Emerging Treatments and Technologies ORIGINAL ARTICLE

Accuracy of an Electrochemical Sensor for Measuring Capillary Blood Ketones by Fingerstick Samples During Metabolic Deterioration After Continuous Subcutaneous Infusion Interruption in Type 1 Diabetic Patients

1 1 BRUNO GUERCI, MD, PHD SEBASTIEN FOUGNOT, MD he Control and Complica- 1 2 MURIEL BENICHOU, MD PATRICIA FRANCK, MD tions Trial (DCCT) has demon- 1 1 MICHELE` FLORIOT, MD PIERRE DROUIN, MD 1 strated that intensive diabetes HILIP OHME MD T P B , management, multiple daily injections (MDI) in two-thirds of cases, and contin- uous subcutaneous insulin infusion (CSII) in one-third of cases could reduce the risk of development and progression OBJECTIVE — This study was designed to test the accuracy of capillary ketonemia for diag- of long-term complications of type 1 dia- nosis of after interruption of insulin infusion. betes (1). CSII provides the most physio- RESEARCH DESIGN AND METHODS — A total of 18 patients with logical pattern of insulin delivery cur- treated by external pump were studied during pump stop for 5 h. Plasma and capillary ketone- rently available (2). However, it is now mia and ketonuria were determined every hour from 7:00 A.M. (time 0 min ϭ T0) to 12:00 P.M. clearly established that CSII is associated (time 300 min ϭ T300). Plasma ␤-hydroxybutyrate (␤-OHB) levels were measured by an with a substantial increase in risk of dia- enzymatic end point spectrophotometric method, and capillary ␤-OHB levels were measured by betic ketoacidosis (DKA) compared with an electrochemical method (MediSense Optium meter). Ketonuria was measured by a semiquan- MDI (3,4). In a meta-analysis of 14 stud- titative test (Ketodiastix). Positive ketosis was defined by a value of Ն0.5 mmol/l for ketonemia Ն ies, odds ratios (95% CI) were 7.20 and 4 mmol/l (moderate) for ketonuria. (2.95–17.58) for exclusive use of insulin RESULTS — After stopping the pump, concentrations of ␤-OHB in both plasma and capillary pumps compared with 1.13 (0.15–8.35) blood increased significantly at time 60 min (T60) compared with T0 (P Ͻ 0.001), reaching for MDI (4). Technical problems with maximum levels at T300 (1.30 Ϯ 0.49 and 1.23 Ϯ 0.78 mmol/l, respectively). Plasma and CSII (pump failure, catheter occlusion, capillary ␤-OHB values were highly correlated (r ϭ 0.94, P Ͻ 0.0001). For diagnosis of ketosis, skin infection) may reduce insulin deliv- capillary ketonemia has a higher sensitivity and negative predictive value (80.4 and 82.5%, ery or insulin absorption and cause DKA respectively) than ketonuria (63 and 71.8%, respectively). For plasma levels Ն250 (5). Currently, the use of short-acting in- mg/dl, plasma and capillary ketonemia were found to be more frequently positive (85 and 78%, sulin analogs in external pumps is becom- respectively) than ketonuria (59%) (P ϭ 0.017). The time delay to diagnosis of ketosis was Ϯ Ϯ ϭ ing more widespread because some significantly higher for ketonuria than for plasma ketonemia (212 67 vs. 140 54 min, P studies have shown that the use of such 0.0023), whereas no difference was noted between plasma and capillary ketonemia. analogs provides better glycemic control CONCLUSIONS — The frequency of screening for ketosis and the efficiency of detection of and stability than regular insulin (6–8). ketosis definitely may be improved by the use of capillary blood ketone determination in clinical Considering the pharmacokinetic charac- practice. teristics of these analogs, a short interrup- tion of their infusion in CSII is associated Diabetes Care 26:1137–1141, 2003 with an earlier and greater metabolic de- terioration, probably due to a limited in- sulin depot in subcutaneous adipose ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● tissue (9). In a previous unpaired study, From the 1Service de Diabe´tologie, Maladies Me´taboliques & Maladies de la Nutrition, CIC-INSERM, this was not the case after nighttime insu- 2 Hoˆpital Jeanne d’Arc, Nancy, France; and the Laboratoire de Biochimie, Hoˆpital Central, Nancy, France. lin interruption (10). Address correspondence and reprint requests to Dr. Bruno Guerci, Service de Diabe´tologie, Maladies Me´taboliques & Maladies de la Nutrition, CIC-INSERM/CHU Nancy, Hoˆpital Jeanne d’Arc, Centre Hospi- Consequently, reliable self-monitor- talo-Universitaire de Nancy, B.P. 303, 54201 Toul Cedex, France. E-mail: [email protected]. ing of ketone body levels has become a Received for publication 31 July 2002 and accepted in revised form 23 December 2002. priority in detecting insulin infusion in- P.D. is deceased. terruption. Urinary ketone levels in type 1 Abbreviations: ␤-OHB, ␤-hydroxybutyrate; CSII, continuous subcutaneous insulin infusion; DKA, di- abetic ketoacidosis; MDI, multiple daily injections. diabetic patients are classically used to A table elsewhere in this issue shows conventional and Syste`me International (SI) units and conversion screen for impending DKA, because de- factors for many substances. lays in the diagnosis and treatment of

DIABETES CARE, VOLUME 26, NUMBER 4, APRIL 2003 1137 Capillary during CSII interruption

DKA are associated with an increase in reasons of insufficient time and/or de- (BioRad, Richmond, CA) with a normal morbidity and mortality (11,12). How- mands of the experimental protocol. In- range of 4–6%. ever, commercial tests for detecting uri- sulin was infused into the abdomen, for Blood samples for determination of nary ketones are associated with well- which the infusion site was changed every ketone bodies were collected and placed known difficulties in their role as diag- 3 days. None of the patients had experi- on crushed ice. Plasma was immediately nostic and management tools for DKA enced any recent episode of DKA or re- obtained by centrifugation at 4°C. The (13). Recently, inexpensive quantitative quired a daily insulin dose Ͼ1.5 units concentration of 3-hydroxybutyrate was Ϫ Ϫ tests of ␤-hydroxybutyrate (␤-OHB) lev- kg 1 day 1. Patients with hyperlipid- determined by an enzymatic end point els have become available for use with emia, thyroid or disease, diabetic or spectrophotometric method using 3-hy- small blood samples, offering new options nondiabetic renal disease, urinary tract droxybutyrate dehydrogenase (3-HBDH), for monitoring and treating diabetes. Previ- infection, pregnancy, or acute or chronic normal range of 0.06– 0.17 mmol/l ous studies (14,15) have evaluated the ac- inflammatory syndrome were excluded (KONE Delta Automatic Analyzer). This curacy of this new electrochemical sensor from the study. None of the patients had method was used as the reference method. for measuring capillary blood ␤-OHB in or , macro- The intra-assay coefficient of variation comparison to reference plasma ketonemia vascular complications, retinopathy, or was 4.9% (17). determination. hypertension. Finally, patients were in- Capillary plasma glucose was mea- The aim of the present study was to structed to follow a weight-maintenance sured using a Medisense Optium meter determine whether use of capillary blood diet (15% of calories as protein, 35% as (MediSense/Abbott Laboratories, Abing- ketonemia was superior to that of ketonu- , and 50% as ) taken as ton, U.K.), which is a combined glucose ria for detection of ketosis in terms of de- three main meals and one to three snacks and ketone sensor that produces an elec- layed diagnosis in patients treated by per day. trical current proportional to blood pump after deliberate interruption of ␤-OHB concentration. The Optium strips CSII. Study protocol were plasma calibrated on a YSI reference Patients arrived at the Clinical Research glucose analyzer (Yellow Springs Instru- RESEARCH DESIGN AND Center (CIC/INSERM-CHU de Nancy) ments, Yellow Springs, OH). METHODS the day before the interruption of CSII, at For determination of capillary blood 8:00 P.M., for a calibrated meal. The de- ketone levels, an electrochemical strip Patients sign of the protocol was similar to that was inserted into the sensor to which 5 ␮l The study comprised 18 patients with previously described (9). No hypoglyce- of capillary blood was applied. The type 1 diabetes, diagnosed according to mia episode was tolerated during the 24 h ␤-OHB, in the presence of hydroxybu- American Diabetes Association criteria before interruption of CSII. Patients tyrate deshydrogenase, was oxidized to (16), who were C-peptide negative (Ͻ0.3 fasted for 11 h, until 7:00 A.M. (no break- acetoacetate with the concomitant reduc- nmol/l, 6 min after intravenous adminis- fast), at which time the pump was tion of NADϩ to NADH. The NADH was tration of 1 mg glucagon). The patients stopped and the catheter was discon- reoxidized to NADϩ by a redox mediator, were treated with infusion of insulin via nected. Every hour from 7:00 A.M. (time 0 such that the current generated was di- an external pump (MiniMed Infusor 507, min ϭ T0) to 12:00 P.M. (time 300 min ϭ rectly proportional to the 3-hydroxybu- 507C, or 508; MiniMed Technologies, T300), plasma glucose levels and ketone- tyrate concentration. After 30 s, the Northridge, CA) and a deconnectable mia as well as capillary blood glucose lev- 3-hydroxybutyrate concentration in a catheter, either the Tender catheter (Dis- els and ketonemia were measured, along sample was displayed. This system is ac- etronic Medical Systems, Burgdorf, Swit- with ketonuria. The pump of each patient curate for 3-hydroxybutyrate levels from zerland) or the Sofset QR (MiniMed). The was then reactivated at 12:00 P.M. (T300) 0 to 6 mmol/l. On three different levels of type of insulin used in the pumps was at the patient’s usual basal rate, at which ␤-OHB (low at the mean of 0.43, moder- velosulin (Velosulin HM, U-100; Novo- time the patients ate a calibrated lunch ate at 1.08, and high at 3.55 mmol/l), the Nordisk, Boulogne-Billancourt, France) and activated their usual prelunch insulin intra-assay coefficients of variation (cal- in eight patients and lispro (Humalog boluses. Additional insulin boluses were culated on three determinations for each U-100; Lilly France, Saint-Cloud, France) also given each hour, according to blood sample) were 10.5, 5.5, and 3.2%, in the other 10 patients. No randomiza- glucose levels, until the disappearance of respectively. tion was performed, and the patients in ketosis. ketone bodies were measured the study were selected from 245 type 1 Informed written consent was ob- by a semiquantitative test (Ketodiastix; diabetic patients treated by CSII at our tained from all patients. The study proto- Bayer Diagnostics, Stoke Poges, Slough, outpatient clinic. All these patients had col was approved by the Ethics Com- U.K.; trace to Ն16 mmol/l [large]). Posi- received CSII for Ͼ1 year, had hypogly- mittee of The Center Hospitalier Univer- tive ketosis was defined as values Ն0.5 cemia awareness, and had good compli- sitaire de Nancy (France). mmol/l for capillary blood ketonemia and ance with self-monitoring of blood Ն4 mmol/l (moderate) for ketonuria glucose (more than four times per day). Biochemical determinations (18). The patients were regularly followed (ev- Plasma glucose was measured by the glu- ery 3 months) by the same investigator to cose oxidase method (Beckman Glucose Statistical analyses Ϯ evaluate metabolic control and to make Analyzer, Beckman, Fullerton). HbA1c Results are expressed as means SD. The insulin dose adjustments. Some patients was measured by high-performance liq- distribution of variables was tested for ap- declined to participate in the study for uid chromatography on Biorex resins proximation to a Gaussian distribution

1138 DIABETES CARE, VOLUME 26, NUMBER 4, APRIL 2003 Guerci and Associates

(normality) using the Kurtosis and Skew- Table 1—Clinical and metabolic character- cantly lower than levels measured in the ness test. Data were compared using the istics of patients in study plasma from T0 to T240 (Fig. 1). Never- nonparametric Wilcoxon’s test for two theless, the capillary blood ␤-OHB levels Sex ratio (men/women) 9/9 paired groups (plasma versus capillary were highly and significantly correlated Ϯ determinations) and versus the value re- Age (years) 41.1 7.1 with plasma ␤-OHB levels (r ϭ 0.94, P Ͻ 2 Ϯ corded at 7:00 A.M. (T0). BMI (kg/m ) 24.8 2.7 0.0001).When we subdivided the pa- Ϯ Univariate linear regression was used Duration of diabetes (years) 20.8 9.6 tients according to the type of insulin Ϯ to calculate the correlation coefficients Duration of CSII (months) 66.4 44.2 used (velosulin or lispro), the kinetics of Ϯ and equations for the regression slopes. Systolic blood pressure 132 9 ketone body levels were very similar for The frequency of positive ketosis was (mmHg) capillary and plasma ␤-OHB levels (data 2 Ϯ checked by ␹ test. Diastolic blood pressure 74 5 not shown). For determination of the accuracy of (mmHg) ␮ Ϯ At T300, 16 of the 18 patients (89%) the electroenzymatic method, we ana- Serum ( mol/l) 105.5 8.8 had a positive reading for plasma ␤-OHB lyzed the results by calculating the speci- Anti-insulin antibodies (%) 32.6 Ϯ 19 Ն Ϯ levels ( 0.5 mmol/l), whereas 15 patients ficity and sensitivity of the capillary and Total cholesterol (mg/dl) 201 34 (83%) had a positive reading for capillary urinary methods to correctly detect ke- Triglyceride (mg/dl) 78 Ϯ 41 ␤ Ն Ϯ -OHB levels ( 0.5 mmol/l). Con- tone bodies by comparison with the LDL cholesterol (mg/dl) 126 27 versely, ketonuria was negative in all pa- ␤ HDL cholesterol (mg/dl) 61 Ϯ 12 plasma 3- hydroxybutyrate concentra- tients at T0 and positive (Ն4 mmol/l HbA (%, normal range 7.34 Ϯ 0.99 tion determined by the spectrophotomet- 1c [moderate]) at T300 only in 13 patients ric method for all time points from T0 to 4–6%) (72%). Using the plasma ␤-OHB value as Ϯ T300. With regard to the positive ketosis Data are means SD. a reference, the sensitivity and negative criteria defined above, we assigned for predictive value of capillary blood ke- each of the sensor and urinary values the T300 (316 Ϯ 79 mg/dl, P ϭ 0.0003). The tonemia were higher than those of keto- term of true positive (TP), false positive capillary plasma glucose levels were sig- nuria (sensitivity 80.4 vs. 63%; negative (FP), true negative (TN), or false negative nificantly correlated with venous plasma predictive value 82.5 vs. 71.8, respec- (FN), according to the plasma ketone glucose levels (r ϭ 0.98, P Ͻ 0.0001). tively) (Table 2). When we analyzed the body value simultaneously measured data for capillary glucose values Ͼ2.50 g/l higher or lower than 0.5 mmol/l. Sensitiv- Ketone body levels only (the limit at which the detection of ity and specificity were defined by the fol- After stopping the pump, the concentra- ketone bodies must be identified), we lowing equations: sensitivity ϭ TP/[TP ϩ tion of plasma ketone bodies (determined found that the sensitivity of capillary FN], and specificity ϭ TN/[TN ϩ FP]. by the reference enzymatic spectrophoto- plasma ketonemia was increased (86.6%) Significance was implied at P Ͻ 0.05. metric method) was increased signifi- and remained higher than the sensitivity StatView software (version V; St. Abacus cantly at T60 compared with T0 (0.28 Ϯ of ketonuria (68.8%). Concepts, Brain Power, Calabasas, CA) 0.26 vs. 0.18 Ϯ 0.18 mmol/l, P ϭ 0.01) During CSII interruption, we re- was used for all calculations. and reached a maximum at T300 (1.30 Ϯ corded a total of 41 plasma glucose values Ϯ ϭ Ն250 mg/dl, a level leading to the detec- RESULTS 0.49 vs. 0.18 0.18 mmol/l, P 0.0002) (Fig. 1). Capillary blood ketone body lev- tion of ketone bodies in clinical practice. els were also increased significantly at In these 41 cases of hyperglycemia, we Clinical and metabolic T60 compared with T0 (0.13 Ϯ 0.29 vs. compared the frequency of positive val- characteristics of patients 0.05 Ϯ 0.15 mmol/l, P ϭ 0.05); the max- ues for plasma ketonemia, capillary blood in the study imum was reached at T300 (1.23 Ϯ 0.78 ketonemia, and ketonuria. Positive keto- The main clinical and biological charac- vs. 0.050 Ϯ 0.15 mmol/l, P ϭ 0.0003). sis was more frequently observed with teristics of the patients in the study are However, these levels remained signifi- plasma ketonemia (85%, 35 of 41 pa- summarized in Table 1. The average age of the 18 patients (9 women and 9 men) and the duration of diabetes (mean Ϯ SE) were 41.1 Ϯ 7.1 and 20.8 Ϯ 9.6 years, respectively.

Glucose levels Compared with T0, plasma glucose con- centrations were increased significantly at T120 after the pump had been stopped Figure 1—Changes in plasma (197 Ϯ 68 vs. 149 Ϯ 60 mg/dl, P ϭ (squares) and capillary blood ke- 0.0012) and reached a maximum at T300 tonemia (circles) during a 5-h in- Ϯ ϭ terruption of CSII (from T0 to (285 68 mg/dl, P 0.0002). Similarly, T300). Results are shown as capillary plasma glucose levels were sig- means Ϯ SD. *P Ͻ 0.05, **P Ͻ nificantly increased at T120 compared 0.001 between plasma and capil- with T0 (213 Ϯ 76 vs. 166 Ϯ 68 mg/dl, lary ketonemia; P Ͻ 0.05–0.001 P ϭ 0.004) and reached a maximum at versus 7:00 A.M. (T0).

DIABETES CARE, VOLUME 26, NUMBER 4, APRIL 2003 1139 Capillary ketone bodies during CSII interruption

Table 2—Sensitivity, specificity, and positive and negative predictive values of capillary ke- ketone testing. However, its routine use tonemia and ketonuria requires that normal values be distin- guished from pathological values and that Positive Negative a threshold of ketosis positivity be de- Sensitivity Specificity predictive value predictive value fined. Levels of circulating ketone bodies vary across populations of normal indi- Ketonuria 63 100 100 71.8 viduals, even after controlling for age and Capillary blood ketonemia 80.4 100 100 82.5 duration of fasting, presumably as a result of variations in basal metabolic rate, he- patic stores, and differences in tients) and capillary blood ketonemia measurement of ketosis according to the mobilization of amino acids from (78%, 32 of 41 patients) than with keto- ketonuria. muscle proteins (18). Most investigators nuria (59%, 24 of 41 patients) (P ϭ The development of inexpensive agree that normal levels of ketone bodies 0.017). quantitative tests of capillary blood are Ͻ0.5 mmol/l; hyperketonemia is de- When we compared the time delay to ␤-OHB levels was recently motivated by fined as levels Ͼ1 mmol/l, and DKA is confirm the diagnosis of ketosis (as de- the unreliability of the urinary ketone defined as levels Ͼ3 mmol/l (18). We fined by a ␤-OHB value Ն0.5 mmol/l or tests in their role as DKA diagnosis and have thus applied these levels of positivity ketonuria Ն4 mmol/l [moderate]), we management tools (19,20). Indeed, sev- in the present work. found a significant difference in the mean eral difficulties are associated with uri- Our study demonstrated clearly that time between the three methods of detec- nary ketone tests. First, patients perceive the hand-held ketone sensor accurately tion (P ϭ 0.0089). This was due to a signif- urinary ketone testing to be an unpleasant measures capillary blood ␤-OHB with a icant difference in time for measurement of and time-consuming experience, particu- strong correlation between the sensor val- ketosis according to plasma ketonemia larly in this era of promoting blood glu- ues and those measured by the venous compared with that of urinary ketone cose monitoring. Consequently, the rates plasma method. Available evidence sug- bodies (140 Ϯ 54 vs. 212 Ϯ 67 min, P ϭ of noncompliance are exceedingly high gests that this test could be helpful in 0.0023). Conversely, no significant dif- (13). Second, these urinary tests can pro- detection of underinsulinization and rec- ference between plasma and capillary ke- vide misleading information in the diag- ognition of impending DKA delayed by tonemia was noted for the mean time to nosis and management of DKA or failure of urinary ketone reagent strips observe ketosis positivity (140 Ϯ 54 vs. impending DKA, because they detect only (22). 180 Ϯ 60 min). After restarting the acetoacetate (AcAc) but not ␤-OHB, The sensitivity and negative predic- pump, the resolution of ketosis was which is the predominant ketone body tive value of capillary ketonemia mea- marked and rapid; all patients returned to produced in the case of DKA. Further- sured here were higher than those of an ␤-OHB value Ͻ0.5 mmol/l or ketonu- more, urinary ketone tests have been re- ketonuria. It is worth noting that for ria Ͻ4 mmol/l within 180 min of insulin ported to give false-positive results in the plasma glucose values Ն250 mg/dl, the infusion. Although the mean time to con- presence of drugs containing sulfhydryl main situation in which the presence ke- firm ketosis resolution seemed longer for groups such as the antihypertensive drug tone bodies must be identified, ketosis ketonuria (113 Ϯ 58 min) than for captopril, as well as mesna, N-acetylcys- positivity was more frequently reported plasma ketonemia bodies (93 Ϯ 37 min) teine, dimercaprol, and penicillamine with capillary ketonemia than with keto- and for capillary ketonemia bodies (87 Ϯ (21). False-negative readings also have nuria. Likewise, the mean time of positive 37 min), we were unable to find any sig- been reported when test strips have been detection for ketonuria was significantly nificant difference between the three exposed to air for an extended period of longer than for plasma ketonemia, sug- methods of ketosis detection (P ϭ 0.19). time or when urine specimens are highly gesting an increased risk in the delay of acidic, such as after the ingestion of large diagnosis of ketosis. Therefore, similar to CONCLUSIONS — In this study, we quantities of ascorbic acid (22). Health the fact that self-monitoring of blood glu- have examined the accuracy of the capil- care professionals should be aware that cose by fingerstick has replaced urine glu- lary blood method compared with the currently available urine ketone tests are cose testing, it is tempting to speculate plasma venous method for diagnosing ke- not reliable for diagnosing DKA or in that fingerstick determinations of ␤-OHB tosis in patients treated by pump infusion monitoring its treatment. Consequently, levels might increase patient compliance of insulin as a function of time after delib- the American Diabetes Association rec- with recommendations for ketone testing. erate interruption of CSII for 5 h. We ommends the use of blood ketone testing In this way, patients would have an earlier demonstrated that the sensor method for methods rather than urine ketone testing warning mechanism for detecting the de- measuring capillary blood ketone body for diagnosis and monitoring of DKA velopment of metabolic deterioration, levels gave results similar to the spectro- (23). which should theoretically bring about a photometric method used to measure The use of a capillary blood ketone reduced incidence of DKA observed in di- plasma ketone body levels and that capil- sensor seems to have many advantages abetic patients treated by external pump. lary blood ␤-OHB levels were highly and over ketonuria testing. Its ease of use, This could provide substantial savings in significantly correlated with plasma small sample volume, short test time (30 health costs for patients, particularly ␤-OHB. We also showed that the ketone s), automatic timing, and digital display those treated by pump (4). sensor has a higher sensitivity and pro- make the ketone sensor a simple method A previous study showed that ketone vided an earlier diagnosis of ketosis than that may increase patient compliance for body measurements using a reflectance

1140 DIABETES CARE, VOLUME 26, NUMBER 4, APRIL 2003 Guerci and Associates meter (Stat-Site meter; GDS Diagnostics) P: Risk of adverse effects of intensified of monitoring to diabetes. Diabetes Metab could enhance the management of DKA treatment in insulin-dependent diabetes Rev 15:412–426, 1999 (24). When acidosis had resolved, all pa- mellitus: a meta-analysis. Diabet Med 14: 14. Byrne HA, Tieszen KL, Hollis S, Dornan tients had ␤-OHB levels Ͻ0.5 mmol/l, 919–928, 1997 TL, New JP: Evaluation of an electro- whereas more than half of the patients 5. Peden NR, Braaten JT, McKendry JB: Di- chemical sensor for measuring blood ke- abetic ketoacidosis during long-term had positive ketonuria for up to 24 h after tones. Diabetes Care 23:500–503, 2000 treatment with continuous subcutaneous 15. Laffel L, Brink S, Kaufman F, Bergenstal R, correction of DKA, suggesting that keto- insulin infusion. Diabetes Care 7:1–5, nuria fails to detect the overall improve- Fineberg SE, Jenkins M: Frequency of el- 1984 evation in blood ␤-hydroxybutyrate (␤- ment and may lead to unnecessary and 6. Zinman B, Tildesley H, Chiasson JL, Tsui OHB) during home monitoring and potentially dangerous increases in insulin E, Strack T: Insulin lispro in CSII: results association with glycemia in insulin- of a double-blind crossover study. Diabe- dose delivery. treated children and adults (Abstract). Di- tes 46:440–443, 1997 In conclusion, the accuracy of ketosis abetes 49 (Suppl. 1):A92, 2000 screening and frequency of ketosis detec- 7. Melki V, Renard E, Lassmann-Vague V, Boivin S, Guerci B, Hanaire-Broutin H, 16. Report of the Expert Committee on the tion definitely may be improved by the Diagnosis and Classification of Diabetes use of this capillary blood ketone deter- Bringer J, Belicar P, Jeandidier N, Meyer L, Blin P, Augendre-Ferrante B, Tauber JP: Mellitus. Diabetes Care 20:1183–1197, mination method in clinical practice. 1997 Improvement of HbA1c and blood glucose stability in IDDM patients treated with lis- 17. McMurray CH, Blanchflower WJ, Rice pro insulin analog in external pumps. Di- DA: Automated kinetic method for D-3- Acknowledgments— This study was sup- abetes Care 21:977–982, 1998 hydroxybutyrate in plasma or serum. Clin ported by Abbott France-Division Medisense 8. Renner R, Pfutzner A, Trautmann M, Har- Chem 30:421–425, 1984 for the distribution of electrochemical strips. zer O, Sauter K, Landgraf R: Use of insulin 18. Mitchell GA, Kassovska-Bratinova S, We thank the staff of the Center d’Investi- lispro in continuous subcutaneous insu- Boukaftane Y, Robert MF, Wang SP, Ash- gation Clinique du CHU de Nancy for their lin infusion treatment: results of a multi- marina L, Lambert M, Lapierre P, Potier E: clinical help and technical assistance. center trial. German Humalog-CSII Study Medical aspects of ketone body metabo- This study is dedicated to Professor Pierre Group. Diabetes Care 22:784–788, 1999 lism. Clin Invest Med 18:193–216, 1995 Drouin, deceased 21 October 2002. 9. Guerci B, Meyer L, Salle A, Charrie A, 19. Sulway MJ, Malins JM: Acetone in dia- Dousset B, Ziegler O, Drouin P: Compar- betic ketoacidosis. Lancet 2:736–740, ison of metabolic deterioration between 1970 References insulin analog and regular insulin after a 20. Foster DW, McGarry JD: The metabolic 1. Diabetes Control and Complications Trial 5-hour interruption of a continuous sub- derangements and treatment of diabetic Research Group: The effect of intensive cutaneous insulin infusion in type 1 dia- ketoacidosis. N Engl J Med 309:159–169, treatment of diabetes on the development betic patients. J Clin Endocrinol Metab 84: 1983 and progression of long-term complica- 2673–2678, 1999 21. Csako G: Causes, consequences, and rec- tions in insulin-dependent diabetes mel- 10. Attia N, Jones TW, Holcombe J, Tambor- ognition of false-positive reactions for ke- litus: the Diabetes Control and Com- lane WV: Comparison of human regular tones. Clin Chem 36:1388–1389, 1990 plications Trial Research Group. N Engl and lispro after interruption of 22. Rosenbloom AL, Malone JI: Recognition J Med 329:977–986, 1993 continuous subcutaneous insulin infu- 2. Bode BW, Steed RD, Davidson PC: Reduc- sion and in the treatment of acutely de- of impending ketoacidosis delayed by ke- tion in severe with long- compensated IDDM. Diabetes Care tone reagent strip failure. JAMA 240: term continuous subcutaneous insulin 21:817–821, 1998 2462–2464, 1978 infusion in type I diabetes. Diabetes Care 11. Lebovitz HE: Diabetic ketoacidosis. Lan- 23. American Diabetes Association: Tests of 19:324–327, 1996 cet 345:767–772, 1995 glycemia in diabetes. (Position State- 3. Diabetes Control and Complications Trial 12. Singh RK, Perros P, Frier BM: Hospital ment). Diabetes Care 25:S97–S99, 2002 Research Group: Implementation of treat- management of diabetic ketoacidosis: are 24. Umpierrez GE, Watts NB, Phillips LS: ment protocols in the Diabetes Control clinical guidelines implemented effec- Clinical utility of ␤-hydroxybutyrate de- and Complications Trial. Diabetes Care tively? Diabet Med 14:482–486, 1997 termined by reflectance meter in the man- 18:361–376, 1995 13. Laffel L: Ketone bodies: a review of phys- agement of diabetic ketoacidosis. Diabetes 4. Egger M, Davey Smith G, Stettler C, Diem iology, pathophysiology and application Care 18:137–138, 1995

DIABETES CARE, VOLUME 26, NUMBER 4, APRIL 2003 1141