Life-long antimicrobial therapy: where is the evidence?

Abstract

The decision to prescribe long-term or ‘life-long’ antibiotics in patients requires careful consideration by the treating clinician. While several guidelines exist to help assist in this decision, the long-term consequences are yet to be well studied. In this review, we aim to provide a summary of the available evidence for patient populations where long-term antibiotic therapy is currently recommended in clinical practice. We will also discuss the pitfalls of this approach, including medication adverse effects, economic cost and any possible contribution to the emerging epidemic of microbial resistance.

Introduction

Since the introduction of early agents such as penicillin, streptomycin, sulphonamides and chloramphenicol around the middle of the twentieth century,¹ antibiotics have significantly reduced the morbidity and mortality of infectious diseases that were previously uncontrollable, or even lethal.²

Prolonged courses of antibiotics have been used for prophylaxis and suppression of infection since 1962 when penicillins were first used in managing sickle cell disease,^3,4 and 1944, when a prolonged course of sulfadiazine successfully treated an actinomycosis superinfection.⁵ In modern medicine, antibiotics are used for long-term therapy in a number of other ways in addition to suppression of ‘incurable’ infections: antibiotic prophylaxis in immunosuppression such as caused by HIV, post-transplantation and post-splenectomy, and for patient groups where antibiotics are used for purposes ostensibly not related to infection but as immune-modulating agents, such as minocycline in dermatology, azithromycin for bronchiectasis, or ciprofloxacin and metronidazole for inflammatory bowel disease.⁶

The rise of antibiotic use in medicine, and a corresponding increase in use in agriculture, has resulted in an evolutionary response from microbes in the form of antimicrobial resistance.^7,,8 Multiresistant organisms are now responsible for ∼25 000 deaths each year in both the United States and the European Union.⁹ In addition, the financial cost of infections by resistant organisms has been estimated to be at least €1.5 billion each year in the European Union alone.¹⁰ The threat of antimicrobial resistance is so significant that international governments, including those of Australia¹¹ and the United States,¹² have followed the advice of the WHO¹³ and developed a national strategy to tackle this issue. It is unclear if long-term human users of antibiotics moving in and out of hospitals and other parts of the healthcare system have acted as a catalyst for this widespread development of resistance in the same way as has agricultural use in non-human animals.

Other potential costs of life-long therapy include medication side effects, the risk of infection with multiresistant organisms and Clostridium difficile, drug–drug interactions, toxicity in pregnancy and effects on adherence to other medications. Metabolic effects and microbiome-related changes have not been well described. In part, this is difficult because of the limited study of long-term suppression as a therapeutic strategy. Problems not directly related to the individual may include the effect on the microbial milieu of healthcare institutions and the broader community. The financial costs of long-term therapy also need to be weighed up against the risk of recrudescence of any suppressed infection.

In this review, we summarize the evidence for commonly prescribed long-term antimicrobial therapy, and discuss the risks and benefits of such practices.

Methods

A search of the literature was performed focusing on international guidelines and recommendations, where available. We identified conditions where antibacterial agents were prescribed, using a duration of >12 months to define ‘long-term’. Searches of PubMed were conducted for articles published in English, from January 1966 to October 2016, using the terms ‘suppressive’, ‘prophylaxis’, ‘long-term’ and ‘antibiotics’. Relevant articles published between 1940 and 1966 were identified through searches in the authors’ personal files, and Google Scholar. Relevant references cited in those articles were reviewed.

Owing to the extensive scope of the topic of prolonged antibacterial therapy, a typical systematic review with a complete, exhaustive analysis of published guidelines and recommendations would be very difficult. The aim of this review was to summarize the variety of reported usage and the level of evidence to support prolonged antibacterial therapy for various indications, to generate areas for potential further study.

In this narrative review, we have divided long-term antibacterial therapy into three broad descriptive groups (Table 1): (i) suppressive therapy following an infection deemed ‘incurable’ (e.g. prosthetic device infection); (ii) prophylaxis (primary and secondary prophylaxis); and (iii) indications other than an antimicrobial effect (e.g. immunomodulation in chronic respiratory diseases), based on a cross-sectional study of antibiotic dispensing records in our own institution.¹⁴

Where applicable, published clinical experience in using prolonged antibiotics in each scenario was summarized. In assessing the pitfalls of long-term therapy, we searched for side effects of long-term medication, including adverse effects. Where available, recommendations were graded using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) system¹⁵ (Table 2), and evidence for these recommendations was graded using the Oxford Levels of Evidence¹⁶ criteria (Table 3). Table 4 summarizes the recommendations and their levels of evidence.

Suppressive antimicrobial therapy

Prosthetic joint infection

Even after >40 years of improvements in surgical techniques and the use of antimicrobial prophylaxis, the rates of prosthetic joint infection remain at ∼1%–2% per procedure.¹⁷ Despite extensive research, the role of suppressive antibiotics in the management of chronic prosthetic joint infections has been difficult to define, and the majority of recommendations are based on expert opinion, rather than clinical evidence.¹⁸ The IDSA Guidelines for Diagnosis and Management of Prosthetic Joint Infection state that clinical cure of a Staphylococcus aureus prosthetic joint infection is achievable with early debridement and a prolonged course of antibiotics, or with a two-stage revision. In situations where neither of these options is appropriate, indefinite oral suppressive therapy options include cefalexin, dicloxacillin and clindamycin.^18–21 For MRSA, minocycline, trimethoprim/sulfamethoxazole, pristinamycin or rifampicin in combination with fusidic acid²² can be utilized depending on antimicrobial susceptibility test results. In non-staphylococcal prosthetic joint infections, oral antimicrobials should be based on in vitro susceptibility, allergies and intolerances.¹⁹ Careful monitoring for efficacy and toxicity is recommended.¹⁹

The International Consensus on Peri-prosthetic Joint Infection concluded that upon review of the available evidence, they ‘do not recommend administration of antibiotics and open debridement alone without removing the implant in chronic Prosthetic Joint Infection’.²³ In cases where the prosthesis cannot be removed, chronic antibiotic suppressive therapy using antibiotic monotherapy with a good safety profile and high oral bioavailability is recommended, following a potent initial induction phase to control the infection.²³ There was ‘no consensus about the length of time that patients should receive suppressive antibiotic therapy’.²³

There is limited reported clinical experience about the use of long-term suppressive antibiotics; however, small studies have reported success rates of up to 86%.^24,,25 A smaller study looking specifically at elderly patients (80 years and older) quoted a successful suppression rate of 60%.²⁶ In that series, failure of suppressive antibiotics was most often associated with staphylococci compared with other bacteria. Other risk factors for failure included low serum albumin and presence of a sinus tract.²⁶ Cure was more likely when there was a shorter duration of symptoms and time to diagnosis, and if the joint affected was a hip.²⁷ One study reported complications related to suppressive antibiotics in 22% of their cases;²⁴ however, these were not significant enough to warrant discontinuation. Tornero et al.²⁸ devised a score (KLIC score: Kidney, Liver, Index surgery, Cemented prosthesis and C-reactive protein value) to predict early prosthetic joint infection treatment failure after debridement. It was recommended that patients with a high KLIC score be treated with multiple debridements, removal of prosthesis (with one- or two-stage exchange) or utilizing more potent biofilm-active antimicrobial agents.²⁸

Vascular graft infection

The incidence of infections after vascular graft procedures has been reported to be between 1% and 6%.²⁹ The American Heart Association Guidelines classify vascular graft infection into five classes, Samson I–V,³⁰ reflecting how extensive the infection is, and involvement of the graft. At least 6 weeks of intravenous therapy is recommended for infections involving the graft, followed by 6 months of an appropriate oral agent based on microbiology results. Long-term suppressive therapy should be considered for infections caused by multiresistant organisms or Candida species, or in complex surgical cases.³⁰ Long-term therapy may be used in aortic graft infections as an adjunct, in combination with other conservative management options, e.g. surgical debridement or percutaneous drainage.³¹ There are only small case series describing the use of suppressive antibiotics, with varying reports on successful suppression and long-term survival.^32–35 An Australian case series from 2010 described five cases of abdominal aortic aneurysm graft infection that were managed with long-term suppressive antibiotic therapy, owing to significant medical comorbidities. All of the patients survived to a median follow-up time of 32 months, the longest having survived for 6 years at the time of writing.³² In another review of 51 cases of patients on chronic suppressive therapy for intravascular device infections, three patients developed relapsing infection while on antibiotic therapy and three cases of adverse drug effects were reported: Candida species superinfection, rash and neutropenia.³⁶

Cardiovascular implantable electronic device infection

Cardiovascular implantable electronic devices (CIEDs) include implantable cardiac defibrillators, cardiac resynchronization therapy devices and permanent pacemakers. The incidence of implantable cardiac device infections is estimated to be <2%³⁷ and cardiac device infections (including device lead infections) now make up ∼10% of all endocarditis cases.³⁸ Guidelines recommend complete removal of CIEDs in patients with device-related infections. However, comorbidities or other factors may make device removal problematic, and following initial intravenous antibiotic therapy, long-term oral suppressive antimicrobial therapy can be attempted.³⁷

There are limited data on outcomes of patients receiving suppressive antimicrobial therapy for CIED infections. Tan et al.³⁹ reviewed 660 cases of infected implantable cardiac devices and identified 48 patients who did not have their device explanted, and were managed with chronic antibiotic suppression. Twelve patients had died either during their index admission or within 1 month of discharge, highlighting the high mortality faced when managing such infections. At 1 year follow-up, 6 of 33 patients (18%) developed relapsed infection and 6 (18%) reported adverse effects of prolonged antibiotic therapy, including rash and C. difficile infection.

Prophylaxis

Antibiotics may be used to prevent an initial infection (primary prophylaxis), or to prevent a recurrence, or reactivation of an infection (secondary prophylaxis).⁶ Antibiotics may be used for brief courses for surgical prophylaxis, and post-exposure prophylaxis (e.g. contacts of patients with Neisseria meningitidis or Bordetella pertussis infections).⁶ In this review we focus on antibiotic prophylaxis in the context of long-term use, defined as a duration of >12 months. The review is limited to the most common and best-supported uses of antibacterial long-term prophylaxis.

Prophylaxis post-splenectomy

Hyposplenic patients are at risk of severe infections with encapsulated organisms such as Streptococcus pneumoniae.^40–42 International guidelines recommend antibiotic prophylaxis for at least 2 years post-splenectomy in adults, coupled with vaccination to prevent severe infections from encapsulated organisms.⁴³ This is largely based on the use of antibiotics in children with sickle cell diseases,⁴⁴ and retrospective evidence that outcomes are worse if post-splenectomy antibiotic prophylaxis,⁴⁵ in addition to vaccination and education, are not utilized.⁴⁶

This is mirrored in Australian and UK guidelines,^41,,47 with penicillin-based prophylaxis recommended for at least 2 years in adults, and life-long if the patient is otherwise immunocompromised, or considered high risk for pneumococcal infection. Antibiotic choices include 250–500 mg oral amoxicillin daily, or 250–500 mg phenoxymethylpenicillin twice daily. A macrolide such as roxithromycin or erythromycin can be used if there is a penicillin allergy.⁴¹ In some guidelines longer-term prophylaxis in adults is generally not advised due to the potential to select for resistance and reduced efficacy due to non-compliance.^46,,48 Penicillin-resistant S. pneumoniae is a concern and has been associated with prophylactic antibiotic use.^49,,50 Apart from in children with sickle cell disease, evidence for efficacy of post-splenectomy prophylactic antibiotics is limited.⁵¹

Pneumocystis pneumonia prophylaxis (primary and secondary)

Pneumocystis jirovecii pneumonia (PJP) prophylaxis in patients with AIDS does not meet our review definition of long-term therapy of >12 months. Prophylaxis in this population is recommended when the CD4+ T cell count is <200 cells/mm³, and should be continued until the CD4 count is >200 cells/mm³ for at least 3 months after commencing ART (level A evidence, ‘high’ GRADE recommendation).^52,53 This review will specifically discuss the role of PJP prophylaxis in patients with haematological malignancies, solid organ transplants and immunosuppressive agents.

In haematopoietic stem cell transplants, prophylaxis for PJP is recommended for all patients after engraftment for at least 6 months and for the duration of immunosuppressive therapy for chronic graft-versus-host disease. The drug of choice for PJP prophylaxis is trimethoprim/sulfamethoxazole.⁵⁴ This is also recommended for prophylaxis after solid organ transplants,⁵⁵ and may provide protection against other infections including Toxoplasma and Listeria species.⁵⁶ Optimal duration of prophylaxis varies according to international guidelines, from 4 to 12 months post-transplantation.^55,57,58

Corticosteroids, cytotoxic agents such as cyclophosphamide and cyclosporine, and biological agents such as TNF-α inhibitors and monoclonal antibodies have been associated with an increased risk of PJP.⁵⁹ Prophylaxis is recommended for patients taking ≥20 mg of prednisolone daily (or equivalent) for >1 month who also have another cause of immunocompromise.^60–64 Patients on the monoclonal antibody alemtuzumab or purine analogue chemotherapy such as fludarabine should be offered PJP prophylaxis until a minimum of 2 months after cessation of therapy, or until the CD4 count is >200 cells/mm³.⁶² Patients taking TNF inhibitors in combination with high-dose glucocorticoids or other intensive immunosuppressive agents should also be offered chemoprophylaxis.⁶⁵ While PJP prophylaxis is not recommended for patients on methotrexate monotherapy, prophylaxis is suggested when co-administered with high-dose glucocorticoids in granulomatosis with polyangiitis;⁶⁶ however, there are theoretical concerns about myelosuppression with trimethoprim/sulfamethoxazole being used in conjunction with methotrexate.

A Cochrane review found that trimethoprim/sulfamethoxazole prophylaxis in non-HIV patients reduced PJP infections by 91% [relative risk (RR) 0.09, 95% CI 0.02–0.32], with a number needed to treat of 15, and therefore should be considered in the setting of haematological malignancies, bone marrow transplant and solid organ transplants.⁶⁷ Adverse effects of prolonged trimethoprim/sulfamethoxazole used for PJP prophylaxis have been reported, including bone marrow suppression and deranged liver function tests;⁶⁸ however, a meta-analysis of four randomized control trials found no significant increase in these adverse events in the groups on trimethoprim/sulfamethoxazole compared with those on no prophylaxis.⁶⁷

Urinary tract infection (primary and secondary)

The use of long-term daily antibiotics as prophylaxis against recurrent uncomplicated urinary tract infections (UTIs) in women has been found to be effective.^46,,69,,70 A threshold of three UTIs in 12 months has been suggested to signify recurrent infections in which antibiotic prophylaxis could be considered.⁶⁹ Suggested antimicrobial agents include daily trimethoprim/sulfamethoxazole, trimethoprim, ciprofloxacin, cefalexin and norfloxacin.^46,,69,,70 Evidence is less robust for the continuous use of antibiotics for recurrent UTIs in other populations, including patients with long-term catheters in place,⁷¹ where guidelines recommend against systemic antibiotic prophylaxis. Long-term antibiotic use for UTI prophylaxis has been better studied in children, where a Cochrane review found a small, but significant, benefit in preventing repeat symptomatic UTIs.⁷² Two randomized controlled trials reported adverse effects associated with prolonged antibiotic consumption and three trials reported emergence of antimicrobial resistance in the treatment arms.⁷²

Rheumatic fever

Rheumatic fever is the most common cause of heart disease in children in developing countries,⁷³ and valvulopathy can progress over years to decades and may lead to severe heart failure if not managed appropriately.⁷³ Because the damage caused by rheumatic fever to the heart valves may worsen with each recurrent infection, the major long-term strategy in management is prevention of recurrent infections with secondary prophylaxis.

The WHO and American Heart Association recommend secondary prophylaxis for all patients who have had an attack of rheumatic fever, even if they do not have residual rheumatic valvular heart disease.^74,,75 The purpose of this is to prevent colonization of the upper respiratory tract with group A β-haemolytic streptococci, and the development of recurrent attacks of rheumatic fever. Penicillin is the recommended antimicrobial of choice, or sulphonamides or erythromycin if there is a penicillin allergy.⁷⁶ Duration of recommended prophylaxis ranges from 5 years after the last attack to life-long, depending on various risk factors including residual valvular heart disease.⁷⁴ A Cochrane review found there was a reduction of recurrence of rheumatic fever by 55% when penicillin was compared with no treatment, and favoured intramuscular over oral penicillin.⁷⁷ The emergence of penicillin-resistant streptococci in the oral flora of patients on penicillin prophylaxis was reported as early as 1949.⁷⁸ As such, the American Heart Association recommends use of another antibiotic, such as azithromycin or clindamycin, in high-risk patients if they require prophylaxis for dental procedures.⁷⁹

Spontaneous bacterial peritonitis

The European Association for the Study of the Liver recommends prophylactic antibiotics for spontaneous bacterial peritonitis (SBP) be restricted to use in three high-risk categories: (i) patients with acute gastrointestinal haemorrhage (in the setting of oesophageal varices); (ii) patients with low total protein content in ascitic fluid; and (iii) patients with a previous history of SBP.⁸⁰ Guidelines published by the American Association for the Study of Liver Diseases agree, recommending long-term daily prophylaxis for patients with previous SBP with daily norfloxacin or trimethoprim/sulfamethoxazole.^46,,81 In patients with cirrhosis and ascites, long-term prophylaxis can be considered if the ascitic fluid total protein is ≤1 g/dL or serum bilirubin is >2.5 mg/dL.⁸¹ A Cochrane review found that patient groups on SPB prophylaxis had a significantly lower infection rate, as well as a significant prevention of mortality.⁸² Adverse effects included nausea, oro-oesophageal candidiasis, diarrhoea, rash and drug fever; however, they were not found to be significantly greater than that among patients treated with either placebo or no treatment.⁸² Fluoroquinolone resistance has been reported with prophylactic use of norfloxacin, and was found to develop as early as 14 days into therapy.⁸³

Antimicrobial therapy for non-antimicrobial indications

Cystic fibrosis

The American Thoracic Society recommends chronic use of azithromycin in patients with cystic fibrosis (CF) with Pseudomonas aeruginosa persistently isolated from respiratory specimens.⁸⁴ The role of macrolide antibiotics in patients with CF is believed to be immunomodulatory rather than antibacterial.⁸⁵ A Cochrane review looked at 31 randomized controlled trials comparing short- and long-term use of macrolide antibiotics compared with placebo or other antimicrobial classes. Patients receiving at least 6 months of azithromycin had improved pulmonary function and reduced exacerbation episodes.⁸⁵ A meta-analysis of six studies on the adverse effects of long-term azithromycin use in patients with chronic lung diseases raised concerns regarding antibiotic resistance and potential hearing impairment.⁸⁶ A Cochrane review (updated 2017) has assessed the role of antistaphylococcal prophylaxis for children with CF, finding it resulted in reduced isolation of S. aureus in sputum, and fewer hospitalizations;⁸⁷ however, the number of studies reviewed was small. A trend towards higher rates of P. aeruginosa isolation in the later years of follow-up was noted; however, this was not statistically significant, and the authors noted that follow-up was not adequate to comment on long-term effects of prophylaxis.⁸⁷

Chronic airways disease

Prolonged courses of macrolide antibiotics, such as roxithromycin, clarithromycin and azithromycin, have been shown to reduce the signs and symptoms of diffuse panbronchiolitis, as well as improve survival, in chronic airways diseases such as asthma and non-CF bronchiectasis; however, there is insufficient evidence to support its use.⁸⁸

A retrospective cohort study of 92 576 patients with chronic airways disease in the UK demonstrated that 0.61% were on long-term antibiotics, with tetracyclines and penicillins being used most commonly, followed by macrolides.⁸⁹ Recent studies showed that pulsed moxifloxacin or daily macrolides reduced the number of and duration of exacerbations, while improving quality of life.^90–92 A study in patients with uncontrolled asthma found that azithromycin administered thrice weekly significantly reduced exacerbations per patient-year and improved asthma-related quality of life; however, diarrhoea was more frequent in patients treated with azithromycin.⁹³ Because of the side effect burden and risk for development of resistance, current guidelines do not recommend long-term prophylaxis, and it should only be considered in select high-risk patients.^88,,89

Acne

Tetracyclines have multiple anti-inflammatory actions, including reducing neutrophil chemotaxis and inhibiting pro-inflammatory cytokines.⁹⁴ Therefore its role in acne vulgaris is as an anti-inflammatory, as well as having direct antibacterial effects.⁹⁴ A Cochrane review⁹⁵ of 27 randomized controlled trials compared minocycline with other tetracyclines, isotretinoin, topical therapy and hormonal therapy in the treatment of inflammatory acne vulgaris. There was no reliable evidence that minocycline was better than any other acne treatment, and antibiotic resistance in Propionibacterium acnes has been noted as an emerging problem.⁹⁶ In light of this, long-term therapy is not recommended.

Inflammatory bowel disease

Two meta-analyses of randomized controlled trials, comparing antibiotics with placebo, found that antibiotics were more effective than placebo for induction and maintenance of remission in Crohn’s disease, as well as being associated with an improvement in clinical symptoms.^97,,98 The ideal duration of antibiotic therapy for maintenance of remission or prevention of disease progression is not known, and antibiotic regimens described in these two systematic reviews were heterogeneous.^97,,98 Prolonged courses of antibiotics in Crohn’s disease have been studied in three randomized trials, all using antimycobacterial agents, for durations ranging from 9 months to 2 years, demonstrating a significant reduction in relapses (RR of relapse 0.62, 95% CI 0.46–0.84, number needed to treat=4);^99,,101 however, this prolonged use to maintain remission is not recommended in international guidelines.¹⁰²

Rifaximin, ciprofloxacin and metronidazole have roles in short-term therapy for bacterial overgrowth in Crohn’s disease; however, long-term courses have not been studied.¹⁰³ While a meta-analysis has shown an improvement in clinical symptoms when comparing antibiotics with placebo for ulcerative colitis,⁹⁸ the longest course of antibiotics studied was 3 months. Long-term antibiotic therapy has not been established in ulcerative colitis. Prolonged courses of ciprofloxacin and metronidazole are not recommended due to toxicities associated with prolonged use, and are described later in this article.

Summary

There is a range of conditions in which long-term antibiotics are prescribed, but, in many cases, the practice is not supported by high-quality evidence. This is particularly apparent in conditions using antimicrobials as anti-inflammatory agents, such as in pulmonary disease, acne and inflammatory bowel disease. The practice with the least supporting evidence is the use of suppressive antimicrobial therapy for infected prosthetic material that cannot be removed. This will increasingly be a challenge as more elderly and frail patients receive prosthetic implants. Until surgical techniques and antimicrobial therapy can be improved to facilitate a cure in such infections, or unless evidence highlights a greater harm than benefit, suppressive long-term antibiotics remain a reasonable therapeutic option.

Adverse effects

The use of long-term antibiotics has been well studied for only certain indications. Antibiotics have often been studied for short-term use and the long-term side effects are often less well described. It is likely however that the side effects, real and theoretical, from short-term use, probably apply to prolonged use.

Potential risks of long-term suppressive antibiotics include the emergence of resistance, medication side effects and toxicity, and C. difficile infections. Careful consideration and clinical judgement should be utilized when choosing which patients should be placed on long-term oral suppressive therapy.⁹⁰ Known complications from prolonged antibiotic therapy may include tendinopathy with fluoroquinolones,¹⁰⁴ photosensitivity with doxycycline,¹⁰⁵ peripheral neuropathy with metronidazole¹⁰⁶ and cytopenias with linezolid.¹⁰⁶ Indeed, prolonged courses of metronidazole and linezolid are usually not used for these reasons. Serious, potentially life-threatening side effects, such as arrhythmias with clarithromycin and erythromycin, need to be appreciated, particularly in patients with underlying cardiac disease.^108–110

Moreover, in short courses, azithromycin has been reported to be associated with an increased risk of fatal cardiac arrest, even in patients without underlying cardiac arrhythmias.¹¹¹ This is a rare side effect of azithromycin therapy, and was exemplified when a review looked at ∼350 000 prescriptions of the drug and found an increased risk of cardiovascular death when compared with no antibiotics and amoxicillin (hazard ratio 2.88; 95% CI 1.79–4.63; P < 0.001, and hazard ratio 2.49; 95% CI 1.38–4.50; P = 0.002 respectively).¹¹¹ Large numbers (>65 000) were also required to demonstrate a short-term increased risk of neuropsychiatric events associated with Helicobacter pylori treatment containing clarithromycin.¹¹² Case reports have described episodes of torsades de pointes associated with voriconazole use,^113,,114 and concern has been raised about the possible link between long-term azole antifungals and skin cancers.¹¹⁵ These studies reflect the difficulties of assessing rare side effects, and as there are few trials studying long-term antimicrobial therapy, rare side effects may not be noticed due to small participant numbers.

Change to the microbiome

The term microbiota is defined as the organisms that live in symbiosis in a human and can consist of up to 100 trillion micro-organisms per person, primarily made up of bacteria in the gastrointestinal tract.^116,,117 The microbiome includes these micro-organisms, but also their genes and their surrounding environment.¹¹⁸ In addition to conventional dose-related or idiosyncratic complications of prolonged therapy there are some more indirect or less well-established side effects, including effects to the microbiome. A well-accepted effect is that of promoting resistant commensal organisms in the gut,¹¹⁹ discussed in further detail below. More speculative are other long-term effects of antibiotics that include increasing mitochondrial dysfunction and oxidative damage¹²⁰ and altering the microbiome even with only short courses of therapy¹²¹ with unclear consequences.¹²² Since microbiome ecology has been associated with, inter alia, psychiatric states and body mass index these changes may have a significant effect on an individual’s morbidity and mortality.^123,,124

Antimicrobial resistance

In addition to the evidence of emerging antimicrobial resistance highlighted in the specific conditions above, multiple studies have shown a significant link between prior or current antibiotic use, and isolation of resistant Enterobacteriaceae,^125,,126 enterococci¹²⁵ and Pseudomonas.¹²⁷ Increasingly widespread use of macrolide antibiotics has resulted in the emergence of macrolide-resistant S. aureus.^128–130 It has also been noted that the strength of the association increased as the duration of use increased.¹²⁷ The escalating problem of antimicrobial resistance is attributed to use in healthcare and agriculture; however, the effect of prolonged use of antibiotics as prophylaxis or for suppression on microbial resistance patterns has not been studied. Although this intervention could change microbial resistance patterns, the extent of this change is not known.⁸⁸

Drug–drug interaction

Antibiotics can also interact with other regular medications, affecting serum drug levels by inhibiting or inducing cytochrome P450,¹³¹ a host of liver and small intestinal enzymes that have a number of roles, the most important of which aids in the metabolism of drugs.¹³² For example, rifampicin is a potent inducer of cytochrome P450, and can reduce blood concentrations of warfarin, the oral contraceptive pill, cyclosporine, corticosteroids and methadone, just to name a few.¹³³ Drugs can act as an enzyme substrate, an inducer, or an inhibitor, and can thus cause a number of adverse effects, including drug toxicities and reduced pharmacological effect.¹³¹ Medication reviews should be performed before commencing long-term antibiotics, with careful attention paid to whether the drugs prescribed can cause significant drug–drug interactions.

Cost

In general, a low cost threshold will have to be set to make antibiotics cost-effective. For example, a hospital day in Australia has a nominal cost of >$A600.¹³⁴ A week of inpatient therapy would be >$A4000. In Australia, a pack of 20 amoxicillin tablets costs $A8.53, so 1 week of inpatient therapy is ∼9000 days of prophylaxis in cost. Previously this sort of prophylaxis has been shown to be cost-effective in the setting of post-splenectomy care.¹³⁵ That example uses the cheapest alternative. The most expensive antibiotic in our hands at the moment is pristinamycin, which is about $A4 a tablet. The point is that prophylaxis, if it works, is clearly effective in terms of cost of medicines—it is only the cost of side effects including promotion of resistance that are likely to make it unattractive.

Adherence

Long-term treatment of any sort may be difficult to take due to perceived side effects and other factors. In the cardiovascular literature where a chronic asymptomatic condition is offset against the difficulty of taking regular medications and their side effects the data show many patients only adhere to 50%–60% of their therapy.^136,,137 With specific respect to antimicrobials, 42% of patients who have had a splenectomy had evidence of penicillin in their urine in one British study,⁴⁸ 40% compliance was recorded in a study on children on prophylactic antibiotics for UTIs¹³⁸ and only two-thirds of participants on long-term azithromycin for CF were compliant with treatment.¹³⁹ Similar studies in children, where adherence might be suspected to be greater because it would be imposed by adults, found penicillin in the urine of only 10 of 31 children with sickle cell anaemia.¹⁴⁰ Low rates of adherence have also been seen with secondary prophylaxis for rheumatic fever,¹⁴¹ where having had the disease did not seem to encourage greater adherence, and for antimalarial prophylaxis.¹⁴² Similar findings have been found in the context of HIV early in the ART era.¹⁴³ Our review could not find any data on adherence in populations of patients on long-term suppressive antibiotics for prosthetic joint infections or vascular graft infections. In circumstances where therapeutic drug level testing cannot be performed, novel methods to assess adherence, such as looking for drug presence in other tissues such as hair,¹⁴⁴ could be considered.

Trial of stopping

A study from the UK¹⁴⁵ looked at outcomes after prosthetic joint infection managed with debridement, antibiotics and implant retention, in particular the association with duration of antibiotics. The risk of treatment failure increased by 4-fold after oral antibiotics were ceased, with the greatest risk being in the first 4 months after discontinuation of antibiotics. A large case series of prosthetic joint infection with S. aureus found the success rate of debridement, antibiotics and implant retention to be 55%,¹⁴⁶ but this low rate may reflect the heterogeneity of surgical and antibiotic therapy.¹⁹ Previously revised joints that had become infected and joints washed out arthroscopically, rather than with open debridement, were associated with a significant risk of treatment failure.¹⁴⁵ It was noted that the majority of patients who discontinued their chronic suppression did not suffer from treatment failure, suggesting that many patients are cured without the use of chronic suppression, but defining that group of patients can be difficult.¹⁹ There are no comparative trials to guide clinicians as to when to stop antimicrobials in other aspects of suppressive treatment such as pacemaker infections.^147,,148

Conclusions

The use of antibiotics for long-term therapy is infrequent, but their role was described as early as a few years after penicillin was first used in clinical practice. Prolonged, if not life-long courses are used for primary and secondary prophylaxis, as well as having non-infective purposes such as anti-inflammatory or prokinetic. Evidence supporting the use of long-term antibiotics in these conditions ranges from strong, to absent. The long-term effects, particularly in terms of changes in microbiome, influences on hospital ecology including the development of resistance, effectiveness, chronic side effects and adherence are not well studied. Clinicians should not prescribe long-term or indefinite antibiotic therapy for indications that lack sound clinical and scientific evidence. Ultimately, the final decision on antibiotic therapy lies with the healthcare practitioner who must weigh up the risks and benefits to both the individual patient, as well as the public at large.

No review article can summarize the evidence for long-term antibiotics for all indications; however, there is a great deal of scope for studying the long-term risks and benefits of the indefinite use of antibiotics more systematically with reference across indications. Standardized guidelines for how to follow-up or monitor patients on long-term therapy should be developed, and further research should be conducted into the plausibility of ceasing long-term therapy.

Transparency declarations

None to declare.

References

Author notes