“Shared understanding is the central challenge of medicine in this century.” Read this week’s research reviews by Richard Lehman.
NEJM 6 Apr 2017 Vol 376
Body weight fluctuations in coronary disease
“In this post hoc analysis involving patients with established coronary artery disease who participated in the TNT trial, fluctuations in body weight were strongly associated with the risk of cardiovascular events and even death.” The TNT trial was a randomised controlled trial of atorvastatin involving 9509 participants with CAD. The fluctuations in weight referred to were those recorded every three months in the first year and then every six months over a total median period of 4.9 years. In other words, this was not a study of day to day weight variation but of medium term fluctuation. The relative risk figures given are hard to follow, but the graphics show a steady increase in CV events with each quintile of weight fluctuation, with the risk of stroke or death doubled between the lowest and the highest quintiles. These were mostly white male people aged a little over 60. Should our advice to all of them be “keep your weight steady or you may die sooner?” Like so many observational studies, this one raises a whole range of questions, and it isn’t clear which of them it would be possible or worthwhile to investigate further. I’m inclined to declare that more research is not needed.
Endovascular treatment for stroke at two years
It took a long time for thrombolysis to become an established treatment for stroke, but I suspect that endovascular clot retrieval may be adopted more quickly if the resources can be made available, because the benefits are very clear. Mind you, this does not sound like easy work: a stroke centre would need a 24 hour team capable of accessing the anterior brain circulation as quickly as possible, identifying the thrombus, putting in a platinum mesh stent to encase it, and then removing it. But a follow-up study of the MR CLEAN randomised trial shows that the beneficial effect of endovascular treatment in patients with acute ischemic stroke caused by a proximal intracranial occlusion of the anterior circulation was sustained during the course of at least two years. This benefit includes major functional differences as well as reduced mortality.
“Subclinical hypothyroidism” hits the bin
The thyroid gland could be described as the playground of overdiagnosis. The hormones it produces are the simplest in the body: little molecules with three or four iodine atoms attached. If you measure these, and the amount of thyroid stimulating hormone (TSH or thyrotropin), you have a nice little trio of biochemical parameters to play with and for people to attach symptoms to. Here is a trial of levothyroxine supplementation in people of mean age 74 with an elevated serum thyrotropin level and a serum free thyroxine level within the reference range. This has come to be called “subclinical hypothyroidism.” The treated group received levothyroxine at a dose sufficient to bring their thyrotropin levels into the “normal” range, whereas the others were cunningly subjected to sham blood testing and dose adjustments to placebo tablets. When assessed by a Hypothyroid Symptoms score and a Tiredness score after a year, the Scottish, Irish, Dutch, and Swiss participants all felt the same, regardless of treatment. Let’s hear no more of subclinical hypothyroidism.
JAMA 4 Apr 2017 Vol 317
Prime Time for Shared Decision Making
Do you feel weary at the mention of shared decision making? I often do. Now read this wonderfully focused piece, which sets out the processes that are needed to make it a reality in clinical practice. It is by Erica Spatz, a young cardiologist at Yale who writes with Harlan Krumholz (our shared mentor), and Ben Moulton, who works on shared decision tools. There is nothing fuzzy or rhetorical here: “Shared decision making has arrived, too early and too late—too late for the need, and too early for the level of preparation among clinicians and their clinical practices.” For clinicians to achieve a shared understanding of medicine with patients, we need a radically better structure of skills and knowledge tools. To me, shared understanding is the central challenge of medicine in this century.
Thyroid cancer in the US: a true increase?
I’ve already labelled the thyroid gland as the playground of overdiagnosis, but does this apply to thyroid cancer as well as to subclinical hypothyroidism? If you detect thyroid nodules by ultrasound, and then biopsy them, you will find malignant looking cells in many of them, resulting in huge increases in the diagnosis of “thyroid cancer.” South Korea is the world’s leading example at present. The same appeared to be true of the US, according to a much cited graph comparing the increasing rate of diagnosis with the flat line of mortality from thyroid cancer between 1975 and 2005. However, a new study based on a cancer registry of 77 276 patients in the US diagnosed between 1974 and 2013 does not show a flat line in mortality, and while the diagnosis of papillary thyroid cancer is increasing by 3.6% per year, death from thyroid cancer is increasing in relation to incidence by 1.1% per year. Unfortunately, the data in this paper are not very easy to follow, either in the text or in the graphics. But the message is clear: not all the increase in thyroid cancer in the US is an artefact of overdiagnosis.
Ann Intern Med 4 Apr 2017 Vol
Weight history counts
Back to weight variation. It’s not your current body mass index that relates most closely to mortality, but your highest ever BMI. This applies to you directly if you are a male doctor or a female nurse in the US, as these figures come from the Nurses’ Health studies and the Health Professionals’ Follow-Up Study. It applies even more if you are under 70 and a never smoker. But don’t worry: the association is not particularly strong and you are going to die some time anyway.
O MAN, BMI is a lousy measure
Those who like to dabble in cardiovascular risk factors have long been sceptical about the usefulness of BMI, a measurement which persists because it has a long history and is easy to calculate from widely held data. Here is yet another study showing that it is a bad correlate for other measures of cardiovascular risk, especially in non-white populations. In two multi-ethnic databases, there was a high prevalence of MAN, which stands for metabolic abnormality but normal weight. The metabolic abnormalities were at least two out of a list of high fasting glucose, high LDL-C, high triglycerides, and/or high blood pressure. Compared with white people, all racial/ethnic minority groups had a statistically significantly higher prevalence of MAN, which was not explained by demographic, behavioural, or ectopic fat measures. Using a BMI criterion for overweight to screen for cardiometabolic risk may result in a large proportion of racial/ethnic minority groups being overlooked.
The Lancet 7 Apr 2017 Vol 389
Flushed with hype
It’s a salesman’s dream: a drug for every woman who gets flushes at or after the menopause. “Treatment with a neurokinin 3 receptor antagonist (MLE4901) could be practice changing as it safely and effectively relieves hot flush symptoms without the need for oestrogen exposure.” But curiously this assertion does not come from Astra Zeneca, which supplied the drug, but from independent academic authors funded by two British public bodies, the Medical Research Council and the National Institute for Health Research. So how safe and how effective is this substance? Well, it was moderately effective over four weeks in a tiny crossover trial with complete data from 28 participants, reducing the number of flushes by 45%. It caused a major rise in liver transaminases in three participants, which later settled. Dosing millions of women for menopausal symptom relief was standard practice until we became aware of the true risks of oestrogen replacement therapy in massive, long term randomised trials. Hold the advertising copy for neurokinin 3 receptor antagonists till then.
“In our TRINITY study, treatment with extrafine fixed triple therapy had clinical benefits compared with tiotropium in patients with symptomatic COPD, FEV1 of less than 50%, and a history of exacerbations.” So who does the “our” refer to? Step up Chiesi Pharmaceuticals, who have produced a new triple inhaler containing old ingredients. “The funder was responsible for the design and analysis of the study, oversaw its conduct, and was responsible for preparing the study report.” And here it is, in one of the world’s top two medical journals. How does it advance medical science? Well, it tells us that the argument for fixed dose combination inhalers is just plain wrong: they do not improve adherence in a way that improves outcomes. When compared with open triple therapy using different inhalers, the Chiesi product did somewhat worse. When compared with tiotropium alone, it did a little better, as you might expect, since it also contains beclometasone and formoterol. But then again, what do any of these inhalers actually do for individual patients in the long term? If you tried to convert this trial into a decision aid, it would show that you would have to use the Chiesi inhaler instead of tiotropium for 10 years in order to have one less exacerbation in your COPD. Of course, you would never know if that applied to you. Some other chap might have two fewer, and you might be dead by then.
Be careful of what BP you wish for
Two trials in the early 2000s hoped to show that telmisartan would improve outcomes in 31 546 patients aged 55 years or older with a history of coronary artery disease, peripheral artery disease, transient ischaemic attack, stroke, or diabetes mellitus complicated by organ damage. The trials were called ONTARGET and TRANSCEND and failed to show any difference. Here is a new analysis of the data from these trials to see what happened to the patients whose BP dropped the most. Mean achieved systolic BP less than 120 mm Hg during treatment was associated with increased risk of cardiovascular outcomes except for myocardial infarction and stroke. Similar patterns were observed for diastolic BP less than 70 mm Hg, plus increased risk for myocardial infarction and hospital admission for heart failure. Treat your highly responsive, high risk patients with care. Stopping drugs may do them more good than starting new ones.
The BMJ 7 Apr 2017 Vol 357
Prognosis of undiagnosed chest pain
If you follow Bayesian logic in a diagnostic pathway, you usually feel happy when you have reached a rule-out level above 95%. British general practice works on a similar basis of “satisficing”: reaching the point where the doctor will carry the risk while the patient can go away reassured, albeit with some kind of safety net in place. We GPs are risk sumps. We do Bayesian thinking all day long, for better or for worse. Here is a study which suggests that we should do it better. It’s an analysis of 172 180 adults aged ≥18 from 223 general practices presenting with a first episode of recorded chest pain, using the CALIBER database. Most go away from their first consultation without a clear diagnosis in the notes, and of these, 4.7% go on to have a cardiovascular diagnosis over the next 5.5 years. Maybe we are too easily satisfied.
Finding individual participant data from trials
When advocating for open data over the past six years, I’ve often paused to admire the hard work of those who actually work with it rather than follow my easy path of just writing about it. One such is Catrin Tudur Smith, who is last author on this paper about the difficulties of getting full individual participant data (IPD) for meta-analysis. This sums up the current state of the art, on which we depend for accurate information to guide the treatment of patients across the world: “IPD meta-analyses are considered to be the ‘gold standard’ for the synthesis of data from clinical research studies; however, only 25% of published IPD meta-analyses have had access to all IPD.” The article was based on a survey of 760 such meta-analyses. Medicine, the most humanly important of the sciences, lags behind most of the rest in its sharing of data.
Plant of the Week: Vinca minor
Through primrose tufts, in that green bower,
The periwinkle trailed its wreaths;
And ’tis my faith that every flower
Enjoys the air it breathes.
So wrote the young Wordsworth in the Lake District 220 years ago: it is good to know that he got past the daffodils and on to the primroses and periwinkles.
Primroses are best left in the state of nature: in our garden, I get rid of any that show signs of hybridisation. But periwinkles large and small (Vinca major and V minor) have been improved to varying degrees by clonal selection. Our particular favourite is a double blue form of the lesser periwinkle called “Double Bowles.”
There is virtually no spot in the garden where you could not use this as ground cover, and it will flower through most of the season and sometimes in winter. However, it does look at its best when well hydrated, so for dry spots it’s probably better to go for a single white flowered variety, or a dark maroon-purple.
Every green bower should be full of little periwinkles trailing their wreaths. Among primroses, preferably.