Lumbiganon, Pisake MD; Villar, Jose MD; Laopaiboon, Malinee PhD; Widmer, Mariana MSc; Thinkhamrop, Jadsada MD; Carroli, Guillermo MD; Duc Vy, Nguyen MD; Mignini, Luciano MD; Festin, Mario MD; Prasertcharoensuk, Witoon MD; Limpongsanurak, Sompop MD; Liabsuetrakul, Tippawan MD, PhD; Sirivatanapa, Pannee MD; for the World Health Organization Asymptomatic Bacteriuria Trial Group
The prevalence of asymptomatic bacteriuria in pregnancy ranges from 2–11%, similar to data in nonpregnant populations. However, whereas asymptomatic bacteriuria is relatively benign in nonpregnant populations, 20–30% of culture-positive pregnant women if untreated develop symptomatic disease, often pyelonephritis and its complications.1–3 There is also evidence of increases in both low birth weight and preterm delivery and in the subsequent development of preeclampsia if pregnant women with asymptomatic bacteriuria are not treated.4 Thus, many authorities recommend screening all pregnant women for asymptomatic bacteriuria and treating those with evidence of infection.5 In this respect, antibiotic treatment has been demonstrated to be effective in reducing the risk of subsequent pyelonephritis in women with asymptomatic bacteriuria in pregnancy.6 However, achieving this level of care may be elusive, especially in developing nations.
A Cochrane review updated in October 2006 assessed the efficacy of length of treatment on eradicating asymptomatic bacteriuria in pregnancy. The authors concluded that there was insufficient evidence to evaluate whether 1-day therapy with an antibiotic or treatment for a longer duration (traditionally 7 days) had similar efficacy in eradication asymptomatic bacteriuria of pregnancy.7
Antibiotics used to treat asymptomatic bacteriuria in pregnancy include ampicillin, amoxicillin, trimethoprim-sulfamethoxazole, cephalexin, and nitrofurantoin. However, resistance of Escherichia coli and other uropathogens to ampicillin, trimethoprim-sulfamethoxazole, and first-generation cephalosporins has been steadily increasing, currently approaching 40%.8 Fortunately, little or no changes in bacterial resistance to nitrofurantoin have been noted.9 We designed a World Health Organization (WHO) multicenter trial to compare whether a 1-day treatment with nitrofurantoin (100 mg twice per day) is as effective as a 7-day regimen (100 mg twice per day) in eradicating asymptomatic bacteriuria during pregnancy.
MATERIALS AND METHODS
A hospital-based, multicenter, double-blind, placebo-controlled, randomized controlled trial was conducted in antenatal clinics in Thailand, the Philippines, Viet Nam, and Argentina. Participating centers were Maharaj Nakorn Chiang Mai Hospital (Chiang Mai University), King Chulalongkorn Memorial Hospital (Chulalongkorn University), Songklanagarind Hospital (Prince of Songkla University) and Srinagarind Hospital (Khon Kaen University) in Thailand; Centro Rosarino de Estudios Perinatales, in Rosario, Argentina; National Hospital for Obstetrics and Gynecology, Hanoi, in Viet Nam; and Philippines General Hospital (University of the Philippines), Manila, in the Philippines. All these centers are part of the WHO Maternal and Perinatal Research Network, each having extensive trial experience. This network involves centers throughout the world from both developed and developing countries.
Pregnant women seeking antenatal care visits at the participating centers between March 2004 and March 2007 and capable of giving informed consent were recruited at gestational ages between 12 weeks and 32 weeks. Eligibility was asymptomatic bacteriuria with the microorganism sensitive to nitrofurantoin, plus absence of symptoms suggesting urinary tract infection (UTI), no history of treatment for UTI during the current pregnancy, presence of an underlying disease that required continuous steroids and/or antibiotic treatment, antibiotic hypersensitivity, or any hematologic disease including glucose-6-phosphate dehydrogenase (G6PD) deficiency. An additional exclusion criteria required by the Argentinean Center was the ineligibility of women with asymptomatic bacteriuria, due to Streptococcus Group B, where a specific antibiotic treatment regimen was required.
Instructions (written and oral) were provided by trained research assistants at each participating center and included how to appropriately collect midstream urine samples into sterile containers. After collection, a microbiologist or a trained research assistant tested the urine sample with Orion Diagnostica's Uricult dipslide (Orion Diagnostica Oy, Espoo, Finland), incubating the dipslide at 35°C to 37°C for 24 hours while the remainder of the sample was refrigerated. Dipslide culture was chosen because it provides results comparable to standard loop inoculation methods.10 We also evaluated the diagnostic performance of the dipslide, comparing it with urine culture using the data from Argentina, where all pregnant women had standard urine culture performed.11 The dipslide was evaluated by the microbiologist the following day. If the results were positive (100,000 or more colony-forming units per milliliter), the pathologic findings were corroborated by plating a sample from refrigerated urine on both Blood and MacConkey agar plates, the colonies isolated tested for antibiotic susceptibility. These women were then formally diagnosed as having asymptomatic bacteriuria, became eligible for random assignment. All participating microbiologists were trained in the dipslide technique, and their laboratories were routinely audited. Note that glucose-6 phosphate dehydrogenase (G6PD) deficiency is relatively common in Southeast Asia, particularly in Thailand, where the reported prevalence is 11.1% in male and 5.8% in female cord blood samples.12
Recruited women with asymptomatic bacteriuria were screened (Berstein assay) for glucose-6 phosphate dehydrogenase deficiency before receiving any treatment and were excluded from randomization if positive. Before implementing this assay all laboratories were trained and certified for the assay.
Culture-positive women without G6PD deficiency who consented were randomly allocated to receive either 1-day, (intervention group) or 7-day (control group) course of 100-mg capsules of nitrofurantoin to be taken twice daily. Women in the intervention group received a treatment package containing two 100-mg nitrofurantoin capsules for day 1 and 12 of identical placebo, all capsules to be taken, orally, twice daily with food. Women in the control group received an identical treatment package in which all fourteen capsules contained 100-mg nitrofurantoin to be taken twice daily for 7 days. All participants were requested to return to antenatal care clinic 14 days later, at which time a dipslide and regular urine culture were repeated and compliance determined by recording the number of the capsules remaining in the treatment box. Adverse effects, identified through interviews, were recorded on standard forms. Additional therapy, if required, was dictated by the new culture and sensitivity results, treatment continuing until cultures were negative. All participating women received standard antenatal care provided by the clinics, and their clinical courses were monitored until delivery to assess for pyelonephritis, preterm delivery, and low birth weight.
Randomization codes were conducted independently for each study site by the statistical unit at the Department of Reproductive Health and Research, the World Health Organization, Geneva, Switzerland. The random allocation sequence was generated using computer-generated random numbers with randomly varying blocks of 6–8 (SAS software, SAS Institute, Inc., Cary, NC). The random allocation was concealed by using sealed, opaque treatment boxes numbering sequentially using the sequence described above.
The primary maternal outcome was bacteriologic cure after antibiotic treatment, which was assessed by a urine culture 14 days after the initiation of treatment. Secondary outcomes included the incidences of symptomatic UTI, preterm delivery, and adverse effects. The neonatal outcomes were low birth weight and congenital malformations. Pyelonephritis was diagnosed when systemic symptoms or signs such as fever, chills, nausea, vomiting, and flank pain occurred in the presence of bacteriuria. Symptoms of lower tract infection (ie, frequency and dysuria) might or might not be present.13 Preterm delivery was documented when delivery occurred before 37 completed weeks.14 Low birth weight was defined as birth weight less than 2,500 g.14
Sample size was calculated for the primary outcome. Considering this trial as an equivalence one, we estimated that regimens would be 80% effective. At 95% confidence level and a power of 90%, an acceptable margin of noninferiority in effectiveness of treatment was 7.5%, for a one-tailed test. The required number of women per treatment group and the number of women to be screened (based on 5% prevalence) were 446 and 17,840, respectively.15 We chose to use a one-tailed test because we want to confirm that a 1-day regimen of nitrofurantoin is not inferior to the 7-day regimen. A sample size of screening 18,000 pregnant women was planned.
All centers adhered to the standardized trial procedures, including the use of data collection forms. Completed data collection forms from all participating hospitals except CREP, Rosario, Argentina were sent to Khon Kaen University monthly; data entry and audit were performed continuously. All data were entered and cleaned and queries checked immediately with the local investigators. At CREP, data entry and audit were also performed continuously and update databases were sent to Khon Kaen University every three months.
Descriptive statistics were used to describe characteristics of the participants in each arm. The results were reported as intention to treat and with comparisons between 1-day and 7-day regimens of nitrofurantoin treatment. Relative risk (RR) (or cure rate ratio) and risk difference with 95% confidence intervals (CIs) were used to compare the differences of categorical outcomes. Mean differences with 95% CI were used for continuous outcomes. The trial was approved by the Scientific and Ethical Review Group of the UNDP/UNFPA/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction, the WHO Ethics Review Committee and the institutional review boards of the participating centers.
A total of 25,326 women were screened, of whom 24,430 were eligible. There were 1,827 women with positive dipslide for whom urine culture confirmation was sought. Of these women's cultures, 1,248 returned positive, giving the overall prevalence of 5.1%. However, only 778 of 1,248 women were recruited into the trial. Main reasons for failure to recruit eligible women included nitrofurantoin resistance (32.2%), loss to follow-up before recruitment could be initiated (22.9%), failure to be in a gestational age range of 12–32 weeks (12.5%), presence of G6PD deficiency or the screening result was unavailable (11.2%), and ineligibility criteria discovered only after the culture had been performed (10.6%). The prevalence of G6PD deficiency among women with positive urine culture was approximately 4%.
We allocated 386 and 392 women to 1-day and 7-day regimens, respectively. Three hundred seventy-seven (97.7%) women in the 1-day regimen group and 382 (97.4%) women in the 7-day regimen group returned for their 2-week follow-up urine culture. Delivery outcomes were available in 354 (91.7%) and 349 (89.0%) women in the 1-day and 7-day regimen group, respectively (Fig. 1).
The sociodemographic characteristics of each group including age, gravidity, and parity were very similar at entry. Compliance was both similar and excellent; 98% of the women in both groups ingested all their tablets (Table 1).
The causative pathogens for asymptomatic bacteriuria at recruitment in both groups were also similar. Escherichia coli was the most common causative bacteria at close to 50%, followed by Staphylococcus saprophyticus, Enterococcus spp, Streptococcus Group B, Staphylococcus aureus, Enterobacter spp, and Klebsiella spp (Table 2). However, when viewing the data, recall that Argentinean women with asymptomatic bacteriuria due to Streptococcus Group B were excluded from the trial, because their protocols required all such patients to receive a full 7-day course of antibiotics. After carefully reviewing the data, we realized that the causative bacteria in 23 recruited patients was Staphylococcus coagulase negative, which according to the protocol should not be considered as pathogens. Therefore these 23 patients were inappropriately recruited.
Bacteriologic cure rates evaluated at 14 days after treatment were 75.7% and 86.2% for 1-day and 7-day antibiotic treatment regimens, respectively. The cure rate difference was –10.5% (95% CI −16.1% to −4.9%) (Table 3 and Fig. 2) and the cure rate ratio was 0.88 (95% CI 0.82–0.94) (Table 3). Sensitivity analysis, in which all contaminated urine culture results were assigned as either cure or not cure did not change the cure rate difference or cure rate ratio. Sensitivity analysis performed by excluding 23 inappropriately recruited patients did not change the cure rate difference or cure rate ratio. The 1-day regimen was thus statistically significantly less effective than the standard 7-day regimen. To follow the intention-to-treat principle, we did not exclude these 23 inappropriately recruited patients from the analysis.
During the 2 weeks that followed initiation of treatment, five women in the 1-day and six women in the 7-day group developed symptomatic UTI. The difference was not statistically significant. Similarly, from day 14 through delivery, there were 10 and 12 cases of symptomatic UTI in the 1-day and 7-day regimen groups, respectively. Again, the difference was not statistically significant. Overall, symptomatic UTIs were diagnosed in 15 (5+10) and 18 (6+12) women in the 1-day and 7-day regimen respectively, RR 0.86 (95% CI 0.44–1.67).
Incidences of preterm delivery (11.0% [39 of 353] compared with 8.9% [31 of 349]] and low birth weight (13.2% [48 of 364] compared with 8.0% [28 of 350]) were higher in the 1-day regimen group but the differences were not statistically significant. There was also no statistically significant difference in the risk of congenital malformations (1.4% [5 of 364] compared with 1.1% [4 of 350]) between the two groups. However, mean birth weight and mean gestational age at delivery were statistically significantly lower in the 1-day regimen group, 3,059 g compared with 3,159 g (mean difference –100, 95% CI –181 to –18) and 38.4 weeks compared with 38.7 weeks (mean difference –0.3, 95% CI –0.7 to –.005), respectively (Table 4). There was less nausea, headache, and other adverse effects but more flatulence in the 1-day regimen group, but the differences were not statistically significant (Table 5).
These data demonstrate that 1-day nitrofurantoin therapy is significantly less effective than the standard 7-day regimen in achieving bacteriologic cure when both approaches were evaluated 14 days after initiating treatment. Patients in the 1-day regimen group also had higher incidences of preterm delivery and low birth weight (not statistically significant), and lower mean birth weight and mean gestational age at delivery (statistically significant). However, the rates of symptomatic UTI were similar in both groups. Interestingly, the rate of clinical UTI was below 5%. Adverse effects were slightly lower in the 1-day regimen group (not statistically significant).
The strength of our findings are underscored by the study design, a multicenter double-blinded placebo-controlled randomized trial. Furthermore, all microbiologists were trained in dipslide technique, the laboratories and data regularly audited to assure high quality, and site visits were performed by WHO officials. There were also two investigators' meetings where trial problems were discussed and resolved. Loss to follow-up after randomization and the number of contaminated specimens was minimal, the latter problem being also controlled with a sensitivity analysis, where the assumed best (all contaminated negative) or worst (all contaminated positive) did not change the result of primary outcome. All the factors summarized above combine to strengthen the validity of our conclusions.
We could not reach our target sample size of 992 women because we could not recruit 32% of the potential eligible subjects, as already described in the results section. However, we recruited 778 women, which represents 85% of our target sample size, permitting us demonstrate that 1-day nitrofurantoin regimen is significantly less effective than the standard 7-day regimen.
We update the Cochrane review7 by incorporating our data into the meta-analysis. Comparing “no cure rate” between 1-day and 4–7 day regimens using the random effects model, adding the data from our study changes the RR and 95% CI from 1.41 (0.80–2.50) to 1.42 (0.89–2.28). Although the confidence interval is narrower, we cannot conclude from the new meta-analysis that the 4–7 day regimen is significantly more effective than the 1-day regimen. There was significant heterogeneity in these meta-analyses, and unfortunately, only two included studies (Masterton et al16 1985 and ours) had adequate allocation concealment, whereas all the other nine studies were classified as inadequate, unclear, or not used. We therefore calculated the pooled RR for these two trials with adequate methodology using a fixed effects model and found that the risk of no cure in the 1-day regimen is significantly higher than the 7-day regimen (RR 2.10, 95% CI 1.44–3.05).
In summary, our trial demonstrates that the 1-day regimen of nitrofurantoin is significantly less effective than the 7-day regimen. Adding these results to the other trial with adequate methodology (Masterton et al16 1985 demonstrates that the 1-day regimen is significantly less effective than the 7-day regimen. We therefore conclude that women with asymptomatic bacteriuria in pregnancy should be treated by the standard 7-day regimen of nitrofurantoin, if such treatment is selected, until more data becomes available testing 7-day compared with 3- or 5-day regimens.
1. Andrews W, Cox S, Gilstrap LC. Urinary tract infections in pregnancy. Int Urogynecol J 1990;1:155–63.
2. Stenqvist K, Dahlén-Nilsson I, Lidin-Janson G, Lincoln K, Oden A, Rignell S, et al. Bacteriuria in pregnancy. Frequency and risk of acquisition Am J Epidemiol 1989;129:372–9.
3. Sweet RL. Bacteriuria and pyelonephritis during pregnancy. Semin Perinatol 1977;1:25–40.
4. Conde-Agudelo A, Villar J, Lindheimer M. Maternal infection and risk of preeclampsia: systematic review and metaanalysis. Am J Obstet Gynecol 2008, 198: 7–22.
5. American College of Obstetricians and Gynecologists. Antimicrobial therapy for obstetric patients. ACOG Educational Bulletin 245. Washington (DC): ACOG; 1998.
6. Smaill F, Vazquez JC. Antibiotics for asymptomatic bacteriuria in pregnancy. Cochrane Database of Systematic Reviews 2007, Issue 2. Art. No.: CD000490. DOI: 10.1002/14651858.CD000490.pub2.
7. Villar J, Widmer M, Lydon-Rochelle MT, Gülmezoglu AM, Roganti A. Duration of treatment for asymptomatic bacteriuria during pregnancy. Cochrane Database of Systematic Reviews 2000, Issue 2. Art. No.: CD000491. DOI: 10.1002/14651858. CD000491.
8. Gupta K, Hooton TM, Stamm WE. Increasing antimicrobial resistance and the management of uncomplicated community-acquired urinary tract infections. Ann Intern Med 2001;135:41–50.
9. Guay DR. An update on the role of nitrofurans in the management of urinary tract infections. Drugs 2001;61:353–64.
10. European Confederation of Laboratory Medicine. European urinalysis guidelines. Scand J Clin Lab Invest Suppl. 2000;231: 1–86.
11. Mignini L, Carroli G, Abalos E, Widmer M, Lumbiganon P, Amigot Set al. Accuracy of diagnostic tests to detect asymptomatic bacteriuria during pregnancy. In press 2008.
12. Nuchprayoon I, Sanpavat S, Nuchprayoon S. Glucose-6-Phosphate dehydrogenase (G-6-PD) mutation in Thailand: G6PD Viangchan (871G>A) is the most common deficiency variant in the Thai population. Hum Mutat 2002;19:185.
13. Delzell JE Jr, Lefevre ML. Urinary tract infections during pregnancy. Am Fam Physician 2000;61:713–21.
14. Department of Reproductive Health and Research. Managing complications in pregnancy and childbirth: A guide for midwives and doctors. Geneva (Switzerland): WHO; 2000.
15. Jones B, Jarvis P, Lewis JA, Ebbutt AF. Trials to assess equivalence: the importance of rigorous methods [published erratum appears in BMJ 1996;313:550]. BMJ 1996;313:36–9.
16. Masterton RG, Evans DC, Strike PW. Single-dose amoxycillin in the treatment of bacteriuria in pregnancy and the puerperium—a controlled clinical trial. Br J Obstet Gynaecol 1985;92: 498–505.
© 2009 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.