Truelove and Witts first described the use of oral corticosteroids in the treatment of ulcerative colitis in 1955 [27] and again demonstrated a similar efficacy with intravenous steroids for the treatment of acute severe colitis [28]. The natural history of patients with UC who require corticosteroids is largely unknown with very few population-based studies. Generally active UC patients requiring steroids show three different patterns of response—one group will have good results achieving steroid-free prolonged remission, a second group will have initial response but lose benefit as treatment is tapered or stopped, and a third group who will have no response to steroids. In Olmsted County, Minnesota, among an inception cohort of 183 UC patients from the prebiologic era, 63 (34 %) received corticosteroid therapy—54 % achieved complete remission, 30 % had partial remission, and 16 % had no response at all after 30 days [29]. Approximately 1 year after corticosteroid initiation, 49 % had noted prolonged response, 22 % were steroid dependent, and 29 % had undergone colectomy [29]. Similar response rates were noted in a hospital-based cohort study of 136 UC patients from Edinburgh, United Kingdom—51, 31, and 18 % had complete, partial, or no response to steroid therapy at 30 days, respectively [30]. A recent Italian study reported early (3 months) and late outcomes (5 years) in 157 patients with UC who required their first systemic steroid therapy within 12 months of diagnosis [31]. Female gender was the only predictor of a better clinical outcome at 3 months, and a complete clinical and endoscopic remission at 3 months after corticosteroid therapy significantly predicted the decreased risk of hospitalizations, immunosuppressive usage, and colectomy at 5 years [31]. It may be relevant therefore to routinely assess the disease response endoscopically after 3 months of corticosteroid therapy in order to risk stratify patients and to plan for early introduction of immunosuppressive treatments in those who fail to achieve endoscopic remission. Antitumor necrosis factor (TNF) agents like infliximab are effective in inducing and maintaining remission in patients with moderately to severely active UC, but the steroid-sparing effects of these biological drugs will need additional prospective studies.

The natural course of ulcerative colitis is that of exacerbation interspersed with periods of remission. Hospitalization therefore accounts for a large part of the costs involved in the care of ulcerative colitis [32, 33]. Recent data from the Nationwide Inpatient Sample suggests that medical hospitalizations for UC in the United States have increased from 56,911 in 1998 to 86,611 in 2007 [34]. In a recent study, patients with UC who required hospitalization for medical reasons were five times more likely to require colectomy, even after adjusting for other factors [35]. Another important factor that may predict hospitalization is the early endoscopic response to corticosteroids. Patients who failed to achieve clinical and endoscopic remission at 3 months after corticosteroid treatment were more likely to undergo hospitalization within 5 years of follow-up (hazard ratio, 3.634; 95 % CI, 2.193–51.039; P = 0.0033) [31]. Among Olmsted County residents diagnosed with UC between 1970 and 1999, corticosteroids were given during 53 % of UC-related hospitalizations, and 33 % of these patients required more than one hospitalization for inpatient steroids [36]. However, there has been considerable variability in hospitalization rates across North America. In a population-based study from Canada, approximately one-fifth of UC patients had more than one hospital stay every year during the period between 1994 and 2001 [37]. Overall the hospitalization rate for UC was stable at 12.6–13.3 per 100,000 population during the 7-year study period [37]. Approximately 55 % of UC-related hospitalizations involved major surgery, of which the commonest was incision, excision, and anastomosis of the intestine [37]. A multilevel study conducted among 3.2 million members of Kaiser Permanente (Northern California) found that hospitalization rates for UC declined by 29 % between 1998 and 2005 [38]. In the Olmsted County population-based study, 270 UC patients were followed up for 3,458 person-years and 114 (42 %) were hospitalized at least once during the median follow-up of 12.3 years—33 % of all these hospital admissions were surgical [36]. Crude hospitalization rates (per 1,000 patient-years) decreased from 124 for 1970–1980 to 71 for 1990–2001. The 10-year cumulative risk of hospitalization was 49 % in those with extensive colitis compared to 33 % and 29 % for left-sided disease and proctitis, respectively (Fig. 3.2). Hospitalization data in UC could therefore provide vital information on the cohort of patients that might have an aggressive clinical course and appropriate treatment could be initiated from the outset in these cases.

The cancer risk in UC was recognized as early as the 1920s [39, 40]. Although the true risk of CRC in UC in the modern era remains uncertain, it is probably far lower than previously estimated. The risks reported in studies from tertiary referral centers, which often include patients with disproportionately severe disease, generally overestimate the cancer risk [41–43]. For example, in a 1971 study of 396 children, the risk of colon cancer was 20 % for every decade of life beginning 10 years after the disease diagnosis [43]. A meta-analysis of 116 studies from a wide array centers and geographic sites estimated the cumulative risk of CRC in UC to be 1.6 % at 10 years, 8.3 % at 20 years, and 18.4 % at 30 years [44]. The cumulative increase in CRC risk with time detected in this meta-analysis is probably explained by the inclusion of variety of studies with different designs including referral center studies. Population-based studies from Sweden [45, 46] and Israel [47] have shown increased relative risks of CRC in UC ranging from 1.4 to 6. However, some of these studies have been based on patients diagnosed as far back as the 1920s. Conversely, several recent population-based cohort studies have reported no significant increased risk of CRC when compared to the background population [48–51]. For example, in a population-based study from Copenhagen County, 1,160 UC patients were observed for a total of 22,290 person-years with a median follow-up period of 19 years (range, 1–36 years) and were found to have no increased risk of CRC [48]. A total of 13 patients developed CRC within the study period compared to the expected number of 12.42 (standardized morbidity ratio, 1.05; 95 % CI, 0.56–1.79) [48]. The cumulative probability of CRC was 0.4 % by 10 years, 1.1 % by 20 years, and 2.1 % by 30 years of disease [48]. Similarly, in Olmsted County, Minnesota, 378 UC patients with 5,567 person-years of follow-up for the study period 1940–2001 had a cumulative incidence of CRC of 0.4 % at 15 years and 2.0 % at 25 years after UC diagnosis [49]. The number of colon cancers was not increased in the study patients when compared to background population [49]. In this same study, none of the patients in Olmsted County who were diagnosed with UC after 1980 had developed CRC [49]. A slightly higher risk of CRC was found in a population study from Hungary—the cumulative risk of CRC was 0.6 % after 10 years, 5.4 % after 20 years, and 7.5 % after 30 years [52]. The presence of dysplasia on any colonic biopsy (before CRC diagnosis), disease duration of over 10 years, extensive colitis, and coexisting PSC all significantly increased the risk of colon cancer [52]. It is obvious that the incidence of CRC in UC is decreasing over time. Whether this is due to the potentially chemoprotective effect of widespread maintenance therapies including aminosalicylates [53], more aggressive endoscopic surveillance regimens, or more aggressive surgical intervention strategies remains unclear.

Many patients with UC will need surgery during the course of their disease. Early population-based studies from Europe reported 10-year cumulative colectomy rates of over 20 % [23, 24, 50]. In a large inception cohort of 1,586 patients with UC from Stockholm County, the cumulative colectomy rate was 20 % at 5 years, 28 % at 10 years, and 45 % at 25 years [24]. Ten percent of all patients had a colectomy during the first year after diagnosis, while the rates of colectomy decreased in the subsequent years [24]. The extent of disease at diagnosis was the main factor affecting the colectomy rates; the 5-year, 10-year, and 25-year colectomy rates for patients with pancolitis were 32 %, 42 %, and 65 %, respectively [24]. A similar study from Copenhagen County reported a 25-year colectomy rate of 32.4 %, and the disease extent at diagnosis was predictive of subsequent colectomy, with extensive colitis increasing the risk of surgery substantially [50].

In Olmsted County, 316 incident cases of UC were followed up for 3,698 person-years. Overall, 53 patients (17 %) underwent surgery for UC and of these, 70 % underwent more than one surgical procedure [54]. The cumulative risk of colectomy from diagnosis was 16.6 % at 10 years and 19.6 % at 20 years (Fig. 3.3) [54]. Male gender and early need (<90 days) for steroids were significantly associated with time to colectomy. The most common surgeries undertaken for UC in Olmsted County were total proctocolectomy with ileal pouch-anal anastomosis (TPC-IPAA) (62 %) and TPC with ileostomy (30 %) [54]. The 10-year cumulative risk of subsequent unplanned surgery was 44.1 % overall, but the risk varied depending on the primary surgery undertaken. Patients who underwent TPC-IPAA were almost twice more likely to have unplanned follow-up surgeries when compared to TPC-ileostomy [54]. More recent estimates of 10-year colectomy rates in UC from European population cohorts have been lower than previously reported [19, 55, 56]. Patients in the European Collaborative Study Group of Inflammatory Bowel Disease (EC-IBD) were recruited during a 2-year period between October 1, 1991 and September 30, 1993 from 20 treatment centers and distributed over 12 European countries [55]. The overall cumulative 10-year colectomy rate was 8.7 %; in the northern centers (Denmark, Norway, and Netherlands), the colectomy rates were significantly higher, 10.4 %, compared to 3.9 % in the southern centers of Europe (Greece, Israel, Italy, and Spain) [55]. This geographic difference in colectomy rates could indicate that patients in the northern centers have more severe disease compared to the southern centers (36.3 % of patients in the northern centers had extension of their colitis at follow-up compared to 28.7 % in the southern centers, P < 0.05). Among the IBSEN cohort, 423 patients completed 10-year follow-up, and the crude colectomy rate was 3.5 %, 7.6 %, and 9.8 % at 1, 5, and 10 years, respectively [19]. An ESR ≥30 mm and extensive colitis at diagnosis were the only independent risk factors for colectomy [19].

Recent data has suggested that patients who are treated with purine antimetabolites had decreased elective colectomy rates. In contrast, emergent colectomy rates were stable, which was hypothesized to have been due to rapid progression of disease activity [57].

Superimposed infections remain a risk factor for hospitalizations and colectomy in patients with UC. A recent study from the Cleveland Clinic illustrated that patients with C. difficile infection had significantly more UC-related emergency room visits in the year following initial infection (37.8 % vs. 4 %) in addition to significantly higher rates of colectomy 1 year following the initial infection associated admission (35.6 % vs. 9.9 %) compared to those patients who did not have infection [58]. Another report from Mount Sinai Hospital in New York highlighted that the rate of UC-related hospitalizations (58 visits vs. 27 visits, P = 0.001) and colectomy rates (44.6 % vs. 25 %, P = 0.04) 1 year after initial hospitalization were higher in patients who were afflicted with C. difficile [59] infection than those without infection.

Population-based studies on the long-term survival of patients with UC have shown some conflicting data, with older studies reporting a reduced overall survival [60, 61], while newer ones showing either an equivalent or even an improved survival [62–65]. In addition, two large Swedish population-based studies comprised of patients with UC between the 1950s and 1980s reported a slightly increased mortality [66, 67]. In the largest population-based study by Ekbom et al. from Uppsala, Sweden, the standardized mortality ratio (SMR) for UC patients was 1.4 (95 % CI, 1.2–1.5), while the presence of coexisting respiratory diseases like bronchitis, asthma, and emphysema further increased their mortality risk (SMR, 1.5; 95 % CI, 1.1–2.2) [67]. Similar findings were noted by Persson et al. reporting on a cohort of UC patients between 1955 and 1984 from Stockholm, Sweden [66]. However, more recent studies from North America and Europe have shown that the overall mortality from UC is not greater than that of the general population. This may be due to favorable disease pattern or to improvements in the treatment of UC in the recent decades. For example, in Olmsted County, an inception cohort of 378 UC patients had decreased mortality when compared to the general population (SMR, 0.80; 95 % CI, 0.6–1.0) [62]. When stratified by calendar year at diagnosis, survival appeared to improve over time—SMR was 0.80 for patients diagnosed between 1940 and 1959 which improved to 0.50 for the period between 1990 and 2001 [62]. Similarly, an EC-IBD study comprised of UC patients reported an overall mortality risk no higher than the general population (SMR 1.09) [68]. There was, however, a trend toward higher mortality for northern European centers (SMR 1.19) compared to the southern centers (SMR 0.82) [68]. In a meta-analysis of all population-based inception cohort studies in UC, the overall SMR was 1.1 (95 % CI, 0.9–1.2; P = 0.42) [69]. Although the overall mortality in UC did not differ from the background population, certain subgroups were at greater risk of dying. Mortality was significantly increased in patients during the first few years after diagnosis, in those with extensive disease and patients from Scandinavian countries [69]. The pooled SMR for the five Scandinavian studies included in the meta-analysis was 1.2, compared to 0.8 for non-Scandinavian studies [69]. The UC-related mortality accounted for 17 % of all deaths, and cause-specific analyses revealed increased mortality from respiratory diseases, colorectal cancer, gastrointestinal, and liver diseases, while mortality from pulmonary cancer was decreased [69]. Similarly, a study on IBD subjects drawn from Kaiser Permanente Medical Care Program, California, found no overall increase in mortality with UC (SMR, 1.0; 95 % CI, 0.9–1.2), but they had a higher risk of dying from digestive diseases other than IBD (SMR, 3.9; 95 % CI, 2.4–6.0) (e.g., liver diseases and colorectal cancer) [70]. Nonetheless those UC patients needing immunosuppressive treatment with azathioprine, 6-mercaptopurine, methotrexate, or infliximab have a higher mortality compared to those who did not, and this probably reflects the increased usage of these treatment regimes in patients with more severe disease [62, 69, 71]. In a study using the General Practice Research Database (GPRD) from the United Kingdom, the mortality in UC patients was increased with current usage (HR, 2.81; 95 % CI, 2.26–3.50) and recent usage (HR, 2.49; 95 % CI, 1.65–3.75) of corticosteroids [71].