that patients who switched to a generic from a reference product incurred higher mean risk-adjusted mental health-related and MDD-related costs for the 6 months following the switch (between-group differences $219 [95% CI = $51–$369], p = .028; and $222 [95% CI = $119–$305], p < .001, respectively). A similar pattern was observed for mental health-related and MDD-related costs (p < .001 for both) in a subgroup analysis of switchers from escitalopram (Wu et al., 2011).

In a retrospective cohort study of claims data from the MarketScan Commercial Claims and Encounters database, Vlahiotis et al. (Vlahiotis, Devine, Eichholz, & Kautzner, 2011) assessed costs during the 6 months following initiation of treatment with generic or reference SSRIs or SNRIs. Of the 16,659 new SSRI/SNRI users, 7955 (47.8%) initiated using a reference product and 8704 (52.2%) initiated using a generic product. Rates of discontinuation at 6 months were similar between users of reference (46.8%) and generic (44.2%) versions. Adjusted mean (95% CI) 6-month all-cause health care costs were lower in patients who initiated with a generic version ($3660 [3538–3787]) than with a reference version ($4587 [4422–4757]); in addition, adjusted mean (95% CI) antidepressant (SSRI/SNRI) costs were 43.7% lower in patients who initiated with a generic compared with those who initiated with a reference version ($174 [169–180] vs $309 [300–319], respectively) (Vlahiotis et al., 2011).

In the Spanish study of generic versus reference gabapentin and venlafaxine (detailed above) (Sicras-Mainar et al., 2015), adjusted mean (95% CI) per-patient health care costs were higher with generic versus reference gabapentin (€1277 [1213–1341] vs. €1057 [959–1155], respectively; difference €220; p < .001) and generic versus reference venlafaxine (€1110 [1006–1215] vs. €928 [814–1042]; difference €182; p = .020). Across both drugs, these differences favoured the reference product and remained significant regardless of age (<65 years: difference €221 [95% CI =59–382]; ≥65 years: difference €217 [51–382]; p < .01 for both comparisons) or gender (men: €197 [63–328], p = .005; women: €239 [96–397], p = .004) (Sicras-Mainar et al., 2015; Navarro-Artieda, Rejas-Gutierrez, Perez-Paramo, & Sicras-Mainar, 2016).

Conclusions

In this updated review of published literature between March 2010 and June 2017, results are mixed in regards to potential clinical differences resulting from taking generic versus reference psychotropic products. Studies evaluating the effects of initiation of reference products versus initiation of generic products suggest that, under these conditions, reference and generic products may be associated with similar clinical outcomes (Erickson et al., 2011; Yamada & Welty, 2011; Talati et al., 2012; Gagne et al., 2015; Italiano et al., 2015; Lessing et al., 2015a, Lessing et al., 2015b; Vari et al., 2016; Markoula et al., 2017). However, several studies have documented negative consequences associated with reference (branded and/or innovator) to generic product switches, including increased total health care costs (Helmers et al., 2010; Wu et al., 2011; Sicras-Mainar et al., 2015; Navarro-Artieda et al., 2016) and resource use (Labiner et al., 2010; Wu et al., 2011; Talati et al., 2012; Hsu et al., 2017), reduced treatment persistence (Rathe 2015; Sicras-Mainar et al., 2015; Navarro-Artieda et al., 2016) and adherence (Oluboka et al., 2010; Sicras-Mainar et al., 2015; Navarro-Artieda et al., 2016), and evidence of clinical deterioration (Murawiec et al., 2015), which are all of clinical concern. These results suggest that it is the switch itself, rather than the particular product prescribed, which is problematic. The findings generally (but not in all cases (Sicras-Mainar et al., 2015)) support the idea that initiating treatment on a generic product—and continuing on that same product—can reduce cost to the payer without reducing clinical benefit, whereas switching between products during treatment risks reducing treatment efficacy and/or tolerability without achieving cost benefit. An important limitation of the available data on switching between medication products is the lack of comparisons between two or more different generic products, which could introduce even greater variability than a switch between reference generic and products. The potential impact of generic to generic substitution has not been adequately studied to date.

This review outlines issues regarding substitution that deserve continued attention from clinicians, patients, and other stakeholders. First, the process time course and associated costs for generic product approval are a fraction of those for their reference predecessors. Bioequivalence of the generic product compared with the reference product can be established via a single positive bioequivalence study. However, bioequivalence studies are typically conducted in healthy volunteers with low doses, instead of within the target population for the drug, with bioequivalence established according to the rate and extent of absorption, without clinical efficacy data. Because many psychotropic drugs are modulators of hepatic cytochrome P450 enzymes, bioequivalence demonstrated after single-dose studies may not be operant under steady-state conditions. Failure to assess the impact of product-specific (e.g., excipients) and patient-specific (e.g., comorbidities, concurrent medications, smoking status) factors during the approval process for generic products may set up a situation where bioequivalence may not translate into therapeutic equivalence.

Pharmacoeconomic studies provide valuable evidence regarding the risk:benefit ratio achieved with the switch from reference to generic or between generic products, while taking into account the clinical course after such substitutions, rather than looking at cost as the only variable. The accessibility of such claims databases for pharmacoeconomic studies, their reliability in illuminating multiple aspects of the care experience, along with the potentially large number of available patients, can provide persuasive arguments for or against switching that may not be available in anecdotal accounts of AEs and clinical deterioration. This real-world assessment provides detail on health care resource use and costs, including variables suggestive of efficacy loss (e.g., hospitalisations, ED visits, provider visits, requirements for adjunctive medications) and reduced adherence/persistence (Emanuele 2011). However, these studies cannot capture the quality of life impact of any loss of clinical efficacy and increase in AEs (and consequently tolerability) that might result from switches between products in severely ill patients.

Health care providers must continue to be advocates for their patients, especially those with underlying psychiatric conditions. Because the potential consequences of reduced or lost therapeutic efficacy with reference to generic or generic to generic switching in patients with psychiatric disorders may be severe (including increased risk for suicide and harm to others), providers should be consulted before any ‘automatic’ switches are made, and the concerns of the patient must also be respected. No substitution should be permitted without the knowledge of the patient and health care provider, and any proposed substitution should be accompanied by a comprehensive discussion between the clinician and the patient about the anticipated benefits (e.g., lower cost) and the possible risks (e.g., loss of efficacy, increasing AEs and health care burden). Being an advocate for the patient and facilitating a consistent and continued supply of the drug product (whether reference or generic) that provides stability and avoids clinical deterioration should be the treatment goal.

We look forward to and encourage future well-designed studies regarding the impact of generic substitution, and to further engagement of the multiple stakeholders involved in the treatment of psychiatric and CNS disorders in the policy discussions emerging from the findings of such studies.

Key points

• Approval of generic products requires demonstration of bioequivalence with the reference (branded) product. However, there may be greater variability between two generic products with the same reference than between either generic and the reference product itself. Further, additional variability is introduced over the lifecycle of each product.

• Some studies of psychotropic medications reviewed here suggest that some generic products may be therapeutically equivalent to their reference counterparts, especially if used at initiation of treatment. However, other studies provide evidence of reduced efficacy (reduced treatment persistence/adherence), and/or increased adverse event burden during switches from reference to generic products.

• Results of pharmacoeconomic studies indicate that patients switched between products during the treatment of psychiatric illness may utilize greater health care resources.

• Although there are immediate economic incentives in terms of medication cost for switching patients between branded and generic products during the course of their treatment, such incentives may not be apparent over the long-term.

• Advocating for the patient and facilitating a consistent and continued supply of the drug product (whether reference or generic) that provides stability and avoids clinical deterioration should be the quintessential treatment goal.

Disclosures of interest

P.B. Honoraria for advisory board participation, consultancy, grant funding, and/or giving lectures for Allergan, Bristol-Myers Squibb, Janssen, Lundbeck, Otsuka, Pfizer, Sunovion, Takeda.

H.M. Honoraria/paid speaker: BMS, Janssen, Lundbeck, Otsuka, Sunovion; Research Support: Lundbeck, Janssen; Consultant: Janssen, Lundbeck, Otsuka, Perdue, Pfizer, Shire, Sunovion; No stocks or employment

E.A.W. Consultant for Pfizer

M.B. Employee of Pfizer Canada Inc. and owns Pfizer Inc. stock and stock options

Acknowledgment

This analysis was funded by Pfizer. Medical writing support was provided by Kathleen M. Dorries, PhD and John H. Simmons, MD of Peloton Advantage and was funded by Pfizer.