Start stopping smartly
BMJ 2016; 353 doi: https://doi.org/10.1136/bmj.i3209 (Published 09 June 2016) Cite this as: BMJ 2016;353:i3209
All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
Even if you can understand the maths, it is challenging to adopt a 'neutral perspective' when you apply statistics to yourself.
Many bombing raids during the second world war, would suffer about 5% loss rates. All of the airmen knew this, but it seems that the majority countered it with a 'firm(ish) belief' that the one aircraft in twenty which would be shot down, 'wouldn't be us'. If bomber crews really accepted, mentally, the attrition figures, the chance of completing a tour of operations was so low, as to often lead to a sort of 'nervous collapse' in those individuals who didn't somehow 'blank out' the statistics.
Once loss rates routinely reached about 10%, the RAF would call a halt to that type of operation, because it became impossible for the crews to 'come to terms with' such a loss rate.
It isn't just a case of whether patients understand the maths - although, from what I can gather, the average person's grasp of maths is pretty poor: it is much more complex than that, because a lot of psychology is involved.
Competing interests: No competing interests
When presenting options, both words and numbers can certainly confuse, but so can the perceived basis of the choice. Some patients seem convinced that in reality they face a Hobson's choice; one influenced by the disease rather than the doctor. Others seem to regard the decision as similar to selecting the right option in the gameshow 'Who wants to be a millionaire?': the quizmaster can help but is not allowed to reveal the correct answer ahead of time. Perhaps we should collect annual statistics that would allow us, for common decisions at least, to tell patients how others recently in a similar predicament had chosen; and how they had felt afterwards when faced with the consequences of that decision. Collection of this 'phone a friend' data would be difficult, but would better inform patients, and may well rapidly reduce demand for interventions identified as by them as being either 'useless' or 'more trouble than they were worth'.
Competing interests: No competing interests
Thanks once again to Dr Godlee for raising this issue.
“the public often struggled to grasp numerical information, creating ample scope for miscommunication and bedevilling efforts to achieve truly shared and personalised decisions”.
The BMJ has gone to some trouble to involve patients in open discussion, even to the point of having patient reviewers examine papers before publication. As one who has been allowed to review reports pre-publication, one of my major points revolves around the use of numerical data so I fully support the above statement.
The problem stems from the statistical need to demonstrate a significant benefit to the “herd” (which no patient is) and the ability to relate this to the individual patient when consulting his/her doctor; two very different things with different requirements. The significant benefit to the “herd” is really the only way to demonstrate benefit of a drug/therapy etc, for which ODDS/Hazard ratios provide a very sensitive procedure. However the total number of patients involved is negatively associated with the number benefiting.
On the other hand the patient wants to know what the probability of his/her individual benefit or no benefit which ORs/HRs do not provide unless the actual numbers are provided so that individual probabilities can be calculated. An example is given below:
χ2 square test
Outcome 1 Outcome 2 Total
Group 1 20 36 56
Group 2 1980 1964 3944
Total 2000 2000 4000
Fisher's exact test
The two-tailed P value equals 0.0426
The association between rows (groups) and columns (outcomes) is considered to be statistically significant.
Odds Ratio Confidence Interval Calculation For 2x2 Contingency Table
Check odds ratio
observed NO 0.357
no TRT Any TRT Relative Rate 0.556
Observed any TrT 0.643
NO TREAT 20 36 56
any trt 1980 1964 3944
Total 2000 2000 4000 No Trt Any Trt
0.010 0.018
Result
True Difference = 0.008 0.80%
Odds ratio 0.55 relative rate= 0.556
95% confidence interval from 0.32 to 0.96 Significant
Real %age difference is 0.8% but the ODs ratio is 55% difference on this example- or nearly a 70 fold inflation. Yet in this paper the raw data is not available for examination.
In short an ODs ratio benefit (ie a ratio of two ratios) looks good but would benefit less than one in 100 patients; in my view pretty long odds for the individual patient
Competing interests: No competing interests
Less Medicine More Health
Dear Fiona,
Decades ago I started a new concept called “Step Down” treatment of hypertension. This occurred to me when I first started work in an Indian Government hospital in the 1960s after having been at the National Heart and The Middlesex Hospitals, London and the Peter Brent Brigham in Boston. I used to religiously record the data on the small slips that patients carry. I used to see lots of hypertensive patients in a large outpatient department. Almost all patients seemed to respond to the treatment, only beta-blockers and diuretics then, during the follow up visit which was a week later. I used to record that also. I used to encourage them as they had responded.
One day one patient had a sarcastic smile on his face when I told him “you have done well by taking medicines as advised correctly and your BP has come down.” When asked as to why he did that he told me the whole story. The government hospitals have a limited drug budget which gets exhausted by June and then till the next financial year drugs are not available regularly. This patient did not get the drugs but did follow my advice on life style changes correctly. I was nonplussed! Then more and more patients followed and I thought of doing a prospective observational study with controls. This confirmed my belief that probably mild-moderate hypertensives without target organ damage will do well on life style modifications alone. Then I started the step down treatment where I used to taper off anti-hypertensive drugs in those patients whose BPs were normal to see what happens to them. More than 50% of them the BP remained normal without drugs after that. Rest needed drugs in smaller doses to keep the BP under control. It was then that Late Professor Sir George Pickering’s wisdom dawned on me: “more people in this world make a living off hypertension than dying of it.”
In this beautiful book, Less Medicine More Health, Dr. Gilbert Welch very wisely depicts the pitfalls of too much medicines, tests and surgeries. I couldn’t agree more. I have described them more graphically in my own book What Doctors Don’t Get to Study in Medical School. I would like to add a word or two to your sentence “delegates concluded that equipping people to evaluate claims about treatments should start in schools; that numbers, words, and pictures were all necessary to cater for different situations and individuals; that doctors needed to communicate uncertainty where evidence was unclear or disputed; that patients differ in their attitudes to taking pills, whatever the evidence indicates; and that information should be used to inform rather than persuade.” In my opinion every doctor needs to give his/her patients just two bits of important information before they embark on any drug treatment. Any drug has two aspects which are important for the patient. The drug’s NNT and ADR risk. None of the drugs that we prescribe help every patient. It is a lottery. Number Needed to Treat (NNT) is a statistical term. To give a concrete example the NNT of drugs in mild-moderate hypertension in the MRC study published in 1985 in your journal puts the NNT at 850. To help one patient with an imaginary stroke in the next five years one has to unnecessarily treat 850 healthy individuals with drugs that will not benefit them but will carry the ADR risk of 5%. If you inform the patient these facts most of them will opt for trying life style modifications to drugs. His chance of getting benefit is one in eight hundred fifty but his chance of getting side effect is 45. Come to think of it our drugs help one and harm 45! The same for statins, NNT is 300 and side effects (ADR) 10%. Taking statins might help one in three hundred but will make 30 of them diabetics in one year and has many other dangerous side effects!
One sentence can sum up the mess that we have created for the elderly by poly-pharmacy. Less medicine for the old elderly gives them better health for them to enjoy life. It will add life to their years while it might also add years to their life!
Yours ever,
bmhegde
Competing interests: No competing interests