Antibióticos para la bacteriuria asintomática en el embarazo
Resumen
Antecedentes
La bacteriuria asintomática es una infección bacteriana de la orina sin los síntomas típicos que se asocian con una infección urinaria, y ocurre entre el 2% y el 15% de los embarazos. Si nos e trata, hasta el 30% de las madres desarrollarán pielonefritis aguda. La bacteriuria asintomática se ha asociado con el bajo peso al nacer y el parto prematuro. Ésta es una actualización de una revisión publicada por última vez en 2015.
Objetivos
Evaluar el efecto del tratamiento antibiótico para la bacteriuria asintomática sobre el desarrollo de pielonefritis y el riesgo de bajo peso al nacer y de parto prematuro.
Métodos de búsqueda
Para esta actualización se hicieron búsquedas en el registro de ensayos del Grupo Cochrane de Embarazo y Parto (Cochrane Pregnancy and Childbirth), ClinicalTrials.gov, en la Plataforma de registros internacionales de ensayos clínicos (ICTRP) de la OMS (4 de noviembre de 2018) y en listas de referencias de estudios recuperados.
Criterios de selección
Ensayos controlados aleatorizados (ECA) que compararon el tratamiento con antibióticos con placebo o ningún tratamiento, en embarazadas con bacteriuria asintomática encontrada durante el cribado prenatal. Los ensayos que utilizaron un diseño de ECA grupal y cuasialeatorizado fueron elegibles para inclusión, al igual que los ensayos publicados en forma de resumen o carta, no así los estudios cruzados (cross‐over).
Obtención y análisis de los datos
Dos autores de la revisión, de forma independiente, evaluaron los ensayos para inclusión y el riesgo de sesgo, extrajeron los datos y verificaron su exactitud. La calidad de la evidencia se evaluó con los criterios GRADE.
Resultados principales
Se incluyeron 15 estudios con 2000 mujeres. El tratamiento con antibióticos, comparado con placebo o ningún tratamiento, puede reducir la incidencia de pielonefritis (riesgo relativo [RR] promedio 0,24; intervalo de confianza [IC] del 95%: 0,13 a 0,41; 12 estudios, 2017 mujeres; evidencia de certeza baja). El tratamiento con antibióticos se puede asociar con una reducción en la incidencia de partos prematuros (RR 0,34; IC del 95%: 0,13 a 0,88; tres estudios, 327 mujeres; evidencia de certeza baja) y de recién nacidos con bajo peso al nacer (RR promedio 0,64; IC del 95%: 0,45 a 0,93; seis estudios, 1437 recién nacidos; evidencia de certeza baja). Puede haber una reducción de la bacteriuria persistente en el momento del parto (RR promedio 0,30; IC del 95%: 0,18 a 0,53; cuatro estudios; 596 mujeres), pero los resultados no fueron concluyentes en cuanto a los resultados neonatales adversos graves (RR promedio 0,64; IC del 95%: 0,23 a 1,79; tres estudios; 549 recién nacidos). Hubo datos muy limitados para calcular el efecto de los antibióticos sobre otros resultados del recién nacido, y en pocas ocasiones se describieron los efectos adversos maternos.
En general, solo un ensayo se consideró con bajo riesgo de sesgo en todos los dominios; los otros 14 estudios se evaluaron como riesgo de sesgo alto o incierto. Muchos estudios carecieron de una descripción adecuada de los métodos y solo fue posible valorar el riesgo de sesgo como incierto, pero en la mayoría de los estudios al menos un dominio se evaluó como alto riesgo de sesgo. La calidad de las evidencia para los tres resultados primarios se evaluó mediante el programa informático GRADE y se encontró evidencia de baja certeza para la pielonefritis, el parto prematuro y el peso al nacer inferior a 2500 g.
Conclusiones de los autores
El tratamiento con antibióticos puede ser efectivo para reducir el riesgo de pielonefritis en el embarazo, pero la confianza en la estimación del efecto es limitada debido a la baja certeza de la evidencia. Puede haber una reducción del parto prematuro y del bajo peso al nacer con el tratamiento con antibióticos, lo que es consistente con las teorías acerca de la función de la infección en los resultados adversos del embarazo, pero, nuevamente, la confianza en el efecto es limitada debido a la baja certeza de la evidencia.
Las implicaciones de investigación identificadas en esta revisión incluyen la necesidad de una evaluación actualizada de la relación coste‐efectividad de los algoritmos de diagnóstico y más evidencia para saber si existe un grupo de mujeres de bajo riesgo con pocas probabilidades de beneficiarse del tratamiento de la bacteriuria asintomática.
PICO
Resumen en términos sencillos
Antibióticos para la infección bacteriana en la orina durante el embarazo cuando no hay síntomas
¿Cuál es el problema?
¿Puede la administración de antibióticos a las embarazadas que tienen una infección urinaria, pero que no presentan síntomas, mejorar los resultados para las mujeres y sus recién nacidos?
¿Por qué es esto importante?
La infección bacteriana de la orina sin los síntomas típicos que se asocian con una infección urinaria (bacteriuria asintomática) ocurre en cierto número de embarazos (del 2% al 15%). Debido a los cambios que se producen en su cuerpo, las embarazadas tienen mayores probabilidades de desarrollar una infección renal (pielonefritis) si tienen una infección urinaria. La infección también puede contribuir a que el recién nacido sea prematuro (parto antes de las 37 semanas) o con bajo peso al nacer (pesa menos de 2500 g [5,5 libras]).
¿Qué evidencia se encontró?
Se encontraron 15 estudios controlados aleatorizados que incluyeron más de 2000 embarazadas con infección urinaria, pero sin síntomas. Los antibióticos pueden ser efectivos para reducir la incidencia de infección renal en la madre (12 estudios, 2017 mujeres) y eliminar la infección de la orina (cuatro estudios, 596 mujeres). También pueden reducir la incidencia de partos prematuros (tres estudios, 327 mujeres) y de recién nacidos con bajo peso al nacer (seis estudios, 1437 recién nacidos). Ninguno de los estudios evaluó adecuadamente los efectos adversos del tratamiento con antibióticos para la madre o el recién nacido, y a menudo la forma en la que se realizó el estudio no se describió bien.
Se evaluaron las tres medidas de resultado principales con el enfoque GRADE y se encontró evidencia de certeza baja de que el tratamiento con antibióticos puede prevenir la pielonefritis, el parto prematuro y el peso al nacer inferior a 2500 g.
¿Qué significa esto?
El tratamiento con antibióticos puede reducir el riesgo de infecciones renales en las embarazadas que presentan una infección urinaria, pero que no muestran síntomas de infección. Los antibióticos también pueden reducir la probabilidad de que un recién nacido nazca demasiado pronto o que tenga un bajo peso al nacer. Sin embargo, debido a la baja certeza de las evidencia es difícil establecer conclusiones; se necesitan más estudios de investigación.
Authors' conclusions
Summary of findings
| Antibiotics compared to no treatment for asymptomatic bacteriuria in pregnancy | ||||||
| Patient or population: pregnant women with asymptomatic bacteriuria Setting: hospital‐based clinics in North America, UK and Ireland, Australia; hospital and community midwifery practices in the Netherlands Intervention: antibiotics Comparison: no treatment | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | № of participants | Certainty of the evidence | Comments | |
| Risk with no treatment | Risk with antibiotics | |||||
| Development of pyelonephritis | Study population | RR 0.24 | 2017 | ⊕⊕⊝⊝ | ||
| 199 per 1000 | 48 per 1000 | |||||
| Preterm birth < 37 weeks | Study population | RR 0.34 | 327 | ⊕⊕⊝⊝ | ||
| 174 per 1000 | 59 per 1000 | |||||
| Birthweight < 2500 g | Study population | RR 0.64 | 1437 | ⊕⊕⊝⊝ | ||
| 136 per 1000 | 87 per 1000 | |||||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| aWe downgraded 1 level for serious limitations in study design: most of the trials contributing outcome data either had important design limitations related to lack of allocation concealment and lack of blinding, or there were insufficient details provided in the report to assess risk of bias. Many of the studies were performed in the 1960s and 1970s, prior to more rigorous study designs and reporting standards. | ||||||
Background
Description of the condition
Asymptomatic bacteriuria, generally defined as true bacteriuria in the absence of specific symptoms of acute urinary tract infection, is a common finding, and occurs in 2% to 15% of all pregnancies (Ipe 2013).
While rates from recent observational studies fall within this range (Abdel‐Aziz 2017; Bandyopadhyay 2005; Celen 2011; Fatima 2006; Kazemier 2015; McIsaac 2005; McNair 2000; Mohammad 2002; Olamijulo 2016; Tugrul 2005), rates of over 20% are reported in studies from some low‐income countries (Ajayi 2012; Rizvi 2011; Tadesse 2014). The prevalence of asymptomatic bacteriuria was reported to be as high as 86.6% in a population from Nigeria that included Staphylococcus aureus, a possible contaminant, as a uropathogen (Akerele 2001). Rates were reported to be higher in HIV positive women in Nigeria (Awolude 2010; Ezechi 2013), but not in a study from South Africa (Widmer 2010). In a large retrospective study, the strongest predictor of bacteriuria was an antepartum urinary tract infection (Pastore 1999). A study from an electronics factory in China found an association between urinary tract infections in pregnancy and frequency of voiding; voiding three or more times a shift was protective (Su 2009). In a study from Iran, there was an association between infection, frequency of sexual intercourse, and genital hygiene practices (Amiri 2009). The prevalence of infection is related to socioeconomic status (Haider 2010; Turck 1962; Whalley 1967), although this may not always apply (Awoleke 2015; Kovavisarach 2009). Other contributing factors, recognised as associated with an increased risk for bacteriuria, include diabetes and anatomical abnormalities of the urinary tract.
The original criterion for diagnosing asymptomatic bacteriuria was a count of more than 100,000 bacteria/mL on two consecutive clean‐catch samples (Kass 1960a). The detection of more than 100,000 bacterial/mL in a single voided midstream urine sample is accepted as an adequate and more practical alternative, although there is only an 80% probability the woman has true bacteriuria; this increases to 95% if two or more consecutive cultures are positive for the same organism (Kass 1960a). Because the performance of rapid urine screening tests in pregnancy is poor, quantitative culture remains the gold standard for diagnosis (Bachman 1993; Garingalo‐Molina 2000; McNair 2000; Mignini 2009; Rogozińska 2016).
Escherichia coli is the most common pathogen associated with asymptomatic bacteriuria, representing up to 80% of isolates (Ipe 2013). Other organisms include other gram‐negative bacteria, e.g. Klebsiella spp., Proteus mirabilis, and group B streptococci. These bacteria colonise the vaginal introitus and periurethral area. Uropathogenic gram‐negative bacteria possess specific virulence factors that enhance both colonisation and invasion of the urinary tract, for example, the P‐fimbriae of certain strains of E. coli that allow for adherence to uroepithelial cells (Eisenstein 1988; Stenqvist 1987). Some strains of E. coli isolated from pregnant women with asymptomatic bacteriuria have a similar virulence pattern to strains from women with symptomatic infections (Lavigne 2011), but this does not always hold true (Stenqvist 1987). While Staphylococcus saprophyticus is recognised as a urinary pathogen, other species of Staphylococci, including Staphylococcus aureus, may reflect contamination rather than a true infection, and prevalence data where the number of Staphylococcus spp. is high, are difficult to interpret. Maternal urinary tract infection with group B streptococci is associated with vaginal colonisation with the organism, and antibiotic treatment during labour is recommended to prevent early onset neonatal group B streptococcal disease (Allen 2012).
While asymptomatic bacteriuria in non‐pregnant women is generally benign, obstruction to the flow of urine in pregnancy leads to stasis, and increases the likelihood that pyelonephritis will complicate asymptomatic bacteriuria (Nicolle 2014). Mechanical compression from the enlarging uterus is the principal cause of hydroureter and hydronephrosis, but smooth muscle relaxation induced by progesterone may also play a role (Sobel 1995). Differences in urine pH and osmolality and pregnancy‐induced glycosuria and aminoaciduria may facilitate bacterial growth. If asymptomatic bacteriuria in pregnancy is untreated, it has generally been accepted that up to 20% to 30% of mothers will develop acute pyelonephritis (Nicolle 2015). Current estimates are difficult to identify because there is almost universal implementation of screening and treatment, however, a recent study from a low‐risk population in the Netherlands, where screening never became standard, reported a rate of pyelonephritis of 2.4% (Kazemier 2015). Clinical signs of pyelonephritis include fever, chills, costovertebral tenderness, dysuria, and frequency. Nausea and vomiting are common, and if infection is associated with bacteraemia, women may present with high fever, shaking chills, and low blood pressure. Maternal complications include maternal respiratory insufficiency, septicaemia, renal dysfunction, and anaemia (Hill 2005; Wing 2014). In the pre‐antibiotic era, acute pyelonephritis was associated with a 20% to 50% incidence of preterm birth. A prospective longitudinal study, in the era of routine screening, over a two‐year period from 2000 to 2001 in Texas, reported an incidence of acute pyelonephritis in pregnancy of 1.4% (Hill 2005). From an 18‐year retrospective review, in an era of routine screening and treatment for asymptomatic bacteriuria, the incidence of acute pyelonephritis in pregnancy was 0.5%, and pyelonephritis was associated with preterm birth (odds ratio (OR) 1.3, 95% confidence interval (CI) 1.2 to 1.5); women with pyelonephritis were more likely to be black or Hispanic, young, less educated, initiate prenatal care late, and smoke (Wing 2014). An association between acute pyelonephritis and preterm birth was described in a retrospective study of 219,612 deliveries from Israel (OR 2.6, 95% CI 1.7 to 3.9; Farkash 2012).
An association between asymptomatic bacteriuria, low birthweight, and preterm birth has been described since the earliest studies of Kass (Kass 1960a), but population‐based studies have produced conflicting results. A retrospective study from Israel, which controlled for confounders, showed an association between asymptomatic bacteriuria and preterm birth (OR 1.9, 95% CI 1.7 to 2.0; Sheiner 2009); in contrast, findings from the Cardiff Birth Survey reported that asymptomatic bacteriuria, adjusted for demographic and social factors, was not associated with preterm birth (OR 1.2, 95% CI 0.9 to 1.5; Meis 1995). However, when preterm deliveries were categorised into those medically indicated because of complications of pregnancy (e.g. antepartum haemorrhage, eclampsia, or renal disease) and spontaneous preterm births, there was a significant association between bacteriuria and medically‐indicated preterm deliveries (OR 2.03, 95% CI 1.5 to 2.8), but not for spontaneous preterm births (OR 1.07, 95% CI 0.78 to 1.46). The authors concluded that if asymptomatic bacteruria does not progress to pyelonephritis, it is not associated with preterm birth (Meis 1995a).
Description of the intervention
The goal of treatment for asymptomatic bacteriuria is to treat and clear the infection. The urinary bacterial isolate should be susceptible to the antibiotic chosen, the length of treatment should be adequate, adherence should be assured, and the drug should have favourable pharmacokinetic parameters. The treatment should be safe in pregnancy, for both the mother and developing fetus. Many antibiotics have been used to treat bacteriuria, including sulphonamides or sulphonamide‐containing combinations, penicillins, cephalosporins, fosfomycin, and nitrofurantoin. However, not all the antibiotics previously evaluated are currently available, e.g. certain sulphonamides and methenamine, or recommended during pregnancy, e.g. tetracycline. Increasing bacterial resistance of urinary pathogens can make it difficult to select an appropriate regimen, especially in under‐resourced settings, where facilities for urine culture and antimicrobial susceptibility testing are limited (Assefa 2008; Enayat 2008; Hernandez Blas 2007; Rizvi 2011; Tadesse 2014). There is no evidence that non‐pharmacological interventions, e.g. cranberry juice, are effective (Wing 2008), although no data exist to suggest the use of cranberry has any harmful effects on pregnancy (Heitmann 2013).
How the intervention might work
Urinary pathogens causing asymptomatic bacteriuria are similar to those causing pyelonephritis; antibiotic treatment and eradication of bacteriuria is expected to prevent ascending urinary tract infection and the development of clinical pyelonephritis.
The relationship between asymptomatic bacteriuria, low birthweight, and preterm birth is controversial, since a biological mechanism for an association between preterm labor and asymptomatic bacteriuria has not been established. Microbial‐induced preterm labor is mediated by an inflammatory process (Goldenberg 2000; Romero 2014). Microorganisms and their products are sensed by pattern‐recognition receptors, such as toll‐like receptors (TLRs), which induce the production of chemokines, prostaglandins, and proteases, leading to the onset of labour. While this mechanism has been well defined for ascending intra‐amniotic infection, there has been no recent research to explore the mechanisms through which asymptomatic bacteriuria might exert adverse pregnancy outcomes.
Why it is important to do this review
Screening for, and treating asymptomatic bacteriuria in pregnancy, has become a standard of obstetric care. While most antenatal guidelines include routine screening for asymptomatic bacteriuria, questions have been raised about the quality of the evidence on which these guidelines are based, and the lack of data on the harms of screening (Angelescu 2016; Moore 2018). Using a decision analysis, screening for, and treating asymptomatic bacteriuria to prevent pyelonephritis, has been shown to be cost‐effective over a wide range of estimates, although the cost‐benefit is diminished if the rate of asymptomatic bacteriuria is less than 2% (Rouse 1995; Wadland 1989). The low prevalence of infection in certain populations, the cost of different screening tests, and uncertainty about the benefits of treatment in decreasing adverse outcomes of pregnancy have been used to argue against screening and treatment as universal recommendations; preventing unnecessary antibiotic use has become an important aspect of programmes to decrease the development of antimicrobial resistance. A rigorous evaluation of studies of the effect of treatment of asymptomatic bacteriuria could provide clarity around these issues. This is an update of a review last published in 2015 (Smaill 2015).
Objectives
To assess the effect of antibiotic treatment for asymptomatic bacteriuria on the development of pyelonephritis and the risk of low birthweight and preterm birth.
Methods
Criteria for considering studies for this review
Types of studies
We included randomised controlled trials and quasi‐randomised trials (e.g. alternation). Cluster‐randomised trials were eligible for inclusion. Cross‐over trials were not eligible for inclusion. Trials published in abstract form only, or as a letter, were eligible for inclusion.
Types of participants
Pregnant women found, on antenatal screening, to have asymptomatic bacteriuria, as defined by the study authors, at any stage of pregnancy.
Types of interventions
We included studies if any antibiotic regimen was compared with placebo or no treatment for asymptomatic bacteriuria.
Types of outcome measures
Primary outcomes
-
Development of pyelonephritis
-
Preterm birth less than 37 weeks
-
Birthweight less than 2500 g
Secondary outcomes
-
Persistent bacteriuria
-
Neonatal mortality or other serious adverse neonatal outcome
-
Maternal side effects
-
Costs, as defined by trial authors
-
Birthweight
-
Gestational age
-
Women's satisfaction, as measured by trial authors
Persistent bacteriuria was defined as bacteriuria persisting at the time of delivery.
We used the World Health Organization's definition of prematurity: a baby born before 37 completed weeks of gestation (Blencowe 2012).
Search methods for identification of studies
The following methods section of this review is based on a standard template used by Cochrane Pregnancy and Childbirth.
Electronic searches
For this update, we searched Cochrane Pregnancy and Childbirth’s Trials Register, by contacting their Information Specialist (4 November 2018).
The Register is a database containing over 25,000 reports of controlled trials in the field of pregnancy and childbirth. It represents over 30 years of searching. For full, current search methods used to populate Pregnancy and Childbirth’s Trials Register, including the detailed search strategies for CENTRAL, MEDLINE, Embase, and CINAHL; the list of handsearched journals and conference proceedings; and the list of journals reviewed via the current awareness service; please follow this link.
Briefly, Cochrane Pregnancy and Childbirth’s Trials Register is maintained by their Information Specialist, and contains trials identified from:
-
monthly searches of the Cochrane Central Register of Controlled Trials (CENTRAL);
-
weekly searches of MEDLINE Ovid;
-
weekly searches of Embase Ovid;
-
monthly searches of CINAHL EBSCO;
-
handsearches of 30 journals and the proceedings of major conferences;
-
weekly current awareness alerts for a further 44 journals, plus monthly BioMed Central email alerts.
Search results are independently screened by two people, and the full text of all relevant trial reports identified through the searching activities described above is reviewed. Based on the intervention described, each trial report is assigned a number that corresponds to a specific Pregnancy and Childbirth review topic (or topics), and is then added to the Register. The Information Specialist searches the Register for each review using this topic number rather than keywords. This results in a more specific search set that has been fully accounted for in the relevant review sections (Included studies; Excluded studies; Studies awaiting classification; Ongoing studies).
In addition, we searched ClinicalTrials.gov and the WHO International Clinical Trials Registry Platform (ICTRP) for unpublished, planned, and ongoing trial reports on 4 November 2018, using the search methods detailed in Appendix 1.
Searching other resources
We searched the reference lists of retrieved studies.
We did not apply any language or date restrictions.
Data collection and analysis
For methods used in the previous versions of this review, seeSmaill 1993, Smaill 2007 and Smaill 2015.
For this update, we used the following methods to assess the reports that were identified as a result of the updated search, based on a standard template used by Cochrane Pregnancy and Childbirth.
Selection of studies
Two review authors independently assessed for inclusion all the potential studies identified as a result of the search strategy. We resolved any disagreement through discussion.
Data extraction and management
We designed a form to extract data. If any new studies were included, both review authors independently extracted the data, using the agreed form. We resolved discrepancies through discussion. We entered data into Review Manager 5 software, and checked for accuracy (Review Manager 2014).
If information regarding any of the above had been unclear, we had planned to contact authors of the original reports to provide further details.
Assessment of risk of bias in included studies
Two review authors independently assessed risk of bias for each study, using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). Any disagreement was resolved by discussion.
(1) Random sequence generation (checking for possible selection bias)
For each included study, we described the method used to generate the allocation sequence in sufficient detail to allow an assessment of whether it should produce comparable groups.
We assessed the method as:
-
low risk of bias (any truly random process, e.g. random number table; computer random number generator);
-
high risk of bias (any non‐random process, e.g. odd or even date of birth; hospital or clinic record number);
-
unclear risk of bias.
(2) Allocation concealment (checking for possible selection bias)
For each included study, we described the method used to conceal allocation to interventions prior to assignment, and assessed whether intervention allocation could have been foreseen in advance of, or during recruitment, or changed after assignment.
We assessed the methods as:
-
low risk of bias (e.g. telephone or central randomisation; consecutively numbered sealed opaque envelopes);
-
high risk of bias (open random allocation; unsealed or non‐opaque envelopes, alternation; date of birth);
-
unclear risk of bias.
(3.1) Blinding of participants and personnel (checking for possible performance bias)
For each included study, we described the methods used, if any, to blind study participants and personnel from knowledge of which intervention a participant received. We considered that studies were at low risk of bias if they were blinded, or if we judged that the lack of blinding was unlikely to affect results. We assessed blinding separately for different outcomes or classes of outcomes.
We assessed the methods as:
-
low, high, or unclear risk of bias for participants;
-
low, high, or unclear risk of bias for personnel.
(3.2) Blinding of outcome assessment (checking for possible detection bias)
For each included study, we described the methods used, if any, to blind outcome assessors from knowledge of which intervention a participant received. We assessed blinding separately for different outcomes or classes of outcomes.
We assessed methods used to blind outcome assessment as:
-
low, high, or unclear risk of bias.
(4) Incomplete outcome data (checking for possible attrition bias due to the amount, nature and handling of incomplete outcome data)
For each included study, and for each outcome or class of outcomes, we described the completeness of data, including attrition and exclusions from the analysis. We stated whether attrition and exclusions were reported, and the numbers included in the analysis at each stage (compared with the total randomised participants), reasons for attrition or exclusion where reported, and whether missing data were balanced across groups, or were related to outcomes. Where sufficient information was reported, or could be supplied by the trial authors, we planned to re‐include missing data in the analyses that we undertook.
We assessed methods as:
-
low risk of bias (e.g. no missing outcome data; missing outcome data balanced across groups);
-
high risk of bias (e.g. numbers or reasons for missing data imbalanced across groups; ‘as treated’ analysis done with substantial departure of intervention received from that assigned at randomisation);
-
unclear risk of bias.
(5) Selective reporting (checking for reporting bias)
For each included study, we described how we investigated the possibility of selective outcome reporting bias, and what we found.
We assessed the methods as:
-
low risk of bias (where it was clear that all of the study’s prespecified outcomes and all expected outcomes of interest to the review had been reported);
-
high risk of bias (where not all the study’s prespecified outcomes had been reported; one or more reported primary outcomes were not prespecified; outcomes of interest were reported incompletely and so could not be used; study failed to include results of a key outcome that would have been expected to have been reported);
-
unclear risk of bias.
(6) Other bias
Where identified, we described bias due to problems not covered elsewhere.
(7) Overall risk of bias
We made explicit judgements about whether studies were at high risk of bias, according to the criteria given in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). With reference to (1) to (6) above, we planned to assess the likely magnitude and direction of the bias, and whether we considered it was likely to impact the findings. We planned to explore the impact of the level of bias by undertaking sensitivity analyses ‐ seeSensitivity analysis.
Assessment of the quality of the evidence, using GRADE
For this update, we assessed the quality of the evidence using the GRADE approach (Schünemann 2009). We assessed the quality of the body of evidence relating to the following outcomes, for the main comparison of antibiotic versus no treatment.
-
Development of pyelonephritis
-
Preterm birth less than 37 weeks
-
Birthweight less than 2500 g
We used GRADEpro GDT software to import data from Review Manager 5 software, to create a ’Summary of findings’ table (GRADEpro GDT; Review Manager 2014). We produced a summary of the intervention effect and a measure of quality for each of the above outcomes, using the GRADE approach. The GRADE approach uses five considerations (study limitations, consistency of effect, imprecision, indirectness, and publication bias) to assess the quality of the body of evidence for each outcome. The evidence can be downgraded from 'high quality' by one level for serious (or by two levels for very serious) limitations, depending on assessments for risk of bias, indirectness of evidence, serious inconsistency, imprecision of effect estimates, or potential publication bias.
Measures of treatment effect
Dichotomous data
For dichotomous data, we presented results as summary risk ratio with 95% confidence intervals.
Continuous data
We used the mean difference if outcomes were measured in the same way between trials. In future updates, we plan to use the standardised mean difference to combine trials that measure the same outcome, but use different methods.
Unit of analysis issues
Cluster‐randomised trials
If any were identified, we planned to include cluster‐randomised trials in the analyses along with individually‐randomised trials. In future updates of this review, if any cluster‐randomised trials are included, we will adjust their sample sizes using the methods described in the Cochrane Handbook for Systematic Reviews of Interventions, Section 16.3.4 or 16.3.6, using an estimate of the intracluster correlation co‐efficient (ICC) derived from the trial (if possible), from a similar trial, or from a study of a similar population. If we use ICCs from other sources, we will report this, and conduct sensitivity analyses to investigate the effect of variation in the ICC. If we identify both cluster‐randomised trials and individually‐randomised trials, we plan to synthesise the relevant information. We will consider it reasonable to combine the results from both if there is little heterogeneity between the study designs, and the interaction between the effect of the intervention and the choice of randomisation unit is considered to be unlikely.
We will also acknowledge heterogeneity in the randomisation unit, and perform a sensitivity analysis to investigate the effects of the randomisation unit.
Dealing with missing data
For included studies, we noted levels of attrition. In future updates, if more eligible studies are included, we will explore the impact of including studies with high levels of missing data in the overall assessment of treatment effect, by completing a sensitivity analysis.
For all outcomes, we carried out analyses, as far as possible, on an intention‐to‐treat basis, and where reasonable, we attempted to include all participants randomised to each group in the analyses.
Assessment of heterogeneity
We assessed statistical heterogeneity in each meta‐analysis using the I² and Chi² statistics, and Tau². We regarded heterogeneity as substantial, if an I² was greater than 30%, and either a Tau² was greater than zero, or there was a low P value (less than 0.10) in the Chi² test for heterogeneity. If we identified substantial heterogeneity (above 30%), we explored it by prespecified subgroup analysis.
Assessment of reporting biases
Where there were 10 or more studies in the meta‐analysis, we investigated reporting biases (such as publication bias) using funnel plots. We assessed funnel plot asymmetry visually, and at any suggestion of asymmetry, we planned to perform exploratory analyses to investigate it.
Data synthesis
We carried out statistical analysis using the Review Manager 5 software (Review Manager 2014). We used fixed‐effect meta‐analysis for combining data where it was reasonable to assume that studies were estimating the same underlying treatment effect, i.e. where trials were examining the same intervention, and the trials’ populations and methods were judged sufficiently similar.
If there was clinical heterogeneity sufficient to expect that the underlying treatment effects differed between trials, or if substantial statistical heterogeneity was detected, we used random‐effects meta‐analysis to produce an overall summary, if an average treatment effect across trials was considered clinically meaningful. The random‐effects summary was treated as the average range of possible treatment effects, and we planned to discuss the clinical implications of treatment effects differing between trials. If the average treatment effect was not clinically meaningful, we would not have combined trials. Where we used random‐effects analyses, we presented the results as the average treatment effect with 95% confidence intervals, and the estimates of Tau² and I².
Subgroup analysis and investigation of heterogeneity
If we identified substantial heterogeneity, we investigated it using subgroup analysis and sensitivity analysis. We considered whether an overall summary was meaningful, and if it was, we used random‐effects analysis to produce it.
We carried out the following subgroup analysis to determine whether there was an effect of the duration of antibiotic therapy on the outcomes.
-
Single dose versus no treatment
-
Short course (three to seven days) versus no treatment
-
Intermediate course (three to six weeks) versus no treatment
-
Continuous antibiotic therapy until delivery versus no treatment
We conducted subgroup analyses for the following outcomes.
-
Development of pyelonephritis
-
Preterm birth less than 37 weeks
-
Birthweight less than 2500 g
We assessed subgroup differences by interaction tests available within RevMan 5 (Review Manager 2014). We reported the results of the subgroup analysis quoting the Chi² statistic and P value, and the interaction test I² value.
Sensitivity analysis
We had planned to carry out sensitivity analysis to explore the effect of risk of bias on the overall results, by excluding studies where the overall risk of bias was high or there was insufficient detail provided to judge risk of bias from the analysis, for the primary outcomes.
Results
Description of studies
Results of the search
See: Figure 1.
Study flow diagram
We assessed four new trial reports (Kazemier 2015; NCT03274960;NCT03275623;NL2921), and reassessed one trial report that was awaiting classification from the previous version of the review (Kazemier 2012). We excluded NCT03274960 because the intervention was screening for asymptomatic bacteriuria versus no screening. NCT03275623 is a study of treatment versus no treatment for low level bacteriuria and is ongoing. We included one new study (Kazemier 2015). Kazemier 2012 and NL2921 are additional reports of that study.
Included studies
We included 15 studies, in 22 reports, involving over 2000 women. For details, see Characteristics of included studies. One study enrolled only women with group B streptococci in the urine (Thomsen 1987). Where there was more than one published report that in the opinion of the review authors referred to the same study, we abstracted information from whichever report provided the most detailed information. The earliest published report was from 1960 (Kass 1960), and the most recent from 2015 (Kazemier 2015). Most studies (N = 11) enrolled women prior to 1970. Only Kazemier 2015 reported any potential conflicts of interest. Two studies did not report a funding source (Foley 1987; Mulla 1960). Local or national research funding sources were reported for the majority of the other studies, and several studies described in‐kind support, where the antibiotic was provided by a pharmaceutical company. We did not find any cluster‐randomised trials.
Participants
Most women were enrolled from hospital‐based clinics in North America (Elder 1966; Elder 1971; Gold 1966; Kass 1960; Mulla 1960), the UK and Ireland (Brumfitt 1975; Foley 1987; Little 1966; Williams 1969), and Australia (Furness 1975; Kincaid‐Smith 1965; Wren 1969); the recent study from the Netherlands was set in the Dutch Obstetric Consortium, and enrolled women from university, teaching and non‐teaching hospitals, ultrasound centres, and midwifery practices (Kazemier 2015).
The majority of studies enrolled women at the first antenatal visit (Brumfitt 1975; Elder 1971; Foley 1987; Kass 1960; Kincaid‐Smith 1965; Little 1966; Williams 1969; Wren 1969). Some studies enrolled women at the second antenatal visit (Furness 1975), between 16 and 22 weeks' gestation (Kazemier 2015), before 24 weeks (Pathak 1969), between 27 and 31 weeks (Thomsen 1987), between 30 and 32 weeks (Mulla 1960), any gestational age < 32 weeks (Elder 1966), or at any prenatal visit (Gold 1966).
Two studies did not specify microbiological criteria for enrolment (Brumfitt 1975; Mulla 1960). Where there were microbiological criteria, bacteriuria was usually defined as at least one clean‐catch, midstream, or catheterised urine specimen with more than 100,000 bacteria/mL on culture. Some studies required one positive culture of > 100,000 bacteria/mL (Foley 1987; Furness 1975; Kazemier 2015); several studies required confirmation with a second culture (Furness 1975; Gold 1966; Kincaid‐Smith 1965; Little 1966; Pathak 1969; Williams 1969; Wren 1969), and some required a third culture (Elder 1966; Elder 1971; Kass 1960). One study included women with a lower colony count of more than 10,000 bacteria/mL on two occasions (Furness 1975); one enrolled women with any growth of group B streptococcus on a mid‐stream urine culture (Thomsen 1987).
Interventions
Several different antibiotic regimens were used for treatment (seeCharacteristics of included studies for details), including the study of group B streptococci, which compared penicillin to placebo (Thomsen 1987). Treatment varied as well; antibiotics were given in a single dose (Brumfitt 1975), for three to seven days (Foley 1987; Kazemier 2015; Mulla 1960; Thomsen 1987; Williams 1969), for three weeks (Pathak 1969), for six weeks (Elder 1971), continued until delivery (Elder 1966; Furness 1975; Gold 1966; Kass 1960; Kincaid‐Smith 1965), or up to six weeks after delivery (Little 1966; Wren 1969). In four studies, a repeat antibiotic course with the same drug was administered if the infection persisted (Kazemier 2015; Mulla 1960; Pathak 1969; Thomsen 1987). In several studies, an alternative agent was used for persisting or resistant organisms (Foley 1987; Kass 1960; Kincaid‐Smith 1965; Little 1966; Williams 1969). Most studies used antibiotics that are no longer routinely used for treating bacteriuria, including certain sulphonamides (Brumfitt 1975; Elder 1966; Foley 1987; Gold 1966; Kass 1960; Kincaid‐Smith 1965; Little 1966; Mulla 1960), tetracycline (Elder 1971), and methenamine (Furness 1975). Some studies used nitrofurantoin, as either first line treatment (Kazemier 2015; Little 1966; Pathak 1969), or for failures (Elder 1971; Kass 1960; Kincaid‐Smith 1965; Little 1966; Williams 1969). In other studies, ampicillin was used for failures (Kincaid‐Smith 1965; Little 1966; Williams 1969). In one study, women received a fixed rotation of nitrofurantoin, ampicillin, sulphurazole, and nalidixic acid(Wren 1969). In only one study were data on antimicrobial susceptibility used to select the antibiotic (Foley 1987).
Outcomes
Most studies (N = 12) included the outcome of pyelonephritis (Brumfitt 1975; Elder 1971; Foley 1987; Furness 1975; Gold 1966; Kass 1960; Kazemier 2015; Kincaid‐Smith 1965; Little 1966; Mulla 1960; Pathak 1969; Williams 1969).
Six studies reported the outcome of birthweight < 2500 g (Brumfitt 1975; Elder 1971; Kass 1960; Kincaid‐Smith 1965; Little 1966; Wren 1969). In many of the studies conducted during the 1960s, the standard definition of preterm birth was low birthweight, defined as birthweight less than 2500 g, rather than a gestational age less than 37 weeks. Two studies defined preterm birth as a gestational age of less than 37 weeks (Thomsen 1987; Wren 1969); Kazemier 2015 included results for gestational ages < 37 weeks, < 34 weeks, and < 28 weeks. One study did not provide a definition of preterm birth (Gold 1966), and Furness 1975 used a definition of less than 38 weeks. While mean birthweight in the two groups were reported by Brumfitt 1975, Elder 1971, Furness 1975, and Kazemier 2015, and mean gestational age by Kass 1960, Kazemier 2015, and Thomsen 1987, not all authors reported the standard deviation of the mean, and we could not include these data in the analyses.
Four studies defined persistent bacteriuria as a positive culture at delivery or the last prenatal visit (Elder 1966; Elder 1971; Gold 1966; Pathak 1969); two defined it as a positive culture at six weeks to three months postpartum (Furness 1975; Kincaid‐Smith 1965); one did not define it (Foley 1987). Three studies also measured long‐term rates of bacteriuria: one between three and nine months postpartum (Pathak 1969), one at six months (Kincaid‐Smith 1965), and one at 10 to 14 years (Kass 1960).
Only two studies specifically commented on maternal side effects: Mulla 1960 stated there were no side effects necessitating discontinuation of treatment, and Gold 1966 reported "no toxic manifestations in women in the treatment group".
Neonatal mortality or other serious adverse neonatal outcomes were incompletely reported, and there were no standard definitions. Neonatal deaths were reported by Elder 1971 and Kass 1960, and Kazemier 2015 defined and reported on severe neonatal morbidity (presence of one or more of the following: severe respiratory distress syndrome, bronchopulmonary dysplasia, periventricular leukomalacia > grade 1, intracerebral haemorrhage > grade 2, necrotizing enterocolitis > grade 1, or proven sepsis).
No studies reported on women's satisfaction with the intervention.
Excluded studies
Six studies, in seven reports, were excluded because they did not meet the inclusion criteria. In one study, we could not ascertain whether the women had been randomly allocated to treatment or no treatment (Calderon‐Jaimes 1989), and Mohammad 2002 was an observational study, with no treatment intervention. In Sanderson 1984, only women who had been successfully treated initially were randomised to continue treatment. LeBlanc 1964 included symptomatic women in the outcomes; the asymptomatic group was not reported separately. The intervention in NCT03274960 was screening versus no screening, rather than treatment versus no treatment, and the women in Rafalskiy 2013 were randomised to one of two treatment groups, without a no treatment group. SeeCharacteristics of excluded studies table for details.
Risk of bias in included studies
For most studies, there was only a brief and incomplete description of the research methods, which made it difficult to assess the risk of bias in the studies. Kazemier 2015 described the methods in detail, and published the study protocol.
See Figure 2. The description of the characteristics of the study groups was generally poor. In only one study, were the similarities in age, parity, and socioeconomic status between the treatment and no treatment groups adequately described (Thomsen 1987); Kass 1960a described the racial distribution of the two groups, which was comparable; in four other studies, the urinary bacterial isolates for the two groups were listed (Elder 1966; Elder 1971; Gold 1966; Mulla 1960), but otherwise, there was no attempt to demonstrate the comparability of the study groups. Kazemier 2015 described the baseline characteristics of the study populations, including age, body mass index, education level, smoking status, alcohol use, parity, pre‐existing hypertension, gestational age at screening, and pregnancy occurring after fertility treatment, but included the untreated group (N = 163) with the women given placebo (N = 45) in the comparison with the asymptomatic bacteriuria‐positive women who received nitrofurantoin. No other study included the rates of maternal smoking, a recognised risk for low birthweight. There was no description of the presence of co‐existing genital infections, although one study excluded women with positive serology for syphilis (Pathak 1969). Details on the management of recurrent urinary tract infection or persistent infection, the treatment of symptomatic lower urinary tract infection (cystitis), and concurrent antibiotic administration were generally incomplete. Some studies included twin deliveries, while other studies excluded these.
'Risk of bias' summary: review authors' judgements about each risk of bias item for each included study
There was no consistent application of standard definitions for the measured outcomes. Pyelonephritis usually referred to symptoms of loin pain, fever, dysuria, or frequency, with or without a significant urine culture. While rates of low birthweight were usually reported, most studies described this as prematurity. For studies that reported rates of preterm births, the definition of preterm birth was inconsistent, and there were insufficient data presented in most of the studies to compare gestational ages and birthweight between treatment and control groups.
Allocation
In only one study was group assignment based on a web‐based application with a computer‐generated list (Kazemier 2015); in another study, the random component in the sequence generation was described as a coin toss (Foley 1987). For the other studies, there was either no description of the method of randomisation, or the method was clearly inadequate: in four studies, women were allocated to treatment by alternation, so we assessed them at high risk of bias (Elder 1971; Gold 1966; Kass 1960; Wren 1969).
In the majority of studies, there was inadequate description of concealment of allocation to judge selection bias. One study referred to the use of sealed envelopes, but the envelope was drawn from a pool of sealed envelopes rather than a consecutively numbered pile, so we assessed this study as unclear risk of bias (Little 1966). One study described allocation as centrally controlled (Kincaid‐Smith 1965); the remaining studies made no such statement.
Blinding
In 10 of the 15 studies, the control group received a placebo (Brumfitt 1975; Elder 1966; Elder 1971; Gold 1966; Kass 1960; Kazemier 2015; Kincaid‐Smith 1965; Little 1966; Pathak 1969; Thomsen 1987). Two of these studies were described as placebo‐controlled, without any further details (Gold 1966; Little 1966); four of the studies were described as double‐blinded, again, without further details (Brumfitt 1975; Elder 1966; Pathak 1969; Thomsen 1987); and Elder 1971 reported the placebo was 'identical appearing'. However there was insufficient information provided in the reports of these seven studies to know whether the blinding could have been broken, and therefore, we classified them at unclear risk of bias. In three studies, there was a specific mention that neither the women nor treating physician was aware of allocation to treatment group, and we judged them at low risk of performance bias (Kass 1960; Kazemier 2015; Kincaid‐Smith 1965). No treatment was given to the control group in the other five studies, so we judged them at high risk of bias, given the outcomes may have been influenced by lack of blinding (Foley 1987; Furness 1975; Mulla 1960; Williams 1969; Wren 1969).
Only Kazemier 2015 specifically commented that the researchers remained blind to group allocation, although in Kincaid‐Smith 1965, it was reported that "a code of instructions to the pharmacist ensured that the trial remained double‐blind..." We judged these two studies at low risk of detection bias. Although four of the studies were described as double‐blinded, there were insufficient details provided to know whether indeed the outcome assessment was blinded (Brumfitt 1975; Elder 1966; Pathak 1969; Thomsen 1987). We judged these four studies, along with the other nine studies where no information was provided, at unclear risk of bias. However, it is likely that in the five studies where there was no use of placebo, the risk of detection bias was high.
Incomplete outcome data
We judged seven studies as low risk for attrition bias: in four studies, there was no loss to follow‐up and outcomes were reported for all enrolled women (Gold 1966; Little 1966; Mulla 1960; Thomsen 1987); in two studies, information was provided on women lost to follow‐up, and missing outcome data were reasonably balanced across groups (Elder 1966; Pathak 1969), and in Kazemier 2015, five women, enrolled in error, were included in the intention‐to‐treat analysis. We judged five studies to have high risk of attrition bias, given that the missing data may have introduced a clinically important effect on the estimate of treatment. There was no explanation provided for missing outcome data in four studies (Brumfitt 1975; Foley 1987; Furness 1975; Kass 1960), while in Kincaid‐Smith 1965, outcomes were not reported for women excluded because of poor compliance. In three studies, we were unclear about the risk of attrition bias: there was no explanation for the differences in group sizes in Williams 1969, and while the reasons for excluding women were provided in two studies, details on the allocation group were not (Elder 1971; Wren 1969).
Selective reporting
In only one study was a study protocol published and the study's prespecified outcomes reported; we judged this study at low risk of selective reporting bias (Kazemier 2015). For several studies, pregnancy outcomes of interest were either not reported (Elder 1966; Foley 1987; Mulla 1960; Pathak 1969; Williams 1969), or there was no definition of prematurity (Gold 1966); we judged these studies at high risk of reporting bias. We judged studies that failed to include the primary outcome of pyelonephritis (Brumfitt 1975; Elder 1966; Wren 1969), studies where the outcome of pyelonephritis in pregnancy was not reported for all women allocated to treatment (Brumfitt 1975; Furness 1975), or those in which there was no clear definition of pyelonephritis (Foley 1987; Kass 1960; Mulla 1960), at high risk of reporting bias. For the remaining four studies, there was insufficient information to permit judgement; we classified these as unclear.
Other potential sources of bias
We did not identify any other obvious sources of bias, and so we judged this category at low risk, for each of the studies.
Overall risk of bias
We judged five studies at high overall risk of overall bias (Elder 1971; Foley 1987; Furness 1975; Williams 1969; Wren 1969), and one study at low overall risk (Kazemier 2015). While we assessed the overall risk of bias as unclear for the other nine studies, we assessed at least one domain at high risk of bias in each of these studies.
Effects of interventions
1. Antibiotics versus no treatment for asymptomatic bacteriuria
Primary outcomes
Antibiotic treatment may reduce the incidence of pyelonephritis in women with asymptomatic bacteriuria (average risk ratio (RR) 0.24, 95% confidence interval (CI) 0.13 to 0.41; 12 studies, 2017 women; I² = 60%; low‐certainty evidence; Analysis 1.1). There was significant heterogeneity among the studies, and we used a random‐effects analysis. For most studies, a beneficial effect was seen with treatment. We assessed reporting biases (such as publication bias) using a funnel plot. There was no strong evidence of funnel plot asymmetry by visual assessment (Figure 3).
Funnel plot of comparison: 1 Antibiotic versus no treatment for asymptomatic bacteriuria, outcome: 1.1 Development of pyelonephritis
Antibiotic treatment may reduce the incidence of preterm birth, when this was defined as a gestational age of less than 37 weeks (RR 0.34, 95% CI 0.13 to 0.88; 3 studies, 327 women; I² = 32%; low‐certainty evidence; Analysis 1.2; Kazemier 2015; Thomsen 1987; Wren 1969). Thomsen 1987 only enrolled women with group B streptococcal bacteriuria.
Antibiotic treatment may reduce the incidence of birthweight less than 2500 g (average RR 0.64, 95% CI 0.45 to 0.93; 6 studies, 1437 babies; I² = 28%; low‐certainty evidence; Analysis 1.3).
We used a random‐effects meta‐analysis for preterm birth and low birthweight because we predicted there would be real differences in treatment effect from study to study given the clinically important differences in study populations and interventions.
Secondary outcomes
Antibiotic treatment is probably effective in clearing asymptomatic bacteriuria (average RR 0.30, 95% CI 0.18 to 0.53; 4 studies, 596 women; I² = 76%; Analysis 1.4). Without treatment, bacteriuria was present at the time of delivery in 66% of women. Although there was significant statistical heterogeneity among trials, likely explained by differences in study design and intervention, the direction of the effect was consistent. Treatment with antibiotics probably has no effect on the long‐term incidence of bacteriuria (one study reported results at between three and nine months postpartum, one reported at six months, and one at 10 to 14 years).
It is uncertain whether antibiotic treatment has any effect on serious adverse neonatal outcomes, because there was insufficient information from a small number of trials with few events (RR 0.64, 95% CI 0.23 to 1.79; 3 studies, 549 babies; Analysis 1.5).
Information on maternal adverse events was incompletely reported and we could not analyse the results.
There were too few studies that reported on mean birthweight and gestational age for us to draw any conclusions. The results were inconclusive for mean birthweight (mean difference (MD) 21.03, 95% CI ‐83.65 to 125.70; 2 studies, 495 babies; Analysis 1.6), and mean gestational age (MD 1.00, 95% CI 0.01 to 1.99; 1 study, 203 babies; Analysis 1.7)
None of the studies measured cost or women's satisfaction with the treatment.
2. Antibiotic treatment versus no treatment: subgroups by duration of treatment
There was a reduction in the incidence of pyelonephritis for all subgroups, regardless of duration of treatment (Analysis 1.1)
There were too few studies to make a meaningful interpretation of the effect of duration of treatment on the risk of preterm birth (Analysis 1.2).
On visual inspection of the graphs, it appeared that there was a difference in the incidence of low birthweight with duration of treatment although the interaction test did not suggest a difference (test for subgroup differences: Chi² = 2.82, df = 2 (P = 0.24), I² = 29.1%; Analysis 1.3), and there were too few studies to be confident that a longer duration of therapy was associated with a better outcome.
We judged that in all but one of the studies the overall risk of bias was low; therefore, we did not perform a sensitivity analysis based on risk of bias.
Discussion
Summary of main results
While the results of these studies are consistent, suggesting there may be a reduction in the incidence of pyelonephritis, low birthweight, and preterm birth with treatment of asymptomatic bacteriuria, these conclusions are based on low‐certainty evidence. There was significant heterogeneity observed among the studies, which may be explained by study design or quality, type of antibiotic used, and the changes in obstetrical practice in the past five decades between the earliest and the latest study. Duration of antibiotic treatment did not appear to explain any heterogeneity.
The overall incidence of pyelonephritis in the untreated group was 20%, but ranged from 2.2% to 36%. While different definitions of pyelonephritis could explain some of this variation, there may be other factors, for example, type of infecting organism, socioeconomic status, other care given in pregnancy, which if defined, could identify groups of women with asymptomatic bacteriuria with different risks of developing pyelonephritis. However, in the absence of this type of information, the presence of asymptomatic bacteriuria itself defines a population at risk of pyelonephritis.
Overall completeness and applicability of evidence
The studies reported here (with only three exceptions) date from the 1960s and 1970s; microbiological methodology for the diagnosis of bacteriuria has not significantly changed over this interval. Although not all of the antibiotics used in these studies remain available currently, and the use of tetracycline is now contraindicated in pregnancy, it is valid to assume that the results are applicable to other antibiotics active against urinary pathogens that are safe in pregnancy. A Cochrane Review of treatments for symptomatic urinary tract infections during pregnancy concluded that although antibiotic treatment is probably effective for the cure of urinary tract infections, there are insufficient data to recommend any specific regimen (Vazquez 2011); there were similar conclusions for the treatment of asymptomatic bacteriuria (Guinto 2010). The choice of a sulphonamide or sulphonamide‐containing combination, a penicillin, cephalosporin, fosfomycin, or nitrofurantoin, based on the results of susceptibility testing, may be appropriate regimens for the management of asymptomatic bacteriuria. However, increasing antibiotic resistance complicates the choice of empiric regimens, and can make it difficult to select an appropriate regimen (Assefa 2008; Enayat 2008; Hernandez Blas 2007; Tadesse 2014). In India, the presence of extended‐spectrum ß‐lactamases (ESBL), making the strain resistant to all penicillins and cephalosporins, was described in 47% of isolates of E coli, and 36.9% of isolates of Klebsiella pneumoniae (Rizvi 2011). However, a recent case‐control study from Israel reported no difference in obstetric outcomes between women with bacteriuria caused by ESBL versus non‐ESBL isolates of Enterobacteriaceae (Yagel 2018). Surveys of antibiotic susceptibility in pathogens causing community‐acquired uncomplicated urinary tract infections demonstrate considerable regional variation: resistance to ampicillin in E. coli in a survey of European countries and Canada averaged 29.8%, but was as high as 53.9% in Spain (Kahlmeter 2003). In the most recent study, 99% of the strains of E. coli from a low‐risk population of pregnant women in the Netherlands were sensitive to nitrofurantoin (Kazemier 2015),
Both continuous treatment and short‐course therapy strategies may show a benefit in the reduction of pyelonephritis. A small randomised study that compared intermittent therapy with continuous treatment suggested that both strategies may be equally effective (Whalley 1977). While short‐course therapy of asymptomatic bacteriuria has become accepted practice, the optimal duration of treatment is unknown; a three to seven day treatment regimen is currently recommended (Widmer 2011).
There may be an association between antibiotic treatment and preterm birth (low‐certainty evidence), but only three studies reported this outcome, one of which only included women with group B streptococcus bacteriuria. Although we chose to combine data from the three studies, given the very different populations and interventions, the effect of treatment on preterm birth is very uncertain. While preterm births are associated with low birthweight, some low birthweight infants are small‐for‐gestational age as a consequence of intrauterine growth retardation, for which there are many possible etiologies. The reduction in the incidence of low birthweight with antibiotic treatment of asymptomatic bacteriuria is consistent with current theories about the role of infection as a cause of adverse pregnancy outcomes, but a greater understanding of the basic mechanisms by which the treatment of bacteriuria could lead to a reduction in low birthweight is required. Prevention of pyelonephritis, which in studies conducted prior to the availability of effective antimicrobial therapy was associated with preterm birth, may be a factor, but treatment of bacteriuria with antibiotics may also eradicate organisms colonizing the cervix and vagina that are associated with adverse pregnancy outcomes. The relationship between genital infections, such as bacterial vaginosis, and preterm labor was not recognised when most of these studies on the treatment of asymptomatic bacteriuria were originally designed.
Quality of the evidence
We assessed and rated the quality of the evidence for the three primary outcomes using GRADEpro GDT software and the GRADE approach. See summary of findings Table for the main comparison. We rated the evidence for pyelonephritis as low certainty; the certainty of the evidence was downgraded by important design limitations leading to a high risk of bias (including lack of allocation concealment and blinding) and inconsistency (heterogeneity in the results and important differences in the population and intervention). We rated the evidence for preterm birth and for birthweight less than 2500 g as low certainty. For preterm birth, there were important differences in the population and intervention, and in one of the studies contributing data, important design limitations. For birthweight less than 2500 g, there were important design limitations (including lack of allocation concealment and blinding) and differences in the intervention.
Many of the studies only contributed data to one or two of the outcomes; data are missing for most of the outcomes. Most studies did include the outcome of pyelonephritis. When these studies were being designed, there was already an awareness of a possible association between asymptomatic bacteriuria and low birthweight and preterm birth (Kass 1960a), but we found no explanation for why these outcomes were not systematically collected and reported in most of the trials. None of the studies systematically collected or reported the adverse effects of antibiotics. They neither systematically collected the incidence of allergic reactions, vaginal yeast infections, gastrointestinal side effects, or the development of bacterial resistance, nor did they systematically collect neonatal outcomes. While it is not possible to compare the benefits versus the disadvantages of antibiotic therapy from these studies, it is unlikely that the expected side effects from a short course of antibiotics would be significant, although increasingly, there are concerns about the effect of antibiotics on the human microbiome and the developing immune system.
Potential biases in the review process
The inclusion criteria for this review were broad, and included any antibiotic regimen, with the aim of being able to include all of the possible evidence, but this did lead to trials that differed in important ways, in respect to the treatment intervention, which could not be resolved by subgroup analysis.
We acknowledge that there was the potential for bias in the reviewing process. For the earliest iterations of this review, there was no predefined protocol, and the methods for including trials, extracting data, and assessing bias were not well described. However, we did address this in later versions of the review, by ensuring two authors independently assessed all the studies for inclusion, extracted data, and assessed risk of bias. But updates to the review cannot be done blinded to the knowledge of previous outcomes, and although the review has become more methodologically robust, there remains the potential to introduce subjective and unconscious biases.
Agreements and disagreements with other studies or reviews
Results of a meta‐analysis of 17 cohort studies showed an association between asymptomatic bacteriuria and low birthweight and preterm birth, but failed to resolve the question of whether or not asymptomatic bacteriuria was merely a marker for low socioeconomic status, which is associated with low birthweight (Romero 1989). The authors of a recent systematic review concluded there was no reliable evidence to support routine screening for asymptomatic bacteriuria, given the serious methodological shortcomings of the studies identified, and the low number of outcomes reported from the more recent, high‐quality study (Angelescu 2016). The recent publication from the Canadian Task Force on Preventive Health Care gave screening for asymptomatic bacteriuria in pregnancy a weak recommendation, based on very low‐quality evidence, because of the small but uncertain benefit, variation in women's values and preferences, and the judgment that harms were likely minimal (Moore 2018).
Study flow diagram
'Risk of bias' summary: review authors' judgements about each risk of bias item for each included study
Funnel plot of comparison: 1 Antibiotic versus no treatment for asymptomatic bacteriuria, outcome: 1.1 Development of pyelonephritis
Comparison 1 Antibiotics versus no treatment for asymptomatic bacteriuria, Outcome 1 Development of pyelonephritis.
Comparison 1 Antibiotics versus no treatment for asymptomatic bacteriuria, Outcome 2 Preterm birth < 37 weeks.
Comparison 1 Antibiotics versus no treatment for asymptomatic bacteriuria, Outcome 3 Birthweight < 2500 g.
Comparison 1 Antibiotics versus no treatment for asymptomatic bacteriuria, Outcome 4 Persistent bacteriuria.
Comparison 1 Antibiotics versus no treatment for asymptomatic bacteriuria, Outcome 5 Serious adverse neonatal outcome.
Comparison 1 Antibiotics versus no treatment for asymptomatic bacteriuria, Outcome 6 Birthweight.
Comparison 1 Antibiotics versus no treatment for asymptomatic bacteriuria, Outcome 7 Gestational age at delivery.
| Antibiotics compared to no treatment for asymptomatic bacteriuria in pregnancy | ||||||
| Patient or population: pregnant women with asymptomatic bacteriuria Setting: hospital‐based clinics in North America, UK and Ireland, Australia; hospital and community midwifery practices in the Netherlands Intervention: antibiotics Comparison: no treatment | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | № of participants | Certainty of the evidence | Comments | |
| Risk with no treatment | Risk with antibiotics | |||||
| Development of pyelonephritis | Study population | RR 0.24 | 2017 | ⊕⊕⊝⊝ | ||
| 199 per 1000 | 48 per 1000 | |||||
| Preterm birth < 37 weeks | Study population | RR 0.34 | 327 | ⊕⊕⊝⊝ | ||
| 174 per 1000 | 59 per 1000 | |||||
| Birthweight < 2500 g | Study population | RR 0.64 | 1437 | ⊕⊕⊝⊝ | ||
| 136 per 1000 | 87 per 1000 | |||||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| aWe downgraded 1 level for serious limitations in study design: most of the trials contributing outcome data either had important design limitations related to lack of allocation concealment and lack of blinding, or there were insufficient details provided in the report to assess risk of bias. Many of the studies were performed in the 1960s and 1970s, prior to more rigorous study designs and reporting standards. | ||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Development of pyelonephritis Show forest plot | 12 | 2017 | Risk Ratio (M‐H, Random, 95% CI) | 0.24 [0.13, 0.41] |
| 1.1 Single dose | 1 | 173 | Risk Ratio (M‐H, Random, 95% CI) | 0.44 [0.21, 0.92] |
| 1.2 Short course (3 to 7 days) | 4 | 568 | Risk Ratio (M‐H, Random, 95% CI) | 0.32 [0.11, 0.92] |
| 1.3 Intermediate course (3 to 6 weeks) | 2 | 433 | Risk Ratio (M‐H, Random, 95% CI) | 0.17 [0.08, 0.37] |
| 1.4 Continuous treatment | 5 | 843 | Risk Ratio (M‐H, Random, 95% CI) | 0.16 [0.04, 0.57] |
| 2 Preterm birth < 37 weeks Show forest plot | 3 | 327 | Risk Ratio (M‐H, Random, 95% CI) | 0.34 [0.13, 0.88] |
| 2.1 Single dose | 0 | 0 | Risk Ratio (M‐H, Random, 95% CI) | 0.0 [0.0, 0.0] |
| 2.2 Short course (3 to 7 days) | 2 | 154 | Risk Ratio (M‐H, Random, 95% CI) | 0.36 [0.05, 2.72] |
| 2.3 Intermediate course (3 to 6 weeks) | 0 | 0 | Risk Ratio (M‐H, Random, 95% CI) | 0.0 [0.0, 0.0] |
| 2.4 Continuous treatment | 1 | 173 | Risk Ratio (M‐H, Random, 95% CI) | 0.36 [0.14, 0.95] |
| 3 Birthweight < 2500 g Show forest plot | 6 | 1437 | Risk Ratio (M‐H, Random, 95% CI) | 0.64 [0.45, 0.93] |
| 3.1 Single dose | 1 | 413 | Risk Ratio (M‐H, Random, 95% CI) | 0.65 [0.36, 1.18] |
| 3.2 Short course (3 to 7 days) | 0 | 0 | Risk Ratio (M‐H, Random, 95% CI) | 0.0 [0.0, 0.0] |
| 3.3 Intermediate course (3 to 6 weeks) | 1 | 278 | Risk Ratio (M‐H, Random, 95% CI) | 1.09 [0.55, 2.14] |
| 3.4 Continuous treatment | 4 | 746 | Risk Ratio (M‐H, Random, 95% CI) | 0.54 [0.33, 0.87] |
| 4 Persistent bacteriuria Show forest plot | 4 | 596 | Risk Ratio (M‐H, Random, 95% CI) | 0.30 [0.18, 0.53] |
| 5 Serious adverse neonatal outcome Show forest plot | 3 | 549 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.64 [0.23, 1.79] |
| 6 Birthweight Show forest plot | 2 | 498 | Mean Difference (IV, Fixed, 95% CI) | 21.03 [‐83.65, 125.70] |
| 7 Gestational age at delivery Show forest plot | 1 | 203 | Mean Difference (IV, Fixed, 95% CI) | 1.0 [0.01, 1.99] |
