Urine dipsticks are not accurate for detecting mild ketosis during a severely energy restricted diet

Summary Background Detection of the mild ketosis induced by severely energy‐restricted diets may be a clinically useful way to monitor and promote dietary adherence. Mild ketosis is often assessed using urine dipsticks, but accuracy for this purpose has not been tested. Objective To determine the accuracy of urine dipsticks to detect mild ketosis during adherence to a severely energy‐restricted diet. Methods Two hundred and sixty three (263) fasting urine and 263 fasting blood samples were taken from 50 women (mean [standard deviation, SD] age 58.0 [4.3] years and body mass index 34.3 [2.4] kg/m2) before and at six time points during or for up to 10 weeks after 16 weeks of severe energy restriction, achieved with a total meal replacement diet. The amount of ketones (acetoacetate) in the urine was classified as ‘0 (Negative)’, ‘+/− (Trace)’, ‘+ (Weak)’ or ‘++ (Medium)’ by urine dipsticks (Ketostix, Bayer). The concentration of ketones (β‐hydroxybutyrate) in the blood was measured with our reference method, a portable ketone monitor (FreeStyle Optium, Abbott). The diagnostic accuracy of the urine dipsticks was assessed from the percent of instances when a person was actually ‘in ketosis’ (as defined by a blood β‐hydroxybutyrate concentration at or above three different thresholds) that were also identified by the urine dipsticks as being from a person in ketosis (the percent ‘true positives’ or sensitivity), as well as the percent of instances when a person was not in ketosis (as defined by the blood monitor result) was correctly identified as such with the urine dipstick (the percent ‘true negatives’ or specificity). Thresholds of ≥0.3mM, ≥0.5mM or ≥1.0mM were selected, because mean blood concentrations of β‐hydroxybutyrate during ketogenic diets are approximately 0.5mM. Sensitivity and specificity were then used to generate receiver operating characteristic curves, with the area under these curves indicating the ability of the dipsticks to correctly identify people in ketosis (1 = perfect results, 0.5 = random results). Results At threshold blood β‐hydroxybutyrate concentrations of ≥0.3mM, ≥0.5mM and ≥1.0mM, the sensitivity of the urine dipsticks was 35%, 52% and 76%; the specificity was 100%, 97% and 78%; and the area under the receiver operating characteristic curves was 0.67, 0.74 and 0.77, respectively. These low levels of sensitivity mean that 65%, 48% or 24% of the instances when a person was in ketosis were not detected by the urine dipsticks. Conclusion Urine dipsticks are not an accurate or clinically useful means of detecting mild ketosis in people undergoing a severely energy‐restricted diet and should thus not be recommended in clinical treatment protocols. If monitoring of mild ketosis is indicated (eg, to monitor or help promote adherence to a severely energy‐restricted diet), then blood monitors should be used instead.

Results: At threshold blood β-hydroxybutyrate concentrations of ≥0.3mM, ≥0.5mM and ≥1.0mM, the sensitivity of the urine dipsticks was 35%, 52% and 76%; the specificity was 100%, 97% and 78%; and the area under the receiver operating characteristic curves was 0.67, 0.74 and 0.77, respectively. These low levels of sensitivity mean that 65%, 48% or 24% of the instances when a person was in ketosis were not detected by the urine dipsticks.
Conclusion: Urine dipsticks are not an accurate or clinically useful means of detecting mild ketosis in people undergoing a severely energy-restricted diet and should thus not be recommended in clinical treatment protocols. If monitoring of mild ketosis is indicated (eg, to monitor or help promote adherence to a severely energy-restricted diet), then blood monitors should be used instead.  1,2 Severely energy-restricted diets are a broad term used to describe very low energy diets (VLEDs, <3,300 kJ or 800 kcal per day) and low energy diets (LEDs, <5,000 kJ or 1,200 kcal per day). VLEDs and LEDs often involve total or partial replacement of food intake with specially formulated meal replacement products. 3,4 Despite such a severe restriction of energy intake, adherence is surprisingly high with severely energyrestricted diets, as evidenced by the rapid and substantial weight loss achieved. [5][6][7] High adherence to the severe restriction of energy intake is thought to be due at least in part to the appetite supressing effect shown to be associated with these diets. 8,9 This appetite-suppressing effect may be due to the associated mild ketosis. 10,11 Ketosis is a condition where hepatic production of ketone bodies or 'ketones' (β-hydroxybutyrate, acetoacetate and acetone) is increased, leading to elevations in their concentrations in various body fluids. Ketosis is a coordinated metabolic response that provides ketones as an alternative fuel source for the brain and other organs when glucose is in short supply. The mild ketosis associated with a severely energy-restricted diet, where blood β-hydroxybutyrate concentrations rise to an average of about 0.5mM, 10 is distinct from the life-threatening state of ketoacidosis (where blood β-hydroxybutyrate concentrations are >3mM), which occurs only in people with diabetes and is accompanied by concomitant hyperglycaemia and a decrease in plasma pH. 12 Given the purported association between mild ketosis and appetite suppression, the detection of ketosis during a severely energy-restricted diet may be a clinically useful way to monitor and promote adherence.
A strong body of evidence suggests that higher level of adherence to an energy-restricted diet, regardless of the type of diet, is an important factor in weight loss success. [13][14][15][16] It follows then that strategies that help to promote adherence to an energy-restricted dietary intervention could improve weight loss success. There are numerous factors that may influence an individual's level of adherence to an energy-restricted diet, and clinicians and consumers will likely need access to a range of strategies to promote adherence.
In energy-restricted dietary interventions, adherence is commonly monitored by recording food intake or body weight. As well as monitoring adherence, these methods are thought to help promote adherence, and thereby weight loss. To our knowledge, no studies have investigated whether monitoring adherence by assessing ketosis during a severely energy-restricted diet, or to other ketogenic diets, promotes adherence.
There are two types of testing commercially available for monitoring ketosis-urine dipsticks and blood monitors. Urine dipsticks detect the presence of acetoacetate in urine by a colorimetric reaction with nitroprusside. Urine dipsticks have been in clinical use for more than 50 years, initially to detect life-threating ketoacidosis in individuals with diabetes. 17 In contrast, blood monitors detect the concentration of β-hydroxybutyrate in blood, typically capillary blood from a finger prick. Blood monitors have overcome some of the limitations of urine dipsticks, most notably that urine dipsticks only provide a surrogate marker of the more clinically relevant β-hydroxybutyrate, as well as the difficulty of obtaining a urine sample from an unconscious patient. Blood ketone monitors have been commercially available to the general public for approximately 10 years, 17 but they remain predominantly used by clinicians rather than consumers.
There are a number of reasons why clinicians or consumers with diabetes may prefer to use urine dipsticks instead of blood monitors. Compared to blood monitors, urine dipsticks are equally able to identify 'true positive' cases of ketoacidosis in people with diabetes (high sensitivity), albeit they have a lower ability to identify 'true negatives' among people with diabetic ketoacidosis (lower specificity). [17][18][19][20] As well as being less invasive than blood ketone monitors (some people may be averse to pricking their finger), urine dipsticks are at least seven times cheaper than blood monitors, at around US$0.10 per dipstick compared with around US $0.70 per test trip for the blood monitors. Furthermore, the blood ketone monitor itself typically costs at least another US$25.00.
In light of the above-mentioned benefits of urine dipsticks over blood monitors for gauging ketosis, some clinical treatment protocols and resources for severely energy-restricted diets administered using total 21 or even partial 22

| Severely energy-restricted diet
Details of the study protocol and 12-month intervention for the severely energy-restricted arm of the TEMPO Diet Trial have been reported in detail previously. [23][24][25] Briefly, participants were randomized to 4 months (16 weeks) of severe energy restriction of 65% to 75% relative to estimated energy expenditure, using a total meal replacement diet, followed by moderate energy restriction of 35% to 25% for an additional 8 months (36 weeks). The total meal replacement products used in the severely energy-restricted diet provided between 2,630 (630 kcal) and 3,260 kJ (780 kcal) per day in this group of participants. Participants were also allowed to consume two cups of non-starchy vegetables and a teaspoon of oil each day.

| Statistical analysis
The accuracy of urine dipsticks to detect ketosis was determined by calculating sensitivity and specificity, which were used to generate receiver operating characteristic curves. Sensitivity is the ability of the urine dipstick (test method) to detect the presence of acetoacetate in the urine when a person is in ketosis, as defined by their blood β-hydroxybutyrate concentration (measured using the blood ketone monitor, the reference method) being above a particular threshold. In other words, sensitivity is the proportion of 'true positives' that are detected by the urine dipstick. Specificity is the ability of the urine dipstick to not detect the presence of acetoacetate in the urine when a person is not in ketosis, as defined above. That is, specificity is the proportion of 'true negatives' that are detected by the urine dipstick.
A receiver operating characteristic curve is a graphical representation of the relationship, or trade-off, between sensitivity (shown on the yaxis) and specificity (represented on the x-axis as 100-specificity). The greater the area under the receiver operator curve, the greater the likelihood that the urine dipstick will detect 100% of true positives and 100% of true negatives, with 0% false positives and 0% false negatives (a perfect diagnostic test). An area under the receiver operating characteristic curve of 0.5 means that there is a not only 50% chance that the urine dipstick will detect true positives and true negatives but also a 50% chance it will detect false positives and false negatives (random results). Thus, the greater the area under the receiver operating characteristic curves (or the closer it is to 1), the more accurate the diagnostic test. In this study, the area under the receiver operating characteristic curve tells us how accurate the urine dipstick is at identifying whether or not someone is in ketosis.
The blood ketone monitor was considered to be the reference method as it has been shown to be highly accurate in comparison with laboratory-determined blood β-hydroxybutyrate concentrations. 18,19 Three different thresholds of blood β-hydroxybutyrate concentrations were used to define whether or not a person was in ketosis: ≥0.3mM, ≥0.5mM and ≥1.0mM. The rationale for this is that although the average blood concentrations of β-hydroxybutyrate for individuals during a severely energy-restricted diet has been shown to be approximately 0.5mM, 10 other research has shown appetite suppression in association with lower levels of ketosis (0.3mM) during a severely energyrestricted diet. 8 Therefore, it would be clinically relevant to be able to determine the ability of the urine dipsticks to act as a diagnostic for ketosis using both of these concentrations as thresholds for ketosis. A higher threshold of ≥1.0mM blood β-hydroxybutyrate was also included, because these concentrations are not uncommon during a severely energy-restricted diet in healthy individuals, 10    0.34 to 1.50) mM, respectively. This demonstrates that although there was an up to two-fold difference in the blood β-hydroxybutyrate concentration at times when urine did and did not have acetoacetate detected, there was no clear relationship between approximate concentration of acetoacetate in the urine and blood β-hydroxybutyrate concentrations. Table 1 shows the sensitivity and specificity of the urine dipsticks to detect ketosis at the three different thresholds of blood β-hydroxybutyrate concentrations used to define a state of ketosis.
These data show that at blood β-hydroxybutyrate levels of ≥0.3mM, ≥0.5mM and ≥1.0mM, the sensitivity of the urine dipstick was approximately 35%, 52% and 76%, respectively. In other words, the urine dipsticks correctly identified only 35% to 76% of the individuals who were in ketosis, as defined by their blood β-hydroxybutyrate concentrations being ≥0.3 to ≥1.0mM. In contrast to this low sensitivity (low rate of true positives), the urine dipsticks had high specificity.
In other words, the urine dipsticks correctly identified 78% to 100% of the individuals who were not in ketosis, as determined by their blood β-hydroxybutyrate concentrations being <1.0 to <0.3mM. Figure 2 shows the sensitivity and specificity for each of the three different thresholds of blood β-hydroxybutyrate concentrations used to define ketosis, as receiver operating characteristic curves. The area under the receiver operating characteristic curves increased slightly with higher thresholds for ketosis (ie, from 0.67 for blood β-hydroxybutyrate ≥0.3mM, to 0.77 for blood β-hydroxybutyrate ≥1.0mM). An area under the curve of 0.67 corresponds to a 67% chance of detecting true positives and true negatives, whereas a value of 0.77 corresponds to a 77% chance of detecting true positives and true negatives. This means that the urine dipsticks were slightly more accurate at detecting ketosis as defined as ≥1.0mM blood β-hydroxybutyrate concentrations than at ≥0.5 or ≥0.3mM.

| DISCUSSION
This study showed that urine dipsticks are not an accurate or clinically useful way of detecting mild ketosis in people undergoing severely energy-restricted diets. Urine dipsticks had low sensitivity for identifying when participants were in mild ketosis (as defined by blood β-hydroxybutyrate concentrations being ≥0.3 and ≥0.5mM), because in only 35% and 52% of those occasions, respectively, did their urine test positive for the presence of ketones with the dipstick. This means that for 48% to 65% of the times when participants were actually in ketosis, as defined by their blood β-hydroxybutyrate concentrations being at or over these 0.3 and 0.5mM thresholds, the urine dipstick indicated that they were not in ketosis. That is, the rate of 'false negative' results were 48% to 65%. Even at a higher threshold for defining ketosis (ie, ≥1.0mM β-hydroxybutyrate in blood), urine dipsticks still had unacceptably low sensitivity: They only detected the presence of acetoacetate in 76% of cases. This means that 24% of the times when participants had a blood β-hydroxybutyrate concentration of ≥1.0mM, the highest level normally associated with severely energy-restricted diets, 10 the urine dipstick result suggested that they were not in ketosis at all (false negative). A false negative rate of 24% to 65% is not good enough for a test used in clinical practice. These data indicate that urine dipsticks should not be recommended in clinical treatment protocols to monitor or promote adherence to a severely energy-restricted diet.
Our findings extend those of previous studies, which have compared the accuracy of urine dipsticks with blood ketone monitors to diagnose ketoacidosis in people with diabetes. 20 A recent systematic review of this topic identified five studies-three of which showed urine dipsticks to have high sensitivity (≥89%, range 89% to 98%) and variable specificity (35% to 100%). 20 In one of the other two of five studies, sensitivity of the urine dipsticks was low (33%), but specificity was high (94%), and in the remaining study, sensitivity and specificity were not reported. 20 In contrast, in the present study of participants with varying states of mild ketosis, the sensitivity of the urine dipsticks was low, but the specificity was high. Therefore, although urine dipsticks may have been found to be acceptably accurate in the context of detecting diabetic ketoacidosis in the majority of studies, the current study has shown that they are not accurate at low levels of ketosis.
Our finding that urine dipsticks are not accurate at detecting mild levels of ketosis in people without diabetes is likely explained by the fact that urine dipsticks detect different ketones (acetoacetate) from those detected by blood monitors (β-hydroxybutyrate), and these ketones differ in their rates of production, utilization and excretion (see Laffel 12 for an in-depth review of ketone physiology). For instance, the ketone ratio (defined as the ratio of blood β-hydroxybutyrate to blood acetoacetate concentrations) increases during ketosis, as β-hydroxybutyrate is produced in greater quantities than acetoacetate. 12 It appears therefore that while during diabetic ketoacidosis, the production of acetoacetate appears to be sufficient to enable detection of ketones in the urine, during mild ketosis, acetoacetate is not produced in sufficient quantitates to be consistently detectable in the urine.
If urine dipsticks should not be used to monitor or promote adherence to a severely energy-restricted diet, does this mean that blood ketone monitors, which provide an accurate indication of mild ketosis, 18 should be used? Putting aside issues of invasiveness (some people may be averse to pricking their finger for capillary blood collection) and cost (some people may not wish to spend money on them), the outstanding question is whether monitoring adherence to severe energy restriction promotes adherence to severe energy restriction.
The literature surrounding the benefits of other forms of self-monitoring-such as recording food intake or measuring body weight, which have proven benefits for promoting adherence-suggests that it might be useful to monitor blood β-hydroxybutyrate concentrations using F I G U R E 1 The distribution of blood β-hydroxybutyrate concentrations measured using a blood ketone monitor (FreeStyle Optium, Abbott), according to whether acetoacetate was detected in the urine using a urine dipstick (Ketostix, Bayer). Boxes represent the 25th and 75th percentiles of the distribution around the median. Whiskers represent 1.5 times the interquartile range, and isolated points represent values lying outside this range Sensitivity and specificity of the urine dipstick (Ketostix, Bayer) to detect the presence of acetoacetate in the urine at three different thresholds of ketosis, as defined by the blood β-hydroxybutyrate concentrations shown in the left-most column  18 Thirdly, the assessment of ketones in the urine with the use of a dipstick is subjective, as it involves comparison of the colour on the dipstick with the colour scale on the back of the packet. This limitation was contained as much as possible by limiting the assessment of urine acetoacetate levels to two trained researchers. Another limitation is that venous blood was collected from a cannula and not capillary blood collected via a finger prick. However, previous studies have shown that the concentration of β-hydroxybutyrate tends to be higher in capillary blood than in venous blood. 26 This suggests that the sensitivity of the urine dipsticks may be even lower than our observations, further supporting our conclusions that urine dipsticks are not accurate for detecting mild ketosis.
In summary, this study showed that urine dipsticks should not be recommended in clinical treatment protocols to monitor adherence to severely-energy restricted diets, as they do not accurately reflect blood ketone levels typically found in people on these diets. If the detection of mild ketosis is indicated as a way to monitor or promote adherence, blood monitors should be used. Given the purported benefits of appetite suppression associated with mild ketosis, as well as the known benefits of self-monitoring to improve weight loss success, it could be of benefit to investigate the use of blood monitors during a severely energy-restricted diet to improve adherence.