 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
* Center for Primary Care and Outcomes Research, Stanford University School of Medicine, Stanford, California
Department of Pediatrics, Stanford University School of Medicine, Stanford, California
Department of Management Science and Engineering, School of Engineering, Stanford University, Stanford, California
|| Graduate School of Business, Stanford University, Stanford, California
¶ Veterans Affairs Palo Alto Health Care System, Palo Alto, California
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
Participants. A hypothetical cohort of pregnant women and their newborns.
Interventions. Screening strategies for group B streptococci using the new polymerase chain reaction technique, the 35- to 37-week culture, or maternal risk factors.
Outcome Measures. Infant infections averted, infant deaths, infant disabilities, costs, and societal benefits of healthy infants.
Results. A screening strategy using the new polymerase chain reaction test generates a net benefit of $7 per birth when compared with the maternal risk-factor strategy. For every 1 million births, 80 700 more women would receive antibiotics, 884 fewer infants would become infected with early-onset group B streptococci, and 23 infants would be saved from death or disability. The polymerase chain reaction-based strategy generates a net benefit of $6 per birth when compared with the 35- to 37-week prenatal culture strategy and results in fewer maternal courses of antibiotics (64 080 per million births), fewer perinatal infections with early-onset group B streptococci (218/million), and a reduction in 6 infant deaths and severe infant disability per million births. The benefits hold over a wide range of assumptions regarding key factors in the analysis.
Conclusions. Although additional clinical trials are needed to establish the accuracy of this new polymerase chain reaction test, initial studies suggest that strategies using this test will be superior to the other 2 strategies. Using the rapid polymerase chain reaction test becomes less attractive as the cost of the test increases. The test’s greatest strengths lie in its ability to identify women and infants at risk at the time of labor, thereby decreasing the number of false-positives and false-negatives seen with the other 2 strategies and allowing for more accurate and effective intrapartum prophylaxis.
Key Words: Streptococcus agalactiae • streptococcal infections • mass screening • cost-benefit analysis
Abbreviations: GBS, group B ß-hemolytic Streptococcus • EOGBS, early-onset group B streptococcal (disease) • AAP, American Academy of Pediatrics • CDC, Centers for Disease Control and Prevention • PCR, polymerase chain reaction • NICU, neonatal intensive care unit • PPV, positive predictive value • NPV, negative predictive value • LPCH, Lucile Salter Packard Children’s Hospital • IV, intravenous • OSHPD, Office of Statewide Health Planning and Development
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
These guidelines offer physicians the choice of either a maternal screening test (rectovaginal culturing at 35–37 weeks) or a maternal risk-factor-based approach. The value and practicality of both strategies has been debated in the literature. At issue are potential overtreatment (in the case of the culture strategy) and undertreatment (in the case of the risk-factor strategy) of patients, as well as the associated costs.15–20 For example, the standard screening test, a rectovaginal culture taken at 35 to 37 weeks, has been controversial because it may not accurately predict genital tract colonization at time of labor (with an estimated sensitivity of 87%–91% and specificity of 89%–96%).21,22 The risk-factor method, on the other hand, would target treatment toward the mothers believed to be at greatest risk but would miss many colonized mothers and at-risk infants.22 Despite their limitations, both approaches are effective in reducing the EOGBS infection rate in infants,4,23–27 although more widespread implementation of the guidelines is needed.3,4
A recent study looked at a new rapid polymerase chain reaction (PCR) assay as an alternative screening tool allowing rapid identification of GBS-colonized women on admission for delivery.28 In the rapid PCR technique, a sample is analyzed using fluorescent hybridization probes and fluorescence detection, which is quicker than the standard gel electrophoresis. This technique also allows for rapid cycling during the amplification process, speeding up this aspect of PCR testing as well. Thus, for many clinical facilities this test offers a feasible alternative screening method, as results are available in <45 minutes.28
This new PCR technique study involved 112 women. These women had vaginal swabs done at time of admission for labor. The samples were tested with the rapid PCR technique, conventional PCR technique, and standard culture (used as the gold standard comparison). Rapid PCR had a sensitivity of 100% and specificity of 98.9% in tests run on vaginal swabs when compared with culture.
One of the potential advantages of using this test for screening is that it could be used at time of admission for labor, therefore offering better sensitivity and specificity for predicting colonization in women at the time of labor when compared with the currently favored screening technique (ie, a culture done prenatally at 35–37 weeks). In addition, rapid processing makes it possible to administer chemoprophylaxis promptly to women before delivery. Finally, mothers who did not receive prenatal care could be screened with the PCR technique at time of admission to the hospital, whereas screening with the conventional culture would not be possible for such women. Moreover, PCR does not depend on the timely transmission of medical records from outside clinics to the hospital of delivery, which both limits accurate identification and treatment of women. The PCR promises to identify women at greatest risk for delivering an infected infant, ie, those with true vaginal colonization at time of labor. This, in turn, would allow for more targeted and effective antibiotic prophylaxis—treatment of a smaller group of women (when compared with 35- to 37-week rectovaginal culture) and potential prevention of illness in a greater number of infants. Although additional studies and testing are needed before PCR can be accepted as a clinical diagnostic technique on its own, these initial results are promising. However, it is uncertain whether the potential benefits of the improved screening ability justify the costs of the test.
To answer this question, we performed a cost-benefit analysis of this new PCR method. We examined the potential health benefits, costs, and savings associated with 3 alternative strategies for identifying and treating (with intrapartum antibiotics) those mothers at risk for passing GBS on to their infants: 1) a hypothetical screening strategy using the new rapid PCR at time of hospital admission for labor, 2) a screening strategy using the standard 35- to 37-week culture, and 3) a strategy using maternal risk factors at time of labor.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
In the model, women-infant pairs can be treated (with intrapartum antibiotics) or not, according to the GBS prevention strategy they are following. Maternal treatment alters the probability that an infant will become infected. Those that remain healthy (infection-free) are assumed to reach their full economic potential. Those that become infected are admitted to the neonatal intensive care unit (NICU) for treatment and subsequently can recover fully (and also achieve full economic potential), be disabled by the illness (and have no economic potential), or die as the result of the infection (Fig 1).
Probabilities
The population of interest consists of women delivering at term and their infants. Estimates of the prevalence of GBS colonization as tested by rectovaginal culture in this population range from 15% to 40%.1–3 A recent review of GBS literature done by Benitz et al19,22 estimated the rectovaginal colonization rate, when tested by culture at 35 to 37 weeks, at 22.8%. Actual vaginal colonization rate at delivery (also identified by culture) was estimated to be 14.7%.19,22 Rectovaginal cultures performed at 36 weeks were estimated to have a sensitivity of 90.8% and specificity of 88.9% when used to predict vaginal colonization at time of labor (positive predictive value [PPV] of 58.6% and negative predictive value [NPV] of 98.2%, assuming a prevalence of 14.7% for vaginal and 22.8% for rectovaginal colonization with GBS).22 This review also estimated that the prevalence of maternal risk factors at labor was 7.56%; that the probabilities of infant infection given maternal vaginal colonization status at time of labor were 1.47% if colonized, 0.0017% if not; that the probabilities of infant infection were 1.4% if maternal risk factors were present at time of labor, and 0.12% if not; and that expected infant infection rates were decreased by 80.2% with maternal intrapartum ampicillin prophylaxis.19,22,48
The new PCR test, when used for vaginal testing at time of labor, has a sensitivity of 100% and specificity of 98.9% (PPV: 95.2%, NPV: 100%), when compared with cultures obtained at the same time.28 Based on its reported sensitivity and specificity, and given a vaginal colonization rate of 14.7%,19 PCR done on vaginal swabs from women in labor would be positive in 
15.6% of women.
We used data from the California Department of Health Services on the availability of quality prenatal care to estimate the probabilities for women arriving in labor without available culture results. Women receiving "adequate" prenatal care would have the test done prenatally at 35 to 37 weeks under the culture strategy. Women without prenatal care constitute 1% of the pregnant population; the prevalence of poor prenatal care (defined by time of initiation of care and number of visits) is estimated at 30%.50 Although the definition of poor prenatal care is often based on time of initiation of care, a significant number of these women will have inconsistent care with missed visits; we therefore estimated our baseline value for "inadequate" prenatal care, ie, no available culture, to be 5%.
We assumed that all mothers with positive tests or positive risk factors would receive timely and appropriate antibiotics.
Outcomes
Our model focused on key perinatal outcomes: infant wellness (without infection), infant death after infection, infant disability after infection, and infant recovery and wellness after infection. When an infant became ill with EOGBS, the outcomes (ie, probability of death and disability) were assumed to be the same across strategies, regardless of screening test, or rate of maternal prophylaxis.27 Published estimates of death and disability rates from EOGBS in term infants vary considerably; our analysis used results from studies conducted in the 1990s.1,4,20,34 All outcome probabilities were varied in sensitivity analysis. We assumed that disabled infants would be severely and permanently disabled (with cerebral palsy, developmental delay, and/or paralysis) with no earning potential as adults.1,31–34 Well infants, both with and without history of illness, were assumed to have full earning potential.
Costs and Benefits
Using standard methods, we calculated the direct costs, or costs associated with medical tests and treatments, and indirect costs, or lost earnings associated with disease. Direct costs included screening test costs, maternal intrapartum antibiotics cost, costs for well-infant care, expenditures associated with treating infants with EOGBS in the NICU, and long-term care costs for well and disabled children. Indirect costs included the child’s potential earnings lost because of death or disability.
The benefits were obtained by subtracting the net present value of an unaffected child’s lifetime health care costs from the net present value of expected lifetime earnings. All costs used are in 2001 US dollars.
The cost of a rectovaginal culture screening test (done at 35–37 weeks) was derived from Stanford University Medical Center (Lucile Salter Packard Children’s Hospital [LPCH]) data. Cost estimates for the new PCR test were calculated from LPCH data on lab equipment and personnel costs, as well as cost data from the local manufacturer and distributor of the PCR equipment (Cepheid, Sunnyvale, CA).a
When any screening test was positive, or the mother was febrile and/or had prolonged rupture of membranes, we assumed that the women were treated with antibiotic therapy during labor. The therapy in our analysis consisted of an initial dose of 2 g of ampicillin intravenously, followed by 1 g every 4 hours at the cost of $63 per course of therapy.b This included the LPCH costs of the antibiotics, intravenous (IV) line supplies, nursing time, as well as the costs of treatment of potential maternal allergic reactions, anaphylaxis, and death.16–20 We excluded the costs and benefits of the treatment for the maternal outcomes of postpartum bacteremia and endometritis from our calculations and focused instead on infant-related costs and benefits.
Generally, the infant of a mother who is identified as a carrier of GBS would undergo observation in the hospital for 2 days (a standard length of stay for a well infant delivered vaginally). If the infant is free of EOGBS infection, he or she is discharged after 2 days, or 4 days if delivered via cesarean section. This is a standard length of stay for infants of mothers not colonized by GBS as well. Based on LPCH data, we estimated that the total cost of caring for the unaffected neonates was $436 per day. Therefore, the average, direct, cost of caring for an unaffected neonate would be $1100, based on a 22% cesarean section rate.51
If the infant developed signs or symptoms of a GBS infection, he or she would be transferred to the NICU. Infants infected with EOGBS require treatment in a NICU (10–14 days for an uncomplicated case). Published estimates of NICU costs vary.15–17,20 For this study, we analyzed data from the California Office of Statewide Health Planning and Development (OSHPD) neonatal dataset. The criteria used to analyze the data were: an infant born between 1995 and 1998, an International Classification of Diseases, Ninth Revision code corresponding to GBS, and, for those that survived, a stay of at least 10 days (to exclude those infants that were never admitted to the NICU or simply admitted for ruling out infection). For the base case, we calculated that $30 100 was the mean NICU cost and assumed this was the same whether the infant survived, survived with disability, or died.
The discounted cost of long-term care for an infant disabled by an EOGBS infection is calculated from long-term health care costs and life-expectancy data in Waitzman et al.52 Estimated lifetime health costs for disabled infants were based on costs for health care for children with cerebral palsy, as the more severe outcomes of GBS often involved cerebral palsy or cerebral palsy-like symptoms.
Finally, for the analysis the overall potential value of a well infant to society was assumed to equal the difference between the potential earnings of this infant and the lifetime health care expenditures (based on the estimate for the general population). The discounted potential earnings of a well infant were calculated from National Vital Statistics life tables and US Census Bureau income estimates.53,54 Lifetime health care expenditures were based on the Consumer Expenditure Survey, age-specific mortality from US life tables, and were discounted at a rate of 3%.54,55 The expected societal benefit that a well infant can bring through her or his life, therefore, was the difference between the present value of lifetime earnings—$411 200—and the present value of lifetime health care expenditures—$33 300, or $377 900 in 2001 US dollars.
Analysis
In our model, the test and subsequent intervention were administered only once—during labor. The final outcomes (ie, infant death, disability, or wellness) and their economic values were treated as known at the time of our decision analysis and were listed in terms of 2001 dollars.
This analysis does not explicitly model quality-of-life effects, concentrating instead on the financial impacts of alternative strategies. Our decision analysis was done using Decision Maker 1999 (Version Beta 0.99.14.0a, Rutgers University, NJ).
Sensitivity Analysis
Sensitivity analyses were performed on key costs: cost of PCR, cost of maternal antibiotics, cost of acute NICU care, cost of well-infant in-hospital care, and cost of long-term care for well and disabled children. Ranges for the sensitivity analyses on costs were derived from published studies16–18,20,52,54–56 and our calculations. The sensitivity analyses also evaluated the effects of variation in potential lifetime income (ranges based on literature review),53,54,56 percentage of women receiving adequate prenatal care (ie, having a history of a culture being performed) (ranges based on California data),50 and PCR sensitivity and specificity (ranges assumed).
Prevalence of overall GBS colonization was varied together with prevalence of maternal risk factors (as an increase in maternal GBS colonization would lead to an increase in maternal fevers resulting from GBS). The probability of EOGBS illness in infants was varied as well. These numbers were varied by ± 20%. Finally, the probabilities of infant death and disability1,4,20 and infant infection rates after maternal prophylaxis19,48,49 were varied to encompass the range reported in the literature.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
Sensitivity Analysis
When comparing the values of the net present value of lifetime earnings less the net present value of the direct costs for the PCR strategy and the risk-factor strategy, we found the difference between these 2 values was most sensitive to NICU costs, the probability of disability, and the cost of the PCR test. In the base case, NICU costs were assumed to be $30 100 per stay. If the cost is $22 000, the difference between the strategy values becomes zero, and at $12 000 per stay the risk-factor overall value is $9 greater than the PCR overall value. At the other end of the range of values, with a NICU stay equal to $128 000, the value of the PCR strategy exceeds the value of the risk-factor strategy by $94. In the base case, probability for disability was set at 1.6%, if the probability for disability is zero, then the value of the risk factor strategy value is $2 greater than the value of the PCR strategy. At 10% risk for disability after infection, the PCR value is $57 dollars greater. In the base case, the cost of the PCR test was $26, at $54 per test the risk-factor method value is $21 greater than the PCR, but at a PCR test cost of $16 the PCR method value is $17 greater (Fig 2).
|
90% (Fig 3).
|
We performed a 2-way sensitivity analysis on the costs and benefits of the PCR screening strategy as a function of PCR sensitivity and specificity (Figs 4 and 5). Maintaining all other values at baseline, if the PCR sensitivity dropped to 94%, the PCR specificity would have to be 94% or higher for the PCR strategy to be equal or superior to the culture strategy. Similarly, if PCR sensitivity dropped to 93% its specificity would have to remain at roughly 94% or higher for the PCR to provide greater benefits than the risk factor strategy.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
There are several limitations to our model. Estimates of the sensitivity and specificity of PCR are sparse; we relied on figures reported in the Bergeron et al article,28 which might be relatively favorable. Larger clinical trials of the test are needed to clarify its accuracy. However, according to our baseline values, if both the sensitivity and specificity of PCR exceed 94%, the overall value of the PCR strategy is greater than that of the other 2 options. The costs of the PCR test in specific settings might be either lower or higher than we assume at the baseline, because costs of equipment, reagents, labor, and the volume of testing may vary. However, the sensitivity analysis showed that the cost of the PCR would have to increase to $32 per test for the overall method value to drop to that of the culture method baseline, and $33 per test for the overall method value to drop to that of the risk-factor method baseline.
The effectiveness of any GBS screening program depends on timely administration of antibiotics to mothers in the hospital. Even if 1 of the 3 screening methods is used, a short labor and brief delivery, poor communication, errors, and many other events may interfere with prompt administration of antibiotics. In addition, the time needed to collect and transport the sample and the 30 to 45 minutes required to process the PCR could delay antibiotic administration when compared with the other 2 strategies. The rate at which antibiotics can be delivered quickly and effectively will need to be studied better to improve the success of all screening strategies for GBS. Furthermore, the consequences of delay in terms of actual infant infection rates remain primarily unknown.48,57 Alternatively, the preadmission culture screening results may allow for those women who are positive for GBS to arrive earlier in their labors and receive more timely treatment, a benefit that might be lost if all testing were only done at time of admission. In summary, real world practice will differ from our assumption of perfect implementation for all three strategies and could impact the outcomes in our model.
In addition, although we have attempted to calculate and include the costs of the additional lab personnel needed to run the PCR tests 24 hours/day, we have presented the PCR model with the assumption of an on-site and capable laboratory in the hospital using the test. This narrows the field of hospitals addressed by our model baseline to larger, predominantly academic institutions (although estimates were varied in sensitivity analysis), and these results may not apply in other settings. Indeed, because the per-test cost would be higher, PCR is likely to be a less attractive alternative in lower-volume hospitals.
In other respects, our analysis might underestimate the potential value of PCR. We may have overestimated the cost to the purchasing hospital of providing PCR, because the PCR equipment would likely be used for other clinical lab tests along with the test for GBS. Because our cost assumptions are based on GBS-only use, a hospital using PCR for many uses is likely to be able to spread the cost of the machine over many more tests, resulting in a lower cost per test.
We also did not include the potential benefits from using PCR as a screening technique for certain preterm deliveries. Even in pregnancies as early as 31 weeks, reliable intrapartum screening tests can identify women and infants at greatest risk and allow for effective treatment, without treating all mothers delivering before 37 weeks.19 This could reduce the number of women receiving antibiotics inappropriately, without any adverse effects on their infants.
In addition, we have not looked at the indirect costs of overuse of antibiotics, eg, resistance of GBS. Although there has been no clinical evidence of GBS resistance to penicillins2,3 the higher rate of antibiotic use under the culture screening strategy remains a point of concern for many practitioners. The potentially more focused antibiotic use under the PCR strategy would help decrease the risk of resistance.
Because we focused on cost consequences, our analysis does not incorporate parental disutilities for losing an infant or raising a disabled child. Although our analysis focused on infant outcomes, parental disutilities and lost parental income, if included, would increase the overall societal value of the PCR and strengthen its cost-effectiveness.
Finally, wider adoption of infant postpartum prophylaxis19,48 as a EOGBS-reducing intervention would make the PCR strategy more valuable. Those infants whose mothers tested positive on PCR would be candidates for the chemoprophylaxis, resulting in fewer infants being treated than under the 35- to 37-week culture strategy, but with greater accuracy than under either the culture or risk-factor strategy.
GBS and EOGBS infections continue to be a significant source of morbidity and mortality in the United States. With good screening techniques and timely antibiotic prophylaxis, the majority of these neonatal infections can be prevented. Current strategies for identifying at risk mothers and infants have been criticized. The new rapid PCR technique is a promising way to identify these mothers because it is both speedy and accurate. Our analysis also suggests that its benefits exceed its costs.
|
ACKNOWLEDGMENTS |
|---|
We thank Dr Ellen Jo Baron, Kevin Krave, Mary McIntyre, and Peter Spivak, MS, for providing cost data from Lucile Salter Packard Children’s Hospital and Stanford Hospital, and Susan Schmitt for her help with the OSHPD data.
|
FOOTNOTES |
|---|
Address correspondence to Corinna A. Haberland, MD, Center for Primary Care and Outcomes Research, 117 Encina Commons, Stanford University, Stanford, CA 94305. Email: haber{at}healthpolicy.stanford.edu
a The machine cost ($57 000) was distributed as a per test cost, with a 4-year depreciation and 10% salvage value, based on approximately 4000 deliveries per year (approximate LPCH average). Included in the cost were all necessary supplies, and costs of controls (total = $8.86), as well as the fixed lab cost per test ($5.04). As the test requires approximately 15 minutes of technician time (it will be necessary for a technician to be available 24 hours per day), the hourly rate of a senior technician, plus overhead, was divided by 4 (result = $9.38) and added to the other cost estimates to obtain the estimate listed in Table 1. 
b Included were: LPCH costs of the antibiotics (we assumed an average of 3 doses per patient, 2 g = $2.90, and 1 g = $1.90), IV line start supplies ($2.30), IV fluid ($41.00, and assumed about 50% of women had IV lines already in place for other reasons, result $20.50), and nursing time (approximately one-half hour at a rate of approximately $60 per hour).
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  M. Busetti, P. D'Agaro, and C. Campello Group B streptococcus prevalence in pregnant women from North-Eastern Italy: advantages of a screening strategy based on direct plating plus broth enrichment J. Clin. Pathol., October 1, 2007; 60(10): 1140 - 1143. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2002 American Academy of Pediatrics. All rights reserved. |
||
|
||
