Association between prior antibiotic therapy and subsequent risk of community-acquired infections: a systematic review

Association between prior antibiotic therapy and subsequent risk of community-acquired... Abstract Background Antibiotic use can have negative unintended consequences including disruption of the human microbiota, which is thought to protect against pathogen overgrowth. We conducted a systematic review to assess whether there is an association between exposure to antibiotics and subsequent risk of community-acquired infections. Methods We searched MEDLINE, EMBASE and Web of Science for studies published before 30 June 2017, examining the association between antibiotic use and subsequent community-acquired infection. Infections caused by Clostridium difficile and fungal organisms were excluded. Studies focusing exclusively on resistant organism infections were also excluded. Results Eighteen of 22588 retrieved studies met the inclusion criteria. From these, 16 studies reported a statistically significant association between antibiotic exposure and subsequent risk of community-acquired infection. Infections associated with prior antibiotic use included Campylobacter jejuni infection (one study), recurrent furunculosis (one study), invasive Haemophilus influenzae type b infection (one study), infectious mastitis (one study), meningitis (one study), invasive pneumococcal disease (one study), Staphylococcus aureus skin infection (one study), typhoid fever (two studies), recurrent boils and abscesses (one study), upper respiratory tract infection and urinary tract infection (one study) and Salmonella infection (five studies), although in three studies on Salmonella infection the effect was of marginal statistical significance. Conclusions We found an association between prior antibiotic use and subsequent risk of a diverse range of community-acquired infections. Gastrointestinal and skin and soft tissue infections were most frequently found to be associated with prior antibiotic exposure. Our findings support the hypothesis that antibiotic use may predispose to future infection risk, including infections caused by both antibiotic-resistant and non-resistant organisms. Introduction Antibiotics are an important weapon in our fight against bacterial infections. Antibiotic use is associated with recognized harms, including the emergence of antibiotic-resistant organisms.1 There is concern that straightforward infections that are currently easy to control may become untreatable in the future. The growing threat of antimicrobial resistance has been declared a global public health crisis.2 Much attention has been focused on limiting inappropriate antibiotic usage as a strategy to control drug resistance, particularly in primary care, where 90% of antibiotic prescriptions are issued.1 Even so, in the UK, most general practices continue to prescribe antibiotics at rates in excess of what is clinically justified.3 Antibiotic therapy can have further unintended consequences, other than selection of resistant microorganisms. The human microbiota is a complex community of up to a hundred trillion microorganisms lining the epithelial surfaces of their hosts and that are exposed to the outside world.4 The microbiota influences human health through its role in a diverse range of physiological functions,4 including a protective role in defence against pathogens. The microbiota is thought to contribute to development of the host’s immune system and its response to infection.5 The huge range of organisms in the microbiota, commensals and potential pathogens, compete with each other for attachment sites and nutrients, thereby preventing pathogen overgrowth.6 Exposure to antibiotics can change the composition of the microbiota, reducing microbial diversity and allowing the overgrowth of potentially harmful microorganisms.6 Short-term antibiotic treatment for Helicobacter pylori eradication, for example, reduces microbial diversity in the human throat and gut microbiota, which in some cases persists for up to 4 years.7 The question arises whether antibiotics prescribed to treat an acute bacterial infection could paradoxically predispose individuals to future infections due to collateral damage inflicted on the microbiota. Clostridium difficile-associated diarrhoea is a well-recognized example of an infection resulting from pathogenic colonization of a microbial community disrupted by recent antibiotic use.8,9 Antibiotic treatment is also associated with increased risk of fungal infection.10–12 Are there other medium- and long-term infection risks associated with antibiotic use unrelated to increased resistance, or infection with C. difficile or opportunist fungi? This systematic review sought evidence of the association between antibiotic use and subsequent risk of infection in the community setting, unrelated to these previously known infection risks. Objectives To examine whether exposure to antibiotic therapy is associated with subsequent increased risk of community-acquired infections. Methods Procedures used in this review were consistent with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.13 Protocol and registration A review protocol was submitted in advance to PROSPERO, which is a database of systematic review protocols (registration ID: CRD42016048521).14 Eligibility criteria Our inclusion criteria were: (i) studies, of any design, evaluating the risk of community-acquired infection associated with prior antibiotic usage; (ii) human participants, of any age, in the community setting; (iii) studies that assessed risk of infection associated with prior antibiotic exposure compared with the risk of infection without prior antibiotic exposure, in the same period; and (iv) an endpoint of any community-acquired infection. Our exclusion criteria were: (i) studies focused on immune-compromised patient groups (e.g. patients with human immunodeficiency virus) as these would not be representative of the general population; (ii) studies focusing on infections that are known to be associated with prior antibiotic exposure including fungal infection10–12 and C. difficile infection;9 (iii) studies focusing exclusively on infections caused by antibiotic-resistant organisms as these have been the subject of a previous systematic review15 (furthermore, these studies would not provide a true estimate of the overall risk of infection following antibiotic exposure because infections resulting from non-resistant organisms would be excluded); (iv) studies assessing the risk of nosocomial infections because their relationship with prior antibiotic use could be confounded by other factors (e.g. vulnerable patients, infectious contacts, invasive procedures); (v) non-English language articles; (vi) reviews, letters, editorials and case reports; (vii) studies assessing the impact of prior antibiotic use on complications resulting from a specific infection (e.g. mortality rate); and (viii) studies examining the use of antibiotics as a prophylactic measure to prevent a specific infection (e.g. post-operative wound infection) because such studies are not designed to assess the impact of antibiotic exposure on the subsequent risk of acquiring other unrelated infections. Search strategy The initial literature search was performed on 29 April 2016. The search was repeated on 30 June 2017 to identify further articles published since the date of the initial search. The databases searched were MEDLINE, EMBASE and Web of Science. No restrictions were imposed on the publication period. Search terms included both text words and MESH terms. Detailed search strategies are presented in Table S1 (available as Supplementary data at JAC Online). The reference lists of eligible studies and review articles were screened to identify further studies eligible for inclusion. Study selection Covidence software was used to facilitate the screening and selection of studies.16 The first reviewer (U. M.) conducted the literature search, removed duplicate articles, and screened titles and abstracts with respect to the eligibility criteria. Full-text articles of potentially relevant studies were independently assessed for eligibility by two reviewers (U. M. and P. W.). Any disagreements were resolved through discussion and consensus. Data extraction One reviewer (U. M.) extracted data from the full texts of included studies. Extracted data were summarized in a predesigned format and cross-checked by a second reviewer (P. W.). Disagreements were resolved through discussion and consensus. Data collected included the first author, publication year, country of study, study design, study objectives, participant characteristics, exposure definition (i.e. antibiotic class, length of treatment, time from exposure to infection), exposure interval (length of time prior to the infection during which antibiotic exposure was assessed), exposure ascertainment method, infection type, definition of cases and the comparison group, sample size, main outcome measure and confounding variables. The primary outcome of interest was the OR for the association of infection with prior antibiotic exposure. Quality assessment The Newcastle Ottawa Scale (NOS), a tool for assessing the quality of non-randomized studies,17 was applied independently to each study by two reviewers (U. M. and P. W.). Any disagreements were resolved through discussion and consensus. Each study could be assigned a maximum of nine points; four for the selection domain, two for the comparability domain and three for the exposure domain. Studies assigned NOS scores of ≥7 were considered high quality, 5–6 were considered moderate quality and <5 were considered low quality. Data synthesis and analysis Our research question is broad, aiming to examine the association between past antibiotic consumption and a diverse range of infections, at different body sites, caused by a variety of different microorganisms. Quantitative pooling of results through a meta-analysis was not justified owing to the high degree of clinical heterogeneity expected between studies in terms of the infection types being investigated and differences in antibiotic exposures. Any observed association between antibiotic use and subsequent infection was considered unlikely to be universal for all infection types. We have therefore reported the findings of the systematic review as a narrative (descriptive) synthesis. Results Study selection The literature search retrieved 22582 publications. Six additional publications were identified from the reference list of retrieved articles. After adjusting for duplicates, 21439 publications remained. After screening by title and abstract, 21053 publications were excluded. A full-text review was conducted of 386 publications, from which 18 met the eligibility criteria and were included in the review.18–35 The detailed rationale behind the exclusion of studies is presented in Figure 1. Figure 1 View largeDownload slide PRISMA flow chart. Figure 1 View largeDownload slide PRISMA flow chart. Study design and participant characteristics Tables 1 and 2 summarize the main characteristics of the included studies. Sixteen studies were case–control studies and two were cohort studies. Table 1 Data extraction table for case–control studies (in alphabetical order of infection studied) First author  Publication year; country  Infection studied  Antibiotic; exposure interval  Case definition  Comparison group  Sample size  Fully adjusted outcome (95% CI)  Confounder variables adjusted for  Effler26  2001; USA  C. jejuni  any; 28 days  lab confirmed; age: not available  age and telephone exchange matched  211 cases; 211 controls  OR 3.3 (1.1–9.6)  consumption of various food items, acid-suppressing drugs and contact with live chickens  El-Gilany28  2009; Egypt  furunculosis (recurrent)  any; 1 year  clinic attendance with ≥3 attacks of boils within the previous 12 months; all ages  attendance of the same clinic diagnosed with furunculosis for the first time  74 cases; 74 controls  OR 16.6 (2.2–66.0)  family history, diabetes mellitus, anaemia, previous hospitalization, personal hygiene, skin diseases and number of lesions  McVernon32  2008; UK  H. influenzae type b (invasive)  frequency of previous antibiotic use  lab confirmed cases; age 5 years to 9 years and 11 months  matched by date of birth and region  136 cases; 295 controls  frequent use: OR 1.51 (1.06–2.13)  sex, prematurity, breastfed, past illness, family demographic factors, bedroom sharing, smoking, central heating, home ownership and vaccination status  Mediano35  2014; Spain  mastitis (infectious)  any; during pregnancy  lab confirmed; lactating females  healthy breastfeeding women with no clinical symptoms of mastitis and negative milk culture  368 cases; 148 controls  OR 5.38 (2.85–10.14)  age, personal and family history, infection, comorbidities, drugs and pregnancy, childbirth and breastfeeding-related factors  Armstrong18  2016; UK  meningitis  any; 1 year  identified from GP records; all ages  matched on age, sex, GP practice and index date  7346 cases; 29384 controls  OR 2.04 (1.91–2.18)  13 variables including demographic factors, lifestyle, comorbidities and medications  Chun19  2015; USA  pneumococcal disease (invasive)  any; 3 months  lab confirmed; age 0–12 years  matched by age, health plan membership and length of membership  171 cases; 342 controls  OR 1.57 (1.06–2.33)  sex, race, age, risk status, health plan membership and pneumococcal vaccination  Doorduyn20  2016; the Netherlands  Salmonella  any; 4 weeks  lab confirmed; all ages  matched for age, sex, degree of urbanization and season  193 cases; 3119 controls  OR 1.9 (1.0–3.4)  age, sex, degree of urbanization and education  Gradel29  2008; Denmark  Salmonella  any; 1 year  lab confirmed; age 1–99 years  residents of same county matched for specimen date, gender and age  1882 cases; 18820 controls  OR 1.59 (1.43–1.77)  gender, antibiotic score,a patient group, age, NTS infection month and serovar  Delarocque- Astagneau21  2000; France  Salmonella  any; 1 month  lab confirmed; age <14 years  matched for age and place of residence  101 cases; 101 controls  OR 2.3 (1.0–5.5)  consumption of various food items  Banatvala25  1999; UK  Salmonella  any; 1 month  lab confirmed; all ages  (1) case nominated and matched for age, gender and area of residence; (2) randomly selected from London area  209 cases and matched controls; 854 random controls  matched: OR 1.3 (0.6–2.8); unmatched: OR 1.3 (0.6–2.6)  not available  Neal23  1994; UK  Salmonella  any; 1 year  lab confirmed and notified; age ≥45 years  next two patients in the practice records system matched for age and sex  188 cases; 376 controls  past year: OR 1.4 (1.0–2.1)  gastric surgery, H2 antagonist treatment and other drug use  Kass22  1992; USA  Salmonella  any; ‘recent’  lab confirmed and notified; age ≥10 years  matched for region  120 cases; 265 controls  OR 1.96 (0.86–4.37)  prior health problems and prior medical therapies  Pavia24  1989; USA  Salmonella  any; 30 days  lab confirmed; age >1 year  matched for age, neighbourhood and telephone exchange  35 cases; 70 controls  OR 3.8 (1.2–11.9)  immunosuppression and use of antacids or H2-blocking agents  Early27  2012; Hawaii  S. aureus skin infection  any; 6 months  lab confirmed; age 6 months to 17 years  matched for age, clinician and date of clinician visit  71 cases; 146 controls  OR 2.90b (1.29–6.61)  household contact, abrasions/wounds, skin disorders, weight and sharing bed linens and towels  Srikantiah33  2007; Uzbekistan  typhoid fever  any; 2 weeks  lab confirmed; all ages  age- and community-matched controls  97 cases; 192 controls  OR 12.2 (4.0–37.0)  occupation and consumption of various food and drink items  Luby30  1998; Pakistan  typhoid fever  any; 2 weeks  lab confirmed; all ages  neighbourhood and age matched  100 cases; 200 controls  OR 5.7 (2.3–13.9)  consumption of various food and drink items  First author  Publication year; country  Infection studied  Antibiotic; exposure interval  Case definition  Comparison group  Sample size  Fully adjusted outcome (95% CI)  Confounder variables adjusted for  Effler26  2001; USA  C. jejuni  any; 28 days  lab confirmed; age: not available  age and telephone exchange matched  211 cases; 211 controls  OR 3.3 (1.1–9.6)  consumption of various food items, acid-suppressing drugs and contact with live chickens  El-Gilany28  2009; Egypt  furunculosis (recurrent)  any; 1 year  clinic attendance with ≥3 attacks of boils within the previous 12 months; all ages  attendance of the same clinic diagnosed with furunculosis for the first time  74 cases; 74 controls  OR 16.6 (2.2–66.0)  family history, diabetes mellitus, anaemia, previous hospitalization, personal hygiene, skin diseases and number of lesions  McVernon32  2008; UK  H. influenzae type b (invasive)  frequency of previous antibiotic use  lab confirmed cases; age 5 years to 9 years and 11 months  matched by date of birth and region  136 cases; 295 controls  frequent use: OR 1.51 (1.06–2.13)  sex, prematurity, breastfed, past illness, family demographic factors, bedroom sharing, smoking, central heating, home ownership and vaccination status  Mediano35  2014; Spain  mastitis (infectious)  any; during pregnancy  lab confirmed; lactating females  healthy breastfeeding women with no clinical symptoms of mastitis and negative milk culture  368 cases; 148 controls  OR 5.38 (2.85–10.14)  age, personal and family history, infection, comorbidities, drugs and pregnancy, childbirth and breastfeeding-related factors  Armstrong18  2016; UK  meningitis  any; 1 year  identified from GP records; all ages  matched on age, sex, GP practice and index date  7346 cases; 29384 controls  OR 2.04 (1.91–2.18)  13 variables including demographic factors, lifestyle, comorbidities and medications  Chun19  2015; USA  pneumococcal disease (invasive)  any; 3 months  lab confirmed; age 0–12 years  matched by age, health plan membership and length of membership  171 cases; 342 controls  OR 1.57 (1.06–2.33)  sex, race, age, risk status, health plan membership and pneumococcal vaccination  Doorduyn20  2016; the Netherlands  Salmonella  any; 4 weeks  lab confirmed; all ages  matched for age, sex, degree of urbanization and season  193 cases; 3119 controls  OR 1.9 (1.0–3.4)  age, sex, degree of urbanization and education  Gradel29  2008; Denmark  Salmonella  any; 1 year  lab confirmed; age 1–99 years  residents of same county matched for specimen date, gender and age  1882 cases; 18820 controls  OR 1.59 (1.43–1.77)  gender, antibiotic score,a patient group, age, NTS infection month and serovar  Delarocque- Astagneau21  2000; France  Salmonella  any; 1 month  lab confirmed; age <14 years  matched for age and place of residence  101 cases; 101 controls  OR 2.3 (1.0–5.5)  consumption of various food items  Banatvala25  1999; UK  Salmonella  any; 1 month  lab confirmed; all ages  (1) case nominated and matched for age, gender and area of residence; (2) randomly selected from London area  209 cases and matched controls; 854 random controls  matched: OR 1.3 (0.6–2.8); unmatched: OR 1.3 (0.6–2.6)  not available  Neal23  1994; UK  Salmonella  any; 1 year  lab confirmed and notified; age ≥45 years  next two patients in the practice records system matched for age and sex  188 cases; 376 controls  past year: OR 1.4 (1.0–2.1)  gastric surgery, H2 antagonist treatment and other drug use  Kass22  1992; USA  Salmonella  any; ‘recent’  lab confirmed and notified; age ≥10 years  matched for region  120 cases; 265 controls  OR 1.96 (0.86–4.37)  prior health problems and prior medical therapies  Pavia24  1989; USA  Salmonella  any; 30 days  lab confirmed; age >1 year  matched for age, neighbourhood and telephone exchange  35 cases; 70 controls  OR 3.8 (1.2–11.9)  immunosuppression and use of antacids or H2-blocking agents  Early27  2012; Hawaii  S. aureus skin infection  any; 6 months  lab confirmed; age 6 months to 17 years  matched for age, clinician and date of clinician visit  71 cases; 146 controls  OR 2.90b (1.29–6.61)  household contact, abrasions/wounds, skin disorders, weight and sharing bed linens and towels  Srikantiah33  2007; Uzbekistan  typhoid fever  any; 2 weeks  lab confirmed; all ages  age- and community-matched controls  97 cases; 192 controls  OR 12.2 (4.0–37.0)  occupation and consumption of various food and drink items  Luby30  1998; Pakistan  typhoid fever  any; 2 weeks  lab confirmed; all ages  neighbourhood and age matched  100 cases; 200 controls  OR 5.7 (2.3–13.9)  consumption of various food and drink items  NTS infection, non-typhoid Salmonella infection. a Antibiotic score (0–5); antibiotics were classified according to their potential impact on the intestinal flora, with higher scores being associated with increasing impact on the intestinal flora. b Association reached significance for the Native Hawaiian and Pacific Islander (NHPI) ethnic category, but not for the non-NHPI ethnic category (OR = 1.93, 95% CI = 0.93–4.01). Table 2. Data extraction table for cohort studies First author  Publication year; country  Data source and study population  Infection studied  Exposure definition  Case definition and follow-up  Comparison  Sample size  Fully adjusted outcome (95% CI)  Confounder variables adjusted for  Shallcross34  2015; UK  registered with THIN- participating GP practice, any age and sought care for a boil, abscess, carbuncle or furuncle  recurrent boil or abscess  antibiotic prescription in the 6 months prior to the date of index consultation  second consultation for a boil or abscess within 3 weeks–12 months  patients without a repeat consultation for boil or abscess  cohort of 164461 patients; 10% developed a repeat boil or abscess  RR 1.4 (1.3–1.4)  age, sex, BMI, diabetes, skin disease, prior antibiotic and smoking status  Margolis31  2005; UK  registered with GPRD- participating GP practice, aged 15–35 years and recorded diagnosis of acne vulgaris  (1) upper respiratory tract infection (URTI); (2) urinary tract infection (UTI).  prescription >6 weeks of oral erythromycin or an oral tetracycline or topical erythromycin or clindamycin or a combination of both  UTI or URTI within 12 months after enrolment  patients without acne antibiotic use were considered unexposed  84997 exposed; 33519 unexposed  URTI: OR 2.23 (2.12–2.34); UTI: OR 1.10a (1.01–1.19)  age, year of diagnosis, sex, contraceptive use or counselling (only for UTIs), practice, diabetes, asthma, visit frequency and the number of prescriptions for acne antibiotics  First author  Publication year; country  Data source and study population  Infection studied  Exposure definition  Case definition and follow-up  Comparison  Sample size  Fully adjusted outcome (95% CI)  Confounder variables adjusted for  Shallcross34  2015; UK  registered with THIN- participating GP practice, any age and sought care for a boil, abscess, carbuncle or furuncle  recurrent boil or abscess  antibiotic prescription in the 6 months prior to the date of index consultation  second consultation for a boil or abscess within 3 weeks–12 months  patients without a repeat consultation for boil or abscess  cohort of 164461 patients; 10% developed a repeat boil or abscess  RR 1.4 (1.3–1.4)  age, sex, BMI, diabetes, skin disease, prior antibiotic and smoking status  Margolis31  2005; UK  registered with GPRD- participating GP practice, aged 15–35 years and recorded diagnosis of acne vulgaris  (1) upper respiratory tract infection (URTI); (2) urinary tract infection (UTI).  prescription >6 weeks of oral erythromycin or an oral tetracycline or topical erythromycin or clindamycin or a combination of both  UTI or URTI within 12 months after enrolment  patients without acne antibiotic use were considered unexposed  84997 exposed; 33519 unexposed  URTI: OR 2.23 (2.12–2.34); UTI: OR 1.10a (1.01–1.19)  age, year of diagnosis, sex, contraceptive use or counselling (only for UTIs), practice, diabetes, asthma, visit frequency and the number of prescriptions for acne antibiotics  THIN, The Health Improvement Network; GPRD, General Practice Research Datalink, now known as Clinical Practice Research Datalink (CPRD); RR, relative risk. a The OR for UTI is statistically significant; however, the authors of the study concluded that it was not clinically meaningful. The studies involved a total of 349085 participants. Study sample sizes ranged from 10524 to 16446134 participants. In 13 studies the hypothesis that antibiotic use would increase subsequent risk of infections was not stated19–22,24,26–28,30,32–35 and earlier antibiotic use was one of many other variables assessed as potential risk factors for infection. Exposure assessment and definition Six studies ascertained exposure status from recorded prescription data.18,19,23,29,31,34 Ten studies determined exposure status through participant interview or questionnaire.20–22,26–28,30,32,33,35 One study interviewed both participants and their physician,24 and in one study medication data was obtained from the physician only when the participant was unsure.25 Seventeen studies defined exposure as the consumption of any antibiotic medication. One study defined exposure in terms of specific antibiotic classes.31 In one study the exposure status was defined as an antibiotic prescription of 6 weeks or more for acne.31 In the remaining 17 studies, exposure was not defined in terms of the duration of antibiotic course. The length of time during which antibiotic exposure was assessed prior to the infection (exposure interval) was specified in 14 out of 18 studies.18–21,23–30,33,34 In these studies, the exposure interval ranged from 2 weeks to 1 year. The median exposure interval was 2 months. Case definition Fourteen studies defined infection on the basis of a positive microbiological specimen.19–27,29,30,32,33,35 Three studies defined infection on the basis of routinely recorded primary care data.18,31,34 One study on recurrent furunculosis did not specify how the cases were defined.28 Comparison group In case–control studies the control group consisted of participants who did not have a confirmed history of the infection of interest. In the cohort study on patients with acne, risk of infection was compared between those who were prescribed long-term antibiotics for acne, versus those who were not.31 In the cohort study on risk of recurrent boil and abscess, antibiotic prescription rates were compared amongst participants who had a repeat consultation for a boil or abscess within 12 months, versus those who did not.34 Associations between exposure and outcome Positive and significant associations between prior antibiotic use and subsequent risk of infection were reported in 16 out of 18 studies, including one study on Campylobacter jejuni infection,26 one study on recurrent furunculosis,28 one study on invasive Haemophilus influenzae type b infection,32 one study on infectious mastitis,35 one study on meningitis,18 one study on invasive pneumococcal disease,19 one study on Staphylococcus aureus skin infection,27 two studies on typhoid fever,30,33 one study on recurrent boils and abscesses,34 one study on upper respiratory tract infection and urinary tract infection31 and five studies on Salmonella infection.20,21,23,24,29 The outcome measure and 95% CI from each study is presented in Tables 1 and 2. Salmonella infection Nine out of 18 studies assessed the association between antibiotic use and subsequent risk of Salmonella infection.20–25,29,30,33 As this was the most frequently studied infection the results are described here in further detail. Two out of nine studies focused exclusively on cases of typhoid fever.30,33 Both studies reported a positive and statistically significant association between prior antibiotic use and subsequent risk of typhoid fever. The remaining seven studies assessing the risk of Salmonella infection did not restrict the definition of cases to typhoid fever.20–25,29 Five out of seven studies reported a positive and significant association between prior antibiotic use and risk of Salmonella infection.20,21,23,24,29 From these five studies, three studies reported 95% CIs for the OR that included 1.0 at the lower end.20,21,23 In these studies the effect was of marginal statistical significance. We found two negative studies, which did not demonstrate a statistically significant association between prior antibiotic use and subsequent risk of Salmonella infection.22,25 Timing of antibiotic exposure Four studies assessed the effect of timing of antibiotic exposure on subsequent risk of infection.18,19,23,29 Increased risk of community-acquired infection after antibiotic exposure was documented up to 1 year in three18,23,29 out of four studies except for one study on pneumococcal disease19 that showed an increased risk only in the following month. In the case of meningitis,18 the association between antibiotic exposure and subsequent infection risk remained statistically significant over the 12 month exposure period prior to the infection, although as the interval between exposure and subsequent infection increased, the effect size generally diminished (0–7 days: OR = 4.23, 95% CI = 3.56–5.04; 8–30 days: OR = 2.12, 95% CI = 1.86–2.42; 31–90 days: OR = 1.88, 95% CI = 1.70–2.08; 91–180 days: OR = 1.74, 95% CI = 1.56–1.94; 181–365 days: OR = 1.93, 95% CI = 1.76–2.13). The ORs for the association between antibiotic exposure and risk of pneumococcal disease19 were reported separately for exposures occurring between 0 and 30 days, 31 and 60 days, and 61 and 90 days before the infection and were 1.9 (95% CI = 1.1–3.2), 1.6 (95% CI = 0.89–3.0) and 1.2 (95% CI = 0.60–2.5), respectively. Gradel et al.29 reported a higher antibiotic consumption rate amongst cases of Salmonella infection, compared with controls, for a 1 year period prior to infection. The excess antibiotic consumption amongst cases remained constant during the first 30 weeks of the 1 year exposure period, after which it increased steadily until 2 weeks preceding the infection. In the study by Neal et al.23 the risk of Salmonella infection was greater for antibiotic exposure in the preceding month (1.8, 0.9–3.8) compared with exposure in the past year (1.4, 1.0–2.1). Three studies assessed antibiotic exposure in the 12 months preceding infection, but excluded exposure occurring within the previous 7 days to account for the possibility of reverse causation. These studies demonstrated an association between previous antibiotic use and recurrent furunculosis (OR = 16.6, 95% CI = 2.2–66.0),28 non-typhoid Salmonella (OR = 1.59, 95% CI = 1.43–1.77)29 and meningitis (OR = 2.04, 95% CI = 1.91–2.18).18 Number of antibiotic prescriptions A dose–response relationship was reported between antibiotic exposure and risk of meningitis.18 Patients receiving ≥4 antimicrobials in the preceding 12 months had a higher risk of meningitis (2.85, 2.44–3.34) compared with those receiving one antimicrobial prescription (1.74, 1.62–1.88). The association between invasive pneumococcal disease19 and antibiotic usage in the preceding 3 months was stronger for patients who had ≥2 antibiotic prescriptions (2.1, 1.2–3.8), compared with those who had a single prescription (1.4, 0.88–2.3). Quality assessment Seven studies were deemed to be of high quality, ten studies of moderate quality and one study of low quality (see Tables S2 and S3). There was considerable variation between studies in the selection of confounding variables. The most common confounding variables were age, gender and location. Case–control studies scored poorly on method of ascertainment of exposure, mostly relying on written self-reports from participants. Non-response rates were also poorly reported in case–control studies. Discussion We have found that previous antibiotic use was associated with 12 different community-acquired infections, including infections of viral and bacterial origin. The associated infections ranged from common ailments such as upper respiratory tract infection (URTI) and infectious diarrhoea to relatively rare, but potentially life-threatening, infections such as meningitis. Our findings provide evidence that harms due to antibiotic use could extend beyond the widely recognized threat of promotion of drug-resistant organisms, to include an increased subsequent risk of infections caused by a range of microorganisms and at many different anatomical sites. Prior antibiotic consumption and subsequent infection risk was not the primary research question in most of the included studies. Most studies assessed several potential risk factors for infection, and in these the discovery of an association between antibiotic use and subsequent infection was an unexpected finding. The novelty of this finding was not commented upon by many authors. Despite inclusion as a potential risk factor in several studies, the hypothesis that prior antibiotic therapy could increase future risk of community-acquired infection has received little attention in the published literature. We did not find evidence of an association between prior antibiotic use and subsequent infection in any randomized controlled clinical trials of antibiotics. This was probably for two reasons. Firstly, prior antibiotic use was not an acknowledged confounding factor. Secondly, infections occurring at sites different to the target of the intervention would not have been seen as an adverse effect of the use of antibiotics. Recurrence of the target infection would have been interpreted as intervention failure, not an adverse effect of the intervention. The association between exposure to antibiotics and future infection may be explained by the disruptive effect of antibiotic therapy on the microbiota. Antibiotic-induced alteration of the microbiota could diminish the local suppressive effect of commensal organisms on pathogen overgrowth. There is evidence that the microbiota stimulates host epithelial cells to produce antimicrobial peptides36 and promotes antimicrobial activity of local immune cells.5,37 Furthermore, the gut microbiota may modulate systemic immunity and influence host susceptibility to infections distant from the gut.38 Experimental studies have shown that microbiota-depleted mice have reduced production of inflammatory cytokines, decreased bactericidal activity of macrophages, reduced production, extravasation and bactericidal activity of neutrophils and impaired antibody response to viral infections.38 If alteration of the microbiota was responsible for the increased infection risk following antibiotic therapy it might be expected that the risk of infection would return to baseline as the microbiota recovered to its pretreatment state. Four studies analysed variation in the strength of association between prior antibiotic therapy and subsequent infection in relation to the timing of antibiotic exposure.18,19,23,29 In three out of four studies there was a trend for weakening of the association between prior antibiotic therapy and subsequent infection as the interval between antibiotic use and diagnosis of infection increased.18,19,29 In the case of pneumococcal disease the observed association between antibiotic exposure and risk of infection weakened to the extent that the association failed to reach statistical significance for distant exposures, i.e. those occurring >30 days prior to the infection.19 Together these findings support the hypothesis that transient alteration of the microbiota may be responsible for mediating an increased infection risk following antibiotic exposure. One study on Salmonella infection was an exception since it did not demonstrate a weakening of the association between antibiotic exposure and subsequent infection as the interval between antibiotic exposure and diagnosis of infection increased.23 The risk of meningitis and invasive pneumococcal disease was found to increase with an increasing number of antibiotics to which the patient was exposed prior to the diagnosis of infection.18,19 Repeated exposure to antibiotics within the exposure interval would be expected to delay the recovery of the microbiota to its pretreatment state. This further strengthens the suspicion that antibiotic-induced alteration of the microbiota may be responsible for increased infection risk following antibiotic therapy. On the other hand the observed association between previous antibiotic exposure and subsequent infection could reflect excessive prior antibiotic consumption amongst persons with increased susceptibility to infections that was independent of the effect of antibiotic therapy on the microbiota. This susceptibility to infections could have been inherited or acquired. If this were the case, frequent antibiotic consumption above a specific threshold would be a clinically valuable marker in identifying a subset of patients who warrant further investigation for a previously unrecognized immune vulnerability. A further explanation for the observed association may lie in the variation between patients of the threshold for seeking a medical consultation for symptoms. Patients with frequent healthcare-seeking behaviour may be more likely to be diagnosed with infections, and to be prescribed antibiotics. One study controlled for frequency of medical care in their analysis and found an increased risk of URTI in patients exposed to long-term antibiotics for acne, even after adjusting for consultation frequency.31 Limitations and future work To our knowledge, this is the first systematic review to demonstrate the association between prior antibiotic therapy and subsequent risk of community-acquired infections, other than infections caused by antibiotic-resistant organisms, fungal organisms and C. difficile. The association between antibiotic use and subsequent risk of infection should be interpreted with caution. The observational nature of the included studies means it is not possible to establish causality. Most studies did not exclude antibiotic exposure occurring in the days leading up to the infection, and hence reverse causation should be considered as an alternative explanation for the observed association. Antibiotic exposure could have occurred as a result of medical consultation due to prodromal symptoms of the later infection being studied, rather than as a distant exposure due to an unrelated illness, which then increased susceptibility to the current infection. A further limitation is the possibility of confounding by indication resulting from underlying host susceptibility to infection, as described earlier. This may be addressed by designing studies that take this into account or adjust for infection susceptibility. The use of retrospective self-reports of antibiotic exposure in some studies is likely to have introduced recall bias. There was insufficient data on adherence to the prescribed course of antibiotics, which could have resulted in misclassification of exposure status. Most studies did not include sufficient prescribing details to establish whether there was a dose–response relationship between antibiotic usage and subsequent infection, or whether the relationship was stronger with broad-spectrum antibiotics, a finding that might have provided more support for the hypothesis of collateral microbiota damage. The studies on Salmonella infection provided inconclusive evidence for whether there was an association between infection and prior antibiotic therapy. This warrants further investigation, possibly through a meta-analysis, which could provide a more precise estimate of the overall effect size. A further limitation of our review was the exclusion of non-English-language papers and those that met the inclusion criteria but for which the full-text article was irretrievable. Conclusions Prior antibiotic therapy was associated with a diverse range of community-acquired infections. This included infections caused by antibiotic-resistant and non-resistant organisms. The association between antibiotic exposure and subsequent infection became weaker with increasing time from antibiotic exposure. The risk of infection increased with increasing number of antibiotics to which the patients were exposed. Whilst antibiotic therapy is often necessary for the treatment of bacterial infections, our findings highlight the continued need to limit inappropriate antibiotic prescriptions in primary care, both to reduce the consequences of bacterial resistance and possibly also to reduce the risk of future infections. The observed association may help clinicians in dissuading their patients from insisting on an antibiotic prescription when this is deemed not to be clinically indicated. New research may help discover whether other infection types, not examined by the studies included in our review, are also associated with prior antibiotic therapy. Further studies are also required to examine the mechanisms underlying the observed association, particularly whether this association could be mediated through antibiotic-induced collateral damage to the microbiota. Acknowledgements Conferences where work has previously been presented: (i) The Society for Academic Primary Care, London Annual Scientific Meeting, Cambridge, UK, 2016 (Abstract Number 81); (ii) Society of Academic Primary Care (SAPC), 45th Annual Scientific Meeting, Dublin, Republic of Ireland, 2016 (Talk Code EP3A.06); and (iii) 19th International Conference on Human Microbiome, Singapore, 2017 (Paper Code 17SG030091). Funding This study was conducted as part of our routine work. Transparency declarations None to declare. Author contributions U. M., D. A., M. A., L. M. and P. W. conceived the study. U. M. carried out the literature search and extracted the data. U. M. and P. W. independently selected eligible articles from potentially suitable articles identified. P. W. cross-checked extracted data. U. M. wrote the final manuscript. D. A., M. A., A. D., L. M., V. L., M. M., U. M. and P. W. contributed to the interpretation and revised the final manuscript. P. W. is guarantor for the study. Supplementary data Tables S1–S3 are available as Supplementary data at JAC Online. References 1 Llor C, Bjerrum L. Antimicrobial resistance: risk associated with antibiotic overuse and initiatives to reduce the problem. Ther Adv Drug Saf  2014; 5: 229– 41. Google Scholar CrossRef Search ADS PubMed  2 WHO. The Evolving Threat of Antimicrobial Resistance—Options for Action. 2012. http://www.who.int/patientsafety/implementation/amr/publication/en/. 3 Gulliford MC, Dregan A, Moore MV et al.   Continued high rates of antibiotic prescribing to adults with respiratory tract infection: survey of 568 UK general practices. BMJ Open  2014; 4: e006245. Google Scholar CrossRef Search ADS PubMed  4 Ursell LK, Metcalf JL, Parfrey LW et al.   Defining the human microbiome. Nutr Rev  2012; 70 Suppl 1: S38– 44. Google Scholar CrossRef Search ADS PubMed  5 Ubeda C, Pamer EG. Antibiotics, microbiota, and immune defense. Trends Immunol  2012; 33: 459– 66. Google Scholar CrossRef Search ADS PubMed  6 Levy J. The effects of antibiotic use on gastrointestinal function. Am J Gastroenterol  2000; 95 Suppl: S8– 10. Google Scholar CrossRef Search ADS PubMed  7 Jakobsson HE, Jernberg C, Andersson AF et al.   Short-term antibiotic treatment has differing long-term impacts on the human throat and gut microbiome. PLoS One  2010; 5: e9836. Google Scholar CrossRef Search ADS PubMed  8 Chang JY, Antonopoulos DA, Kalra A et al.   Decreased diversity of the fecal microbiome in recurrent Clostridium difficile-associated diarrhea. J Infect Dis  2008; 197: 435– 8. Google Scholar CrossRef Search ADS PubMed  9 Brown KA, Khanafer N, Daneman N et al.   Meta-analysis of antibiotics and the risk of community-associated Clostridium difficile infection. Antimicrob Agents Chemother  2013; 57: 2326– 32. Google Scholar CrossRef Search ADS PubMed  10 Xu J, Schwartz K, Bartoces M et al.   Effect of antibiotics on vulvovaginal candidiasis: a MetroNet study. J Am Board Fam Med  2008; 21: 261– 8. Google Scholar CrossRef Search ADS PubMed  11 Spinillo A, Capuzzo E, Acciano S et al.   Effect of antibiotic use on the prevalence of symptomatic vulvovaginal candidiasis. Am J Obstet Gynecol  1999; 180: 14– 7. Google Scholar CrossRef Search ADS PubMed  12 MacDonald TM, Beardon PH, McGilchrist MM et al.   The risks of symptomatic vaginal candidiasis after oral antibiotic therapy. Q J Med  1993; 86: 419– 24. Google Scholar PubMed  13 Moher D, Liberati A, Tetzlaff J et al.   Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ  2009; 339: b2535. Google Scholar CrossRef Search ADS PubMed  14 Malik U, Armstrong D, Ashworth M et al.   Association between antibiotic use and subsequent risk of community-acquired infections: a systematic review. PROSPERO: International prospective register of systematic reviews. 2016. CRD42016048521. http://www.crd.york.ac.uk/prospero/display_record.asp?ID=CRD42016048521. 15 Costelloe C, Metcalfe C, Lovering A et al.   Effect of antibiotic prescribing in primary care on antimicrobial resistance in individual patients: systematic review and meta-analysis. BMJ  2010; 340: c2096. Google Scholar CrossRef Search ADS PubMed  16 Covidence—Accelerate Your Systematic Review. 2016. https://www.covidence.org/. 17 Wells GA, Shea B, O'Connell D et al.   The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. 2014. http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp. 18 Armstrong D, Ashworth M, Dregan A et al.   The relationship between prior antimicrobial prescription and meningitis: a case-control study. Br J Gen Pract  2016; 66: e228– 33. Google Scholar CrossRef Search ADS PubMed  19 Chun CS, Weinmann S, Riedlinger K et al.   Passive cigarette smoke exposure and other risk factors for invasive pneumococcal disease in children: a case-control study. Perm J  2015; 19: 38– 43. Google Scholar CrossRef Search ADS PubMed  20 Doorduyn Y, Van Den Brandhof WE, Van Duynhoven YT et al.   Risk factors for Salmonella Enteritidis and Typhimurium (DT104 and non-DT104) infections in The Netherlands: predominant roles for raw eggs in Enteritidis and sandboxes in Typhimurium infections. Epidemiol Infect  2006; 134: 617– 26. Google Scholar CrossRef Search ADS PubMed  21 Delarocque-Astagneau E, Bouillant C, Vaillant V et al.   Risk factors for the occurrence of sporadic Salmonella enterica serotype typhimurium infections in children in France: a national case-control study. Clin Infect Dis  2000; 31: 488– 92. Google Scholar CrossRef Search ADS PubMed  22 Kass PH, Farver TB, Beaumont JJ et al.   Disease determinants of sporadic salmonellosis in four northern California counties. A case-control study of older children and adults. Ann Epidemiol  1992; 2: 683– 96. Google Scholar CrossRef Search ADS PubMed  23 Neal KR, Briji SO, Slack RC et al.   Recent treatment with H2 antagonists and antibiotics and gastric surgery as risk factors for Salmonella infection. BMJ  1994; 308: 176. Google Scholar CrossRef Search ADS PubMed  24 Pavia AT, Shipman LD, Wells JG et al.   Epidemiologic evidence that prior antimicrobial exposure decreases resistance to infection by antimicrobial-sensitive Salmonella. J Infect Dis  1990; 161: 255– 60. Google Scholar CrossRef Search ADS PubMed  25 Banatvala N, Cramp A, Jones IR et al.   Salmonellosis in North Thames (East), UK: associated risk factors. Epidemiol Infect  1999; 122: 201– 7. Google Scholar CrossRef Search ADS PubMed  26 Effler P, Ieong MC, Kimura A et al.   Sporadic Campylobacter jejuni infections in Hawaii: associations with prior antibiotic use and commercially prepared chicken. J Infect Dis  2001; 183: 1152– 5. Google Scholar CrossRef Search ADS PubMed  27 Early GJ, Seifried SE. Risk factors for community-associated Staphylococcus aureus skin infection in children of Maui. Hawaii J Med Public Health  2012; 71: 218– 23. Google Scholar PubMed  28 El-Gilany AH, Fathy H. Risk factors of recurrent furunculosis. Dermatol Online J  2009; 15: 16. Google Scholar PubMed  29 Gradel KO, Dethlefsen C, Ejlertsen T et al.   Increased prescription rate of antibiotics prior to non-typhoid Salmonella infections: a one-year nested case-control study. Scand J Infect Dis  2008; 40: 635– 41. Google Scholar CrossRef Search ADS PubMed  30 Luby SP, Faizan MK, Fisher-Hoch SP et al.   Risk factors for typhoid fever in an endemic setting, Karachi, Pakistan. Epidemiol Infect  1998; 120: 129– 38. Google Scholar CrossRef Search ADS PubMed  31 Margolis DJ, Bowe WP, Hoffstad O et al.   Antibiotic treatment of acne may be associated with upper respiratory tract infections. Arch Dermatol  2005; 141: 1132– 6. Google Scholar CrossRef Search ADS PubMed  32 McVernon J, Andrews N, Slack M et al.   Host and environmental factors associated with Hib in England, 1998-2002. Arch Dis Child  2008; 93: 670– 5. Google Scholar CrossRef Search ADS PubMed  33 Srikantiah P, Vafokulov S, Luby SP et al.   Epidemiology and risk factors for endemic typhoid fever in Uzbekistan. Trop Med Int Health  2007; 12: 838– 47. Google Scholar CrossRef Search ADS PubMed  34 Shallcross LJ, Hayward AC, Johnson AM et al.   Incidence and recurrence of boils and abscesses within the first year: a cohort study in UK primary care. Br J Gen Pract  2015; 65: e668– 76. Google Scholar CrossRef Search ADS PubMed  35 Mediano P, Fernandez L, Rodriguez JM et al.   Case-control study of risk factors for infectious mastitis in Spanish breastfeeding women. BMC Pregnancy Childbirth  2014; 14: 195. Google Scholar CrossRef Search ADS PubMed  36 Gallo RL, Hooper LV. Epithelial antimicrobial defence of the skin and intestine. Nat Rev Immunol  2012; 12: 503– 16. Google Scholar CrossRef Search ADS PubMed  37 Ivanov II, Atarashi K, Manel N et al.   Induction of intestinal Th17 cells by segmented filamentous bacteria. Cell  2009; 139: 485– 98. Google Scholar CrossRef Search ADS PubMed  38 Brown RL, Clarke TB. The regulation of host defences to infection by the microbiota. Immunology  2017; 150: 1– 6. Google Scholar CrossRef Search ADS PubMed  © The Author 2017. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please email: journals.permissions@oup.com. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Antimicrobial Chemotherapy Oxford University Press

Association between prior antibiotic therapy and subsequent risk of community-acquired infections: a systematic review

Loading next page...
 
/lp/ou_press/association-between-prior-antibiotic-therapy-and-subsequent-risk-of-obBQO1Ohs3
Publisher
Oxford University Press
Copyright
© The Author 2017. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please email: journals.permissions@oup.com.
ISSN
0305-7453
eISSN
1460-2091
D.O.I.
10.1093/jac/dkx374
Publisher site
See Article on Publisher Site

Abstract

Abstract Background Antibiotic use can have negative unintended consequences including disruption of the human microbiota, which is thought to protect against pathogen overgrowth. We conducted a systematic review to assess whether there is an association between exposure to antibiotics and subsequent risk of community-acquired infections. Methods We searched MEDLINE, EMBASE and Web of Science for studies published before 30 June 2017, examining the association between antibiotic use and subsequent community-acquired infection. Infections caused by Clostridium difficile and fungal organisms were excluded. Studies focusing exclusively on resistant organism infections were also excluded. Results Eighteen of 22588 retrieved studies met the inclusion criteria. From these, 16 studies reported a statistically significant association between antibiotic exposure and subsequent risk of community-acquired infection. Infections associated with prior antibiotic use included Campylobacter jejuni infection (one study), recurrent furunculosis (one study), invasive Haemophilus influenzae type b infection (one study), infectious mastitis (one study), meningitis (one study), invasive pneumococcal disease (one study), Staphylococcus aureus skin infection (one study), typhoid fever (two studies), recurrent boils and abscesses (one study), upper respiratory tract infection and urinary tract infection (one study) and Salmonella infection (five studies), although in three studies on Salmonella infection the effect was of marginal statistical significance. Conclusions We found an association between prior antibiotic use and subsequent risk of a diverse range of community-acquired infections. Gastrointestinal and skin and soft tissue infections were most frequently found to be associated with prior antibiotic exposure. Our findings support the hypothesis that antibiotic use may predispose to future infection risk, including infections caused by both antibiotic-resistant and non-resistant organisms. Introduction Antibiotics are an important weapon in our fight against bacterial infections. Antibiotic use is associated with recognized harms, including the emergence of antibiotic-resistant organisms.1 There is concern that straightforward infections that are currently easy to control may become untreatable in the future. The growing threat of antimicrobial resistance has been declared a global public health crisis.2 Much attention has been focused on limiting inappropriate antibiotic usage as a strategy to control drug resistance, particularly in primary care, where 90% of antibiotic prescriptions are issued.1 Even so, in the UK, most general practices continue to prescribe antibiotics at rates in excess of what is clinically justified.3 Antibiotic therapy can have further unintended consequences, other than selection of resistant microorganisms. The human microbiota is a complex community of up to a hundred trillion microorganisms lining the epithelial surfaces of their hosts and that are exposed to the outside world.4 The microbiota influences human health through its role in a diverse range of physiological functions,4 including a protective role in defence against pathogens. The microbiota is thought to contribute to development of the host’s immune system and its response to infection.5 The huge range of organisms in the microbiota, commensals and potential pathogens, compete with each other for attachment sites and nutrients, thereby preventing pathogen overgrowth.6 Exposure to antibiotics can change the composition of the microbiota, reducing microbial diversity and allowing the overgrowth of potentially harmful microorganisms.6 Short-term antibiotic treatment for Helicobacter pylori eradication, for example, reduces microbial diversity in the human throat and gut microbiota, which in some cases persists for up to 4 years.7 The question arises whether antibiotics prescribed to treat an acute bacterial infection could paradoxically predispose individuals to future infections due to collateral damage inflicted on the microbiota. Clostridium difficile-associated diarrhoea is a well-recognized example of an infection resulting from pathogenic colonization of a microbial community disrupted by recent antibiotic use.8,9 Antibiotic treatment is also associated with increased risk of fungal infection.10–12 Are there other medium- and long-term infection risks associated with antibiotic use unrelated to increased resistance, or infection with C. difficile or opportunist fungi? This systematic review sought evidence of the association between antibiotic use and subsequent risk of infection in the community setting, unrelated to these previously known infection risks. Objectives To examine whether exposure to antibiotic therapy is associated with subsequent increased risk of community-acquired infections. Methods Procedures used in this review were consistent with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.13 Protocol and registration A review protocol was submitted in advance to PROSPERO, which is a database of systematic review protocols (registration ID: CRD42016048521).14 Eligibility criteria Our inclusion criteria were: (i) studies, of any design, evaluating the risk of community-acquired infection associated with prior antibiotic usage; (ii) human participants, of any age, in the community setting; (iii) studies that assessed risk of infection associated with prior antibiotic exposure compared with the risk of infection without prior antibiotic exposure, in the same period; and (iv) an endpoint of any community-acquired infection. Our exclusion criteria were: (i) studies focused on immune-compromised patient groups (e.g. patients with human immunodeficiency virus) as these would not be representative of the general population; (ii) studies focusing on infections that are known to be associated with prior antibiotic exposure including fungal infection10–12 and C. difficile infection;9 (iii) studies focusing exclusively on infections caused by antibiotic-resistant organisms as these have been the subject of a previous systematic review15 (furthermore, these studies would not provide a true estimate of the overall risk of infection following antibiotic exposure because infections resulting from non-resistant organisms would be excluded); (iv) studies assessing the risk of nosocomial infections because their relationship with prior antibiotic use could be confounded by other factors (e.g. vulnerable patients, infectious contacts, invasive procedures); (v) non-English language articles; (vi) reviews, letters, editorials and case reports; (vii) studies assessing the impact of prior antibiotic use on complications resulting from a specific infection (e.g. mortality rate); and (viii) studies examining the use of antibiotics as a prophylactic measure to prevent a specific infection (e.g. post-operative wound infection) because such studies are not designed to assess the impact of antibiotic exposure on the subsequent risk of acquiring other unrelated infections. Search strategy The initial literature search was performed on 29 April 2016. The search was repeated on 30 June 2017 to identify further articles published since the date of the initial search. The databases searched were MEDLINE, EMBASE and Web of Science. No restrictions were imposed on the publication period. Search terms included both text words and MESH terms. Detailed search strategies are presented in Table S1 (available as Supplementary data at JAC Online). The reference lists of eligible studies and review articles were screened to identify further studies eligible for inclusion. Study selection Covidence software was used to facilitate the screening and selection of studies.16 The first reviewer (U. M.) conducted the literature search, removed duplicate articles, and screened titles and abstracts with respect to the eligibility criteria. Full-text articles of potentially relevant studies were independently assessed for eligibility by two reviewers (U. M. and P. W.). Any disagreements were resolved through discussion and consensus. Data extraction One reviewer (U. M.) extracted data from the full texts of included studies. Extracted data were summarized in a predesigned format and cross-checked by a second reviewer (P. W.). Disagreements were resolved through discussion and consensus. Data collected included the first author, publication year, country of study, study design, study objectives, participant characteristics, exposure definition (i.e. antibiotic class, length of treatment, time from exposure to infection), exposure interval (length of time prior to the infection during which antibiotic exposure was assessed), exposure ascertainment method, infection type, definition of cases and the comparison group, sample size, main outcome measure and confounding variables. The primary outcome of interest was the OR for the association of infection with prior antibiotic exposure. Quality assessment The Newcastle Ottawa Scale (NOS), a tool for assessing the quality of non-randomized studies,17 was applied independently to each study by two reviewers (U. M. and P. W.). Any disagreements were resolved through discussion and consensus. Each study could be assigned a maximum of nine points; four for the selection domain, two for the comparability domain and three for the exposure domain. Studies assigned NOS scores of ≥7 were considered high quality, 5–6 were considered moderate quality and <5 were considered low quality. Data synthesis and analysis Our research question is broad, aiming to examine the association between past antibiotic consumption and a diverse range of infections, at different body sites, caused by a variety of different microorganisms. Quantitative pooling of results through a meta-analysis was not justified owing to the high degree of clinical heterogeneity expected between studies in terms of the infection types being investigated and differences in antibiotic exposures. Any observed association between antibiotic use and subsequent infection was considered unlikely to be universal for all infection types. We have therefore reported the findings of the systematic review as a narrative (descriptive) synthesis. Results Study selection The literature search retrieved 22582 publications. Six additional publications were identified from the reference list of retrieved articles. After adjusting for duplicates, 21439 publications remained. After screening by title and abstract, 21053 publications were excluded. A full-text review was conducted of 386 publications, from which 18 met the eligibility criteria and were included in the review.18–35 The detailed rationale behind the exclusion of studies is presented in Figure 1. Figure 1 View largeDownload slide PRISMA flow chart. Figure 1 View largeDownload slide PRISMA flow chart. Study design and participant characteristics Tables 1 and 2 summarize the main characteristics of the included studies. Sixteen studies were case–control studies and two were cohort studies. Table 1 Data extraction table for case–control studies (in alphabetical order of infection studied) First author  Publication year; country  Infection studied  Antibiotic; exposure interval  Case definition  Comparison group  Sample size  Fully adjusted outcome (95% CI)  Confounder variables adjusted for  Effler26  2001; USA  C. jejuni  any; 28 days  lab confirmed; age: not available  age and telephone exchange matched  211 cases; 211 controls  OR 3.3 (1.1–9.6)  consumption of various food items, acid-suppressing drugs and contact with live chickens  El-Gilany28  2009; Egypt  furunculosis (recurrent)  any; 1 year  clinic attendance with ≥3 attacks of boils within the previous 12 months; all ages  attendance of the same clinic diagnosed with furunculosis for the first time  74 cases; 74 controls  OR 16.6 (2.2–66.0)  family history, diabetes mellitus, anaemia, previous hospitalization, personal hygiene, skin diseases and number of lesions  McVernon32  2008; UK  H. influenzae type b (invasive)  frequency of previous antibiotic use  lab confirmed cases; age 5 years to 9 years and 11 months  matched by date of birth and region  136 cases; 295 controls  frequent use: OR 1.51 (1.06–2.13)  sex, prematurity, breastfed, past illness, family demographic factors, bedroom sharing, smoking, central heating, home ownership and vaccination status  Mediano35  2014; Spain  mastitis (infectious)  any; during pregnancy  lab confirmed; lactating females  healthy breastfeeding women with no clinical symptoms of mastitis and negative milk culture  368 cases; 148 controls  OR 5.38 (2.85–10.14)  age, personal and family history, infection, comorbidities, drugs and pregnancy, childbirth and breastfeeding-related factors  Armstrong18  2016; UK  meningitis  any; 1 year  identified from GP records; all ages  matched on age, sex, GP practice and index date  7346 cases; 29384 controls  OR 2.04 (1.91–2.18)  13 variables including demographic factors, lifestyle, comorbidities and medications  Chun19  2015; USA  pneumococcal disease (invasive)  any; 3 months  lab confirmed; age 0–12 years  matched by age, health plan membership and length of membership  171 cases; 342 controls  OR 1.57 (1.06–2.33)  sex, race, age, risk status, health plan membership and pneumococcal vaccination  Doorduyn20  2016; the Netherlands  Salmonella  any; 4 weeks  lab confirmed; all ages  matched for age, sex, degree of urbanization and season  193 cases; 3119 controls  OR 1.9 (1.0–3.4)  age, sex, degree of urbanization and education  Gradel29  2008; Denmark  Salmonella  any; 1 year  lab confirmed; age 1–99 years  residents of same county matched for specimen date, gender and age  1882 cases; 18820 controls  OR 1.59 (1.43–1.77)  gender, antibiotic score,a patient group, age, NTS infection month and serovar  Delarocque- Astagneau21  2000; France  Salmonella  any; 1 month  lab confirmed; age <14 years  matched for age and place of residence  101 cases; 101 controls  OR 2.3 (1.0–5.5)  consumption of various food items  Banatvala25  1999; UK  Salmonella  any; 1 month  lab confirmed; all ages  (1) case nominated and matched for age, gender and area of residence; (2) randomly selected from London area  209 cases and matched controls; 854 random controls  matched: OR 1.3 (0.6–2.8); unmatched: OR 1.3 (0.6–2.6)  not available  Neal23  1994; UK  Salmonella  any; 1 year  lab confirmed and notified; age ≥45 years  next two patients in the practice records system matched for age and sex  188 cases; 376 controls  past year: OR 1.4 (1.0–2.1)  gastric surgery, H2 antagonist treatment and other drug use  Kass22  1992; USA  Salmonella  any; ‘recent’  lab confirmed and notified; age ≥10 years  matched for region  120 cases; 265 controls  OR 1.96 (0.86–4.37)  prior health problems and prior medical therapies  Pavia24  1989; USA  Salmonella  any; 30 days  lab confirmed; age >1 year  matched for age, neighbourhood and telephone exchange  35 cases; 70 controls  OR 3.8 (1.2–11.9)  immunosuppression and use of antacids or H2-blocking agents  Early27  2012; Hawaii  S. aureus skin infection  any; 6 months  lab confirmed; age 6 months to 17 years  matched for age, clinician and date of clinician visit  71 cases; 146 controls  OR 2.90b (1.29–6.61)  household contact, abrasions/wounds, skin disorders, weight and sharing bed linens and towels  Srikantiah33  2007; Uzbekistan  typhoid fever  any; 2 weeks  lab confirmed; all ages  age- and community-matched controls  97 cases; 192 controls  OR 12.2 (4.0–37.0)  occupation and consumption of various food and drink items  Luby30  1998; Pakistan  typhoid fever  any; 2 weeks  lab confirmed; all ages  neighbourhood and age matched  100 cases; 200 controls  OR 5.7 (2.3–13.9)  consumption of various food and drink items  First author  Publication year; country  Infection studied  Antibiotic; exposure interval  Case definition  Comparison group  Sample size  Fully adjusted outcome (95% CI)  Confounder variables adjusted for  Effler26  2001; USA  C. jejuni  any; 28 days  lab confirmed; age: not available  age and telephone exchange matched  211 cases; 211 controls  OR 3.3 (1.1–9.6)  consumption of various food items, acid-suppressing drugs and contact with live chickens  El-Gilany28  2009; Egypt  furunculosis (recurrent)  any; 1 year  clinic attendance with ≥3 attacks of boils within the previous 12 months; all ages  attendance of the same clinic diagnosed with furunculosis for the first time  74 cases; 74 controls  OR 16.6 (2.2–66.0)  family history, diabetes mellitus, anaemia, previous hospitalization, personal hygiene, skin diseases and number of lesions  McVernon32  2008; UK  H. influenzae type b (invasive)  frequency of previous antibiotic use  lab confirmed cases; age 5 years to 9 years and 11 months  matched by date of birth and region  136 cases; 295 controls  frequent use: OR 1.51 (1.06–2.13)  sex, prematurity, breastfed, past illness, family demographic factors, bedroom sharing, smoking, central heating, home ownership and vaccination status  Mediano35  2014; Spain  mastitis (infectious)  any; during pregnancy  lab confirmed; lactating females  healthy breastfeeding women with no clinical symptoms of mastitis and negative milk culture  368 cases; 148 controls  OR 5.38 (2.85–10.14)  age, personal and family history, infection, comorbidities, drugs and pregnancy, childbirth and breastfeeding-related factors  Armstrong18  2016; UK  meningitis  any; 1 year  identified from GP records; all ages  matched on age, sex, GP practice and index date  7346 cases; 29384 controls  OR 2.04 (1.91–2.18)  13 variables including demographic factors, lifestyle, comorbidities and medications  Chun19  2015; USA  pneumococcal disease (invasive)  any; 3 months  lab confirmed; age 0–12 years  matched by age, health plan membership and length of membership  171 cases; 342 controls  OR 1.57 (1.06–2.33)  sex, race, age, risk status, health plan membership and pneumococcal vaccination  Doorduyn20  2016; the Netherlands  Salmonella  any; 4 weeks  lab confirmed; all ages  matched for age, sex, degree of urbanization and season  193 cases; 3119 controls  OR 1.9 (1.0–3.4)  age, sex, degree of urbanization and education  Gradel29  2008; Denmark  Salmonella  any; 1 year  lab confirmed; age 1–99 years  residents of same county matched for specimen date, gender and age  1882 cases; 18820 controls  OR 1.59 (1.43–1.77)  gender, antibiotic score,a patient group, age, NTS infection month and serovar  Delarocque- Astagneau21  2000; France  Salmonella  any; 1 month  lab confirmed; age <14 years  matched for age and place of residence  101 cases; 101 controls  OR 2.3 (1.0–5.5)  consumption of various food items  Banatvala25  1999; UK  Salmonella  any; 1 month  lab confirmed; all ages  (1) case nominated and matched for age, gender and area of residence; (2) randomly selected from London area  209 cases and matched controls; 854 random controls  matched: OR 1.3 (0.6–2.8); unmatched: OR 1.3 (0.6–2.6)  not available  Neal23  1994; UK  Salmonella  any; 1 year  lab confirmed and notified; age ≥45 years  next two patients in the practice records system matched for age and sex  188 cases; 376 controls  past year: OR 1.4 (1.0–2.1)  gastric surgery, H2 antagonist treatment and other drug use  Kass22  1992; USA  Salmonella  any; ‘recent’  lab confirmed and notified; age ≥10 years  matched for region  120 cases; 265 controls  OR 1.96 (0.86–4.37)  prior health problems and prior medical therapies  Pavia24  1989; USA  Salmonella  any; 30 days  lab confirmed; age >1 year  matched for age, neighbourhood and telephone exchange  35 cases; 70 controls  OR 3.8 (1.2–11.9)  immunosuppression and use of antacids or H2-blocking agents  Early27  2012; Hawaii  S. aureus skin infection  any; 6 months  lab confirmed; age 6 months to 17 years  matched for age, clinician and date of clinician visit  71 cases; 146 controls  OR 2.90b (1.29–6.61)  household contact, abrasions/wounds, skin disorders, weight and sharing bed linens and towels  Srikantiah33  2007; Uzbekistan  typhoid fever  any; 2 weeks  lab confirmed; all ages  age- and community-matched controls  97 cases; 192 controls  OR 12.2 (4.0–37.0)  occupation and consumption of various food and drink items  Luby30  1998; Pakistan  typhoid fever  any; 2 weeks  lab confirmed; all ages  neighbourhood and age matched  100 cases; 200 controls  OR 5.7 (2.3–13.9)  consumption of various food and drink items  NTS infection, non-typhoid Salmonella infection. a Antibiotic score (0–5); antibiotics were classified according to their potential impact on the intestinal flora, with higher scores being associated with increasing impact on the intestinal flora. b Association reached significance for the Native Hawaiian and Pacific Islander (NHPI) ethnic category, but not for the non-NHPI ethnic category (OR = 1.93, 95% CI = 0.93–4.01). Table 2. Data extraction table for cohort studies First author  Publication year; country  Data source and study population  Infection studied  Exposure definition  Case definition and follow-up  Comparison  Sample size  Fully adjusted outcome (95% CI)  Confounder variables adjusted for  Shallcross34  2015; UK  registered with THIN- participating GP practice, any age and sought care for a boil, abscess, carbuncle or furuncle  recurrent boil or abscess  antibiotic prescription in the 6 months prior to the date of index consultation  second consultation for a boil or abscess within 3 weeks–12 months  patients without a repeat consultation for boil or abscess  cohort of 164461 patients; 10% developed a repeat boil or abscess  RR 1.4 (1.3–1.4)  age, sex, BMI, diabetes, skin disease, prior antibiotic and smoking status  Margolis31  2005; UK  registered with GPRD- participating GP practice, aged 15–35 years and recorded diagnosis of acne vulgaris  (1) upper respiratory tract infection (URTI); (2) urinary tract infection (UTI).  prescription >6 weeks of oral erythromycin or an oral tetracycline or topical erythromycin or clindamycin or a combination of both  UTI or URTI within 12 months after enrolment  patients without acne antibiotic use were considered unexposed  84997 exposed; 33519 unexposed  URTI: OR 2.23 (2.12–2.34); UTI: OR 1.10a (1.01–1.19)  age, year of diagnosis, sex, contraceptive use or counselling (only for UTIs), practice, diabetes, asthma, visit frequency and the number of prescriptions for acne antibiotics  First author  Publication year; country  Data source and study population  Infection studied  Exposure definition  Case definition and follow-up  Comparison  Sample size  Fully adjusted outcome (95% CI)  Confounder variables adjusted for  Shallcross34  2015; UK  registered with THIN- participating GP practice, any age and sought care for a boil, abscess, carbuncle or furuncle  recurrent boil or abscess  antibiotic prescription in the 6 months prior to the date of index consultation  second consultation for a boil or abscess within 3 weeks–12 months  patients without a repeat consultation for boil or abscess  cohort of 164461 patients; 10% developed a repeat boil or abscess  RR 1.4 (1.3–1.4)  age, sex, BMI, diabetes, skin disease, prior antibiotic and smoking status  Margolis31  2005; UK  registered with GPRD- participating GP practice, aged 15–35 years and recorded diagnosis of acne vulgaris  (1) upper respiratory tract infection (URTI); (2) urinary tract infection (UTI).  prescription >6 weeks of oral erythromycin or an oral tetracycline or topical erythromycin or clindamycin or a combination of both  UTI or URTI within 12 months after enrolment  patients without acne antibiotic use were considered unexposed  84997 exposed; 33519 unexposed  URTI: OR 2.23 (2.12–2.34); UTI: OR 1.10a (1.01–1.19)  age, year of diagnosis, sex, contraceptive use or counselling (only for UTIs), practice, diabetes, asthma, visit frequency and the number of prescriptions for acne antibiotics  THIN, The Health Improvement Network; GPRD, General Practice Research Datalink, now known as Clinical Practice Research Datalink (CPRD); RR, relative risk. a The OR for UTI is statistically significant; however, the authors of the study concluded that it was not clinically meaningful. The studies involved a total of 349085 participants. Study sample sizes ranged from 10524 to 16446134 participants. In 13 studies the hypothesis that antibiotic use would increase subsequent risk of infections was not stated19–22,24,26–28,30,32–35 and earlier antibiotic use was one of many other variables assessed as potential risk factors for infection. Exposure assessment and definition Six studies ascertained exposure status from recorded prescription data.18,19,23,29,31,34 Ten studies determined exposure status through participant interview or questionnaire.20–22,26–28,30,32,33,35 One study interviewed both participants and their physician,24 and in one study medication data was obtained from the physician only when the participant was unsure.25 Seventeen studies defined exposure as the consumption of any antibiotic medication. One study defined exposure in terms of specific antibiotic classes.31 In one study the exposure status was defined as an antibiotic prescription of 6 weeks or more for acne.31 In the remaining 17 studies, exposure was not defined in terms of the duration of antibiotic course. The length of time during which antibiotic exposure was assessed prior to the infection (exposure interval) was specified in 14 out of 18 studies.18–21,23–30,33,34 In these studies, the exposure interval ranged from 2 weeks to 1 year. The median exposure interval was 2 months. Case definition Fourteen studies defined infection on the basis of a positive microbiological specimen.19–27,29,30,32,33,35 Three studies defined infection on the basis of routinely recorded primary care data.18,31,34 One study on recurrent furunculosis did not specify how the cases were defined.28 Comparison group In case–control studies the control group consisted of participants who did not have a confirmed history of the infection of interest. In the cohort study on patients with acne, risk of infection was compared between those who were prescribed long-term antibiotics for acne, versus those who were not.31 In the cohort study on risk of recurrent boil and abscess, antibiotic prescription rates were compared amongst participants who had a repeat consultation for a boil or abscess within 12 months, versus those who did not.34 Associations between exposure and outcome Positive and significant associations between prior antibiotic use and subsequent risk of infection were reported in 16 out of 18 studies, including one study on Campylobacter jejuni infection,26 one study on recurrent furunculosis,28 one study on invasive Haemophilus influenzae type b infection,32 one study on infectious mastitis,35 one study on meningitis,18 one study on invasive pneumococcal disease,19 one study on Staphylococcus aureus skin infection,27 two studies on typhoid fever,30,33 one study on recurrent boils and abscesses,34 one study on upper respiratory tract infection and urinary tract infection31 and five studies on Salmonella infection.20,21,23,24,29 The outcome measure and 95% CI from each study is presented in Tables 1 and 2. Salmonella infection Nine out of 18 studies assessed the association between antibiotic use and subsequent risk of Salmonella infection.20–25,29,30,33 As this was the most frequently studied infection the results are described here in further detail. Two out of nine studies focused exclusively on cases of typhoid fever.30,33 Both studies reported a positive and statistically significant association between prior antibiotic use and subsequent risk of typhoid fever. The remaining seven studies assessing the risk of Salmonella infection did not restrict the definition of cases to typhoid fever.20–25,29 Five out of seven studies reported a positive and significant association between prior antibiotic use and risk of Salmonella infection.20,21,23,24,29 From these five studies, three studies reported 95% CIs for the OR that included 1.0 at the lower end.20,21,23 In these studies the effect was of marginal statistical significance. We found two negative studies, which did not demonstrate a statistically significant association between prior antibiotic use and subsequent risk of Salmonella infection.22,25 Timing of antibiotic exposure Four studies assessed the effect of timing of antibiotic exposure on subsequent risk of infection.18,19,23,29 Increased risk of community-acquired infection after antibiotic exposure was documented up to 1 year in three18,23,29 out of four studies except for one study on pneumococcal disease19 that showed an increased risk only in the following month. In the case of meningitis,18 the association between antibiotic exposure and subsequent infection risk remained statistically significant over the 12 month exposure period prior to the infection, although as the interval between exposure and subsequent infection increased, the effect size generally diminished (0–7 days: OR = 4.23, 95% CI = 3.56–5.04; 8–30 days: OR = 2.12, 95% CI = 1.86–2.42; 31–90 days: OR = 1.88, 95% CI = 1.70–2.08; 91–180 days: OR = 1.74, 95% CI = 1.56–1.94; 181–365 days: OR = 1.93, 95% CI = 1.76–2.13). The ORs for the association between antibiotic exposure and risk of pneumococcal disease19 were reported separately for exposures occurring between 0 and 30 days, 31 and 60 days, and 61 and 90 days before the infection and were 1.9 (95% CI = 1.1–3.2), 1.6 (95% CI = 0.89–3.0) and 1.2 (95% CI = 0.60–2.5), respectively. Gradel et al.29 reported a higher antibiotic consumption rate amongst cases of Salmonella infection, compared with controls, for a 1 year period prior to infection. The excess antibiotic consumption amongst cases remained constant during the first 30 weeks of the 1 year exposure period, after which it increased steadily until 2 weeks preceding the infection. In the study by Neal et al.23 the risk of Salmonella infection was greater for antibiotic exposure in the preceding month (1.8, 0.9–3.8) compared with exposure in the past year (1.4, 1.0–2.1). Three studies assessed antibiotic exposure in the 12 months preceding infection, but excluded exposure occurring within the previous 7 days to account for the possibility of reverse causation. These studies demonstrated an association between previous antibiotic use and recurrent furunculosis (OR = 16.6, 95% CI = 2.2–66.0),28 non-typhoid Salmonella (OR = 1.59, 95% CI = 1.43–1.77)29 and meningitis (OR = 2.04, 95% CI = 1.91–2.18).18 Number of antibiotic prescriptions A dose–response relationship was reported between antibiotic exposure and risk of meningitis.18 Patients receiving ≥4 antimicrobials in the preceding 12 months had a higher risk of meningitis (2.85, 2.44–3.34) compared with those receiving one antimicrobial prescription (1.74, 1.62–1.88). The association between invasive pneumococcal disease19 and antibiotic usage in the preceding 3 months was stronger for patients who had ≥2 antibiotic prescriptions (2.1, 1.2–3.8), compared with those who had a single prescription (1.4, 0.88–2.3). Quality assessment Seven studies were deemed to be of high quality, ten studies of moderate quality and one study of low quality (see Tables S2 and S3). There was considerable variation between studies in the selection of confounding variables. The most common confounding variables were age, gender and location. Case–control studies scored poorly on method of ascertainment of exposure, mostly relying on written self-reports from participants. Non-response rates were also poorly reported in case–control studies. Discussion We have found that previous antibiotic use was associated with 12 different community-acquired infections, including infections of viral and bacterial origin. The associated infections ranged from common ailments such as upper respiratory tract infection (URTI) and infectious diarrhoea to relatively rare, but potentially life-threatening, infections such as meningitis. Our findings provide evidence that harms due to antibiotic use could extend beyond the widely recognized threat of promotion of drug-resistant organisms, to include an increased subsequent risk of infections caused by a range of microorganisms and at many different anatomical sites. Prior antibiotic consumption and subsequent infection risk was not the primary research question in most of the included studies. Most studies assessed several potential risk factors for infection, and in these the discovery of an association between antibiotic use and subsequent infection was an unexpected finding. The novelty of this finding was not commented upon by many authors. Despite inclusion as a potential risk factor in several studies, the hypothesis that prior antibiotic therapy could increase future risk of community-acquired infection has received little attention in the published literature. We did not find evidence of an association between prior antibiotic use and subsequent infection in any randomized controlled clinical trials of antibiotics. This was probably for two reasons. Firstly, prior antibiotic use was not an acknowledged confounding factor. Secondly, infections occurring at sites different to the target of the intervention would not have been seen as an adverse effect of the use of antibiotics. Recurrence of the target infection would have been interpreted as intervention failure, not an adverse effect of the intervention. The association between exposure to antibiotics and future infection may be explained by the disruptive effect of antibiotic therapy on the microbiota. Antibiotic-induced alteration of the microbiota could diminish the local suppressive effect of commensal organisms on pathogen overgrowth. There is evidence that the microbiota stimulates host epithelial cells to produce antimicrobial peptides36 and promotes antimicrobial activity of local immune cells.5,37 Furthermore, the gut microbiota may modulate systemic immunity and influence host susceptibility to infections distant from the gut.38 Experimental studies have shown that microbiota-depleted mice have reduced production of inflammatory cytokines, decreased bactericidal activity of macrophages, reduced production, extravasation and bactericidal activity of neutrophils and impaired antibody response to viral infections.38 If alteration of the microbiota was responsible for the increased infection risk following antibiotic therapy it might be expected that the risk of infection would return to baseline as the microbiota recovered to its pretreatment state. Four studies analysed variation in the strength of association between prior antibiotic therapy and subsequent infection in relation to the timing of antibiotic exposure.18,19,23,29 In three out of four studies there was a trend for weakening of the association between prior antibiotic therapy and subsequent infection as the interval between antibiotic use and diagnosis of infection increased.18,19,29 In the case of pneumococcal disease the observed association between antibiotic exposure and risk of infection weakened to the extent that the association failed to reach statistical significance for distant exposures, i.e. those occurring >30 days prior to the infection.19 Together these findings support the hypothesis that transient alteration of the microbiota may be responsible for mediating an increased infection risk following antibiotic exposure. One study on Salmonella infection was an exception since it did not demonstrate a weakening of the association between antibiotic exposure and subsequent infection as the interval between antibiotic exposure and diagnosis of infection increased.23 The risk of meningitis and invasive pneumococcal disease was found to increase with an increasing number of antibiotics to which the patient was exposed prior to the diagnosis of infection.18,19 Repeated exposure to antibiotics within the exposure interval would be expected to delay the recovery of the microbiota to its pretreatment state. This further strengthens the suspicion that antibiotic-induced alteration of the microbiota may be responsible for increased infection risk following antibiotic therapy. On the other hand the observed association between previous antibiotic exposure and subsequent infection could reflect excessive prior antibiotic consumption amongst persons with increased susceptibility to infections that was independent of the effect of antibiotic therapy on the microbiota. This susceptibility to infections could have been inherited or acquired. If this were the case, frequent antibiotic consumption above a specific threshold would be a clinically valuable marker in identifying a subset of patients who warrant further investigation for a previously unrecognized immune vulnerability. A further explanation for the observed association may lie in the variation between patients of the threshold for seeking a medical consultation for symptoms. Patients with frequent healthcare-seeking behaviour may be more likely to be diagnosed with infections, and to be prescribed antibiotics. One study controlled for frequency of medical care in their analysis and found an increased risk of URTI in patients exposed to long-term antibiotics for acne, even after adjusting for consultation frequency.31 Limitations and future work To our knowledge, this is the first systematic review to demonstrate the association between prior antibiotic therapy and subsequent risk of community-acquired infections, other than infections caused by antibiotic-resistant organisms, fungal organisms and C. difficile. The association between antibiotic use and subsequent risk of infection should be interpreted with caution. The observational nature of the included studies means it is not possible to establish causality. Most studies did not exclude antibiotic exposure occurring in the days leading up to the infection, and hence reverse causation should be considered as an alternative explanation for the observed association. Antibiotic exposure could have occurred as a result of medical consultation due to prodromal symptoms of the later infection being studied, rather than as a distant exposure due to an unrelated illness, which then increased susceptibility to the current infection. A further limitation is the possibility of confounding by indication resulting from underlying host susceptibility to infection, as described earlier. This may be addressed by designing studies that take this into account or adjust for infection susceptibility. The use of retrospective self-reports of antibiotic exposure in some studies is likely to have introduced recall bias. There was insufficient data on adherence to the prescribed course of antibiotics, which could have resulted in misclassification of exposure status. Most studies did not include sufficient prescribing details to establish whether there was a dose–response relationship between antibiotic usage and subsequent infection, or whether the relationship was stronger with broad-spectrum antibiotics, a finding that might have provided more support for the hypothesis of collateral microbiota damage. The studies on Salmonella infection provided inconclusive evidence for whether there was an association between infection and prior antibiotic therapy. This warrants further investigation, possibly through a meta-analysis, which could provide a more precise estimate of the overall effect size. A further limitation of our review was the exclusion of non-English-language papers and those that met the inclusion criteria but for which the full-text article was irretrievable. Conclusions Prior antibiotic therapy was associated with a diverse range of community-acquired infections. This included infections caused by antibiotic-resistant and non-resistant organisms. The association between antibiotic exposure and subsequent infection became weaker with increasing time from antibiotic exposure. The risk of infection increased with increasing number of antibiotics to which the patients were exposed. Whilst antibiotic therapy is often necessary for the treatment of bacterial infections, our findings highlight the continued need to limit inappropriate antibiotic prescriptions in primary care, both to reduce the consequences of bacterial resistance and possibly also to reduce the risk of future infections. The observed association may help clinicians in dissuading their patients from insisting on an antibiotic prescription when this is deemed not to be clinically indicated. New research may help discover whether other infection types, not examined by the studies included in our review, are also associated with prior antibiotic therapy. Further studies are also required to examine the mechanisms underlying the observed association, particularly whether this association could be mediated through antibiotic-induced collateral damage to the microbiota. Acknowledgements Conferences where work has previously been presented: (i) The Society for Academic Primary Care, London Annual Scientific Meeting, Cambridge, UK, 2016 (Abstract Number 81); (ii) Society of Academic Primary Care (SAPC), 45th Annual Scientific Meeting, Dublin, Republic of Ireland, 2016 (Talk Code EP3A.06); and (iii) 19th International Conference on Human Microbiome, Singapore, 2017 (Paper Code 17SG030091). Funding This study was conducted as part of our routine work. Transparency declarations None to declare. Author contributions U. M., D. A., M. A., L. M. and P. W. conceived the study. U. M. carried out the literature search and extracted the data. U. M. and P. W. independently selected eligible articles from potentially suitable articles identified. P. W. cross-checked extracted data. U. M. wrote the final manuscript. D. A., M. A., A. D., L. M., V. L., M. M., U. M. and P. W. contributed to the interpretation and revised the final manuscript. P. W. is guarantor for the study. Supplementary data Tables S1–S3 are available as Supplementary data at JAC Online. References 1 Llor C, Bjerrum L. Antimicrobial resistance: risk associated with antibiotic overuse and initiatives to reduce the problem. Ther Adv Drug Saf  2014; 5: 229– 41. Google Scholar CrossRef Search ADS PubMed  2 WHO. The Evolving Threat of Antimicrobial Resistance—Options for Action. 2012. http://www.who.int/patientsafety/implementation/amr/publication/en/. 3 Gulliford MC, Dregan A, Moore MV et al.   Continued high rates of antibiotic prescribing to adults with respiratory tract infection: survey of 568 UK general practices. BMJ Open  2014; 4: e006245. Google Scholar CrossRef Search ADS PubMed  4 Ursell LK, Metcalf JL, Parfrey LW et al.   Defining the human microbiome. Nutr Rev  2012; 70 Suppl 1: S38– 44. Google Scholar CrossRef Search ADS PubMed  5 Ubeda C, Pamer EG. Antibiotics, microbiota, and immune defense. Trends Immunol  2012; 33: 459– 66. Google Scholar CrossRef Search ADS PubMed  6 Levy J. The effects of antibiotic use on gastrointestinal function. Am J Gastroenterol  2000; 95 Suppl: S8– 10. Google Scholar CrossRef Search ADS PubMed  7 Jakobsson HE, Jernberg C, Andersson AF et al.   Short-term antibiotic treatment has differing long-term impacts on the human throat and gut microbiome. PLoS One  2010; 5: e9836. Google Scholar CrossRef Search ADS PubMed  8 Chang JY, Antonopoulos DA, Kalra A et al.   Decreased diversity of the fecal microbiome in recurrent Clostridium difficile-associated diarrhea. J Infect Dis  2008; 197: 435– 8. Google Scholar CrossRef Search ADS PubMed  9 Brown KA, Khanafer N, Daneman N et al.   Meta-analysis of antibiotics and the risk of community-associated Clostridium difficile infection. Antimicrob Agents Chemother  2013; 57: 2326– 32. Google Scholar CrossRef Search ADS PubMed  10 Xu J, Schwartz K, Bartoces M et al.   Effect of antibiotics on vulvovaginal candidiasis: a MetroNet study. J Am Board Fam Med  2008; 21: 261– 8. Google Scholar CrossRef Search ADS PubMed  11 Spinillo A, Capuzzo E, Acciano S et al.   Effect of antibiotic use on the prevalence of symptomatic vulvovaginal candidiasis. Am J Obstet Gynecol  1999; 180: 14– 7. Google Scholar CrossRef Search ADS PubMed  12 MacDonald TM, Beardon PH, McGilchrist MM et al.   The risks of symptomatic vaginal candidiasis after oral antibiotic therapy. Q J Med  1993; 86: 419– 24. Google Scholar PubMed  13 Moher D, Liberati A, Tetzlaff J et al.   Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ  2009; 339: b2535. Google Scholar CrossRef Search ADS PubMed  14 Malik U, Armstrong D, Ashworth M et al.   Association between antibiotic use and subsequent risk of community-acquired infections: a systematic review. PROSPERO: International prospective register of systematic reviews. 2016. CRD42016048521. http://www.crd.york.ac.uk/prospero/display_record.asp?ID=CRD42016048521. 15 Costelloe C, Metcalfe C, Lovering A et al.   Effect of antibiotic prescribing in primary care on antimicrobial resistance in individual patients: systematic review and meta-analysis. BMJ  2010; 340: c2096. Google Scholar CrossRef Search ADS PubMed  16 Covidence—Accelerate Your Systematic Review. 2016. https://www.covidence.org/. 17 Wells GA, Shea B, O'Connell D et al.   The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. 2014. http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp. 18 Armstrong D, Ashworth M, Dregan A et al.   The relationship between prior antimicrobial prescription and meningitis: a case-control study. Br J Gen Pract  2016; 66: e228– 33. Google Scholar CrossRef Search ADS PubMed  19 Chun CS, Weinmann S, Riedlinger K et al.   Passive cigarette smoke exposure and other risk factors for invasive pneumococcal disease in children: a case-control study. Perm J  2015; 19: 38– 43. Google Scholar CrossRef Search ADS PubMed  20 Doorduyn Y, Van Den Brandhof WE, Van Duynhoven YT et al.   Risk factors for Salmonella Enteritidis and Typhimurium (DT104 and non-DT104) infections in The Netherlands: predominant roles for raw eggs in Enteritidis and sandboxes in Typhimurium infections. Epidemiol Infect  2006; 134: 617– 26. Google Scholar CrossRef Search ADS PubMed  21 Delarocque-Astagneau E, Bouillant C, Vaillant V et al.   Risk factors for the occurrence of sporadic Salmonella enterica serotype typhimurium infections in children in France: a national case-control study. Clin Infect Dis  2000; 31: 488– 92. Google Scholar CrossRef Search ADS PubMed  22 Kass PH, Farver TB, Beaumont JJ et al.   Disease determinants of sporadic salmonellosis in four northern California counties. A case-control study of older children and adults. Ann Epidemiol  1992; 2: 683– 96. Google Scholar CrossRef Search ADS PubMed  23 Neal KR, Briji SO, Slack RC et al.   Recent treatment with H2 antagonists and antibiotics and gastric surgery as risk factors for Salmonella infection. BMJ  1994; 308: 176. Google Scholar CrossRef Search ADS PubMed  24 Pavia AT, Shipman LD, Wells JG et al.   Epidemiologic evidence that prior antimicrobial exposure decreases resistance to infection by antimicrobial-sensitive Salmonella. J Infect Dis  1990; 161: 255– 60. Google Scholar CrossRef Search ADS PubMed  25 Banatvala N, Cramp A, Jones IR et al.   Salmonellosis in North Thames (East), UK: associated risk factors. Epidemiol Infect  1999; 122: 201– 7. Google Scholar CrossRef Search ADS PubMed  26 Effler P, Ieong MC, Kimura A et al.   Sporadic Campylobacter jejuni infections in Hawaii: associations with prior antibiotic use and commercially prepared chicken. J Infect Dis  2001; 183: 1152– 5. Google Scholar CrossRef Search ADS PubMed  27 Early GJ, Seifried SE. Risk factors for community-associated Staphylococcus aureus skin infection in children of Maui. Hawaii J Med Public Health  2012; 71: 218– 23. Google Scholar PubMed  28 El-Gilany AH, Fathy H. Risk factors of recurrent furunculosis. Dermatol Online J  2009; 15: 16. Google Scholar PubMed  29 Gradel KO, Dethlefsen C, Ejlertsen T et al.   Increased prescription rate of antibiotics prior to non-typhoid Salmonella infections: a one-year nested case-control study. Scand J Infect Dis  2008; 40: 635– 41. Google Scholar CrossRef Search ADS PubMed  30 Luby SP, Faizan MK, Fisher-Hoch SP et al.   Risk factors for typhoid fever in an endemic setting, Karachi, Pakistan. Epidemiol Infect  1998; 120: 129– 38. Google Scholar CrossRef Search ADS PubMed  31 Margolis DJ, Bowe WP, Hoffstad O et al.   Antibiotic treatment of acne may be associated with upper respiratory tract infections. Arch Dermatol  2005; 141: 1132– 6. Google Scholar CrossRef Search ADS PubMed  32 McVernon J, Andrews N, Slack M et al.   Host and environmental factors associated with Hib in England, 1998-2002. Arch Dis Child  2008; 93: 670– 5. Google Scholar CrossRef Search ADS PubMed  33 Srikantiah P, Vafokulov S, Luby SP et al.   Epidemiology and risk factors for endemic typhoid fever in Uzbekistan. Trop Med Int Health  2007; 12: 838– 47. Google Scholar CrossRef Search ADS PubMed  34 Shallcross LJ, Hayward AC, Johnson AM et al.   Incidence and recurrence of boils and abscesses within the first year: a cohort study in UK primary care. Br J Gen Pract  2015; 65: e668– 76. Google Scholar CrossRef Search ADS PubMed  35 Mediano P, Fernandez L, Rodriguez JM et al.   Case-control study of risk factors for infectious mastitis in Spanish breastfeeding women. BMC Pregnancy Childbirth  2014; 14: 195. Google Scholar CrossRef Search ADS PubMed  36 Gallo RL, Hooper LV. Epithelial antimicrobial defence of the skin and intestine. Nat Rev Immunol  2012; 12: 503– 16. Google Scholar CrossRef Search ADS PubMed  37 Ivanov II, Atarashi K, Manel N et al.   Induction of intestinal Th17 cells by segmented filamentous bacteria. Cell  2009; 139: 485– 98. Google Scholar CrossRef Search ADS PubMed  38 Brown RL, Clarke TB. The regulation of host defences to infection by the microbiota. Immunology  2017; 150: 1– 6. Google Scholar CrossRef Search ADS PubMed  © The Author 2017. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please email: journals.permissions@oup.com.

Journal

Journal of Antimicrobial ChemotherapyOxford University Press

Published: Feb 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off