Prescribing of antibiotics for upper respiratory tract infections (URTI) varies substantially in primary care.
To describe and compare antibiotic and antipyretic/anti-inflammatory drugs use, URTI symptoms' resolution and occurrence of potentially-associated infections in patients seeking care from general practitioners (GPs) who exclusively prescribe conventional medications (GP-CM), regularly prescribe homeopathy within a mixed practice (GP-Mx), or are certified homeopathic GPs (GP-Ho).
The EPI3 survey was a nationwide population-based study of a representative sample of 825 GPs and their patients in France (2007–2008). GP recruitment was stratified by self-declared homeopathic prescribing preferences. Adults and children with confirmed URTI were asked to participate in a standardized telephone interview at inclusion, one-, three- and twelve-month follow up. Study outcomes included medication consumption, URTI symptoms' resolution and potentially-associated infections (sinusitis or otitis media/externa) as reported by patients. Analyses included calibration to account for non-respondents and groups were compared using multivate analyses adjusting for baseline differences with a propensity score.
518 adults and children with URTI (79.3% rhinopharyngitis) were included (36.9% response rate comparable between groups). As opposed to GP-CM patients, patients in the GP-Ho group showed significantly lower consumption of antibiotics (Odds ratio (OR) = 0.43, 95% confidence interval (CI): 0.27–0.68) and antipyretic/anti-inflammatory drugs (OR = 0.54, 95% CI: 0.38–0.76) with similar evolution in related symptoms (OR = 1.16, 95% CI: 0.64–2.10). An excess of potentially-associated infections (OR = 1.70, 95% CI: 0.90–3.20) was observed in the GP-Ho group (not statistically significant). No difference was found between GP-CM and GP-Mx patients.
Patients who chose to consult GPs certified in homeopathy used less antibiotics and antipyretic/anti-inflammatory drugs for URTI than those seen by GPs prescribing conventional medications. No difference was observed in patients consulting GPs within mixed-practice. A non-statistically significant excess was estimated through modelling for associated infections in the GP-Ho group and needs to be further studied.
Citation: Grimaldi-Bensouda L, Bégaud B, Rossignol M, Avouac B, Lert F, Rouillon F, et al. (2014) Management of Upper Respiratory Tract Infections by Different Medical Practices, Including Homeopathy, and Consumption of Antibiotics in Primary Care: The EPI3 Cohort Study in France 2007–2008. PLoS ONE 9(3): e89990. https://doi.org/10.1371/journal.pone.0089990
Editor: C. Mary Schooling, CUNY, United States of America
Received: January 16, 2013; Accepted: January 28, 2014; Published: March 19, 2014
Copyright: © 2014 Grimaldi-Bensouda et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This study was funded by Laboratoires Boiron, France. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have the following interests: LGB, BA and MR are employees of LA-SER, the company conducting the study. LG-B was the recipient of a research fellowship from INSERM (French National Institute of Health and Medical Research) at the time of the study. LA is a stockholder in LA-SER. GD is employed by Cyklad Group. FR and DG received a consulting fee or honorarium from LA-SER for the submitted work. This study was funded by Laboratoires Boiron, France. There are no patents, products in development or marketed products to declare. This does not alter the authors' adherence to all the PLoS ONE policies on sharing data and materials.
High antibiotic consumption is said to be associated with the emergence and dissemination of multi-resistant bacteria in the community . Demands and expectations for antibiotics in common upper respiratory tract infections [URTI]) are important drivers of antibiotic overprescribing in primary care . Many countries have initiated programs targeted at physicians and the general public to reduce antibiotic prescribing . Most evaluated programs have recorded some success even though the effect on resistance to antimicrobial drugs, and particularly on dissemination of antibiotic-resistant pneumococci, remains uncertain . Substantial heterogeneity in antibiotic prescribing among French GPs has been observed . Despite the modest decrease in ambulatory antibiotic prescribing for respiratory tract infections between 2001 and 2009, France remains a country with one of the highest antibiotic consumption rates in Europe , .
While there is evidence that homeopathy has little effect on URTI or flu-like symptoms , , its potential for reducing antibiotic consumption has been proposed . In France, homeopathic medicines are partially reimbursed by the National Health Insurance and are prescribed exclusively by a physician. Besides, patients must choose a ‘treating physician’, who will be responsible for follow-up and referral to specialists. This treating physician may be a physician specializing in homeopathy. This context provided a unique opportunity to observe homeopathic prescribing practices in the management of patients with URTI in primary care.
The objectives of this one-year population-based cohort study was to describe and compare antibiotic and antipyretic/anti-inflammatory drugs use, resolution of URTI symptoms and occurrence of potentially associated infections in patients who seek care for URTI from general practitioners (GPs) showing different prescribing preferences for homeopathy: strictly prescribers of conventional medications reluctant to prescribe homeopathic medicines (GP-CM), regular prescribers of homeopathic medicines in an otherwise conventional medical practice (“mixed prescribing” or GP-Mx), and certified homeopathic GPs (GP-Ho), who also prescribe conventional medications.
Study design and population
The EPI3 survey was a nationwide survey of primary care practice conducted in a representative sample of GPs from across France and their patients between 2007 and 2008 . The sample was drawn using a two-stage sampling process. First, a random sample of GPs was drawn from the French national directory of physicians in primary care. Sampling of GPs was stratified according to self-declaration of prescribing preferences obtained by telephone at the time of recruitment and categorized into three groups: strictly prescribers of conventional medications (GP-CM) who declared never using homeopathy, or only at the patient's request; regular prescribers of homeopathic medicines within a mixed prescribing practice (GP-Mx); and certified homeopathic GPs (GP-Ho). Second, a one-day survey of all patients attending the medical practice of each participating GP was conducted where a trained research assistant surveyed all patients in the waiting room. For this cohort study, the first 5 (minimum) to 15 (maximum) consenting adult patients and guardians of children were invited to participate if (a) the attending physician was declared by patients as their regular physician (designated “treating physician” according to French regulation) and (b) one of the clinical diagnosis declared by the physician at that visit included one of the following ICD-9 (International Classification of Diseases 9th revision)  codes: acute nasopharyngitis [common cold] (code 460), acute upper respiratory infections of multiple or unspecified sites (code 465); acute bronchitis and bronchiolitis or bronchitis, not otherwise specified (codes 466 and 490, respectively), acute pharyngitis (code 462) and acute laryngitis and tracheitis (code 464).
At inclusion, GPs completed a medical questionnaire for each patient included in the cohort with the main reason diagnosis, a standardized history of respiratory diagnoses in the previous year and of respiratory symptoms in the current episode of URTI, up to five other diagnoses (comorbidities) and all drugs prescribed that day. Diagnoses were coded according to the ICD-9 classification by a trained research assistant. All consenting patients completed a self-administered questionnaire at inclusion, in the waiting room, collecting information on lifestyle and history of medical consultations and hospitalizations in the previous year. The follow-up telephone interview at one month included the inventory of URTI symptoms obtained via patients' self-assessment of changes in those symptoms from baseline (cleared, much improved, slightly improved, no change or worsened). Interviews at one, three and twelve months spanned the patient's history since the previous interview with regard to the occurrence of infections associated with the URTI, defined as patients' self-report of a diagnosis (with or without treatment) of otitis (media or externa) and/or sinusitis, and any drug consumption (conventional and homeopathy). This calendar was used to aid patients' recall during the one-year follow-up. Drug consumption, whether prescribed or obtained over-the-counter or from the family pharmacy, was assessed using a standardized method named Progressive Assisted Backward Active Recall (PABAR) previously validated against medical prescriptions , . Briefly, patients received at the time of their recruitment a booklet detailing the interview, including a list of commonly used drugs for URTIs, and were instructed to collect all their prescriptions. Trained interviewers helped patients recall past exposure to drugs, starting with recent history and progressing back in time to identify events at key dates. Drugs were automatically recorded using the anatomical therapeutic chemical classification index (ATC), 2009 revision. Particular emphasis was put on antibiotics [antibacterial for systemic use (class J01), neuraminidase inhibitors (class J05AH)], antipyretics [acetaminophen (class N02BE01), acetylsalicylic acid (class N02BA01)] and non-steroidal anti-inflammatory drugs (NSAIDs) [ibuprofen (class M01AE01)], as well as homeopathic drugs commonly used in URTI. Patients were asked to specifically report consumption of any drug from a list of 41 products after they had spontaneously reported all drugs used. Patients also reported the occurrence of diagnoses (with or without treatment) of otitis (media or externa) and/or sinusitis. Those two diagnoses were used as proxies for potentially associated infections.
Four outcomes were evaluated as yes/no responses (Table 1). Consumption of antibiotics and antipyretic/anti-inflammatory drugs for URTI was defined at each interview interval as the proportion of patients who declared using at least one drug (since the previous interview) from any of the ATC classes listed above. Utilization of antibiotics and antipyretic/anti-inflammatory drugs was then defined as at least one utilization for URTI at any of the one, three or twelve-month interviews. Resolution of the URTI was defined following either patients' self-report of complete resolution or significant improvement of baseline symptoms at the one-month interview. Infections potentially associated to the URTI were defined following patients' self-report of at least one declaration of a diagnosis of otitis media, otitis externa or sinusitis (with or without treatment) at any of the one, three or twelve-month interviews.
Participants and nonparticipants in the cohort study were compared using information collected from all surveyed patients at baseline. Characteristics of patients not participating in the survey (gender, age and main reasons for consultation) were used to calibrate the final sample using a method known as the CALMAR procedure . In this method, differences between participants and nonparticipants were compensated by attributing a specific weight to each patient in the analysis, inversely proportional to the participation rate of similar patients at baseline among all patients surveyed. Differences at baseline between GP-CM, GP-Mx and GP-Ho groups were estimated using multivariate logistic regression analyses. A propensity score was computed for each participant in the study indicating their probability of belonging to either GP-Mx or GP-Ho groups compared to the GP-CM group according to all variables listed in Table 2. The score was intended to adjust for confounding by differences between the groups in all subsequent analyses.
The three groups were compared for the four binary outcomes using the GP-CM group as the reference group using logistic regression analyses adjusted for baseline characteristics (propensity score) and the number of URTI symptoms at baseline (as listed in Table 3). Clustering effects resulting from recruiting several patients consulting the same GP and autocorrelation between responses to the four consecutive interviews were controlled using Generalized Estimating Equations (GEE) in the multivariate models. Sample size was estimated for the EPI3 survey as a whole so as to provide accurate estimates of prevalence (at ±2%) for each group of diagnoses seen in primary care, including URTI. All analyses were performed using SAS version 9.1 (SAS Institute, Inc., Cary, North Carolina).
The study was approved by the French National Data-Protection Commission (CNIL) and the French National Medical Council (CNOM). In accordance with CNIL regulation, written consent was obtained from each participating adult patient and from one of the parents accompanying each participating child. Participating physicians received compensation fees for their participation but not patients.
The general EPI3 health survey included 825 GPs and 8,559 participants (ten patients per GP on average, minimum 5, maximum 15). Of the latter, 1,402 children and adults fulfilled the specific inclusion criteria for the URTI cohort, of which 699 (49.9%) agreed to participate with 518 (36.9%) responding to all three follow-up interviews and therefore included in the analysis. Participants were slightly more often females (59.8%) compared to nonparticipants (54.2%), more likely to belong to the 50+ years age group (24.5% and 19.8%, respectively), to have completed high school education (57.1% and 53.4%), and less likely to be a current smoker (29.7% and 39.0%, respectively), all differences statistically significant (Table 2). Of participants under 20 years of age (41.9%), two thirds (28.1%) were 6 years old or younger. Among participants, patients who consulted a GP-Ho were more often non-smoking females who completed high school education compared to the GP-CM group, differences that were statistically significant after taking into account all other factors, and were similar otherwise (Table 2). Patients in the GP-Mx group were comparable to the GP-CM group.
With regard to types of URTI at baseline, there were little differences between the three groups. The most commonly reported was rhinopharyngitis (73.9%), followed with bronchitis (28.0%), flu-like symptoms (12.7%), Strep-A negative viral angina (8.7%) and bronchiolitis (5.2%) (Table 3). For symptoms reported by patients, those who consulted a GP-Ho were less likely to have fever (37.3%), nasal obstruction (52.7%) and cough (64.0%) than those consulting a GP-CM (53.3%, 63.0% and 76.4%, respectively; all three statistically significant). Patients in the GP-Mx were comparable to those in the GP-CM group.
Prescribing preferences of physicians in the three groups were confirmed at baseline by their respective prescribing rates of homeopathic drugs, which were 0.6%, 9.4% and 61.3%, respectively, in the GP-CM, GP-Mx and GP-Ho groups (Table 3). Conversely, antibiotic and antipyretic/anti-inflammatory drug prescriptions, which were relatively comparable between the GP-CM and the GP-Mx groups with rates above 40%, were much lower in the GP-Ho group with rates at or below 20%.
Table 1 shows the multivariate analyses on drug utilization, resolution of the URTI symptoms and occurrence of potentially associated infections. For antibiotic and antipyretic/anti-inflammatory drug utilization, the lower consumption observed at baseline in the GP-Ho group was maintained over the twelve-month follow-up with an adjusted probability of patients' reporting significantly lower than the GP-CM group both for antibiotics (OR = 0.43; 95% CI: 0.27–0.68) and for antipyretic/anti-inflammatory drugs (OR = 0.54; 95% CI: 0.38–0.76). There was no difference in drug utilization between the groups GP-Mx and GP-CM.
For resolution of the URTI symptoms at one-month follow-up, patients declared similar improvement in the three groups (77.0%, 81.1% and 77.7% respectively in the GP-CM, GP-Mx and GP-Ho groups) with no statistically significant differences in multivariate analyses adjusting for baseline characteristics (OR = 1.16; 95% CI: 0.64–2.10). On the other hand, there was a slightly higher rate of potentially associated infections (otitis media/externa and sinusitis) reported by patients in the GP-Ho group over twelve months follow-up (17.7%) compared to the GP-CM group (16.9%), a difference that was not statistically significant (OR = 1.71; 95% CI: 0.90–3.20). Results on symptoms resolution and occurrence of potentially associated infections exhibited wide confidence intervals.
This population-based prospective cohort study described and compared clinical management and evolution of patients consulting for URTI between three groups of physicians with different levels of prescribing preferences for homeopathy. At baseline, patients who chose to be seen by GP-Ho for URTI declared to have used half the amount of antibiotics and antipyretic/anti-inflammatory drugs compared to patients seen by conventional medicine practitioners. This lower consumption of conventional medications in the GP-Ho group was sustained over the 12-month follow-up. At the same time, no difference in the resolution of the URTI symptoms was observed between groups but confidence intervals were wide indicating lack of statistical power for that outcome. Similarly, the excess rate of potentially associated infections observed in the GP-Ho group, although non-statistically significant, cannot be ruled out. No difference was seen in patients from the GP-Mx group, which was comparable to the GP-CM group on all outcomes.
Previous observational studies conducted in several countries have shown an antibiotic-sparing effect resulting from management by GPs using homeopathy without increase in complication rates of URTI , . Patients' education, including appropriate indication for antibiotic use, infection prognosis, and alternative treatment recommendations, may contribute to lower patients' expectations toward antibiotics while improving satisfaction . This has been described in France during the 2009–2010 influenza season . Authors have pointed out the difficulty of sorting out patients' expectations/motivation and homeopathic care itself, including their providers , .
The rise in bacterial resistance to antibiotics is widely recognized as a major threat to public health , . Antibiotic prescribing for URTI varies widely within and across countries ,  suggesting that further control of antibiotic prescribing is possible. Many countries have implemented policies aimed at reducing inappropriate prescribing of antimicrobials in primary care , . In that context, our results are not unexpected and can contribute to reinforce the motivation of decision makers to pursue these policies –.
Our results could be explained in part by the different characteristics of patients seen by GPs who practice homeopathy and by the lower rate of fever, nasal obstruction and cough in the GP-Ho group at baseline compared to the two other groups. Adjustment by severity of URTI and other potential confounders did not alter the results but residual confounding cannot be excluded. As for our observation of a small non-statistically significant excess in the occurrence of potentially associated infections in the GP-Ho group, it may be due to chance or to a lack of protection against these infections. The latter instance cannot be ruled out as the study lacked statistical power to distinguish between the two interpretations.
The participation rate in this URTI cohort study was only 36.9% of eligible patients, which is comparatively equivalent to what is seen in general health surveys where patients are asked to participate in a long follow-up . Given that this study was appended to a general population health survey, contributed at reducing the risk of selection bias of physicians and patients. The overall prevalence of URTI in this survey (12.4%; 95% CI: 11.5–13.2) was compatible with statistics on GP consultations in France . Precautions taken to calibrate the final sample so as to ensure representatively of the eligible population contributed at reducing sampling bias but without ruling it out entirely. The results may also be subject to residual confounding because the propensity score may have not accounted for all the differences between patients who seek treatment from different types of physicians.
Another potential limitation is related to the nature of URTI diagnoses that have not been validated against a disease management guideline. No such attempt was made to preserve the authenticity of primary care practice in real life. This is partially why diagnoses of bronchitis and bronchiolitis were included in this cohort as they may represent co-occurrences of URTI. The standardized collection of symptoms allowed a partial control for severity of URTI at inclusion. Two conditions, sinusitis and otitis, were studied as proxies for the occurrence of infections potentially associated to the URTI. Diagnoses were obtained from patients' self-declaration over the telephone and should not be interpreted strictly. It is not known whether they represent true complications or URTI and/or represent associated infections as a result of no antibiotic treatment. This should be studied, particularly in view of the apparent excess of infections observed in the GP-Ho group. However, the lack of diagnostic confirmation should not bias the comparison between the groups but may bias the results toward the null and thus reducing the statistical significance of the observation. In view of the different characteristics of patients in the GP-Ho group at inclusion, the lower frequency of symptoms reported that group might be explained by a lower threshold of these patients to consult a physician rather than a true difference in the diagnoses makeup of the group.
Finally, results on resolution of URTI symptoms were underpowered to show non-inferiority between groups as illustrated by the wide confidence intervals. The estimates however were close to unity in both GP-Mx and GP-Ho groups, indicating similarity for self-declaration of symptoms resolution at one month between patients from both groups. Sample size was sufficient to show an Odds ratio superior to 1.22 (or inferior to 0.82) for the main outcomes. Strengths of this study included the length of follow-up and the quality of the data which combined medical and patient information collected from physicians and patients. Drug consumption was obtained from patients interviews using a validated approach ,  that allowed the identification of prescription drugs as well as those obtained over-the-counter or from the family pharmacy, the latter being known to be an important source of self-treatment for URTI .
In conclusion, this cohort study showed that patients with URTI who choose to consult homeopathy-certified GPs in primary care, had a lower consumption of antibiotics and antipyretic/anti-inflammatory drugs as compared to patients seen by physicians who use conventional medicine. This difference may be due to specific attributes of either physicians or patients but also interactions between the two. No difference was observed for patients consulting GPs with mixed prescribing habits. The non-significant excess of potentially associated infections in the GP-Ho group esteemed through modelling may be due to chance alone or driven by less use antibiotics. Further studies are needed to clarify this effect. Other large studies are needed to establish the long-term consequences of different prescribing practices in primary care.
Conceived and designed the experiments: LG-B BB MR BA FL FR JB JM GD A-MM LA DG. Performed the experiments: LG-B. Analyzed the data: LG-B MR DG. Wrote the paper: MR DG. Developed and approved the study protocol and the analyses plan: LG-B BB MR BA FL FR JB JM GD A-MM LA DG. Discussed and interpreted the results: LG-B BB MR BA FL FR JB JM GD A-MM LA DG. Revised the manuscript: LG-B BB MR BA FL FR JB JM GD A-MM LA DG.
- 1. van de Sande-Bruinsma N, Grundmann H, Verloo D, Tiemersma E, Monen J, et al. (2008) Antimicrobial drug use and resistance in Europe. Emerg Infect Dis 14: 1722–1730 DOI: 10.3201/eid1411.070467.
- 2. Filipetto FA, Modi DS, Weiss LB, Ciervo CA (2008) Patient knowledge and perception of upper respiratory infections, antibiotic indications and resistance. Patient Prefer Adherence 2: 35–39 DOI http://dx.doi.org/10.2147/PPA.S.
- 3. Davey P, Sneddon J, Nathwani D (2010) Overview of strategies for overcoming the challenge of antimicrobial resistance. Expert Rev Clin Pharmacol 3: 667–686 DOI: 10.1586/ecp.10.46.
- 4. Huttner B, Goossens H, Verheij T, Harbarth S (2010) CHAMP consortium (2010) Characteristics and outcomes of public campaigns aimed at improving the use of antibiotics in outpatients in high-income countries. Lancet Infect Dis 10: 17–31 DOI: 10.1016/S1473-3099(09)70305-6.
- 5. Pulcini C, Lions C, Ventelou B, Verger P (2013) Drug-specific quality indicators assessing outpatient antibiotic use among French general practitioners. Eur J Public Health 23: 262–264 DOI: 10.1093/eurpub/cks100.
- 6. Chahwakilian P, Huttner B, Schlemmer B, Harbarth S (2011) Impact of the French campaign to reduce inappropriate ambulatory antibiotic use on the prescription and consultation rates for respiratory tract infections. J Antimicrob Chemother 66: 2872–2879 DOI: 10.1093/jac/dkr387.
- 7. Dommergues MA, Hentgen V (2012) Decreased paediatric antibiotic consumption in France between 2000 and 2010. Scand J Infect Dis 44: 495–501 DOI: 10.3109/00365548.2012.669840.
- 8. Altunç U, Pittler MH, Ernst E (2007) Homeopathy for childhood and adolescence ailments: systematic review of randomized clinical trials. Mayo Clin Proc 82: 69–75.
- 9. Kirkby R, Herscu P (2010) Homeopathic trial design in influenza treatment. Homeopathy 99: 69–75 DOI: 10.1016/j.homp.2009.09.001.
- 10. Trichard M, Chaufferin G, Nicoloyannis N (2005) Pharmacoeconomic comparison between homeopathic and antibiotic treatment strategies in recurrent acute rhinopharyngitis in children. Homeopathy 94: 3–9.
- 11. Grimaldi-Bensouda L, Begaud B, Lert F, Rouillon F, Massol J, et al. (2011) Benchmarking the burden of 100 diseases: results of a nationwide representative survey within general practices. BMJ Open 1: e000215 DOI: 10.1136/bmjopen-2011-000215.
- 12. World Health Organization (1977) Statistical classification of diseases and causes of death, 9th revision. Geneva: World Health Organization.
- 13. Grimaldi-Bensouda L, Rossignol M, Aubrun E, Benichou J, Abenhaim L, et al. (2012) Agreement between patients' self-report and physicians' prescriptions on nonsteroidal anti-inflammatory drugs and other drugs used in musculoskeletal disorders: the international Pharmacoepidemiologic General Research eXtension database. Pharmacoepidemiol Drug Saf 21: 753–759 DOI: 10.1002/pds.3194.
- 14. Grimaldi-Bensouda L, Rossignol M, Aubrun E, El Kerri N, Benichou J, et al. (2010) Agreement between patients' self-report and physicians' prescriptions on cardiovascular drug exposure: the PGRx database experience. Pharmacoepidemiol Drug Saf 19: 591–595 DOI: 10.1002/pds.1952.
- 15. Deville JC, Sarndal CE (1992) Calibration estimators in survey sampling. Journal of the American Statistical Association 87: 376–382.
- 16. Haidvogl M, Riley DS, Heger M, Brien S, Jong M, et al. (2007) Homeopathic and conventional treatment for acute respiratory and ear complaints: A comparative study on outcome in the primary care setting. BMC Complement Altern Med 7: 7 DOI:10.1186/1472-6882-7-7.
- 17. Vincent S, Demonceaux A, Deswarte D, Scimeca D, Bordet MF (2013) Management of influenza-like illness by homeopathic and allopathic general practitioners in France during the 2009–2010 influenza season. J Altern Complement Med 19: 146–152 DOI: 10.1089/acm.2011.0706.
- 18. Williams-Piehota PA, Sirois FM, Bann CM, Isenberg KB, Walsh EG (2011) Agents of change: how do complementary and alternative medicine providers play a role in health behavior change? Altern Ther Health Med 17: 22–30.
- 19. Welschen I, Kuyvenhoven M, Hoes A, Verheij T (2004) Antibiotics for acute respiratory tract symptoms: patients' expectations, GPs' management and patient satisfaction. Fam Pract 21: 234–237 DOI: 10.1093/fampra/cmh303.
- 20. Costelloe C, Metcalfe C, Lovering A, Mant D, Hay AD (2010) Effect of antibiotic prescribing in primary care on antimicrobial resistance in individual patients: systematic review and meta-analysis. BMJ 340: c2096 DOI: 10.1136/bmj.c2096.
- 21. Ashworth M, Charlton J, Ballard K, Latinovic R, Gulliford M (2005) Variations in antibiotic prescribing and consultation rates for acute respiratory infection in UK general practices 1995–2000. Br J Gen Pract 55: 603–608.
- 22. Ferech M, Coenen S, Malhotra-Kumar S, Dvorakova K, Hendrickx E, et al. (2006) European Surveillance of Antimicrobial Consumption (ESAC): outpatient quinolone use in Europe. J Antimicrob Chemother 58: 423–427 DOI: 10.1093/jac/dkl183.
- 23. Rossi E, Crudeli L, Endrizzi C, Garibaldi D (2009) Cost-benefit evaluation of homeopathic versus conventional therapy in respiratory diseases. Homeopathy 98: 2–10 DOI: 10.1016/j.homp.2008.11.005.
- 24. Jain A (2003) Does homeopathy reduce the cost of conventional drug prescribing? A study of comparative prescribing costs in General Practice. Homeopathy 92: 71–76.
- 25. Kooreman P, Baars EW (2011) Patients whose GP knows complementary medicine tend to have lower costs and live longer. Eur J Health Econ 13: 1–8 DOI: 10.1007/s10198-011-0330-2.
- 26. Labarthe G (2004) Les consultations et visites des médecins généralistes. Un essai de typologie. DREES Etudes et Résultats (315) 1–11.
- 27. Smith SM, Schroeder K, Fahey T (2012) Over-the-counter (OTC) medications for acute cough in children and adults in ambulatory settings. Cochrane Database Syst Rev 8: CD001831 DOI: 10.1002/14651858.CD001831.pub4.