Diabetes Research amtCIhfical Practk'e. 5 (1988) 55-61 Elsevier

DRC00204

55

Capillary glucose determination in the screening of gestational diabetes

Ernesto Meriggi, Gian Franco Trossarelli, Quirico Carta ~, Guido Menato, Maria Angela Porta, Rita Bordon and Leone Gagliardi

lstituto di Ginecologia e Oslelricia dell'Universit~'t di Torim~. aml ~ Cenlro Anlidiahetico dell'OslJe&tle Molinette di Torhm. Turhz. Italy

(Received 6 October 1987, revision received 4 January 1988, accepted 5 January 1988)

K<v word,',': Gestational diabetes: Diabetes screening: Capillary glucose

Summary

Paired capillary-venous blood samples were obtained from 418 pregnant women undergoing an oral glucose challenge test (GCT) for the screening of gestational diabetes (GD). The relationship between capillary and plasma glucose concentrations was investigated in order to establish a capillary GCT threshold. Plasma glucose was assayed by the glucose oxidase method and capillary glucose using Reflocheck Glucose strips and a Reflocheck reflectance meter. During GCT the capillary values exceeded plasma glucose values by a mean difference of 10-12 mg/dl fasting and 22 24 mg/dl after 1 h. A high correlation between the glucose values of the two techniques was found, particularly for those at I h, with corresponding capillary deter- minations being 20 mg/dl above plasma values. Tile sensitivity, specificity and predictive value of the various capillary thresholds investigated in detecting GD corresponded substantially to the accuracy of plasma thresholds 20 mg/dl lower. The receiver operator characteristic curves of the plasma and capillary thresholds were similar in shape and the optimal cut-off point for performing a diagnostic test was set at 135 and 155 mg/dl, respectively. These cut-off values should be reconsidered in tile light of the costs and perinatal out- come.

Introduction

The early detection and the correct treatment of gestational diabetes (GD) are very important not only for the improvement of the perinatal outcome

Address for correspondence: Dr. Ernesto Meriggi, Istitt.to di Ginecologia e Ostetricia deH'Universit~i, Via Venlimiglia, 3. 10126 Turin, Italy.

but also for tile future health of mothers and their babies. Unfortunately an ideal protocol for tile management of GD is not yet available. However, the recommendations of tile Second International Workshop Conference oll GD [1] should be a useful point of reference for screening, diagnostic, and therapeutic programs. It was established that screening is indicated in all pregnant women with or without clinical and/or ananmestic risk factors

0168-8227/88/$03.50 ,i' 1988 Elsevier Science Publishers B.V. (Biomedical Division l

56

for GD and that the screening test should be per- formed by administering a 50-g oral glucose load and assaying blood glucose alter 1 h. Many clinical studies have shown the advantages of this test, in- cluding advantages fl'om tile cost/benefit point of view. However, some doubts still exist concerning tile timing of the blood specimen and the threshold value for test positivity [2-7]. Recently the reflec- tance meter and the reagent strip of the assay of capillary blood glucose have become popular home monitoring procedures for diabetic disease, since they provide rapid results, are easy to use, have a low cost and are very accurate compared to labo- ratory glucose determinations. Several studies have confirmed a good correlation between glucose con- centration values measured by a variety of reflec- tance meters and the values obtained by enzymatic techniques on capillary or plasma specimens, the best results being observed with the self-calibrating instruments [8 10]. The advantages of home blood glucose monitoring in diabetic pregnant women are well known and recently attempts have been made to better understand tile relationship between cap- illary and venous glucose concentrations during pregnancy as well as to construct capillary equiva- lents for venous-derived norrns [11].

The aims of this study are (a) to verify, following the guidelines of the screening protocol we adopted for GD, the plasma glucose threshold for glucose challenge test (GCT) positivity, (b) to determine the differences and correlations between glucose wdues obtained simultaneously fl'om plasnla by labora- tory techniques and from capillary blood with a re- flectance meter during GCT and (c) to determine the capillary threshold for GCT positivity in order to facilitate the use of the reflectance meter in GD screening programs for all pregnant women.

Materials and methods

The study population consisted of women receiving prenatal care at the Department of Obstetrics and Gynecology of Turin University. A total of 418 patients who were 25 years of age or older were given a GCT for GD screening between June, 1985

and December, 1986. In all cases the fasting plasma glucose levels assayed a few days before the test were below 95 mg/dl. In those cases with clinical and/or anamnestic risk factors plasma glucose val- ues assayed 2 h after lunch were all below 120 mg/dl. In the 122 cases without risk factors (group 1) the GCT was performed only between the 24th and the 28th week, while ill the cases with risk fac- tors the first GCT was performed between the 12th and the 16th week, and, if negative, it was repeated between the 24th and the 28th week. Out of 296 cases with risk factors, 116 patients (group 2) were seen ill time for GCT bookings to be made as in- dicated ill the protocol, while 180 patients (group 3) came late and were submitted to the G CT only once between the 24th and the 28th week. The test was performed in all cases in the morning after an overnight fast of at least 12 h and without dietary preparation in the preceding days. Paired venous and capillary blood specimens were drawn fasting and at 1 h after an oral glucose load of 50 g in 200 ml of water. The GCT was considered positive if the plasrna glucose value was equal to or above 135 mg/dl as suggested by Carpenter and Coustan [3]. The patients with a positive G CT underwent an oral glucose tolerance test (GTT) in the morning after an overnight fast of at least 12 h and after 3 days of unrestricted diet (> 250 g carbohydrates) and physical activity. First, a fasting blood sample for plasma glucose was obtained, then a 100-g oral glucose load was given in 400 ml of water and ad- ditional blood samples were drawn at 1, 2 and 3 h after the drink. A definitive diagnosis of GD re- quired that two or more of the following plasma glucose concentrations be met or exceeded: 95, 180, 155 and 140 mg/dl for the fasting, l-h, 2-h and 3-h assays respectively as proposed by Carpenter and Coustan [3].

Plasma glucose was measured oil a IL 919 ana- lyzer (Instrumentation Laboratories, Milan, Italy) which uses the glucose oxidase method. Capillary glucose was measured using a Reflocheck Glucose strip (glucose oxidase:peroxidase technique) and a Reflocheck reflectance meter (both from Boehrin- ger Mannheim Diagnostic, Inc.). All capillary glu- cose meast, rements were performed by three of the

authors (M.A.P., R.B., G.M.) specifically trained in the appropriate technique, lnterassay variation among operators during the training period was within 5%.

Statistical analysis was performed with the use of regression analysis, the t-test for paired data and the chi-square test when appropriate. The receiver operator characteristic curve (ROC) was also con- structed to determine an optirnal plasma and cap- illary blood glucose threshold for performing a GTT [12].

Results

A total of 530 GCTs were performed on 418 patients. In group 1, 20 cases (16.4%) were positive to GCT and 5 (4.1%) positive to GTT. In group 2, 19 cases (16.4%) were positive to the first GCT (12-16 weeks) and four (3.4%) positive to GTT; 37 cases (33%) were positive to the second GCT (24- 28 weeks) and nine (8%) were positive to GTT. In group 3, 54 cases (30%) were positive to GCT and nine (5%) were positive to GTT. In all, 111 cases (26.5%) were found to be GCT-positive and 27 (6.5%) to be GTT-positive. The different incidences of GTT-positive cases in the various groups were not statistically significant. Among the cases with GD, only five (18.5%) had a severely impaired G TT (two values above 3 SD) and 16 (59.2%) were obese; the treatment of GD cases was only dietetic, the length of pregnancy was 38.6 4- 1.9 weeks, ma- ternal cornplications (toxemia) occurred in three cases ( 11.1%), the cesarean section rate was 14.8%,

57

n. cases

430. 420. 410 "

400- i 7O !

60-

50-

40 -

3 0 -

20"

1 0 -

0 < 1 3 5

n. cases

429 ~ GCT-

E~'~ GCT + GTr+

20169

1124 4123

135/'39 140t44 145/49 :>150 PLASMA GLUCOSE mg/dl

<155

~ GCT-

] OT + GTT+

18/61

2/19 3111

155159 160m4 165/s9 :>170 CAPILLARY GLUCOSE

Fig. I. Incidence of gestaiional diabetes (GD) in groups with different plasma or capillary values of the glucose challenge test.

All cases with GD ( G T T + ) had a capillary glucose value equal

or above 155 mg/dl and the incidence of GD in the various

groups indicated a dilTerence of about 20 mg/dl l\~r the capilhlry

deterrnir~ations.

the neonatal morbidity (hypoglycemia, hypocal- cemia, hyperbil irubinemia)was 18.5% and three (11.1%) macrosomic newborns ( >i4000 g) were ob- served. Fig. 1 shows the distribution of cases de- pending on negativity oz" positivity to GCT and, for the latter group, the incidence of GTT-positive cas- es when different plasma or capillary threshold levels are used. As regards the plasma glucose de- terminations, 24.5% of cases had values equal to or above 135 mg/dl and in the various subgroups with glucose values of 135 139, 140 144, 145 149, and ~> 150 mg/dl the incidence of GD was respectively 4.2%, 17.4%, 14.3% and 29'7o. As concerns the capillary glucose determinations, 19.1% of cases had values equal to or above 155 mg/dl and in the various subgroups with glucose values of 155 159,

TABLE 1

A C C U R A C Y OF THE VARIOUS PLASMA T H R E S H O L D S OF THE G L U C O S E C H A L L E N G E TEST IN D E T E C T I N G GES- T A T I O N A L DIABETES

m g d l ~> 135 ~> 140 /> 145 ~> 150

Sensibility (%) 100.0 96.2 g 1.5 74. I

Specificity (%) 79.5 84. I 87.8 90.2 Predictive value abnormal test (%) 20.7 24.5 26.5 28.9

Predictive vahie normal lest (%) 100.0 99.7 98.8 94.4

B8

TABLE 2

A C C U R A C Y OF THE VARIOUS C A P I L L A R Y T H R E S H O L D S OF THE G L U C O S E C H A L L E N G E TEST IN D E T E C T I N G

G E S T A T I O N A L DIABETES

The characteristics of the capillary thresholds are substantially similar to the corresponding plasma thresholds that arc 20 mg/dl lower.

mg/dl /> 155 /> 160 >/165 >~ 170

Sensibility (%) 100.0 88.9 81.5 70.3

Specilicity (%) 85.3 86.8 90.2 91.5

Predictive value abnormal test ('%) 26.7 26.4 30.9 30.6

Predictive value normal test ('%) I00.0 99.3 98.9 98.3

160 164, 165 169, and ~> 170 mg/dl the incidence of GD was respectively 27.3%, 10.5%, 40% and 29.5%. The accuracy and predictive values of the various plasma and capillary GCT thresholds in detecting GD are reported in Tables 1 and 2. The values of the various plasma thresholds and those of the corresponding capillary thresholds 20 mg/dl higher are substantially similar.

The differences between the mean values of plas- ma and capillary glucose determinations fasting and 1 h after glucose load in the various groups of pregnant women are given in Table 3. Mean cap- illary values were significantly above (P < 0.001) plasma values both fasting and after 1 h in all groups, with mean differences respectively 10-12 mg/dl and 22-24 mg/dl. The correlation between the plasma and capillary fasting blood glucose val- ues was significant in all groups (P < 0.001), with

a regression coefficient value of 0.33 for group l, 0.42 for group 2 and 0.45 for group 3. Fig. 2 shows the correlation between plasma and capillary glu- cose values 1 h after GCT in cases with and without risk factors. The line of the best fit for the latter cases was capillary glucose = 0.86 plasma glucose + 36, and capillary glucose = 0.98 plasma glucose + 17 for the former cases. The regression coeffi- cient was high and statistically significant (P < 0.001) both for cases with (r = 0.91) and without (r = 0.85) risk factors and the slopes of the two lines were not significantly different.

The ROC technique provides a clear, graphic analysis of the performance of OCT, a quantitative method for determining what cut-off point to use and the level of efficacy one can expect in detecting GD. Graphically, Fig. 3 shows clearly that a plasma glucose value of 135 mg/dl and a capillary value of

TABLE 3

C A P I L L A R Y A N D PLASMA G L U C O S E C O N C E N T R A T I O N S D U R I N G G L U C O S E C H A L L E N G E TEST

Results arc given as mean :t: SD at fasting and I I1 after a 50-g oral glucose load. Mean capillary wdues are signilicantly higher than

plasma values in all groups with mean differences (C - P) of 13'¼, for the fasting and of 18°/,, for the 1-11 determinations.

Group I (n = 122) G r o u p 2 ( i t = 112) Group 3 (n = 180)

fasting I h fasting I 11 fasting 1 h

Capillary (mg,'dl) 84.8 8.5" 132 25.7* 84.3 8.6* 135 27.3* 82.8 17.7" 137 29.9*

P l a s m a ( m g d l ) 73.2 7.1 I10 25.8 75.1 7.6 117 27.7 73.5 11.2 121 27.8

C - P(nag/dl) 13.1 5.7 22 11.3 10.7 4.7 23 11.5 11.2 5.1 24 12.1

* P < 0.001: statistical evaluation perlbrmed with Student 's /-test for paired data.

o

.-Jul O ::"- O ~...i

5~ .JIl~

mg/dl

250"

2OO

150

100

50-

0

without R.F. n = 1 2 2 ~ r = 0.85 ~ , ~ p < .001 / /+

/ r = 0.91 / p < .001

, rag/dr

PLASMA GLUCOSE

Fig. 2. Linear regression between plastna and capillary glucose concentrations at I h of the glucose challenge test. The regres- sion coefficient value is high and statistically significant both for cases with and without risk factors (R.F.) for GD and the slopes

of the two lines are not statistically difl+crent.

155 mg/dl represent optimal cut-off points for rec- ommending a diagnostic test (GTT). In practice the choice of this cut-off point is also influenced by clin- ical judgement regarding epidemiological, eco- nomic, psychological and many other factors which may be different for each clinical situation [12]. In other words, cut-off points are deeply influenced by

1.00 ,

,96

.81

I- .74

.50

PLASMA GLUCOSE

135 140 ~ 1.00"

';/°Y + ~ .81

.70

- - RICHAROSON

.50

,10 .12 .16 .20 1.00

1-- SPECIFICITY

CAPILLARY GLUCOSE

155 160 P 148

/ •

~165 I 1 170

64

-- ,7. LANDON

59 • • • • Jf~'l

.08.10 .13 .15 1.00

1-- SPECIFICITY

Fig. 3. Rccciver opcrator characteristic curve (ROC) of tfic plas- ma and capillary thresholds of the glucose challenge test (GCT). The figure shows thc comparison bctwccn our ROCs (dotted lines) and those of Richardson et al. [12] and of Landon ct al. [13] (continuotts lines)Ibr GCT phtsma and capilhtry thresholds, respectively. The reference fc, r our plasma tfircsl|old ROC is a positive GTT and fc, r capillary threshold ROC is a plasrna glu-

cose wthte ~> 135 mg.dl.

59

the different weights attributed to the diagnostic or the therapeutic aspects of the test.

Discussion

The data of this study confirm the already described utility of a screening program for GD that includes testing of all pregnant women, with or without risk factors, over 24 years of age [2,3,5,14]. The pro- portion of GD cases observed in the group without risk factors (4.1%) is not significantly different from that in the group with risk factors (7.4%) and these proportions are very similar to those given by others [3,13] who used similar criteria for establish- ing GCT and G TT positivity. Our incidence of GD is obviously higher than that obtained by authors [2,5,16] who adopted higher thresholds for GCT and G TT than ours. We believe that it is advisable to perform early screening (12 16 weeks) in cases with risk factors, since about one third of GTT- positive cases can be diagnosed and suitably treat- ed.

Our data relative to plasma and capillary glucose determinations fasting and 1 h after GCT confirm the observations of those authors who used labo- ratory techniques and reflectometric systems similar to ours [11,13]. Since about 92% of the capillary glucose values were higher than those of venous plasma and the correlation coefficient between the values of the two determinations 1 h after GCT was highly significant, a mean difference of 20 mg/dl for the capillary values seems acceptable, in particular for those values in a range of 80-250 mg/dl that usually are observed 1 h after a 50-g oral glucose load. The capillary threshold corresponding to the plasma threshold will therefore be 20 mg/dl higher. The evaluation of the characteristics of GCT plas- ma and capillary thresholds, as well as the shape of their respective ROCs, confirms these observations. Capillary threshold accuracy percentages (Table 2) in wtrious subsets of glucose values are substan- tially similar to the corresponding plasma thresh- olds which are 20 mg/dl lower (Table 1). As for the evalttation of the ROCs, which have been con- structed by us for GCT plasma and capillary

60

thresholds, we must stress the similarity of their shapes (Fig. 3). More specifically, in comparing our ROC for plasma thresholds with that of Richard- son et al. [12], we noticed an exact correspondence of the values of our series at 135 mg/dl with the cut-off point of Carpenter and Coustan [3] and, at 145 mg/dl, with the cut-off point of O'Sullivan et al. [14]. Our ROC for capillary thresholds, on the other hand, turned out to be different from that of Landon et al. [13], in that it shows a higher inci- dence of false-positive results for all points, but a higher sensitivity, above all at 165 and 170 mg/dl. Moreover, Landon et al. [13] emphasize that the capillary threshold of 160 mg/dl, if compared with the plasma threshold of 135 mg/dl, represents an optimal reference point for performing a GTT and the analysis of the costs for a GD screening pro- gram with the reflectance meter shows a saving of about 80% with respect to the laboratory enzy- matic technique. This reduction of the GCT costs is similar to ours, since the methods for the plasma and capillary glucose determination were the same in both studies.

As mentioned before, the definition of an optimal threshold for GCT is a controversial point at the moment. In particular there are detailed studies on the costs of a screening program for GD that show savings of about 50% for each case of GD detected if the screening is limited to all pregnant women whose age is over 24 years, though in this manner over 20% of GD cases are lost [4,5]. Marquette et al. [18] believe that, if the plasma GCT threshold is set at 150 mg/dl, about 50% fewer GTTs are per- formed than would be needed with a 130-mg/dl threshold, independently of the age of the mothers. Furthermore, if GCT sensitivity is increased with a second plasma glucose determination after 2 h, the cost of each GD detected is reduced by a further 32% [7]. There are authors [19] who have shown that even minor abnormalities of carbohydrate metabolism during pregnancy may affect the preg- nancy outcome and others [15] who have shown that there is only an increased risk of macrosomic newborns in cases with abnormal GCT and normal

G TT (11.9%) compared to cases with normal GCT (6.4%) and that the patient at risk may benefit from dietary control and more intensive evaluation of fasting and postprandial glucose levels.

It is also our impression that, considering the lit- erature data and those of our case series, in par- ticular the data concerning type of therapy and peri- natal outcome of GD cases (out of the three patients with macrosomic babies only one had a GCT plasma value below 150 mg/dl and all cases of maternal toxemia and neonatal complications occurred in patients with GCT plasma values equal to or above 150 mg/dl), if we elevate the plasma GCT threshold from 135 to 150 mg/dl the loss of test sensitivity is not of practical significance. A suitable diet can be given to patients with GCT thresholds between 135 and 149 mg/dl and a second GCT should be performed after 6-8 weeks. As pre- viously said, the choice of cut-off points differs, de- pending on whether GCT is being employed as a preliminary selection procedure in order to reach a further diagnostic refinement or in order to enroll the positive cases in a suitable therapeutic protocol. Therefore, the less frequently the implementation of more expensive and refined diagnostic tests (i.e., GTT) is inspired by false-positive results, the more optimal the situation is,. This means that priority will be given to those cut-off points which guarantee a higher specificity. Conversely, the more frequently the implementation of suitable therapeutic proto- cols is inspired by true-positive results, the more optimal the situation is.

In conclusion, we believe the reflectance meter has definite advantages over the laboratory enzy- matic techniques in the screening of G D and that its simplicity, precision of blood glucose determi- nations and high efficiency make it a suitable in- strument for the screening of G D in all pregnant women. A larger population study is needed to con- firm the accuracy of the plasma G CT threshold of 150 mg/dl, already used by other authors [2,16,20], and to verify, in the light of the maternal and neo- natal outcome, the safety and validity of a corre- sponding capillary threshold of 170 mg/dl.

Acknowledgement

We are deeply grateful to Dr. F. Bottasso, Chief of the Epidemiological Unit of the Regione Piemonte, for the statistical evaluation of the data and for the revision of the manuscript.

References

I Second International Workshop Conference on Gestational Diabetes Mellitus (1985) Summary and recommendations. Diabetes 34 (Suppl. 2), 123 126.

2 Lavin, J.P., Barden, T.P. and Miodovnik, M. (198l) Clinical experience with a screening program for gestational diabetes. Am. J. Obstet. Gynec. 141,491-494.

3 Carpenter, M.W. and Coustan, D.R. (1982) Criteria for screening tests for gestational diabetes. Am. J. Obstet. Gy- nec. 144, 768-773.

4 Reed, B.D. (1984) Screening for gestational diabetes. Analy- sis by screening criteria. J. Family Pract. 19, 751-755.

5 Marquette, J.P., Klein, V.R. and Niebyl, J.R. ([ 985) Etficacy of screening for gestational diabetes. Am. J. Perinatol. 2. 7-9.

6 Coustan, D.R., Widness, J.A., Carpenter, M.W., Rotondo, L., Pratt, D.C. and Oh, W. (1986) Should the fifty-gram, one-hour plasma glucose screening test for gestational dia- betes be administered in fasting or fed state? Am. J. Obstet. Gynec. 154, 1031 1035.

7 Weiner, C.P., Fraser, M.M., Burns, J.M., Schndor, D., Her- rig, J. and Whitaker, L.A. (1986) Cost e[ficacy of routine screening for diabetes in pregnancy: I-h versus 2-h specimen. Diabetes Care 9, 255 259.

8 Shapiro, B., Sawtge, P.J., Lomatch, D. et al. (198l) A com- parison of accuracy and estimated cost of methods for home blood glucose monitoring. Diabetes Care 4, 396-403.

9 Koschinsky, T.. Gries. F.A., Rosak, C. et al. (1984) A new

61

instrument for determination of blood glucose. Dtsch. Med. Wschr. 109, 97-100.

10 Rayman, G., Spencer, P.D., Tyller, C.R. and Wise, P.H. (1984) Evaluation o['a self-calibrating blood glucose moni- tor. Diabetes Care 7, 378-380.

I 1 Weiner, C.P., Faustich, M., Burns, J., Fraser, M., Whitaker, L. and Klugman, M. (1986) The relationship between cap- illary and venous glucose concentration during pregnancy. Am. J. Obstet. Gynec. 155, 61-64.

12 Richardson, D.K., Schwartz, J.S., Weinbaum, P.J. and Gabbe, S.G. (1985) Diagnostic test in obstetrics: a method for improved evaluation. Am. J. Obstet. Gynec. 152, 613- 618.

13 Landon, M.B., Cembrowski, G.S. and Gabbc, S.G. (1986) Capillary blood glucose screening for gestational diabetes: a preliminary investigation. Am. J. Obstet. Gynec. 155, 717-721.

14 O'Sullivan, J.B., Mahan, C.M., Charles, D. and Dandrow, R.V. (1973) Screening criteria for high-risk gestational dia- betic patients. Am. J. Obstet. Gynec. 116, 895 900.

15 Leikin, E.L., Jenkins, J.H., Pomerantz, G.A. and Klein, L. ([987) Abnormal glucose screening tests in pregnancy: it risk [actor for fetal macrosomia. Obstct. Gynec. 69, 570-573.

16 Jovanovic, L. and Peterson. C.M. (1985) Screening for ges- tational diabetes. Optimum tinting and criteria for retesting. Diabetes 34 (Suppl. 2), 21 23.

17 National Diabetes Data Group (1979) Classification and di- agnosis of diabetes mellitus and other categories of glucose intolerance. Diabetes 28, 1039-1057.

18 Marquette, G.P., Klein, V.R., Repke, J.T. and Niebyl, J.R. (1985) Cost-effective criteria for glucose screening. Obstet. Gynec. 66, 181-184.

19 Tallarigo, L., Giampietro, O., Penno, G., MJccoli, R., Gre- gori, G. and Nowdesi, R. (1986) Relation of glucose toler- ance to complications of pregnancy in nondiabelic women. N. Engl. J. Med. 315, 989-992.

20 Cousins, L., Dattel, B., Hollingsworlh, D., Hulbert, D. and Zettner, A. (1985) Screening for carbohydrate intolerance in pregnancy: a comparison of two tests and reassessment of a common approach. An1. J. Obstet. Gynec. 153, 381-385.