Regulatory review time and post-market safety events for novel medicines approved by the EMA between 2001 and 2010: a cross-sectional study

Abstract

Aims

Regulatory review time has been associated with post-market medication safety issues in the United States. Our objective was to evaluate whether regulatory review time and near deadline approval are associated with post-market safety events (PMSEs) for novel medicines approved by the European Medicines Agency (EMA).

Methods

We performed a cross-sectional analysis of all novel medicines approved by the EMA through the centralized authorization procedure between 2001 and 2010. PMSEs were defined as withdrawals and communications identified through Dear Healthcare Professional Communications (DHPCs). Regulatory review time was defined as the time that elapsed between the start of the assessment procedure and approval. Near regulatory deadline approval was defined as approval within the 30 days before the EMA’s 210 day regulatory deadline.

Results

Among 161 eligible medicines, PMSEs were identified for 49 (30.4%), 44 of which were DHPCs, five of which were withdrawals. Median regulatory review time was 337 days (IQR 276–406) and was not associated with PMSEs (P = 0.57). However, when categorized by regulatory review speed tertile, there were differences in risk of PMSEs, with higher rates among medicines in the middle tertile (25 of 55, 45.4%; P = 0.01). Finally, 26 medicines were approved near the 210 day regulatory deadline, but were not more likely to have PMSEs (38.5% vs. 28.7%; P = 0.32).

Conclusions

Neither faster EMA regulatory review speed nor approval near regulatory deadlines was associated with greater likelihood of PMSEs among recently approved novel medicines.

What is Already Known About this Subject

• Pharmaceutical regulatory agencies face the constant challenge of balancing competing priorities, facilitating rapid access to novel medicines while ensuring public safety through adequate assessment of medication harms.

• For the US Food and Drug Administration, the approval of new drugs within 60 days of the agency’s regulatory deadline was associated with increased post-market safety risk.

• Few other studies have examined timeliness of regulatory review of new drug applications and post-market safety risk and none has focused on novel medicines authorized by the European Medicines Agency.

What this Study Adds

• For the European Medicines Agency, neither faster regulatory review speed nor approval near regulatory deadlines was associated with greater likelihood of post-market safety events among recently approved novel medicines, including either post-market safety communications or withdrawals.

• These findings provide some reassurance that the European Medicines Agency is able to facilitate rapid access to novel medicines while ensuring public safety through adequate pre-market assessment of medication harms.

Introduction

Pharmaceutical regulatory agencies face the constant challenge of balancing competing priorities, facilitating rapid access to novel medicines while ensuring public safety through adequate assessment of medication harms. At times, these priorities are at odds with one another, as determination of whether medicines are safe and efficacious can require time, which in turn delays access. On the other hand, if regulatory reviews are hastened in an attempt to facilitate rapid access to novel medicines, regulators may not have sufficient time to review thoroughly all of the data and involve all of the relevant experts to evaluate comprehensively the safety of medicines. The European Medicines Agency (EMA) is acutely aware of these competing priorities and its Senior Medical Officer has commented on this dilemma publicly 1, acknowledging the inherent challenge and suggesting ways it can be addressed 2.

In order to facilitate timely regulatory review, most agencies operate under deadlines. For instance, the United States enacted the Prescription Drug User Fee Act (PDUFA) in 1992, requiring that the Food and Drug Administration (FDA) make a certain proportion of regulatory decisions within predefined periods after submission, such as within 6 months for priority reviews of new drug applications or 10 months for standard reviews (although the deadline was 12 months initially) 3. In the European Union (EU), the EMA is responsible for the regulatory review, market authorization and coordination of safety monitoring of all marketed medicines and similarly must make regulatory decisions within predefined periods 4. The EMA is generally required to make a decision on marketing authorization within 210 days, excluding the time taken by the manufacturer to address the agency’s questions 5.

Implementation of these regulatory review time requirements have led to both the FDA and EMA making regulatory review decisions more quickly, although the FDA is slightly more timely 6. However, the consequences of more rapid regulatory decision making on our understanding of medication safety at approval are not fully understood. An examination of all medicines approved by the FDA between 1993 and 2005 found that approval decisions were concentrated in the weeks preceding the deadlines and that medicines authorized within 60 days of the deadline were associated with a significantly increased risk of post-market safety communications and withdrawals 7. This work has been updated and enhanced in a subsequent article using additional data, extending the period of observation and examining four measures of post-market safety events. Nevertheless, the findings remain consistent, a concentration of approvals in the weeks preceding the deadlines and an increased risk of post-market safety events for the medicines approved close to the deadline 8. No studies have examined whether this same risk is present among medicines approved by the EMA, although the agency is increasingly making decisions within similarly short time periods 6. Moreover, the EMA now has a similar scope of authority, as it is responsible for marketing authorization for all human medicines submitted to the centralized authorization procedure which, once granted by the European Commission, is valid for all European Union (EU) member states.

Accordingly, our research objective was to examine whether the time required by the EMA for approval of novel medicines, as well as whether approvals granted near the regulatory deadline, were associated with identification of safety concerns after approval, once the drug is available for use. Post-market safety events (PMSEs) were defined as medication withdrawals and safety communications, identified using safety-focused Direct Healthcare Professional Communications (DHPCs). We hypothesized that, similar to the FDA experience, novel medicines approved more quickly or approved near the regulatory deadline would be more likely to have PMSEs.

Methods

Study sample

We created a sample of medicines approved by the EMA from January 1 2001 to December 31 2010. In brief, we identified all novel products approved during this period, including small molecules and biologics through the EMA’s website and their public records. Then, reformulations, combinations therapies and non-therapeutic agents, such as radiographic dye, were excluded. Generics and biosimilar products were not included. To be explicit, we only included novel medicines approved by the EMA through the centralized authorization procedure. The Central Authorization Procedure represents the most frequently used, but not the only, pathway for new medicine approval in the EU. Current rules require that the vast majority of medicines obtain marketing authorization through this pathway [http://www.ema.europa.eu/ema/index.jsp?curl = pages/about_us/general/general_content_000109.jsp&mid = WC0b01ac0580028a47]. From among this sample of 184 novel medicines, we excluded medicines approved after more than one review cycle, because they were few in number (n = 7) and were likely to be distinct as applications in comparison with the larger majority of applications approved after one review cycle. We also excluded medicines for which administrative dates were missing on the documents retrieved from the EMA website (n = 16), since we would be unable to calculate EMA review time for these applications. Thus, our final sample included 161 novel medicines for which all needed data were complete (Figure 1).

Figure 1.

Identification of the sample of novel medicines for the current analysis

Data source

Two investigators (JDZ and HB) obtained information for every medicine approved by the EMA through the EMA’s website, which makes European Public Assessment Reports (EPARs) publicly available. In brief, EPARs provide a summary of scientific reviews and list notable regulatory events for all medicines currently approved by the EMA, as well as for medicines that were refused or withdrawn for any reason. Therefore, EPARs contain all needed information on each approved medicine that we collected for our study, including drug name (molecule and brand name), manufacturer, small molecule or biological and orphan status, as well as regulatory submission date, procedure start, manufacturer inquiry and response and approval dates.

Main outcome variable: Post-market safety events (PMSEs)

Our main outcome of interest was identification of safety events after the medicine was approved for use, including either a withdrawal or safety communication. Post-market safety communications otherwise known as DHPCs and including those leading to a withdrawal of the medicine, were thought to be the most relevant and reliable proxy for safety events because the EMA does not apply black box warnings to the labels of approved medicines, as is done in the United States. In order to identify DHPCs, we first examined safety communications and materials made available by the EMA but determined (and confirmed through personal communication with EMA representatives from the Information Department, March 2013) that the EMA does not maintain a comprehensive or public listing of issued DHPCs. Therefore, as described in prior work 9, we conducted a cross-country examination of four large EU pharmaceutical regulatory agencies in France, Netherlands, Spain and the United Kingdom that make DHPCs and medicine withdrawals publicly available. Because national regulators mostly issue DHPCs and withdraw medicines in response to EMA communications, communications from these four agencies were the best available proxy of PMSEs. All safety communications (DHPCs and withdrawals) were collected for the period from January 1 2001 to October 31 2013, allowing a minimum of nearly 3 years of observation after approval among the medicines included in our sample. All DHPCs and withdrawals were reviewed to ensure that they were communicating safety issues. DHPCs related to technical problems (dosage, route of administration, quality defects), supply issues, or efficacy issues were not categorized as safety communications and were subsequently excluded. DHPCs communicating both safety and efficacy issues were included.

Main independent variables: regulatory review time and deadlines

Our main independent variables of interest were regulatory review time and near regulatory deadline approvals. For each medicine, regulatory review time was calculated as the time elapsing between the start of the assessment procedure (not the receipt of the application) and approval. We examined three different aspects of review time: 1) regulatory review time, as described above, our primary independent variable of interest, 2) EMA review time, which represents the time taken solely by the EMA to review the application and was a secondary variable of interest and 3) manufacturer regulatory response time, which represents the time taken by the manufacturer to respond to EMA inquiries during the review process and was also a secondary variable of interest. Thus, regulatory review time represents the sum of the EMA review time and the manufacturer regulatory response time. We made the decision to examine multiple components of the review process because it seemed relevant to study not only the time taken by the EMA to review the application but also the time taken by manufacturers to address the questions and issues raised by the agency, since it can reflect the effort required to define clearly the benefit–risk profile of a medicine.

For purposes of analysis, regulatory review time was examined first continuously and then categorized into tertiles as a secondary analysis. We assumed that after categorizing the sample into tertiles, medicines approved more rapidly would be associated with a greater risk of post-market safety communications, and that medicines approved more slowly (last tertile) might also be associated with a higher safety risk, since the longer time taken by the regulator could signal difficulties in assessing the benefit–risk profile of those medicines. The sample was also categorized as follows: 6 months or fewer, 7 months and 8 months or longer, reflecting the EMA 210 day regulatory deadline. However, as results were generally consistent, we only present results of the tertile analysis.

Similarly, EMA review time and manufacturer regulatory response time were examined first continuously and then categorized as secondary analyses. EMA review time was categorized as less than 7 months, 7 months, more than 7 months, such that the first period corresponds to an EMA review time where the decision was made well ahead of the regulatory deadline, the second period near the regulatory deadline, and the last period after the deadline has passed. Manufacturer regulatory response time was categorized as tertiles rounded up to a whole number of weeks, as no deadline is defined for the manufacturer to address the EMA’s queries.

Near regulatory deadline approval was defined as an approval granted within the 30 days before the 210 days deadline for a decision on market authorization by the EMA, again excluding the time taken by the manufacturer to address the agency’s questions.

Exploratory analyses

Because regulatory action is a complex multifaceted process, we explored whether the association between regulatory review time and post-market safety risk differed on the basis of several key medicine and application characteristics, including whether the medicine was evaluated in the accelerated assessment procedure or granted orphan designation, year of approval (specifically in reference to the Risk Management Plan requirements instituted in 2005 9), whether the EMA had approved the medicine for market prior to the FDA, whether it was a small molecule or biologic, whether the medicine was indicated for acute, intermediate or chronic duration of treatment as a measure or morbidity, and presence of a pre-approval safety concern. To identify pre-approval safety concerns, one author (JDZ) reviewed the EPARs for discussions of medicine safety, including the ‘Public assessment report’ and/or ‘Scientific discussion’ section of the EPARs. When no concern was identified, documents were searched using the term ‘concern’. Free text was extracted and reviewed by four authors (JDZ, NSD, HB and JSR) who categorized each medicine as having or not any pre-approval safety concern. Disagreements were resolved by discussion.

Statistical analysis

Descriptive statistics were used to characterize the sample. Chi-square tests were used to examine risk of PMSEs by key medicine and application characteristics. Distribution of regulatory review, EMA review and manufacturer response times were assessed using the Shapiro–Wilk Ttest. Based on the non-normal distribution, we used Wilcoxon’s test for non-parametric data to examine the association between the three time dependent variables (regulatory review time, EMA review time and manufacturer regulatory response time) and risk of PMSEs. Regulatory review time was examined overall and stratified for the key medicine and application characteristics described previously. Prior to stratifying these analyses, an interaction term between regulatory review time and each key medicine and application characteristic was tested to determine if the key characteristic modified the relationship between review time and risk of PMSEs. In addition, because the novel medicines included in our sample had differing periods of market availability during which post-market safety events could take place, we also plotted Kaplan–Meier curves and report post-market safety events within 2, 5 and 10 years of market authorization. The log-rank test was used to examine the association between the three regulatory review times and occurrence of a PMSE. Finally, we used Chi Square tests to examine the association between near deadline regulatory approval and risk of PMSEs. All tests were two-sided at 0.05 significance level and all analyses have been performed using JMP software version 9.0.1 (SAS Institute, Inc., Cary, NC, USA).

Results

Sample characteristics

Between January 1 2001 and December 31 2010, 161 novel medicines were approved by the EMA through the centralized authorization procedure and included in our sample. Approximately three-quarters were small molecule medicines and one-quarter were biologics, while seven (4.3%) were authorized through an accelerated assessment procedure (Table1). One-third were approved between 2001 and 2005 and two-thirds between 2006 and 2010. Nearly two-thirds were first approved by the FDA. A pre-approval safety concern was identified for three-fifths of our sample of novel medicines. Median regulatory review time, EMA review time and manufacture regulatory response time were 337 days (interquartile range [IQR] 276–406), 219 days (IQR 208–237) and 113 days (IQR 70–173), respectively (Table2, Figure 2).

Table 1.

Characteristics of the 161 novel medicines approved by the EMA between 2001 and 2010, including the prevalence of post-market safety events stratified by key medicine and application characteristics

Characteristics	n (%)	Post-market safety event (%)
Overall cohort	161 (100%)	49 (30.4%)
Year of approval
2001–2005	61 (37.9%)	21 (34.4%)
2006–2010	100 (62.1%)	28 (28.0%)
Therapeutic type
Small molecule	120 (74.5%)	35 (29.2%)
Biologic	41 (25.5%)	14 (34.2%)
First regulator to approve
FDA	102 (63.4%)	35 (34.3%)
EMA	59 (36.7%)	14 (23.7%)
Expected duration of treatment
Acute	26 (16.1%)	9 (34.6%)
Intermediate	45 (28.0%)	17 (37.8%)
Chronic	90 (55.9%)	23 (25.6%)
Orphan designation
Yes	48 (29.8%)	14 (29.2%)
No	113 (70.2%)	35 (31.0%)
Accelerated assessment procedure
Yes	7 (4.3%)	0 (0%)
No	154 (95.7%)	49 (31.8%)
Pre-approval safety concern
Yes	98 (60.9%)	35 (35.7%)
No	63 (39.1%)	14 (22.2%)

EMA review, manufacturer response and regulatory review times for 161 novel medicines approved by the EMA between 2001 and 2010, overall and stratified by whether there was a post-market safety event as well as regulatory review time stratified by key medicine and application characteristics

	Median (interquartile range), days			P value
All medicines	Medicines with safety events	Medicines with no safety event
EMA review time	337 (276–406)	348 (312–401)	331 (267–412)	P = 0.23
Manufacturer response time	219 (208–237)	219 (206–233)	219 (209–240)	P = 0.93
Regulatory review time	1143 (70–173)	117 (86–170)	112 (61–174)	P = 0.57
Regulatory review time, stratified by key medicine and application characteristics
Accelerated assessment pathway
Yes (n = 7)	183 (176–253)	NA	183 (176–253)	NA
No (n = 154)	341 (295–414)	348 (313–401)	337 (288–419)	P = 0.79
Designated by EMA as orphan drug at approval
Yes (n = 48)	348 (274–456)	337 (274–367)	359 (269–491)	P = 0.28
No (n = 113)	337 (276–404)	349 (316–428)	330 (267–401)	P = 0.08
Approved in 2005 or later, after implementation of EMA’s risk management plan
Yes (n = 100)	331 (294–399)	337 (299–371)	330 (279–411)	P = 0.84
No (n = 61)	360 (269–426)	373 (335–449)	331 (250–413)	P = 0.19
Therapeutic type
Small molecule (n = 120)	337 (276–412)	337 (295–380)	331 (267–430)	P = 0.99
Biologic (n = 41)	360 (284–431)	373 (328–485)	331 (269–393)	P = 0.11
Pre-approval safety concern identified by EMA
Yes (n = 98)	358 (295–424)	344 (316–395)	365 (269–449)	P = 0.82
No (n = 63)	330 (268–395)	365 (276–445)	329 (267–393)	P = 0.24
Approved by the FDA prior to EMA approval
Yes (n = 102)	334 (268–408)	358 (316–418)	329 (267–405)	P = 0.13
No (n = 59)	360 (302–405)	341 (306–380)	367 (292–419)	P = 0.56
Medication’s expected duration of treatment
Acute (n = 26)	334 (290–393)	380 (343–448)	315 (268–387)	P = 0.03
Intermediate (n = 45)	331 (258–393)	331 (240–366)	334 (267–412)	P = 0.34
Chronic (n = 90)	348 (293–439)	358 (316–428)	337 (269–449)	P = 0.45

EMA, European Medicines Agency; FDA, Food and Drug Administration; NA, not applicable.

^*P value for difference in median regulatory review time between medicines for which there was and was not a post-market safety event.

†P for interaction test = 0.04.

‡P for interaction test > 0.10.

§Not applicable because no medicines approved via the accelerated assessment pathway were found to have post-market safety events.

Figure 2.

Kaplan–Meier curve of occurrence of safety events according to time after EMA approval

Post-market safety events

Post-market safety events were identified for 49 (30.4%) medicines between January 1 2001 and October 31 2013. Of these 44 (89.8%) were safety DHPCs and five (10.2%) withdrawals. Safety events occurred after a median of 39.5 months (IQR 22.3–55.0) months. Median duration of follow-up after approval was 65.1 months (range 11–152) and more than 90% of the reported PMSEs (n = 45) occurred during the 66 months after approval. At 2, 5 and 10 years following market authorization, PMSEs had been identified for 7.7%, 27.7% and 35.4% of medicines, respectively (Figure 2). No medicine approved through an accelerated assessment procedure was subsequently affected by a DHPC or withdrawal.

Regulatory review time and post-market safety events

There was no association between regulatory review time, EMA review time or manufacturer regulatory response time and subsequent risk of PMSEs when examined as continuous variables (P values > 0.20; Table2). Median regulatory review time among the 49 medicines for which PMSEs were identified was 348 days (IQR 312–401), whereas it was 331 days (IQR 267–412) among the 112 medicines for which no PMSEs were identified (P = 0.23). Median EMA review times for medicines for which PMSEs were identified and not identified were very similar (219 days, IQR 206–233 and 219 days, IQR 209–240, respectively; P = 0.93), as were median manufacturer regulatory response times (117 days, IQR 86–170 and 112 days IQR, 61–174, respectively; P = 0.57).

When review speed was categorized into tertiles, there were differences across regulatory review time tertiles in risk of PMSEs (Figure 3). There were higher rates of PMSEs 5 years after market authorization among medicines with review speeds in the middle tertile (42%) and lower rates among medicines in the lowest tertile (fastest approvals, 21.8%) and highest tertile (slowest approvals, 21.7%); (P = 0.0102; Figure 4). However, there were no differences across EMA review time categories or manufacturer regulatory response time tertiles in risk of PMSEs (P values > 0.44; Figures 5 and 6).

Figure 3.

Occurrence of safety events according to the regulatory time, EMA review time and manufacturer response time divided in tertiles (P value gave results of Chi-square test). , Product with safety event; , Products with no safety event

Figure 4.

Kaplan–Meier curves showing the occurrence of post-market safety communications by regulatory review time among medicines approved by the EMA. , 8–45 weeks; , 46–54 weeks; , 55 weeks or more

Figure 5.

Kaplan–Meier curves showing the occurrence of safety communications after medicines’ approval depending on the time taken by the EMA. , 0–180 days; , 181–210 days; , > 210 days

Figure 6.

Kaplan–Meier curves showing the occurrence of safety communications after medicines’ approval depending on the time taken by the manufacturer regulatory response time. , 0–100 days; , 101–170 days; , > 170 days

Near regulatory deadline approval and post-market safety events

There were 44 medicines (27.3%) that were approved before and 117 medicines (72.7%) approved after the 210 days regulatory deadline. There was no significant difference in risk of PMSEs (34.1% vs. 29.1% respectively, P = 0.38). Among the 44 medicines approved prior to the deadline, 26 (16.0%) were approved in the 30 days prior to the regulatory deadline. Thus, near regulatory deadline approval was not associated with significant differences in risk of PMSEs (38.5% vs. 28.7%; odds ratio 1.55, 95% CI 0.65, 3.72; P = 0.32).

Exploratory analyses

We found no differences in the risk of PMSEs by any of the key medicine and application characteristics we examined (Table1), including approval via the accelerated assessment procedure (0% vs. 31.8%; P = 0.07), orphan designation (29.2 % vs. 31.0%; P = 0.82), the year of approval as a marker of the applicability of the Risk Management Plan instituted in 2005 (34.4% vs. 28.0%; P = 0.39), whether it was a small molecule or biologic (29.2% vs. 34.2%; P = 0.55), whether a medicine was first approved by the FDA or by the EMA (34.3% vs. 23.7%; P = 0.16), whether the drug was indicated for acute, intermediate or chronic duration of treatment (34.6% vs. 37.8% vs. 25.6%; P = 0.31) and presence of a pre-approval safety concern (35.7% vs. 22.2%; P = 0.07). In addition, we found that no key medicine or application characteristic modified the relationship between regulatory review time and post-market safety risk (P for interaction tests > 0.10; Table2), with the potential exception of accelerated assessment pathway (P for interaction test = 0.04).

Discussion

In our study of novel medicines approved by the EMA between January 2001 and December 2010, we found that neither more rapid regulatory reviews nor approvals within 30 days of the regulatory deadline were associated with a higher risk of post-market safety communications or withdrawals. As the agency strives to facilitate rapid access to novel medicines while ensuring public safety through adequate assessment of medication harms, our findings provide some reassurance. In addition, as the EMA’s regulatory review speed varied among the medicines approved, our findings suggest the agency is appropriately taking longer or shorter periods of time in order to evaluate medications adequately, at least with respect to safety, prior to market approval.

Our findings raise two important issues worth considering about the EMA’s efforts to promote timely drug approval. First, we found that many medicines were approved after the EMA’s regulatory deadline. This suggests that there are opportunities to speed approvals further and perhaps explains why EMA’s regulatory review times have been shown to be approximately 2 months slower than the FDA’s 6. Second, the accelerated assessment procedure was rarely used by the EMA, although the FDA’s similarly designed Priority Review pathway has been used successfully in the United States and elsewhere to speed drug approval 10. Of note, there are two non-mutually exclusive approaches that can be used to facilitate earlier access to novel medicines. Accelerated assessment imposes shorter regulatory review deadlines, whereas Conditional Approval shortens development time by allowing medicines addressing unmet needs to reach the market with encouraging clinical data that are not yet comprehensive. The approval is renewed every year until post-market studies have been conducted with confirmatory results thus allowing the EMA to convert the Conditional Approval into a normal one. Of note, we found in our exploratory analyses that use of the accelerated assessment pathway potentially modified the relationship between regulatory review time and post-market safety risk, with an interaction test P value of 0.04. However, there were no PMSEs identified among these seven medications and the finding deserves further scrutiny.

Our study stands in contrast to prior works that demonstrated that medicines approved within 60 days of the deadlines required by PDUFA in the USA were associated with greater risk of post-market safety problems 7,8. However, there are several possible explanations for the discrepancy between USA studies and ours. First, the EMA faces different pressures than does the FDA because new drugs are more often approved first for use in the USA 6, potentially allowing the EMA to have a better understanding of medication risk during pre-market evaluation and modify its assessment in response to actions taken by the FDA. Second, the deadlines followed by the FDA and EMA appear to be operationalized differently. By our estimates, only 25% of medicines were effectively approved by the EMA before its regulatory deadline, whereas 95% or more approval decisions are made by the FDA before its regulatory deadlines 11. Whereas Carpenter et al. had observed a concentration of authorizations granted just before the FDA deadline after implementation of PDUFA 7, this phenomenom was not observed among EMA authorizations, leading to only a few being made within 30 days of the deadline. Third, the threshold for safety communications may differ between the two agencies and the FDA has additional tools at its disposal, including black box warnings, which do not exist in the EU. This dissimilarity between the two regulators is likely to reflect important differences regarding processes related to safety communication. Finally, we focused on the period 2001–2010, as opposed to 1993–2005, and drug safety evaluations may have changed over this time. For instance, some have speculated that after PDUFA, there was a ‘backlog evacuation’ of new drug applications in the USA 12. By examining only more recent approvals, it is possible that we have observed a trend of increased scrutiny and more robust premarket safety evaluations, before medicines are available on the market. Also, conversely to Carpenter et al., our study could not specifically address the comparison of medicine approval before and after the implementation of FDA’s 1992 deadline policy, since this rule predates the creation of the EMA in 1995. In addition, while prior work had used propensity score methods, such an approach was not necessary in our analysis because there were no differences in medication characteristics observed between drugs that did and did not have post-market safety events. Last, another more recent study similarly found an increased risk of black-box warnings and market withdrawals after passage of PDUFA 13. However, this ecological study did not specifically correlate review speed and post-market safety risk. In addition, other factors that were a consequence of PDUFA, such as increased post-market vigilance and enhanced post-market regulatory requirements, may explain their findings.

We found no consistent evidence to suggest that regulatory review speed was associated with post-market safety risk, regardless of whether it was examined by overall regulatory review speed, manufacturer response time or EMA review speed. Also, we found that no key medicine or application characteristic modified the relationship between regulatory review time and post-market safety risk. However, we did find increased risk of safety events among medicines approved with overall regulatory review speeds in the middle tertile, neither quickly nor slowly. While this finding may simply be spurious, there are potential explanations worth considering, such as an adaptation of the EMA to some features of the medicines submitted, influencing the characterization of the safety profile and acceptance. This finding bears further scrutiny.

We used safety communications, including both DHPCs and drug withdrawals, as the primary marker of PMSEs, a method utilized by prior studies on drug safety 7,14–17. However, there are inherent challenges in using these communications. First, no comprehensive listing of DHPCs is available from the EMA. In order to identify DHPCs, we screened and compiled all publicly-available safety communications issued by four of the largest EU national regulators 9, an exhaustive approach that is representative of EU safety communications although we cannot rule out the possibility that other DHPCs were issued by other national regulators within the study period. Second, DHPCs reflect the regulator’s tolerance for drug safety and risk, which might vary among European regulators, as well as among patients and clinicians. Nevertheless, DHPCs represent the best proxy available for post-market drug safety concerns in the EU. Our study was also limited by its focus on medicines approved through the EMA’s centralized authorization procedure. We cannot rule out the possibility that other medicines with different characteristics and risks were submitted and approved through other procedures, such as mutual recognition procedure and decentralized procedure. However, since the centralized authorization procedure is in practice the principal means of approval, this limitation would be expected to have little impact, if any, on our findings.

Our study has other important limitations to consider. First, several factors inherent to the regulatory system can affect both regulatory review time and safety assessment, such as political pressures, pending lawsuits, hidden data and organizational factors, including staff training and resources, but are difficult to quantify. Second, many features of the regulatory process were unchanged during the study period, including the 210 day deadline, but there have also been changes that may have impacted either the regulatory review process or safety assessments, including the introduction of guidelines and reorganization within the EMA. The most prominent change, implementation of the Risk Management Plan in 2005, was not found to have changed the association between regulatory review time and post-market safety. We did not have access to data related to review panels. While our findings were consistent across the 10 year period of study, we cannot know whether there were changes to the review panels and their composition. Third, even over a 10 year time period, there were only 161 medications in our sample, only 26 of which were approved near the regulatory deadline. Although we observed 49 post-market safety events, our study may have had limitations in its power to detect differences in post-market safety risk. Finally, regulatory action is a complex multifaceted process and regulators are likely to be influenced by the actions of another. While we could not observe communications between agencies, particularly whether safety communications issued by the FDA influenced the EMA, we did examine whether the first approval of medicines by the FDA changed the association between EMA regulatory review time and post-market safety and found no effect.

In conclusion, neither more rapid regulatory reviews nor approvals within 30 days of the regulatory deadline of 210 days were associated with a higher risk of post-market safety communications or withdrawals among novel medicines approved by the EMA between January 2001 and December 2010. These findings provide some reassurance that the EMA is able to facilitate rapid access to novel medicines while ensuring public safety through adequate pre-market assessment of medication harms.

Financial support

This project was not supported by any external grants or funds. Drs Downing and Ross receive support from The Pew Charitable Trusts to examine FDA regulatory issues. Dr Ross receives support from Medtronic, Inc. and Johnson and Johnson to develop methods of clinical trial data sharing, from the Centers of Medicare and Medicaid Services (CMS) to develop and maintain performance measures that are used for public reporting, and from the Food and Drug Administration (FDA) to develop methods for post-market surveillance of medical devices and is also supported by the National Institute on Aging (K08 AG032886) and by the American Federation for Aging Research through the Paul B. Beeson Career Development Award Program.

Author responsibility

All authors had full access to all of the data in the study (including statistical reports and tables) and can take responsibility for the integrity of the data and the accuracy of the data analysis.

Competing Interests

All authors have completed the Unified Competing Interest form at http://www.icmje.org/coi_disclosure.pdf. We declare the following interests: Dr Ross reports that he is a member of a scientific advisory board for FAIR Health, Inc. Dr Zeitoun reports he serves as an external consultant for several consulting firms and communication companies linked with pharmaceutical industry (Cepton, Oliver Wyman, McCann Healthcare, Omnicom, Canal 55, Grey Healthcare, Saatchi and Saatchi Healthcare, Sudler & Hennessey, TBWA, inVentiv Health France). He also reports collaboration with Merck, Teva and Menarini, unpaid consultancy for Allurion Technologies and EY, conducting workshops funded by Amgen and being invited to a French medical congress by AbbVie. He is an Associate in House at The Family, an investment company located in London and Paris. Dr Lefèvre reports that he received fees from Vifor Pharma and being invited to an international medical congress by Sanofi. Henri Bergeron reports training activities in the context of Sciences Po University Professional Training Center for Bayer and Sanofi-Pasteur.

Contributors

Drs Zeitoun and Ross were responsible for the conception and design of this work. Dr Zeitoun drafted the manuscript and was responsible for acquisition of data. Dr Lefevre conducted the statistical analysis. Henri Bergeron and Dr Ross provided supervision. All authors participated in the analysis and interpretation of the data and critically revised the manuscript for important intellectual content.