|
|
| |||||||||||||||||||||||||||||||||||||||
Background Nosocomial infections are a major cause of morbidity and mortality in premature infants. As a rule, their low serum gamma globulin levels at birth subsequently decline to hypogammaglobulinemic values; hence, prophylactic administration of intravenous immune globulin may reduce the rate of hospital-acquired infections.
Methods In this prospective, multicenter, two-phase controlled trial, 2416 infants were stratified according to birth weight (501 to 1000 g and 1001 to 1500 g) and randomly assigned to an intravenous immune globulin group (n = 1204) or a control group (n = 1212). Control infants were given placebo infusions during phase 1 of the study (n = 623) but were not given any infusions during phase 2 (n = 589). Infants weighing 501 to 1000 g at birth were given 900 mg of immune globulin per kilogram of body weight, and infants weighing 1001 to 1500 g at birth were given a dose of 700 mg per kilogram. The immune globulin infusions were repeated every 14 days until the infants weighed 1800 g, were transferred to another center, died, or were sent home from the hospital.
Results Nosocomial infections of the blood, meninges, or urinary tract occurred in 439 of the 2416 infants (18.2 percent): 208 (17.3 percent) in the immune globulin group and 231 (19.1 percent) in the control group (relative risk, 0.91; 95 percent confidence interval, 0.77 to 1.08). Septicemia occurred in 15.5 percent of the immune globulin recipients and 17.2 percent of the controls. During phase 1 the rate of nosocomial infections was 13.4 percent in the immune globulin group and 17.8 percent in the control group; the respective rates during phase 2 were 21.0 percent and 20.4 percent. The predominant organisms included gram-positive cocci (53.0 percent), gram-negative bacilli (22.4 percent), and candida species (16.0 percent). Adverse reactions were rarely observed during the infusions. Immune globulin therapy had no effect on respiratory distress syndrome, bronchopulmonary dysplasia, intracranial hemorrhage, the duration of hospitalization, or mortality. The incidence of necrotizing enterocolitis was 12.0 percent in the immune globulin group and 9.5 percent in the control group.
Conclusions Prophylactic use of intravenous immune globulin failed to reduce the incidence of hospital-acquired infections in very-low-birth-weight infants. .
1.5 kg) continue to increase,1 nosocomial infections remain a major cause of morbidity and mortality. Prolonged hospitalization with exposure to resistant organisms and multiple invasive procedures, in the presence of immunologic immaturity,2 renders these infants vulnerable to hospital-acquired infections3. Profound hypogammaglobulinemia may result from low levels of IgG at birth (IgG is largely acquired transplacentally in the latter half of the third trimester), degradation of maternally acquired IgG, and delayed production of IgG after birth4. The use of pooled IgG has been suggested as a possible means of reducing this deficiency and preventing nosocomially acquired infections. Prior studies testing the ability of intravenous immune globulin to prevent nosocomial infections in premature infants have varied in design and sample size5,6,7,8,9,10,11,12. Despite differences in the rates of observed infection, immune globulin preparations, doses, and infusion intervals, a meta-analysis of published reports suggests that nosocomial infections may be diminished by the prophylactic infusion of IgG13. The National Institute of Child Health and Human Development (NICHD) Neonatal Research Network therefore performed a prospective multicenter randomized trial to test the hypothesis that the intravenous administration of immune globulin to infants with birth weights between 501 and 1500 g would reduce the incidence of nosocomial infections.
Methods
A controlled clinical trial was conducted at eight participating centers between January 1, 1988, and March 31, 1991. Patients were randomly assigned to an intravenous immune globulin group or a control group. There were two phases to the study (see below). During phase 1 the control infants received infusions of placebo. During phase 2 the control infants received no infusion therapy. The protocol, study design, and consent forms were approved by the institutional review board at each study center. An independent data safety and monitoring committee was appointed by the NICHD to monitor the study.
Study Population
All neonates with birth weights of 501 to 1500 g were eligible for randomization. Infants were excluded if they were more than 72 hours old; they were one of three or more fetuses from a multiple pregnancy; they had infections associated with toxoplasma, rubella, cytomegalovirus, and herpes simplex viruses (the TORCH complex),14 a major congenital malformation, an identifiable syndrome, or a chromosomal abnormality; they were considered nonviable; or parental consent could not be obtained.
The infants were screened and parental consent was obtained within 72 hours of birth. Randomization, with an unbiased coin design,15 was stratified according to center and birth weight (501 to 1000 g and 1001 to 1500 g). Base-line data were obtained from the medical records of the mothers and infants at the time of enrollment. Additional clinical information, laboratory data, and morbidity data were collected daily with the use of standardized study forms. Episodes of sepsis were prospectively evaluated, and the symptoms, results of diagnostic tests, culture results, and treatments were recorded.
Infusions
A lyophilized human immune globulin product (Sandoglobulin, Sandoz Pharmaceuticals, East Hanover, N.J.) reconstituted with normal saline or 5 percent glucose (in very-low-birth-weight infants) was used in the trial. Each batch of immune globulin was derived from the pooled plasma of at least 8000 U.S. volunteer donors who were nonreactive for the human immunodeficiency virus and hepatitis B serum antigen. Four lots of immune globulin, distributed among the clinical centers, were used during the course of the trial. All lots were marked and recorded by code. The placebo infusion (phase 1) consisted of an equal volume of 5 percent albumin solution in the same vehicle prepared by the manufacturer of the immune globulin.
The infants received their first dose of study drug within 24 hours of randomization. To achieve a target level of 700 mg of immune globulin per deciliter, infants weighing 501 to 1000 g were given 900 mg of immune globulin per kilogram of body weight and infants weighing 1001 to 1500 g were given 700 mg per kilogram. The infusions were repeated every two weeks until the infants weighed 1800 g, were transferred to another hospital, died, or were sent home.
After an initial dose of 0.5 mg per kilogram per minute over a period of 30 minutes, the infusions were administered at a rate of 3 mg per kilogram per minute over a period of 4 to 6 hours. Vital signs, including blood pressure, were monitored throughout the infusion. Serum IgG levels were measured in blood obtained by a heel stick from all infants before the first infusion, seven days after the first dose, and before each additional dose. The serum samples were separated, frozen, and analyzed in a single laboratory by nephelometry (V. Hemming, M.D., Uniformed Services University of the Health Sciences, Bethesda, Md.).
Study Outcomes
The primary outcome variable in the trial was the development of confirmed nosocomial infection, including septicemia, meningitis, or urinary tract infection, during the first 120 days of life. A case of sepsis (septicemia) was considered to be confirmed if a positive blood culture for bacteria or fungi was obtained at least 96 hours after birth and before 120 days of life from an infant with symptoms compatible with infection. The diagnosis of sepsis due to organisms often regarded as commensal, predominantly coagulase-negative staphylococci, required two positive blood cultures for the same organism obtained no more than four days apart, either from a central catheter and a peripheral venipuncture or two peripheral venipunctures. A diagnosis of probable sepsis was given to symptomatic infants treated for more than three days with antibiotic therapy who had one positive blood culture for a commensal organism and either no subsequent cultures or a negative subsequent culture. The diagnosis of meningitis required a positive culture of cerebrospinal fluid. The diagnosis of urinary tract infection required a pure culture of more than 105 organisms per milliliter from a urine specimen obtained by a catheter or a culture of more than 102 organisms per milliliter from a specimen obtained by suprapubic puncture.
Secondary outcomes included mortality and neonatal morbidity, which was assessed in terms of the duration of ventilator support, frequency of bronchopulmonary dysplasia, and duration of hospitalization. Other outcomes that were assessed included local infections, necrotizing enterocolitis (with the stage determined according to Bell's modified classification),16 and specific complications of immune globulin or placebo infusion.
Study Design and Statistical Analysis
The study was designed to test the hypothesis that the use of immune globulin would reduce the incidence of confirmed nosocomial infections by at least 33 percent. The assignment of a 10 percent incidence of nosocomial infection in the placebo group was based on data from several of the participating clinical sites. We calculated that the study required a total sample of 2408 infants, assuming a power of 90 percent and a one-tailed type I error of 0.05.
Questions were raised midway through the trial about whether there was an increased incidence of necrotizing enterocolitis in the study population. The data and safety monitoring committee reviewing the data noted no significant difference in the incidence of necrotizing enterocolitis between the immune globulin and the placebo recipients. There were, however, fewer nosocomial infections among the immune globulin recipients; hence, the committee recommended continuing the study, but since the placebo infusion could not be beneficial and could instead be harmful, the committee mandated that the placebo infusion be discontinued and that the study continue in an unblinded fashion. During phase 2 (October 1989 through April 1991), 1198 infants were randomly assigned in an unblinded fashion to receive either immune globulin or no infusion.
The primary outcome -- the incidence of nosocomial infection in the two study groups -- was compared by the Mantel-Haenszel statistic, adjusted for clinical center and study phase. Asymptotic confidence intervals around the estimated relative risk were also calculated. The analysis of all patients was based on their assigned randomization group in an intention-to-treat analysis. Base-line comparisons were performed with standard t-tests and chi-square tests, as appropriate. Except for the primary outcome, all evaluations were assessed with the use of a nominal significance level of 0.05. Continuous data were expressed as means ±1 SD. Statistical analyses were performed with the SAS computer program (version 6.07)17. The method of Lan and DeMets18 was used to adjust for the effects of interim analyses on the probability of a type I error. A P value of 0.044 was therefore required to indicate statistical significance in the final analysis.
To evaluate the effectiveness of each lot of immune globulin, the treatment group was split into four subgroups, each containing all the infants who received immune globulin from a particular lot. Infants in the placebo group were assigned to the same subgroup as the infants in the immune globulin group from the same center with the closest randomization date. Infants who received infusions from more than one lot of immune globulin and their matched controls were excluded from this analysis (n = 196). The relative risk of nosocomial infection was computed, after adjustment for the subgroup, according to the method of Mantel and Haenszel.
Results
Screening and Enrollment
A total of 5498 infants were screened for eligibility. Of these, 1013 were ineligible for the following reasons: 63 percent were deemed unlikely to survive, 18 percent had major malformations, 11 percent had TORCH infections, and 8 percent were the product of a multiple gestation of more than twins. An additional 2069 infants were ineligible mainly because their parents either were unavailable (18 percent) or refused permission for enrollment (68 percent). Thus, 2416 infants were randomized, 1218 before the protocol modification (phase 1: 595 to immune globulin and 623 to placebo) and 1198 after the protocol modification (phase 2: 609 to immune globulin and 589 to no infusion). The infants were randomized a mean of 44 ±25 hours after birth.
Characteristics of the Infants and Their Mothers
Table 1 presents base-line maternal and neonatal characteristics. There were no statistically significant or clinically meaningful differences between the groups with regard to any of these factors. The incidence of respiratory distress syndrome and the level of IgG were similar in each of the groups before the initial infusion. Surfactant therapy was available for this study population only during phase 2; its use was evenly distributed among the study groups.
|
During phase 1, the immune globulin and placebo groups received their first infusion at 55.4 ±27 and 54.6 ±28 hours of life, respectively. During phase 2, the immune globulin group received their first infusion at 60.0 ±27 hours. The mean number of infusions given during phase 1 was 3.0 ±1.6 in the immune globulin group and 3.1 ±1.6 in the placebo group; during phase 2, a mean of 3.3 ±1.7 infusions of immune globulin were given. During phase 1, 71 percent of the immune globulin group and 74 percent of the placebo group received infusions until they reached a weight of 1800 g or were sent home. In each group, 8 percent died, 9 percent were transferred from the study center, 3 percent were withdrawn from the study, and 7 percent did not complete the course of infusions because of the temporary suspension of the trial. During phase 2, 77 percent of the infants assigned to immune globulin completed the infusion course, 8 percent died, 8 percent were transferred, and 7 percent were withdrawn before the course of infusions was completed.
The infusions were well tolerated, with few untoward reactions noted. Monitoring of hematologic, renal, and hepatic variables in the first 200 infants who received at least three infusions revealed no evidence of toxicity attributable to immune globulin. The infusions were discontinued in less than 1 percent of infants (10 in the immune globulin group and 11 in the placebo group) because of tachycardia or acute changes in blood pressure.
The serum IgG levels are presented in Figure 1 according to birth-weight category and treatment group. The levels were determined before and one week after the first infusion; thereafter, trough levels were determined before all subsequent infusions (Figure 1). Serum IgG levels in the immune globulin and placebo groups were similar in both birth-weight categories before the first infusion: 376 ±137 and 377 ±144 mg per deciliter, respectively, in infants with birth weights of 501 to 1000 g, and 542 ±216 and 546 ±218 mg per deciliter, respectively, in infants with birth weights of 1001 to 1500 g. Base-line values were missing for 4.7 percent of the infants. In the immune globulin group, 153 infants (13 percent) had IgG values above 700 mg per deciliter at birth. After the first infusion, 585 (49 percent) had trough levels that exceeded 700 mg per deciliter. In 151 infants (13 percent) in the immune globulin group, all trough levels met or exceeded the target level. In the control group, 155 infants (13 percent) were born with values above 700 mg per deciliter, but the target IgG level was sustained in only 18 of these infants. The immune globulin group had significantly higher values than the control group one week after the first infusion (751 ±269 vs. 481 ±195 mg per deciliter; P<0.01) and before each subsequent infusion (P<0.01).
|
During the study, 17.3 percent of the immune globulin group and 19.1 percent of the control group had confirmed nosocomial infections (septicemia, meningitis, or urinary tract infection) (relative risk, 0.91; 95 percent confidence interval, 0.77 to 1.08; P = 0.25 by the log-rank chi-square test) (Table 2). Meningitis was confirmed in 2 percent of each group, whereas 2 percent of the immune globulin group and 3 percent of the control group had urinary tract infections. During phase 1, 11.6 percent of the immune globulin group had septicemia, as compared with 16.4 percent of the placebo group. In phase 2, the incidence of septicemia increased to 19.2 percent in the immune globulin group and 18.2 percent in the control group. The final combined analysis (15.5 percent of the immune globulin group and 17.2 percent of the control group) revealed no statistically significant differences between the study groups (relative risk, 0.90; 95 percent confidence interval, 0.75 to 1.08). A second infection was noted in 2.7 percent of the infants in the immune globulin group and 3.2 percent of the control group. More than two infections per infant were uncommon. An additional 8 percent of each group was given a diagnosis of probable sepsis. There was no significant difference between the immune globulin and the control groups in the average length of time to the onset of the first infection (22 ±17 vs. 24 ±19 days). Infants weighing 501 to 1000 g at birth had a higher incidence of confirmed nosocomial infections than infants weighing 1001 to 1500 g at birth, regardless of treatment, in both phases of the trial (Table 3).
|
|
|
|
|
Intravenous immune globulin has been reported to be an effective adjunct to therapy in a number of diverse disorders, including primary immunodeficiencies, idiopathic thrombocytopenic purpura, the acquired immunodeficiency syndrome, and Kawasaki's syndrome19. Studies published before 1990 suggested that prophylactic immunotherapy also reduced nosocomial infections in very-low-birth-weight infants13. However, these studies enrolled small numbers of patients; employed varied designs, preparations, and doses; and included diverse study populations. In this large multicenter, randomized controlled trial, the repeated prophylactic administration of intravenous immune globulin failed to reduce the incidence of nosocomial infections significantly in premature infants weighing 501 to 1500 g at birth. Furthermore, there were no significant differences in morbidity, mortality, or the duration of hospitalization between infants given gamma globulin and infants given no infusion or an infusion of placebo.
There is no obvious explanation for the association between necrotizing enterocolitis and infusions of immune globulin or placebo. Necrotizing enterocolitis is a complex, multifaceted disorder variously attributed to ischemia, infection, and feeding practices, alone or in combination20. To explore the relation between the infusions and necrotizing enterocolitis, a matched case-control study is under way comparing detailed approaches to fluid management and feeding in all the infants in whom necrotizing enterocolitis developed during the trial.
The lack of efficacy of repeated infusions of immune globulin in reducing nosocomial infections in very-low-birth-weight infants conflicts with the results of Baker et al.,5 who reported a significant reduction in the number of first infections and a shortened hospital stay among gamma globulin recipients with infections during the first 56 days of life. However, their reported infection rates of 35 percent (104 of 297) in controls and 24 percent (70 of 287) in immune globulin recipients far exceed the rates of 19 percent (controls) and 17 percent (immune globulin group) in our cohort. When their data were expressed as the number of nosocomial infections per 100 patient-days (according to the Centers for Disease Control and Prevention guidelines for comparing interhospital rates of nosocomial infection),21 Baker et al.5 documented a rate of 1.08 per 100 patient-days in the placebo group as compared with a rate of 0.86 per 100 patient-days in the immune globulin group; these rates differ from the respective rates of 0.35 and 0.31 per 100 patient-days recorded in the NICHD trial. Differences in the characteristics of the cohorts as well as in the antibody profiles of the immune globulin preparations may also account for the disparity in the findings of these two large trials. Magny et al.,22 Weisman et al.,23 and Kinney et al.24 have also reported that immune globulin therapy failed to reduce infection rates. In addition, Magny et al. reported increased rates of necrotizing enterocolitis among the immune globulin recipients22.
The increased susceptibility of premature infants to infections coincides with their reduced serum immunoglobulin levels, but they also have quantitative cellular and humoral defects2. The kinetics of intravenous immune globulin in neonates, as measured by the half-life of the serum IgG level, are difficult to predict; hence, neither the optimal dose nor the protective serum concentration of IgG has been established25,26,27,28. Kyllonen et al.26 achieved a serum IgG level of 700 mg per deciliter, the lower "normal" limit in term newborns, by infusing 900 mg per kilogram in infants weighing less than 1 kg and 700 mg per kilogram in those weighing 1 kg or more. These levels were sustained for two weeks in 40 percent of the smaller infants and 8 percent of the larger infants. Clapp et al.7 documented IgG levels below 400 mg per deciliter at the time of diagnosis in all infected infants. In the NICHD study, however, despite the fact that mean trough serum IgG levels exceeded 500 mg per deciliter at each measurement in the immune globulin group (Figure 1), the rate of nosocomial infections was not reduced.
When the efficacy of the four lots of immune globulin administered in our trial was compared, only one lot, given to 79 infants (6.6 percent), significantly reduced the rate of nosocomial infections. Despite the use of many thousand donors for each processed batch of immune globulin, a substantial batch-to-batch variation has been noted in the antibody profile29. No functional activity against coagulase-negative staphylococcus, a common pathogen, could be demonstrated in any of the four batches used in the trial. However, protective antibody against group B streptococci was demonstrated in all four lots of immune globulin, and late-onset infections due to group B streptococci were significantly reduced in the group given immune globulin. We were technically unable to test for activity against candida species (Hill H: personal communication).
The principle of the use of immune globulin therapy to supplement falling IgG levels in very-low-birth-weight infants appears sound. The potential effectiveness of commercial preparations may be limited by the absence or low levels of functional antibody against the specific pathogens causing most neonatal nosocomial infections. Differences in immune globulin activity may also explain the variable responses to different batches from the same manufacturer as well as between different products. Since hypogammaglobulinemia is only one of the defects in the immature infant's immune system, the role of intravenous immune globulin may be further limited. The development of targeted immune globulin preparations against the most common neonatal pathogens may have a place in the prevention of nosocomial infections in low-birth-weight infants. The potential benefit of such preparations must be balanced against the apparent risk of precipitating disorders such as necrotizing enterocolitis.
Supported by the National Institute of Child Health and Human Development through Cooperative Agreements (U10 HD21364, U10 HD21415, U10 HD19897, U10 HD21385, U10 HD21397, U10 HD21373, U10 HD27890, and U10 HD21466).
We are indebted to Jill E. Baley, M.D., Melvin Berger, M.D., D. Wade Clapp, M.D., Val Hemming, M.D., Harry Hill, M.D., Robert M. Kliegman, M.D., Barbara Stenzel, M.B.A., and Anne Willoughby, M.D., for their generous assistance in the design and conduct of this trial; to Joel Verter, Ph.D., and Naji Younes, M.A., for their assistance in the data analysis and interpretation; and to Sandoz Pharmaceutical for donating the immune globulin.
Source Information
From Case Western Reserve University, Cleveland (A.A.F.); the University of Tennessee at Memphis, Memphis (S.B.K.); the National Institute of Child Health and Human Development, Bethesda, Md. (L.L.W., C.S.C.); the Biostatistics Center, George Washington University, Rockville, Md. (E.C.W.); Wayne State University, Detroit (R.L.P., S.S.); the University of Miami, Miami (C.B.B.); the University of Texas Southwestern Medical Center, Dallas (J.E.T); the University of Alabama at Birmingham, Birmingham (J.B.P.); Dartmouth-Hitchcock Medical Center, Hanover, N.H. (W.E.); the University of Vermont, Burlington (J.F.L.); and Brown University, Providence, R.I. (W.O.). Presented in part at the Society for Pediatric Research Annual Meeting, Baltimore, May 4, 1992.The members of the National Institute of Child Health and Human Development Neonatal Research Network are listed in the Appendix.
Address reprint requests to Dr. Fanaroff at the Division of Neonatology, Rainbow Babies and Children's Hospital, 2074 Abington Rd., Rm. 3100, Cleveland, OH 44106.
References
The members of the National Institute of Child Health and Human Development Neonatal Research Network are as follows: Brown University: W. Oh (Chairman); University of Alabama at Birmingham: J.B. Philips III, G. Cassady, and K. Braune; Case Western Reserve University: A.A. Fanaroff, M. Hack, and N.S. Newman; Dartmouth-Hitchcock Medical Center: W. Edwards, G. Little, and C. Nattie; George Washington University: R.P. Bain, J. Verter, E.C. Wright, N. Younes, and S. Hawes; Wayne State University: R.L. Poland, S. Shankaran, and G. Muran; University of Miami: C.R. Bauer, E.S. Bandstra, and S. Martinez; National Institute of Child Health and Human Development: S.J. Yaffe, C.S. Catz, L.L. Wright, and M. Malloy; University of Tennessee at Memphis: S.B. Korones, R. Cooke, and J. Moore; University of Texas Southwestern Medical Center: J.E. Tyson, R. Uauy, and J. Burchfield; and University of Vermont: J.F. Lucey, J.D. Horbar, and K. Leahy.
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
Multiple-Antibiotic-Resistant Bacteria
Bloch A. B., Onorato I. M., Castro K. G., Tomasz A.
Extract |
Full Text
N Engl J Med 1994;
331:678-679, Sep 8, 1994.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
