Descriptive analysis
The normal test of the general information in three groups showed that age does not fit the normal distribution (supplementary file1). So we used median and quartile range to describe the concentration and dispersion trend of age. The median of age of the case group was 63, 25 percentiles and 75 percentiles were 55 and 72, respectively. And the corresponding values of negative control group and positive control group were shown in Table 1. Then we used chi square test to compare the factors (gender, age, and past medical history) among the three groups. As shown in Table 2, among the three groups, the proportion of patients with hypertension in case group, negative control group and positive control group were 16.8%, 18.3% and 18.4%, respectively, and the chi square value was 2.391, p value was 0.303, the difference was not statistically significant. In addition, the difference of coronary disease among the three groups was not statistically significant. These evidence indicated that there was no difference in the history of hypertension and coronary disease among the three groups. However, there were significant differences in age, gender and diabetes history among the three groups.
Table 1
Concentration and dispersion trend of age of patients with cholangiocarcinoma and control subjects.
Age(years) |
Group | Median | Percentile(25) | Percentile(75) |
Case group | 63 | 55 | 72 |
Negative control group | 56 | 43 | 72 |
Positive control group | 61 | 53 | 70 |
Table 2
Chi square analysis of the factors (gender, age, hypertension, diabetes, coronary disease) among the three groups.
Group | Case group (n = 1749) | Negative control group (n = 2863) | Positive control group (n = 5087) | χ2 p value |
N | Percentage (%) | N | Percentage (%) | N | Percentage (%) | |
Age Gender | 240.564 | 0.000 |
184.847 | 0.000 |
Male | 1043 | 59.6% | 1317 | 45.3% | 3074 | 60.4% | |
Female | 706 | 40.4% | 1591 | 54.7% | 2013 | 39.6% |
Hypertension | 2.391 | 0.303 |
Absent | 1455 | 83.2% | 2375 | 81.7% | 4149 | 81.6% | |
Present | 294 | 16.8% | 533 | 18.3% | 936 | 18.4% |
Diabetes | 12.914 | 0.002 |
Absent | 1607 | 91.9% | 2626 | 90.3% | 4525 | 89.0% | |
Present | 142 | 8.1% | 282 | 9.7% | 561 | 11.0% |
Coronary disease | 1.674 | 0.433 |
Absent | 1692 | 96.7% | 2801 | 96.3% | 4887 | 96.1% | |
Present | 57 | 3.3% | 107 | 3.7% | 200 | 3.9% |
Chi-square Test Of The Rate Of Cholecystectomy
As shown in Table 3, the number of patients with cholecystectomy in the case group was 239, accounting for 13.7%; the number in the negative control group was 164, accounting for 5.6%; the number in the positive control group was 515, accounting for 10.1%. The chi square value was 88.586 and p value was 0.000, the difference of the rate of cholecystectomy among the three groups was statistically significant. The data indicated that effect of cholecystectomy on cholangiocarcinoma is different from that on the other diseases (colon cancer and pancreatic cancer, femoral fracture and hepatic hemangioma). Furthermore, the rate of cholecystectomy was compared in the case group and the positive control group, and compared in the case group and the negative control group separately, and the test level was adjusted with the partitions of X2 method. Comparing the rate of cholecystectomy between the case group and the positive control group, the chi square value was 16.631 and p value was 0.000, which means the difference was statistically significant (Table 4). Similarly, the difference of the rate of cholecystectomy between the case group and the positive control group was statistically significant too (Table 5).
Table 3
Chi square analysis of the rate of cholecystectomy among the three groups.
Group | | Cholecystectomy | χ2 p value |
Absent | Present |
| | 88.586 | | 0.000 |
Case group | N | 1510 | 239 | |
Percentage | 86.3% | 13.7% |
(%) | | |
Negative control group | N | 2744 | 164 |
Percentage | 94.4% | 5.6% |
(%) | | |
Positive control group | N | 4572 | 515 |
Percentage | 89.9% | 10.1% |
(%) | | |
Sum | N | 8826 | 918 |
Percentage | 90.6% | 9.4% |
(%) | | |
Table 4
Chi square analysis of the rate of cholecystectomy between the case group and the positive control group.
Group | | Cholecystectomy | χ2 p value |
Absent | Present |
| | 16.631 | | 0.000 |
Case group | N | 1510 | 239 | |
Percentage | 86.3% | 13.7% |
(%) | | |
Positive control group | N | 4572 | 515 |
Percentage | 89.9% | 10.1% |
(%) | | |
Sum | N | 6082 | 754 |
Percentage | 89.0% | 11.0% |
(%) | | |
Table 5
Chi square analysis of the rate of cholecystectomy between the case group and the negative control group.
Group | | Cholecystectomy | χ2 p value |
Absent | Present |
| | 88.985 | | 0.000 |
Case group | N | 1510 | 239 | |
Percentage | 86.3% | 13.7% |
(%) | | |
Negative control group | N | 2744 | 164 |
Percentage | 94.4% | 5.6% |
(%) | | |
Sum | N | 4254 | 403 |
Percentage | 91.3% | 8.7% |
(%) | | |
Logistic Univariate Regression Analysis
Then we analyzed a series of confounding factors such as gender, age, past medical history (hypertension, diabetes, coronary disease, cholecystolithiasis) and cholecystectomy by logistic univariate regression analysis, from which we screened out the differential factors among the groups for the subsequent logistic multivariate regression analysis. As shown in Table 6, we can conclude that except for coronary disease there were significant differences in gender, age, cholecystectomy, cholecystolithiasis, hypertension and diabetes among the groups. These evidence indicated that all these factors besides coronary disease may affect the incidence of cholangiocarcinoma. So we included all the factors except coronary disease into the logistic multivariate regression analysis to observe which were the risk factors of cholangiocarcinoma actually.
Table 6
Logistic univariate regression analysis.
Group | likelihood ratio test |
-2 Log likelihood | χ2 | DOF | p |
Gender | 6244.024 | 235.361 | 2 | 0.000 |
Age | 6293.053 | 284.391 | 2 | 0.000 |
Cholecystectomy | 6126.656 | 117.993 | 2 | 0.000 |
Cholecystolithiasis | 6252.102 | 243.439 | 2 | 0.000 |
Hypertension | 6028.325 | 19.662 | 2 | 0.000 |
Diabetes | 6026.402 | 17.739 | 2 | 0.000 |
Coronary disease | 6014.403 | 5.741 | 2 | 0.057 |
DOF: degree of freedom |
Logistic Multivariate Regression Analysis
Logistic multivariate regression analysis between the case group and the positive control group.
Firstly, we compared the six factors (gender, age, cholecystectomy, cholecystolithiasis, hypertension and diabetes) between the case group and the positive control group by logistic multivariate regression analysis. As shown in Table 7, we find that the OR value of gender was 0.976 (95%CI = 0.871–1.093) and p value was 0.661, the difference was not statistically significant. Because there were more male patients in case group and positive control group, it can be considered that male patients were more likely to suffer from cholangiocarcinoma, colon cancer and pancreatic cancer than female patients. Although the difference in age was statistically significant (0.000), however the OR value was 1.011 (95%CI = 1.007–1.016), basically showing no significant difference. Since the median age in the case group and the positive control group were 63 and 61, respectively, indicating that the risk of cholangiocarcinoma, colon cancer and pancreatic cancer increased with age. Patients with hypertension were more likely to suffer from colorectal cancer and pancreatic cancer than cholangiocarcinoma. The OR value of hypertension was 0.851 (95%CI = 0.730–0.991) and p value was 0.037. That is to say, compared with the incidence of colon cancer and pancreatic cancer, hypertension has less influence on the incidence of cholangiocarcinoma. Next, we compared the diabetes between the case group and the positive control group. The OR value of diabetes was 0.675 (95%CI = 0.553–0.824) and p value was 0.000. The data showed that patients with diabetes have a slightly higher risk of colorectal and pancreatic cancer than those with cholangiocarcinoma. Refer to the difference of cholecystectomy, the risk of cholangiocarcinoma in patients with cholecystectomy was 1.553 times higher than that of colon cancer and pancreatic cancer (OR = 1.553, 95%CI = 1.311–1.840), and the difference was statistically significant. And the risk of cholangiocarcinoma was higher in patients with cholecystolithiasis, which was 2.46 times higher than that of colorectal cancer and pancreatic cancer (OR = 2.460, 95%CI = 2.093–2.890).
Table 7
Logistic multivariate analysis between the case group and the positive group.
| 95% CI |
Group | B | SE | Wald | DOF | P | OR | Lower | Upper |
Gender | -0.025 | 0.058 | 0.183 | 1 | 0.669 | 0.976 | 0.871 | 1.093 |
Age | 0.011 | 0.002 | 27.438 | 1 | 0.000 | 1.011 | 1.007 | 1.016 |
Cholecystectomy | 0.440 | 0.086 | 25.917 | 1 | 0.000 | 1.553 | 1.311 | 1.840 |
Cholecystolithiasis | 0.900 | 0.082 | 119.512 | 1 | 0.000 | 2.460 | 2.093 | 2.890 |
Hypertension | -0.162 | 0.078 | 4.331 | 1 | 0.037 | 0.851 | 0.730 | 0.991 |
Diabetes | -0.393 | 0.101 | 14.997 | 1 | 0.000 | 0.675 | 0.553 | 0.824 |
B: regression coefficient, SE: standard error, DOF: degree of freedom, OR: odds ratio |
Logistic multivariate regression analysis between the case group and the negative control group.
Subsequently, we compared the six factors (gender, age, cholecystectomy, cholecystolithiasis, hypertension and diabetes) between the case group and the negative control group by logistic multivariate regression analysis. As shown in Table 8, we find that the OR value of gender was 0.484 (95%CI = 0.427–0.549) and p value was 0.000, the difference was statistically significant. Which means compared with the incidence of cholangiocarcinoma, women were more likely to suffer from hepatic hemangioma or femoral fracture. Although the difference in age was statistically significant (0.000), however the OR value was 1.037 (95%CI = 1.032–1.042), basically showing no significant difference. Since the median age in the case group and the negative control group were 63 and 56, respectively, indicating that the risk of cholangiocarcinoma, hepatic hemangioma and femoral fracture increased with age. Patients with hypertension were more likely to suffer from hepatic hemangioma or femoral fracture than cholangiocarcinoma. The OR value of hypertension was 0.688 (95%CI = 0.580–0.817) and p value was 0.000. Then we compared the diabetes between the case group and the negative control group. The OR value of diabetes was 0.659 (95%CI = 0.527–0.824) and p value was 0.000. The data showed that patients with diabetes have a higher risk of hepatic hemangioma or femoral fracture than those with cholangiocarcinoma. Refer to the difference of cholecystectomy, the risk of cholangiocarcinoma in patients with cholecystectomy was 3.181 times higher than that of hepatic hemangioma or femoral fracture (OR = 3.181, 95%CI = 2.561–3.951), and the difference was statistically significant. And the risk of cholangiocarcinoma was higher in patients with cholecystolithiasis, which was 5.426 times higher than that of hepatic hemangioma or femoral fracture (OR = 5.426, 95%CI = 4.325–6.809).
Table 8
Logistic multivariate analysis between the case group and the negative group.
| 95% CI |
Group | B | SE | Wald | DOF | P | OR | Lower | Upper |
Gender | -0.726 | 0.064 | 127.686 | 1 | 0.000 | 0.484 | 0.427 | 0.549 |
Age | -0.036 | 0.002 | 224.211 | 1 | 0.000 | 1.037 | 1.032 | 1.042 |
Cholecystectomy | 1.157 | 0.111 | 109.559 | 1 | 0.000 | 3.181 | 2.561 | 3.951 |
Cholecystolithiasis | 1.691 | 0.116 | 213.362 | 1 | 0.000 | 5.426 | 4.325 | 6.809 |
Hypertension | -0.374 | 0.087 | 18.301 | 1 | 0.000 | 0.688 | 0.580 | 0.817 |
Diabetes | -0.417 | 0.114 | 13.388 | 1 | 0.000 | 0.659 | 0.527 | 0.824 |
B: regression coefficient, SE: standard error, DOF: degree of freedom, OR: odds ratio |
Comparison between cholecystectomy and cholecystolithiasis by Chi-square test in case group
Based on logistic multivariate regression analysis, we found that patients with cholecystolithiasis were more likely to suffer from cholangiocarcinoma than those with cholecystectomy. Therefore, we compared cholecystectomy and cholecystolithiasis by Chi-square test in the case group to determine if there was any difference between them. As shown in Table 9, the number of patients with cholecystolithiasis in the case group was 304, accounting for 17.4%; the number of patients with cholecystectomy in the case group was 239, accounting for 13.7%. The chi square value was 45.063 and p value was 0.000, the difference between cholecystectomy and cholecystolithiasis was statistically significant. These data showed patients with cholecystolithiasis has a higher risk of cholangiocarcinoma than those with cholecystectomy.
Table 9
Comparison between cholecystectomy and cholecystolithiasis by Chi-square test in case group.
Group | Cholecystolithiasis | Cholecystectomy | χ2 | p |
| 45.063 | 0.000 |
Presence | 304(17.4%) | 239(13.7%) | |
Absence | 1445(82.6%) | 1510(86.3%) |
Distribution of time interval between cholecystectomy and the diagnosis of cholangiocarcinoma the first time after cholecystectomy
Based on the above evidence, we can conclude that cholecystectomy has a close relationship with the incidence of cholangiocarcinoma. To further elucidate whether the risk of cholangiocarcinoma related to the interval between cholecystectomy and the diagnosis of cholangiocarcinoma the first time after cholecystectomy, we divided the interval into three groups: “one to five years” group, “five to ten years” group, and “more than ten years” group. As shown in Table 10, the number of the patients in the “one to five years group” was 105, accounting for 43.93%; the number of the patients in the “five to ten years group” was 63, accounting for 26.36%; and the number of the patients in the “more than ten years group” was 71, accounting for 29.71%. We compared the “one to five years group” with another two groups separately, and the difference was statistically significant. These data indicated that the incidence of cholangiocarcinoma is the highest within one to five years after cholecystectomy.
Table 10
Distribution of time interval between cholecystectomy and the diagnosis of cholangiocarcinoma the first time after cholecystectomy.
Time interval | Number | Proportion | P |
1–5 years | 105 | 43.93% | |
5–10 years | 63 | 26.36% | |
> 10 years | 71 | 29.71% | |
| 0.000 |