Current Development

An analysis of antenatal tests to detect infection in preterm premature rupture of the membranes Arne Ohlsson, MD: and Elaine Wang, MSc, MD b

Toronto, Ontario, Canada The purpose of this study was to critically review published studies regarding sensitivity, specificity, and positive and negative predictive values of antenatal tests to diagnose chorioamnionitis or fetal-neonatal sepsis in preterm premature rupture of the membranes. A Medline Data-Base computer program search from 1980 to 1988 identified 39 studies, 23 of which were accepted after independent review with preset criteria. An ideal test to predict chorioamnionitis or neonatal sepsis was not found. The low success rate for amniocentesis and the need for repeat taps preclude the acceptance of tests on the basis of amniotic fluid. Single, small studies, the precision of which has never been tested, show good indices for repeatedly increased serum levels of C-reactive protein (>20 mg/L), a high level of C-reactive protein >40 mg/L, or a day-to-day coefficient of variation for C-reactive protein of >30% in the prediction of histologic or clinical chorioamnionitis. Ultrasonographic observation of fetal activity, if published study results are confirmed, may be of value to predict amniotic fluid bacterial colonization. (AM J OSSTET GVNECOL 1990;162:809-18.)

Key words: Fetal membranes, premature rupture, preterm, chorioamnionitis, amniocentesis, ultrasonographic diagnosis, fetus-infant, sepsis Pre term premature rupture of the membranes (PROM) occurs in approximately I % of all pregnancies. I It is associated with 30% to 40% of preterm births and is one of the most common underlying causes of preterm delivery and perinatal death. 2. 5 Current trends favor expectant management in an effort to maximize the benefits of increasing fetal maturity and avoid potential harm to the fetus and mother. Maternal-fetal infection and fetal distress as a result of cord complications or abruptio placentae6 7 may occur, or respiratory distress syndrome as a result of premature delivery despite expectant management may occur. Infection acquired in utero is a major threat to the fetus in pregnancies complicated by preterm PROM. In one neonatal unit, congenital infection was associated with a 50% mortality rate. s Deformations and pulmonary hypoplasia may develop if preterm PROM From the DIVisIOn of Neonatology, Unwerstty of Toronto ReglOnal Pennatal Umt, Women's College Hospital," and DIvISIOn of InfeetlOus DISeases, Hospttal for Sick Chzldren.' Received for publicatIOn June 15, 1989; revISed July 17, 1989; accepted August 7, 1989. Reprint requests: Arne Ohlsson. MD, DwislOn of Neonatology, Regzonal Pennatal Unit, Women's College HospItal, 76 GrenvIlle St., Toronto, Ontario, Canada M5S 1B2. 611115887

occurs before 28 weeks' gestation and is associated with marked oligohydramnios. 9 . 10 The optimal management of preterm PROM as it relates to fetal-neonatal outcome remains a controversial issue. II In a recently performed review, no evidence was found from published randomized controlled trials for the use of tocolysis or prophylactic antibiotics in preterm PROM. 12 The use of antenatal steroids in preterm PROM significantly decreases the risk of respiratory distress syndrome, significantly increases the risk of endometritis, and is associated with a trend toward an increase in neonatal infections. 12 Perinatal mortality rates in preterm PROM are not significantly reduced by the use of antenatal steroids. 12 The risks of both respiratory distress syndrome and infection decrease as the duration of membrane rupture increases, as the gestational age at rupture of the membranes increases, and as the gestational age at delivery increases, which supports an expectant approach to the management of preterm PROM. n However, the mortality rate has been observed to be 7 times higher from neonatal bacterial infection (36%) than from respiratory distress syndrome (5%).1:1 Chorioamnionitis is associated with a fourfold increase in neonatal mortality. 14 809

810 Ohlsson and Wang

Table I. Inclusion criteria 1. At least part of the study population was pregnant

women with preterm PROM. Preterm was defined as a gestation of 30% Positive latex (= 8 mg/L with laser method)

Gold standard

Histologic chorioamnionitis Histologic chorioamnionitis Histologic chorioamnionitis HisLOlogic chorioamnionitis Histologic chorioamnionitis

Sermtivity

(%)

Specificity

(%)

Positive predJctive value

Negative predJctJve value

(%)

(%)

Prevalence

88.5

96.2

95.8

89.3

0.50

80.0

68.4

40.0

92.9

0.21

67.0

81.0

90.0

50.0

0.63

50.0

81.0

79.0

53.1

0.59

9!.7

80.0

78.6

92.3

0.44

Histologic chorioamnionitis Histologic chorioamnionitis

36. 7

100.0

100.0

52.5

0.59

75.0

100.0

100.0

83.3

0.44

Clinical chorioamnionitis Clinical chorioamnionitis Clinical chorioamnionitis Clinical chorioamnionitis Clinical chorioamnionitis

55 .6

72.7

45.5

80.0

0 .29

85.7

81.8

42.9

97.3

0.14

82.0

55.0

36.0

91.0

0 .18

60.0

64.0

10.0

96.0

0.07

100.0

100.0

100.0

100.0

0.44

88.0

44.0

12.0

98.0

0.08

Clinical chorioamnionitis

bacterial colonization. However, others have noted the association between recovery of organisms from the chorioa mnion and histologic chorioamnionitis. 15 Neonatal sepsis as defined in Table I is an acceptable gold standard. However, congenital neonatal sepsis is usually a late manifestation of the infectious process associated with preterm PROM and carries a mortality rate of at least 50% despite appropriate postnatal treatment.s Ideally, one would therefore like to diagnose the fetus at risk of infection and intervene before fetal sepsis occurs. Possible neonatal sepsis was not accepted as a gold standard in this review as the definition used by Vintzileos et aJ.2" and others"" 24 is likely to result in over diagnosis. Others have proposed a stricter definition of the term probable infection to avoid too much overcall. 12. 16 The term infectious outcome as used by Vintzileos et al. 22 includes "possible neonatal sepsis" and therefore was rejected as a gold standard. A combination of maternal and neonatal "infectious morbidity" was not accepted as a gold standard.

Results

Thirty-nine studies were identified for review"'·5"; 23 were accepted for inclusion and 16 were excluded.* The excluded studies and the reasons for their exclusion are listed in Table II. Data from 22 of the 23 accepted studies are presented in Tables III to X. The results with regard to different tests are presented under separate headings. Tests performed on maternal blood White blood cell count, differential count, band count, and erythrocyte sedimentation rate. Daily total white blood cell count, differential count, and band counts are usually performed on mothers with preterm PROM. As outlined in Table 111,28.>0." although these tests generally have a high specificity, they have low sensitivity and positive predictive value for both histologic and clinical chorioamnionitis. In one study of 52 patients (26 normal and 26 with chorioamnionitis) an erythrocyte sedimentation rate >60 mm/hr had a high specificity and *26,29,31,36,37,42-44,46,47,49.50.52,54,55,57

Volume 162 Number 3

Tests to detect infection in preterm PROM

813

Table V. Bacterial colonization of amniotic fluid and clinical chorioamnionitis

Reference No.

Nu. of pallents

53

29

56

30

Telt

Gold standard

Gram stain or culture Culture

Clinical ,horioamnionitis Clinical chorioamnionitis

SellHtlvlty

SpecijzClly

POlltlve predictIVe value

(%)

(%)

(%)

(%)

Prevalence

66.7

85.0

66.7

85.0

0.31

85.7

86.9

66.7

95 .2

0.23

NegatIVe predictIVe value

Table VI. Analysis of amniotic fluid for bacteria by gram stain or microscopy for white blood cells

Reference No.

Nu. of patients

25 23 24* 24t 51 27

44 171 114 114 53 18

Test

Gold 5tandard

Gram stain Gram stain Gram stain Gram stain Gram stain Microscopy > 2 white blood cells per oil immersion field x 1000

Culture Culture Culture Culture Culture Culture

Sens ltlVlt~

Specifzrtt.~

POSItIVe predictIVe value

NegatIVe predictive value

(% )

(%)

( %)

(%)

Prevalence

53.9 36.2 44.8 51.7 60.0 80.0

83.3 94.7 97.6 96.5 94.7 92.3

82.4 77.8 86.7 83 .3 81.8 80.0

55.6 74.3 83.8 85.4 85.7 92.3

0.59 0.34 0.25 0.25 0.28 0.28

*Upspun sample. tSpun sample.

Table VII. Leukocycte esterase test

Reference No.

No. of patIents

Gold standard

25 23 27 23

44 171 20 171

Culture Culture Culture Clinical chorioamnionitis

Sermtwl(V

SpecifzClty

POSItive predictive value

(%)

( 'k )

( %)

N egatIVe predIctive value r%)

80.8 19.0 83 .3 8.3

100.0 86.7 85.7 84.3

100.0 42.3 71.4 3.9

78.3 67.6 92 .3 92.4

pOSItIve predictive value of 100%, but the senSItiVIty and negative predictive value for histologic chorioamnionitis was only 65% and 74 % , respectively.") C-reactive protein. C-reactive protein is a substance m serum that reacts with soma tic C-polysaccharide of pneumococci in vitro. It is a beta globulin that occurs frequently in mal"kedly increased amounts in sera of patients with inflammatory, neoplastic, or necrotizing processes. The laboratory test for the presence of Creactive protein constitutes a nonspecific test for the presence of inflammation or tissue injury.hu Six studies regarding C-reactive protein were included in this review (Table IV)."" JU 32· 3 ' As can be seen in Table IV, considerable variation in the sensitivity and specificity of these tests is reported by different autho1"S. This may be because of the use of different cut-off points for abnormal values (12.5 to 40 mg / L), different methods, or different gold standards. Fisk et al., \) in a blinded prospective study with a membrane roll-technique to diagnose histopathologic chorioamnionitis,bl reported 100% specificity and positive predictive value but low

Prevalence 0.59 0.34 0.30 0.07

sensitivity and negative predictive value for a cut-off level of 40 mg/L. Fisk et al." also found that C-reactive protein elevation often preceded delivery or clinical infection by several days. A total of 13 of 14 patients with two daily estimations > 20 mg/L and 12 of 13 with consecutive estimations >20 mg/L had histologic chorioamnionitis. In a study of 16 patients Romem and Artal"" found that a day-to-day coefficient of variation for C-reactive protein of > 30% was 100% accurate in the prediction of clinical chorioamnionitis during follow-up. Tests performed on amniotic fluid. The reported success rate for obtainment of fluid by amniocentesis varies from 45.0% to 96.9%. * Analysis of amnIOtIC flUId for bacteria by Gram stain orfor white blood cells by mIcroscopy in relation to positive cultureclinical chorlOarnnionitis. Gram stain has a moderately low sensitivity for clinical chorioamnionitis (Table V)."" As outlined in Table VI ,."'; " Gram stain has low *24,27,33,45,50,51,53,55.56

814 Ohlsson and Wang

Am

Marc h 1990

J Obstet Gynecol

Table VIII. Gas-liquid chro matography

Reference No .

No. of patients

58

38

59 21

47 25

Test

Abnormal gas liquid chromatography pattern Presence of acetate Abnormal gas liquid chromatography pattern

Sensitivity (%)

Specificity (%)

P ositive predictive value (%)

Culture

93 .3

91.3

87 .5

95 .5

0.40

Culture Culture

66.7 28.6

73.7 55.6

37.5 20.0

90.3 66.7

0.19 0.28

Gold standard

sensitivity but high specificity and positive predictive value with respect to results of amniotic fluid culture. Identification of more than two white blood cells per oil immersion field x 1000 was found to be more sensitive with a high negative predictive value in ruling out a positive culture and in maintaining high specificity and positive predictive value, according to one studyY The culture result was quite sensitive and specific for clinical chorioamnionitis (Table V) .'6 Leukocyte esterase test. The leukocyte esterase activity reflects the presence of white blood cells in a biologic fluid and the test was initially developed to detect pyuria. 62 Test strips that change color in the presence of leukocyte esterase are commercially available and can be read within minutes after exposure to amniotic fluid. Three studies that correlate positive leukocyte esterase test with either positive amniotic fluid culture or clinical chorioamnionitis were identified (Table VII).23. 2'. 27 Specificity was consistently high but sensitivity varied from a low of 8.3% to 83.3% in the different studies. Gas-liquid chromatography. Bacteria produce metabolites such as volatile fatty acids and nonvolatile organic acids that are detectable in body fluids by gas-liquid chromatography.63 The excellent indices presented by Gravett et al.· 8 have not been reproduced by two other groups (Table VIII)!,",9 Fibronectin in amniotic fluid. Fibronectin describes a family of high molecular weight glycoproteins that are present on many cell surfaces, in extra cellular fluids , in connective tissues, and in most basement membranes. Its biologic activities include regulation of cell adhesion, spreading, and locomotion.'" Fibronectin circulates in plasma where it helps to maintain vascular integrity and promotes reticuloendothelial clearance of particulate maUer. Decreased levels of fibronectin in neonates are thought to be one of the abnormalities of their host defense system. 65 One study (N = 30) that correlates amniotic fluid concentrations of fibronectin with intraamniotic infection has been published .· s A decrease in fibronectin was studied with positive amniotic fluid culture as gold standard. No difference was found in the concentrations of fibronectin in the amniotic fluid in the in-

N egatIVe predictive value ( %)

Prevalence

fected (mean ± SD = 26.1 ± 13.1 mg/nl) versus noninfected group (mean ± SD = 25.4 ± 12.9 mg/nl). Noninvasive tests with ultrasonographic techniques Ultrasonographically estimated amniotic fluid pocket of reactive protein with premature rupture of membranes and premature labor. Obstet Gynecol 1983;62:49-51. Ismail MA, Zinaman MJ, Lowensohn RI, Moawad AH. The significance of C-reactive protem levels in women with premature rupture of membranes. AM J OBSTET G)NECO) 1985; 151 :54 1-4. Fisk NM, Fysh]. Child AG, Gatenby PA, ]effer>' H ,Bradfield AH. Is C-reactive protein reall y useful in preterm premature rupture of the membranes? Br J Obstet CynaecoI1987;94:159-64. Ernest ]M, Swain M, Block SM. Nelso n LH, Ha~is CG, Meis PJ. C-reactive pro tein: a limited test for managing patients with preterm labor or preterm rupture of membranes? AM] OBSTET GV:\F.COL 1987; 156:449-54. el-Mekkawi TM, Abdel Aziz AA, el-Shlemy RE, Abdel Rahman S, Otaifi G. C-reactive protein as a predictor of chorioamnionitis with rupture of membranes. ] Egypt Public Health Assoc 1985;60:287-99. Potkul RK, Moawad AH , Ponto KL. The association of subclinical infection with preterm la bor: the role of Creactive protein . AM] OBS'IF) GYNE Lol.I985;153:642-S . Vintzileos AM, Campbell WA, Nochimson D], Weinbaum PJ, Escoto DT, Mirochnick MH. Qualitative amniotic fluid volume versus amnioce ntesis in predicting infection in preterm premature rupture of the membranes. Obstet Gynecol 1986;67:579-83. Vintzlleos AM. Campbe ll WA, Nochimson DJ. Weinbaum PJ. Degree of oligoh ydramnios and pregnancy outcome in patients with premature rupture of the membranes. Obstet GynecoI1985;66:162-7.

Tests to detect infection in preterm PROM

817

40. Gonik B, Bottoms SF. Cotton DB. Amniotic fluid volume as a risk factor in preterm premature rupture of the membranes. Obstet Gyn ecol 1985;65:456-9. 41. Vintzileos AM, Campbell WA, Nochimson DJ, Weinbaum PJ, Mirochnick MH, Escoto DT. Fetal biophysical profile versus amniocentesis in predicting infection in preterm prema ture rupture of the membranes. Obstet Gvnecol 1986;68:488-94. 42. Vintzileos AM. Campbell WA. Nochimson DJ. Weinba um PJ. The use of the nonstress test in patients with premature rupture of the m embranes. AMJ Oss') u GY"' ECOI. 1986; 155: 149-53. 43. Vintzileos AM. Bors-Koefoed R, Pelega no .IF, Campbell WA, RodisJF, Nochimsoll DJ, Kontopo ulos VG. The use of fetal biophysical profile improves pregnancy outcome in premature rupture of the membranes. A~ J OBSTET GYNECOL 1987;157 :236-40. 44. Augustsson P, Patel NB. The predictive value of feta l breathing movements in the diagnosis of preteI'm labour. BrJ Obstet Gynaecol 1987;94:860-3 . 45. Goldstein I, Romero R, Merrill S, et al. Fetal body and breathing movements as predictors o f intraamniotic infectio n in preterm premature rupture of membranes. AM .I OBSTF.T GYNECOL 1988; 159:363-t!. 46. Vintzileos AM, Campbell WA, Nochimson DJ. Weinbaum P]. Fetal breathing as a predictor of infection in premature rupture of the membranes. Obstet Gynecol 1986; 67 :8 13-7. (Erratum in Obstet Gynecol 1986;68:559.) 47. Mo berg LJ , Garite TJ, Freeman RK. Fetal heart rate patterns a nd fetal distress in patients with preteI'm premature rupture of the membranes. Obstet Gynecol 1984;64: 60-4 . 48. Robert JA. Romero R, Costigan K. Amniotic flUId concentrations of fibronectin and intraamniotic infection. Am .I Perinatol 1988;5 :26-8. 49. Spinnato .lA , Shaver DC, Bray EM, Lipshitz J. Preterm premature rupture of the membranes with fetal pulmonary maturity present: a prospective study. Obstet GynecoI1987;69:196-201. 50. Feinstein SJ, Vintzileos AM. Lodeiro JG. Campbell WA , Weinba um PJ, Nochimsol1 D.J. Amniocentesis with premature rupture of membranes. Obstet Cynecol 1986;6t!: 147-52 . 51. Broekhuizen FF, Gilman M. Hamilton PR oAmniocentesis for Gram stain and culture in preterm premature rupture of the membranes. Obstet Gynecol 1985;66:316-21. 52. Gonik B, Cotton DB. The u~e of amniocentesis in preterm premature rupture o f membranes. AmJ Perinatol 1985; 2:21-4. 53. Ziatnik FJ , Cruikshank DP, Petzold C R, Calask RP. Amniocentesis in the identifica tion of inapparent infection in preteI'm patients with premature rupture of the membranes . .I Reprod Med 1984;29:656-60. 54. Cotton DB, Gonik B, Bottoms SF. Conservative versus aggress ive management of preteI'm rupture of membranes. AmJ PerinalOl 1984; 1:322-4. 55. Cotton DB, Hill LM . Strassner HT, Platt LD. Ledger W.J. Use of amniocentesis in preteI'm gestation with ruptured membranes. Obstet Gynecol 1984;63:38-43 . 56. Garne TJ. Freeman RK. Lmzey EM, Braly P. The use of amnioce ntesis in patie nts with premature rupture of membra nes. Obstet Gynecol 1979;54:226-30. 57. Morales WJ, Lim D. Reduction of group B streptococcal materna l and neonata l infections in preteI'm pregnancies with premature rupture of membranes through a rapid ide ntifi cation test. AM] OSSTET GY:'