Collectively, we characterized the disease trajectory of over one million patients with IBD, including 462,502 CD and 589,118 UC individuals. Tables 1, 2 and Supplementary Tables 2–4 present information on the size, age and sex distributions, and observation time of individuals in the CD, UC, IBD and IBDU cohorts, respectively, in each database (and, as context, compare to each database characteristic in Supplementary Table 1); extended tables are available online (https://data.ohdsi.org/IbdTable1/). Figure 1 shows the percentage of individuals with first IBD-related event (diagnosis or prescription) within each index-year stratum, indicating that most patients, in nearly all DBs (with the exception of MDCR and KDH), had their recorded onset event no earlier than 2010. Of note, temporal and age coverage differ considerably between DBs (see Supplementary Tables 1).

Percentage of patients with incident IBD, per database (y-axis), within each index-year stratum (x-axis); left: Crohn’s disease, right: ulcerative colitis

The percentage of female patients with incident CD and UC in Asia (Japan’s JMDC, South Korea’s AUSOM and KDH: 27.4%–38.6% and 35.6%–39.8%, respectively; Table 1 and 2) is considerably lower than in the European and USA cohorts (53.8%–65.2% and 49.1%–65.9%; and see Supplementary Tables 5 and 6 for female percentage in index-age strata); as well as in Asian sub-populations at western countries (CD: 53.2% in AMB-EMR and 57.5% in Optum-EHR; UC: 47.3% in IMRD-UK, 52.8% in AMB-EMR, and 54.1% Optum-EHR). CD female proportion decreased over time in Europe, USA, and Australia (e.g., CCAE: 59.1% and 51.5% in 2000 and 2020, respectively) but increased in Japan (JMDC: 21.4% in 2010 to 30.5% in 2020; Supplementary Table 7); UC female to male ratio decreased in most US databases (noticeably, MDCD and JHM) but remained stable in Germany, UK, and Japan (Supplementary Table 8).

The most common cohort index event, in both CD and UC, is the corresponding primary diagnosis, accounting for 25–80% and 28–84.6% of cases, respectively (Supplementary File 2); followed by more specific diagnoses, e.g., CD of small intestine or CD of large bowel (e.g., up to 19.5% and 16%, respectively, in AUSOM) and chronic ulcerative pancolitis (19.7% in JHM) or ulcerative proctocolitis (18.1% in IMRD-UK). Specific formulations—in particular of mesalamine, prednisone, and prednisolone—have been prescribed, in some databases, to at most 25% of individuals on the index date.

Overall, there has been a noticeable decrease in the average age at which CD is diagnosed across various populations and regions, as indicated in Supplementary Table 9. This trend is particularly prominent in Europe and North America (e.g., average age-of-diagnosis decreased from 47.4 years in the 2005–2010 period to 44.3 years after 2020 in the Optum-EHR database). In Japan, age at CD diagnosis has been stable (average age of 33.7 years in the 2010–2015 period and 33.9 years after 2020). A similar, though weaker, decrease in diagnosis age has been observed in the UC cohorts (Supplementary Table 10); e.g., average age of diagnosis decreased from 52.5 years in the 2005–2010 period to 49.5 years after 2020 in the Optum-EHR database and 52.2 to 45.1 years in 2005–2010 and after 2020, respectively, in the MDCD database.

Next, we investigated the symptoms recorded for patients in the year leading up to their IBD diagnoses. In US insurance claims databases, as well as in IMRD-UK, about one-third of patients with CD had a diagnostic code of abdominal pain during that timeframe (e.g., 38.1% in the Optum-CDM dataset in 2020), while much lower rates have been recorded in all other databases (e.g., 15.5% in GERMANY; Supplementary Table 11). Overall, abdominal pain prevalence increased over time, though in some cases only modestly. Similar trends have been observed in UC, with the exception of IMRD-UK abdominal pain rates being much lower in patients with UC than CD (15.1% versus 28.8% in 2015–2020; Supplementary Table 12).

For rectal bleeding among patients with UC, considerable differences were noted between individual datasets (Supplementary Table 13), such as 31.8% and 34.7% in IMRD-UK versus 11.7% to 11.1% in the MDCR, during 2000–2005 and 2015–2020, respectively. While rates increased in most databases over the years, some showed an opposite trend (e.g., Optum-CDM decreasing from 20.7% in 2000–2005 to 14.8% in 2015–2020).

Psychological symptoms prior to IBD diagnosis were also examined. Across most European and US datasets, there was an increase in the proportion of patients receiving an anxiety diagnosis before their CD diagnosis (Supplementary Table 14). In the IMRD dataset, the prevalence rose from 13.8% in 2000 to 27.3% during 2015–2020, and in the Optum-CDM dataset it increased from 21.3% to 29.3% during the same periods. Conversely, the increase in depression diagnosis rates was less pronounced (23.9% to 35.4% and 26.5% to 29.2%, respectively, for the same populations; Supplementary Table 15) and several datasets exhibited a decrease in depression rates over time.

Anxiety diagnosis rates also increased among patients with UC since the early 2000s across various regions and populations (Supplementary Table 16), with some databases nearly doubling in prevalence. Despite substantial differences among datasets, this upward trend remained consistent: 5.1% and 21.6% in 2000–2005 to 12.3% and 33.6% after 2020, in the CUIMC and Optum-CDM datasets, respectively. Depression rates, though less prominent, also showed an increase among these populations over the past two decades (6.9% and 27.2% to 14.9% and 34.1% from 2000 to 2020 in CUIMC and Optum-CDM, respectively; Supplementary Table 17).

Sex differences in diagnosis rates of both physical and psychological symptoms were evident among patients with IBD across the datasets. Specifically, for CD (Supplementary Table 18), anxiety was diagnosed in 8.7% to 44.8% of females, whereas in males, it ranged from 6.9% to 32.8%, and the proportion of females with anxiety was consistently higher across all datasets (Paired t-test P-value < 0.001). Similarly, for depression, the range was 10.1% to 46.4% in females and 6.9% to 32.9% in males, again showing a higher proportion among females in all datasets (Paired t-test P-value < 0.001). Recorded complaints of abdominal pain ranged from 7.1% to 51.2% among females and 7.2% to 44.4% among males, with most datasets indicating higher proportions among females (Paired t-test P-value < 0.001). Diagnosis of diarrhea occurred at similar proportions between sexes, ranging from 0.7% to 29.9% in females and 0.5% to 25.9% in males, with no significant differences noted in the individual datasets (Paired t-test P-value = 0.1).

Similar patterns were identified in the UC population (Supplementary Table 19), wherein anxiety and depression diagnosis rates ranged from 10.7% to 48.8% and 8.4% to 50.7% among females, and 7.5% to 39.8% and 7.9% to 41.2% among males, respectively. In all datasets, a higher proportion of anxiety and depression diagnoses were observed in females (for both diagnoses, paired t-test P-value < 0.001). Abdominal pain occurrence varied from 6.5% to 48.7% in females and 4.8% to 42.6% in males across the datasets, consistently demonstrating a higher proportion among females (Paired t-test P-value < 0.001). Conversely, diarrhea diagnosis rates were similar between sexes across all datasets, ranging from 0.8% to 31.6% in females and 0.9% to 32.8% in males (Paired t-test P-value = 0.3).

Regarding surgical outcomes for patients with CD, most US databases (with the exception of AMB-EMR) demonstrated stable or increasing rates of partial colonic resection within one year from diagnosis, over the past two decades (e.g., 7.9% and 8.9% in the CCAE database and 11.2% to 14% in the MDCR database, during 2005-2010 and 2015-2020, respectively; Supplementary Table 20). Notably, rates in IMRD-UK decreased considerably from 6.9% in 2000–2005 to 4.1% in 2015–2020. A similar trend is observed when examining the colonic resection rates 3 years after diagnosis (CCAE: 10.3% to 12.8%, MDCD: 10% to 13.7%, and, conversely, IMRD-UK: 11% to 6.7%, in 2005 and after 2015, respectively; Supplementary Table 21).

Small bowel resection rates had demonstrated similar course over time among patients with CD: stable or increasing in most US databases (e.g., 3.3% and 4.1% at the early 2000s, 5.3% and 4.4% in 2015–2020 in the CCAE and MDCR data sets, respectively; Supplementary Table 22); but decreasing in AMB-EMR (1% in 2005–2010 to 0.4% in 2015–2020), IMRD-UK (8% in 2000–2005 to 4.7% in 2015–2020) and JMDC (4.8% in 2010–2015 to 2.4% after 2020). These same trends were kept when examining small bowel resection rate 3 years after diagnosis (Supplementary Table 23).

Partial colectomy rates for patients with ulcerative colitis varied over time and across cohorts. Within one year from diagnosis (Supplementary Table 24), resection rates increased between 2000 and 2020 (e.g., MDCR database from 8.8% in 2000 to 14.1% in 2020). When examining colectomy rates within three years following diagnosis (Supplementary Table 25), we observed a similar trend: In the CUIMC database, colectomy rate was 5.8% in 2000–2005 and 7.2% in 2015–2020; in the Optum-CDM cohort, the rates were 12.8% and 16.9% for 2000 and 2020, respectively. The most significant increase in colectomy rates was evident in the CCAE and MDCR cohorts, where rates were 10.1% and 15.6% in 2000, respectively, and then increased to 14% and 20.5% by 2020.