Capetonian Peter Venn lost 22kg on a low-carbohydrate diet in 2014. Then he started eating carbohydrates again and regained the weight.
He said friends and colleagues usually quit their low-carbohydrate diets within three weeks.
A new study supports Venn's observation: few people adhere to low-carb diets for up to six months.
The study, compiled by Johannesburg Centre for Diabetes and Endocrinology dietician Hamish van Wyk and two Welsh dieticians, was recently printed in the British journal Diabetic Medicine.
The trio examined more than 153 studies of diabetic patients eating either low-carb or high-carb diets to see which one was better in helping control blood sugar.
Only 12 studies met their criteria of being randomised control studies in which two groups of diabetic people on different diets were followed and compared for four weeks or longer. Dieters also had to report what they ate at the end of their study.
But researchers found no difference in blood sugar control between low-carb and high-carb dieters – possibly because low-carb dieters did not exist.
In one diet, the difference between how much carbohydrates each group ate was only 8g a day.
When low-carb dieters were asked after six months what they had eaten, their average daily carb intake was 166g – roughly the same as the high-carb dieters' consumption.
A low-carb diet is usually considered to contain less than 50g of carbs a day.
Professor Tim Noakes made the low-carb diet popular with his book The Real Meal Revolution, claiming the regimen helped prevent diabetes.
Van Wyk said his research showed strict low-carb diets might not be attainable because even dieters taking part in a study who visited a dietician 18 times over six months did not stick to their diet all of the time.
The American Diabetes Association has identified carbohydrates are the number-one nutrient that influences sugar control in diabetics and said that a reduced carb intake should lower blood sugar and allow better control. But Van Wyk noted that studies differed in their conclusions: some found a low carb intake was better for diabetics, while others found high-carb diets to be equally good.
Such confusion may be due to a mixture of low-quality and high-quality studies, he suggested, adding that higher-quality studies had more consistent results.
In one of the 12 diets that met the researchers' criteria, people lost more weight on the low-carb diet but their waist circumference measurements were not smaller than those of high-carb dieters.
A high percentage of waist fat is a predictor of diabetes and heart disease risk.
The study concluded that calorie intake – energy, food and drinks consumed – was the best predictor of body weight.