The ketogenic diet (KD) is a very low-carbohydrate (<30–40 g/day, ≈ 5 % of energy), high-fat (>60 % of energy) and adequate-protein diet that without limiting calories (eucaloric) induces a metabolic condition called “physiological ketosis”, thus distinguishing it from the severe pathological ketosis (ketoacidosis) often observed in uncontrolled diabetes [1, 2]. The lower insulin level via low carbohydrate intake raises the serum glucagon level and induces lipolysis, leading to the augmented availability of fatty acids [3]. These are metabolised by the mitochondria of liver cells to two water-soluble types of ketones (ketogenesis): acetoacetate (AA) and ß-hydroxybutyrate (BHB) [4]. A third ketone body, acetone, is produced by the enzymatic decarboxylation of AA, and is largely exhaled unused. AA and BHB become the primary energy source for the brain and other tissues [4].

For the optimal clinical management of children with epilepsy on a KD, the International Ketogenic Diet Study Group recommends monitoring compliance by urine testing for ketosis several times a week [5]. Except for one study that did not test for ketosis [6], the majority took daily measurements of urinary ketone bodies without [7–9] or with contrary recommendations on the precise time of the day for measuring: 8:00 [10] or in the afternoon [11]. Furthermore, a study in healthy adults who consumed four ketogenic meals over 12 h revealed that intra-day blood and urinary concentrations of ketone bodies fluctuate strongly [12] and can be affected by several factors [13].

At the beginning of our KetoPerformance study in healthy adults investigating the impact of a 6-week KD without caloric restriction on physical performance, we were unable to give concrete advice to our subjects concerning when it is best to test for the presence of ketosis in urine during the day. Therefore, we aimed to investigate in a portion of the study population the intra-day course of blood and urinary concentrations of ketone bodies during a KD.

Regular urine testing for ketosis is the most common and feasible approach to monitoring a KD's adherence, and it should be an essential factor in all low-carbohydrate and KD intervention studies. Daily self-testing of urinary ketones is the norm in the literature [7, 9–11, 15–17], except for one study investigating the capacity for moderate endurance exercise after a KD, where urine ketones were measured twice daily [18]. A few others did collect 24-h urine [19, 20], while others only measured blood ketones [21, 22] or did not even test for ketosis [6]. The International KD Study Group on the optimal clinical management of children following a KD recommends urine ketosis evaluation routinely, several times per week, but they give no advice on when to measure [5]. Hence, studies in children with medically refractory epilepsy treated with a KD had the families check urine ketones daily and thus also without specifying the time of day [7, 9, 16]. KD studies in adults without epilepsy usually did not address this issue either [8, 17, 18]. There is widespread disagreement regarding measurements of blood BHB or urinary AA in epilepsy patients, when beside diet compliance its efficacy should be monitored. Two studies found that blood BHB correlated better with seizure control than urinary testing of AA [7, 23]. However, the present trial was not designed to investigate this issue. An important finding of our study is that the course of blood and urine ketone bodies was very similar throughout the period of 24 h.

We identified just two publications giving recommendations on the precise time of the day for subjects to measure [10, 11]. The present substudy reveals that ketonuria in subjects during the sixth week of a KD and with stable ketosis can be most reliably detected in the first morning urine and several hours after dinner late in the evening. The slightly inferior detection rate of the second morning urine could be because subjects were obliged to urinate already 4 h previously at 3:00 during the night, yielding a 100 % detection rate. These results endorse the 8:00 recommendation to healthy adults put on a 6-week KD made by Sharman et al. [10].

The least favourable period of time according to our data was from 10:00 till 19:00, which stands in contrast to the recommendation by Klement et al. [11] (namely to use the urine reagent strips preferably in the afternoon). The considerably higher false negative rate during daytime measurements could be due to two opposing reasons: (1) meals and (2) physical activity. We detected continuous decreases in blood BHB and urinary AA lasting up to 3 h after each principal meal, except for urinary AA increasing after dinner. This concurs with the results in healthy adults published by Musa-Veloso et al. [12], where the time elapsed after each ketogenic meal revealed a significant negative effect on blood BHB and AA concentrations and that blood ketone bodies are rapidly metabolised for energy. Schwartz et al. [24] obtained similar results and advise children with epilepsy to consume more frequent meals and snacks to keep blood ketone concentrations more stable and to avoid the acute rises and falls resulting from ketogenic meals. Some (≈40 %) of our subjects performed no additional physical activities, whereas the others engaged in activities of light to vigorous intensity, factors that may also influence temporary daytime fluctuations in blood and urinary ketones, as muscles consume this energy source rapidly [25].

One of our study’s limitations is that we did not measure initial concentrations of ketone bodies under the usual high carbohydrate diet conditions. A further limitation is our study population's wide variation in additional physical activities.

However, our study has several strengths, including the eucaloric character of the diet to eliminate the confounding influence of negative energy balance and the synchronisation of the times of equal fluid intake volumes and food intake. A further strength is that the substudy was conducted during a KD's the sixth week, thus assuring a metabolic shift to stable ketosis and a study period outside the metabolic adjustment period lasting several days upon the initiation of such a diet.

In summary, our results suggest that urine testing for ketosis in the scope of monitoring compliance with a KD in adults is best done in the early morning urine and several hours after dinner late in the evening. Hence study participants in future studies with KDs can be given recommendations on the precise time of the day for measuring ketone bodies in urine. Our results should only apply with caution to children with epilepsy, as additional research is needed for this patient cohort.