"What they see is what you get": Prescribing antibiotics for respiratory tract infections in primary care: Do high prescribers diagnose differently? An analysis of German routine data
"What they see is what you get": Prescribing antibiotics for respiratory tract infections in primary care: Do high prescribers diagnose differently? An analysis of German routine data
Susann Hueber 0 1
Thomas Kuehlein 0 1
Roman Gerlach 1
Martin Tauscher 1
Angela Schedlbauer 0 1
0 Universit aÈtsklinikum Erlangen, Institute of General Practice , Erlangen, Germany, 2 Kassen aÈrztliche Vereinigung Bayern, M uÈnchen , Germany
1 Editor: Chaojie Liu, La Trobe University , AUSTRALIA
Characteristics of high and low prescribers of antibiotics in German primary care were analysed using population data. We aimed to evaluate differences in prescribing rates and factors being associated with high prescribing, and whether high prescribers made the diagnosis of perceived bacterial infections more often.
Data Availability Statement: Data were obtained
from the Bavarian Association of Statutory Health
Insurance Physicians (KassenaÈrztliche Vereinigung
Bayerns, KVB). These data are personal data and
due to legal reasons these data are not publicly
available. Data are maintained in the KVB. For
research use, data will be anonymized in
accordance with data protection laws. Via remote
access, those data can be provided by the KVB.
Every interested researcher is able to apply for and
Routine data were provided by the Bavarian Association of Statutory Health Insurance Phy
sicians. Routine data are delivered by primary care practices on a quarterly basis. We
analysed data from 2011 and 2012. Patients older than 15 years with respiratory tract infections
consulting a primary care physician were selected (6.647 primary care practices). Patient
and physician characteristics associated with high prescribing were identified using
stepwise logistic regression.
Mean prescribing rate of antibiotics was 24.9%. Prescribing rate for high prescribers was
43.5% compared to 8.5% for low prescribers. High prescribers made the diagnosis of
perceived bacterial infections more often (Mhigh = 64.5%, Mlow = 45.2%). In the adjusted
regression model, perceived bacterial infections were strongly associated with high prescribing
(OR = 13.9, 95% CI [10.2, 18.8]). Treating patients with comorbidities was associated with
lower prescribing of antibiotics (OR = 0.6, 95% CI [0.4, 0.8]). High prescribers had a higher
practice volume, a higher degree of prescribing dominance, and were situated more often in
deprived areas and in rural settings.
obtain the data in the same way as we did. For
further information please see: https://www.kvb.de/
Abteilung GeschaÈftsfuÈhrung Strategie, General
Management and Strategy, Email to
, Fax: +49 89
Funding: The authors received no specific funding
for this study.
Competing interests: The authors have declared
that no competing interests exist.
Compared to findings of studies in other European countries, prescribing rates were low.
There was a considerable difference between prescribing rates of high and low prescribers.
Diagnostic labelling was the best predictor for high prescribing. Current guidelines recommend considering antibiotic treatment for patients with co-morbidities. In our study, treating a large number of high-risk patients was not associated with high prescribing.
Antibiotics are still overprescribed for respiratory tract infections (RTI) [
]. Most RTI are of
viral origin and antibiotics are rarely indicated. A recent analysis of routine data of German
ambulatory care showed that antibiotics were prescribed for 30% of all patients with
respiratory tract infections consulting a GP [
]. In outpatient care, clinical signs are often unspecific
causing considerable diagnostic uncertainty in the differentiation between viral and bacterial
infections. Due to the inevitable uncertainty of the diagnosis, guidelines recommend a more
generous indication of antibiotic treatment for older patients and patients with comorbidities
. Clinical factors, as well as patients' and physicians' characteristics influence antibiotic
prescribing. Poor general health seems to be associated with higher prescribing, whereas age
under 60 is associated with lower prescribing . Prescribing rates increase with physician's
age and time in practice [5±7], high practice volume [6, 8±10], rural practice location , low
population density  and deprivation of the catchment area of the practice . Prescribing
rates generally differ widely between physicians, with studies suggesting a considerable
influence of personal overall preference on prescribing behaviour [14±16]. Patients diagnosed with
acute bronchitis were much more likely to receive an antibiotic compared to patients
diagnosed with common cold . Despite the fact that acute bronchitis is mainly of viral origin, it
might be that the diagnosis is perceived as an illness with a potential to develop into
pneumonia. Diagnoses with a potential bacterial cause such as acute sinusitis or bronchitis may serve
as a false justification for antibiotic prescribing and were used by high prescribers more often
[18±21]. We will call these diagnoses perceived bacterial infections in the sections to follow.
In our study, characteristics of high and low prescribers of antibiotics in German primary
care were evaluated. Large population data derived from routine data of ambulatory care were
examined. Specific objectives were: to examine how much high and low prescribers differed in
their antibiotic prescribing rates, whether high prescribers made the diagnosis of perceived
bacterial infections more frequently and to identify physician and patient characteristics being
associated with high prescribing.
Materials and methods
Approval was granted by the Ethics Committee of the Faculty of Medicine of the
FriedrichAlexander University Erlangen-NuÈrnberg (218_14 B).
Claims data and criteria on data selection
To understand the circumstances around data collection, it is important to consider the
characteristics of the German health care system: free provider choice and unregulated access to
health care. Patients have free access to both primary care physicians (PCP) and office based
2 / 11
specialists. Consulting more than one PCP and specialists in parallel is possible. Therefore
patients can receive prescriptions from primary care physicians but also from office-based
specialists in parallel. In rural areas, where there are only few specialists, most prescribing for a
given patient is done by the PCP. In urban setting much of the prescribing is taken over by
specialists. The so-called prescribing dominance is an indicator for the domination of prescribing
activities taking place in primary care and allows evaluating to which degree a primary care
practice is the main prescriber for its patients. It is defined as the proportion of prescriptions
issued by a specific practice divided by all prescriptions being issued for the same patient
Claims data were provided by the Bavarian Association of Statutory Health Insurance
Physicians (KassenaÈrztliche Vereinigung Bayerns, KVB). In Germany, physicians accredited with
statutory health insurances send their reimbursement claims for provided ambulatory medical
services to their corresponding regional KV. Data are delivered on a quarterly basis and do not
contain information on a day to day basis, not allowing a direct link between prescriptions and
The provided data contained an anonymous unique patient identifier, patient's age in
fiveyear intervals, sex and diagnoses encoded according to the International Classification of
Diseases (ICD-10-GM). Unlike the international version, the German modification of ICD-10
allows the doctors to add the strength of diagnostic reasoning such as suspected, assured,
excluded or sequelae. The ICD-10 does not allow for a meaningful aggregation of its classes for
our purposes. Therefore, diagnoses in the data set were transformed and grouped following
the International Classification of Primary Care (ICPC 2-R, ), a classification system
related to the World Health Organization (WHO)±International Family of Classifications. On
top of being an adequate classification for the domain of primary care in itself, the ICPC-2, via
an official mapping between the two classifications, allows for a meaningful aggregation based
on ICD-10 into ICPC-2 codes.
The data set revealed details about physicians' age (in five-year intervals) and sex,
practice location (rural, urban, large city and administrative district), practice type
(singlehanded or group practice, number of physicians per surgery) and specialist training (trained
GP, specialist in internal medicine working as primary care physician or physician without
specialist training; hereinafter all of them together called primary care physicians (PCP)).
Further information was provided on practice volume per quarter, regional deprivation
coded as Bavarian Index of Multiple Deprivation (BIMD, ) and on degree of prescribing
dominance. Patients' and physicians' data were linked to all prescriptions issued by a
specific surgery. The origin of prescriptions cannot be identified beyond the practice level. In
case of group practices, linking prescription data to PCPs' individual data, such as age or
sex, is not possible. Details on medication are encoded following the Anatomical
Therapeutic Chemical (ATC) classification .
Data on physicians and patients were anonymized. German law allows for analysing
anonymized patient data for scientific issues without formal consent of the patients. All relevant
data protection requirements were taken into consideration. Data captured in 2011 and 2012
were analysed (eight accounting quarters). The following data were provided: (1) Patients with
at least one of the following ICPC-diagnoses in one quarter: R72 (strep throat), R74 (acute
upper respiratory infection), R75 (acute/chronic sinusitis), R76 (acute tonsillitis), R77 (acute
laryngitis/tracheitis), R78 (acute bronchitis/bronchiolitis), R80 (influenza), R81 (pneumonia)
and/or (2) patients who received a prescription of antibiotics (ATC J01) and/or of
neuraminidase inhibitors (J05 AH).
3 / 11
Data processing and filtering
Data preparation and selection process is depicted in Fig 1. Patients with a diagnosis of RTI as
defined above who had seen a PCP in Bavaria were selected. Only patients with one diagnosis
of a RTI and/or one antibiotic prescription per quarter were included in order to allow an
association between diagnosis and prescription. Exclusion criteria were: Patients with
incompatible or implausible coding combinations, younger than fifteen years and/or with a diagnosis
of other infectious diseases. Comorbidity was determined using the Charlson comorbidity
]. Primary care practices with less than 200 patients per doctor per quarter and with
less than ten patients with RTIs per doctor per quarter were excluded in order to select
ªtypicalº primary care providers only. The following antibiotics used in the treatment of RTIs were
selected through their ATC-codes: tetracycline (J01A), beta-lactam antibiotics penicillin
(J01C), other beta-lactam antibacterial like first-, second-, third- and fourth generation
cephalosporin (J01DB, J01DC, J01DD, J01DE, respectively) and other cephalosporin (J01DI),
macrolides, (J01F) and quinolones (J01M). Data of patient-physician-contacts in 6.647 PCP
practices were analysed.
Antibiotic prescribing rate per practice was defined as the proportion of patients with an
antibiotic prescription divided by the number of patients with RTI. Data of all eight quarters have
been summarized. PCP sample was stratified for prescribing rate into low prescribers (<25th
percentile) and high prescribers (>75th percentile) . Diagnostic labelling was defined as
the proportion of a specific RTI diagnosis (numerator) and the sum of all RTI diagnoses
(denominator). Also, the proportion of the sum of all perceived bacterial infections of all RTI
diagnoses was calculated. The following diagnoses were counted as perceived bacterial
infections: R72 (strep throat), R75 (acute/chronic sinusitis), R76 (acute tonsillitis), R77 (acute
laryngitis/tracheitis), R78 (acute bronchitis/bronchiolitis), and R81 (pneumonia). Differences
between types of prescriber were tested statistically applying Student's t-test with a significance
level at 5%. Effect sizes using Cohen's d were calculated [
Factors associated with type of prescriber were examined using stepwise binary logistic
regression, the depending variable being the type of prescriber (high vs. low). As we were
interested to contrast factors associated with high prescribing behaviour as compared to factors
associated with low prescribing, only data of high prescribers and low prescribers practices
were included in the analysis. As described above, data were stratified for prescriber type and
therefore we included data of 1.662 high prescriber practices and data of 1.662 low prescriber
practices (in sum N = 3.324 primary care practices). Predictors were: (1) patient
characteristics: proportion of patients with perceived bacterial infections, proportion of patients older
than 65 years and proportion of patients with co-morbidities (indicated by the proportion of
patients with a Charlson index > 0), (2) practice characteristics: practice volume per quarter,
index of regional deprivation (BIMD, ), prescribing dominance, type of practice
(singlehanded vs. group practice) and practice location (rural vs. urban/large cities). All predictors,
except type of primary care practice and practice location, were continuous variables. To
simplify the interpretation of odds ratios in the regression model, all continuous variables were
transformed in categorical variables using quintiles. For each categorical variable, the lowest
quintile acted as the reference category. Crude and adjusted odds ratios (OR) were
determined. In the adjusted model all variables were taken into account. OR described the degree of
increasing or decreasing odds of high prescribers in a specific quintile category as compared to
the odds of high prescribers in the lowest quintile. As physicians' age and gender could only be
linked to prescription data on practice level, these analyses were restricted to single-handed
4 / 11
Fig 1. Flowchart for data selection and filtering and to illustrate characteristics of PCPs and patients.
practices. To analyse the effect of physicians' age, Mann-Whitney-U-Test was applied and
effect size was calculated (formula according to Cohen [
]: r = Z / sqrt (N)). To evaluate,
whether gender has an effect, a Chi-squared test was applied.
5 / 11
Antibiotic prescribing rate: The mean prescribing rate was M = 24.9% (SD = 14%). High
prescribers had a prescribing rate above 33.9%, whereas low prescribers were below 14.9%. Mean
prescribing rate within the high prescriber group was Mhigh = 43.5% (SD = 7.7%) compared to
Mlow = 8.5% within the low prescriber group (SD = 4.5%, t(2681.34) = -158.93, p < .01,
d = 5.63).
Diagnostic labelling: Results and statistical analysis can be seen in Table 1. The most
frequent diagnosis was acute upper respiratory tract infection (R74; M = 44.5%) followed by
acute bronchitis (R78; M = 31.9%). Low prescribers diagnosed significantly more often acute
upper respiratory tract infections (Mlow = 52.4% vs. Mhigh = 33.9%; t(3278.3) = 23.3, p < .01,
d = 0.8), whereas acute bronchitis was diagnosed more often by high prescribers (Mlow =
25.2% vs. Mhigh = 41.1%; t(3292.8) = -23.7, p < .01, d = 0.8). Over all practices, the mean
proportion of perceived bacterial diagnoses was M = 53.3% (SD = 22.9%). Perceived bacterial
infections were diagnosed more often by high prescribers (Mhigh = 64.5%, Mlow = 45.2%, t
(3299.9) = -24.87, p < .01, d = 0.9). Effect sizes indicate that differences are of clinical
Factors associated with high prescribing: The regression analysis model can be seen in
Table 2. Crude OR indicated a strong association between high prescribing and diagnostic
labelling, higher age of patients and comorbidities. Crude odds for being a high prescriber
were eleven times higher in practices with the highest rates of perceived bacterial infections
compared to practices with the lowest rates (crude OR for highest quintile = 10.8, 95% CI
[8.5,13.8]. In the fully adjusted model, adjusted OR for patients' characteristics changed.
Adjusted odds for high prescribers increased to being 14 times higher in practices with the
highest proportion of perceived bacterial infections (adjusted OR = 13.9, 95% CI [10.2, 18.8]).
Interestingly, the OR for practices with the largest number of patients with comorbidities
decreased to 0.6 when being adjusted for other factors in the model (95% CI [0.4, 0.8]. This
means, a 40% decrease of high prescribers in practices with the largest number patients with
comorbidities compared to practices with the lowest number of patients with comorbidities. A
high proportion of patients at old age was not associated with high prescribing (adjusted
OR = 0.9, 95% CI [0.7, 1.2]). In the adjusted model, structural factors such as higher practice
volume, deprived area, higher prescribing dominance and practice in rural area remained
6 / 11
having a strong positive association with high prescribing. Being a physician in a
singlehanded practice strengthened the negative association with high prescribing.
Regarding physicians' age and gender, the difference between high prescribers and low
prescribers was analysed for single-handed practices only. Concerning age, the difference was
significant (Medianlow = 57.8 years, Medianhigh = 60.0 years, p < 0.02), but an effect size of
r = 0.03 indicated that the difference is negligible. The analysis of gender showed that women
were more likely to be in the low prescriber group (women: low prescriber: 56.2%, high
prescriber: 43.8%, men: low prescriber: 50.3%, high prescriber: 49.7%, Χ (1) = 6.4, p = .01).
Across primary care practices in Bavaria antibiotics were prescribed for roughly a quarter of
patients with respiratory tract infections. Prescribing rates differed considerably between
7 / 11
physicians: high prescribers were five times more likely to prescribe antibiotics. High
prescribers diagnosed more perceived bacterial infections, this in turn being the best predictor for high
prescriber type. Being a physician working in a practice with a multimorbid practice
population was the best predictor for low prescribing. High and low prescriber practices differed in
structural factors: high prescribers had a higher practice volume and a higher degree of
prescribing dominance. They were more likely to work in deprived areas and in rural settings.
Compared to our study, a similar prescribing rate was found in an earlier routine data
analysis of German ambulatory care (Bavaria: 27%, [
]). Clearly higher prescribing rates were
found in studies conducted in the Netherlands, the United Kingdom and Sweden [1, 28±31].
There are two possible explanations. Firstly, the finding could be attributed to methodological
differences. We applied restricted inclusion criteria to allow for an association between
diagnosis and prescription, e.g. patients with other infectious diseases and with more than one
diagnosis of RTI were excluded. Eliminating patients with a more severe clinical course could
have resulted in a diluting effect and an underestimation of true prescribing rates. Secondly,
differences in health care systems are possible reasons for inconsistencies in prescribing rates.
In the UK and in Scandinavian countries with strictly implemented primary care systems, GPs
account for nearly all prescriptions for an individual patient, whereas in Germany patients also
receive prescriptions from office-based specialists. We assume that co-prescribing by
specialists was leading to lower prescribing rates for PCPs with a lower prescribing dominance in our
study. This is supported by the fact that higher prescribing rates were associated with a higher
degree of prescribing dominance (see Table 2).
Our results support previous findings of considerable differences in prescribing rates
between physicians [
14, 15, 30, 32, 33
]. The structural factors of practices we identified were
consistent with those found in other studies: higher practice volume [6, 9, 10], regional
deprivation [13, 33] and rural setting  are associated with high prescribing. The strongest
association with high prescriber type was found for diagnostic labelling. This also confirms results
found in earlier studies [18, 19, 21, 34]. In a German study, when asked about reasons to
prescribe antibiotics, physicians reported that patient-related factors such as age and
comorbidities had a strong influence on their decision [
]. Surprisingly, in our study comorbidity was
associated with lower prescribing. Looking at patient factors such as age or comorbidity alone,
the analysis showed that these factors are associated with high antibiotic prescribing rates
(crude OR in highest quintiles = 2.3 and 2.9, respectively; see Table 2). This association
disappeared when adjusting for other relevant factors such as diagnostic labelling (adjusted OR in
highest quintiles 0.9 and 0.6, respectively; see Table 2). This aspect of our findings was
confirmed by other studies in the medical literature. An analysis of medical records by Aspinall
et al. confirmed that both a diagnosis of acute bronchitis and comorbidity were associated
with high antibiotic prescribing . However, there was a much weaker association to
comorbidity . An evaluation of high and low prescribing in another study found that patients with
low comorbidity were equally covered by both low and high prescribers [
]. Sutter et al.
concluded that whether or not a patient received an antibiotic was mainly determined by
physicians' personal attitude and not so much by the clinical picture, and that the tendency to
prescribe medication in general and a defensive attitude were related to antibiotic prescribing
. Other studies found that a diagnosis might serve as a justification for treatment choice
[18, 19]. Patients with acute respiratory tract infections often show nonspecific symptoms
making a valid diagnosis more difficult. So one might ask whether in this case diagnostic
labelling is closely related to prescribers' personal traits. With our data, we cannot prove the
assumption that prescribing behaviour and diagnostic labelling are related to personal
characteristics. Due to the limitations of secondary data, further research using primary data or data
linkage of primary and secondary data should aim to determine the causal relationship
8 / 11
between prescribing, diagnostic labelling and patient characteristics. This could be achieved by
applying a study design linking routine data to questionnaire studies in order to investigate the
role of personal attitudes such as tolerance of ambiguity [
The limitations of secondary data are well-known and mainly caused by the fact that the
original purpose to capture those data was for billing purposes, not for research. This raises the
question of data accuracy, precision and completeness. Inclusion and exclusion criteria were
restricted to patients with one diagnosis and/or one antibiotic prescription to indirectly make
probable a direct link between diagnosis and prescription. As a result, generalisability of our
results may be somewhat reduced.
The rates of antibiotic prescribing were relatively low compared to the UK or to Scandinavian
countries. There was a considerable difference between prescribing rates of high and low
prescribers. Diagnostic labelling was the best predictor for high prescribing. Structural factors of
primary care practices were also strong influential factors. In contrast to what we had expected,
patient comorbidity was not associated with high prescribing rates, when adjusted for other
The authors thank the Bavarian Association of Statutory Health Insurance Physicians
(KassenaÈrztliche Vereinigung Bayerns, KVB) for data provision, for their advice and support during
Conceptualization: Susann Hueber, Thomas Kuehlein, Roman Gerlach, Martin Tauscher,
Data curation: Susann Hueber.
Formal analysis: Susann Hueber, Roman Gerlach, Martin Tauscher, Angela Schedlbauer.
Investigation: Susann Hueber, Thomas Kuehlein, Angela Schedlbauer.
Methodology: Susann Hueber, Thomas Kuehlein, Roman Gerlach, Martin Tauscher, Angela
Project administration: Susann Hueber.
Writing ± original draft: Susann Hueber.
Writing ± review & editing: Thomas Kuehlein, Roman Gerlach, Martin Tauscher, Angela
9 / 11
NICE CfCPa. Respiratory tract infections-antibiotic prescribing: prescribing of antibiotics for self-limiting
respiratory tract infections in adults and children in primary care. 2008.
Coenen S, Michiels B, Van Royen P, Van der Auwera JC, Denekens J. Antibiotics for coughing in
general practice: a questionnaire study to quantify and condense the reasons for prescribing. BMC Fam
Pract. 2002; 3(1):16. https://doi.org/10.1186/1471-2296-3-16 PMID: 12217080; PubMed Central
Gjelstad S, Dalen I, Lindbaek M. GPs' antibiotic prescription patterns for respiratory tract infectionsÐ
still room for improvement. Scand J Prim Health Care. 2009; 27(4):208±15. https://doi.org/10.3109/
02813430903438718 PMID: 19929185; PubMed Central PMCID: PMC3413912.
Cadieux G, Tamblyn R, Dauphinee D, Libman M. Predictors of inappropriate antibiotic prescribing
among primary care physicians. Canadian Medical Association Journal. 2007; 177(8):877±83. https://
doi.org/10.1503/cmaj.070151 PMID: 17923655
10 / 11
WHO Collaborating Centre for Drug Statistics Methodology: Anatomical Therapeutic Chemical
1. Dekker AR , Verheij TJ , van der Velden AW. Inappropriate antibiotic prescription for respiratory tract indications: most prominent in adult patients . Family practice . 2015 ; 32 ( 4 ): 401 ±7. https://doi.org/10. 1093/fampra/cmv019 PMID: 25911505
2. Hering R , Schulz M , BaÈtzing-Feigenbaum J . Entwicklung der ambulanten Antibiotikaverordnungen im Zeitraum 2008 bis 2012 im regionalen Vergleich . Versorgungsatlas . 2014 .
Aspinall SL , Good CB , Metlay JP , Mor MK , Fine MJ . Antibiotic prescribing for presumed nonbacterial acute respiratory tract infections . Am J Emerg Med . 2009 ; 27 ( 5 ): 544 ± 51 . https://doi.org/10.1016/j.ajem.
2008 . 04 .015 PMID: 19497459 .
Gjelstad S , Straand J , Dalen I , Fetveit A , Strøm H , Lindbñk M . Do general practitioners' consultation rates influence their prescribing patterns of antibiotics for acute respiratory tract infections? Journal of antimicrobial chemotherapy . 2011 ; 66 ( 10 ): 2425 ± 33 . https://doi.org/10.1093/jac/dkr295 PMID: 21784782
Hutchinson JM , Foley RN . Method of physician remuneration and rates of antibiotic prescription . Canadian Medical Association Journal . 1999 ; 160 ( 7 ): 1013 ± 7 . PMID: 10207340 Mazzaglia G , Caputi AP , Rossi A , Bettoncelli G , Stefanini G , Ventriglia G , et al. Exploring patient-and doctor-related variables associated with antibiotic prescribing for respiratory infections in primary care .
European journal of clinical pharmacology . 2003 ; 59 ( 8 ±9): 651 ±7. https://doi.org/10.1007/s00228-003 - 0669-0 PMID: 13680039
Linder JA , Singer DE , Stafford RS . Association between antibiotic prescribing and visit duration in adults with upper respiratory tract infections . Clinical therapeutics . 2003 ; 25 ( 9 ): 2419 ± 30 . PMID: 14604741 Blommaert A , Marais C , Hens N , Coenen S , Muller A , Goossens H , et al. Determinants of betweencountry differences in ambulatory antibiotic use and antibiotic resistance in Europe: a longitudinal observational study . Journal of Antimicrobial Chemotherapy . 2014 ; 69 ( 2 ): 535 ± 47 . https://doi.org/10.1093/jac/ dkt377 PMID: 24080501
Koller D , Hoffmann F , Maier W , Tholen K , Windt R , Glaeske G. Variation in antibiotic prescriptions: is area deprivation an explanation? Analysis of 1.2 million children in Germany . Infection. 2013 ; 41 ( 1 ): 121 ±7. https://doi.org/10.1007/s15010-012-0302-1 PMID: 22826031 De Sutter AI , De Meyere MJ , De Maeseneer JM , Peersman WP . Antibiotic prescribing in acute infections of the nose or sinuses: a matter of personal habit? Family practice . 2001 ; 18 ( 2 ): 209 ± 13 . PMID: 11264274
Fiegen H , Sielk M , Brockmann S , Altiner A . Wie haÈ ufig werden bei akutem Husten Reserveantibiotika verordnet . Z Allg Med . 2002 ; 81 : 372 ± 6 .
Mainous AG III , Hueston WJ , Love MM . Antibiotics for colds in children: who are the high prescribers? Archives of pediatrics & adolescent medicine . 1998 ; 152 ( 4 ): 349 ± 52 .
Verheij TJM , Hermans J , Kaptein AA , Wijkel D , Mulder JD . Acute BronchitisÐGeneral-Practitioners Views Regarding Diagnosis and Treatment . Family Practice . 1990 ; 7 ( 3 ): 175 ± 80 . https://doi.org/10.
1093/fampra/7.3.175 PMID: 2245886
Grover ML , Mookadam M , Rutkowski RH , Cullan AM , Hill DE , Patchett DC , et al. Acute respiratory tract infection: a practice examines its antibiotic prescribing habits . The Journal of family practice . 2012 ; 61 ( 6 ): 330 ± 5 . PMID: 22670235
1972 ; 22 ( 118 ): 310 . PMID: 5073372
Hutchinson J , Jelinski S , Hefferton D , Desaulniers G , Parfrey P . Role of diagnostic labeling in antibiotic prescription . Canadian Family Physician . 2001 ; 47 ( 6 ): 1217 ± 24 .
Duijn HJ , Kuyvenhoven MM , Tiebosch HM , Schellevis FG , Verheij TJ . Diagnostic labelling as determinant of antibiotic prescribing for acute respiratory tract episodes in general practice . BMC Family Practice . 2007 ; 8(1):1±5 . https://doi.org/10.1186/ 1471 -2296-8-55 PMID: 17883832 22 . ICPC-2-R: international classification of primary care: Oxford University Press, USA; 2005 .
Maier W , Fairburn J , Mielck A . Regionale Deprivation und Mortalit aÈt in Bayern . Entwicklung eines, Index Multipler Deprivation'auf Gemeindeebene. Das Gesundheitswesen . 2012 ; 74 ( 7 ): 416 ± 25 . https:// doi.org/10.1055/s-0031-1280846 PMID: 22020751
25. Sundararajan V , Henderson T , Perry C , Muggivan A , Quan H , Ghali WA . New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality . J Clin Epidemiol . 2004 ; 57 ( 12 ): 1288 ± 94 . https://doi.org/10.1016/j.jclinepi. 2004 . 03 .012 PMID: 15617955 .
26. Cohen J . Statistical power analysis for the behavioral sciences 2 edition Lawrence Erlbaum . Hillsdale, NJ. 1988 .
27. Schulz M , Kern WV , Hering R , Schulz M , Baetzing-Feigenbaum J . Antibiotikaverordungen in der ambulanten Versorgung in Deutschland bei bestimmten Infektionserkrankungen. Teil 2ÐKrankheitsspezifische Analyse von QualitaÈtsindikatoren auf regionaler Ebene . Zentralinstitut der KassenaÈrztlichen Vereinigung in Deutschland, 2014 20.06. 2014 . Report No.
28. Gulliford MC , van Staa T , Dregan A , McDermott L , McCann G , Ashworth M , et al. Electronic health records for intervention research: a cluster randomized trial to reduce antibiotic prescribing in primary care (eCRT study) . Ann Fam Med . 2014 ; 12 ( 4 ): 344 ± 51 . Epub 2014/07/16. https://doi.org/10.1370/afm.1659 PMID: 25024243; PubMed Central PMCID : PMCPmc4096472 .
29. Tell D , Engstrom S , Molstad S. Adherence to guidelines on antibiotic treatment for respiratory tract infections in various categories of physicians: a retrospective cross-sectional study of data from electronic patient records . Bmj Open . 2015 ; 5 ( 7 ):e008096. https://doi.org/10.1136/bmjopen-2015 -008096 PMID: 26179648
30. Gulliford MC , Dregan A , Moore MV , Ashworth M , Staa T , McCann G , et al. Continued high rates of antibiotic prescribing to adults with respiratory tract infection: survey of 568 UK general practices . BMJ Open . 2014 ; 4 ( 10 ): e006245 . Epub 2014 /10/29. https://doi.org/10.1136/bmjopen-2014-006245 PMID: 25348424; PubMed Central PMCID : PMC4212213 .
31. Adriaenssens N , Coenen S , Versporten A , Muller A , Minalu G , Faes C , et al. European Surveillance of Antimicrobial Consumption (ESAC): outpatient antibiotic use in Europe ( 1997 ± 2009 ). Journal of antimicrobial chemotherapy . 2011 ; 66 ( suppl 6 ):vi3± vi12 .
32. Steffensen FH , ønheyder HC , Ørensen Ht. High prescribers of antibiotics among general practitionersrelation to prescribing habits of other drugs and use of microbiological diagnostics . Scandinavian journal of infectious diseases . 1997 ; 29 ( 4 ): 409 ± 13 . PMID: 9360258
Wang KY , Seed P , Schofield P , Ibrahim S , Ashworth M. Which practices are high antibiotic prescribers? A cross-sectional analysis . Br J Gen Pract . 2009 ; 59 ( 567 ):e315± e20 . https://doi.org/10.3399/ bjgp09X472593 PMID: 19843411
34. Deschepper R , Vander Stichele RH , Haaijer-Ruskamp FM . Cross-cultural differences in lay attitudes and utilisation of antibiotics in a Belgian and a Dutch city . Patient education and counseling . 2002 ; 48 ( 2 ): 161 ± 9 . PMID: 12401419
35. Velasco E , Espelage W , Faber M , Noll I , Ziegelmann A , Krause G , et al. A national cross-sectional study on socio-behavioural factors that influence physicians' decisions to begin antimicrobial therapy . Infection . 2011 ; 39 ( 4 ): 289 ± 97 . https://doi.org/10.1007/s15010-011 -0137-1 PMID: 21717149
36. Barlam TF , Morgan JR , Wetzler LM , Christiansen CL , Drainoni ML . Antibiotics for respiratory tract infections: a comparison of prescribing in an outpatient setting . Infect Control Hosp Epidemiol . 2015 ; 36 ( 2 ): 153 ± 9 . Epub 2015/01/31. https://doi.org/10.1017/ice. 2014 .21 PMID: 25632997 .
37. Gerrity MS , DeVellis RF , Earp JA . Physicians' reactions to uncertainty in patient care: a new measure and new insights . Medical care . 1990 : 724 ± 36 . PMID: 2385142
38. Gerrity MS , White KP , DeVellis RF , Dittus RS . Physicians' reactions to uncertainty: refining the constructs and scales . Motivation and Emotion . 1995 ; 19 ( 3 ): 175 ± 91 .