Long-Term Prognosis of Patients with Ulcerative Colitis in Japan

Even though it is important in improving the quality of life to evaluate the long-term prognosis of patients with ulcerative colitis in comparison with the general population, it is unknown in detail. One hundred and seventeen cases followed-up for 10 years or more were evaluated to define the long-term prognosis of ulcerative colitis by the person-years method. The estimated number of death (E) was 14.5, and the observed number of death (0) was 20. The O/E ratio was 1.38 and confidence interval was 0.84-2.13, showing no significant difference between E and O. Evaluation of change in th O/E ratio at 5-year intervals revealed a decrease in both the males and females, with a significant difference observed between the ratios in 1960 and 1965 (each p<0.05). Generally, the death rate was significantly higher in the patients with ulcerative clitis than in the general population between 1953 and 1965 but did not significantly differ thereafter . On the other hand, the E from malignant tumors was 3.94, and the 0 was 4; the O/E ratio was 1 .02 and 95% confidence interval was 0.27-2.60, showing no significant difference between E and O. J Epidemiol, 2000 ; 10 : 48-54

With recent advances in the methods for diagnosing and treating ulcerative colitis (UC), the number of patients with UC with a long-term course has been increasing Thus, it is important in terms of improving their quality of life to evaluate the long-term prognosis of these patients in comparison with the general population.
We evaluated the outcomes of patients with this disease, and analyzed their long-term prognos is by the person-years method.

Subjects and follow-up
Study subject were 117 patients (70 male and 47 female patients) in the consecutive series who were diagnosed as ulcerative colitis at the First Department of Internal Medicine in Hirosaki University Hospital during the period from January 1, 1953 to August 31, 1981.
The mean age at the time of the onset was 34.0 years (35.1 years in the males and 33.2 years in the females), showing a peak between 20 and 30 years in both males and females (Table 1).The patients were classified according to the extension of the lesion, the severity, and the disease type based on the clinical course by "The Diagnostic Criteria for Ulcerative Colitis" I .Number of patients according to these clinical characteristics are shown in Table 2.

1) Survival analysis
The analysis of the survival rate was performed by the Kaplan-Meier method.The patients' survival was examined as regards the time of onset (before the year 1970 or thereafter) and clinical characteristics such as the extension of the lesion , severity, and clinical course as mentioned above .Because at our department, salazopyrine (SASP) had become to be generally used, and total colonoscopy had been introduced in 1970 .

2) Mortality in comparison with general population
In the analysis of the long-term prognosis by the personyears method, observation was initiated on the day of the diagnosis and terminated on August 31, 1991, when the outcome of all patients could be confirmed.The observed numbers of death (0) were compared with the expected numbers of deaths (E), which were computed by multiplying age-(5 years groups), sex-, and calendar-(5 years intervals) specific personyears at risk by the corresponding death rates in Aomori Prefecture.The death rates were calculated for five successive periods with 5-year intervals from 1953-1991 with use of mortality data for the Aomori Prefecture, which were available in the Aomori Health Statistical Report '), and with the use of the census population data '.The ratio of the observed number of deaths (0) to E (O/E ratio) was calculated.
In addition, the 0 and E was obtained according to the final observation year classified at 5-year intervals --1960, 1965, 1970, 1975, 1980, 1985 and 1990, and the O/E ratio was calculated.
In addition, the 0 and E due to malignant tumor in the patients with UC were obtained by the same method as mentioned above, and their ratio (O/E ratio) was calculated.
3) Statistical analysis Significant differences between E and 0 were determined using Bailar's table 7), assuming the Poisson distribution in term of the ratio of 0 to E values in each group of subjects.Differences at a p value of less than 5% were considered significant.

Clinical characteristics of patients with a long-term course 1) Medical treatment
Medical treatment was performed in 78 patients (43 males and 35 females), of whom 11 (14.1%)died.

2) Surgically treated patients
Enterectomy was performed in 39 patients (27 males and 12 females; 33.3%) because of poor responses to medical treatment or massive bleeding.The mean interval between the onset and surgery was 2 years with a range of 2 months to 19 years.More than 50% of the surgically treated patients underwent surgery within 2 years of the onset (Figure 1).
Surgery was performed in 25 (54%) of the 46 patients in Group A and 14 (20%) of the 71 in Group B.
3) Complications The complications of UC were classified as systemic or local.Number of causes developing local complications in the intestine were: massive bleeding 3, perforation 3, toxic megacolon 1. Perianal abscess and adenoma were also observed I case, respectively.There were no patients showing canceration.Of systemic complications, anemia and hypoproteinemia were frequently observed.The other complications included ureterolithiasis (4 patients), cerebral infarction (3), cerebral hemorrhage (2), myocardial infarction (2), and schizophrenia (3).Malignant tumors were noted in 8 patients (7%): gastric cancer in 2, and, in 1 each, liver cancer, pancreatic cancer, lung cancer, pharyngeal cancer, uterine cancer, and malignant lymphoma.The mean interval between the onset of UC and the development of malignant tumors was 17 years and 1 month.4) Deaths Twenty patients (16 males and 4 females) died.The mean age at the time of death was 55.2 years.The interval from the onset to death ranged from 1 year, 1 month to 32 years (mean, 10 years, 1 month) (Figure 2).The cause of death was the primary disease in 5 patients, pulmonary edema as a complication  that developed during treatment of the primary disease in 2, and other diseases in 13, including a malignant tumor in 4, heart disease in 2, and cerebrovascular disease in 2.
In Group A, 5 patients died within 3 years of the onset, and all of their causes were associated with the primary disease.In Group B, 2 patients died of perforation and pulmonary edema associated with the primary disease within 3 years after the onset.9 patients in Group A died and 4 in Group B died since 3years.The cause of death was unrelated the primary disease.

Analysis of the long-term prognosis 1) Survival rate
The survival rate was significantly higher (p<0.05) in Group B than in Group A. However, the survival rate did not significantly differ among the 3 or 2 groups classified according to the extension of the lesion, according to the severity of UC, or according to the clinical course (Figure 3).2) Person-years method (Table 3) The total E was 14.5 (11.7 in the males and 2.8 in the females), and the total 0 was 20 (16 in the males and 4 in the females).The mean O/E ratio was 1.38 (1.37 in the males and 1.44 in the females), showing no significant difference between E and O. Therefore, the presence of UC itself did not appear to affect the long-term prognosis.
In Group A, the total E was 9.2 (7.8 in the males and 1 .4 in the females) and the total 0 was 14 (10 in the males and 4 in the females); the mean O/E ratio was 1.52 (1.28 in the males and 2.84 in the females).In Group B, the total E was 5.3 (3.9 in the males and 1.4 in the females), and the total 0 was 6 (6 in the males and 0 in the females); the mean O/E ratio was 1.13 (1.54 in the males and 0 in the females).No significant difference was observed between E and 0 in either period, but the O/E ratio was higher in the earlier period than in the later period.
The evaluation of changes in the O/E ratio for the 5-year intervals in 1960, 1965, 1970, 1975, 1980, 1985 and 1990 as the final year of observation revealed that these ratios were higher than 1.0 in all years.But these ratios tended to decrease 3a 3b 3c 3d Figure 3. 3a: The survival rate was significantly higher (p<0.05) in Group B than in Group A.
3bc: The survival rate did not significantly differ among the 3 groups classified according to the extension of the lesion or the severity of UC.
3d: The survival rate did not significantly differ among the 2 groups classified according to the clinical course.The total E from malignant tumors was 3.94 (3.01 in the males and 0.93 in the females), and the total 0 was 4 (4 in the males and 0 in the females); the mean O/E ratio was 1.02 (1.33 in the males and 0 in the females) and 95% confidence interval was 0.27-2.60,showing no significant difference between E and O. Therefore, the death rate from malignant tumors in the UC patients during the long-term course appears to be similar to that in the general population.

DISCUSSION
In the earlier period, medical treatment methods had not been adequately established, and the diagnostic techniques were not well developed.Thus, some patients in the earlier group may not have been diagnosed until surgery, and some, who could have been readily diagnosed and medically treated with presently available resources, may also have been treated by enterectomy.The interval between the onset and surgery was within 2 years in 76% of the surgically treated patients in Group A but in only 21% of those in Group B .Thus, the survival rate was more markedly affected by the time of the onset than by the state of the disease.
A high incidence of cancer of the large intestinal in UC patients after a long course has been reported since the 1960s.Canceration was reported to occur in about 3% of all patients.In particular, the risk for canceration increases 10 years or more after the onset.A recent study revealed a canceration rate of 3% after 15 years, 5% after 20 years, and 9% after 25 years 10.Suzuki et al s) performed endoscopy 261 times in 90 patients with UC and observed dysplasia in 20 patients (22%) including 3 with Dukes A cancer, suggesting the usefulness of endoscopic examination.However, none of the patients treated at our department showed canceration.
Nation-wide statistics issued by the Research Group of the Ministry of Health and Welfare in 1990 cited a death rate of 3.3% in UC patients10).At our department , of 117 patients with UC, 20 (16 males and 4 females) died (17.1%).This marked difference may be associated with the follow-up period and follow-up rate.In the patients treated at our department , the observation period was longer, and the follow-up rate was higher (100%) than those in the nation-wide statistics .There were more patients who died of other diseases in our study .
In this study, 7 patients (6.1%) died of the primary disease; perforation, peritonitis, or postoperative death was observed in 5 patients, and pulmonary edema that developed during treatment of the primary disease in 2. In addition, these 7 patients died within 3 years after the onset, and 5 of them were treated before 1969.In recent years, the number of patients treated with emergency surgery has decreased due to advances in diagnostic techniques including colonoscopy and advances in the medical treatment of severe cases such as high-dose steroid therapy, intraarterial prednisolone administration, and parenteral nutrition, and the death rate from UC has markedly decreased.Therefore, the 7 patients who died may have been cured had they been treated at present.In all patients who died 5 years or more after the onset, the cause of death was other diseases such as malignant tumors (4 patients), heart diseases such as myocardial infarction (1), and cerebrovascular disease (2).These results show that deaths from UC are few at present with advanced treatment techniques.
In western countries, a large-scale survey by the Organization Mondiale de Gastroenterologie revealed a death rate of 3.7% between 1974 and 1984 in UC patients 10 years after the onset "i.When this death rate is compared with that in studies appearing the 1930s and 1972, a marked improvement is observed.In the 1950s, the death rate 10 years after the onset was reported to be 55%12).In the 1960s, the death rate was reported to be less than 12% due to the decreased number of deaths in severe cases after the introduction of corticosteroids 11.In the present study, comparison of the survival rates disclosed a significantly higher survival rate 10 years after the onset in Group B (94%) than in Group A (84%).The analysis of the data for the patients who died suggested that this difference was associated with diagnostic and treatment techniques.On the other hand, Inoue et al.17) noted a higher risk for relapse and surgery, and poorer prognosis in patients with severer disease or more extensive lesion at the time of the onset.In this study, the state of UC affected the hospitalization period and enterectomy status but not the prognosis.
For accurate analysis of the prognosis of UC, comparison with the natural history in the general population is necessary.Hendriksen16) compared data for UC patients and the general population.However, the results of his study are not convincing because the two groups were not matched for sex, the age at the time of the initiation of observation, or the initiation time of observation.In our study, the patient groups and the general population were compared by the person-years method that allows matching at the individual level for all three of these variables.
The evaluation of changes in the OlE ratio for the 5-year intervals in 1960, 1965, 1970, 1975, 1980, 1985 and 1990 as the final year of observation revealed that these ratios were higher than 1.0 in all years.But these ratios tended to decrease annually, and significant diferences were observed in only 1960 and 1965 (each p<0.05).
The observed number of deaths from malignant tumors during the long-term course was similar to the estimated number of deaths.Therefore, it can not be concluded that UC patients frequently die of malignant tumors during the long-term course.
The prognosis of UC was not good until 1965.Since there are established diagnostic and treatment methods at present, the death rate in UC patients including that from malignant tumors is similar to that in the general population.

Figure 1 .
Figure 1.Numbers of surgically-treated patients: interval between the onset and operation.

Figure 2 .
Figure 2. Cause of death.The interval from the onset to death ranged from I year , I month to 32 years (mean, 10 years, I month).
13.14) Sonnenberg et al. 15) analyzed the statistics for deaths due to inflammatory intestinal diseases in England and the United States and found a marked decrease in the death rate from UC starting from the 1950s in each generation, suggesting differences from Crohn's disease showing a decrease in the death rate from the 1970s.In 1985, Hendriksen et al.") carried out a follow-up survey with a mean follow-up period of 7 years and reported similar survival rates in the patient and control groups.

Table 1 .
Distribution of study patients by sex and age class.
a Age at oncet

Table 2 .
Clinical characteristics of study patients.