Screening for Asymptomatic Bacteriuria in Adults: Updated Evidence Report and Systematic Review for the US Preventive Services Task Force

Abstract

Importance:

Screening for asymptomatic bacteriuria (ASB) can identify patients for whom treatment might be beneficial for preventing symptomatic infection and other health outcomes.

Objective:

To systematically review benefits and harms of ASB screening and treatment in adults, including during pregnancy, to inform the US Preventive Services Task Force.

Data Sources:

MEDLINE, PubMed (publisher-supplied records), and Cochrane Collaboration Central Registry of Controlled Trials; surveillance through May 24, 2019.

Study Selection:

Randomized trials and observational studies on benefits and harms of screening for ASB, and randomized trials on benefits and harms of ASB treatment. Eligible populations included unselected, asymptomatic individuals without known urinary tract conditions.

Data Extraction and Synthesis:

Independent critical appraisal and data abstraction by 2 reviewers. Random-effects meta-analysis was conducted to estimate benefits of the interventions.

Main Outcomes and Measures:

Symptomatic infection; function, morbidity, mortality; pregnancy complications and birth outcomes.

Results:

Nineteen studies (N=8443) meeting inclusion criteria were identified. Two cohort studies (N = 5289) found fewer cases of pyelonephritis in the cohorts of screened pregnant women (0.5%) than within retrospective comparisons of unscreened cohorts (2.2% and 1.8%), and the larger study estimated a statistically significant relative risk (relative risk [RR] = 0.30 [95% CI, 0.15, 0.60]). No studies examined screening in nonpregnant populations. Among 12 trials of ASB screening and treatment during pregnancy (n = 2377; one conducted within past 30 years), reduced rates of pyelonephritis (intervention group [IG] range 0 to 16.5%, control group [CG] range 2.2 to 36.4%, pooled risk ratio [RR] = 0.24 [95% CI, 0.14, 0.40], k = 12) and low birthweight (IG range 2.5 to 14.8%, CG range 6.7 to 21.4%, pooled RR = 0.64 [95% CI, 0.46, 0.90], k = 7) were observed. There was no difference in infant mortality (pooled RR = 0.98 [95% CI, 0.29, 3.26], k = 6). Five randomized clinical trials of ASB treatment in nonpregnant adults (n = 777) did not report any significant differences in risk of infection, mobility, or mortality. Limited evidence on harms of screening or treatment was available, and no statistically significant differences were identified.

Conclusions and Relevance:

Screening and treatment for ASB during pregnancy was associated with reduced rates of pyelonephritis and low birth weights, but the available evidence was not current, with only one study conducted in the past 30 years. Benefits of ASB treatment in nonpregnant adult populations were not found. Trial evidence on harms of ASB antibiotic treatment was limited.

Asymptomatic bacteriuria (ASB) is defined as the presence of a significant bacterial colony count in urine (≥10⁵ colony forming units per milliliter of a single bacterial species) in the absence of any of the typical signs or symptoms of a urinary tract infection (UTI).¹ During pregnancy, ASB is present in an estimated 2% to 10% of women and has been associated with an increased risk of symptomatic infection, including pyelonephritis.^2,3 Rates of pyelonephritis during pregnancy are low in the United States, possibly owing to ASB screening and treatment as part of standard prenatal care.^4,5 In general adult populations, women have higher rates of ASB than men, and prevalence rises after menopause for women and older (>90 years) men. In 2008, the US Preventive Services Task Force (USPSTF) concluded there was a high level of certainty that the net benefit of screening pregnant women for ASB was substantial and recommended screening for ASB with urine culture for pregnant women at 12–16 weeks’ gestation, or at the first prenatal visit (A Recommendation). The USPSTF concluded with moderate certainty that the harms of screening men and nonpregnant women for ASB outweigh the benefits and recommended against screening for ASB in men and nonpregnant women (D Recommendation). The aim of this systematic review was to update the evidence on benefits and harms of screening and treatment of screen-detected ASB to inform the USPSTF recommendation update.^6,7

METHODS

Scope of Review

The USPSTF commissioned this review to evaluate evidence on the benefits and harms of screening and treatment of asymptomatic bacteriuria in adults, including pregnant women. Four key questions were drafted to address screening (KQ1, KQ2) and treatment (KQ3, KQ4), as shown in Figure 1. Methodological details, including details on study selection and a list of excluded studies, and detailed result including descriptive tables and forest plots are available in the full evidence report at https://www.uspreventiveservicestaskforce.org/Page/Document/UpdateSummaryFinal/asymptomatic-bacteriuria-in-adults-screening1.

Figure 1.

Analytic Framework

Evidence reviews for the US Preventive Services Task Force (USPSTF) use an analytic framework to visually display the key questions that the review will address to allow the USPSTF to evaluate the effectiveness and safety of a preventive service. The questions are depicted by linkages that relate interventions and outcomes. Refer to the USPSTF Procedure Manual for interpretation of the analytic framework.¹⁰

Data Sources and Searches

In addition to considering all studies from the previous reviews on this topic for inclusion in the current review,^6,8,9 a comprehensive search of MEDLINE, PubMed Publisher-Supplied Records, and the Cochrane Collaboration Central Registry of Controlled Trials for literature published through September 7, 2018, was conducted (eMethods in the Supplement). The reference lists of other previously published reviews, meta-analyses, and primary studies were examined and ClinicalTrials.gov was searched to identify additional relevant studies or ongoing trials. After September 2018, ongoing surveillance continued through article alerts and targeted searches of high-impact journals to identify major studies published in the interim that could affect the conclusions or understanding of the evidence and affect the related USPSTF recommendation. The last surveillance was conducted on May 24, 2019and resulted in the addition of no new studies.

Study Selection

Two reviewers reviewed 4138 abstracts and 288 full-text articles against a priori inclusion and exclusion criteria (Figure 2; eTable 1 in the Supplement). The review was defined to focus on adults (ages ≥18) and pregnant women of any age who were asymptomatic for, and not in specialty care for, conditions of the urinary tract; not immunocompromised; and not undergoing surgical or catheterization procedures. Studies among community-dwelling adults, including those living in independent or assistant living facilities, were included, but studies conducted in hospitalized or institutionalized settings were not. For general adult populations, we included studies conducted in countries categorized as “very high” on the Human Development Index (HDI) in 2016. For pregnant women, the scope was expanded slightly to include studies with an HDI of “high” or “very high” because ASB screening and treatment in pregnancy is standard-of-care, established practice in many “very high” HDI countries and not an active area of research.

Figure 2.

Literature Flow Diagram

^aReasons for exclusion: Relevance: Study aim was not relevant. Design: Study did not use an included design or was not a primary research article. Setting: Study conducted in an eligible setting, including those conducted in a country not classified as high or very/high on the Human Development Index in 2016. Population: Study conducted in an ineligible population, including those conducted in individuals who were institutionalized, symptomatic of a urinary tract infection (UTI), had recurrent UTIs, or where screening was related to a surgical procedure. Screening Test: Ineligible screening test including screening performed via catheter or suprapubic aspiration. Intervention: intervention was out of scope. Comparator: ineligible comparator including comparative effectiveness studies. Outcomes: No relevant outcomes or incomplete outcomes reporting. Language: Publication not in English.

For all KQs, we included randomized clinical trials (RCTs) to assess the benefits and harms of screening (KQs 1 and 2) and treatment (KQs 3 and 4) for ASB in asymptomatic pregnant and nonpregnant adults. In addition, for KQs 1 and 3 among pregnant women, observational cohort studies with a comparator of no screening or no treatment were included because prior evidence and recommendations established a standard practice of ASB screening and treatment, and movement away from trial research for the topic. For KQs 2 and 4, observational cohort studies with or without a comparison group were also included, as well as registry studies.

Data Extraction and Quality Assessment

Included studies were assessed for risk of bias by 2 independent reviewers according to prespecified criteria (eTable 2 in the Supplement),^10,11 and discrepancies were resolved after discussion with a third reviewer. Due to the historical nature of the evidence, a more lenient quality rating of studies was employed to allow for changes in trial reporting standards over time. To support sensitivity analyses, studies were flagged as “high risk of bias” if there were notable problems with reporting or procedures related to: concerns about diagnostic criteria or definitions of outcomes, treatment allocation, baseline characteristics of participants (e.g., missing or unbalanced study groups), or high suspicion of selective outcome reporting. One reviewer extracted key descriptive and outcome data into standardized abstraction forms, and a second reviewer checked the data for accuracy.

Data Synthesis and Analysis

Summary tables describing study design, population, and intervention characteristics were created separately for nonpregnant adult and pregnant populations. Data were synthesized separately for each KQ. Given the small number of included studies, a narrative synthesis was provided of study results for all KQs evaluated in adult, nonpregnant populations. Similarly, for KQ1 and KQ2 in pregnant populations, data were summarized narratively and in tables.

Quantitative pooling of KQ3 and KQ4 outcomes related to ASB treatment in pregnant populations was conducted with random-effects meta-analyses that applied the DerSimonian and Laird¹² (DL) method to calculate pooled differences in mean changes or pooled risk ratios (for binary outcomes). When pooling fewer than 10 studies, sensitivity analyses using a restricted maximum likelihood (REML) model with Knapp-Hartung correction for small samples were calculated and reported if different from the DL result. Details of the data analysis methods are included in the full report. Statistical heterogeneity among the pooled studies was evaluated with standard χ² tests and the I² statistic to estimate the proportion of total variability in point estimates.¹³ Funnel plots were used to examine outcomes for potential small-study effects (a possible indication of publication bias), and the Egger test was conducted if ≥10 studies were available to statistically assess whether study size was associated with study results.¹⁴ Sensitivity analyses were conducted after dropping “high risk of bias” studies from meta-analysis. Stata version 15.1 (StataCorp LP) was used for all analyses. All significance testing was 2-sided, and results were considered statistically significant if the P value was .05 or less.

The strength of the overall body of evidence for each key question was graded as high, moderate, low, or insufficient based on the consistency, precision, and limitations of the body of evidence related to each outcome. For more details on review methods, see the full report.

RESULTS

Two reviewers independently assessed 4318 abstracts and reviewed 288 full-text articles. In total, 19 studies (1 good quality, 18 fair quality; N = 8443) of ASB screening or treatment were included in this review.^15–33 Of these, 14 were conducted with pregnant populations: 2 comparative cohort studies of screening effectiveness and harms (KQ1, KQ2) and 12 trials of the effectiveness and harms of ASB treatment (KQ3, KQ4). Five trials examined the effectiveness and harms of ASB treatment (KQ3, KQ4) among nonpregnant adults, primarily women and older adults.

Key Question #1: Does screening for asymptomatic bacteriuria improve health outcomes among adults, including pregnant women?

Pregnant Populations

Two fair-quality cohort studies (N = 5289)^15,16 compared a screened cohort to an unscreened historical comparison group to assess the association of maternal and perinatal outcomes with implementation of an ASB screening and treatment program (Table 1). In a study conducted in Spain (n = 4917), 77 (4.7%) of screened participants were diagnosed with ASB. There were 9 cases of pyelonephritis (0.5%) diagnosed in this screened cohort and 60 cases (1.8%) in the historical comparison cohort (RR, 0.30 [95% CI, 0.15, 0.60]). In a study conducted in Turkey (n = 372), 17 (9.3%) participants screened ASB positive, and there were 4 pyelonephritis cases (2.2%) in the unscreened cohort and 1 (0.54%) in the screened cohort. No birth or infant outcomes were reported in the Spanish study,¹⁵ and the Turkish study reported no significant difference in the weight of newborns (data NR) or premature births (screened 11.8% [22/186] versus unscreened 9.7% [18/186]).¹⁶ Low-event rates, study size, and reporting inconsistencies limited the ability to draw conclusions for other outcomes in this study (eTable 3 in the Supplement).

Table 1.

Summary of Study Characteristics, by Population

Author, Year	QR	Study design N rand	Country	Population description	Race/Ethnicity (%)	Confirmatory culture requireda	ASB prevalence (N pos./N screened)	Mean age (range)	KQ
Pregnant women
Brumfit, 197526	Fair	RCT 414	GBR	Pregnant women, <32 weeks’ gestation	Asian: 9.7 Other: 10.6b	NR	426/20000 (2.1%)	26.4 (NR)	3
Elder, 197125	Fair	RCT 289	US	Pregnant women, <32 weeks’ gestation	White: 39.9 Other: 60.1	Yes	362/9156c (4.0%)	25.1 (NR)	3, 4
Foley, 198729	Fair	RCT 220	IRL	Pregnant women	NR	Nod	220/6883 (3.2%)	NR	3
Furness, 197527	Fair	RCT 206	AUS	Pregnant women	NR	Noe	226/5256 (4.3%)	NR	3, 4
Gratacos, 199415	Fair	Cohort 4917f	ESP	Pregnant women, <25 weeks gestation	NR	Yesg	77/1652 (4.7%)	NR	1
Gold, 196618	Fair	RCT 65	US	Pregnant women	White: 9.0h Hisp: 6.0i Other: 85.0j	Yes	65/1281 (5.1%)	NR	3, 4
Kazemier, 201533	Fair	RCT 85	NLD	Pregnant women (age ≥ 18 years), 16–22 weeks’ gestation	White: 92.0k	Noe	255/5132 (5.0%)	29k (NR)	3, 4
Kincaid-Smith, 196517	Fair	RCT 116	AUS	Pregnant women, <26 weeks’ gestation	NR	Yes	160/4000l (4.0%)	NR	3, 4
Little, 196619	Fair	RCT 265	GBR	Pregnant women, 12 weeks’ gestation (mean)	NR	Yes	265/5000 (5.3%)	NR (10–40+)	3
Pathak, 196922	Fair	RCT 178	JAM	Pregnant women, <24 weeks’ gestation	NR	Yesm	217/7602 (2.9%)	NR	3, 4
Savage, 196720	Fair	RCT 203	US	Pregnant women, <32 weeks’ gestation	White: 52.0n AA: 46.0n Other: 2.0n	Yes (two confirmatory tests)	245/6327 (3.9%)	NR	3
Uncu, 200216	Fair	Cohort 372g	TUR	Pregnant women, <33 weeks gestation	NR	No	23/247 (9.3%)	27.7 (NR)	1, 2
Williams, 196923	Fair	RCT 163	GBR	Pregnant women, <30 weeks’ gestation	NR	Yes	211/5542 (3.8%)	NR	3, 4
Wren, 196924	Fair	RCT 173	AUS	Pregnant women	NR	Yesf	183/3604 (5.1%)	NR	3
General adults
Abrutyn, 199430	Fair	Nonrand CCT 358	US	Women (mean age: 81.9 years)	NR	Yes	NR	81.9 (NR)	3
Asscher, 196921	Fair	RCT 94	WLS	Women (age 20–65 years)	NR	Yes	107/3578 (3.0%)	NR (20–65)	3, 4
Boscia, 198728	Fair	RCT 124	US	Women (age ≥65 years)	NR	Yes	124/603 (20.6%)	85.8 (70–100)	3, 4
Giamarellou, 199831	Fair	RCT 96	GRC	Older adults (83.9% female; age ≥65 years)	NR	Yes	106/455 (23.3%)	83.3 (NR)	3, 4
Harding, 200232	Good	RCT 105	CAN	Women with diabetes (age >16 years)	NR	Yes	135/1900 (7.1%)	55.3 (NR)	3, 4

^aScreening with culture using a threshold of 10⁵ CFU/ml unless otherwise noted

^bWest Indian

^cNumber invited

^dScreening method NR

^eScreening with dipslide

^fIncluded in cohort

^gCutoff NR

^hBaseline characteristics for entire screened cohort “other white”

ⁱPuerto Rican

^jNon-white

^kIG only, CG age NR

^lCalculated

^mScreening test NR

ⁿEstimated from figure

Abbreviations: AUS = Australia; CAN = Canada; CCT = controlled clinical trial; CFU/ml = colony forming units/milliliter; ESP = Spain; FU = followup; GBR = Great Britain; GRC = Greece; IRL = Ireland; JAM = Jamaica; mg/dL = milligram/deciliter; mos = months; N = number of participants; NLD = Netherlands; NR = not reported; pos = positive; QR = quality rating; rand = randomized; RCT = randomized clinical trial; TUR = Turkey; US = United States; WLS = Wales

Nonpregnant Adult Populations

No studies were identified that addressed the benefits of screening for ASB in the general adult population.

Key Question #2: What are the harms of screening for asymptomatic bacteriuria?

Pregnant Populations

The Turkish cohort study described above reported on potential harms of the screening program (such as fetal abnormalities and intrauterine death), and found no evidence of harms associated with the screening program.¹⁶ Two congenital abnormalities were reported in the unscreened cohort (1.1%) compared with 3 in the screening cohort (1.6%); however, the 3 congenital abnormalities were observed among infants of women who screened ASB negative and presumably were not prescribed antibiotics to treat ASB (eTable 3 in the Supplement).

Nonpregnant Adult Populations

No studies were identified that addressed the harms of screening for ASB in the general adult population.

Key Question #3: Does treatment of screen-detected asymptomatic bacteriuria improve health outcomes?

Pregnant Populations

Twelve trials of pregnant women screened for ASB and randomized to either a treatment or control condition were included (N = 2377).^{17–20,22–27,29,33} All but 2^29,33 were published in the 1960s or 1970s (Table 1). The 2 most recently published studies were conducted in the Netherlands (2015)³³ and Ireland (1987).²⁹ Among the 10 early studies, 3 were conducted in the United States,^18,20,25 and the remainder in Great Britain,^19,23,26 Jamaica,²² and Australia.^17,24,27 Most studies were conducted in the OB/GYN clinics of hospitals, with 7 specifying screening at the first prenatal visit,^{17–19,23,24,26,27} 2 specifying screening by a certain week of gestation in pregnancy,^25,33 and 3 indicating pregnant women with no mention of the timing of study recruitment.^20,22,29 Information on the characteristics of the study participants was sparsely reported. Only 3 studies reported the mean age of randomized women (range 25–29).^25,26,33

Exclusion criteria were generally not specified or limited (e.g., hypertensive, chronic renal insufficiency, recent UTI) in most included studies. The most recent trial had an aim of enrolling a low-risk study population and thus excluded women at risk of preterm birth and other health conditions.³³ The percent of pregnant women screened who were diagnosed with ASB ranged from 2.1% to 5.3% across the included studies. The most recent study identified 5.0% of women who were screened with dipslide as ASB positive.³³ Treatments for screen-detected ASB varied widely across the included studies with respect to timing, dosage, duration, and medication. Sulfonamides were the most common class of antibiotics, but many specific antibiotic formulations are no longer used (e.g., sulfamethizole, sulfadimethoxine). The treatment dosage and duration in most of the studies were higher and longer than what is more common to contemporary practice. Five studies were judged to have particularly high risk of bias due to multiple concerns related to randomization and inconsistencies or a lack of clarity regarding the participant characteristics and outcomes.^{20,23,24,26,27}

The most commonly reported health outcome was pyelonephritis (k = 12); other perinatal health outcomes less consistently reported (Table 2). Rates of pyelonephritis in the control group were much lower in the two most recent studies^29,33 than in older literature (2.2% and 2.5% versus 7%−36%); and 8 of the 12 included studies reporting pyelonephritis rates greater than 20% among women with ASB in the untreated/placebo group. Higher rates of pyelonephritis were observed in the control group (placebo or no treatment) than in the treated group in all but 1 study.²⁹ The pooled effect estimate indicated a 4-fold risk reduction associated with ASB treatment, pooled RR, 0.24 (95% CI, 0.14, 0.40); rates of pyelonephritis in the intervention group ranged from 0 to 16.5% and ranged from 2.2 to 36.4% in the control group. Sensitivity analyses that eliminated 5 studies deemed to have particularly high risk of bias demonstrated a greater pooled risk reduction and lower statistical heterogeneity, pooled RR, 0.19 (95% CI, 0.11 to 0.34). Visual inspection of a funnel plot including all studies reporting this outcome were somewhat asymmetric, suggesting potential reporting bias, with the Egger test showing p = 0.08.

Table 2.

Pooled Effects of ASB Treatment in Pregnancy, by Outcome

Health outcome	Studies (k)	Observations (n)	Outcome event rate or mean (SD) in CG	Pooled RR (95%CI) or Mean difference (95%CI) associated with intervention	I2
Pyelonephritisa	12	2068	212/1023 (20.7%)	0.24 (0.14 to 0.40)	56.9%
Low birthweightb	7	1522	99/753 (13.1%)	0.64 (0.46 to 0.90)	15.8%
Mean birthweight (g)	5	1070	3210.5 g (628.6)	15.51 g (−91.17 to 122.18)	52.5%
Preterm birthc	4	493	27/217 (12.4%)	0.81 (0.38 to 1.73)	34.8%
Perinatal mortalityd	6	1103	21/574 (3.6%)	0.98 (0.29 to 3.26)	52.3%
Hypertensive disorders of pregnancye	5	889	31/465 (6.7%)	1.21 (0.76 to 1.93)	0%
Congenital malformations among infants	5	961	12/472 (2.5%)	0.44 (0.16 to 1.22)	0%

^aDefined by the study

^bBirthweight less than 2500 grams or small for gestational age (below the 10th percentile for gestational age)

^cBirth prior to 37- or 38-weeks’ gestation or study defined “premature birth”

^dPerinatal mortality includes fetal and infant deaths occurring >20 weeks of gestation and <1 week postpartum

^eToxemia, Preeclampsia, HELLP syndrome

Abbreviations: CG = control group; CI = confidence interval; g= grams; RR = relative risk; SD = standard deviation

Seven studies reported differences in low birthweight infants (<2,500 grams or small for gestational age [weight below the 10^th percentile for gestation age]) among women treated or untreated for ASB (Table 2).^{17,19,20,24–26,33} The proportion of low birthweight infants ranged from 2.5% to 14.8% in the intervention groups and from 6.7% to 21.4% in the control groups. A statistically significant reduction in the risk of low birthweight was reported in 2 studies,^20,24 and the pooled estimate was also statistically significant, with low statistical heterogeneity, pooled RR, 0.64 (95% CI, 0.46 to 0.90). Sensitivity analysis removing studies with the highest risk of bias led to exclusion of the statistically significant studies, and a no longer significant pooled effect, pooled RR, 0.86 (95% CI, 0.57 to 1.31). There were too few studies available for this outcome to support the Egger test or assessment of publication bias with a funnel plot. Results from 5 studies reporting mean birthweight^24–27,33 were inconsistent and did not show a statistically significant association. Few small and clinically heterogeneous studies reported results for preterm birth, limiting conclusions that can be drawn from the null pooled estimate.

Six studies reported perinatal mortality.^{17,19,20,24,25,33} Three of the trial effects were in the direction of treatment benefit and the other three in the direction of treatment harms. Owing to small numbers, none of the studies reported statistically significant effects, and the pooled estimate was null (pooled RR, 0.98 [95% CI, 0.29 to 3.26]). The small number and size of the included studies in the sensitivity analysis, however, resulted in an underpowered analysis for evaluating this rare outcome. Event rates ranged from 0 to 6.6% in the intervention group and from 0 to 7.3% in the control group.

Nonpregnant Adult Populations

Five included trials (N = 777) examined the effectiveness of antibiotic treatment in general adult populations with screen-detected ASB (Table 1).^{21,28,30–32} These studies were conducted in the United States (k = 2),^28,30 Canada,³² the United Kingdom,²¹ and Greece.³¹ Two studies of adult women recruited individuals from medical centers,^21,32 with 1 of them limited to women with diabetes.³² Three studies were conducted among older adults (mean age 81.9–85.8 years) residing in independent living facilities.^28,30,31 Two of the 3 studies among older adults were limited to women,^28,30 and the third was mostly women (83.9%).³¹ In general, population characteristics were sparsely reported across studies. The study of ASB treatment in diabetic women was rated as good quality,³² and the 4 remaining studies were rated fair quality.

Four studies reported on the rate of symptomatic infection or pyelonephritis, with none finding a statistically significant difference between treatment and control groups across a range of timepoints (eTable 4 in the Supplement). One study among male and female older adults reported on mobility as an outcome and found no difference at 6 months.³¹ The study excluded individuals who needed help performing activities of daily living at enrollment and used a subjective proxy measure of good mobility (i.e., complete independence) determined by the physician and head nurse of the independent living pavilions. All 3 trials specific to older adults reported mortality by treatment group as a primary or secondary outcome, with no trial finding a statistically significant difference (eTable 5 in the Supplement). One trial conducted among women that analyzed mortality as the primary outcome reported 18% mortality over the course of 100 months of follow-up compared with 20% in the control group (HR, 0.92 [95% CI, 0.50 to 1.47]).³⁰ Two additional studies that measured mortality as a secondary outcome found no statistically significant effect of treatment at 6 months of follow-up.^28,31

Key Question #4: What harms are associated with treatment of screen-detected asymptomatic bacteriuria?

Pregnant Populations

Seven of the included studies for KQ3 described above reported potential harms of treatment of screen-detected ASB.^{18–20,24,25,27,33} Five studies reported on congenital malformations in the intervention and control groups (Table 2).^{19,24,25,27,33} Few cases were reported and more cases were observed in the control groups than the intervention groups in all but 1 study.¹⁹ The pooled estimate was not statistically significant, pooled RR, 0.44 (95% CI, 0.16 to 1.22) – cases of congenital malformation ranged from 0 to 1.6% in the intervention group and from 1.4 to 4.2% in the control groups. Evidence related to other infant and maternal harms of ASB treatment in pregnancy was sparsely and inconsistently reported, and there was a lack of evidence on long-term neonatal outcomes following antibiotic treatment of ASB in pregnancy. Two studies provided information on maternal adverse reactions to medications.^19,24 For ampicillin treatment, vaginitis and diarrhea were reported.²⁴ For nalidixic acid and nitrofurantoin treatment, rashes and nausea were reported.^19,24

Nonpregnant Adult Populations

Two studies of treatment in nonpregnant adult women^21,32 and 2 studies in older adults^28,31 reported on rates of adverse events associated with treatment of ASB. One study among nonpregnant women reported that there were no adverse drug reactions among the 49 women treatment with nitrofurantoin therapy.²¹ The study among diabetic women (N = 105) reported higher rates of treatment-related adverse events (not specified) among those treated with trimethoprim-sulfamethoxazole compared to placebo but the difference was not statistically significant; 18.1% compared with 6.0% of women, RR 3.45 (95% CI 0.90 to 14.29).³² One study among older adults reported that no adverse medication reactions occurred among the 63 women treated with trimethoprim.²⁸ Another reported that 6% (2/32) of women assigned to daily ofloxacin therapy withdrew due to adverse events (vertigo and gastrointestinal tract symptoms).³¹ Treatment was found to not affect hematocrit, serum bilirubin, or blood urea, but a mild reduction in serum creatinine was seen in the treatment groups.

DISCUSSION

A summary of the evidence for this review is reported in Table 3. For general nonpregnant adult populations, no evidence identified health benefits associated with screening or treatment of screen-detected ASB. This is consistent with findings in reviews by others, including the ISDA.^2,34 Concerns about potential overuse of antimicrobial treatment and the development of resistance to treatment are therefore highlighted in discussions regarding clinical practice for asymptomatic people, including the elderly and those with diabetes.

Table 3.

Summary of Evidence by Key Question and Population

Key Question	Population	Studies (k) Observations (n) Study Designs	Summary of Findings	Consistency and Precision	Other Limitations	Strength of Evidence	Applicability
KQ1: Screening effectiveness	Pregnant women	k=2 (n=5,289) Retrospective cohort studies	Fewer cases of pyelonephritis occurred in pregnant women included in a screening cohort compared with retrospective cohort of unscreened women	Direction of effects consistent and one study with adequate precision	Fair-quality studies with risk of bias due to limited information about how cohort was identified, characteristics of women in comparison cohorts, ascertainment bias, selective reporting.	Low for a benefit of screening for prevention of pyelonephritis in pregnancy based on two fair-quality cohort studies	One study conducted in a hospital in Spain 24 years ago and another in Turkey 16 years ago – may not be entirely applicable to current U.S. hospital settings and populations.
KQ1: Screening effectiveness	General adult populations	k=0	NA	NA	NA	NA	NA
KQ2: Screening harms	Pregnant women	k=1 (n=372) Retrospective cohort study	No harms of screening were identified. Number of congenital abnormalities similar between groups and none in screen-positive women.	Consistency NA Imprecise due to small n, few events	Limited reporting on potential harms of screening and treatment, low n for detecting rare harms.	Insufficient for the absence of screening harms based on one fair-quality cohort study	Small study conducted in Turkey 16 years ago – may not be entirely applicable to current U.S. hospital settings and populations
KQ2: Screening harms	General adult populations	k=0	NA	NA	NA	NA	NA
KQ3: Treatment effectiveness	Pregnant women	k=12 (n=2,369) RCTs	Treatment of screen-detected ASB in pregnancy reduced the risk of pyelonephritis (5.5% vs 20.7%; pooled RR, 0.24 [95% CI, 0.14 to 0.41], k=12, n = 2,068) and low birthweight infants (8.3% vs 13.1%; pooled RR, 0.64 [95% CI, 0.46 to 0.90], k=7, n =1,522). Other birth outcomes were less consistently reported, and statistically significant differences were not found in pooled analyses	Consistent and precise for pyelonephritis Consistent but less precise for low birthweight Imprecise and inconsistent for other perinatal outcomes, including perinatal mortality, mean birthweight, and preterm birth	Risk of bias present or difficult to assess in all studies; limited reporting of baseline characteristics; problems with blinding, randomization, selective reporting, and outcome definitions.	Moderate for benefit of treatment on pyelonephritis from 12 fair- quality RCTS (including 5 with high risk of bias) Low for benefit of treatment on low birthweight from 7 fair- quality studies (including 3 with high risk of bias)	Most studies conducted over 40 years ago, and many of the treatment protocols and medications are no longer used in clinical practice Rates of pyelonephritis as much as 10-fold higher in historical trials compared with estimates from modern prenatal care.
KQ3: Treatment effectiveness	General adult populations	k=5 (n=777) 4 RCTs, 1 nonrandomized CCT	Mortality: 3 trials in older adults found no difference in mortality over 6 to 100 months followup Mobility: 1 trial in older adults found no effect on mobility at 6 months Symptomatic Infection/Pyelonephritis: 4 trials (including 2 in older adults) found no difference in the rate of symptomatic infection	Consistent for no benefit Imprecise Treatment ranged from a single dose of treatment to daily treatment over 3 months	Lack of reporting of population characteristics Some studies lacked reporting on randomization, allocation, and outcome assessment	Low	Evidence primarily applies to women (84–100% female in each study) Three studies limited to older adults (2 of 3 limited to older women) One study limited to women with diabetes
KQ4: Treatment harms	Pregnant women	k=7 (n=1,286)	Five studies reported congenital malformations with effect in direction of benefit, but null. All other outcomes sparsely reported and did not provide evidence of or rule out harms.	Inconsistent Imprecise	Risk of bias present or difficult to assess in all studies; limited reporting of baseline characteristics; problems with blinding, randomization, selective reporting, and outcome definitions	Insufficient for absence of treatment harms	Most studies conducted over 40 years ago, and many of the treatment protocols and medications are no longer used in clinical practice
KQ4: Treatment harms	General adult populations	k=4 (n=419) RCTs	Minimal reporting of adverse events. Most studies reported no adverse events or only on those few patients who withdrew from the trials based on adverse events. One study of treatment of women with diabetes found higher rates of adverse events among women in the treatment group.	Inconsistent Imprecise	Limited data reporting. Unclear reporting bias	Insufficient	Evidence is limited to mostly women (84–100% female in each study)

Abbreviations: CCT = controlled clinical trial; KQ = Key Question; n = number of participants; NA = not applicable; RCT = randomized clinical trial

Few studies of ASB screening or treatment in pregnant populations have been conducted in the past 40 years; historical evidence established ASB screening and treatment as standard obstetric practice in the United States. The most consistent finding from the available trials indicates that treatment of ASB in pregnancy was associated with a lower risk of pyelonephritis, even when the studies at highest risk of bias were excluded in sensitivity analysis. Other outcomes were less consistently reported, raising concerns about possible selective reporting or publication bias.

The sparse, less rigorous reporting and research methods seen in the earlier scientific era could contribute to risk of bias. In addition, the applicability of historic evidence to current clinical settings and populations is limited due to changes over time in the population characteristics, delivery of care, and treatment protocols for ASB. The only recently conducted study of the treatment of ASB is a randomized trial conducted in the Netherlands,³³ where ASB screening and treatment are not standard care. However, this study focused on a low-risk population, and event rates and study power were too low to draw conclusions about the effectiveness of treatment from this study alone. It does, however, provide important contextual information about the course of untreated ASB in a modern population.

More recent evidence on the association between ASB and pyelonephritis, and between these infections and pregnancy health outcomes, is important to consider when interpreting the historical evidence. In most of the studies conducted before 1980, rates of pyelonephritis in pregnant women with untreated ASB were at least 10-fold greater than more recently observed. Rates of 2.2% and 2.5% were reported in the 2 most recent trials, yet in 8 of the earlier trials rates were above 20%, with 2 reporting 36% and 33% of women in the control condition developed pyelonephritis.^17,24 The much lower incidence of pyelonephritis in more recent studies may be owing to a range of factors, such as changes in population health status, more stringent diagnostic criteria, better recognition and treatment of lower urinary tract infections, changes in health behaviors, and differences in the infectious microorganisms circulating in the population. Regardless of the reasons, this difference would result in lower absolute differences in risk. meaning higher number needed to treat (NNT) to prevent a single case of pyelonephritis. Assuming 2.5% incidence, for example, 25 women in 1,000 with ASB would develop pyelonephritis in the absence of treatment and applying the pooled RR of 0.24 from this review, 19 cases of pyelonephritis would be prevented for every 1,000 women treated for ASB with antibiotics during pregnancy.

Understanding of the harms of antibiotic use has increased greatly in the 40 years since the seminal trials of ASB treatment in pregnancy were conducted. The emergence of antibiotic-resistant bacteria and the rare but rising incidence of Clostridium difficile infection—including during pregnancy^35–37—have shifted clinical practice toward a more cautious approach to antibiotic use.³⁸ Most recently, research on the microbiome has led to discoveries of protective bacterial colonization, including in the renal system, and growing concern that perturbations caused by antibiotic exposure may influence health.³⁹ In light of this shift in understanding, selection of the type of antibiotic, duration of use, and indications for prescription have become more targeted.

Observational evidence from large health system cohorts have provided complementary evidence that pyelonephritis contributes to poor maternal and fetal health outcomes.^5,40,41 These studies also note that ASB in pregnancy often occurs along with other risk factors associated with poor birth outcomes.^41–44 Associations of ASB and pyelonephritis with poor birth outcomes could therefore also in part arise from shared underlying risk factors. Several risk factors associated with the development of pyelonephritis have also been identified, including younger age, nulliparity, fewer years of education, black or Hispanic race/ethnicity, smoking during pregnancy, late initiation of prenatal care, and pregestational diabetes. Thus, women at risk of developing pyelonephritis in pregnancy, particularly women with limited access to health care, are at risk of poor birth outcomes for a host of reasons. Ensuring adequate screening and interventions for those at high risk for poor outcomes may require system- and policy-level interventions to facilitate early and regular access to prenatal health care.

Limitations

This study has several limitations. First, the review focused on English-language evidence from countries that were ranked “very high” and “high” (for pregnant women) on the Human Development Index, and it is possible that relevant evidence in other languages or settings may exist. Recent evidence reviews on this topic, however, did not identify additional studies that would apply to women obtaining care in the United States.^34,45 Second, the review scope was limited to trials when assessing the effectiveness of treatment. Cohort studies could also have been included in the general adult population, but for this topic cohort studies were expected to have too many threats to internal validity to draw conclusions about effects of ASB treatment in the absence of randomized comparisons. For harms of treatment, general studies of the effects of antibiotic treatment in pregnancy would not be sufficiently guarded against risk of bias from the health consequences of underlying conditions that would require antibiotics. Thus, for treatment benefits and harms, the scope was narrowly defined for study design. Nevertheless, no major cohort studies were identified that would have strengthened the review conclusions if included.

CONCLUSIONS

Screening and treatment for ASB during pregnancy was associated with reduced rates of pyelonephritis and low birth weights, but the available evidence was not current, with only one study conducted in the past 30 years. Benefits of ASB treatment in nonpregnant adult populations were not found. Trial evidence on harms of ASB antibiotic treatment was limited.