The analyzed dataset contained 1,165 pivotal trials (Table 1). Over 80% were Phase III trials (912), and about 16% were Phase II (171). Less than 2% total were Phase I or Phase IV trials (17 and 2, respectively). Oncology was the most common therapeutic area among the trials (16.2%), followed by anti-infective (14.5%), central nervous system (14.4%), endocrine (12.5%), and cardiovascular (11.9%). Table 1 also shows that more than three-quarters of the trials were part of NDAs, and just over 20% were part of BLAs, slightly more than half of the trials were for drugs that were included in one or more special programs (Priority, Orphan, Fast Track, Breakthrough, or Accelerate Approval), and roughly one-third of the trials were for products sponsored by large companies (> $25B revenue in 2021)15. For disparity comparisons, over 60% of trials had indication-specific sex demographic data, nearly 60% had indication-specific racial demographic data, and just more than 55% had indication-specific ethnicity demographic data. For trials without indication-specific demographic data, disparity comparisons were made to US census data.
Table 1
Trial Data Characteristics
Characteristic | n | Percent |
Phase | | |
| Phase I | 17 | 1.5% |
| Phase II | 171 | 15.5% |
| Phase III | 912 | 82.8% |
| Phase IV | 2 | 0.2% |
Therapeutic Area | | |
| Oncology | 189 | 16.2% |
| Anti-Infective | 169 | 14.5% |
| Central Nervous System | 168 | 14.4% |
| Endocrine | 146 | 12.5% |
| Cardiovascular | 138 | 11.9% |
| Immunologic | 73 | 6.3% |
| Respiratory | 64 | 5.5% |
| Diagnostic | 45 | 3.9% |
| Gastrointestinal | 45 | 3.9% |
| Anesthetic / Analgesic | 42 | 3.6% |
| Other | 86 | 7.4% |
Application Type | | |
| BLA | 246 | 21.1% |
| NDA | 919 | 78.9% |
Special Program | | |
| Any | 600 | 51.5% |
| Priority | 509 | 43.7% |
| Orphan Designation | 361 | 31.2% |
| Fast Track | 299 | 25.7% |
| Breakthrough | 178 | 15.3% |
| Accelerated Approval | 108 | 9.3% |
Company Size | | |
| Small / Medium (< $25B Revenue, 2021) | 789 | 67.7% |
| Large (> $25B Revenue, 2021) | 376 | 32.3% |
Indication-Specific Comparison Data Available | | |
| Sex Demographic Data | 711 | 61.0% |
| Racial Demographic Data | 683 | 58.6% |
| Ethnicity Demographic Data | 644 | 55.3% |
While the availability of sex demographic trial data has increased over each of the three time periods used in this analysis (2007–2011, 2012–2016, and 2017–2021), both racial demographic trial data and ethnicity demographic trial data saw a decrease in availability in the second time period, before increasing in the final time period (Table 2). Regardless of the observed dip, the availability of trial demographic data has increased significantly for all three demographic categories.
Table 2
Availability of Trial Demographic Data Over Time
Demographic Category | 2007–2011 | 2012–2016 | 2017–2021 | Odds Ratio | p-value |
Sex Data Available | 244 (84.4%) | 364 (90.1%) | 445 (97.0%) | 1.186 (1.123–1.252) | < .0001 |
Racial Data Available | 222 (76.8%) | 265 (65.6%) | 421 (91.7%) | 1.103 (1.065–1.142) | < .0001 |
Ethnicity Data Available | 101 (35.0%) | 135 (33.4%) | 292 (63.6%) | 1.137 (1.103–1.172) | < .0001 |
Table 3 displays the average demographic distribution of pivotal trial participants. On average, white participants make up more than three-quarters of trial participants, with Black and Asian participants making up less than 10% of trial participants each (8.6% and 9.6%, respectively). Participants of other racial identities made up less than 5% of trial participants, on average. Hispanic and Latino participants made up more than 13% of trial participants, on average. For each of these identities other than white, the median trial percentage was lower than the mean, indicating that for half of trials, trial percentages were much lower.
Table 3
Demographic Distribution of Pivotal Trials
Demographic Category | n | Mean (CoV) | Median |
Sex | | | |
| Male | 1,055 | 49.9% (0.44) | 52.0% |
| Female | 1,055 | 50.1% (0.44) | 48.0% |
Racial Identity | | | |
| White | 774 | 77.2% (0.25) | 81.4% |
| Black | 774 | 8.6% (1.37) | 4.4% |
| Asian | 774 | 9.6% (1.62) | 3.3% |
| Other Racial Identities | 774 | 4.1% (1.64) | 2.1% |
Ethnic Identity | | | |
| Hispanic / Latino | 473 | 13.1% (0.98) | 9.9% |
Table 4 contains the median demographic disparities for trials. Half of the pivotal trials in this sample overrepresented white participants by about 13% or more. At the same time, half of trials underrepresented Black participants by more than 64%, underrepresented Asian participants by about 32% or more, and underrepresented other racial identities by about 73% or more. Half of the trials also underrepresented Hispanic or Latino participants by about 32% or more.
Table 4
Demographic Disparities of Pivotal Trials
Demographic Category | n | Median |
Sex | | |
| Male | 992 | 1.6% |
| Female | 992 | -2.3% |
Racial Identity | | |
| White | 766 | 12.9% |
| Black | 765 | -64.2% |
| Asian | 566 | -31.7% |
| Other Racial Identities | 597 | -73.3% |
Ethnic Identity | | |
| Hispanic / Latino | 467 | -32.1% |
*Disparity is the difference between the trial demographic and the indication/census demographic percentages, as a percentage of the indication/census demographic. |
Table 5 shows the trial demographic disparities over 3 time periods; 2007–2011, 2012–2016, and 2017–2021. The median disparity for Asian participants has fluctuated between time periods, -68.9% (meaning they were underrepresented by 68.9%) in the first time period, 51.5% (meaning they were overrepresented by 51.5%) in the second time period, and − 29.1% in the final time period, and the median disparity for Hispanic and Latino participants has shown a similar pattern to a lesser degree, -51.6% to -19.9% to -32.1%.
Table 5
Demographic Disparities in Pivotal Trials Over Time
| 2007–2011 | 2012–2016 | 2017–2021 |
Demographic Category | n | Median | n | Median | n | Median |
Sex | | | | | | |
| Male | 231 | 2.2% | 351 | 1.6% | 408 | 1.6% |
| Female | 231 | -2.8% | 351 | -2.1% | 408 | -2.3% |
Racial Identity | | | | | | |
| White | 173 | 15.1% | 241 | 11.9% | 352 | 12.6% |
| Black | 173 | -64.0% | 241 | -70.7% | 351 | -58.7% |
| Asian | 124 | -68.9% | 146 | 51.5% | 296 | -29.1% |
| Other Racial Identities | 129 | -65.9% | 189 | -67.2% | 279 | -80.2% |
Ethnic Identity | | | | | | |
| Hispanic / Latino | 88 | -51.6% | 124 | -19.9% | 255 | -32.1% |
*Disparity is the difference between the trial demographic and the indication/census demographic percentages, as a percentage of the indication/census demographic. |
The results of the regression analyses indicate that disparities have changed significantly over time only for white and Black patients (Table 6). Overrepresentation of white participants has decreased over the 15-year time period, while representation of Black participants has improved over the same time period. Other demographic identities have not shown a significant trend.
Table 6
Regression Analysis of Disparities Over Time
Demographic Category | n | Coefficient | Adj. R-Square | p-value |
Sex | | | | |
| Male | 990 | -0.39 | .0005 | .2163 |
| Female | 990 | 0.32 | .0004 | .2449 |
Racial Identity | | | | |
| White | 766 | -2.49 | .0247 | < .0001 |
| Black | 765 | 3.27 | .0160 | .0003 |
| Asian | 566 | 6.71 | .0021 | .1420 |
| Other Racial Identities | 597 | -1.75 | − .0014 | .6747 |
Ethnic Identity | | | | |
| Hispanic / Latino | 467 | 0.60 | − .0019 | .7598 |
Table 7 displays the percentage of pivotal trials that either underrepresent a demographic by more than 20%, overrepresent the demographic by more than 20%, or accurately represent the demographic. Both male and female participants are accurately represented in approximately half of pivotal trials, however male participants are overrepresented more often than they are underrepresented, and female participants are underrepresented more often than they are overrepresented. White participants are either accurately or overrepresented in nearly 90% of trials. Asian and Hispanic or Latino participants are underrepresented in more than half of trials, and Black participants or participants of other racial identities are underrepresented in about 80% of pivotal trials. Looking specifically at oncology trials, the percent of trials underrepresenting male participants decreases, while accurate representation of male participants increases. Underrepresentation of Black and Hispanic or Latino participants increases, and underrepresentation of Asian participants and participants of other racial identities decreases.
Table 7
Representation in Pivotal Trials
Demographic Category | Pivotal Trials that Underrepresent Demographic by > 20% | Pivotal Trials that Accurately Represent Demographic | Pivotal Trials that Overrepresent Demographic by > 20% |
All Trials | | | |
Sex | | | |
| Male (n = 992) | 22.7% | 50.2% | 27.1% |
| Female (n = 992) | 29.4% | 49.2% | 21.4% |
Racial Identity | | | |
| White (n = 766) | 10.2% | 51.8% | 38.0% |
| Black (n = 765) | 77.8% | 8.9% | 13.3% |
| Asian (n = 566) | 55.1% | 6.2% | 38.7% |
| Other (n = 597) | 80.1% | 7.0% | 12.9% |
Ethnic Identity | | | |
| Hispanic / Latino (n = 467) | 56.5% | 17.1% | 26.3% |
Oncology Trials | | | |
Sex | | | |
| Male (n = 153) | 13.7% | 66.7% | 19.6% |
| Female (n = 153) | 28.1% | 51.0% | 20.9% |
Racial Identity | | | |
| White (n = 134) | 9.7% | 56.0% | 34.3% |
| Black (n = 134) | 89.6% | 5.2% | 5.2% |
| Asian (n = 125) | 37.6% | 8.0% | 54.4% |
| Other (n = 110) | 67.3% | 5.5% | 27.3% |
Ethnic Identity | | | |
| Hispanic / Latino (n = 74) | 78.4% | 14.9% | 6.8% |
Non-Oncology | | | |
Sex | | | |
| Male (n = 839) | 24.3% | 47.2% | 28.5% |
| Female (n = 839) | 29.7% | 48.9% | 21.5% |
Racial Identity | | | |
| White (n = 632) | 10.3% | 51.0% | 38.8% |
| Black (n = 631) | 75.3% | 9.7% | 15.1% |
| Asian (n = 441) | 60.1% | 5.7% | 34.2% |
| Other (n = 487) | 83.0% | 7.4% | 9.7% |
Ethnic Identity | | | |
| Hispanic / Latino (n = 393) | 52.4% | 17.6% | 30.0% |
Table 8 indicates that there has been little change in the frequency with which female participants have been underrepresented in trials over the three time periods examined, nor has there been much change in underrepresentation of participants of other racial identities. Underrepresentation of Black participants does show some improvement in the most recent time period, and underrepresentation of Hispanic or Latino participants showed an improvement in the second time period but has increased again in the most recent, approaching the level it was at during the first time period. Asian participants also saw a decrease in underrepresentation in the second time period, which has since increased again, though not to the same degree as the first time period.
Table 8
Underrepresentation Over Time
Demographic Category | Percent of Pivotal Trials that were Underrepresentative 2007–2011 | Percent of Pivotal Trials that were Underrepresentative 2012–2016 | Percent of Pivotal Trials that were Underrepresentative 2017–2021 |
Sex | | | |
| Male | 22.5% | 19.7% | 25.0% |
| Female | 31.2% | 29.9% | 28.2% |
Racial Identity | | | |
| White | 11.6% | 10.0% | 9.7% |
| Black | 81.5% | 82.2% | 72.9% |
| Asian | 75.0% | 40.4% | 54.1% |
| Other | 78.3% | 77.3% | 82.8% |
Ethnic Identity | | | |
| Hispanic / Latino | 63.6% | 50.0% | 57.3% |
Table 9
Association Between Demographic Disparities and Trial Duration
Demographic | n | p-value of model (r-sq.) | p-value of Year Started | p-value of Therapeutic Area | p-value of Number of Countries | p-value of Disparity |
Sex | 633 | < .0001 (.1712) | < .0001 | < .0001 | .0036 | |
| Male | | | | | | .7715 |
| Female | | | | | | .6306 |
Racial Identity | 344 | < .0001 (.1272) | .0077 | < .0001 | .2898 | |
| White | | | | | | .3501 |
| Black | | | | | | .0350 |
| Asian | | | | | | .2686 |
| Other Identity | | | | | | .7432 |
Ethnic Identity | 325 | < .0001 (.1715) | < .0001 | < .0001 | .1580 | |
| Hispanic / Latino | | | | | | .3439 |
Regression models which include the year the trial started, therapeutic area of the product (oncology or non-oncology) and the number of countries included in the trial, indicate that representation of male and female participants are not significant predictors of trial duration (Table 8). When trial duration is modeled using representation of racial demographics instead of sex, only representation of Black participants was a significant predictor, indicating that as representation of Black participants increases, trial duration decreases. No other demographic disparities were found to be significant predictors of trial duration.
Similar models of FDA review duration were conducted, using the year the product was approved, therapeutic area (oncology or non-oncology), and enrollment in a special program, as well as the average demographic disparity for individual demographic subgroups across a product’s pivotal trials (Table 10). These models indicated that demographic disparities do not have a significant impact on FDA review duration.
Table 10
Association Between Demographic Disparities and FDA Review Duration
Demographic | n | p-value of model (r-sq.) | p-value of Year Approved | p-value of Therapeutic Area | p-value of Special Program | p-value of Avg. Disparity |
Sex | 471 | < .0001 (.1122) | .0545 | .0113 | < .0001 | |
| Male | | | | | | .1480 |
| Female | | | | | | .3336 |
Racial Identity | 291 | < .0001 (.1091) | .5804 | .0064 | .0008 | |
| White | | | | | | .3999 |
| Black | | | | | | .1403 |
| Asian | | | | | | .3578 |
| Other Identity | | | | | | .6335 |
Ethnic Identity | 259 | < .0001 (.0893) | .8922 | .0413 | .0004 | |
| Hispanic / Latino | | | | | | .6851 |