As the saying goes, the answer to 'bad science' is not 'no science' but 'better science'. Fortunately there are people who are expert at interpreting published studies and can see where some of the biases lie. Unfortunately, it can be very difficult to do this. One problem in medicine is that pharmaceutical companies may choose not to publish 'negative' results - this gives the impression that a drug is more effective than it really is. Another is to measure large numbers of parameters during the study so that, by chance, one of the parameters seems to improve with the drug - by omitting the list of all the parameters measured it makes it difficult to see that this was likely to be just due to chance.
One way round these problems is to insist that only the results from registered trials can be published and that all the parameters being measured also have to be registered in advance. In this way, results of all trials are seen, not just the ones that show the drug in a positive light.
If you're interested in improving the quality of the information available from clinical trials, you might like to sign this international AllTrials petition to get all clinical trials published
http://www.alltrials.net/petition/