Vous êtes sur la page 1sur 11


DOI: 10.1111/dme.12604

Review Article
Utility of ketone measurement in the prevention,
diagnosis and management of diabetic ketoacidosis

S. Misra1,2 and N. S. Oliver1

Department of Diabetes, Endocrinology and Metabolism, Imperial College London and 2Clincal Biochemistry and Metabolic Medicine, Imperial Healthcare NHS
Trust, London, UK

Accepted 6 October 2014

Ketone measurement is advocated for the diagnosis of diabetic ketoacidosis and assessment of its severity. Assessing the
evidence base for ketone measurement in clinical practice is challenging because multiple methods are available but there
is a lack of consensus about which is preferable. Evaluating the utility of ketone measurement is additionally problematic
because of variability in the biochemical definition of ketoacidosis internationally and in the proposed thresholds for
ketone measures. This has led to conflicting guidance from expert bodies on how ketone measurement should be used in
the management of ketoacidosis. The development of point-of-care devices that can reliably measure the capillary blood
ketone b-hydroxybutyrate (BOHB) has widened the spectrum of applications of ketone measurement, but whether the
evidence base supporting these applications is robust enough to warrant their incorporation into routine clinical practice
remains unclear. The imprecision of capillary blood ketone measures at higher values, the lack of availability of routine
laboratory-based assays for BOHB and the continued cost-effectiveness of urine ketone assessment prompt further
discussion on the role of capillary blood ketone assessment in ketoacidosis. In the present article, we review the various
existing methods of ketone measurement, the precision of capillary blood ketone as compared with other measures, its
diagnostic accuracy in predicting ketoacidosis and other clinical applications including prevention, assessment of severity
and resolution of ketoacidosis.
Diabet. Med. 32, 14–23 (2015)

In the present article, we discuss the biochemical and

analytical aspects of ketone measurement and critically
Diabetic ketoacidosis is a serious, acute complication of review the evidence base for the clinical applications.
diabetes, associated with significant morbidity and mortality
[1]. It results from the accumulation of ketone bodies
Methods of ketone measurement
[b-hydroxybutyrate (BOHB also known as 3-hydroxybuty-
rate) and acetoacetate (AcAc)] and accompanying hydrogen The main ketoacids are AcAc and BOHB, which usually
ions, produced after lipolysis of triglycerides in the absence circulate in low equimolar concentrations (< 0.5 mmol/l in
or deficiency of insulin [2,3]. Diagnosis of ketoacidosis healthy adult blood after an overnight fast) [2]. Acetone,
requires a measure of glycaemia, identification of ketones another ketone body, is formed from the spontaneous
and assessment of acid-base status. The recognition that the decarboxylation of AcAc, is volatile, and present in very
accumulation of ketones and resultant ketoacidosis are the low concentrations. Ketones can currently be measured in
underlying biochemical problem has led to a shift in focus the urine, capillary blood and serum.
away from glucose measurement for both diagnosis and
ongoing management. Ketone detection has traditionally
Urine ketones
relied on urine assessment using qualitative or semi-quanti-
tative dipsticks; however, the development of point-of-care Urine dipsticks (detecting ketones qualitatively or
hand-held analysers capable of measuring capillary blood semi-quantitatively) impregnated with nitroprusside reagent,
ketones has allowed the rapid assessment of blood ketones at react only with AcAc and acetone but not BOHB. During
the bedside and in the home. ketoacidosis, the ratio of AcAc to BOHB favours the latter
and can be as high as 1:10, as BOHB is formed from the
Correspondence to: Shivani Misra. E-mail: smisra@imperial.ac.uk reduction of AcAc [2]. This shift in equilibrium accounts

ª 2014 The Authors.

14 Diabetic Medicine ª 2014 Diabetes UK
Review article DIABETICMedicine

for the discordant results often encountered between urine facility to measure serum BOHB, as until recently its use has
ketone measurement and the severity of ketoacidosis, as been limited to the investigation of neonatal hypoglycaemia.
assessed by pH and bicarbonate, as measurement of AcAc in The availability of laboratory analysis is essential as it has
urine will underestimate the extent of ketonaemia [4]. Urine the facility to dilute samples outside of the linear range and
ketone concentrations also reflect the total accumulated to confirm analysis of capillary blood ketone outside the
since the last void and levels measured may therefore become range of point-of-care devices.
uncoupled from real-time changes.
Urine ketone assessment is also limited by the paradoxical
Capillary blood ketones
increase in urinary AcAc during resolution of ketoacidosis,
which is attributable to the conversion of excess BOHB back The measurement of capillary blood ketones at the point of
to AcAc, as acidosis resolves. Laboratory guidelines from the care has recently been advocated [10]. The advantages of any
American Diabetes Association state that for use by patients point-of-care test apply to this method, in that a rapid
in the home setting, control materials would be desirable to quantitative result can be obtained in real time, allowing
ensure accuracy of test results, but these are not commer- immediate changes in management. This is advantageous in
cially available [4]. the diabetes setting, where devices can be used to empower
These limitations, coupled with the semi-quantitative and self-management.
subjective interpretation of the test, the lag time between There are currently two approved meters capable of
serum and urine metabolic changes and difficulties with measuring capillary blood ketone using dry-chemistry meth-
serial urine samples in a dehydrated unwell patient, have led odology: the FreeStyleOptium (Precision Xceed; Abbott
to the suggestion that blood ketone measurement is superior Diabetes Care, Maidenhead, UK) and the Glucomen LX Plus
to urine assessment. This is further supported by the (Menarini, Florence, Italy). Versions of these meters are also
apparent delay in urine dipstick assessment compared with available for inpatient monitoring with connectivity. The
blood ketone assessment in those presenting acutely with devices measure BOHB using BOHB dehydrogenase coupled
hyperglycaemia [5,6] and by the fact that many acutely with electrochemical detection. These methods and older
unwell patients are unable to provide a urine sample in the variants have been evaluated against reference enzymatic
emergency department [7]. spectrophotometric assays and generally show good correla-
tion [6,11–14]. The Abbott devices are perhaps the most
Blood ketone measurement
Older-generation point-of-care devices, such as the Med-
Ketones may be measured in serum, plasma or whole isense Optium (Abbott Diabetes Care), have now been
blood samples using laboratory-based analysers, or in superseded by newer devices. It is worth noting that many of
finger-prick samples using capillary blood. Several different the studies underpinning current guidance incorporate data
assay methods (colorimetric, gas chromatography, capillary generated using older devices. The main difference between
electrophoresis and enzymatic) have been described for older and newer devices is a shorter measurement time, with
specific BOHB detection, but enzymatic methods appear to smaller capillary blood volume and an improved coefficient
be the most widely used [4]. The measurement of AcAc of variation at lower range measurements: 5.9% at
using nitroprusside in blood is not advocated for diagnos- 0.68 mmol/l (Precision Xceed pro) vs 8.2–14.6% at concen-
tic purposes, as BOHB will not be detected. It is important trations below 1 mmol/l (Medisense Optium) [12–16].
to make the distinction between laboratory measurement Whilst these advances are important developments, the
of serum BOHB and point-of-care measurement of coefficient of variation outside of the normal range (in the
capillary ketones (BOHB) in whole blood. This is analo- ketoacidotic range) has not improved considerably. In
gous to the difference between plasma glucose and addition, evaluations of the newer devices show a loss of
capillary blood glucose, and the same advantages and linearity and imprecision at high levels of BOHB, which seem
disadvantages apply. not to have been reported for earlier versions, except in one
study [17]. Manufacturers quote linearity up to ketone values
of 8 mmol/l, after which a ‘Hi’ reading is reported, but this
Serum b-hydroxybutyrate
contrasts with reports validating meters in clinical settings,
A variety of enzyme-based assays are available for the which show a loss of linearity above levels of 3 mmol/l
detection of BOHB in serum. The quoted reference method [18,19].
(Stanbio reagents on Cobas 501 analyser) is linear up to A recent study also compared the Precision Xceed meter
4.5 mmol/l, but levels above this can be diluted into the with a non-enzymatic reference method using liquid chro-
linear range. Coefficients of variation are quoted as being matography tandem mass spectrometry traceable to weighed
between 1 and 4%, but higher outside the linear range [8,9]. standard solutions, using dried blood spots. They found
Many laboratory-based analysers are capable of measuring linearity up to concentrations of 6 mmol/l (r = 0.97), after
BOHB, but it is unclear how many laboratories have the which the Abbott method showed saturation, and was

ª 2014 The Authors.

Diabetic Medicine ª 2014 Diabetes UK 15
DIABETICMedicine Ketone measurement in diabetic ketoacidosis  S. Misra and N. S. Oliver

non-linear; however, inter-individual coefficients of variation is a facility to perform laboratory ketone testing when meter
were higher at ~ 6 mmol/l, leading the authors to concur that readings are invalid, and that there is sufficient connectivity
capillary blood ketone BOHB levels should not be reported to electronic patient records [10]. These standards are a
above 3 mmol/l [20]. This contrasts with the manufac- challenge to maintain, but if capillary blood ketone testing is
turer-stated range of up to 8 mmol/l. to become widespread in and outside of hospital, the
These findings are important if capillary blood ketone provision of these services must be taken into consideration.
measurement is used to monitor resolution of ketoacidosis in Reassuringly both capillary blood ketone and serum BOHB
conjunction with other biochemical variables, as advised in measurement have external quality assurance schemes, in
current guidance, because reported values of > 3 mmol/l which laboratories can participate. There is therefore infra-
may, in reality, be higher [10]. This is especially important if structure in place to ensure quality is maintained [22,23].
higher levels of capillary blood ketone are used to guide level
of care; for example, blood ketones > 6 mmol/l to guide
Guideline recommendations
admission to high dependency areas. Guidance also advo-
cates a target BOHB reduction of 0.5 mmol/l/h, and to Published consensus criteria from the Joint British Diabetes
achieve this reliably, levels outside of the range of linearity of Societies [10], and American Diabetes Association [24],
point-of-care devices need to be determined [10]. identify a role for ketone measurement in the diagnosis of
It could be argued that the absolute value of capillary ketoacidosis, and also advocate its use for monitoring of
blood ketones is not essential for the clinical management of ketonaemia resolution, representing a shift in focus from the
ketoacidosis and differentiation of high from low levels traditional variables assessed, such as bicarbonate, glycaemia
would be sufficient, but this needs to confirmed in high-qual- and pH. However, the guidance available lacks consensus on
ity studies that not only assess diagnostic accuracy, but also both the method of ketone measurement and the clinical
clinical utility and impact on patient outcomes. applications; the Joint British Diabetes Societies recommends
Most ketone meters also have recommended operational use of capillary blood or urine ketones for diagnosis, and
haematocrit ranges. For the PrecisionXceed the upper limit is capillary blood ketones for monitoring severity and resolu-
set at 60%, above which it is recommended that capillary tion. The American Diabetes Association guidance stipulates
blood ketone measurement is not used; however, dehydra- the use of serum and urine ketone measurements for
tion and hyperosmolality are clinical states nearly always diagnosis, and direct BOHB measurement for monitoring
seen in ketoacidosis, and haematocrits may be well above the resolution, but do not specify how this should be measured.
60% threshold. As with any test, these limitations should be The laboratory guidelines from the American Diabetes
taken into consideration when the measurement is being Association [4] contradict the American Diabetes Associa-
made, although it should be noted that bedside ketone tion ketoacidosis guidance and recommend blood, but not
monitoring is one method of several to aid making the urine, ketone measurements for diagnosis of ketoacidosis and
diagnosis. state that urine ketone measurement should not be used to
There are clear potential advantages of capillary blood monitor treatment. Meanwhile the International Society for
ketone measurement over urine ketone assessment, both Pediatric and Adolescent Diabetes clinical practice guidelines
biochemically and analytically. A recent systematic review recommend capillary blood ketone use only if laboratory
comparing urine AcAc with blood BOHB (serum or capillary measures are unavailable [25]. Table 1 provides a summary
blood ketones) concluded that blood BOHB assessment was of the recommendations.
associated with reduced frequency of hospitalization, shorter
time to recovery from diabetic ketoacidosis and lower costs;
Clinical applications of ketone
however, only four studies met the inclusion criteria for this
review and the authors acknowledged the quality of evidence
(by National Health and Medical Research Council stan-
Diagnosis of ketoacidosis
dards) underpinning these conclusions was grade ‘C’ or
satisfactory, whilst the studies themselves were heteroge- Biochemically, ketosis and acidosis do not always accom-
neous in terms of design and outcomes assessed [21]. pany each other because, in the early stages of ketosis, excess
Furthermore, it is important to distinguish between studies hydrogen ions may be buffered [15]. Bicarbonate and pH
using serum or capillary blood ketone BOHB, as there are may be affected by factors such as vomiting, respiratory
clear practical differences. compensation and pre-existing acid base disturbances [15]
The cost and lack of availability of widespread capillary and, although glucose values have been incorporated into
blood ketone analysis, mean that in some centres urine diagnostic criteria, the recognition of euglycaemic ketoaci-
ketone testing may still be the favoured choice. Point-of-care dosis and the decoupling of ketosis from hyperglycaemia,
devices must also be compliant with biochemistry laboratory means that hyperglycaemia alone cannot be diagnostic of
protocols, to ensure staff training is adequate, that meters ketoacidosis [10]. Since ketosis is the underlying abnormality
undergo rigorous and frequent quality assurance, that there and given the non-specific nature of changes in bicarbonate

ª 2014 The Authors.

16 Diabetic Medicine ª 2014 Diabetes UK
Review article DIABETICMedicine

Table 1 Summary of guidelines advocating the use of ketone assessment in diabetic ketoacidosis

Definition of Ketones in Laboratory

ketoacidosis Ketones for Ketones for monitoring measurement
Guideline (where given) diagnosis severity resolution advocated

American Diabetes Nitroprusside-based No BOHB for monitoring No

Association on  Glucose reactions for resolution. No
management of > 13.8 mmol/l diagnosis. indication of
hyperglycaemic Qualitative whether serum or
crises, 2009 [19]  pH < 7.3 result in urine capillary blood
or serum ketone assessment
< 18 mmol/l

 Urine or serum ke
tones by nitroprus
side reaction

Joint British Diabetes Capillary Yes, a level Yes, aim for a Yes, if values on
Society, 2011 [9]  Glucose blood > 6 mmol/l reduction of at least capillary blood
> 11 mmol/l or ketone measurement 0.5 mmol/l/h BOHB ketone meter ‘
of BOHB as measured in out-of-range’
known diabetes
or, if unavailable, capillary blood
 Bicarbonate semi-quantitative
assessment of urine
< 15 mmol/l and/
or pH < 7.3

 Ketones > 3 mmol/

l or > 2+ on urine

American Diabetes N/A Urine ketones Not specified Specific BOHB Not specified
Association should not recommended but
laboratory guidance be used to not specified
on ketones in diagnose or whether capillary
diabetes, 2011 [3] monitor ketoacidosis. blood ketone or
Specific BOHB preferred laboratory-based
for diagnosis assessment
International Society Urine ketones No Yes, until urine Yes. Capillary blood
for Paediatric and  Glucose or BOHB ketones cleared ketone assessment
Adolescent Diabetes > 11 mmol/l or if available only if laboratory
guidelines, 2009 BOHB every 2 h unable to provide
[20]  Bicarbonate timely results
< 15 mmol/l and/
or pH < 7.3

 Ketonaemia and

BOHB, b-hydroxybutyrate; AcAc, acetoacetate.

and pH, the focus of diagnosis has shifted to ketone problematic. In addition, there is no clear international
measurement. consensus on the diagnostic variables for ketoacidosis, with
The current guidelines for ketoacidosis diagnosis in the UK the American Diabetes Association suggesting a bicarbonate
include; ketonaemia > 3 mmol/l or > 2+ ketones on urine level of < 18 mmol/l, no specific numerical thresholds for
dipstick, blood glucose > 11 mmol/l or known diabetes ketone measurement and a glucose level of > 14 mmol/l [24].
mellitus and venous bicarbonate < 15 mmol/l or pH level In a retrospective study of 304 patients, serum BOHB
< 7.3 [10]. Many previous studies have compared the measurements were compared with serum bicarbonate levels
sensitivity of capillary ketone testing with that of urine to determine BOHB levels corresponding to the American
ketone, bicarbonate, glycaemia and pH in the diagnosis of Diabetes Association-recommended bicarbonate threshold of
ketoacidosis [6,7,9,26–28]. These studies are heterogeneous 18 mmol/l [9]. A curvilinear relationship between bicarbon-
and use different numerical thresholds, making comparison ate and BOHB was shown, with a BOHB level of 3.8 mmol/l

ª 2014 The Authors.

Diabetic Medicine ª 2014 Diabetes UK 17
DIABETICMedicine Ketone measurement in diabetic ketoacidosis  S. Misra and N. S. Oliver

(3 mmol/l in children) corresponding to a bicarbonate level urine ketone assessment, and that the latter, although equally
of 18 mmol/l. That study suggests that a BOHB minimum sensitive in diagnosing ketoacidosis, may be more valuable in
concentration of 3.8 mmol/l could be incorporated into excluding it and has lower specificity [6,32–34]. It is worth
diagnostic criteria, but the authors do not advocate the use of considering that the two methods of measurement reflect not
point-of-care devices, recognizing the limited precision of only different ketone bodies, but also variable time points,
such analysers above 3 mmol/l. and discrepancies between urine and blood ketones will
In a prospective study, 171 patients presenting to the reflect either resolving ketonaemia (urine ketones high, but
emergency department with hyperglycaemia (> 11 mmol/l) capillary blood ketone normal), or incipient ketoacidosis
and capillary blood ketone > 0.1 mmol/l were assessed [29]. (high capillary blood ketone, with undetectable urine
Ketoacidosis was defined as a glucose level > 11 mmol/l, with ketones) [35]. Furthermore, although capillary blood ketone
serum BOHB > 0.42 mmol/l and a pH level < 7.3. Urine, is superior to urine ketone assessment, it is unclear if it is
serum and capillary ketones were compared between those the best method to diagnose ketoacidosis in a patient with
meeting the criteria for ketoacidosis, and those individuals haemoconcentration, as laboratory testing is the ‘gold
with ketonaemia alone. Capillary blood ketone had a 72% standard’ and could be used as a diagnostic test to support
sensitivity and 82% specificity for the detection of ketoaci- a positive urine dipstick. Whatever the sensitivity and
dosis, compared with 66 and 78%, respectively, using urine specificity of capillary blood ketone measurement, the
ketone assessment. Although capillary and serum ketones did provision of a laboratory-based reference method, through
not differ significantly, the correlation between the values was which higher levels can be diluted, must be available as is the
weak but significant (r = 0.488, p < 0.0001), suggesting that case with glucose. If the limitations of bedside ketone meters
capillary blood ketone could be an alternative to laboratory are fully appreciated, laboratories can be mobilized to
ketone testing, and that it was superior to urine testing in the provide alternative methods. Whilst these are not widely
recognition of ketoacidosis; however, the threshold for available at present, further research into the area will enable
diagnosis of ketoacidosis was low at 0.42 mmol/l, which is a case to be made at a local level.
in the normal range, making it a poor discriminator of
ketoacidosis and comparison with other studies a challenge.
Assessing the severity of ketoacidosis
More recently, a study comparing urine ketone assessment
with capillary blood ketone measurement in the diagnosis of Although there may be an emerging role for ketone assess-
ketoacidosis (defined as glucose ≥14 mmol/l, anion gap > 10 ment in the diagnosis or identification of ketoacidosis, its role
mmol/l, bicarbonate < 18 mmol/l and pH < 7.3 mmol/l) in the assessment of severity of ketoacidosis is more
found that capillary blood ketone had a 98% sensitivity and controversial. Serum BOHB levels have been correlated with
78.6% specificity compared with a urine dipstick sensitivity serum bicarbonate with a modest association; Sheikh-Ali
of 98.1% and a specificity of 35.1% for ketoacidosis [30]. et al. [9] found a correlation coefficient (r = 0.8) between
The proposed capillary blood ketone threshold of 1.5 mmol/l initial serum bicarbonate and BOHB in adults, and Fulop
in that study, however, again contrasts with other studies et al. [36] prospectively measured serum BOHB levels from
that set a higher threshold of BHOB for the diagnosis of patients with ketoacidosis and compared them with serum
ketoacidosis of > 3 mmol/l and, in addition, the study bicarbonate levels (r = 0.69), anion gap (r = 0.52) and
recommended an optimum threshold of 2 mmol/l on arterial pH (r = 0.38). The correlations using only initial
receiver–operator curve analysis. This highlights the diffi- measurement of BOHB were weaker suggesting that BOHB
culty in comparing studies with differences in the definition measurements should not be routinely used to assess initial
of ketoacidosis and proposed thresholds. The similar sensi- ketoacidosis severity. Glucose levels are also poorly corre-
tivities found for capillary blood ketone and urine testing lated to BOHB levels [37]. The correlation between capillary
may mean that a negative urine dipstick remains a valuable blood ketone and ketoacidosis severity at time of presenta-
test in excluding ketoacidosis. tion in 54 patients [30] was substantially weaker, with
The above-mentioned studies show that blood ketone correlation coefficients between capillary blood ketone and
measurement has a role to play in the diagnosis of ketoac- pH, bicarbonate and anion gap of 0.33, 0.25 and 0.16,
idosis, with high positive predictive values at some thresh- respectively. Along with other studies [34,38], these findings
olds. Further work is needed to address the lack of consensus suggest capillary blood ketone values cannot reliably assess
on a diagnostic threshold, the wide intra- and inter-individ- ketoacidosis severity.
ual variability of the test and the heterogeneity of study The guidance from the Joint British Diabetes Society
design, which currently make the data difficult to interpret suggests the presence of biochemical variables such as blood
for clinical practice. This difficulty is further compounded by (capillary or serum unspecified) ketones > 6 mmol/l, bicar-
the variable levels of capillary blood ketone in people with bonate < 5 mmol/l, pH level < 7.1 or anion gap > 16 mmol/l
hyperglycaemia and euglycaemia without ketoacidosis [31]. should prompt referral to a high dependency unit [10];
The literature suggests there is superior utility in serum however, the poor correlation between BOHB and the other
ketone or capillary blood ketone assessment compared with markers of severity, suggests that capillary blood ketone

ª 2014 The Authors.

18 Diabetic Medicine ª 2014 Diabetes UK
Review article DIABETICMedicine

should not be used to guide the level of care, particularly ketoacidosis, particularly if a decrement of 0.5 mmol/l per
when the precision of capillary blood ketone measurement is h is advised, as the coefficient of variation of the meters may
lost above 3 mmol/l, making the value of 6 mmol/l difficult be beyond this. Unless capillary blood ketone measurements
to assess in the absence of laboratory confirmation. Other are routinely confirmed with laboratory serum analysis, the
clinical variables outlined in Joint British Diabetes Societies limitations in this context negate its regular use, for
guidance [10], such as urine output, haemodynamic stability, assessment of severity [41].
Glasgow Coma score and airway protection may be more
important determinants of level of care.
Prevention of ketoacidosis

While the utility of capillary blood ketone in the acute setting

Monitoring for resolution of ketoacidosis
may be limited by its poor precision at high values and in the
It is unclear which variables should be monitored to assess dehydrated patient, there is a promising role in the outpatient
resolution of ketoacidosis. The Joint British Diabetes Soci- setting to prevent ketoacidosis. Self-monitoring of ketones at
eties guidance states that capillary blood ketones should be home aims to increase awareness of possible impending
measured serially, while traditionally, pH, bicarbonate, urine ketoacidosis, offering patients the opportunity to increase
ketones and glycaemia have all been used [10]. The latter is insulin and reduce ketogenesis during hyperglycaemia [42].
now considered less of an indicator because of the recogni- A study of young adults and children as part of ‘sick-day’
tion of euglycaemic ketoacidosis and because suppression of management at home [43] randomized participants to either
lipolysis may require more insulin than is needed to restore capillary blood ketone or urine ketone assessment in
euglycaemia. The limitations of urine ketones have already conjunction with blood glucose monitoring, after intensive
been discussed and centre on the paradoxical increase in education [43]. After 6 months of follow-up, the 123
urine ketones (AcAc) during ketoacidosis resolution. Several participants were more likely to test capillary blood ketones
studies have considered the use of capillary blood ketone to than urine ketones (90.8 vs 61.3%) and the capillary blood
track ketoacidosis resolution and determine when i.v. insulin ketone group were significantly less likely to be admitted for
can be switched to s.c. insulin or when patients can be hospitalization (38 per 100 patient-years vs 75 per 100
discharged [17,39]. patient-years). The outcome in this study was hospital
A small study measured serum BOHB (4-hourly), capillary admission for any cause and not diagnosis of ketoacidosis,
blood ketone (hourly) and urine ketones in the resolution of so it cannot be concluded that capillary blood ketone
ketoacidosis, and compared their current endpoint (pH > 7.3 assessment prevented ketoacidosis. Of the 33 hospital
and urine ketones negative) to a proposed new endpoint (pH admissions recorded, the vast majority were related to
> 7.3 and two successive capillary blood ketone < 1 mmol/l), hyperglycaemia or ‘metabolic derangement’ but some were
in a paediatric population with ketoacidosis [8]. Out of 11 also attributable to hypoglycaemia. While the majority of
samples that had laboratory BOHB values > 6 mmol/l, only participants preferred capillary blood ketone testing to urine
three were reported as ‘Hi’ on the capillary blood ketone testing, further work is needed to determine if the cost
meter, with the remaining reported between 1 and 7 mmol/l incurred from providing routine outpatient capillary blood
lower than the actual value. The study found a median lag ketone testing offsets the financial burden of hospital
time of 11 h between their current endpoint for cessation of admission. However, although patient acceptability is an
i.v. insulin and the proposed capillary blood ketone end- important factor, if patients complied equally with urine and
point, with the latter allowing earlier discharge from a high capillary blood ketone testing, the difference in hospitaliza-
dependency unit, suggesting a role for capillary blood ketone tion may diminish.
measurement in a paediatric population; however, earlier Another study analysed the utility of capillary blood
discharge was dependent on hourly sampling for ketones, a ketone assessment in an outpatient setting in 15 patients and
resource-intensive exercise. found that, in the absence of intercurrent illness, capillary
In another paediatric study comparing serial measure- blood ketone measurements did not exceed 1 mmol/l in spite
ments of capillary blood ketone with serum BOHB and of glucose levels up to 33.1 mmol/l [15]. In individuals with
other biochemical variables such as pH and bicarbonate, a intercurrent illness some individuals noted levels of capillary
good correlation was shown between capillary blood blood ketone > 1 mmol/l; however, self-monitoring
ketone and resolution of ketoacidosis [40]. Uncoupling of prompted increases in insulin dosage and resolution of
this relationship was observed at higher BOHB concentra- ketonaemia. A similar paediatric study examined capillary
tions, which may also account for the discrepancy in levels blood ketone levels in children with Type 1 diabetes, and
of ketones, pH and bicarbonate often encountered at suggested that levels above 0.4 mmol/l should be considered
diagnosis. abnormal [31]. These studies suggest a preventative role for
Since capillary blood ketones cannot reliably be measured capillary blood ketone measurement, and larger randomized
above 6 mmol/l and may be unreliable above 3 mmol/l, it is studies are required to confirm these findings. Other studies
difficult to recommend their use in the monitoring of have examined the role of capillary blood ketone

ª 2014 The Authors.

Diabetic Medicine ª 2014 Diabetes UK 19
DIABETICMedicine Ketone measurement in diabetic ketoacidosis  S. Misra and N. S. Oliver

measurement in detecting interruptions of continuous s.c. ketone-testing strips to people with Type 1 diabetes against
insulin infusion in patients on insulin pump therapy [44]. the cost of urine ketone strips, which are almost 10 times
Sick-day diabetes management requires serial measure- cheaper and may be used less frequently. The potential
ment of blood glucose, and ketone measurement provides a savings from preventing admission may offset this cost but
useful adjunct to assessment of glucose in the hyperglycaemic we do not yet have these data.
state but is in no way a substitute. A measure of ketosis in Large, well-designed studies are needed to assess conclu-
hyperglycaemia can help to differentiate hyperglycaemia sively the utility of capillary blood ketone measurement and
secondary to a carbohydrate and insulin mismatch from to determine the cost-effectiveness of such an approach. In an
evolving insulin deficiency. Whilst trends of capillary blood inpatient setting, studies to date suggest that capillary blood
ketone may be useful, a single reading which may be falsely ketone measurement may facilitate early discharge from
high or low could provide false reassurance or cause anxiety higher dependency environments in paediatric patients, and
and it is unclear how often capillary blood ketones should be may be a cost-effective tool; however, the provision of
measured. Ketonuria assessment has limitations but does at laboratory-based serum ketone measurement into these
least confirm ketosis over a time period of hours. cost-analyses must also be included.
It is critical that intensive education accompanies self-mon-
itoring of capillary blood ketones and the added value of
home capillary blood ketone monitoring, in addition to Discussion
intensive and structured diabetes education, should be As ketones are the primary derangement in ketoacidosis,
considered. In the absence of continuing clinical support their measurement is a logical step to take in unwell people
there is a risk that home capillary blood ketone testing may with diabetes, but clinical practice has jumped an important
paradoxically increase hospital attendance in patients who step by attempting to validate the test in a point-of-care
would otherwise self-manage hyperglycaemia. Ketonuria capacity before fully appreciating its value with precise
testing at least, on sick-days with input from specialist laboratory measurement. Multiple clinical applications are
nurses, was shown to potentially reduce admissions, but increasingly reported, with no consensus on thresholds or
capillary blood ketone may also be useful in this context frequency of measurement.
[45]. Results are awaited on a large study comparing urine Capillary blood ketone measurement represents an ana-
and blood ketone assessment in young people with Type 1 lytical and technological advance on traditional qualitative
diabetes, although preliminary results suggest the latter may or semi-quantitative urine ketone measurement, with many
improve glycaemic control in insulin pump treatment [46]. potential applications. The evidence base evaluating poin-
t-of-care ketone measurements in ketoacidosis is still in its
infancy, with heterogeneous studies published, a paucity of
Cost-effectiveness of capillary blood ketone measurement
high-quality randomized diagnostic accuracy studies and
The availability of point-of-care technology represents an limited data on whether changes in diagnostic accuracy
exciting technological development, and is often proposed to impact on patient outcome, a common issue when new tests
be cheaper in the long term than laboratory-based alterna- are introduced [48]. The studies reviewed have been con-
tives; however, this argument may not hold true for ketone ducted over the last 20 years, with some evaluating early
assessment as routine laboratory-based analysis is seldom versions of ketone meters that have now been superseded by
undertaken. Evidence suggests potential cost-saving in its use next-generation devices that may or may not be more
vs urine ketone testing in a paediatric intensive care setting. accurate. This aspect of the evidence base warrants further
Thirty three children with severe ketoacidosis were random- consideration. It is worth questioning why so few studies
ized to be monitored either with hourly capillary blood have robustly assessed utility. This may reflect the regulatory
ketone or urine ketone assessment. The capillary blood frameworks for devices, which do not require efficacy data
ketone group cleared ketosis between 4 and 9.5 h earlier for marketing (in contrast to that for medicines), and only
than the ketonuria group, meaning an earlier discharge from assess safety and minimum performance standards; however,
the intensive care unit and a saving of nearly 3000 Euros diagnostic tests in general are commonly inadequately
[47]. Nearly half of the children in the urine monitoring assessed and proposals to ensure high-quality evaluation
group, however, did not produce any urine because of exist, but are not routinely followed [49].
dehydration. In addition, in that study, a good correlation Despite the technological advances, deciding if capillary
was found between resolving ketonaemia, pH and glucose blood ketone is a clinically useful improvement needs further
levels. One could therefore argue that resolution of acidosis study and, until studies that formally assess clinical utility are
and hyperglycaemia could be justifiable endpoints without undertaken with associated laboratory measurements that
incurring the excess cost of ketone measurement. can provide robust transferable reference ranges, there will
Calculations of ketone measurement cost-effectiveness remain confusion over its use, diminishing its value and
need to take into account the potential cost of providing implementation.

ª 2014 The Authors.

20 Diabetic Medicine ª 2014 Diabetes UK
Review article DIABETICMedicine

For the diagnosis of ketoacidosis, studies are conflicting, ing results, means that there is a lack of clarity about how
with some showing good correlation of capillary blood the technology should be applied and what the result means
ketones with other biochemical variables such as bicarbonate or offers above more routinely available methods. The
and pH levels, whilst others are more limited. Since caution routine use of capillary blood ketones in the absence of a
should be exercised when measuring capillary blood ketones concomitant laboratory service is not ideal and there is a
in dehydrated patients, who are invariably volume-depleted danger that a simple point-of-care urine ketone test with
and have haemoconcentration, the validity of using capillary well-known limitations is being replaced with another test
blood ketones alone as the measure of ketosis to diagnose in which limitations are not widely appreciated. Exact
ketoacidosis is questionable. Until there are consensus guidance on clinical usage is lacking consensus and a firm
criteria on the definition of ketoacidosis and an appropriate evidence base, and, until these are available, capillary blood
and evidence-based capillary blood ketone threshold, it will ketone measurement should be used and interpreted with
be difficult to design and carry out appropriately powered caution.
prospective studies.
One of the biggest limitations hindering routine use of Funding sources
capillary blood ketones lies in the poor precision above
3 mmol/l. Ketone levels above this threshold may falsely None.
underestimate ketonaemia and levels starting above 3 mmol/l
at diagnosis, cannot be reliably tracked for improvements in Competing interests
ketonaemia as advised in Joint British Diabetes Societies
None declared.
guidance. This single issue impedes routine use of capillary
blood ketone in ketoacidosis in the absence of labora-
tory-based corroborative analysis. Ketone assessment is, References
however, one variable used to diagnose ketoacidosis, and it 1 Basu A, Close CF, Jenkins D, Krentz AJ, Nattrass M, Wright AD.
could be argued that as long as the above limitations are Persisting mortality in diabetic ketoacidosis. Diabet Med 1993; 10:
considered there is still a place for capillary blood ketone 782.
assessment. Uptake of Joint British Diabetes Societies guid- 2 Laffel L. Ketone bodies: a review of physiology, pathophysiology
and application of monitoring to diabetes. Diabetes Metab Res Rev
ance in the UK has been widespread and it is therefore
1999; 15: 412–426.
important that all clinical users are aware of the pros and cons 3 English P, Williams G. Hyperglycaemic crises and lactic acidosis in
of the test [50]. diabetes mellitus. Postgrad Med J 2004; 80: 253–261.
In the same way that glucometers are not routinely used to 4 Sacks DB, Arnold M, Bakris GL, Bruns DE, Horvath AR, Kirkman
diagnose diabetes, but are corroborated with plasma glucose MS et al. Guidelines and recommendations for laboratory analysis
in the diagnosis and management of diabetes mellitus. Diabetes
measurement, capillary blood ketone measurement without
Care 2011; 34: e61–99.
concomitant serum assessment cannot be advocated. 5 Taboulet P, Haas L, Porcher R, Manamani J, Fontaine J-P, Feugeas
Although many studies show a reasonable correlation J-P et al. Urinary acetoacetate or capillary beta-hydroxybutyrate for
between capillary blood ketones and serum BOHB, this the diagnosis of ketoacidosis in the Emergency Department setting.
relationship is uncoupled at values above 3 mmol/l. These Eur J Emerg Med 2004; 11: 251–258.
6 Voulgari C, Tentolouris N. The Performance of a Glucose-Ketone
limitations need to be understood by clinical staff undertak-
Meter in the Diagnosis of Diabetic Ketoacidosis in Patients with
ing the test and all staff require a suitable level of training to Type 2 Diabetes in the Emergency Room. Diabetes Technol Ther
carry out these tests. 2010; 12: 529–535.
Given the lower concentrations of BOHB involved and the 7 Charles RA, Bee YM, Eng PHK, Goh SY. Point-of-care blood
reduced need for rigid thresholds, the most promising utility ketone testing: screening for diabetic ketoacidosis at the emergency
department. Singapore Med J 2007; 48: 986–989.
for capillary blood ketone assessment lies in self-management
8 Noyes KJ, Crofton P, Bath LE, Holmes A, Stark L, Oxley CD et al.
and ketoacidosis prevention, just as capillary blood glucose Hydroxybutyrate near-patient testing to evaluate a new end-point
measurement is most potent for self-management, and this for intravenous insulin therapy in the treatment of diabetic ketoac-
warrants further investigation. Intensive diabetes education idosis in children. Pediatr Diabetes 2007; 8: 150–156.
to empower self-management, is often the most useful tool in 9 Sheikh-Ali M, Karon BS, Basu A, Kudva YC, Muller LA, Xu J et al.
Can serum beta-hydroxybutyrate be used to diagnose diabetic
preventing ketoacidosis and it will be difficult to delineate the
ketoacidosis? Diabetes Care 2008; 31: 643–647.
effects of this from capillary blood ketone testing. 10 Savage MW, Dhatariya KK, Kilvert A, Rayman G, Rees JA,
Courtney CH et al. Joint British Diabetes Societies guideline for the
management of diabetic ketoacidosis. Diabet Med 2011; 28: 508–
Conclusion 515.
11 Umpierrez GE, Watts N, Phillips L. Clinical Utility of b3-Hydrox-
There is undoubtedly a role for the use of capillary blood
ybutyrate Determined by Reflectance Meter in the Management of
ketone assessment in the diagnosis, management and pre- Diabetic Ketoacidosis. Diabetes Care 1995; 18: 137–138.
vention of ketoacidosis, but the heterogeneous studies, often 12 Guerci B, Benichou M, Floriot M, Bohme P, Fougnot S, Franck P
non-randomized, incorporating small numbers with conflict- et al. Accuracy of an electrochemical sensor for measuring

ª 2014 The Authors.

Diabetic Medicine ª 2014 Diabetes UK 21
DIABETICMedicine Ketone measurement in diabetic ketoacidosis  S. Misra and N. S. Oliver

capillary blood ketones by fingerstick samples during metabolic gency-department triage: {beta}-hydroxybutyrate versus the urine
deterioration after continuous subcutaneous insulin infusion dipstick. Diabetes Care 2011; 34: 852–854.
interruption in type 1 diabetic patients. Diabetes Care 2003; 31 Samuelsson U, Ludvigsson J. When should determination of
26: 1137–1141. ketonemia be recommended? Diabetes Technol Ther 2002; 4:
13 Chiu RWK, Ho CS, Tong SF, Ng KF, Lam CW. Evaluation of a 645–650.
new handheld biosensor for point-of-care testing of whole blood 32 Harris S, Ng R, Syed H, Hillson R. Near patient blood ketone
beta-hydroxybutyrate concentration. Hong Kong Med J 2002; 8: measurements and their utility in predicting diabetic ketoacidosis.
172–176. Diabet Med 2005; 22: 221–224.
14 Byrne HA, Tieszen KL, Hollis S, Dornan TL, New JP. Evaluation of 33 Taboulet P, Deconinck N, Thurel A, Haas L, Manamani J, Porcher
an electrochemical sensor for measuring blood ketones. Diabetes R et al. Correlation between urine ketones (acetoacetate) and cap-
Care 2000; 23: 500–503. illary blood ketones (3-beta-hydroxybutyrate) in hyperglycaemic
15 Wallace TM, Meston NM, Gardner SG, Matthews DR. The hospital patients. Diabetes Metab 2007; 33: 135–139.
and home use of a 30-second hand-held blood ketone meter: 34 Mackay L, Lyall M, Delaney S, McKnight J, Strachan M. Are
guidelines for clinical practice. Diabet Med 2001; 18: 640–645. blood ketones a better predictor than urine ketones of acid base
16 Yu H-YE, Agus M, Kellogg MD. Clinical utility of Abbott Precision balance in diabetic ketoacidosis? Pract Diabetes Int 2010; 27:
Xceed Pro ketone meter in diabetic patients. Pediatr Diabetes 396–399.
2011; 12: 649–655. 35 Mesa J, Salcedo D, Calle Hde L, Delgado E, N ovoa J, Hawkins F et
17 Ham MR, Okada P, White PC. Bedside ketone determination in al. Detection of ketonemia and its relationship with hyperglycemia
diabetic children with hyperglycemia and ketosis in the acute care in type 1 diabetic patients. Diabetes Res Clin Pract 2006; 72: 292–
setting. Pediatr Diabetes 2004; 5: 39–43. 297.
18 Khan ASA, Talbot JA, Tieszen KL, Gardener EA, Gibson JM, New 36 Fulop M, Murthy V, Michilli A, Nalamati J, Qian Q, Saitowitz
JP. Evaluation of a bedside blood ketone sensor: the effects of A. Serum beta-hydroxybutyrate measurement in patients with
acidosis, hyperglycaemia and acetoacetate on sensor performance. uncontrolled diabetes mellitus. Arch Intern Med 1999; 159:
Diabet Med 2004; 21: 782–785. 381–384.
19 Yu H-YE, Agus M, Kellogg MD. Clinical utility of Abbott Precision 37 Prisco F, Picardi A, Iafusco D, Lorini R, Minicucci L, Martinucci
Xceed Pro ketone meter in diabetic patients. Pediatr Diabetes ME et al. Blood ketone bodies in patients with recent-onset type
2011; 12: 649–655. 1 diabetes (a multicenter study). Pediatr Diabetes 2006; 7:
20 Janssen MJW, Hendrickx BHE. Habets-van Der Poel CD, Van Den 223–228.
Bergh JPW, Haagen AAM, Bakker JA. Accuracy of the Precision 38 Arora S, Probst MA, Agy C, Menchine M. Point-of-care beta-hy-
point-of-care ketone test examined by liquid chromatography droxybutyrate testing for assessing diabetic ketoacidosis severity
tandem-mass spectrometry (LC-MS/MS) in the same fingerstick prior to treatment in the emergency department. Diabetes Res Clin
sample. Clin Chem Lab Med 2010; 48: 1781–1784. Pract 2011; 94: e86–88.
21 Klocke A, Phelan H, Twigg S, Craig M. Systematic Review: Blood 39 Wiggam M, O’Kane M, Harper R, Atkinson A, Hadden D, Trimble
beta-hydroxybutyrate vs. urine acetoacetate testing for the preven- E et al. Treatment of diabetic ketoacidosis using endpoint of
tion and management of ketoacidosis in Type 1 diabetes. Diabet emergency management. A randomized controlled study. Diabetes
Med 2013; 30: 818–824. Care 1997; 20: 1347–1352.
22 National External Quality Assurance Scheme, Clinical Chemistry. 40 Rewers A, McFann K, Chase HP. Bedside monitoring of blood
Available at: http://www.ukneqas.org.uk/content/PageServer.asp? beta-hydroxybutyrate levels in the management of diabetic ketoac-
S=606515537&C=1252&Type=N&AID=16&SID=99 website. idosis in children. Diabetes Technol Ther 2006; 8: 671–676.
Last accessed 10 July 2014. 41 Loh TP, Saw S, Sethi SK. Bedside monitoring of blood ketone for
23 Welsh External Quality Assurance Scheme for point-of-care testing. management of diabetic ketoacidosis: proceed with care. Diabet
Available at: http://www.weqas.com/poct/index.html. Last Med 2012; 29: 827–828.
accessed 10 July 2014. 42 Wallace TM, Matthews DR. Recent advances in the monitoring
24 Kitabchi AE, Umpierrez GE, Miles JM, Fisher JN. Hyperglycemic and management of diabetic ketoacidosis. QJM 2004; 97: 773–
crises in adult patients with diabetes. Diabetes Care 2009; 32: 780.
1335–1343. 43 Laffel LMB, Wentzell K, Loughlin C, Tovar A, Moltz K, Brink S.
25 Wolfsdorf J, Craig ME, Daneman D, Dunger D, Edge J, Lee W Sick day management using blood 3-hydroxybutyrate (3-OHB)
et al. Diabetic ketoacidosis in children and adolescents with compared with urine ketone monitoring reduces hospital visits in
diabetes. Pediatr Diabetes 2009; 10(Suppl. 1): 118–133. young people with T1DM: a randomized clinical trial. Diabet Med
26 Turan S, Omar A, Bereket A. Comparison of capillary blood ketone 2006; 23: 278–284.
measurement by electrochemical method and urinary ketone in 44 Orsini-Federici M, Akwi JA, Canonico V, Celleno R, Ferolla P,
treatment of diabetic ketosis and ketoacidosis in children. Acta Pippi R et al. Early detection of insulin deprivation in continuous
Diabetol 2008; 45: 83–85. subcutaneous insulin infusion-treated patients with type 1 diabetes.
27 Arora S, Henderson SO, Long T, Menchine M. Diagnostic accuracy Diabetes Technol Ther 2006; 8: 67–75.
of point-of-care testing for diabetic ketoacidosis at emer- 45 Evans NR, Richardson LS, Dhatariya KK, Sampson MJ. Diabetes
gency-department triage: beta-hydroxybutyrate versus the urine specialist nurse telemedicine: admissions avoidance, costs and
dipstick. Diabetes Care 2011; 34: 852–854. casemix. Pract Diabetes 2012; 29: 25–29.
28 Naunheim R, Jang TJ, Banet G, Richmond A, McGill J. 46 Bismuth E, Svoren B, Volkening L, Butler D, Laffel L. Impact of
Point-of-care Test Identifies Diabetic Ketoacidosis at Triage. Acad insulin pump therapy and blood BOHB ketone monitoring on
Emerg Med 2006; 13: 683–685. glycemic outcomes in youth with type 1 diabetes. Diabetes 2007;
29 Bektas F, Eray O, Sari R, Akbas H. Point of care blood ketone 56: pA485.
testing of diabetic patients in the emergency department. Endocr 47 Vanelli M, Chiari G, Ghizzoni L, Costi G, Giacalone T, Chiarelli F.
Res 2004; 30: 395–402. Effectiveness of a prevention program for diabetic ketoacidosis in
30 Arora S, Henderson SO, Long T, Menchine M. Diagnostic accuracy children. An 8-year study in schools and private practices. Diabetes
of point-of-care testing for diabetic ketoacidosis at emer- Care 1999; 22: 7–9.

ª 2014 The Authors.

22 Diabetic Medicine ª 2014 Diabetes UK
Review article DIABETICMedicine

48 Misra S, Barth JH. Clinica Chimica Acta How good is the evidence 50 Sampson M. Adoption and outcomes of Joint British Diabetes
base for test selection in clinical guidelines ? Clin Chim Acta 2014; guidelines for inpatient care. Present Diabetes UK 2013. Available
432: 27–32. at: http://www.diabetes.org.uk/upload/DPC_Speakers2013/Thur/
49 Lijmer JG, Leeflang M, Bossuyt PMM. Proposals for a phased 11.10_Thu_Sampson_Hall.pdf. Last accessed 10 July 2014.
evaluation of medical tests. Med Decis Making 2009; 29: E13–E21.

ª 2014 The Authors.

Diabetic Medicine ª 2014 Diabetes UK 23
Copyright of Diabetic Medicine is the property of Wiley-Blackwell and its content may not be
copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for
individual use.