If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Address correspondence to: Chintan V. Dave, Department of Medicine, Brigham and Women’s Hospital, Division of Pharmacoepidemiology and Pharmacoeconomics, 1620 Tremont Street, Suite 3030, Boston, MS 02120.
Program On Regulation, Therapeutics, And Law (PORTAL), Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital, Boston, MA, USA
Program On Regulation, Therapeutics, And Law (PORTAL), Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital, Boston, MA, USA
Program On Regulation, Therapeutics, And Law (PORTAL), Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital, Boston, MA, USA
Program On Regulation, Therapeutics, And Law (PORTAL), Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital, Boston, MA, USA
Prescription drug shortages can disrupt essential patient care and drive up drug prices.
Objective
To evaluate some predictors of shortages within a large cohort of generic drugs in the United States and to determine the association between drug shortages and changes in generic drug prices.
Methods
This was a retrospective cohort study. Outpatient prescription claims from commercial health plans between 2008 and 2014 were analyzed. Seven years of data were divided into fourteen 6-month periods; the first period was designated as the baseline period. The first model estimated the probability of experiencing a drug shortage using drug-specific competition levels, market sizes, formulations (e.g., capsules), and drug prices as predictors. The second model estimated the percentage change in drug prices from baseline on the basis of drug shortage duration.
Results
From 1.3 billion prescription claims, a cohort of 1114 generic drugs was identified. Low-priced generic drugs were at a higher risk for drug shortages compared with medium- and high-priced generic drugs, with odds ratios of 0.60 (95% confidence interval [CI] 0.44–0.82) and 0.72 (95% CI 0.52–0.99), respectively. Compared with periods of no shortage, drug shortages lasting less than 6 months, 6 to 12 months, 12 to 18 months, and at least 18 months had corresponding price increases of 6.0% (95% CI 4.7-7.4), 10.9% (95% CI 8.5-13.4), 14.2% (95% CI 10.6-17.9), and 14.0% (95% CI 9.1-19.2), respectively.
Conclusions
Study findings may not be generalizable to drugs that became generic after 2008 or those commonly used in an inpatient setting. The lowest priced drugs are at a substantially elevated risk of experiencing a drug shortage. Periods of drug shortages were associated with modest increases in drug prices.
Generic drugs make up about 88% of all filled prescriptions in the United States, and have saved the US health care system a trillion dollars in the past decade [
]. However, the phenomena of shortages of generic drugs and cases of rising generic drug prices have the potential to threaten these economic and social gains.
A Government Accountability Office report found a three-fold increase in the number of drugs affected by shortages between 2007 (154) and 2012 (456) [
]. More recently, production disruptions in Puerto Rico—where a significant number of drugs are manufactured—are anticipated to result in severe drug shortages [
]. When patients are forced to switch to a brand-name version of a generic drug experiencing a shortage, they often face therapy disruptions because of requirements of previous authorizations [
] have seen dramatic price increases over short periods of time. Although these cases are commonly attributed to the lack of robust competition levels in the generic drug marketplace [
], patients are faced with the dual problem of reduced drug availability and an expensive remaining supply.
To help understand the origins of generic drug shortages and their subsequent impact on prices, we used a large cohort of generic drugs in the United States to identify the predictors of drug shortages and the relationships between drug shortages and increases in generic drug prices.
Methods
Study Cohort and Data Sources
MarketScan commercial outpatient pharmacy claims data—an employer-sourced database covering 30 million lives annually—from 2008 to 2014 were used to identify a cohort of generic drugs available for the entirety of the study period. The outpatient pharmacy files contain patient-level drug dispensing data, including patient and third-party payment information. Pertinent non–payment-related variables include drug fill dates, quantity of drug dispensed, active ingredient names, administration form, drug strength, manufacturer name, and whether the dispensed drug was a generic or brand-name medication. Individual study drugs were defined as distinct combinations of active ingredient, administration forms, and drug strength.
The University of Utah Drug Information Services (UUDIS) database between 2008 and 2014 was used to elicit information on drug shortage status [
], and compiles information on the drug shortage start date, end date, and—whenever possible—reason for the shortage. Once a drug has been voluntarily reported to be in short supply, clinicians at the UUDIS verify the authenticity of the report and validate the shortage.
The study was approved by the Institutional Review Board at Brigham and Women’s Hospital.
Predictors of Drug Shortages
The 7-year period between 2008 and 2014 was divided into 14 six-month intervals. The first half of 2008 was designated as the baseline period during which baseline predictors were assessed; time-varying predictors were assessed thereafter. Predictors of drug shortages were grouped into four categories: market competition level, market size, administration form, and baseline drug price.
Herfindahl-Hirschman Index (HHI) is a measure of competition levels and has been used in previous research in this field [
], and is calculated by summing the squares of individual firms’ (in this case, generic manufacturers’) market shares. For a generic drug produced by four manufacturers with market shares of 10%, 20%, 30%, and 40%, the estimated HHI would be 102 + 202 + 302 + 402 or 3000; higher HHI values are therefore indicative of a less competitive market. A manufacturer’s market share was estimated by dividing the number of prescriptions produced by a manufacturer by the total number of dispensed prescriptions for a given generic drug in a study period.
In our study, HHI was used in two ways. The first assessment was the baseline (time-stable) HHI, measured during the first half of 2008. Using baseline HHI thresholds of 5000 and 8000 (for rationale on how the thresholds were selected, see Appendix in Supplemental Materials found at doi:10.1016/j.jval.2018.04.1826), drugs were grouped into high, medium, and low levels of baseline market competition. Second was the change in HHI from its baseline value, which was estimated by calculating the difference in HHI in the previous study period and its baseline value. Lagged values of HHI ensured that a change in HHI preceded a change in the outcome. Because HHI explicitly incorporates the number of manufacturers into its calculation, while more accurately reflecting the marketplace dynamics of a drug, we modeled HHI instead of number of generic manufacturers; as sensitivity analysis, we modeled the number of generic manufacturers instead of HHI.
Baseline market size was estimated by dividing the number of dispenses for an individual drug (e.g., lisinopril 20 mg tablets) by the total number of dispensings for the entire cohort of generic drugs in the baseline period. Using percentiles, the study drugs were grouped into small, medium, and large markets. Drugs were also grouped into seven administration form categories: tablets, extended-release tablets, capsules, extended-release capsules, creams/lotions/gels, solutions, and miscellaneous. Because extended-release formulations may be more difficult to produce, they were modeled separately. Using percentiles of estimated baseline prices (see next section), and within administration forms, we grouped generic drugs into the low, medium, and high baseline price groups.
Finally, using the UUDIS database, we determined whether a study drug was in shortage during a given study period.
Association between Drug Shortages and Change in Price
Patient out-of-pocket costs—which include co-pays, co-insurances, and deductibles—and third-party costs were summed to calculate the overall price of a dispensed medication. Patient and third-party costs were combined to account for potential cost-shifting from payers to patients over time. This overall price of a dispensed medication was divided by the quantity of the drug supplied (e.g., number of capsules) to estimate drug prices by quantity supplied.
Drug-specific prices were estimated for all 14 study periods. Drug prices estimated during the baseline period were adjusted as the baseline value of the outcome variable, and prices calculated in the 13 subsequent study periods were modeled as the dependent variable. In other words, the outcome was the change in drug price from its baseline value (first half of 2008) over time.
The estimand of interest was the association between shortage duration and change in drug price. Before the start of each period, we estimated the cumulative number of months that a drug was in shortage (accounting for multiple shortage episodes), and categorized the duration into the following time-varying categorical variables: no shortage, 0 to less than 6 months of shortage, 6 to less than 12 months of shortage, 12 to less than 18 months of shortage, and 18 or more months of shortage. The model also controlled for market competition levels (baseline and time-varying), baseline market size, and administration forms.
Statistical Analysis
All analyses were conducted using SAS (version 9.4, SAS Institute, Cary, NC). See Appendix for full model specification.
The first study objective was to identify the determinants of drug shortages. Accordingly, using the GENMOD procedure in SAS, we used a generalized estimating equation modeling drug shortage status (dichotomous) as the dependent variable and market competition levels, market size, administration form, and drug prices as independent variables.
For the second study objective, we used the MIXED procedure to estimate a linear mixed effects model, which modeled the change in generic drug prices from the baseline value. Drug prices were log-transformed to account for skewness and violations of normality. Standard errors were corrected using the Kenward-Roger degrees of freedom approximation [
For both models, time was modeled nonparametrically as a categorical factor to avoid assumptions of linear trends, and an unstructured covariance matrix for the residuals was fit. The models also incorporated a three-tier nested correlation structure (see Appendix).
Results
After analyzing data for 1.3 billion prescription claims over the 7-year period, we identified a cohort of 1114 generic drugs that met our inclusion criteria, grouped into 119 drug classes. Examples of drug classes represented in the cohort included angiotensin-converting enzyme inhibitors, macrolides, quinolones, and statins.
The prevalence of drug shortages increased on a yearly basis after 2008, peaking at 11% in the first half of 2011, declining thereafter to 4% in the second half of 2014 (Fig. 1). Over the 7-year period, 309 drugs (28%) experienced shortages (see Appendix Table 1 in Supplemental Materials found at doi:10.1016/j.jval.2018.04.1826); the median duration of a shortage episode was 8.4 months (interquartile range 4.7-16.7). When the reason for a shortage was known (32%), almost three-quarters of drug shortages were attributed to manufacturing issues, including production delays or difficulties in the procurement of raw materials, whereas one-quarter were attributed to an increase in demand exceeding the available supply. Lower-priced drugs and drugs with extended-release formulations were more likely to experience shortages. Drugs that did not experience a shortage on average saw a price increase of 44% over the study period, whereas drugs that experienced a shortage saw a 55% increase (see Appendix Table 2 in Supplemental Materials found at doi:10.1016/j.jval.2018.04.1826).
Fig. 1Prevalence of shortages in a cohort of generic drugs, 2008–2014.
Table 1 lists the shortage status and prices for select drugs in the cohort. Salsalate 500 mg tablets were in shortage for 17 months during the study period, and increased in price by over 650%. However, a shortage was not always associated with an increase in drug prices. For example, even though clonazepam 0.25 mg disintegrating tablets were in shortage for 6 months, the price decreased by 14% over the study period.
Table 1Price and shortage information for select drugs.
Using high levels of baseline market competition levels as the referent category, generic drugs with medium and low competition levels did not have a significantly elevated risk for drug shortages (odds ratio [OR] 1.05; 95% confidence interval [CI] 0.78-1.41 and OR 1.00; 95% CI 0.68-1.48, respectively; Table 2). Similarly, a decrease in market competition levels (measured by an increase in HHI over time) was not associated with a risk of a drug shortage (OR 0.97; 95% CI 0.92-1.03). Market size was also not a significant determinant of drug shortage status; compared with drugs serving small patient populations (i.e., small markets), drugs serving medium and large markets did not have elevated risks for drug shortages (OR 0.98; 95% CI, 0.71-1.35 and OR, 1.08; 95% CI 0.77-1.49, respectively). Solutions and extended-release capsules were associated with higher risks of drug shortages when compared with tablets (OR 1.67; 95% CI 1.00-2.79 and OR 2.64; 95% CI, 1.58-4.42, respectively).
Table 2Adjusted association between generic drug characteristics and shortages
Market size defined as the percentage of the total generic drug market held by a study drug during the baseline period. Using tertiles, drugs were grouped into one of three market categories.
Using percentiles of baseline prices (measured during 2008, first half), drugs were grouped into low, medium, and high baseline price groups.
Low
Reference
Medium
0.60 (0.44–0.82)
High
0.72 (0.52–0.99)
CI, confidence interval; HHI, Herfindahl-Hirschman Index, a measure of market competition, see text for explanation.
Using thresholds of 5000 and 8000, drugs were classified into three baseline competition groups.
† Per 1000 increase in HHI from baseline, signifying a decrease in market competition levels over time.
‡ Market size defined as the percentage of the total generic drug market held by a study drug during the baseline period. Using tertiles, drugs were grouped into one of three market categories.
§ Using percentiles of baseline prices (measured during 2008, first half), drugs were grouped into low, medium, and high baseline price groups.
Baseline prices were associated with the risk of shortages. Generic drugs with medium and high prices had significantly lower risks for drug shortages compared with lower-priced drugs (OR 0.60; 95% CI 0.44-0.82 and OR 0.72; 95% CI 0.52-0.99, respectively).
In the sensitivity analysis in which we modeled the number of manufacturers instead of market competition levels, after adjustment, the number of manufacturers was not associated with a significant risk of a shortage (OR 1.01; 95% CI 0.97-1.04).
Association between Drug Shortages and Change in Price
We found a stepwise percentage increase in prices as a function of shortage duration (Table 3). Compared with drugs that did not experience shortages, drugs with shortages lasting less than 6 months were expected to see price increases of 6.0% (95% CI 4.7-7.4). For shortage durations of 6 to less than 12 months, 12 to less than 18 months, and 18 or more months, the corresponding increases in prices were 10.9% (95% CI 8.5-13.4), 14.2% (95% CI 10.6-17.9), and 14.0% (95% CI 9.1-19.2), respectively.
Table 3Association between shortage status and change in drug price
In our study of commonly used outpatient generic drugs available between 2008 and 2014, we found that the prevalence of shortages decreased from its peak in 2011-2012 to less than 4% of the overall sample by the end of 2014. We also found that market competition levels and market sizes were not associated with the likelihood of drug shortages. Rather, generic drug prices were strongly predictive of drug shortages; compared with their high-priced generic counterparts, low-priced generic drugs were more likely to experience shortages. Drug shortages were associated with a modest increase in prices.
Our finding that low-priced generic drugs were at an increased risk for drug shortages could be due to many different factors. First, because low-priced generic drugs are generally less profitable than high-priced ones, manufacturers may prioritize continued production of more profitable drugs within their portfolios. Other experts have attributed the higher incidence of drug shortages in the United States—compared with other developed regions such as Europe—to the higher reimbursement rates for generic drugs outside the United States [
]. Another possibility is that smaller margins in low-priced generic drugs may induce manufacturers to maximize profits by cutting corners in ways that compromise production quality, predisposing these agents to plant closures and production disruptions.
In an effort to increase the predictability of drug shortages, the Food and Drug Administration (FDA) Safety and Innovation Act of 2012 required manufacturers to report anticipated disruptions in supply to the FDA; however, the effect of this provision of the law is unknown [
]. When the FDA is considering a plant closure, the agency routinely takes steps to avoid precipitating a drug shortage. For example, in part to avoid a shortage of a needed medication, the FDA recently decided to allow the continued distribution of a generic injectable drug containing glass fragments, advising health care professionals to use a filter before administering the drug [
]. More innovative approaches that balance the safety risks of substandard drugs with the possibility of drug shortages such as changing the pass/fail system that the FDA currently uses to letter grades have also been proposed [
FDA allows temporary saline imports to deal with shortages caused by Hurricane Maria.2017; (Available from: http://www.raps.org/Regulatory-Focus/News/2017/10/11/28653/FDA-Allows-Temporary-Saline-Imports-to-Deal-With-Shortages-Caused-by-Hurricane-Maria/. [Accessed October 18, 2017])
]. This latter approach could be expanded further if deflationary pressures on certain generic drugs lead to future US shortages.
Although drug shortages have been associated with some well-publicized examples of rising generic drug prices, the low prevalence of drug shortages in recent years coupled with their modest association with price increases in our study demonstrate that other factors such as market competition levels are more important contributors to this phenomenon [
]. Although we found that drug prices do not rise substantially following a shortage, modest price increases may still happen. Because a drug in shortage may not have an appropriate therapeutic substitute, a patient cannot always change their medication regimen in an event of a shortage, allowing manufacturers to increase drug prices without risking substantial reductions in prescription numbers. In other cases, when the FDA has recommended a plant closure [
], the resulting decrease in competition levels could allow the remaining manufacturers more leeway to increase drug prices.
The real costs of shortages to patients are likely higher than what we report because in addition to their impact on drug prices, drug shortages can increase health care costs in other ways. For example, the costs associated with switching patients to a more expensive therapy [
The study combined disparate data sources to assess several pertinent drug-specific variables including drug shortage status, drug prices, and competition levels, and analyzed a recent and large commercial claims database to identify a representative cohort of more than a 1000 generic drugs. We were able to characterize for the first time the secular trends of drug shortages in generic drugs, identify the predictors of drug shortages, and link drug shortages to changes in generic drug prices. However, the findings are limited to drugs commonly used in an outpatient setting (mostly oral and topical formulations), and may not generalize to commonly used inpatient drugs (e.g., injectable formulations). Although the cohort inclusion criteria required drugs to be available as generics for the entirety of the study period, there is little reason to believe that these findings would not extrapolate to more recently approved generic drugs.
Drug shortages are challenging public health crises that contribute to patient mortality [
]. In recent years, there has also been worry that drug shortages can lead to generic drug price increases, and we found that modest price increases were associated with the outpatient drug shortages in our cohort. Policymakers interested in avoiding spikes in certain generic drug prices should focus on the growing number of factors—such as market competition levels—that have been demonstrably linked to price changes [
We also found that drug shortages were associated with the segments of the market featuring the lowest generic drug prices. Some commentators have proposed ways to bolster the generic drug market, which could result in increased generic drug prices [
], and our data suggest that such moves may work to reduce the risk of some generic drug shortages. However, the secondary effects of such proposals should be considered carefully given the substantial gains to patient health and the health care system that have emerged as a result of the low prices achieved by direct competition among interchangeable brand name and generic drugs in the United States in the last 30 years. In the short-term, the FDA should recognize that very low generic drug prices might be a factor in subsequent shortages and continue to consider the risk of shortages when taking steps to promote the safety and quality of the generic drug marketplace.
Source of financial support: The Laura and John Arnold Foundation supported the study. Dr. Kesselheim’s work is also supported by the Engelberg Foundation and the Harvard Program in Therapeutic Science.
FDA allows temporary saline imports to deal with shortages caused by Hurricane Maria.2017; (Available from: http://www.raps.org/Regulatory-Focus/News/2017/10/11/28653/FDA-Allows-Temporary-Saline-Imports-to-Deal-With-Shortages-Caused-by-Hurricane-Maria/. [Accessed October 18, 2017])
Depending on the route of administration, reported prices are per tablet or actuation.
32148-X&id=gr1.jpg){kind=link}