Inappropriate Use of Antimicrobials for Lower Respiratory Tract Infections in Elderly Patients: Patient- and Community-Related Implications and Possible Interventions
The elderly are more susceptible to infections, which is reflected in the incidence and mortality of lower respiratory tract infections (LRTIs) increasing with age. Several aspects of antimicrobial use for LRTIs in elderly patients should be considered to determine appropriateness. We discuss possible differences in microbial etiology between elderly and younger adults, definitions of inappropriate antimicrobial use for LRTIs currently found in the literature, along with their results, and the possible negative impact of antimicrobial therapy at both an individual and community level. Finally, we propose that both antimicrobial stewardship interventions and novel rapid diagnostic techniques may optimize antimicrobial use in elderly patients with LRTIs.
Reports on (in)appropriate antimicrobial use lack a reference standard for defining and measuring appropriateness of treatment.
Quinolones or macrolides should be restricted to selected cases empirically, given the low incidence of atypical pathogens in elderly patients and higher risks of adverse drug events and drug–drug interactions.
The use of low-dose computed tomography scanning, point-of-care ultrasonography, or point-of-care polymerase chain reaction testing for viral pathogens are promising research areas to decrease the inappropriate use of antimicrobials.
Elderly people (adults over 65 years of age) comprised one-fifth of the total population in Europe in 2016, and this proportion may further increase to 25% in 2030 [1, 2]. The elderly are more susceptible to infections and their sequelae than younger adults [3, 4], which is reflected in the incidence and mortality of lower respiratory tract infections (LRTIs) increasing with age [5, 6, 7]. Next to increased incidences of comorbidities, it is thought that age-related altered immune regulation, often referred to as ‘immunosenescence’, also contributes to this . Several aspects of LRTIs in the elderly make it increasingly difficult to determine the most appropriate antimicrobial therapy for individual patients. First, the etiology of LRTIs in elderly patients could be slightly different compared with younger adults, which would require adjusted empirical antimicrobial therapy. In addition, the diagnosis of LRTIs in the elderly could be more challenging, which might lower the threshold for prescribing antimicrobials. Furthermore, with advancing age, the human body changes in composition and organ function, resulting in alteration of the pharmacokinetics and pharmacodynamics of antimicrobials . When combined with the increasing frequency of comorbidities and/or polypharmacy, this facilitates the occurrence of adverse drug events (ADEs) and drug–drug interactions (DDIs).
We discuss the etiology, currently used definitions for appropriate use of antimicrobials, and different negative consequences of antimicrobial therapy for LRTIs in elderly patients, in both individual patients and the community. Finally, we propose targeted interventions to improve antimicrobial prescribing in these patients.
2 Microbiological Etiology
Identified causal pathogen in studies with head-to-head comparison between younger and elderly patients
Young patientsa,c (%)
Elderly patientsb,c (%)
Total viral pathogens
Respiratory syncytial virus
Higher incidences of viral pathogens in elderly patients have been reported [12, 14, 16], although this was not demonstrated consistently [10, 11, 15]. Studies showing higher incidences of viral pathogens were mostly population-based, including outpatients from general practitioners, while contradicting studies were hospital-based, which may explain conflicting results. In addition, bias could have been introduced because viral testing was not uniformly performed and age may have influenced the decision to test . Two recent cohort studies of hospitalized community-acquired pneumonia (CAP) patients, one from the US (N = 2259) and one from The Netherlands (N = 408), routinely performed testing and found a virus as the only pathogen in 22% and 13% of cases, respectively. In both studies, the most commonly detected viral pathogens were human rhinovirus and influenza virus [20, 21].
The occurrence of a viral etiology was similar for young and elderly patients in the US study , while the study from The Netherlands found even more viral pathogens with increasing age.  The contribution of viral pathogens as a cause for CAP in all age groups might be larger than previously thought.
Elderly patients more often had unknown etiology in LRTIs compared with younger adults [10, 11, 13, 15, 16]. Possible reasons for less detection of the causal pathogen could be a more conservative diagnostic approach in this group of patients or difficulty in obtaining good-quality material for culturing .
In recent years, healthcare-associated pneumonia (HCAP) has been proposed as a separate entity from CAP. HCAP may be more frequently caused by gram-negatives and multidrug-resistant organisms (MDROs), and is associated with higher mortality [22, 23, 24, 25]. However, a meta-analysis of 24 studies concluded that HCAP criteria do not accurately predict MDROs, although low study quality and heterogeneous designs preclude a firm conclusion . European studies tend to report community-like etiology, while studies from Asia and the US show increased MDRO rates in HCAP patients [25, 26, 27, 28]. To date, the clinical relevance of HCAP remains unclear [23, 24]. In fact, both Infectious Diseases Society of America (IDSA) and European Society of Clinical Microbiology and Infectious Diseases (ESCMID) guidelines do not address HCAP, which leaves a gap in the recommendations regarding treatment for patients from long-term care facilities (LTCFs).
The microbial etiology does not justify routine empirical coverage of Legionella pneumophila as the low incidence is further decreased in elderly patients to approximately 1–5% (Table 1). As members of our group have previously suggested, β-lactam monotherapy, preferably aminopenicillins, should generally be the first choice of empirical therapy [29, 30]. Naturally, the severity of disease, local epidemiologic data, prior cultures or known colonization of individual patients, comorbidities, or allergies could lead to an alternative antibiotic choice. Doxycycline, the addition of a macrolide to β-lactam therapy, or the newer fourth-generation fluoroquinolones are potent therapies, but higher risks of ADEs and DDIs should be considered.
3 Inappropriate Use of Antimicrobials
Whether to start antimicrobial therapy for LRTIs, and choosing the specific class, depends on multiple factors in daily clinical practice. Most factors in this decision pathway are dependent on clinical judgement, which interferes with evaluating appropriateness in a standardized way. Deviation from protocol for empirical therapy, definitive drug selection, and duration of therapy might be justifiable for individual patients, for reasons that are not captured in the guidelines.
Examples from the literature of different definitions of (in)appropriate antimicrobial use for LRTIs in the elderly
Definition of inappropriate antimicrobial use
Appropriateness of RTI treatment
Tobia et al., 2008
at emergency department [N = 153]
Medication Appropriateness Index (MAI)
1. Indication (e.g. presence of symptoms)
4. Directions (e.g. route)
5. Practicality (e.g. adherence)
6. Drug–drug interactions
7. Drug–disease interactions
8. Unnecessary duplication
10. Expensiveness (least expensive alternative)
A, appropriate; B, marginal; C, inappropriate
n = 99 (65%)
Van Buul et al., 2015
Long-term care facility [N = 208]
Algorithm for RTI based on guidelines and national expert panel
Distinction between: (1) acute cough; or (2) no acute cough but fever; or (3) no cough and fever
Then presence/absence of abnormalities on lung auscultation, COPD, CRP results, other airway and non-airway symptoms, and certain risk factors
A, appropriate; B, probably appropriate; C, probably inappropriate; D, inappropriate; E, insufficient information
n = 180 (86.5%; range 60.0–96.2)
Vergidis et al., 2011
Long-term care facility [N = 752]
Appropriate (with/without antimicrobial prescription)
With: when effective drug was used
Without: when use of an antimicrobial was not indicated
With: when a more-effective drug was indicated
With: use of any antimicrobial was not indicated
Without: when use of an antimicrobial was indicated
Insufficient information for categorization
n = 592 (79%)
Loeb et al., 2001
Long-term care facility [N = 646]
Assessment of prescriptions to see if they fulfilled the diagnostic criteria
At least three of the following: (1) new/increased cough; (2) new/increased sputum production; (3) fever; (4) pleuritic chest pain; (5) new or increased physical findings on chest examination; (6) new/increased shortness of breath or respiratory rate more than 25/min, or worsening mental or functional status
n = 375 (58%)
It is clear that a reference standard for measuring inappropriate antimicrobial use is currently lacking, which was also concluded by a specific review of this subject . More work on definitions and standardization is heavily needed to ensure evaluation of appropriateness will become more reliable and less dependent of the interobserver variation inherent to expert evaluation .
Although it has been suggested that certain potential inappropriate prescriptions could be automatically recognized by an electronic health records system, we think this is unfeasible as human judgement is almost always needed . For example, a patient with a documented allergy to guideline-recommended treatment could be automatically flagged by such a system as appropriate deviation, yet this is dependent on accurate interpretation and registration of allergy data, with the latter often being incorrect .
Despite the heterogeneity in definitions, and resulting heterogeneity in inappropriateness rates, there is a clear consensus on the adverse consequences of inappropriate antimicrobial use at the patient and community level [55, 56]. Both will be addressed in the following section.
4 Negative Impact of Antimicrobials in the Elderly
Treating patients with antimicrobials is not without risk. Antimicrobials can have a negative impact on both individual users and the community. For individuals, there is a risk of different ADEs, including DDIs and drug–comorbidity interactions. In elderly patients, the age-related changes in pharmacokinetics, frequent concurrent use of medication, and higher prevalence of comorbidities, all contribute to an increased risk of such events. Faulkner et al. provide an extensive review of antimicrobial ADEs in the elderly . In the community, there is a risk of spread of Clostridium difficile infections (CDIs) and antimicrobial resistance.
4.1 Adverse Drug Events (ADEs)
In a cohort study evaluating antimicrobial-associated ADEs in hospitalized patients (N = 1488; median age 59 years), 20% of patients experienced at least one ADE, with gastrointestinal, renal, and hematologic abnormalities being the most frequent. Furthermore, 20% of the reported antimicrobial-related ADEs were due to unnecessary antimicrobial use . In another cohort evaluating ADEs in nursing-home residents, antimicrobials were the second most often reported drug class to cause an ADE (20%), after antipsychotics. The majority of observed ADEs were rashes and CDIs .
Certain ADEs are serious enough that elderly patients have to visit the emergency room (ER); 10.6% of elderly patient ER visits were due to an ADE, with antimicrobials being one of the most frequently implicated medication classes (16.7% of all ADEs, and 25% of definite or probable ADEs) . The most frequent (serious) ADEs in elderly patients who use β-lactam antimicrobials for LRTIs are drug fever, interstitial nephritis, and blood dyscrasias. For patients who use macrolides, the most frequent ADE is QT prolongation, while for patients taking trimethoprim-sulfamethoxazole, the most frequent ADEs are hyperkalemia, blood dyscrasias, and drug fever . Certain very rare ADEs, such as tendinitis and tendon rupture during fluoroquinolone use, are relatively more frequent in the elderly and can be further increased by concomitant glucocorticoid use or renal failure [61, 62], although the absolute risks remain low .
4.2 Drug–Drug Interactions
The elderly have an increased risk for DDIs since polypharmacy is more frequent . It is estimated that 51% (1380/2707) of elderly patients receive six or more drugs per day . The most serious DDIs in elderly patients using antimicrobials for LRTIs are trimethoprim-sulfamethoxazole in combination with vitamin K antagonists (increases anticoagulant effect), trimethoprim-sulfamethoxazole with potassium-sparing diuretics (risk of hospitalization due to hyperkalemia), and clarithromycin/erythromycin with drugs that are deactivated by cytochrome P450 3A4 enzymes (e.g. risk of rhabdomyolysis by increased concentrations of atorvastatin) .
4.3 Drug–Comorbidity Interactions
Macrolides are associated with an increased risk for cardiovascular events and deaths, especially in (elderly) patients with a higher baseline risk for cardiovascular events [64, 65, 66, 67, 68, 69]. These associations should caution the use of macrolides for LRTIs unless strictly necessary. Renal insufficiency is more frequent in the elderly, which results in a decreased elimination of certain (hydrophilic) antimicrobials (e.g. cephalosporins, fluoroquinolones) and increases the risk of ADEs. Therefore, a dose adjustment is recommended for patients with impaired renal function [9, 70].
4.4 Clostridium Difficile Infection
Broad-spectrum antimicrobials disturb the gastrointestinal flora, contributing to an increased risk of CDIs. Increased age, recent hospitalization, immune suppression, malignancy, chronic renal failure, and use of proton pump inhibitors have been identified as independent risk factors of CDIs and are highly prevalent in the elderly [70, 71]. Macrolides are more strongly associated with CDIs than doxycycline, and physicians may therefore choose the latter when atypical coverage is deemed necessary . Still, the risk of developing CDIs with macrolides appears smaller than with fluoroquinolones, clindamycin, or broad-spectrum β-lactams .
4.5 Antimicrobial Resistance
There is clear consensus that inappropriate antimicrobial use contributes to antimicrobial resistance, potentially leading to increased morbidity, mortality, and healthcare costs [73, 74]. Antimicrobial resistance rates may increase with age, as reported in a Canadian surveillance study, further increasing the risk for the elderly population . This could be explained by elderly people more often residing in LTCFs, needing hospitalization, receiving healthcare at home, and receiving antimicrobials, which are all risk factors for developing antimicrobial resistance .
5 Optimizing Appropriate Antimicrobial Use in Elderly Patients
5.1 Antimicrobial Stewardship Interventions
Antimicrobial stewardship interventions aim at reducing inappropriate use of antimicrobials while maintaining good clinical outcomes. Elderly patients might especially benefit from antimicrobial stewardship as they may have the highest risk for worse clinical outcomes due to both overtreatment (e.g. antimicrobial side effects, CDIs) or undertreatment (e.g. infectious complications). The risk for and possible harm due to treatment differs per patient and depends on patient factors such as comorbidities, immunological status, comedication, previous antibiotic treatment, recent hospitalizations, and the severity of the LRTI. Therefore, an individualized approach where individual patient risks are balanced with possible collateral damage in the form of selecting for antimicrobial resistance is recommended.
Commonly used antimicrobial stewardship interventions include periodic or individual patient audits and feedback, decision support, education (educational meetings, educational materials), and formulary restriction of antimicrobials [73, 76].
To date, the majority of studies on stewardship interventions are performed on adult or pediatric patients, or on specific hospital settings, e.g. intensive care unit (ICU). In a recent systematic review of stewardship interventions in hospitalized patients, only 1.8% (4/211) of studies specifically targeted elderly patients [73, 77, 78, 79, 80]. Two controlled before-after studies showed a reduction in 30-day mortality after introducing a pneumonia guideline with clinical decision support [77, 78]. Two interrupted time series showed a reduction in the incidence of CDIs after implementation of an audit and feedback program and a restrictive antimicrobial policy [79, 80]. Although promising, these four studies used non-randomized designs, risking confounding bias. There is a great need for high-quality studies in elderly patients. Currently, the effects of antimicrobial stewardship interventions in elderly patients may be over- or underestimated.
As stated earlier, the appropriate prescription of antimicrobials in LTCFs is often impeded by the frailty of elderly residents, limited clinical evaluation, sometimes only consisting of a nurse’s assessment and telephone supervision by a physician, and lack of diagnostic testing. In nursing homes, the majority of antimicrobials are prescribed after telephone contact with nursing staff, highlighting the importance of involving nurses in antimicrobial stewardship programs . In a recent review of stewardship interventions in LTCFs, the following approaches were identified to be the most effective: multidisciplinary education; pre-prescriptive data collection tools; post-prescriptive prescriber recommendations; and introducing consultation by infectious diseases experts . When designing antimicrobial stewardship interventions, it is important to consider the setting and corresponding prescription process. This was illustrated in nursing-home interventions designed to improve communication, which were only effective in reducing antimicrobial use when nursing staff were involved [83, 84]. Appealing antimicrobial stewardship targets to improve antimicrobial use for LRTIs in LTCFs include improving the indication for starting antimicrobials, optimizing the use of available diagnostics, limiting the use of fluoroquinolones given their strong association with CDIs, and ensuring proper dosing and duration of antimicrobial therapy [83, 85].
5.2 New Diagnostic Techniques
As infections can be difficult to diagnose in elderly patients, they may be initially treated with broad-spectrum antimicrobials. Novel diagnostic techniques to establish the diagnosis of LRTIs may particularly reduce unnecessary antimicrobials in elderly patients. The current cornerstone for the radiological diagnosis of pneumonia is the demonstration of an infiltrate by conventional chest X-ray; however, the estimated sensitivity of a chest X-ray is only 60–70% in patients with a clinical diagnosis of CAP [29, 86, 87]. In addition, certain comorbidities that are common in elderly patients (e.g. heart failure and chronic obstructive pulmonary disease) may impede the detection of pulmonary infiltrates . The increasing availability of computed tomography (CT) scans and the possibility of low- or ultra-low-dose scanning seems a promising alternative. Recently, CT scans changed the diagnosis of 58.6% (95% confidence interval 53.2–64.0%) of consecutive CAP patients, potentially leading to optimization of management in 25% of patients . These results need to be validated in clinical practice to demonstrate improved patient outcomes, while reducing antimicrobial use for non-infectious alternative diagnoses. In settings where CT scans are not readily available, chest ultrasonography may be a valuable alternative. It has high diagnostic accuracy for pneumonia, can be performed at the bedside, is highly reproducible in trained professionals, and the costs are relatively low .
5.2.2 Microbiological Testing
Established microbiological techniques to determine the causative pathogen include respiratory cultures or polymerase chain reaction (PCR), blood cultures, and urinary antigen tests for Legionella and pneumococcus; however, the sensitivity of these tests are limited and in 60–70% of patients suspected of CAP no causative pathogen will be identified [20, 29].
When viral CAP is suspected, point-of-care PCR (PoC-PCR) for respiratory viruses might play a role in optimizing antimicrobial therapy. Previous studies using regular respiratory PCR have failed to show an effect on antimicrobial use, possibly because it is difficult to stop antimicrobials after they have been administered for 1–2 days when PCR results become available [91, 92]. PoC-PCR may allow withholding or rapid discontinuation of antimicrobials if a viral pathogen is identified as results can be available in 1–2 h. The effects and cost effectiveness of PoC-PCR on antimicrobial use and patient outcome have not yet been investigated, but a cluster randomized clinical trial evaluating the clinical effects of both low-dose CT and PoC-PCR in patients with CAP admitted to a non-ICU ward is underway (ClinicalTrials.gov identifier: NCT01660204).
We have addressed several issues on the appropriate use of antimicrobials in elderly patients with LRTIs. As the microbial etiology is only slightly different compared with the younger population, the mainstay of treatment should consist of β-lactam monotherapy. Extended coverage of gram-negatives could be considered in LTCF or nursing-home residents. Quinolones or macrolides should be restricted to selected cases empirically, given the low incidence of atypical pathogens in elderly patients and higher risks of ADEs and DDIs. Next to these ADEs, inappropriate use of antimicrobials contributes to CDIs and antimicrobial resistance. A reference standard for measuring inappropriate antimicrobial use is currently lacking. Future work on definitions and standardization will hopefully increase validity and generalizability of reports on (in)appropriate antimicrobial therapy. Well-designed antimicrobial stewardship interventions could improve antimicrobial prescribing, but studies specifically targeting elderly patients are needed. These programs should generally consist of multiple components, depending on the specific clinical setting, such as improving diagnostics (i.e. indication for starting antimicrobials) and ensuring proper dosing and duration of therapy. We argue that new diagnostic techniques such as low-dose CT scanning or PoC-PCR testing for viral pathogens could potentially reduce inappropriate use of antimicrobials. Such techniques and interventions can hopefully decrease the inappropriate use of antimicrobials in the near future.
Compliance with Ethical Standards
This research received no specific grants from any funding agency in the public, commercial, or not-for-profit sectors.
Conflicts of interest
Inger van Heijl, Valentijn A. Schweitzer, Lufang Zhang, Paul D. van der Linden, Cornelius H. van Werkhoven, and Douwe F. Postma declare that they have no conflicts of interest.
- 1.United Nations, Department of Economic and Social Affairs, Population Division. World Population Ageing 2015. 2015 (ST/ESA/SER.A/390).Google Scholar
- 2.World Bank Group. Development Knowledge: Data on population ages 65 and above. 2016. https://data.worldbank.org/indicator/SP.POP.65UP.TO.ZS. Accessed 29 Nov 2017.
- 13.Sahuquillo Arce J, Menéndez R, Méndez R, Amara Elori I, Zalacain R, Capelastegui A, Aspa J, Borderías L, Martín Villasclaras J, Bello S, Alfageme I, de Castro F, Rello J, Molinos L, Ruiz Manzano J, Torres A. Age-related risk factors for bacterial aetiology in community-acquired pneumonia. Respirology. 2016. https://doi.org/10.1111/resp.12851.PubMedGoogle Scholar
- 14.Capelastegui A, España P, Bilbao A, Gamazo J, Medel F, Salgado J, Gorostiaga I, de Goicoechea L, Jose M, Gorordo I, Esteban C, Altube L, Quintana J. Etiology of community-acquired pneumonia in a population-based study: link between etiology and patients characteristics, process-of-care, clinical evolution and outcomes. BMC Infect Dis. 2012. https://doi.org/10.1186/1471-2334-12-134.
- 15.Gutiérrez F, Masiá M, Rodríguez JC, Mirete C, Soldán B, Padilla S, Hernández I, De Ory F, Royo G, Hidalgo AM. Epidemiology of community-acquired pneumonia in adult patients at the dawn of the 21st century: a prospective study on the Mediterranean coast of Spain. Clin Microbiol Infect. 2005. https://doi.org/10.1111/j.1469-0691.2005.01226.x.Google Scholar
- 21.Huijskens EGW, Koopmans M, Palmen FMH, van Erkel, Adriana JM, Mulder PGH, Rossen JWA. The value of signs and symptoms in differentiating between bacterial, viral and mixed aetiology in patients with community-acquired pneumonia. J Med Microbiol. 2014. https://doi.org/10.1099/jmm.0.067108-0.
- 43.Willemsen I, van der Kooij T, van Benthem B, Wille J, Kluytmans J. Appropriateness of antimicrobial therapy: a multicentre prevalence survey in the Netherlands, 2008–2009. Euro Surveill. 2010.Google Scholar
- 53.Baclet N, Ficheur G, Alfandari S, Ferret L, Senneville E, Chazard E, et al. Explicit definitions of potentially inappropriate prescriptions of antibiotics in older patients: a compilation derived from a systematic review. Int J Antimicrob Agents. 2017. https://doi.org/10.1016/j.ijantimicag.2017.08.011.PubMedGoogle Scholar
- 54.Sogn DD, Evans R, Shepherd GM, Casale TB, Condemi J, Greenberger PA, et al. Results of the National Institute of Allergy and Infectious Diseases Collaborative Clinical Trial to test the predictive value of skin testing with major and minor penicillin derivatives in hospitalized adults. Arch Intern Med. 1992;152(5):1025–32.CrossRefPubMedGoogle Scholar
- 56.Barlam T, Cosgrove S, Abbo L, MacDougall C, Schuetz A, Septimus E, et al. Implementing an antibiotic Stewardship Program: guidelines by the infectious diseases society of america and the society for healthcare epidemiology of America. Clin Infect Dis. 2016. https://doi.org/10.1093/cid/ciw118.PubMedCentralGoogle Scholar
- 61.van der Linden PD, Sturkenboom MCJM, Herings RMC, Leufkens HMG, Rowlands S, Stricker BHC. Increased risk of achilles tendon rupture with quinolone antibacterial use, especially in elderly patients taking oral corticosteroids. Arch Intern Med. 2003. https://doi.org/10.1001/archinte.163.15.1801.Google Scholar
- 72.UpToDate. Antimicrobial agents that may induce Clostridium difficile diarrhea and colitis. 2017. https://www.uptodate.com/contents/clostridium-difficile-infection-in-adultstreatment#subscribeMessage. Accessed 23 Dec 2017.
- 74.Dellit T, Owens R, McGowan J, Gerding D, Weinstein R, Burke J, et al. Infectious Diseases Society of America and the Society for Healthcare Epidemiology of America guidelines for developing an institutional program to enhance antimicrobial stewardship. Clin Infect Dis. 2007. https://doi.org/10.1086/510393.PubMedGoogle Scholar
- 91.Oosterheert J, van Loon A, Schuurman R, Hoepelman AIM, Hak E, Thijsen S, et al. Impact of rapid detection of viral and atypical bacterial pathogens by real-time polymerase chain reaction for patients with lower respiratory tract infection. Clin Infect Dis. 2005. https://doi.org/10.1086/497134.PubMedGoogle Scholar
- 92.Hernes SS, Hagen E, Quarsten H, Bjorvatn B, Bakke PS. No impact of early real-time PCR screening for respiratory viruses on length of stay and use of antibiotics in elderly patients hospitalized with symptoms of a respiratory tract infection in a single center in Norway. Eur J Clin Microbiol Infect Dis. 2014. https://doi.org/10.1007/s10096-013-1963-0.PubMedGoogle Scholar
- 93.Tobia C, Aspinall S, Good C, Fine M, Hanlon J. Appropriateness of antibiotic prescribing in veterans with community-acquired pneumonia, sinusitis, or acute exacerbations of chronic bronchitis: a cross-sectional study. Clin Ther. 2008. https://doi.org/10.1016/j.clinthera.2008.06.009.PubMedPubMedCentralGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.