Intended for healthcare professionals

Endgames Case Review

A patient request for some “deprescribing”

BMJ 2015; 351 doi: (Published 03 August 2015) Cite this as: BMJ 2015;351:h4023
  1. David Unwin, general practitioner1,
  2. Simon Tobin, general practitioner1
  1. 1Norwood Surgery, Southport PR9 7EG, UK
  1. Correspondence to: D Unwin unwin5{at}

A 52 year old man with a history of type 2 diabetes for 14 years and hypertension for nine years presented to his general practitioner. He was a non-smoker with an alcohol intake of eight units a week. He had been experiencing bloating, abdominal pains, and erratic motions for more than a year. Because he drove about 12 000 miles a year for his job he found the loose motions “a real worry.” He wondered whether any of his problems might be caused by his drugs and asked if he could cut down on any if they weren’t all needed. He admitted to being afraid that his diabetic control might deteriorate and that he might need insulin, like some of his relatives who also had diabetes.

He was taking aspirin 75 mg once daily, metformin 500 mg three times daily, perindopril 4 mg daily, and simvastatin 40 mg at night.

On examination his weight was 108.8 kg (steady at this for 10 years), body mass index was 34.4, waist circumference was 113 cm, and his blood pressure was 130/80 mm Hg (steady at this level for some years). His abdominal examination was normal, except that he had central obesity.

Glycated haemoglobin (HbA1c) was 52 mmol/mol (reference range 0-41), bilirubin was 7 µmol/L (0-20), alanine aminotransferase was 53 U/L (5-37), and γ-glutamyl transferase (GGT) was 59 U/L (0-50). In addition, his estimated glomerular filtration rate was 100 mL/min/1.73m2 (90-120), total cholesterol was 3.7 mmol/L (desirable ≤4.0), high density lipoprotein-cholesterol was 1.3 mmol/L (>1.0), and triglycerides were 1.3 mmol/L (<1.7).


  • 1. What syndrome does this patient have?

  • 2. Which of the drugs he is taking would be the most likely to be causing his abdominal symptoms?

  • 3. What are the possible causes of his raised GGT?

  • 4. How could his request to cut down on drugs be handled?


1. What syndrome does this patient have?

Short answer

The metabolic syndrome. Definitions vary but are based on insulin resistance, hypertension, obesity, and dyslipidaemia.


The metabolic syndrome has several slightly different definitions but they all feature insulin resistance, hypertension, obesity, and dyslipidaemia.1 The syndrome was first properly described in 1988.2 In the presence of obesity, the liver, fatty tissues, and voluntary muscle become resistant to insulin, resulting in a hyperinsulinaemic state with worsening obesity, hypertension, and liver induced dyslipidaemia.

Hypertension, type 2 diabetes, central obesity, abnormal liver function tests, and dyslipidaemia are linked by the metabolic syndrome and the common causative factor is obesity. The best remedies are therefore weight loss and exercise—highly effective interventions that are not always given the prominence they merit.

When thinking about guiding patients with the metabolic syndrome it is perhaps best to concentrate on lifestyle changes before considering the individual, pharmacologically based protocols associated with hypertension and diabetes.

Evidence suggests that the features of the metabolic syndrome can be improved by the low carbohydrate diet that this patient went on to choose.3 4 Patients with obesity and type 2 diabetes should avoid sugar, as well as bread, pasta, cereals, and rice—the starch in which is composed of conjoined glucose molecules.

2. Which of the drugs he is taking would be the most likely to cause his abdominal symptoms?

Short answer



In our clinical practice the most common side effect of metformin is altered bowel habit. Symptoms such as flatulence, nausea, vomiting, and abdominal discomfort are common. We know of many patients who have undergone gastrointestinal investigations only to find that their symptoms settled when metformin was withdrawn. One study found that 26% of patients taking metformin had adverse gastrointestinal adverse effects and 18% experienced diarrhoea.5 The extended release formulation of metformin was found to be associated with fewer side effects than immediate release metformin.5

Metformin has been the mainstay of treatment for type 2 diabetes since 1998 when the UK Prospective Diabetes Study (number 34) showed reduced mortality with metformin use compared with diet alone.6 Recently a French meta-analysis of 13 random controlled trials questioned the central role of metformin in the care of patients with diabetes.7 In this meta-analysis, in which 9560 patients were given metformin and 3550 were given conventional treatment or placebo, metformin did not significantly affect the primary outcomes of all cause mortality or cardiovascular mortality. The secondary outcomes—myocardial infarction, stroke, heart failure, peripheral vascular disease, leg amputation, and microvascular complications—were also unaffected by treatment with metformin.

3. What are the possible causes of his raised GGT?

Short answer

The most likely cause is non-alcoholic fatty liver disease, thought to be currently affecting 20% of the developed world. Many doctors assume alcohol to be the cause of raised serum GGT, but other causes include drugs such as statins, methotrexate, steroids, and amiodarone.


Alcohol (whether disclosed or not) is a common cause of raised GGT, but it is important to remember that there are many others. Patients with raised GGT who don’t drink often become frustrated by repeated questioning on the subject.

Other causes include viral hepatitis; Wilson’s disease; lipodystrophy; starvation; and drugs such as statins, methotrexate, steroids, and amiodarone.8 A particularly common cause that is relevant to this case, mainly because alanine aminotransferase was also raised, is non-alcoholic fatty liver disease, which affects 20% of the developed world and most patients with type 2 diabetes.9

Non-alcoholic fatty liver disease is a continuum of liver disease that includes hepatic steatosis, steatohepatitis, fibrosis, and ultimately cirrhosis. It is unclear what tests are appropriate for this condition. Experts differ about when an ultrasound scan or a liver biopsy is indicated. Given that the best treatment consists of weight loss and exercise,8 10 it could be argued that a sensible response to a raised GGT is a change in lifestyle and ongoing monitoring of liver function tests. What might the link be between obesity and non-alcoholic fatty liver disease given that the best remedy is weight loss? Once the liver’s limited capacity to store glucose as glycogen has been exceeded excess glucose is stored in the liver cells as triglyceride.11 This helps explain how diabetes could lead to a fatty liver and why a low carbohydrate diet can be so effective.

The Framingham Heart Study provided evidence of the importance of a raised GGT concentration—after 19 years of follow-up in 3451 people, those GGT value was in the highest quarter had a 67% increase in the incidence of cardiovascular disease.12 This suggests that the common clinical finding of raised GGT in the absence of excessive alcohol consumption could be seen as a chance to investigate further for cardiovascular risk factors (HbA1c, weight, blood pressure, and waist circumference). A raised GGT result can also be a useful stimulus for both the doctor and patient to think about serious lifestyle changes.

4. How could his request to cut down on drugs be handled?

Short answer

Arguably none of his drugs is essential—they have all been prescribed to reduce his risk of cardiovascular events and the complications of diabetes, not to treat an actual disease. These risks also depend on lifestyle choices such as diet, smoking, and exercise, which could be explored as possible alternatives to medication.


We suggest that, rather than prescribing drugs to patients, healthcare providers should try to explain the benefits and harms so that patients and clinicians can come to a shared decision. A recent review discusses possible processes for desprescribing and argues that this should be patient centred and evidence based.13 In this case, several lifestyle and pharmacological interventions are available that could reduce cardiovascular risk. Patients should be involved in making informed decisions.

Several studies suggest that shared decision making improves patient satisfaction with health services,14 15 although evidence that it has an impact on clinical outcomes is less clear. However, one study into shared decision making in patients with poorly controlled asthma showed significant improvements to asthma pharmacotherapy and clinical outcomes.16

Until recently the various risk calculators that could be shared with patients did not properly factor in diet or exercise, and they did not mention side effects or numbers needed to harm (NNH). However, James McCormack, a professor in the faculty of pharmaceutical sciences at the University of British Columbia in Canada, has developed a risk calculator that facilitates discussion of risks and harms with patients (; the studies that he used as a basis for this can be found under FAQ). The calculator can demonstrate to patients the impact of various lifestyle and drug interventions in a similar case to theirs, but before drugs are started.

Using this adapted QRISK2-2014 calculator, a person like our patient with type 2 diabetes of age 52 years, weight 108.8 kg, height 178 cm, systolic blood pressure 155 mm Hg, total cholesterol 5.0 mmol/L, high density lipoprotein 1.0 mmol/L, no family history of note or history of smoking, chronic kidney disease, atrial fibrillation, or rheumatoid arthritis would have an untreated 10 year risk of heart attacks or strokes of about 15% at baseline. Patients can then see how this can be reduced by the various interventions on offer (table).

Ballpark benefits to risk of heart attacks or strokes and harms of specific interventions for our patient*†

View this table:

As shown in the table, physical activity is as effective as low to moderate intensity statins at reducing our patient’s risk of cardiovascular disease and outperforms aspirin. The Mediterranean diet is nearly as effective as metformin. In addition, the major lifestyle interventions mentioned (physical activity and Mediterranean diet) have a low risk of harm compared with the drugs listed and reduce the risk of comorbidities such as osteoarthritis, some cancers, and gallstones. Because our patient had already started treatment these figures are less accurate, but even so they still inform the debate.

An important motivating factor in changing patients’ behaviour is fear of negative consequences (“shroud waving”); another is hope for a better future. It is interesting that much of the placebo effect is explained by patients’ hopes of an improved future.17 When patients can imagine a better future if—for example, they lose weight—they begin to buy into and invest in this preferred future. So when our patient understood that weight loss could help him feel better and give up his drugs he began to investigate which diet would suit him best—in this case the low carbohydrate diet. He also reported that monthly follow-up, with the continuity of seeing the same clinician who believed he could achieve his goal, was very important.

Patient outcome

The patient steadily lost a total of 16 kg over seven months and successfully stopped all four prescribed drugs, thereby achieving his goal of being medication-free. This was accomplished using a low carbohydrate diet—in his words: “more a lifestyle than a diet.” The weight loss enabled him to take more exercise, join a gym, and take up yoga. He has come off sugar altogether and cut out bread (he previously consumed a lot of this), potatoes, pasta, cereals, and rice. This has led to greater consumption of green vegetables, but also eggs, full fat Greek yoghurt, and cheese.

The weight loss has been maintained for a year, so he weighs less now than at any time in his adult life. The goal of coming off all drugs was achieved in a stepwise manner as he lost weight—first metformin, then perindopril, followed by simvastatin and aspirin.

His weight loss has been matched by improvements in other parameters: HbA1c down from 52 mmol/mol to 43 mmol/mol (6.9% to 6.2%), blood pressure from 130/80 mm Hg to 117/70 mm Hg, GGT from 59 U/L to 19 U/L, and alanine aminotransferase from 53 U/L to 20 U/L. Of particular note is that—despite eating more eggs and lots of full fat Greek yoghurt, and stopping statins—his cholesterol:high density lipoprotein ratio has improved slightly from 2.8 to 2.7, and serum triglycerides have improved from 1.3 mmol/L to 1.1 mmol/L.

His bowel problems and abdominal pains ceased within days of stopping metformin, his energy returned, and he now needs an hour and a half less sleep a day.

In general he reports feeling “just much younger again.”


Cite this as: BMJ 2015;351:h4023


  • Competing interests: We have read and understood BMJ policy on declaration of interests and declare the following interests: None.

  • Provenance and peer review: Not commissioned; externally peer reviewed.

  • Patient consent obtained.


View Abstract