Quinolone Consumption by Mothers Increases Their Children’s Risk of Acquiring Quinolone-Resistant Bacteriuria

Abstract

Background

Quinolone resistance has been documented in the pediatric population, although their use is limited in children. This study investigated the effect of maternal quinolone use on gram-negative bacterial resistance to quinolones in their offspring.

Methods

We conducted a population-based, unmatched case-control study during 2010–2017. Cases were all children aged 0.5–17 years with community acquired, gram-negative quinolone-resistant bacteriuria. Controls were similar children with quinolone-sensitive bacteriuria. Only the first positive urine cultures for each child were included. Data on quinolones dispensed to the mother, any antibiotics dispensed to the children, age, sex, ethnicity, and prior hospitalizations were collected. Children with previous quinolone use were excluded.

Results

The study population consisted of 40 204 children. Quinolone resistance was detected in 2182 (5.3%) urine cultures. The median age was 5 years, with 93.7% females and 77.6% Jewish. A total of 26 937 (65%) of the children received any antibiotic and 1359 (3.2%) of the mothers received quinolones in the 6 months preceding bacteriuria. Independent risk factors were quinolone dispensed to the mothers (odds ratio [OR], 1.50 [95% confidence interval {CI}, 1.22–1.85]), Arab ethnicity (OR, 1.99 [95% CI, 1.81–2.19]), and antibiotic dispensed to the child (OR, 1.54 [95% CI, 1.38–1.71]). Compared with children aged 12–17 years, younger children had 1.33–1.43 increased odds for quinolone-resistant bacteriuria.

Conclusions

Quinolone prescription to mothers was linked to increased risk of community-acquired, quinolone-resistant bacteria in their offspring, by about 50%. This is another example of the deleterious ecological effects of antibiotic use and should be considered when prescribing antibiotics.

antibiotics, quinolone resistance, community, mothers, children

Quinolone resistance is a worldwide concern that does not spare the pediatric population. The global prevalence of Escherichia coli resistance to ciprofloxacin in pediatric urinary tract infections is 2% in Organization for Economic Cooperation and Development (OECD) countries and 26% in countries outside the OECD [1].

Traditionally, the major risk factor for resistance to quinolones among adults is previous quinolone exposure [2, 3]. Although quinolones are in limited use in the pediatric population, Qin et al described finding ciprofloxacin-resistant gram-negative bacilli in the fecal microflora of 455 children, none of whom had recent quinolone therapy [4]. This suggested that indirect quinolone exposure is a potential risk factor for quinolone resistance in this population.

A possible source of exposure to quinolones could be other quinolone users in the child’s environment, and the household is the most immediate surrounding. We chose to study the influence of the mother’s quinolone consumption as a test case for household exposure, as she is usually the child’s closest contact. An association between gut colonization with ciprofloxacin-resistant E. coli of healthy children and their mothers was recently described [5], yet the link between antibiotic consumption by mothers and a clinically relevant resistance pattern in their offspring still needs to be established.

Our objective was to evaluate whether quinolone consumption by mothers is a risk factor for quinolone-resistant gram-negative infection in their offspring.

METHODS

Setting

The study was based on electronic medical record data from Clalit Health Services (CHS), which is 1 of 4 nationally mandated health maintenance organizations (HMOs) in Israel. CHS serves 4.5 million persons (54% of the Israeli population). Among those enrolled in CHS, 33% are <18 years of age.

Study Design

This population-based, unmatched, case-control study was conducted from 1 July 2010 to 31 December 2017.

Participants

All pediatric patients aged 0.5–17 years in whom gram-negative bacteriuria was detected were included. Cases were all pediatric patients with quinolone-resistant gram-negative bacteriuria. Controls were all pediatric patients with quinolone-sensitive gram-negative bacteriuria. Only community-acquired bacteriuria cases were included. Exclusion criteria were (1) hospitalization during the 30-day period prior to bacteriuria, to avoid nosocomially acquired resistance; (2) children with no available information regarding antibiotic use among themselves or their mothers 6 months prior to the bacteriuria; (3) children with no consistent HMO coverage to them or to their mothers 6 months prior to the bacteriuria; (4) children who received quinolones during the 6 months preceding bacteriuria; and (5) children who were hospitalized for >1 week in the 6 months preceding the bacteriuria. The last exclusion criterion was introduced to avoid including children with chronic conditions.

Figure 1 describes the process of selecting the study population. The population was stratified into 4 age categories: 0.5-2 years, 3–6 years, 7–11 years, and 12–17 years.

Figure 1.

Flowchart describing the process for patient selection.

Urine Cultures

Community-acquired bacteriuria was defined as a positive urine culture (>10⁴ colony-forming units/mL) taken in the community or emergency department from children who were not hospitalized in the month prior to the culture. According to CHS guidelines, urine culture specimens were obtained via midstream clean catch or catheter, or suprapubic aspiration in incontinent infants. Data were retrieved from community microbiology laboratory records or from hospital microbiology records, in cases of emergency department visit. Only the first episode of gram-negative bacteriuria for each patient was included, regardless of the clinical indication for the culture or the technique by which it was obtained.

Included were urine cultures for which quinolone sensitivity (ciprofloxacin and ofloxacin) testing was performed. Susceptibility patterns were determined with the use of Vitek GN and AST-N098 cards and Vitek II (bioMérieux, Marcy l’Étoile, France). Isolates were considered resistant if they were classified as resistant or intermediate to any of the quinolones, as defined by local laboratories using similar criteria. Urine cultures with >1 organism were excluded (Figure 1). All Clalit HMO microbiology laboratories are International Organozation for Stanrdardizaion 9001: 2015 certified.

Data Source and Variables

The CHS database has nearly 2 decades of detailed, person-level inpatient and outpatient clinical data (diagnoses, laboratory results, medication prescription and dispensing, and imaging studies) [6]. This database was validated previously [7, 8]. For each child, the following information was retrieved from the HMO electronic database: date of birth, sex, ethnic origin, and hospitalizations in the 6 months preceding the first positive urine culture in the child. Any antibiotic dispensed to the child was retrieved from the HMO community pharmacy database for the same period. The number of antibiotic prescriptions dispensed to each child was stratified into none, 1, or ≥2. Each new antibiotic prescription was counted, with no minimum interval between prescriptions. Of note, over-the-counter dispensing is illegal in Israel and to the best of our knowledge, nearly nonexistent. The antibiotics dispensed by all pharmacies (private or CHS) are recorded in the central database. Children and mothers were linked using identification number in the HMO database. For the mothers, we retrieved information regarding hospitalizations during the 6 months before the first positive urine culture in the child and any antibiotic dispensed for the same period. Antibiotics dispensed to mothers were stratified into prescription of quinolones or any other antibiotics and was defined as none or any.

This study was approved by the CHS Ethics Committee.

Statistical Analysis

Statistical analyses were performed using SPSS version 23. Results were considered significant when the P value was <.05, using a 2-sided test.

A contingency table was used to assess crude associations between the variables and the child’s odds of infection with quinolone-resistant bacteria. All variables found to be significant by univariate analysis were considered for inclusion in the multivariate analysis.

Stratified analyses and unconditional logistic regression were used to assess independent association between a mother’s use of quinolones and her children’s odds of infection with gram-negative, quinolone-resistant bacteria. Cross-level and intralevel interactions were checked for significance. To further separate the possible effects of children’s consumption of antibiotics, we conducted a post hoc subgroup analysis of children who had not used antibiotics within 1 year of the index culture.

RESULTS

There were 85 505 episodes of community-acquired bacteriuria among children during the study period. A total of 40 204 met the inclusion criteria and constituted the final study population of 2090 cases of quinolone-resistant bacteriuria and 38 114 controls (Figure 1). The children had a median age of 5 years. There were 37 699 (93.8%) females, and 31 302 (77.9%) were Jewish. We captured hospitalizations in 1003 (2.5%) children and 1478 (3.7%) mothers in the 6 months preceding the bacteriuria. Children hospitalized in the month prior to the bacteriuria were excluded from the study to avoid capturing hospital-acquired bacteriuria.

A total of 26 206 children (65.2%) received an antibiotic during the 6 months preceding the bacteriuria, among whom 11 210 (27.9%) received ≥2 prescriptions. Quinolones were dispensed to 1319 (3.3%) mothers. Other antibiotics were dispensed to 13 515 (33.6%) mothers. The most frequent isolate was E. coli (80.3%), followed by Proteus, Klebsiella, and Pseudomonas species (8.1%, 5.6%, and 2%, respectively).

Quinolone or any other antibiotic dispensed to the mother, at least 2 antibiotic prescriptions dispensed to the child, Arab ethnicity, and age were significantly (P < .001) associated with quinolone-resistant bacteriuria (Table 1).

Table 1.

Characteristics of the Study Population

Characteristic	Total	Quinolone Sensitive	Quinolone Resistant	P Value
Total	40 204	38 114 (94.8)	2090 (5.2)
Sex
Female	37 699 (93.8)	35 739 (93.8)	1960 (93.8)	.995
Male	2505 (6.2)	2375 (6.2)	130 (6.2)
Age group, y
0.5–2	9855 (24.5)	9355 (24.5)	500 (23.9)	< .001
3–6	13 821 (34.4)	13 039 (34.2)	782 (37.4)
7–11	7926 (19.7)	7455 (19.6)	471 (22.5)
12–17	8602 (21.4)	8265 (21.7)	337 (16.1)
Ethnic origin
Jewish	31 302 (77.9)	29 950 (78.6)	1352 (64.7)	< .001
Arab	8902 (22.1)	8164 (21.4)	738 (35.3)
Hospitalization of childa	1003 (2.5)	947 (2.5)	61 (2.9)	.194
No. of antibiotic prescriptions dispensed to child
0	13 998 (34.8)	13 369 (35.1)	629 (30.1)	< .001
1	14 996 (37.3)	14 297 (37.5)	699 (33.4)
≥2	11 210 (27.9)	10 448 (27.4)	762 (36.5)
Quinolone dispensed to mother	1319 (3.3)	1216 (3.2)	103 (4.9)	< .001
Other antibiotics dispensed to mother	13 515 (33.6)	12 743 (33.4)	772 (36.9)	.001
Hospitalization of motherb	1478 (3.7)	1388 (3.6)	90 (4.3)	.116

Data are presented as No. (%) unless otherwise indicated.

^aHospitalization >1 mo and <6 mo prior.

^bHospitalization <6 mo prior.

A multivariable analysis showed that quinolone dispensed to the mother (OR, 1.50 [95% confidence interval {CI}, 1.22–1.85]), at least 2 antibiotic prescriptions dispensed to the child (OR, 1.54 [95% CI, 1.38–1.71]), and Arab ethnicity (OR, 1.99 [95% CI, 1.81–2.19]) were independent risk factors for resistance. Compared with children aged 12–17 years, younger children had 1.33–1.44 increased odds for quinolone-resistant bacteriuria, with increasing odds at younger ages (Table 2).

Table 2.

Multiple Logistic Regression Analysis

Characteristic	Total Population			Children Who Did Not Use Antibiotics
	OR	(95% CI)	P Value	OR	(95% CI)	P Value
Sex
Male	1	…		1	…
Female	1.01	(.83–1.21)	.96	0.99	(.73–1.36)	.97
Age group, y
12–17	1	…		1	…
7–11	1.33	(1.15–1.54)	< .001	1.44	(1.27–1.84)	.004
3–6	1.38	(1.21–1.58)	< .001	1.43	(1.14–1.78)	.002
0.5–2	1.43	(1.24–1.65)	< .001	1.44	(1.28–1.83)	.003
Ethnicity
Jewish	1	…		1	…
Arab	1.99	(1.81–2.19)	< .001	1.78	(1.50–2.11)	< .001
No. of antibiotic prescriptions dispensed to child
0	1	…		…	…
1	1.06	(.95–1.18)	.329	…	…
≥2	1.54	(1.38–1.71)	< .001	…	…
Quinolones dispensed to mother	1.50	(1.22–1.85)	< .001	1.62	(1.10–2.44)	.015
Other antibiotics dispensed to mother	1.05	(.96–1.15)	0.31	…	…

Abbreviations: CI, confidence interval; OR, odds ratio.

A subgroup analysis excluding children who received any antibiotic course showed similar results. Moreover, the odds of quinolones dispensed to the mothers increased from 1.50 to 1.62 (Table 2).

DISCUSSION

In this study of >40 000 mother-child pairs, quinolone consumption by the mother was an independent risk factor for the presence of quinolone-resistant bacteria in the urine of their offspring, with an OR of 1.5. The relative impact of parental antibiotic use on resistance among children is difficult to assess because antibiotics are often prescribed to the pediatric population. This makes it very difficult to differentiate between the impact of parental antibiotic use vs that of the child. Our study design overcame this limitation. As quinolones are seldom used in the pediatric population, we were able to isolate the impact of parental quinolone use. This effect was also found in a subgroup analysis of children who were not prescribed any antibiotics.

A possible hypothesis to explain our results is that quinolone consumption by mothers led to gut colonization of the mother with quinolone-resistant bacteria, which were transmitted to their offspring. This resulted in resistant bacteria colonization in the child, serving as reservoir for ensuing clinical infection.

Although clusters of household members colonized with the same resistant bacteria have been described, the pathway that facilitates this phenomenon is speculative. Common sources such as drinking water [9], pets [10, 11], and dietary preferences [9] have been implicated as contributing to shared resistance.

Several studies reported person-to-person transmission of antibacterial resistance to extended-spectrum β-lactamases (ESBLs) among household members [12, 13]. However, the transmission kinetics of ESBLs, which are via plasmid, might not be applicable to transmission kinetics of quinolone, where resistance is primarily chromosomal.

Studies regarding the relative contribution of antibiotic exposure by household members to antimicrobial resistance within a family are few and have yielded conflicting results. In a study of 18 mother-child pairs, Gurnee et al reported that healthy infants and their mothers commonly harbored ciprofloxacin-resistant E. coli in their gut [5], yet prior antibiotic consumption by the mothers or the children were not a risk factor for acquiring resistance. This lack of association could be attributed to the small size of the study population. In contrast, Stewardson et al investigated the effect of ciprofloxacin on the gut flora of 716 patients and their untreated household members. They found increased prevalence of ciprofloxacin-resistant Enterobacteriaceae among untreated household contacts of treated patients [14]. Similar results were reported by Hannah et al; they looked at 517 study participants and observed that recent use of antibiotics in their households was associated with increased risk for intestinal carriage of nalidixic acid–resistant E. coli [15]. The findings of these 2 small studies are in agreement with our results. Nevertheless, their focus was colonization of gut flora in asymptomatic study participants, in which stool cultures were obtained only for the purpose of the study. The clinical relevance of their findings was not addressed and is speculative because the natural history of gut colonization in the community has not been studied in depth. We purposely chose not to study gut colonization of resistant bacteria because we were interested in the clinical relevance of transmitted resistance. In contrast, our population-based study evaluated urine cultures taken in real life, thus illustrating this phenomenon and its magnitude in a clinical context.

We investigated possible known confounders that might have influenced both quinolone consumption by the mother and the presence of resistant bacteria in their offspring. Prior antibiotic consumption by either mother or the child might reflect a common culture of antibiotic use or healthcare provider. Hospitalization of the mother might result in acquired resistance irrespective of quinolone use. Ethnicity is a marker for socioeconomic and cultural habits.

Prior dispensing of an antibiotic other than quinolones to the children was found to be an independent risk factor for quinolone resistance in the urine, with OR 1.54. It is well documented that antibiotic use is a risk factor for resistance to the same antibiotic [16, 17], as well as to other antibiotics. A specific correlation between use of antibiotics such as aminoglycoside, cephalosporin, amoxicillin-clavulanate, and nitrofurantoin with quinolone resistance was described among adults [18–21]. To the best of our knowledge, this correlation was not previously described in the pediatric population. A systematic review describing global prevalence of urinary antibiotic resistance among the pediatric population tested almost 54 000 isolates for antibiotic resistance [1]. Previous antibiotic exposure was a risk factor for resistant bacteria. Yet, no single study in this meta-analysis specifically assessed the effect of antibiotic use on quinolone resistance. Our study of 40 204 child-mother pairs enabled us to reveal the effect of antibiotic consumption, other than quinolones, on quinolone resistance among children. Notably, this effect was present up to 6 months after antibiotic use. In contrast, use of antibiotics other than quinolones by the mother was not an independent risk factor for quinolone-resistant bacteriuria in the child.

The strongest risk factor for resistant bacteriuria was Arab ethnicity, a minority that comprised 22% of the cohort population and generally has a lower socioeconomic status in Israel [22]. Lower socioeconomic status correlates with high prevalence of antimicrobial resistance worldwide [1, 23–25]. This correlation is mainly mediated by increased antibiotic use. Nonetheless, in our study, ethnicity was found to be an independent risk factor for resistance.

Age was also found to influence the resistance pattern. We observed that in comparison to teens, all other age groups were at increased risk for developing resistance. One possible link is infant and toddler day care attendance, which is common in Israel [26]. Notably, the correlation decreased with increasing age of the child, which might reflect the close contact between mother and child in infancy, which declines as the child grows.

Fan et al found hospitalization to be a risk factor for resistance in children. However, this observation specifically pertained to hospitalization during the previous month [27]. As the focus of our study was resistance patterns in the community, we excluded children hospitalized in the month prior to the bacteriuria event. In our study, previous hospitalization of the child 1–6 months prior was not found to be a risk factor for resistance.

The main limitation of the current study was the lack of data regarding possible confounders such as travel, household crowding, and dietary habits. Of note, as mentioned above, ethnicity might be a surrogate marker for varying dietary habits and crowding. Another limitation is lack of clinical data. Some of the urine cultures might represent asymptomatic bacteriuria. To minimize this risk, we took only the first episode of bacteriuria, thereby avoiding repeat cultures after treatment completion. Moreover, urine cultures are not routinely taken from asymptomatic children. Of note, few of the older children might have had a previous urine culture before study period. Lack of clinical data prohibited us from identifying a subpopulation of children with chronic medical conditions who are frequently exposed to the healthcare system. Therefore, we excluded children with long-term hospitalizations, which might be a marker of a chronic disorder. This assumption was validated using data from a small number of children (data not shown). However, it is not expected to confound the results, as it is not likely to be associated with maternal quinolone consumption. We did not capture the timing of transmitted resistance; thus, colonization of the child prior to quinolone consumption by the mother is possible, but as this is expected to be a nondifferential misclassification, it could only decrease the association.

In conclusion, in this large-scale, population-based study, quinolones prescribed to mothers were linked to increased resistance in their offspring. This result is another example of the deleterious effect of quinolone use. Although we used quinolones as a model, the effect might be wider and project to the use of other antibiotics.

Notes

Author contributions. M. C. had full access to all of the data in the study and had final responsibility for the decision to submit for publication. Faye Schreiber, MS, an employee of Meir Medical Center, edited the manuscript.

Financial support. This work was supported by internal departmental resources.

Potential conflicts of interest. The authors report no potential conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Author notes

B.-S. G. and M. L. contributed equally to this work.