In the present times, the online Medline databases like PubMed have made it very easy for anyone to search all that has been published about a disease or topic of interests over the past 30 years! The PubMed search itself is very simple - much like 'Yahoo' or 'Google' search. You can type in a word or phrase, for example - Macular degeneration. You can use qualifiers to narrow your search, for example - Macular degeneration AND Smoking will get you all articles on macular degeneration that specifically deal with smoking. Using the 'AND' or 'OR' function will get you more targeted results.
Many of the abstracts that come up have links to 'full text' of the published article. You may have to pay to get access to full text articles. You can email or even send old fashioned postal mail to the corresponding author and request a reprint of the article. Even if you are able to access only the abstracts, you will see almost all studies justify their conclusions by providing some or the other statistic and talk about a 'significant' affect. On this page we will attempt to show you how to make sense of the reported statistics. Our hope is that, at the very least, you would view the results of medical studies from an additional perspective i.e. you will question whether the results are likely to make an easily recognizable difference to the course of your disease such that your health or lifestyle are positively impacted.

If you read that a particular drug or treatment improved vision significantly, this does not necessarily mean that in practical everyday real-life sense vision has improved enough to be of considerable value to you. Statistical significance merely means that in the reported study data, it is very unlikely that the difference in the two study groups are not real.
Statistical significance assures you that there indeed was a difference between the two groups, at least in a mathematical sense. We will dwell, at considerable length, upon the meaning of p value and statistical significance as well as provide you with the meaning of other criteria commonly used to report data such as odds ratio and relative risk.
We think it is very important to have at least a basic understanding of how to interpret study data because, given the pressures of 24 hours news reporting; there may be a rush to print and report new data. The recent recommendations on hormone replacement therapy and mammography have come to many as a surprise but provide an important lesson.

The most commonly used level of statistical significance is a p-value less than 0.05. It means that the probability of the reported difference having occurred just by chance or fluke is less than 1 in 20. You may wonder, what's so special about a p value of 0.05? Nothing really. This value is a hangover from the days before computers, when it was difficult to calculate exact p values and people used tables of values for the test statistic corresponding to a few arbitrarily chosen p values, namely 0.05. This value has now become enshrined as the threshold value for declaring statistical significance. A researcher could very easily define a different p value to declare significance, but the journal reviewers would probably insist on adhering to conventional data reporting practices.
If you have 20/200 vision (therefore legally blind) and a study reports vision improvement to 20/100 in a large enough number of patients with some treatment, these results may be statistically significant. However, you may wonder whether this improvement in vision, though welcome, is of any practical value to you if the side effects or cost of the treatment are factors to be considered too. Therefore it is possible that statistically significant results do not impact on your lifestyle or disease in a considerable way. Herein lays another point of interest. What if the data is NOT statistically significant - Is the data worthless then? In a letter to the editor about a study published in the journal Ophthalmology, Romano tackled this very point (Ophthalmology 2002:109;1949-50).

In the study, two different eye muscle surgery techniques are compared. The two surgical techniques are essentially equal in risk, so the risk from making a choice of one or the other is virtually nil. The benefit to be gained, however, is a better result from a treatment, and this is a benefit gained at no cost to the patient or to society, because the risks of the two compared treatments are equal.
In this kind of a situation, one hardly needs a P value less than 0.05 i.e. a probability of 1 in 20 that the difference in results is because of chance, instead of being a real difference (the meaning of p less than 0.05). Literally, any chance at all that one procedure is better than another may, as is the case here, warrant shifting to any demonstrably superior procedure. What sort of a probability does one need then for this situation? Even odds (50/50) would be fine, which is a P value of .50, not .05! And a probability of even less than 0.50 would be acceptable, too.
To translate the statistical meaning of P = 0.178 (reported in the study as not significant because this p value is not less than 0.05), there is a 1 in 5.6 probability (or chance) of this difference in results being the result of happenstance and not a real difference in results, and a 4.6 of 5.6 probability (or chance) that the difference in results is real and the result of the different surgical techniques used. In other words, it is an 82%, or 82 in 100, likelihood that the new surgical technique does give better results than older surgical technique. That is an impressive opportunity at no additional risk and no cost whatsoever to us, the patient, or society. Therefore, P = 0.178 is more than adequate in terms of medical or clinical significance to adopt the new surgical technique in preference to the old one.

For learning more about statistical methods you can visit Hyperstat Online.

Understanding Clinical Trials
There are four phases of clinical research that occur in sequential order: Phase I, Phase II, Phase III, and Phase IV. Information from the earlier phases is used to advance the study of the investigational drug through these phases. At the end of Phase III, there generally is enough accumulated information to prepare and submit an application to market the investigational drug.
Phase I studies are the first studies that evaluate the use of the investigational compound in humans. These studies are generally done across a wide range of potential doses to determine the safety and tolerability of the investigational drug. These studies are not designed to determine the efficacy of the treatment. Generally, the first Phase I studies are conducted in about 80 healthy subjects who do not have the disease target; however, this is not necessarily the case.
Phase II studies (sometimes referred to as proof of concept studies) are the first studies in which patients with the disease of interest are studied. The objective of these studies is to evaluate the safety and tolerability of the investigational drug and to assess whether the drug has the desired clinical effect (benefit) or efficacy. Usually, these studies evaluate a range of doses in a limited number of patients (about 300) who take the drug for a limited period of time.
Phase III studies are the definitive studies of the safety and efficacy of the investigational drug that primarily support the applications to market the new drug. Phase III studies are larger and more expanded versions of Phase II studies. By design, these studies are less restrictive than Phase II studies and include a wider variety of patients taking a wide variety of other medications. The duration of treatment and the size of the studies is highly variable, depending on the expected use of the medication and the endpoints that are being evaluated.
Phase IV studies are generally conducted after the drug has been approved for marketing. Phase IV studies are designed to investigate other potential uses of the drug or to compare the drug to other available treatments. Additionally, Phase IV studies are also conducted to obtain specific information from selected subgroups of patients or to gain additional safety experience in a huge number of patients.