Amniotic fluid glucose concentration: A rapid and simple method for the detection of intraamniotic infection in preterm labor Roberto Romero, MD, Carlos Jimenez, MD, Ajay K. Lohda, MD, Jose Nores, MD, Shuichi Hanaoka, MD, Cecilia Avila, MD, Robert Callahan, BS, Moshe Mazor, MD, John C. Hobbins, MD, and Michael P. Diamond, MD New Haven, Connecticut The purpose of this study was to determine whether amniotic fluid glucose concentrations is of value in the rapid diagnosis of intraamniotic infection. Amniocenteses were performed in 168 patients with preterm labor and intact membranes. Amniotic fluid was cultured for aerobic and anaerobic bacteria, as well as Mycoplasma species. The prevalence of positive amniotic fluid cultures was 13.6% (23/168). Patients with positive amniotic fluid cultures for microorganisms had significantly lower median amniotic fluid glucose concentrations than patients with negative amniotic fluid cultures (median 11 mgldl, range 2 to 30 mgldl vs median 28 mgldl, range 3 to 74, respectively; p < 0.001). Amniotic fluid glucose concentrations below 14 mgldl had a sensitivity of 86.9% (20/23), a specificity of 91.7% (133/145). a positive predictive value of 62.5% (20/32), and a negative predictive value of 97.8% (133/136) in the detection of a positive amniotic fluid culture. Amniotic fluid glucose determination is a rapid, sensitive. inexpensive, and simple test for the detection of intraamniotic infection in women with preterm labor and intact membranes. (AM J OBSTET GVNECOL 1990;163:968-74.)
Key words: Glucose, intraamniotic infection, chorioamnionitis, preterm labor, prematurity, Gram stain, amniotic fluid
Strong evidence supports an association of intraamniotic infection with preterm labor and delivery. 1 Patients in preterm labor with intact membranes and positive amniotic fluid cultures are more likely to have clinical chorioamnionitis and fail to respond to tocolysis. I. 2 Furthermore, neonates born to women with positive amniotic fluid cultures are at an increased risk for infectious complications! The early diagnosis of intraamniotic infection is a desirable clinical goal. However, clinical signs of infection are insensitive and will detect only 12% of patients with positive amniotic fluid cultures in women with preterm labor and intact membranes! While amniotic fluid cultures are widely used to assess the microbiologic state of the amniotic cavity, results are not immediately available and may take up to 5 days. Rapid methods for the detection of infection, such as the Gram stain and the From the Department of Obstetrics and Gynecology, Yale University School of Medicine. Supported by a grant from the Walter Scott Foundation for Medical Research. Dr. Romero is the recipient of a Physician Scientist Award from the National Institute of Child Health and Human Development. Dr. Diamond is a Clinical Associate Physician of the General Clinical Research Center of the NIH at Yale University School of Medicine. Presented at the Tenth Annual Meeting of the Society of Perinatal Obstetricians, January 23-27, 1990. Reprint requests: Roberto Romero, MD, Yale University School of Medicine, Department of Obstetrics and Gynecology, 333 Cedar St., P.O. Box 3333, New Haven, CT 06510. 616122595
968
acridine orange stain of amniotic fluid, are insensitive, detecting only 50% to 60% of all intraamniotic infections.'" Other tests, such as gas-liquid chromatographic analysis of organic acids in amniotic fluid, require expensive equipment and trained personnel and are not available in most medical centers. Thus there is a need for a sensitive, simple, and rapid test for the detection of intraamniotic infection. A decreased glucose concentration in several body fluids (i.e., cerebrospinal fluid, pleural fluid, and synovial fluid) has been used for the diagnosis of infection.'-7 The purpose of this study was to determine whether amniotic fluid glucose concentration could be of value in the diagnosis of intraamniotic infection. The value of amniotic fluid glucose concentration in the diagnosis of intraamniotic infection was first suggested by Dr. Brian Kirshon (personal communication, 1989).
Material and methods Patient population. The study population consisted of 168 consecutive patients who were admitted to YaleNew Haven Hospital with the diagnosis of preterm lahor and intact membranes from March 3, 1988, to May 26, 1989, and who underwent amniocenteses for evaluation of the microbiologic status of the amniotic fluid. Patients with known abnormal glucose tolerance test results or with overt diabetes mellitus were excluded. Preterm labor was defined as the presence of
Volume 163 Number 3
regular uterine contractions with a frequency of at least two every 10 minutes. Rupture of membranes was excluded by testing for pooling, Nitrazine paper reaction, and ferning. Ritodrine, terbutaline, or magnesium sulfate was administered intravenously for tocolysis after amniocentesis. Failure of tocolysis was diagnosed when cervical dilatation progressed beyond 5 cm or delivery occurred. Retrieval of amniotic fluid. In all cases amniotic fluid was retrieved by transabdominal amniocentesis, which was performed under ultrasonographic guidance. Microbiologic culture technique. Amniotic fluid was transported to the laboratory in a capped plastic syringe immediately after collection. These conditions precluded air contact with the specimen. Plating occurred within 30 minutes of collection in all cases. Amniotic fluid was cultured for aerobic and anaerobic bacteria, as well as for Mycoplasma species (Mycoplasma hominis and Ureaplasma urealyticum), according to methods previously described" Gram stain examination. Gram stain examinations were performed by using commercial reagents (crystal violet, saffranin, and Gram's iodine [Difco Laboratories, Detroit]) under standard conditions. Stained slides were then examined by trained technologists, and the presence or absence of microorganisms was noted. The results of the Gram stain examinations were shared with the clinicians. Patients with negative Gram stain examinations for bacteria were given tocolysis. Patients with positive Gram stain examination of amniotic fluid for bacteria were given parenteral antibiotics and allowed to labor. A positive Gram stain for bacteria was an indication for discontinuing tocolysis. Amniotic fluid glucose determinations. Amniotic fluid was centrifuged at 700 g for 10 minutes at 4° C and stored at -70 0 C until assayed. Glucose analysis was performed by the glucose oxidase method with a Beckman glucose analyzer. Samples were run in duplicate. Results were not available for patient management. The coefficient of variation was 5%. Criteria for diagnosis of chorioamnionitis and neonatal sepsis. Intraamniotic infection was defined as the presence of a positive amniotic fluid culture.' Clinical chorioamnionitis was defined according to the criteria proposed by Gibbs et al. 8 Neonatal sepsis was diagnosed by the presence of a positive culture of blood, urine, or cerebrospinal fluid. Statistical analysis. A Mann-Whitney test was used to compare amniotic fluid glucose concentrations between infected and noninfected patients. A receiver operator characteristic curve analysis was conducted with the program True Epistat (Epistat Services, Richardson, Tex.). Diagnostic indices (sensitivity, specificity, and positive and negative predictive values) for the detection of a positive amniotic fluid culture were calculated for both the Gram stain and amniotic fluid glucose.
Amniotic fluid glucose and intraamniotic infection
969
Results
The criteria for inclusion into this study were met by 168 patients, 54.1 % (91/168) of whom went on to have a preterm delivery. The remainder were delivered at term. The prevalence of positive amniotic fluid cultures was 13.6% (23/168). Fig. I displays amniotic fluid glucose concentrations according to culture results. Patients with positive amniotic fluid cultures for microorganisms had significantly lower median amniotic fluid glucose concentrations than patients with negative amniotic fluid cultures (median 11 mgldl, range 2 to 30 mgldl vs median 28 mgldl, range 3 to 74, respectively; p < 0.001, MannWhitney test). To describe the relationship between the sensitivity (true-positive rate) and the false-positive rate for different values of amniotic fluid glucose in the detection of a positive amniotic fluid culture, a receiver operator characteristic curve was constructed. Fig. 2 displays the receiver operator characteristic curve (area under the curve 0.93; SE 0.03; z 14.1; P < 0.001). The curve is well above the 45-degree line, indicating that an abnormal test result (reduced amniotic fluid glucose concentration) increases the probability of a positive amniotic fluid culture. As with any other diagnostic test, there is a trade-off between the sensitivity (true-positive rate) and the false-positive rate (100 - Specificity). We chose to present the diagnostic indices using a cutoff at 14 mgl dl, as this value seems to represent a reasonable compromise between a true-positive and a falsepositive rate. A glucose level below 14 mg 1dl had a sensitivity of 86.9% (20/23), a specificity of 91. 7% (1331145), a positive predictive value of 62.5% (20/32), and a negative predictive value of 97.8% (1331136). On the other hand, the diagnostic indices for the Gram stain examination of amniotic fluid were: sensitivity of 65.2% (15/23), specificity of 99.3% (144/145), positive predictive value of93.7% (15116), and negative predictive value of 94.7% (1441152). When the Gram stain and glucose level (below 14 mgl dl) were used in combination (either test result considered abnormal), the diagnostic indices were sensitivity of 86.9% (20/23), specificity of 91% (1321145), positive predictive value of60.6% (20/33), and negative predictive value of 97.8% (1321l35). There were 131 patients with negative Gram stains and glucose concentrations above 14 mgl dl. Three had positive amniotic fluid cultures (two with Ureaplasma urealyticum and one with Gardnerella vaginalis). All three patients failed to respond to intravenous tocolysis and were delivered of preterm neonates. Table I displays the clinical data of the 12 patients with amniotic fluid glucose concentrations of < 14 mgldl and negative amniotic fluid cultures for bacteria and Mycoplasma species. Although all of these patients
970
Romero et al.
September 1990 Am J Obstet Gynecol
80
• 70
...--.
--E
• • ••
60
Ol
.......Q)
50
•
(Jl
0
•
U
::::l
a "'0
40
'5
u::: u
'';::::;
0
30
•
20
• • •
c
E
«
-----=-..,.111 -----•
=0
10 0
--.....
""
•
Infection N = 23
• e::-
.= • •
•
•
No Infection
N = 145
Fig. 1. Amniotic fluid glucose concentrations in the study population. The median amniotic fluid glucose concentration in patients with intraamniotic infection was significantly lower than in patients with negative amniotic fluid cultures (median 11 mg/dl, range 2 to 30 mg/dl vs median 28 mg/dl, range 3 to 74, respectively; p < 0.0001).
received parenteral tocolysis, 75% (9112) went on to have a preterm delivery. Five of the nine patients with negative amniotic fluid cultures had been treated with antibiotics before undergoing amniocentesis, and one of the neonates subsequently had documented neonatal sepsis. In addition, one neonate was diagnosed to have congenital syphilis, but treponemas were not recovered from the amniotic fluid. Two of the three remaining patients had no clinical evidence of infection (maternal or neonatal), but their placentas showed acute inflammatory lesions (histopathologic chorioamnionitis). Three patients with low glucose concentrations and negative amniotic fluid cultures were delivered at term. One patient had polyhydramnios of an unknown cause. The second patient had fever (101° F) and was treated with erythromycin because of a positive endocervical culture for Chlamydia trachQmatis. The remaining patient had no evidence of an infectious process. Table II displays the clinical course and microbiologic characteristics of patients with positive amniotic fluid cultures. Twenty-one percent (5/23) of patients had polymicrobial infections. Eight patients had negative
Gram stains for microorganisms, and five of these had glucose concentrations below 14 mg/dl. Clinical chorioamnionitis occurred in 2.4% (4/168) of patients. Three patients had positive amniotic fluid cultures for microorganisms and amniotic fluid glucose concentrations below 14 mg/dl, while the remaining patient had a negative culture of the amniotic fluid. This woman had lower abdominal pain, a temperature of 106° F, and leukocytosis (a white blood cell count of 22,300/mm') and underwent a cesarean section because of fetal distress. Placental examination did not show any evidence of inflammation, and the newborn had no evidence of sepsis. The amniotic fluid glucose concentration was 28 mg/dl. The only neonate with sepsis was born to a woman with a positive amniotic fluid culture. The amniotic fluid glucose concentration in this case was 12 mg/dl. Of the 168 patients included in the study, 149 received a trial of tocolysis (15 patients had a positive Gram stain, and four additional patients had tocolysis discontinued because of maternal or fetal indications). A low glucose concentration «14 mg/dl) had a
Volume 163 Number 3
Amniotic fluid glucose and intraamniotic infection
100
31.5
30
21
971
67
14
80 Q)
> ..;::;
·iii
60
0
a.. Q)
::J ..... I-
40
11
O~-----L----~--
o
20
40
____
~
60
____
~
____
80
~
100
False Positive Fig. 2. A receiver operator characteristic curve analysis of the value of the amniotic fluid glucose concentration in diagnosis of a positive amniotic fluid culture (area under the curve 0.93, SE 0.03; z 14.1; P < 0.001). The numbers in the figure represent amniotic fluid glucose concentrations in milligrams per deciliter.
sensltlVlty of 37.9% (11129), a specificity of 88.3% (106/120), a positive predictive value of 44% (11125), and a negative predictive value of 85.4% (1061124) for the prediction of failure to respond to tocolytic therapy. Comment
The data clearly demonstrate that microbial invasion of the amniotic cavity in women with preterm labor with intact membranes is associated with reduced glucose concentration in amniotic fluid (Fig. 1). Similar observations have been made in other body fluids (i.e., cerebrospinal fluid) in the setting of bacterial infection. 5.' Why is microbial invasion of the amniotic cavity associated with a reduced glucose concentration? In cerebrospinal fluid, three mechanisms have been invoked to explain a similar phenomenon; (1) metabolism of glucose by microorganisms, (2) consumption of glucose by activated neutrophils, and (3) changes in glucose transport across the hematoencephalic barrier. The most likely explanation for the low glucose concentration in cerebrospinal fluid is metabolism of glucose by white blood cells and bacteria. The evidence suggesting that glucose catabolism by activated neutrophils is the primary mechanism involved includes the following: (1) Neutrophils obtained from septic cerebrospinal fluid consume glucose at a rate that is two to three times higher than that of neutrophils from noninfected patients 10; (2) incubation of cerebrospinal fluid with a
combination of leukocytes and bacteria results in a lowered glucose concentration, while incubation by either one alone does not have the same effect ' !; (3) irradiated leukopenic dogs do not respond to pneumococcal meningitis with pleocytosis or with a reduction in cerebrospinal fluid glucose.!2 In contrast to bacterial infections, viral infections are not associated with a reduction in glucose concentrations. In viral infections the mononuclear cells are predominant. These cells are known to consume glucose at a lesser rate than neutrophils. An important question that remains to be answered is whether these aforementioned mechanisms are operative in amniotic fluid. The observation that six women who had negative amniotic fluid cultures for microorganisms and who also had acute inflammatory lesions of the placenta (see Table I) had reduced amniotic fluid glucose concentrations indicates that this finding could occur in the absence of viable microorganisms. Therefore, an inflammatory host response is probably associated with reduced glucose concentration in amniotic fluid. In our study amniotic fluid glucose determinations, with a cutoff $14 mg/dl, had a sensitivity of 86.9% in the detection of intraamniotic infection while the sensitivity of the Gram stain was 65.2%. This difference fell short of reaching statistical significance with our sample size (23 patients with intraamniotic infections). The false-positive rate (100 - Specificity) of the Gram stain was lower than that of the amniotic fluid glucose concentration (0.7% vs 8.3%; P < 0.05). However, the
972
Romero et al.
September 1990 Am J Obstet Gynecol
Table I. Clinical information and amniotic fluid concentrations in patients with false-positive results
Patient No.
Amniotic fluid glucose concentration (mgldl)
Gestational age at amniocentesis (wk)
Gestational age at delivery (wk)
Tocolysis response
Antibiotics before amniocentesis
Histologic chorioamnionitis*
1 2
3 4
33 29
33 29
No No
No Yes
Yes Yes
3 4
4 8
34 33.5
35 39
No Yes
Yes Yes
N/A N/A
5 6 7 8
II II II II
31 25 34 27
32 25 40 27
No No Yes No
No Yes No Yes
Yes Yes
N/A
9
II
30
30
No
No
No
10 11
12 12
33 34
33 37
No Yes
No No
N/A
12
14
35
35
No
Yes
N/A
Yes
Yes
Comments
Neonate with congenital syphilis Maternal fever Chlamydia trachomatis endocervicitis Neonatal sepsis Neonatal viral meningitis Foul-smelling amniotic fluid at delivery Massive polyhydramnios Urinary tract infection
*N I A, Placental histologic data were not available.
Table II. Clinical data, Gram stain, and amniotic fluid concentrations in patients with positive amniotic fluid cultures
Patient No.
Gestational age at amniocentesis (wk)
Amniotic fluid glucose concentration (mgldl)
Gram stain*
Culture
1 2
25 25
2 2
+ +
Candida tropicalis
3 4
28.5 32
4 4
+ +
5
26
8
+
6
24
10
+
Fusobacterium nucleatum Mycoplasma hominis Ureaplasma urealyticum Haemophilus influenzae Capnocytophaga species Clostridium species
7 8 9 10 11 12 13 14 15 16 17 18 19
25.5 29 23 23 24 25 28.5 29 29.5 31 33 23 24
20 21 22 23
34 31.5 24 34
*All
10 10
+
II
+ + + +
II
+
11 11 11 11
11 11 11 12 12 13 17 21 30
+ + +
Mixed aerobic flora of 3 varieties
Mixed anaerobic flora
Mycoplasma hominis Streptococcus agalactiae Ureaplasma urealyticum Fusobacterium nucleatum Gardnerella vaginalis Streptococcus agalactiae Fusobacterium nucleatum Ureaplasma urealyticum Gardnerella vaginalis Staphylococcus aureus Ureaplasma urealyticum Fusobacterium species Mycoplasma hominis Gardnerella vaginalis Peptostreptococcus species Peptostreptococcus species Ureaplasma urealyticum Ureaplasma urealyticum Gardnerella vaginalis
Colony count (CFUlmm3 )
Clinical chorioamnionitis
30,000
No No
> 100,000
Yes Yes
> 100,000 > 100,000 > 100,000 > 100,000 > 100,000 >100,000 50,000 > 100,000 1,000 cfu > 100,000 20,000 > 100,000 1,000 > 100,000 >100,000
Yes No No No No No No No No No No No No No No No No No No
patients with positive Gram stains received antibiotics.
false-positive rate of amniotic fluid glucose concentration deserves comment. Seventy-five percent (9112) of patients with low amniotic fluid glucose concentrations went on to deliver a preterm neonate despite intrave-
nous tocolysis. Six of the seven placentas examined from women delivered prematurely had histologic evidence of chorioamnionitis. Thus an infectious process that was not detected by microbiologic examinations of
Volume 163 Number 3
Amniotic fluid glucose and intraamniotic infection
973
168 Patients
/('\ Glucose < 14 mg/dl
(+) Gram Stain 15 (46.8%)
/\
(+) Culture 15
(-) Culture
o
Glucose> 14 mg/dl
/~(B'\
(-) Gram Stain 17 (53.2%)
/\
(+) Culture
5
(-) Culture 12
(+) Gram Stain 1 (0.7%)
/\
(+) Culture
o
(-) Culture 1
(-) Gram Stain 135 (99.3%)
/\
(+) Culture
3
H
Culture 132
Fig. 3. Flow diagram of the results of amniotic fluid glucose determination, Gram stain, and culture of amniotic fluid.
amniotic fluid might have been responsible for the acute inflammatory lesions in the placenta. The failure to recover microorganisms may have resulted from either antibiotic administration before amniocentesis (in five cases) or failure of growth of fastidious microorganisms. The possiblity exists that all nine patients with low amniotic fluid glucose concentrations and negative amniotic fluid cultures had an infectious process that escaped detection with current microbiologic techniques used in our laboratory. There were three patients with low amniotic fluid glucose concentrations (8, 11, and 12 mg/dl, respectively) who were delivered at term. Two patients had clinical conditions that might explain the low concentration of glucose in amniotic fluid. The first patient had polyhydramnios, and this may have had a dilutional effect on amniotic fluid glucose concentration. The second patient had Chlamydia trachomatis endocervicitis and received antibiotics before amniocentesis. Antimicrobial therapy may have precluded successful isolation of a pathogen from amniotic fluid. We could not find any explanation for the reduced amniotic fluid glucose concentration observed in the third patient. Despite the high sensitivity of amniotic fluid glucose determinations, a problem uncovered by this study is a false-positive rate of 8.3%. The seriousness of a falsepositive result is dependent on the action taken after an abnormal test result. If the course of action is to perform a Gram stain of amniotic fluid, then a falsepositive result is not clinically problematic. However, if the course of action is to deliver a pre term neonate believed to be infected, then there is the potential for serious consequences.
To examine the role of amniotic fluid glucose determinations in the rapid diagnosis of intraamniotic infection, we explored two diagnostic schemes: (1) simultaneous utilization with the Gram stain and (2) sequential use of both tests. Our data do not support the simultaneous use of both tests because the improvement in sensitivity is accompanied by a significant increase in the false-positive rate (from 0.7% to 9%). The other alternative is to use the tests in sequence. The flow diagram provided in Fig. 3 displays the results of using a sequential diagnostic testing modality that incorporates amniotic fluid glucose determinations as a first-line test. A low glucose determination would have identified 15 of the 16 cases with positive Gram stains. Furthermore, a high glucose concentration has a high negative predictive value (97.8%). Is there a role for this scheme of sequential testing in clinical practice? The Gram stain examination is the most widely used method for the rapid detection of bacteria in body fluids as it is rapid and inexpensive. However, Gram stain examination of amniotic fluid requires considerable experience. In a study conducted at our institution, considerable disagreement was found between results obtained by obstetric medical staff and trained laboratory personnel. Given the seriousness of the intervention implemented after identification of bacteria in a Gram stain, we concluded that Gram stain examinations of amniotic fluid should be read only by experienced laboratory personnel (Romero R, unpublished observations). In most medical centers trained microbiology technicians are not available around-the-clock. Amniotic fluid glucose determination can assist the clinician in
974
Romero et al.
deciding whether a Gram stain needs to be performed and analyzed immediately or whether it may be postponed until experienced personnel are available. In this context, amniotic fluid glucose determination has several advantages. The test is simple to perform and the required equipment is generally available in most obstetric units. Glucose determination is rapid, inexpensive, and does not require sophisticated interpretation by trained personnel. The rapid diagnosis of infection in clinical medicine is a challenging problem. The approach used in other compartments, such as cerebrospinal fluid, consists of a combination of tests including the Gram stain, white blood cell and differential counts, and glucose and protein concentrations. A similar approach could be used for the detection of intraamniotic infection. Amniotic fluid glucose determinations may serve as an important component of this battery of tests. Further studies are required to explore the advantages and disadvantages of this approach. Other issues that need to be addressed are the effects of changes of amniotic fluid volume and maternal glucose intolerance on the clinical value of this test.
September 1990 Am J Obstet Gynecol
3.
4. 5. 6.
7. 8.
9. 10. 11.
We gratefully acknowledge the assistance of Carol A. Brekus, MSN, CNM, and Valerie W. Kraemer, MFA. REFERENCES 1. Romero R, Mazor M. Infection and preterm labor. Clin Obstet Gynecol 1988;31 :553-84. 2. Romero R, Sirtori M, Oyarzun E, et al. Infection and labor. V. Prevalence, microbiology, and clinical signifi-
12.
cance of intraamniotic infection in women with preterm labor and intact membranes. AM J OBSTET GYNECOL 1989; 161 :817-24. Romero R, Emamian M, Quintero R, et al. The value and limitations of the Gram stain examination in the diagnosis of intraamniotic infection. AM J OBSTET GYNECOL 1988;159:114-9. Romero R, Emamian M, Quintero R, et al. Diagnosis of intra-amniotic infection: the acridine orange stain. Am J Perinatol 1989;6:41-5. Overturf GD. Infections of the central nervous system. In: Hoeprich PD, Jordan MC, eds. Infectious diseases. 4th ed. Philadelphia: JB Lippincott, 1989: 1114-32. Brody JS. Diseases of the pleura, mediastinum, diaphragm and chest wall. In: Wyngaarden JB, Smith LH, J r, eds. Cecil textbook of medicine. 17th ed. Philadelphia: WB Saunders, 1985:447-54. Parker RH. Skeletal infections. In: Hoeprich PD, Jordan MC, eds. Infectious diseases. 4th ed. Philadelphia: JB Lippincott, 1989: 1376-82. Gibbs RS, Blanco JD, St Clair PJ, Castaneda YS. Quantitative bacteriology of amniotic fluid from women with clinical intraamniotic infection at term. J Infect Dis 1982; 145: 1-8. Brooke RD. Alterations in the glucose transport mechanism in patients with complications of bacterial meningitis. Pediatrics 1964;34:491-501. Betz G, Mauer AM. Glucose consumption by polymorphonuclear leukocytes in the cerebrospinal fluid of patients with bacterial meningitis. J Pediatr 1967;70:767-71. Petersdorf RG, Swarner DR, Garcia M. Synergistic action of pneumococci and leukocytes in lowering cerebrospinal fluid glucose. Proc Soc Exp Bioi Med 1960; 103: 380-2. Petersdorf RG, Garcia M, Swarner DR. Mechanism of hypoglycorrhachia in experimental pneumococcal meningitis. Proc Soc Exp Bioi Med 1959;102:669-72.
Bound volumes available to subscribers Bound volumes of the AMERICAN JOURNAL OF OBSTETRICS AND GYNECOLOGY are available to subscribers (only) for the 1990 issues from the Publisher, at a cost of $56.00 ($79.00 international) for Vol. 162 (january-June) and Vol. 163 (july-December). Shipping charges are included. Each bound volume contains a subject and author index and all advertising is removed. Copies are shipped within 60 days after publication of the last issue in the volume. The binding is durable buckram with the JOURNAL name, volume number, and year stamped in gold on the spine. Payment must accompany all orders. Contact Mosby- Year Book, Inc., Circulation Department, 11830 Westline Industrial Drive, St. Louis, Missouri 63146-3318, USA; phone (800) 325-4177, ext. 7351. Subscriptions must be in force to qualify. Bound volumes are not available in place of a regular JOURNAL subscription.
Key words: Glucose, intraamniotic infection, chorioamnionitis, preterm labor, prematurity, Gram stain, amniotic fluid
Strong evidence supports an association of intraamniotic infection with preterm labor and delivery. 1 Patients in preterm labor with intact membranes and positive amniotic fluid cultures are more likely to have clinical chorioamnionitis and fail to respond to tocolysis. I. 2 Furthermore, neonates born to women with positive amniotic fluid cultures are at an increased risk for infectious complications! The early diagnosis of intraamniotic infection is a desirable clinical goal. However, clinical signs of infection are insensitive and will detect only 12% of patients with positive amniotic fluid cultures in women with preterm labor and intact membranes! While amniotic fluid cultures are widely used to assess the microbiologic state of the amniotic cavity, results are not immediately available and may take up to 5 days. Rapid methods for the detection of infection, such as the Gram stain and the From the Department of Obstetrics and Gynecology, Yale University School of Medicine. Supported by a grant from the Walter Scott Foundation for Medical Research. Dr. Romero is the recipient of a Physician Scientist Award from the National Institute of Child Health and Human Development. Dr. Diamond is a Clinical Associate Physician of the General Clinical Research Center of the NIH at Yale University School of Medicine. Presented at the Tenth Annual Meeting of the Society of Perinatal Obstetricians, January 23-27, 1990. Reprint requests: Roberto Romero, MD, Yale University School of Medicine, Department of Obstetrics and Gynecology, 333 Cedar St., P.O. Box 3333, New Haven, CT 06510. 616122595
968
acridine orange stain of amniotic fluid, are insensitive, detecting only 50% to 60% of all intraamniotic infections.'" Other tests, such as gas-liquid chromatographic analysis of organic acids in amniotic fluid, require expensive equipment and trained personnel and are not available in most medical centers. Thus there is a need for a sensitive, simple, and rapid test for the detection of intraamniotic infection. A decreased glucose concentration in several body fluids (i.e., cerebrospinal fluid, pleural fluid, and synovial fluid) has been used for the diagnosis of infection.'-7 The purpose of this study was to determine whether amniotic fluid glucose concentration could be of value in the diagnosis of intraamniotic infection. The value of amniotic fluid glucose concentration in the diagnosis of intraamniotic infection was first suggested by Dr. Brian Kirshon (personal communication, 1989).
Material and methods Patient population. The study population consisted of 168 consecutive patients who were admitted to YaleNew Haven Hospital with the diagnosis of preterm lahor and intact membranes from March 3, 1988, to May 26, 1989, and who underwent amniocenteses for evaluation of the microbiologic status of the amniotic fluid. Patients with known abnormal glucose tolerance test results or with overt diabetes mellitus were excluded. Preterm labor was defined as the presence of
Volume 163 Number 3
regular uterine contractions with a frequency of at least two every 10 minutes. Rupture of membranes was excluded by testing for pooling, Nitrazine paper reaction, and ferning. Ritodrine, terbutaline, or magnesium sulfate was administered intravenously for tocolysis after amniocentesis. Failure of tocolysis was diagnosed when cervical dilatation progressed beyond 5 cm or delivery occurred. Retrieval of amniotic fluid. In all cases amniotic fluid was retrieved by transabdominal amniocentesis, which was performed under ultrasonographic guidance. Microbiologic culture technique. Amniotic fluid was transported to the laboratory in a capped plastic syringe immediately after collection. These conditions precluded air contact with the specimen. Plating occurred within 30 minutes of collection in all cases. Amniotic fluid was cultured for aerobic and anaerobic bacteria, as well as for Mycoplasma species (Mycoplasma hominis and Ureaplasma urealyticum), according to methods previously described" Gram stain examination. Gram stain examinations were performed by using commercial reagents (crystal violet, saffranin, and Gram's iodine [Difco Laboratories, Detroit]) under standard conditions. Stained slides were then examined by trained technologists, and the presence or absence of microorganisms was noted. The results of the Gram stain examinations were shared with the clinicians. Patients with negative Gram stain examinations for bacteria were given tocolysis. Patients with positive Gram stain examination of amniotic fluid for bacteria were given parenteral antibiotics and allowed to labor. A positive Gram stain for bacteria was an indication for discontinuing tocolysis. Amniotic fluid glucose determinations. Amniotic fluid was centrifuged at 700 g for 10 minutes at 4° C and stored at -70 0 C until assayed. Glucose analysis was performed by the glucose oxidase method with a Beckman glucose analyzer. Samples were run in duplicate. Results were not available for patient management. The coefficient of variation was 5%. Criteria for diagnosis of chorioamnionitis and neonatal sepsis. Intraamniotic infection was defined as the presence of a positive amniotic fluid culture.' Clinical chorioamnionitis was defined according to the criteria proposed by Gibbs et al. 8 Neonatal sepsis was diagnosed by the presence of a positive culture of blood, urine, or cerebrospinal fluid. Statistical analysis. A Mann-Whitney test was used to compare amniotic fluid glucose concentrations between infected and noninfected patients. A receiver operator characteristic curve analysis was conducted with the program True Epistat (Epistat Services, Richardson, Tex.). Diagnostic indices (sensitivity, specificity, and positive and negative predictive values) for the detection of a positive amniotic fluid culture were calculated for both the Gram stain and amniotic fluid glucose.
Amniotic fluid glucose and intraamniotic infection
969
Results
The criteria for inclusion into this study were met by 168 patients, 54.1 % (91/168) of whom went on to have a preterm delivery. The remainder were delivered at term. The prevalence of positive amniotic fluid cultures was 13.6% (23/168). Fig. I displays amniotic fluid glucose concentrations according to culture results. Patients with positive amniotic fluid cultures for microorganisms had significantly lower median amniotic fluid glucose concentrations than patients with negative amniotic fluid cultures (median 11 mgldl, range 2 to 30 mgldl vs median 28 mgldl, range 3 to 74, respectively; p < 0.001, MannWhitney test). To describe the relationship between the sensitivity (true-positive rate) and the false-positive rate for different values of amniotic fluid glucose in the detection of a positive amniotic fluid culture, a receiver operator characteristic curve was constructed. Fig. 2 displays the receiver operator characteristic curve (area under the curve 0.93; SE 0.03; z 14.1; P < 0.001). The curve is well above the 45-degree line, indicating that an abnormal test result (reduced amniotic fluid glucose concentration) increases the probability of a positive amniotic fluid culture. As with any other diagnostic test, there is a trade-off between the sensitivity (true-positive rate) and the false-positive rate (100 - Specificity). We chose to present the diagnostic indices using a cutoff at 14 mgl dl, as this value seems to represent a reasonable compromise between a true-positive and a falsepositive rate. A glucose level below 14 mg 1dl had a sensitivity of 86.9% (20/23), a specificity of 91. 7% (1331145), a positive predictive value of 62.5% (20/32), and a negative predictive value of 97.8% (1331136). On the other hand, the diagnostic indices for the Gram stain examination of amniotic fluid were: sensitivity of 65.2% (15/23), specificity of 99.3% (144/145), positive predictive value of93.7% (15116), and negative predictive value of 94.7% (1441152). When the Gram stain and glucose level (below 14 mgl dl) were used in combination (either test result considered abnormal), the diagnostic indices were sensitivity of 86.9% (20/23), specificity of 91% (1321145), positive predictive value of60.6% (20/33), and negative predictive value of 97.8% (1321l35). There were 131 patients with negative Gram stains and glucose concentrations above 14 mgl dl. Three had positive amniotic fluid cultures (two with Ureaplasma urealyticum and one with Gardnerella vaginalis). All three patients failed to respond to intravenous tocolysis and were delivered of preterm neonates. Table I displays the clinical data of the 12 patients with amniotic fluid glucose concentrations of < 14 mgldl and negative amniotic fluid cultures for bacteria and Mycoplasma species. Although all of these patients
970
Romero et al.
September 1990 Am J Obstet Gynecol
80
• 70
...--.
--E
• • ••
60
Ol
.......Q)
50
•
(Jl
0
•
U
::::l
a "'0
40
'5
u::: u
'';::::;
0
30
•
20
• • •
c
E
«
-----=-..,.111 -----•
=0
10 0
--.....
""
•
Infection N = 23
• e::-
.= • •
•
•
No Infection
N = 145
Fig. 1. Amniotic fluid glucose concentrations in the study population. The median amniotic fluid glucose concentration in patients with intraamniotic infection was significantly lower than in patients with negative amniotic fluid cultures (median 11 mg/dl, range 2 to 30 mg/dl vs median 28 mg/dl, range 3 to 74, respectively; p < 0.0001).
received parenteral tocolysis, 75% (9112) went on to have a preterm delivery. Five of the nine patients with negative amniotic fluid cultures had been treated with antibiotics before undergoing amniocentesis, and one of the neonates subsequently had documented neonatal sepsis. In addition, one neonate was diagnosed to have congenital syphilis, but treponemas were not recovered from the amniotic fluid. Two of the three remaining patients had no clinical evidence of infection (maternal or neonatal), but their placentas showed acute inflammatory lesions (histopathologic chorioamnionitis). Three patients with low glucose concentrations and negative amniotic fluid cultures were delivered at term. One patient had polyhydramnios of an unknown cause. The second patient had fever (101° F) and was treated with erythromycin because of a positive endocervical culture for Chlamydia trachQmatis. The remaining patient had no evidence of an infectious process. Table II displays the clinical course and microbiologic characteristics of patients with positive amniotic fluid cultures. Twenty-one percent (5/23) of patients had polymicrobial infections. Eight patients had negative
Gram stains for microorganisms, and five of these had glucose concentrations below 14 mg/dl. Clinical chorioamnionitis occurred in 2.4% (4/168) of patients. Three patients had positive amniotic fluid cultures for microorganisms and amniotic fluid glucose concentrations below 14 mg/dl, while the remaining patient had a negative culture of the amniotic fluid. This woman had lower abdominal pain, a temperature of 106° F, and leukocytosis (a white blood cell count of 22,300/mm') and underwent a cesarean section because of fetal distress. Placental examination did not show any evidence of inflammation, and the newborn had no evidence of sepsis. The amniotic fluid glucose concentration was 28 mg/dl. The only neonate with sepsis was born to a woman with a positive amniotic fluid culture. The amniotic fluid glucose concentration in this case was 12 mg/dl. Of the 168 patients included in the study, 149 received a trial of tocolysis (15 patients had a positive Gram stain, and four additional patients had tocolysis discontinued because of maternal or fetal indications). A low glucose concentration «14 mg/dl) had a
Volume 163 Number 3
Amniotic fluid glucose and intraamniotic infection
100
31.5
30
21
971
67
14
80 Q)
> ..;::;
·iii
60
0
a.. Q)
::J ..... I-
40
11
O~-----L----~--
o
20
40
____
~
60
____
~
____
80
~
100
False Positive Fig. 2. A receiver operator characteristic curve analysis of the value of the amniotic fluid glucose concentration in diagnosis of a positive amniotic fluid culture (area under the curve 0.93, SE 0.03; z 14.1; P < 0.001). The numbers in the figure represent amniotic fluid glucose concentrations in milligrams per deciliter.
sensltlVlty of 37.9% (11129), a specificity of 88.3% (106/120), a positive predictive value of 44% (11125), and a negative predictive value of 85.4% (1061124) for the prediction of failure to respond to tocolytic therapy. Comment
The data clearly demonstrate that microbial invasion of the amniotic cavity in women with preterm labor with intact membranes is associated with reduced glucose concentration in amniotic fluid (Fig. 1). Similar observations have been made in other body fluids (i.e., cerebrospinal fluid) in the setting of bacterial infection. 5.' Why is microbial invasion of the amniotic cavity associated with a reduced glucose concentration? In cerebrospinal fluid, three mechanisms have been invoked to explain a similar phenomenon; (1) metabolism of glucose by microorganisms, (2) consumption of glucose by activated neutrophils, and (3) changes in glucose transport across the hematoencephalic barrier. The most likely explanation for the low glucose concentration in cerebrospinal fluid is metabolism of glucose by white blood cells and bacteria. The evidence suggesting that glucose catabolism by activated neutrophils is the primary mechanism involved includes the following: (1) Neutrophils obtained from septic cerebrospinal fluid consume glucose at a rate that is two to three times higher than that of neutrophils from noninfected patients 10; (2) incubation of cerebrospinal fluid with a
combination of leukocytes and bacteria results in a lowered glucose concentration, while incubation by either one alone does not have the same effect ' !; (3) irradiated leukopenic dogs do not respond to pneumococcal meningitis with pleocytosis or with a reduction in cerebrospinal fluid glucose.!2 In contrast to bacterial infections, viral infections are not associated with a reduction in glucose concentrations. In viral infections the mononuclear cells are predominant. These cells are known to consume glucose at a lesser rate than neutrophils. An important question that remains to be answered is whether these aforementioned mechanisms are operative in amniotic fluid. The observation that six women who had negative amniotic fluid cultures for microorganisms and who also had acute inflammatory lesions of the placenta (see Table I) had reduced amniotic fluid glucose concentrations indicates that this finding could occur in the absence of viable microorganisms. Therefore, an inflammatory host response is probably associated with reduced glucose concentration in amniotic fluid. In our study amniotic fluid glucose determinations, with a cutoff $14 mg/dl, had a sensitivity of 86.9% in the detection of intraamniotic infection while the sensitivity of the Gram stain was 65.2%. This difference fell short of reaching statistical significance with our sample size (23 patients with intraamniotic infections). The false-positive rate (100 - Specificity) of the Gram stain was lower than that of the amniotic fluid glucose concentration (0.7% vs 8.3%; P < 0.05). However, the
972
Romero et al.
September 1990 Am J Obstet Gynecol
Table I. Clinical information and amniotic fluid concentrations in patients with false-positive results
Patient No.
Amniotic fluid glucose concentration (mgldl)
Gestational age at amniocentesis (wk)
Gestational age at delivery (wk)
Tocolysis response
Antibiotics before amniocentesis
Histologic chorioamnionitis*
1 2
3 4
33 29
33 29
No No
No Yes
Yes Yes
3 4
4 8
34 33.5
35 39
No Yes
Yes Yes
N/A N/A
5 6 7 8
II II II II
31 25 34 27
32 25 40 27
No No Yes No
No Yes No Yes
Yes Yes
N/A
9
II
30
30
No
No
No
10 11
12 12
33 34
33 37
No Yes
No No
N/A
12
14
35
35
No
Yes
N/A
Yes
Yes
Comments
Neonate with congenital syphilis Maternal fever Chlamydia trachomatis endocervicitis Neonatal sepsis Neonatal viral meningitis Foul-smelling amniotic fluid at delivery Massive polyhydramnios Urinary tract infection
*N I A, Placental histologic data were not available.
Table II. Clinical data, Gram stain, and amniotic fluid concentrations in patients with positive amniotic fluid cultures
Patient No.
Gestational age at amniocentesis (wk)
Amniotic fluid glucose concentration (mgldl)
Gram stain*
Culture
1 2
25 25
2 2
+ +
Candida tropicalis
3 4
28.5 32
4 4
+ +
5
26
8
+
6
24
10
+
Fusobacterium nucleatum Mycoplasma hominis Ureaplasma urealyticum Haemophilus influenzae Capnocytophaga species Clostridium species
7 8 9 10 11 12 13 14 15 16 17 18 19
25.5 29 23 23 24 25 28.5 29 29.5 31 33 23 24
20 21 22 23
34 31.5 24 34
*All
10 10
+
II
+ + + +
II
+
11 11 11 11
11 11 11 12 12 13 17 21 30
+ + +
Mixed aerobic flora of 3 varieties
Mixed anaerobic flora
Mycoplasma hominis Streptococcus agalactiae Ureaplasma urealyticum Fusobacterium nucleatum Gardnerella vaginalis Streptococcus agalactiae Fusobacterium nucleatum Ureaplasma urealyticum Gardnerella vaginalis Staphylococcus aureus Ureaplasma urealyticum Fusobacterium species Mycoplasma hominis Gardnerella vaginalis Peptostreptococcus species Peptostreptococcus species Ureaplasma urealyticum Ureaplasma urealyticum Gardnerella vaginalis
Colony count (CFUlmm3 )
Clinical chorioamnionitis
30,000
No No
> 100,000
Yes Yes
> 100,000 > 100,000 > 100,000 > 100,000 > 100,000 >100,000 50,000 > 100,000 1,000 cfu > 100,000 20,000 > 100,000 1,000 > 100,000 >100,000
Yes No No No No No No No No No No No No No No No No No No
patients with positive Gram stains received antibiotics.
false-positive rate of amniotic fluid glucose concentration deserves comment. Seventy-five percent (9112) of patients with low amniotic fluid glucose concentrations went on to deliver a preterm neonate despite intrave-
nous tocolysis. Six of the seven placentas examined from women delivered prematurely had histologic evidence of chorioamnionitis. Thus an infectious process that was not detected by microbiologic examinations of
Volume 163 Number 3
Amniotic fluid glucose and intraamniotic infection
973
168 Patients
/('\ Glucose < 14 mg/dl
(+) Gram Stain 15 (46.8%)
/\
(+) Culture 15
(-) Culture
o
Glucose> 14 mg/dl
/~(B'\
(-) Gram Stain 17 (53.2%)
/\
(+) Culture
5
(-) Culture 12
(+) Gram Stain 1 (0.7%)
/\
(+) Culture
o
(-) Culture 1
(-) Gram Stain 135 (99.3%)
/\
(+) Culture
3
H
Culture 132
Fig. 3. Flow diagram of the results of amniotic fluid glucose determination, Gram stain, and culture of amniotic fluid.
amniotic fluid might have been responsible for the acute inflammatory lesions in the placenta. The failure to recover microorganisms may have resulted from either antibiotic administration before amniocentesis (in five cases) or failure of growth of fastidious microorganisms. The possiblity exists that all nine patients with low amniotic fluid glucose concentrations and negative amniotic fluid cultures had an infectious process that escaped detection with current microbiologic techniques used in our laboratory. There were three patients with low amniotic fluid glucose concentrations (8, 11, and 12 mg/dl, respectively) who were delivered at term. Two patients had clinical conditions that might explain the low concentration of glucose in amniotic fluid. The first patient had polyhydramnios, and this may have had a dilutional effect on amniotic fluid glucose concentration. The second patient had Chlamydia trachomatis endocervicitis and received antibiotics before amniocentesis. Antimicrobial therapy may have precluded successful isolation of a pathogen from amniotic fluid. We could not find any explanation for the reduced amniotic fluid glucose concentration observed in the third patient. Despite the high sensitivity of amniotic fluid glucose determinations, a problem uncovered by this study is a false-positive rate of 8.3%. The seriousness of a falsepositive result is dependent on the action taken after an abnormal test result. If the course of action is to perform a Gram stain of amniotic fluid, then a falsepositive result is not clinically problematic. However, if the course of action is to deliver a pre term neonate believed to be infected, then there is the potential for serious consequences.
To examine the role of amniotic fluid glucose determinations in the rapid diagnosis of intraamniotic infection, we explored two diagnostic schemes: (1) simultaneous utilization with the Gram stain and (2) sequential use of both tests. Our data do not support the simultaneous use of both tests because the improvement in sensitivity is accompanied by a significant increase in the false-positive rate (from 0.7% to 9%). The other alternative is to use the tests in sequence. The flow diagram provided in Fig. 3 displays the results of using a sequential diagnostic testing modality that incorporates amniotic fluid glucose determinations as a first-line test. A low glucose determination would have identified 15 of the 16 cases with positive Gram stains. Furthermore, a high glucose concentration has a high negative predictive value (97.8%). Is there a role for this scheme of sequential testing in clinical practice? The Gram stain examination is the most widely used method for the rapid detection of bacteria in body fluids as it is rapid and inexpensive. However, Gram stain examination of amniotic fluid requires considerable experience. In a study conducted at our institution, considerable disagreement was found between results obtained by obstetric medical staff and trained laboratory personnel. Given the seriousness of the intervention implemented after identification of bacteria in a Gram stain, we concluded that Gram stain examinations of amniotic fluid should be read only by experienced laboratory personnel (Romero R, unpublished observations). In most medical centers trained microbiology technicians are not available around-the-clock. Amniotic fluid glucose determination can assist the clinician in
974
Romero et al.
deciding whether a Gram stain needs to be performed and analyzed immediately or whether it may be postponed until experienced personnel are available. In this context, amniotic fluid glucose determination has several advantages. The test is simple to perform and the required equipment is generally available in most obstetric units. Glucose determination is rapid, inexpensive, and does not require sophisticated interpretation by trained personnel. The rapid diagnosis of infection in clinical medicine is a challenging problem. The approach used in other compartments, such as cerebrospinal fluid, consists of a combination of tests including the Gram stain, white blood cell and differential counts, and glucose and protein concentrations. A similar approach could be used for the detection of intraamniotic infection. Amniotic fluid glucose determinations may serve as an important component of this battery of tests. Further studies are required to explore the advantages and disadvantages of this approach. Other issues that need to be addressed are the effects of changes of amniotic fluid volume and maternal glucose intolerance on the clinical value of this test.
September 1990 Am J Obstet Gynecol
3.
4. 5. 6.
7. 8.
9. 10. 11.
We gratefully acknowledge the assistance of Carol A. Brekus, MSN, CNM, and Valerie W. Kraemer, MFA. REFERENCES 1. Romero R, Mazor M. Infection and preterm labor. Clin Obstet Gynecol 1988;31 :553-84. 2. Romero R, Sirtori M, Oyarzun E, et al. Infection and labor. V. Prevalence, microbiology, and clinical signifi-
12.
cance of intraamniotic infection in women with preterm labor and intact membranes. AM J OBSTET GYNECOL 1989; 161 :817-24. Romero R, Emamian M, Quintero R, et al. The value and limitations of the Gram stain examination in the diagnosis of intraamniotic infection. AM J OBSTET GYNECOL 1988;159:114-9. Romero R, Emamian M, Quintero R, et al. Diagnosis of intra-amniotic infection: the acridine orange stain. Am J Perinatol 1989;6:41-5. Overturf GD. Infections of the central nervous system. In: Hoeprich PD, Jordan MC, eds. Infectious diseases. 4th ed. Philadelphia: JB Lippincott, 1989: 1114-32. Brody JS. Diseases of the pleura, mediastinum, diaphragm and chest wall. In: Wyngaarden JB, Smith LH, J r, eds. Cecil textbook of medicine. 17th ed. Philadelphia: WB Saunders, 1985:447-54. Parker RH. Skeletal infections. In: Hoeprich PD, Jordan MC, eds. Infectious diseases. 4th ed. Philadelphia: JB Lippincott, 1989: 1376-82. Gibbs RS, Blanco JD, St Clair PJ, Castaneda YS. Quantitative bacteriology of amniotic fluid from women with clinical intraamniotic infection at term. J Infect Dis 1982; 145: 1-8. Brooke RD. Alterations in the glucose transport mechanism in patients with complications of bacterial meningitis. Pediatrics 1964;34:491-501. Betz G, Mauer AM. Glucose consumption by polymorphonuclear leukocytes in the cerebrospinal fluid of patients with bacterial meningitis. J Pediatr 1967;70:767-71. Petersdorf RG, Swarner DR, Garcia M. Synergistic action of pneumococci and leukocytes in lowering cerebrospinal fluid glucose. Proc Soc Exp Bioi Med 1960; 103: 380-2. Petersdorf RG, Garcia M, Swarner DR. Mechanism of hypoglycorrhachia in experimental pneumococcal meningitis. Proc Soc Exp Bioi Med 1959;102:669-72.
Bound volumes available to subscribers Bound volumes of the AMERICAN JOURNAL OF OBSTETRICS AND GYNECOLOGY are available to subscribers (only) for the 1990 issues from the Publisher, at a cost of $56.00 ($79.00 international) for Vol. 162 (january-June) and Vol. 163 (july-December). Shipping charges are included. Each bound volume contains a subject and author index and all advertising is removed. Copies are shipped within 60 days after publication of the last issue in the volume. The binding is durable buckram with the JOURNAL name, volume number, and year stamped in gold on the spine. Payment must accompany all orders. Contact Mosby- Year Book, Inc., Circulation Department, 11830 Westline Industrial Drive, St. Louis, Missouri 63146-3318, USA; phone (800) 325-4177, ext. 7351. Subscriptions must be in force to qualify. Bound volumes are not available in place of a regular JOURNAL subscription.
