主题：贴，贴， -- 虽远必诛

5. Duration of fever

Five trials reported no difference in the duration of fever between

the intervention and control groups. Two of these trials reported

no difference in the number of participants whose temperatures

were normalised within 24 hours (Chang 2002), 48 hours and 72

hours (Yu 1997): Sanhan Jiere Koufuye versus (Fenghan) Biaoshi

Ganmao Chongji (RR 1.02 95% CI 0.85 to 1.22) (Chang 2002);

and Caichen Qinre Weixin Guanchangji versus virazole and acetaminophen

(RR 1.46, 95% CI 0.89 to 2.40, and RR 0.33, 95%

CI 0.11 to 0.99, respectively) (Yu 1997).

Likewise, another three trials (Chang 2002; Chang 2005; Wang

2008a) showed no difference in duration of fever between the two

groups (WMD -0.29 hours, 95% CI -2.26 to 1.68, WMD -3.91

hours, 95%CI -8.24 to 0.42,WMD 4.20 hours, 95%CI -1.41 to

9.81 and WMD 3.62 hours, 95% CI -2.74 to 9.98, respectively).

Three trials (Chang 2005; Li 1998; Wang 1998) reported a statistically

significantly shorter duration of fever in the treatment

group compared to the control group 2.02, 6.90 and 5.70 hours,

respectively (WMD -2.02, 95% CI -3.86 to -0.18; WMD -6.90,

95% CI -9.94 to -3.86, and WMD -5.70, 95% CI -10.09 to -

1.31, respectively). One study (Li 1998) showed that in the treatment

group the average duration of fever was longer than control

group by 11.60 hours (WMD -11.60, 95% CI -15.65 to -7.55).

Another trial (Wang 2008a) showed no statistically significant differences

in the time to resolution of fever in the treatment group

A and B as compared to the control group (WMD -2.06 hours,

95% CI -5.02 to 0.90, and WMD -1.44 hours, 95% CI -4.70 to

1.82).

6. Adverse events

None of the trials reported any adverse events, according to our

definitions.

7. Additional outcomes

Seven trials (Chang 2002; Chang 2005; Chang 2007; Chen 2004;

Wang 2004; Wang 2008a; Yu 2005) reported effects on TCM

signs. We only analysed data in two studies (Chang 2005; Chen

2004). There were no statistically significant differences between

the two groups in these trials.

D I S C U S S I O N

Summary of main results

Studies of Chinese herbal medicines for the common cold lack

sufficient power to provide reliable estimates of their efficacy, due

to poor study design and methodological quality.

Although Chinese herbal medicines as a treatment for the common

cold and the methods of manufacturing these medicines are

widely accepted in China, most of the constituents of the pharmacologically

prepared drugs used in trials cannot be specified.

This is in marked contrast to the pharmacological agents used

in Western medicine, for which the chemical constituents, their

quantities and the percentage of any impurities or contaminants

are precisely known. In addition, the variation between different

production batches of Western medicines is kept within specified

limits. In contrast, variation between formulations and batches

of pharmacological agents are inevitable in traditional Chinese

medicine (TCM), although the Chinese Government specifies the

acceptable limits of variation. This variation is a factor that may

contribute to any heterogeneity between different study results.

A large number of the trials initially identified claimed to be randomised

controlled trials (RCTs). However, after contacting the

trial authors to request information regarding the method of randomisation

used, we found that more than 95% of the authors

misunderstood the concept of randomisation. In addition to this,

some of the studies were conducted several years ago, and the trial

authors may have forgotten the details of the methodology they

employed. This could lead to a memory bias and undermine the

veracity of information.

Some studies (Chang 2002; Li 1998; Li 1999a; Ma 2000; Wang

2004; Yang 2000; Yu 1997) used unequal arms in their design.

Of these studies, Yu 1997 used a proportion of 3:1, and only 15

patients were included in the control group (43:15). However,

consideration of the sample sizes was not reported in these trials,

which makes it difficult to detect the differences between the two

groups accurately. This may lead to a low test power.

Finally, we considered it pertinent to attempt to compare selected

TCMsigns in different groups of participants. TCMsigns are difficult

to quantify because they use subjective outcomes. For example,

’wu han’ means that the participant feels cold whilst also

having a fever, and this can be divided into low, medium and high

categories, depending on the participants’ subjective view. ’Mai xiang’measures

the pulse; in TCMit is a complex and difficult technique

to judge this accurately and often dependent upon on the

physician’s experience.TCMresearchers and physicians should decide

upon an accurate and consistent method of measuring TCM

signs.

Quality of the evidence

None of the trials included in this review used a placebo as a control:

instead “positive effect drugs” were selected. Two studies (

Chang 2005; Chang 2007) mentioned that the rule of selecting

the control drug was that “the effect was generally acknowledged”.

This may result in false positive findings, as a number of interventions

are considered effective for the common cold, particularly if

trial conductors know that a “positive” drug was used in the trial

and the purpose of the study was to demonstrate the same effect

as the control (so called “equal effect test”). For example, Chang

2005 compared two different forms of the same ingredient. In this

Chinese medicinal herbs for the common cold (Review) 13

Copyright 2009 The Cochrane Collaboration. Published by JohnWiley & Sons, Ltd.

case, “double blinding” would not be a valid term. This may result

in false positive findings: if the trialist knows that a “positive” drug

was used and the study was an “equal effect test” study, there is

a potential risk that the outcome detectors will give same results

for the two groups. For example, the purpose of Chang 2005 was

to compare two different forms of the same drug; one was in a

granule form and the other was in a capsule form and the trialists

were aware of this. Therefore, there was a possibility that the

results were not scrutinised carefully - particularly the subjective

outcomes. In this case “double blinding” does not have any value.

We found that three trials claimed to have used ’double blinding’ (

Chang 2005; Chang 2007; Yang 2000) and one used ’single blinding’

(Chen 2004). These studies found a similar recovery rate as

the “positive” control. Two of the ’double blinding’ studies (Wang

2004; Wang 2008a) found the recovery rate in the intervention

group to be higher than the control group by almost 50% (Wang

2004 RR 1.43; Wang 2008a RR 1.59 and RR 1.67). However,

Wang 2004 lacked a description of how to verify the success of the

blinding. Another trial (Pan 2000) found the marked improvement

rate in the intervention group to be higher than the control

group (RR 1.69). For a disease like the common cold, which can

go into spontaneous remission within one week, it is very difficult

to find that the effect of a new drug is better than the “positive

effect” of the control drugs.We have to point out that the possible

reason that the relative risk is so high is that a high risk of selection

bias, detection bias or both may have existed in the studies where

the effects were similar or higher than “positive effect” controls.

Many trials incorrectly used drugs whose effects have not been

demonstrated as a control. Thus it is difficult to determine the

efficacy of the intervention by comparison.Where the intervention

is equal in efficacy to the control, no conclusion can be made.

For example, the average rate of recovery from the common cold

was the same for those participants taking Jinlian Qinre capsules

(intervention) and JinlianQinre granules (control) (Chang 2005).

Since there is no evidence assessing the efficacy of Jinlian Qinre

granules, we therefore cannot determine the efficacy of Jinlian

Qinre capsules.

Potential biases in the review process

We found the formulations of interventions in five studies (Ma

2000; Pan 2000; Yang 2000; Yu 1997; Yu 2005) were prepared

by either the trial authors or colleagues in their ho