Drug treatment of depression randomized double-blind trial
balance of Empirical Research on the baseline
Author: Yan Yan Hua Lin Zhang belt Rong Fu Ying
【Abstract】 Objective: To understand the drug treatment of depression, double-blind randomized controlled clinical
trials, two groups of baseline indicators of the balance. Methods: "China Hospital Digital Library," Journal of CNKI full text database searching literature, describe the difference between two groups the distribution of baseline
indicators, line sign test, and Meta analysis to determine the difference statistically significant. Results: In 82 selected trials, two groups the number of cases, age, sex ratio and duration of disease more balanced view of the differences in
the distribution, while the value of the distribution of NAMD seems uneven. Sign test, and qualitative analysis of Ridit tests showed that the test group is greater than the value of HAMD-based study of the control group than the control group
is greater than the number of test group the number of differences between two groups statistically significant differences in the distribution. Meta analysis of quantitative analysis showed that the test group than the control value of HAMD-based group Big 0.10,95% CI 0.047 ~ 0.153. Conclusion: The trial group and the control group there is the value of the HAMD-based imbalances; randomized double-blind clinical
trials there may also be uneven baseline situation.
【Key Words】 bias in the baseline depression of baseline information, double-blind trial randomized controlled trial Meta-analysis of empirical evidence-based medicine
In the randomized controlled double-blind clinical trial,
baseline irregularity may be caused by one of the reasons bias
test results. Of a randomized double-blind clinical trial in
the test group and control group baseline indicator values (pre-intervention in critical condition severity, demographic and other prognostic indicators that may affect the value) for
comparisons and statistical analysis showed two very small between the two groups there are significant differences. However, a class of randomized double-blind trial of the trial
group and the control group to describe the baseline index
value, comparison and statistical analysis, may be able to find some problems. Published by the domestic drug treatment of depression, double-blind randomized controlled clinical trials, trial group and the control group the baseline, including the number of cases, age, sex ratio and the Hamilton Depression Rating Scale (HAMD) base value such as baseline comparison and analysis are available from one aspect of the evaluation of the quality of double-blind study group, and to
explore the bias may cause the results of certain factors.
1 Data and methods
1.1 Literature selection and exclusion criteria
Inclusion criteria: Chinese medicine treatment of depression or depressive disorder, double-blind randomized
controlled clinical trial; two kinds of programs to compare test; in order to observe the HAMD scale as the main indicators of measurement tests; from 1994 to 2005 literature.
Exclusion criteria: random assignment of cases not expressly literature; not expressly for the double-blind trial
operation; In addition to a 1:1 ratio, the proportion of the literature does not explicitly configured; three groups or three groups over the study, but the other group was open study, except; self-control study; not briefed the two groups,
the mean baseline HAMD literature; respectively introduced the two groups before and after excluding cases of the mean baseline HAMD (4 group mean) and literature; to safety evaluation as the main indicators of literature; Asia Group analysis or further follow-up class literature; has been
incorporated into a multi-center trial of the sub-point test;
foreign literature; clear summary of class literature; test for non-drug measures in the intervention group literature.
1.2 method of collection of literature
Database: "China Hospital Digital Library"
(192.168.200.235) in the CNKI periodical entire library. Search Type: depression (keywords) AND double-blind (keywords)
OR double-blind (summary). March 2006 implementation of the search.
1.3 Analysis of the content and methods of
Select the number of cases the two groups, the mean age of male and female sex ratio, duration of disease duration and HAMD values (not to HAMA value) as the analysis of the baseline indicators.
Records of the test group is greater than all baseline indicators, equal to and less than the control group the number of trials, description, or the ratio of the mean difference between two sets of baseline distribution and to conduct visual analysis. On the visual impression of the
baseline indicators of suspicious, using sign test (sign test) analysis of the statistical significance of differences between the two groups. On the sign test showed a statistically significant indicators, differences in the distribution of Ridit analysis. The experimental group and control group the mean baseline HAMD, using quantitative data of the Cochran law Meta analysis (by the heterogeneity test homogeneous, and therefore the fixed effects model).
2.1 The search and collection of results
Retrieved 128 documents (6 unable to read the full text), in which compliance literature 82, but there are two base value does not describe the standard deviation of HAMD. Excluding sub-point test 5, four groups the mean one.
2.2 or the ratio of the mean difference between two sets of baseline analysis of the distribution of
2.2.1 visual impression of
In the 82 or the double-blind randomized controlled
clinical trials, each depicting an experimental group and
control group of age, male or female size, duration or stage of disease were 70,65 and 48 (Table 1). For the number of cases, age, sex ratio and the duration of the baseline stage of disease indicators, the test group is greater than the
number of research studies the control group (n) and the test group than the control group study of the number of (n-)
(differential distribution) a more balanced, However, the value of the difference between the two groups HAMD distribution seems uneven, in Table 1. Table 1 sets the value of group-based differences in the distribution of (slightly) Note: The sex ratio: the ratio of the number of cases of men and women; n: the test group the mean or the ratio of the number of trials is greater than the control group; n =: two
groups of equal number of trials; n-: control group,
experimental group is greater than the number of trials. An old soldier cancer it: the elderly, stroke, cancer and other diseases associated with depression.
Express the characteristics of the experimental
literature, multi-center, domestic drug (test group)
differences in the distribution of type of study does not seem balanced, but the elderly, stroke, cancer and other diseases associated with depression in the distribution of type of
study emerged reverse, but the data were less. In addition, the 16,30 and 108 experimental difference between the two data
sets too large, in Table 2. Table 2 double-blind study of drug
treatment of depression the number of cases of the two groups, and HAMD-based value (omitted)
2.2.2 Sign Test
Table 1 HAMD baseline values (n and n-) data on behalf of
the paired data into a simple sign test formula:
χ2 = [| n-n-| -1] 2 / (n n-)
= (52-28-1) 2 / (52 28) = 6.6125
Greater than χ2 (1,0.02) = 5.412 (P