| Objectives: To evaluate the clinical effects of different medical treatment projects for type 2 diabetes mellitus using multiple dimensions of evaluating index system involved clinical effect indexes, medical cost and quality of life, and to explore the effects of missing values on the validity of clinical trials and to compare the appropriate methods handling the missingnees. Synthetical evaluation methods such as data envelopment analysis and TOPSIS method were introduced to assess and optimize effectiveness of clinical trials. This study also was designed to provide physicians with the options to make decision about the priority of clinical treatment projects.Methods: From Dec, 2002 to Dec, 2003, A randomized controlled clinical trial was conducted to follow-up the patients with type 2 diabetes mellitus, who had outpatient care services at the Second Xiangya Hospital of Central South University in Hunan province of China, and were at completely random allocated to different medical treatment projects into conventional therapy, follow-up once three months or intensive therapy with the control target of hemoglobin Alc (HbAlc) below 7.0%, once a month, and three drug treatment projects including sulfonylureas, biguanides and combination of the above two drugs. Questionnaire developed by this study and consulted by an expert panel of clinic physicians and epidemiology was used to collect the data of different treatment projects including baseline characteristics, clinical effects, Quality of Life and medication costs at the beginning of the study and the following periods of 3 months, 6 months, 9 months and 12 months respectively. Medication costs were measured in form of direct cost comprising drug, inspect and transportation costs. The continuous data expressed by geometric mean and geometric standard deviation were analyzed by t-test, ANOVA and nonparametric test and frequency or proportion of the categorical data byχ~2 test in order to compare the baseline characteristics and medical outcomes among treatment projects. Methods such as complete-case analysis, mean substitution, last observation carried forward, expectation-maximization and multiple imputation were used to analyze missing values for 2-h postprandial blood glucose (PBG2h) during the 6-month, 9-month and 12-month follow-up periods respectively. Health economic evaluation methods such as the cost-effectiveness analysis and cost-utility analysis were used in analyzing economic effects of the treatment projects. TOPSIS method and data envelopment analysis were conducted to evaluate the comprehensive treatment effects asosiciated with the inputs and outputs of different treatment projects on type 2 diabetes. The combining method of canonical correlation and data envelopment analysis was performed to make full rank on medical effects over three drug treatment projects. We used the SPSS version 11.5 software to set up the dataset. All analyses were performed using SAS version 9.13 and SPSS version 11.5. All of the statistical tests that we reported were two sided, and the "statistically significant" implied P≤0.05.Results: 1) Baseline characteristics: A total of 183 outpatients (male 89, female 94, mean age of 56.34 years) with newly type 2 diabetes mellitus were recruited from a provincial level teaching hospital from Dec, 2002 to Dec, 2003. 90 patients were received intensive therapy and 93 patients were received conventional therapy, and 36 patients had medication for sulfonylureas, 45 for biguanides and 102 for combination of the above two drugs, respectively. There was homogeneity in baseline characteristics about sociodemographics, household income and the severity of health condition over different treatment projects, except the differences in hemoglobin Alc among three drug therapies.2) Clinical medical effects: The decreased trends on geometric mean of fasting blood glucose (FBG), PBG2h and hemoglobin Alc (HbAlc) among different treatment projects including conventional treatment and intensive treatment had been taken place during the 1-year follow-up period (P＜0.05). There were statistically significant differences of the geometric mean of PBG2h (7.84 mmol/L vs. 9.11mmol/L) and HbAlc (5.42% vs. 5.86%) at the end-point of follow-up between intensive treatment and conventional treatment (P＜0.05) and the lower level of the geometric mean was reported in intensive treatment. The difference for the average level of FBG (6.43mmol/L vs. 6.18mmol/L) was not significant statistically within those two groups. The statistically significant differences of the geometric mean for PBG2h were found among sulfonylureas (6.84mmol/L), biguanides (8.13mmol/L) and combined treatment of sulfonylureas and biguanides (8.84mmol/L), and the level of the geometric mean of PBG2h in sulfonylureas treatment project was lower than that of another projects (P<0.05). There were no statistically differences in the geometric mean of FBG and HbAlc among those three drug treatment projects.3) Quality of life: The patients with intensive treatment had higher level of the scores of quality-adjusted life-years (QALYs) than that of patients with conventional treatment after the 1-year therapy, and the difference in scores of QALYs was statistically significant between intensive treatment (0.8477) and conventional treatment (0.8119), P<0.05. The patients with intensive treatment acquired more scores of QALYs. The quality of life among patients with medication of sulfonylureas, biguanides and combined treatment of the two drugs had been improved before-after treatment (P<0.05) but the no statistically differences in scores of QALYs were found among those three drugs treatment projects.4) Health economic evaluation: During the 1-year follow-up period, total annual medical cost per outpatient with conventional and intensive treatment was 1403.78 and 1460.83 Yuan respectively, and the expenditure resulted from the drug accounted for the main proportion (about 80%) of the total medical cost. The medical cost of reduction of FBG per 1 mmol/L, PBG2h per 1 mmol/L and per 1 unit HbAlc in patients with intensive treatment amounted to 777.04, 315.51 and 632.39 Yuan respectively, and the cost effectiveness ratios of FBG, PBG2h and HbAlc in patients with intensive treatment were all less than that of patients with conventional treatment. Compared to conventional treatment project, the increment cost effectiveness ratios to the much more reduction of 1 mmol/L FBG; PBG2h and 1 unit HbAlc in patients with intensive treatment were 259.32, 54.33 and 158.47 Yuan respectively. The cost utility ratios of QALYs were 1723.29 and 1729.01 Yuan in the intensive treatment and the conventional treatment respectively, which had no statistically significant (P>0.05) between those two groups. The increment cost effectiveness ratios to the much more reduction of 1 unit QALYs for patients with intensive treatment was 1593.58 Yuan contrasted to ones with conventional treatment.The annual medical costs per outpatient with the medication of sulfonylureas, biguanides and combined treatment of the two drugs were 1334.27, 1117.96 and 1636.79 Yuan respectively. The medical cost of reduction in FBG per 1 mmol/L among the three drug treatment projects amounted to 644.59, 540.48 and 1016.64 Yuan separately and the cost per 1 mmol/L PBG2h reduction was 196.22, 310.54 and 426.25 Yuan, and the cost per 1 unit HbAlc reduction was 717.35, 646.22 and 693.56 Yuan, respectively. Cost effectiveness ratios of FBG, PBG2h were highest in the patients with combined therapy of two drugs. Lowest cost effectiveness ratios of FBG, and HbAlc were reported in patients with the medication of sulfonylureas and lowest cost effectiveness ratios of PBG2h in outpatient with the medication of biguanides. The cost utility ratios of QALYs were 1607.55, 1352.48 and 1965.64 Yuan in the treatment of sulfonylureas, biguanides and combined treatment of the two drugs, respectively, and the lowest cost utility ratios of QALYs were found in the treatment project of biguanides.5) Handling the missing values: The hypothesis of missing completely at random (MCAR) for the missing data in this study was not rejected using Little's test for MCAR (x~2=4.127, P=0.903). The handling results for missing data of PBG2h were not fully consistent among different handling methods such as complete-case analysis, last observation carried forward (LOCF), mean substitution, Expectation-Maximization (EM) and multiple imputations (MI). The results of imputation among those methods, especially for EM and MI, were very close to that of complete-case analysis when the proportion of missingness was less than 10%. It was noticed that there was a moderate difference compared complete-case analysis to other methods such as LOCF and mean substitution when the proportion of missingness was greater than 50%, which implied that handling missingness had no practical benefits on data analysis in clinical trials. The difeerences in basic characteristics between subjects with complete data and subjects with missing data were not observed, which allows the results from complete data analyses as final findings reported in this study.6) Comprehensive evaluation: From the analyzing results of TOPSIS method, the overall medical effect for multiple outcomes index in outpatients with intensive treatment was better than that of ones with conventional treatment. At the same time, the overall medical effect for multiple outcomes index in outpatients with sulfonylureas treatment was ranked the top, followed by biguanides and combined treatment of the two drugs. The results of data envelopment analysis (DEA) indicated that the project of intensive treatment had reached a DEA efficiency compared with conventional treatment, and the projects of sulfonylureas and biguanides had also reached DEA efficiency among three drug treatment projects. The combined results of canonical correlation analysis and data envelopment analysis showed that the comprehensive medical effect in outpatients with biguanides treatment was ranked the first place, followed by sulfonylureas and combined treatment of the two drugs.Conclusions: We conclude that medical effects of intensive treatment for type 2 diabetic outpatients in elderly people are more economical and effectual and quality of life in those peoples have been much improved. Sulfonylureas are more effectual drugs to be used in the treatment of type 2 diabetes than that of biguanides. Researchers have to pay attention to missing value issue in clinical trials and do well handling missingness by appropriate method. Multiple outcomes including economic index, clinical index and quality of life introduced in clininal tirals can ben used in analyzing the all-aroud treatment effects for type 2 diabetic patients. Comprehensive evaluation method such as data envelopment analysis and TOPSIS can be well performed in clinical trials and which is introduced to assess the overall medical effects and to optimize the medical treatment project, combined with health economic assessment. |
PDF Full Text Request