Results

Previous methods were applied to colorectal cancer. Data were collected by the SEER Program [SEERD03] during the 1990-1999 period. The material consists of 23334 cases that were followed until the end of 1999. The follow-up cut-off date for this analysis was 31 December 1999. Survival time was defined as the time from diagnosis to death if prior to 1999 or to the end of 1999 . Overall status was defined as unity if the subject was dead before the end of 1999, and zero otherwise. The population used for calculating prevalence rates was provided by SEER program [SEERD03], as well as the U.S mortality rates for the population [SEERM03] and the size of population [SEERP03]. Mortality rates are published in 5-year age intervals.

Table 1 shows the age-specific incidence of colorectal cancer in Connecticut between 1990 and 1999 used for estimating age-specific prevalences by the Transition Rate Method.

For the parametric model, estimates of the parameters of incidence are b equals 5 and a equal L51-12. The estimates of the proportion of cured cases and the parameter of survival of uncured cases were performed using S-plus

Table 1. Table 1 : Colorectal cancer observed incidence rates by age in Connecticut between 1990 and 1999. Rates are per 100,000.

Age range Rate (SE)

40-44

13.7 (0.7)

45-49

28.9 (1.1)

50-54

53.6 (1.6)

55-59

97.1 (2.5)

60-64

163.2 (3.5)

65-69

247.8 (4.4)

70-74

327.9 (5.3)

function (nlminb in order to minimize the negative of the log of the likelihood function). The estimate of proportion of cure is about 63% corresponding to the standard error of 0.028 and the parameter of the exponential survival is about 0.56 year-1 corresponding to the standard error of 0.026.

The age-specific prevalence for people diagnosed between 1990 and 1999 are presented in table 2. The Transition Rates Method estimate of age-specific 10-year partial prevalence of colorectal cancer for people in the 55-59 age range is 466.6 per 100,000. The estimated standard error, 11.1, corresponds to a coefficient of variation of 2.3% [D93]. The parametric method and the counting method yield prevalence estimates in people aged 55-59 of 427.8 and 419.2, respectively, with corresponding standard errors of 11.8 and 15.5 and corresponding coefficients of variation of 2,7% and 3.7%.

Table 2. Colorectal cancer estimates 10-year partial prevalence by age in Connecticut in 1999 using the Transition Rate Method, the Parametric Method and the Counting Method. Rates are per 100,000.

Age range

TRM (SE) PM (SE)

CM (SE)

40-44

61.1 (3.2) 82.9 (1.9)

52.7 (4.3)

45-49

127.7 (4.8) 152.3 (3.8)

121.1 (6.7)

50-54

253.2 (7.4) 262.1 (6.9)

212.7 (9.7)

55-59

466.6 (11.1) 427.8 (11.8)

419.2 (15.5)

60-64

790.1 (15.1) 668.9 (19.1)

734.1 (23.6)

65-69

1223.6 (19.4) 1008.2 (29.9)

1192.9 (31.8)

70-74

1664.6 (22.5) 1473.6 (45.3) 1673.9 (38.24)

As regards the age-specific prevalence of non-recovery, the time required to be cured is assumed to be 5 years. Table 3 presents the estimates of non recovery 10-year partial prevalence of colorectal cancer with the corresponding standard error. The Transition Rate and parametric methods yield estimates of non recovery prevalence in the age range 55-59 of 225.9 and 188.1, re spectively, with corresponding standard errors 6.7 and 5.9. The coefficient of variation are, respectively 2.9% and 3.1%.

Table 3. Colorectal cancer estimates non recovery 10-year prevalence by age in Connecticut in 1999 using the Transition Rate Method and the Parametric Method. Rates are per 100,000.

Age range Estimated non cure prevalence

Table 3. Colorectal cancer estimates non recovery 10-year prevalence by age in Connecticut in 1999 using the Transition Rate Method and the Parametric Method. Rates are per 100,000.

Age range Estimated non cure prevalence

TRM

PM

40-44

29.4 (2)

45.9 (1.2)

45-49

63 (2.7)

77.7 (2.1)

50-54

125.2 (4.6)

123.9 (3.7)

55-59

225.9 (6.7)

188.1 (5.9)

60-64

375.9 (9.4)

274.7 (9.4)

65-69

571.7 (11.4)

388.4 (14.1)

70-74

761.7 (12.7)

534.3 (20.6)

Figure 2 shows the estimates of prevalence and non recovery prevalence using all previous methods. For the under sixties, the estimates of 10-year partial prevalence are similar and for the over sixties, the Parametric Method provides estimates lower than using the two other methods. As regards the non recovery 10-year partial prevalence estimates, the same trend is noted, the Transition rate method provides estimates higher than using the parametric method.

Using the TRM and the parametric method, the non recovery 10-year prevalence represents around 45% of the prevalence (Figure 3). This point is consistent with the estimate of the proportion of cure (63%).

Was this article helpful?

0 0

Post a comment