October 2002 Volume 83 Number 10 "serving the protectors"

Search for the best medical evidence

How do doctors know whether the treatment they provide is any good? Sometimes, they don’t know, but the progress of modern medicine has come from the ability to tell the difference between a treatment that works and one that does not.

The usual plan is to test a treatment and compare it with doing nothing or another treatment.

The outcome is then compared to what we think would happen by chance alone. If there is a difference, then the treatment is said to be significantly better – or worse.

All of these ideas are based on being able to tell what would have happened normally or by chance alone. Because humans are so complex, it can be hard to tell what is normal and, sometimes, cancer can be cured in unexpected circumstances without treatment.

Doctors try to begin with a definition of what would be normal and make choices from there. As an example, what is a normal height? Is someone who is five feet tall that height because they are sick or just a normal height?

To start with, we use a definition of normal, rather than knowing what is normal. Then we say that someone who is not in the normal range is more likely to be sick, but it could be a healthy “normal” thing to be that short.

We define normal as being whatever 97 per cent of all people are. That means that, automatically, 3 per cent of normal people become abnormal by this medical definition. The way doctors use this information is to start looking for things to be wrong in those who are not “normal”.

Research scientists recently released results about hormone replacement therapy (HRT) that spoke of an increased risk – 26 per cent – of breast cancer.

It is immediately important to know what the normal risk is and the normal risk for that group of women.

The number of women with breast cancer increased from 20 per 10,000 to 28 per 10,000, after five years of HRT using combined oestrogen and progesterone.

The normal group must be of the same age otherwise the comparison is not valid. The next thing is to see that the increase is not for 26 per cent of all women but is from 20 to 28. This is an increase, but is it significant? It is for the women who might get the cancer.

Other statistics showed that the risk of fractured hips and bowel cancer decreased, with the overall results showing there was not any increase in the death rate in women with HRT and those without.

The implication of this is the need to put any statistics in context. Don’t jump to conclusions. Get all the details.

Treatments that we test can be proven in a number of ways and the medical profession has what it calls “levels of evidence” to help convince doctors that a treatment is proven.

The description of the best proof is called “level-one evidence”, based on double-blind crossover trials and more than just one person’s participation. To do this, a treatment is given to a person who does not know what he or she is reviewing, so the “placebo” effect is eliminated.

The double-blind comes from good trials when the person giving the treatment does not know whether the drug he or she is giving is real or sugar. In this way, the person’s bias is eliminated.

The crossover refers to the sugar treatment crossing over to the drug treatment during the trial, and no one knowing when that switch occurred. This is the best proof: to show, using such trials, that there is an improvement in the health of the patient.

This level of evidence is expensive and difficult to get sometimes, but provides the most confidence in a treatment or drug.

Other levels of evidence are useful. Level-four evidence is when experts agree but effectiveness has not been proven by the strict trials.

If you want to know about treatment or dangers of medication, it is important to find information for yourself. Know the numbers involved and the types of people on whom trials were conducted so you can see if the comparison with yourself is real. Then, you need to know if it is real evidence or just the idea of someone seeking money or headlines.

If the numbers stack up, and there is good evidence, you will make good choices about your lifestyle and treatments. Alternatively, ask someone you can trust.

Your questions answered

Dr Pearce will answer questions on any health issue important to you. For his response, write to or fax the Police Journal with your question. Those who write need not identify themselves.

• Police Journal, PO Box 6128, Halifax St, Adelaide, SA, 5000
• Internal dispatch, post code 168
• Fax: 8231 0855

If you prefer to correspond by e-mail, send messages to the associate editor
(brettwilliams@policejournalsa.org.au).

PASAweb   Index & Search   Top of Page   Comments   Email to Editor

The Police Journal Online is an official publication of the Police Association of South Australia and is published monthly. Editors of kindred publications can seek permission from the Editor to re-publish any Police Journal Online article. Copyright 2001  The Police Association of South Australia sustance