Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Association between physicians’ interaction with pharmaceutical companies and their clinical practices: A systematic review and meta-analysis

  • Hneine Brax ,

    Contributed equally to this work with: Hneine Brax, Racha Fadlallah

    ‡ These authors are first authors on this work.

    Affiliation Faculty of Medicine, Université Saint Joseph, Beirut, Lebanon

  • Racha Fadlallah ,

    Contributed equally to this work with: Hneine Brax, Racha Fadlallah

    ‡ These authors are first authors on this work.

    Affiliation Center for Systematic Reviews of Health Policy and Systems Research (SPARK), American University of Beirut, Beirut, Lebanon

  • Lina Al-Khaled,

    Affiliation Department of Pediatrics and Adolescent Medicine, Faculty of Medicine, American University of Beirut, Beirut, Lebanon

  • Lara A. Kahale,

    Affiliation Department of Internal Medicine, American University of Beirut, Beirut, Lebanon

  • Hala Nas,

    Affiliation Faculty of Medicine, University of Damascus, Damascus, Syria

  • Fadi El-Jardali,

    Affiliations Center for Systematic Reviews of Health Policy and Systems Research (SPARK), American University of Beirut, Beirut, Lebanon, Department of Health Management and Policy, American University of Beirut, Beirut, Lebanon, Department of Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, ON, Canada

  • Elie A. Akl

    ea32@aub.edu.lb

    Affiliations Center for Systematic Reviews of Health Policy and Systems Research (SPARK), American University of Beirut, Beirut, Lebanon, Department of Internal Medicine, American University of Beirut, Beirut, Lebanon, Department of Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, ON, Canada

Abstract

Background

Pharmaceutical company representatives likely influence the prescribing habits and professional behaviors of physicians. The objective of this study was to systematically review the association between physicians’ interactions with pharmaceutical companies and their clinical practices.

Methods

We used the standard systematic review methodology. Observational and experimental study designs examining any type of targeted interaction between practicing physicians and pharmaceutical companies were eligible. The search strategy included a search of MEDLINE and EMBASE databases up to July 2016. Two reviewers selected studies, abstracted data, and assessed risk of bias in duplicate and independently. We assessed the quality of evidence using the GRADE approach.

Results

Twenty articles reporting on 19 studies met our inclusion criteria. All of these studies were conducted in high-income countries and examined different types of interactions, including detailing, industry-funded continuing medical education, and receiving free gifts. While all included studies assessed prescribing behaviors, four studies also assessed financial outcomes, one assessed physicians’ knowledge, and one assessed their beliefs. None of the studies assessed clinical outcomes. Out of the 19 studies, 15 found a consistent association between interactions promoting a medication, and inappropriately increased prescribing rates, lower prescribing quality, and/or increased prescribing costs. The remaining four studies found both associations and lack of significant associations for the different types of exposures and drugs examined in the studies. A meta-analysis of six of these studies found a statistically significant association between exposure and physicians’ prescribing behaviors (OR = 2.52; 95% CI 1.82–3.50). The quality of evidence was downgraded to moderate for risk of bias and inconsistency. Sensitivity analysis excluding studies at high risk of bias did not substantially change these results. A subgroup analysis did not find a difference by type of exposure.

Conclusion

There is moderate quality evidence that physicians’ interactions with pharmaceutical companies are associated with their prescribing patterns and quality.

Introduction

Promotional activities within pharmaceutical industries were relatively high in the past few years. In 2012, the pharmaceutical industry in the USA spent more than US$27 billion on drug promotion [1]. In Canada, promotional activities were estimated to cost US$30000 per physician per year [2]. Pharmaceutical companies appear to spend much more on promotion than they do on research and development (R&D) [3]. For example, a study based on annual reports of pharmaceutical companies found that ten of the largest global pharmaceutical companies spent a total of US$739 billion on ‘marketing and administration’ between 1996 and 2005 compared to US$288 billion on R&D for the same period [3].

The industry claims that the promotional activities aim to provide health care professionals with scientific and educational information [4]. Also, surveys suggest that many physicians believe that marketing does not influence their prescribing habits or acknowledge that it may have an influence on some physicians but not on themselves [57]. Despite these claims, there is evidence suggesting that the interaction of pharmaceutical companies with physicians may have a negative effect on their clinical practice [811].

The last identified systematic review assessing the interaction of pharmaceutical companies with physicians was published by Spurling et al. in 2010. The population under study included both physicians in practice and residents in training [11]. The review found high degrees of heterogeneity that may have been due to the diverse populations under study (i.e., both practicing and in-training physicians). The review by Spurling et al. could not reach definitive conclusions about the degree to which information from pharmaceutical companies decreases, increases or has no effect on the quality, cost or frequency of prescribing.

Since the publication of Spurling’s review, at least eight original studies have been published [1219]. One of these was a large study of the association between physicians’ receipt of meals from industry and the rates of prescribing the promoted drug to Medicare patients [15]. That study appears to be at lower risk of bias than the previously published studies, and thus would contribute the improving the quality of the evidence. Therefore, the objective of our study was to systematically review the association between physicians’ interactions with pharmaceutical companies and their clinical practice.

Methods

Protocol

We followed a detailed methodology that we describe in the protocol included in S1 Appendix. The review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (S2 Appendix).

Eligibility criteria

The inclusion criteria were:

  • Type of study design: observational design (e.g., cohort, time series analysis, before-after design, case control, cross sectional), and experimental design (non-randomized controlled trials, and randomized controlled trials);
  • Type of participants: practicing physicians as defined in the primary studies
  • Type of exposure: targeted interaction between physicians and pharmaceutical companies, where there is direct interaction with the physician. Direct interactions could include individual invitation to a continuing medical education (CME) event; active presentation of industry-related information to the physician; or provision of gifts to the individual;
  • Type of control: either no interaction or a lower level of interaction. The intention was to capture studies that stratified the levels of interaction of physicians with medical representatives, e.g., according to the number of visits within a specified period of time.
  • Types of outcomes:
    1. ○. Knowledge of physicians (e.g., accuracy of knowledge related to a specific medication);
    2. ○. Attitude of physicians (e.g., perceived influence of information from pharmaceutical company on their behavior);
    3. ○. Behavior of physicians (e.g., prescribing quality; prescribing quantity/frequency; reliance on pharmaceutical companies for drug information; giving drug sample to patients; submitting a formulary request for a drug made by a specific company);
    4. ○. Financial outcomes (e.g., patient out of pocket expenses; prescription costs);
    5. ○. Patients’ clinical outcomes.

The exclusion criteria were:

  • Qualitative studies, ecological studies, econometric studies, editorials, letters to the editor, and non-English studies.
  • Studies focusing on medical students and physicians in training.
  • Studies assessing the relationship between attitudes and behavior
  • Studies assessing non-targeted interactions (e.g., journal advertisement) and research funding
  • Studies assessing interventions to share industry-independent drug information or interventions to reduce interactions between physicians and pharmaceutical companies. The latter has been addressed in a recent systematic review [20].

We did not exclude studies based on date of publication. We did not exclude any study based on risk of bias. Instead, we conducted sensitivity analyses excluding studies at high risk of bias, and also took risk of bias into account when grading the quality of evidence using GRADE approach.

Search strategy

We used OVID interface to electronically search MEDLINE and EMBASE in July 2016. The search included both free text words and medical subject headings. It combined terms for physicians and pharmaceutical industry and did not use any search filter. A medical librarian assisted with designing the search strategy (see supporting file S3 Appendix for full search strategy). In addition, we reviewed the references lists of included and relevant primary studies and literature reviews.

Selection of studies

Two reviewers screened the title and abstracts of identified citations for potential eligibility in duplicate and independently. We retrieved the full text for citations considered potentially eligible by at least one of the two reviewers. The two reviewers then screened the full texts in duplicate and independently for eligibility. The reviewers resolved disagreement by discussion or with the help of a third reviewer. They conducted calibration exercises and used a standardized and pilot tested screening form.

Data collection

Two reviewers abstracted data from eligible studies using a standardized and pilot tested screening form with detailed written instructions. When needed, disagreement was resolved with the help of a third reviewer. The data abstracted included: type of study, funding source, characteristics of the population, exposure, outcomes assessed, and statistical data. For studies including both attending physicians and residents, we attempted to contact authors for data relating to the former group.

Assessment of risk of bias in included studies

Two reviewers assessed the risk of bias in each eligible study in duplicate and independently. They resolved disagreements by discussion or with the help of a third reviewer. We used the tool suggested by the GRADE working group for assessing the risk of bias for observational studies [21].

We calculated the risk of bias using the following criteria:

  • Failure to develop and apply appropriate eligibility criteria (e.g., no clear eligibility criteria, convenient sampling, under- or over-matching in case-control studies, selection of exposed and unexposed in cohort studies from different populations, and low response rate (<60%) with no attempts to compare non-respondents to respondents) [22].
  • Flawed measurement of exposure (e.g., differences in measurement of exposure such as recall bias in case- control studies, and subjective or self-reported assessment of exposure)
  • Flawed measurement of outcome (e.g., differential surveillance for outcome in exposed and unexposed in cohort studies, and subjective or self-reported assessment of outcome)
  • Failure to adequately control confounding (e.g., failure of accurate measurement of all known prognostic factors, absence of control group in a before-after study, failure to match for prognostic factors and/or adjustment in statistical analysis
  • Incomplete follow-up or failure to control for loss-to-follow up

We graded each potential source of bias as high, low or unclear risk of bias. We used unclear when the authors did not report enough information for us to make the judgment.

Data synthesis

We calculated the kappa statistic to assess the agreement between reviewers for full text screening.

We conducted a meta-analysis to pool the results across studies for the association between ‘targeted interactions with physicians’ as the exposure of interest, and ‘changes in physician prescribing behavior’ as the outcome of interest. We contacted the authors of studies that appeared to have measured the outcome and exposure of interest, but did not report data (such as odds ratio or standard error) that we could include directly. We received responses from authors of 7 out of 9 relevant studies. The authors provided us with the needed information for only two out of the seven studies. Please refer to S1 Table for a summary of author contacts.

We used the following a priori plan for choosing which data to include in the meta-analysis:

  • For studies reporting on more than one type of exposure (e.g., gifts, detailing), we treated each exposure as a separate unit of analysis.
  • For studies measuring the same outcome at several points in time, we chose the first time point to avoid any potential confounding effects from subsequent measures.
  • For studies assessing the association of interest for more than one drug (i.e., reporting more than one association), we included the value that is the closest to the mean of all reported values amongst those associations.

We used the generic inverse variance technique with a random-effects model to pool the association measures across included studies that reported the needed statistical data. We carried out statistical analysis using RevMan (version 5.2).

To take into account the heterogeneity introduced by the different types of exposures (i.e., gifts, detailing, and CME), we stratified the meta-analyses by type of exposure. We tested the results for homogeneity using the I2 test and considered heterogeneity present if I2 exceeded 50%.

In addition, we conducted three post-hoc sensitivity analyses by respectively excluding:

  • Studies at high risk of bias;
  • Studies funded by pharmaceutical industry;
  • Studies measuring the outcome of interest as ‘changes in generic prescription’ or ‘formulary request’ (as these were considered indirect measures compared with the ‘changes in the prescribing of promoted drug’).

Although we had planned to construct funnel plots to assess publication bias, the number of included studies in the meta-analyses was too low to allow for that. Indeed, funnel plots are encouraged for interventions that include at least 10 studies, with a substantially higher number required if significant heterogeneity is present.[23].

We used the GRADE approach to assess the quality of the body of evidence [21]. The GRADE methodology involves rating the initial quality of evidence for an association as high (with observational data), followed by downgrading based on five criteria (risk of bias, inconsistency, imprecision, indirectness and publication bias), and upgrading based on three criteria (large effect size, dose-response gradient, and plausible confounding) [24].

We narratively reported any additional results that we were not able to include in the meta-analysis from eligible studies (this includes studies that could have contributed data to the meta-analysis). Whenever provided, we included the p-value to denote significance of results.

Results

Selection of studies

Fig 1 shows the study flow. Of the 12, 400 article titles identified by the electronic literature search, 20 articles reporting on 19 studies met our inclusion criteria (two articles reported on different outcomes for the same study) [25, 26]. A list of the excluded studies along with reasons for exclusion is provided in S2 Table. The kappa statistic value for full text screening was 0.89, suggesting high levels of agreement.

Characteristics of included studies

Table 1 shows the characteristics of the 19 included studies. The design for the majority of studies was cross-sectional (n = 13). Four studies reported using pre-post study designs, [17, 2729] one was a retrospective cohort study [12] and one was a nested case-control study [30]. The sample sizes in these studies varied between 10 and 279, 669 with a median of 206. One study did not provide adequate information on the exact sample size [18]. The included studies were conducted in the USA (n = 13), Australia (n = 1), Spain (1), Denmark (n = 1), Germany (n = 1) and the Netherlands (n = 2). The publication year ranged from 1972 to 2016. The specialties of the physicians in the majority of studies were primary care (including general practitioners, family medicine, internal medicine) (n = 8); obstetrics and gynecology (OB/GYN) (n = 1); dermatologists (n = 1); mix (n = 3); and unclear (n = 6). The types of exposure assessed were sales representatives’ visits or detailing (n = 10), industry-funded continuing medical education including travel funds (n = 4), and receiving free gifts (e.g. drug samples, meals, gifts in the form of office stationery, and grants and payments) (n = 11). One study assessed a mix of exposures without reporting data specific to each exposure [31]. The types of outcomes assessed were physicians’ prescribing behaviors (n = 19), physicians’ beliefs (n = 1), physicians’ knowledge (n = 1) and financial outcomes (n = 4). Physicians’ prescribing behavior was defined as the changes in quantity or quality of prescriptions. None of the studies assessed clinical outcomes.

Risk of bias

The detailed judgments about each risk of bias item for included studies are displayed in Table 2. Fig 2 shows the corresponding risk of bias summary for these studies. For the majority of the studies, the risk of bias was judged to be low for ‘appropriate eligibility criteria’, ‘measurement of intervention’, and ‘measurement of outcome’, except for the ‘completeness of data’ that was judged as unclear. For ‘controlling for confounding’ the risk of bias was judged as low for nine studies, unclear for four studies, and high for six studies.

thumbnail
Fig 2. Risk of bias summary reflecting reviewers’ judgments about each risk of bias item for included studies.

https://doi.org/10.1371/journal.pone.0175493.g002

thumbnail
Table 2. Risk of bias and funding source for each included study.

https://doi.org/10.1371/journal.pone.0175493.t002

Findings of studies

Table 3 provides a summary of the outcomes and the statistical results reported for each included study. Out of the 19 included studies, six reported data in a format that could be included in the meta-analysis of the association between the exposure and the behavior (i.e., reported odds ratio or risk ratio or provided raw data allowing the calculation of an odds ratio) [15, 17, 19, 27, 28, 30]. Below, we present the results of those six studies and their meta-analysis. We then narratively report any additional results that we were unable to include in the meta-analysis from eligible studies.

thumbnail
Table 3. A summary of the outcomes and statistical results of each included study.

https://doi.org/10.1371/journal.pone.0175493.t003

Results of the meta-analysis

The study design of the six included studies were retrospective (n = 2), nested-case control (n = 1), pre-post (n = 1) and cross-sectional studies (n = 2). These studies assessed the following types of interactions (with some studies reporting on more than one type): detailing (n = 2); industry-funded continuing medical education including travel funding (n = 2); and receiving free gifts (drug samples and meals) (n = 3).

Sondergaard et al. conducted a retrospective cohort study and reported a statistically significant effect of the first visit of drug representatives on the general practitioner’s drug preference favoring the marketed drug (odds ratio (OR) 2.39; 95% confidence interval (CI) 1.72–3.32) [17]. The effect on drug preference increased further after the second visit (OR = 1.51; 95% CI: 1.19–1.93), but no significant change was noted after the third visit (OR = 1.06; 95% CI: 0.94–1.20). We considered the data for the first visits only as the analysis for subsequent visits may be confounded by the effect of the previous visits as highlighted by the authors: “the effect of promotional visits could in part be caused by representatives selecting practices with a higher probability of adopting the promoted drug. Although we have controlled for the time until first visit, a selection effect cannot be excluded.” Also, none of the remaining studies included in the meta-analysis reported data for subsequent visits.

Bowman et al. conducted a pre-post survey showing the effects of attendance of three continuing medical education courses, each subsidized by a single but different drug company, and the changes in rate of prescribing by physicians of course-related drugs [27]. We excluded the results for courses I and II, given data were not matched. In course III, Diltiazem was the sponsoring company's drug. Prescribing Diltiazem most frequently to new patients statistically increased from 22.3% to 33.9% pre-post course (p<0.05). In addition, the number of new prescriptions for Diltiazem increased statistically from 31.4% to 50.1% pre-post course (p<0.05%) [27]. We considered the first outcome and conducted sensitivity analysis that demonstrated no important changes in results.

Pinckney et al. examined in a cross-sectional study the interaction between the availability of medication samples in the clinics and the prescription preference of primary care prescribers (stated as the name of the medication) in response to two clinical vignettes [19]. Clinicians who did not have samples in their offices were more likely to prescribe hypertension medication according to clinical practice guidelines (p<0.01), and more likely to prescribe a depression medication that was generic (p = 0.02) [19]. Multivariable regression models were conducted only for the hypertension vignette. The findings showed that clinicians with samples were still less likely to select thiazide diuretic that is the preferred treatment for hypertension (OR = 0.15; 95% CI: 0.04–0.56).

Miller et al. conducted a retrospective study to look at the association of a pre-post removal of drug sample closet and the percentage of medications prescribed to uninsured or Medicaid patients as generics. Following a logistic regression model, the authors found that the absence of the sample closet was associated with uninsured patients receiving a generic prescription (OR = 4.54; 95% CI: 1.37–15.0) [28].

In a nested case-control study, Chren el at found that physicians who had met with pharmaceutical representatives were significantly more likely to have requested that drugs manufactured by specific companies be added to the formulary, than other physicians (OR = 3.4; 95% CI 1.8–6.6) [30]. Similar increased odds of formulary requests were obtained for physicians who had accepted money from those companies to attend educational symposia (OR = 7.9; 95% CI: 1.1–55.6), or to speak at educational symposia (OR = 3.9; 95% CI: 1.2–12.7). We treated each exposure as a separate unit of analysis in the meta-analysis.

Dejong et al. examined in a cross-sectional study the association between the receipt of an industry-sponsored meal promoting the drug of interest and prescribing rates of promoted drugs compared with alternatives in the same class [15]. Physicians who received a single meal promoting the drug of interest had higher rates of prescribing Rosuvastatin over other statins (OR = 1.18; 95% CI: 1.17–1.18), Olmesartan over other ACE inhibitors and ARBs (OR = 1.52; 95% CI: 1.51–1.53), Nebivolol over other β-blockers (OR = 1.70; 95% CI: 1.69–1.72), and Desvenlafaxine over other SSRIs and SNRIs (OR = 2.18; 95% CI: 2.13–2.23). We included the value that is the closest to the mean of all reported values amongst those associations.

We pooled the results for the six studies in a meta-analysis, stratified by type of exposure. Please refer to S3 Table for a summary of all decisions and their rationale with respect to the statistical data included in the meta-analysis. The pooled estimate showed a statistically significant association between interaction with pharmaceutical industry and physicians’ prescribing behaviors (OR = 2.52; 95% CI: 1.82–3.50). The heterogeneity was considered high with I2 of 64% (Fig 3). The test for subgroup effect did not identify any subgroup difference by type of exposure (Test for subgroup differences: P = 0.88).

thumbnail
Fig 3. Forest plot for changes in physician prescribing behavior stratified by type of exposure.

https://doi.org/10.1371/journal.pone.0175493.g003

Fig 4 shows the risk of bias summary for these six studies. We judged the overall risk of bias in four of these studies as low [15, 17, 28, 30]. Following the GRADE methodology we downgraded the quality of evidence for the outcome ‘behavior of physician’ from high to moderate for risk of bias and inconsistency. There was no major concern with imprecision, indirectness of the evidence, or publication bias warranting further downgrading.

thumbnail
Fig 4. Risk of bias summary reflecting reviewers’ judgments about each risk of bias item for studies included in the meta-analysis.

https://doi.org/10.1371/journal.pone.0175493.g004

The first sensitivity analysis excluded two studies judged as having an overall high risk of bias [19, 27]. The analysis found no important change in the pooled effect estimate (OR = 2.55, 95% CI: 1.75–3.71). In the second sensitivity analysis, we excluded the one study funded by pharmaceutical industry [17]. The analysis resulted in a non-substantial increase in size of the pooled effect estimate (OR = 2.74, 95% CI: 1.78–4.23). The third sensitivity analysis excluded two studies measuring the outcome of interest as ‘changes in generic prescription’ [28] and ‘formulary requests’ [30], respectively. The analysis resulted in the pooled effect estimate decreasing slightly in size but remaining statistically significant (OR = 2.11; 95% CI: 1.62–2.74). The details of the analyses are provided in S1, S2 and S3 Figs, respectively.”

Narrative summary of studies not included in the meta-analysis

As noted above, fourteen eligible articles reporting thirteen studies were not included in the meta-analysis [12, 14, 16, 25, 26, 29, 3137]. In addition, we narratively summarized the findings of the study by Pinckney et al (included in the meta-analysis), albeit only for the outcome ‘beliefs of physicians’ [19].

The study designs were cross-sectional (n = 12), pre-post survey (n = 1) [27] and retrospective cohort study (n = 1)[12]. These studies assessed the following types of interactions between physicians and drug representatives (with some studies reporting on more than one type): detailing (n = 8); continuing medical education including travel funding (n = 2); receiving free gifts (e.g. drug samples, meals, grants and payments) (n = 9); and a mix of interactions (n = 1; this study did not report data specific to each exposure). We summarized the results narratively, stratified by type of exposure and within each exposure by risk of bias.

Detailing (or pharmaceutical representative visits).

Nine articles reporting on eight studies evaluated the interactions between physicians and pharmaceutical sales representatives [12, 14, 25, 26, 3437]. All of these studies assessed prescribing behaviors. In addition to assessing prescribing behavior, one assessed financial outcome [25] and one assessed physicians’ knowledge [37]. Five of the eight studies found an association between detailing and increased prescribing frequency, lower prescribing quality, higher expenditure, or earlier awareness of promoted drugs. Three studies found both associations with higher prescribing frequency or expenditures and lack of significant associations, for the different types of drugs or exposures examined. The overall risk of bias was judged as high for two of the eight studies [14, 37].

Figueiras et al. and Caamano et al. conducted a cross-sectional study to examine the association between detailing and the amount, expenditure and quality of drug prescribed by primary care physicians. The quality of prescription was reflected via three indicators which were combined to produce a global indicator variable. Figueiras et al. showed that using information obtained from pharmaceutical representatives was associated with a higher percentage of prescription drug not included in the primary care formulary and with a higher “global indicator variable”, thus reflecting lower prescription quality [26]. Camaano, et al., found that the utilization of the visiting marketers’ information was significantly associated with higher amounts of prescription (p = 0.048) and higher expenditure per physician (p = 0.035) [25]. On the contrary, the number of sales representative visits was not statistically associated with prescription amount or expenditure.

Pedan et al. conducted a retrospective cohort study to assess the impact of detailing on new prescriptions for three different statin brands, namely, Lipitor, Crestor and Vytorin [12]. The authors provided data for promotional activities of ‘own brand’ and the promotion of ‘competitive brand’. The findings for ‘own promotion’ indicated that detailing produced a highly significant positive impact on new prescriptions for Lipitor and Crestor (p<0.05), while results were not significant for Vytorin. ‘Competitive’ detailing had a significant negative impact on new prescription for Vytorin (p<0.05) whereas results were not significant for the other two brands.

Muijrers et al. conducted a cross-sectional study and found that more frequent visits from pharmaceutical industry representatives had a significant negative correlation with the quality of prescribing of general practitioners measured as “adherence to guidelines” (p<0.05) [36]. The latter was calculated as a weighted average score on 20 prescribing indicators based on general practice guidelines of the Dutch College of General Practitioners.

Mizik et al. conducted a pooled time series cross-sectional study to evaluate the association between pharmaceutical sales representative visits per month and the number of new prescriptions issued by a physician per month for three studied drugs [35]. The average number of new prescription per month by drug, were: 1.56 (95% CI: 0.80–2.23) for drug A; 0.32 (95% CI: 0.22–0.43) for drug B; and 0.15 (95% CI: 0.11–0.20) for drug C.

Becker et al. conducted a cross-sectional study and found that the use of detail men as sources of prescribing information concerning new drugs was significantly associated with higher prescription of the drug chloramphenicol by primary care physicians (p<0.01), and a poorer rating of prescription quality (p<0.01) by two panels of experts relative to five common complaints and five common illnesses [34].

The findings of the remaining two studies with overall high risk of bias are discussed below [14, 37].

Peay et al. conducted a cross-sectional study and found that doctors (specialists and general practitioners) who had contact with detail men regarding the promoted drug Temazepam reported earlier awareness of it (p < 0.041), were more likely to rate it as a moderate advance (as opposed to a minor advance or no advance at all) (p< 0.028), were more likely to have prescribed it (p< 0.0031), reported prescribing it earlier (p< 0.005) and were more likely to prescribe it routinely in preference to alternatives (p< 0.014) [37]. In the multivariate analysis, contact with detail man regarding Temazepam and first news of drugs from detail man remained significantly related to four of the five dependent variables correlating with prescription of Temazepam.

In another cross-sectional study, Lieb et al. found that frequently visited practices (i.e. daily or 2–3 times per week) had a significantly higher number of prescriptions (p = 0.005) and total daily doses per patient (p = 0.003) compared to practices visited less frequently by representatives. However, they did not have a higher total expenditure per patient (p = 0.115). The effects of the “frequency” of pharmaceutical sales representatives visits on prescribing behavior were no longer significant after taking into consideration the number of patients per office [14].

Continuing medical education.

Two studies assessed the effects of industry-funded continuing medical education on physician prescribing behaviors [14, 29] and expenditure on off-patent branded drugs [14]. Both studies found significant associations between attending industry-funded continuing medical education and higher prescribing frequency, lower prescribing quality, or increased prescription cost. The overall risk of bias was judged as high for one of them [14].

A pre-post study tracked the prescribing pattern for two drugs before and after physicians attended symposia sponsored by a pharmaceutical company. The ‘expense-paid seminar at a resort’ was associated with a significant increase in the prescribing of the two promoted drugs within a few months of each symposium compared to their use before the symposium (p<0.001) [29].

The cross-sectional study with high risk of bias found that compared to doctors who frequently, occasionally or rarely took part in sponsored CME events, doctors who mentioned that they never took part in such events had a lower number of on patent-branded drug prescriptions per patient (mean ± SD; 1.05±0.35 vs. 1.27±0.55; p = 0.005, a higher proportion of generics (83.28±7.77% vs. 76.34±13.58%; p<0.0005) and lower expenditure on off-patent branded drugs per patient (£27.36±23.23 vs. £43.75±43.22; p = 0.002) [14].

Receiving free gifts.

Nine studies evaluated receiving free gifts as the exposure of interest [12, 14, 16, 18, 19, 32, 33, 35, 37]. All of these studies assessed prescribing behaviors. In addition to assessing prescribing behaviors, two assessed financial outcomes [18, 32] and one assessed physicians’ beliefs [19]. For the latter study, we did not include the findings pertaining to prescribing behavior as these were already included in the meta-analysis. Five of the nine studies found an association between receiving free gifts and increased prescribing frequency, lower prescribing quality, or increased prescription cost. Each of three studies found both associations with higher prescribing frequency and lack of significant associations, for the different types of gifts examined in the studies. The remaining study found that prescribers with access to samples were significantly more likely to believe that samples benefited patients. The overall risk of bias was judged as high for four of the nine studies [14, 19, 33, 37].

Pedan et al. conducted a retrospective cohort study to evaluate the effects of sample dispensing and meals on total monthly number of new prescriptions written by a physician for the three leading statin brands (Crestor, Lipitor and Vytorin) [12]. The authors provided data for promotional activities of ‘own brand’ as well as the promotion of ‘competitive brand’. The findings for ‘own promotion’ showed that sample dispensing had a significant positive effect on new prescription for Crestor (p<0.01) and Vytorin (p<0.05), but results were not significant for Lipitor. They also found that free meals had a significant positive impact on new prescription for all three statin brands: Lipitor (p<0.05), Crestor (p<0.05) and Vytorin (p<0.01). ‘Competitive’ sampling had a significant negative impact on new prescription for Lipitor (p<0.05) and Vytorin (p<0.05) but results were not significant for Crestor. ‘Competitive’ free meal-related promotions had a significant negative impact on Lipitor only (p<0.05) The authors concluded that, while on average the marketing efforts affect the brand share positively, the magnitude of the effects is very brand specific.

Mizik et al. conducted a pooled time series cross-sectional study to examine the association between receiving free sample medications and the number of new prescriptions issued by a physician per month for three drugs [35]. They observed statistically significant but small effects of sample dispensing on prescription behavior for all three types of drugs.

In a cross-sectional study, Symm et al. examined the prescription claims data for 25 medications in one “clinic X” where sample medications were dispensed compared to two clinics, Y and Z, which did not dispense free sample medications [32]. They showed that, first, family physicians in clinic X significantly wrote the largest proportion of prescriptions for study medications (p <0.0001) versus non-study medications. Second, family physicians in clinic X significantly prescribed the lowest proportion of preferred name brands among study medications (p <0.0001). Third, the average cost of a 30-day prescription differed significantly by clinic (p <0.0001), being the highest in clinic X.

Hurley et al. conducted a cross-sectional study and found a strong correlation (r = 0.92) between the increase in provision of samples with a prescription by dermatologists and increased use of branded generic drugs promoted by these samples [18]. For physicians at local academic centers where free samples are prohibited, only 17% (230 of 1364) of the commonly prescribed medications were for branded or branded generic drugs compared to 79% for office-based dermatologists on a national level where free samples are available. Additionally, the national mean total retail cost of prescriptions was “conservatively” estimated to be twice as higher (roughly $465 nationally versus $200 at an academic medical center where samples were prohibited).

Yeh et al. found that among physicians with industry payments in the Massachusetts database, every $1000 in total payments received was associated with a 0.1% increase in the rate of brand-name statin drug prescribing (95% CI: 0.06%-0.13%; P < .001) [16]. Receiving payment for educational training was associated with an average 4.8% increase in prescribing of brand-name drugs (95% CI: 1.55–7.95; p = 0.004), but the other types of payment (i.e. food, grants/education gifts, and bona fide services) were not.

The findings of the four studies with overall high risk of biases are summarized below.

Peay et al. examined in a cross-sectional study the association between receiving a sample of Temazepam and physicians’ prescription rate and preference for it over the alternatives [37]. They demonstrated that physicians (specialists and general practitioners) who had received a sample of Temazepam, compared to those who had not, were more likely to have prescribed it (p < 0.001) and more likely to say that they now usually prescribe it rather than the alternatives (p < 0.006).

Pinckney et al. examined the association between the presence of samples in offices and prescribers’ beliefs about the use of samples. They found that prescribers with samples were significantly more likely to believe that samples: are liked by patients; expedite treatment; help patients who cannot afford their medication; reduce patient costs; and help physicians assess the efficacy of medications. Nonetheless, most prescribers with samples still agreed that samples alter treatment plans and increase the costs of care [19].

In another cross-sectional study, Lieb et al. found that physicians who always or frequently accepted gifts in the form of office stationery prescribed higher daily dose totals per patient (mean ± SD; 491.97±158.95 vs. 420.53±140.57; p = 0.003) and more generics (mean ±SD; 385.52±147.52 vs. 319.43±133.69; p = 0.004) in comparison to physicians who only occasionally, rarely or never accepted stationery [14]. We did not include the other types of gifts as the categorization of answer options was not conducive to interpretation (e.g., the categories ‘rarely’ ‘frequently’ and ‘occasionally’ were grouped together as exposure group and the category ‘never’ as control group for one type of gift whereas for another type of gift, the categories ‘frequently’ ‘occasionally’ ‘rarely’ and ‘never’ were categorized together as control group).

Anderson et al. conducted a cross-sectional study and found that the frequency of eating industry-funded food was associated with greater reliance of obstetrics and gynecology physicians on pharmaceutical representatives for drug information when prescribing new medications [33]. This finding was statistically significant in a first regression analysis (β = 0.16, 95% CI: 0.02–0.31). It became non-significant in a second regression model including as an independent variable “the perceived value of pharmaceutical representatives in helping physicians to learn about new drugs” (β = 0.07, 95% CI: -0.06–0.19).

Mixed exposures.

One cross-sectional study with overall high risk of bias examined the association between rational prescribing of general practitioners and a mix of exposures including “manufacturer’s representatives, drug companies’ mailings, use of samples, drug companies’ journals, drug firm meetings and usefulness of drug company information” [31]. The investigators found that reliance on information from pharmaceutical industry was negatively associated with prescribing rationality (p<0.001).

Discussion

Summary and interpretation of findings

We identified 19 studies looking the association between physicians’ interactions with pharmaceutical companies and their prescribing behaviors. In addition to assessing prescribing behavior, four of the 19 studies assessed financial outcomes [12, 19, 26, 35], one assessed physicians’ knowledge [37], and one assessed their beliefs [19]. None of the included studies assessed clinical outcomes.

Out of the 19 studies, 15 found a consistent association between interactions promoting a medication, and inappropriately increased prescribing rates, lower prescribing quality, and/or increased prescription costs. Each of three studies found both associations with higher prescribing frequency and lack of significant associations, for the different types of exposures and drugs examined in the studies [12, 16, 25]. Only one study, albeit at high risk of bias, found an association between receiving gifts from the industry and increased generic prescribing whereas the results were mixed for the remaining types of exposures and outcomes assessed in that study [14].

A meta-analysis of six of the included studies provided moderate quality evidence showing more than doubling of inappropriate prescribing rate among practicing physicians. The heterogeneity was considered high. Sensitivity analyses excluding the two studies at high risk of bias and the industry-funded study, respectively, did not substantially change these results. A subgroup analysis did not find a difference by type of exposure.

Strengths and limitations of the review

A major strength of the present study is the use of Cochrane methodology for conducting the systematic review. In addition, this is the first systematic review to focus on practicing physicians, as other reviews included residents and physicians in training as their population of interest [11]. A potential limitation of our review is that we searched only two electronic databases. However, we believe our search was sensitive. Also, Spurling et al. did not include a study that was eligible for our review, and that was missed by our search strategy. Another potential limitation is the exclusion of studies published in a language other than English. Also, all included studies were conducted in developed, high-income countries; therefore, there is a chance that we have missed studies conducted in low or middle-income countries that have been published in a non-English language. However, given the consistency of findings, we expect them to be generalizable to those countries. Another limitation relates to the potential misattribution of the exposure category; however, such cases were few. One example is the study conducted by Yeh et al. which considered “payments for CME” as different from “payments for educational training” [16]. Other limitations relate to the observational nature of all included studies and to our inability to include some studies in the meta-analysis because they did not report data for the association between the exposure and the outcome of interest.

Comparison to findings of similar reviews

The latest systematic review addressing the same topic was published in 2010 and included physicians as well as residents [11]. The chosen populations may have contributed to the high level of statistical heterogeneity reported (I2 = 91%). Also, the previous systematic review focused on exposure to information directly provided by pharmaceutical companies, and thus excluded other types of exposures such as gifts, samples, and continuing medical education courses that were funded by unrestricted grants from pharmaceutical companies. On the other hand, our review has included these types of interaction and has covered 8 years of literature since 2008, the year of literature search of the 2010 review. Still, our review included a relatively lower number of studies compared to the previous review due to the stricter eligibility criteria (e.g., we excluded residents and medical trainees; non-targeted interactions such as advertisements in journals or prescribing software, or mailed information; and participation in sponsored clinical trials). Nevertheless, our findings are consistent with those of the previous review in terms of decreased prescribing quality and increased costs associated with exposure to information provided directly by pharmaceutical companies. In addition, we found evidence of similar effects associated with drug samples and industry gifts (both of which were not assessed in the previous review).

Implications for policy and practice

Our findings suggest that the interaction between physicians and pharmaceutical companies should be better managed to reduce any negative effects and promote appropriate drug prescribing. While many policy options have been proposed to manage the interactions between physicians and pharmaceutical companies [38], the level of evidence supporting them varies, as described below.

There is an increasing trend of mandating pharmaceutical companies to disclose payments to physicians. For instance, the Physician Payments Sunshine Act enacted in the USA in 2010 requires pharmaceutical and medical device manufacturers to publicly disclose payments (or transfers of values) exceeding US$10 per instance or US$100 per year made to physicians and teaching hospitals [39]. Similarly, Medicines Australia recently revised its code of conduct, requiring pharmaceutical companies to report on payments to individual health professionals for their services, sponsorships to attend educational events, as well as educational grants [40]. However, so far there is no evidence supporting the effectiveness of mandatory disclosure [20].

As an example of regulatory approach, France introduced in 2004 the French Sales Visit Charter which requires sales representatives to provide physicians with “approved product information” [41]. A report by French National Authority for Health pointed to the ineffectiveness of the Charter and the difficulty in supervising the content of verbal information conveyed during representative visits [42].

A potentially effective option would be to restrict physician-industry interactions, particularly given the evidence that restriction policies may have a positive effect on improving prescribing behavior [20]. This could be achieved, at the institutional level by restricting free samples, promotional material, and meetings with pharmaceutical company representatives. For example, Stanford University banned pharmaceutical sales representatives from its hospitals [43]. Similarly, Memorial Sloan Kettering Cancer Center and Brody School of Medicine at East Carolina University banned all industry involvement (including funding) in CME, reportedly with success [44]. At a higher level, policy makers may consider legislation specifying the types of interactions that are permissible, and those that are not. For example, Minnesota enacted a law that bans pharmaceutical industries from providing gifts to physicians with a total annual combined retail value above US$50 [38].

While some may claim that restricting interactions between physicians and pharmaceutical companies could create an ‘information gap’, several studies conducted in different contexts found that sales representatives often did not state the risks and harmful effects of drugs to physicians [40, 41, 45]. Academic detailing has emerged as an effective alternative to industry-dependent drug information [46, 47]. For instance, Canada and some USA states have established nationally-funded academic detailing programs that rely on similar sales tactics utilized by the pharmaceutical industry to influence physician prescribing according to evidence-based guidelines [48].

Considerations should also be given to educating health care providers about the influence of interactions with the pharmaceutical industry, as well as inclusion of courses on industry marketing techniques and conflict of interest in medical curricula [49]. Existing evidence suggests positive effects of educational programs about industry marketing strategies on medical trainees’ attitudes and behaviors [50, 51].

Implications for research

Given all of the included studies were conducted in high-income countries, future studies should explore the effect of interaction of physicians in low and middle-income countries with pharmaceutical companies on their clinical practices. Also, it would be important to understand the impact of these interactions on clinical outcomes.

Supporting information

S2 Table. A list of the excluded studies along with reasons for exclusion.

https://doi.org/10.1371/journal.pone.0175493.s005

(PDF)

S3 Table. A summary of all decisions and their rationale with respect to the statistical data included in the meta-analysis.

https://doi.org/10.1371/journal.pone.0175493.s006

(PDF)

S1 Fig. Sensitivity analysis- exclusion of high risk of bias studies.

https://doi.org/10.1371/journal.pone.0175493.s007

(TIF)

S2 Fig. Sensitivity analysis-exclusion of industry-sponsored study.

https://doi.org/10.1371/journal.pone.0175493.s008

(TIF)

S3 Fig. Sensitivity analysis- exclusion of studies with indirect measures of outcomes.

https://doi.org/10.1371/journal.pone.0175493.s009

(TIF)

Acknowledgments

We would like to thank Ms. Aida Farha for her valuable help in designing the search strategy as well as the Alliance for Health Policy and Systems Research for supporting the work of our group on systematic reviews related to health policy and health systems.

We would also like to thank Andrea Darzi (MD, MPH) for editing the manuscript.

Author Contributions

  1. Conceptualization: EAA.
  2. Data curation: HB RF LA LAK HN.
  3. Formal analysis: EAA HB RF.
  4. Methodology: EAA HB RF.
  5. Project administration: HB RF FEJ LA LAK.
  6. Resources: EAA.
  7. Supervision: EAA.
  8. Validation: EAA HB RF.
  9. Writing – original draft: EAA HB RF.
  10. Writing – review & editing: EAA HB RF FEJ LA LAK HN.

References

  1. 1. The PEW Charitable Trusts. Persuading the Prescribers: Pharmaceutical Industry Marketing and its Influence on Physicians and Patients. 2013. Accessed from http://www.pewtrusts.org/en/research-and-analysis/fact-sheets/2013/11/11/persuading-the-prescribers-pharmaceutical-industry-marketing-and-its-influence-on-physicians-and-patients.
  2. 2. Lexchin J: Are drugs too expensive in Canada? Yes. Can Fam Physician 2006, 52:573–576, 578–581.
  3. 3. Gagnon MA, Lexchin J: The cost of pushing pills: a new estimate of pharmaceutical promotion expenditures in the United States. PLoS Med 2008, 5(1):e1. pmid:18177202
  4. 4. Code on Interactions with Healthcare Professionals [http://www.phrma.org/code_on_interactions_with_healthcare_professionals]
  5. 5. Rutledge P, Crookes D, McKinstry B, Maxwell SR: Do doctors rely on pharmaceutical industry funding to attend conferences and do they perceive that this creates a bias in their drug selection? Results from a questionnaire survey. Pharmacoepidemiol Drug Saf 2003, 12(8):663–667. pmid:14762982
  6. 6. Morgan MA, Dana J, Loewenstein G, Zinberg S, Schulkin J: Interactions of doctors with the pharmaceutical industry. J Med Ethics 2006, 32(10):559–563. pmid:17012493
  7. 7. Gross GA, Ference JD: Pharmaceutical representative rounds: Teaching resident physician-pharmaceutical representative interactions. Osteopathic Family Physician 2011, 3(3):123–126.
  8. 8. Othman N, Vitry A, Roughead EE: Quality of pharmaceutical advertisements in medical journals: A systematic review. PLoS ONE 2009, 4(7).
  9. 9. Wazana A: Physicians and the pharmaceutical industry: is a gift ever just a gift? Jama 2000, 283(3):373–380. pmid:10647801
  10. 10. Ziegler MG, Lew P, Singer BC: The accuracy of drug information from pharmaceutical sales representatives. Jama 1995, 273(16):1296–1298. pmid:7715044
  11. 11. Spurling GK, Mansfield PR, Montgomery BD, Lexchin J, Doust J, Othman N et al: Information from pharmaceutical companies and the quality, quantity, and cost of physicians' prescribing: a systematic review. PLoS Med 2010, 7(10):e1000352. pmid:20976098
  12. 12. Pedan A, Wu H: Asymmetric responsiveness of physician prescription behavior to drug promotion of competitive brands within an established therapeutic drug class. Health Mark Q 2011, 28(2):133–154. pmid:21590561
  13. 13. Chressanthis GA, Khedkar P, Jain N, Poddar P, Seiders MG: Can Access Limits on Sales Representatives to Physicians Affect Clinical Prescription Decisions? A Study of Recent Events With Diabetes and Lipid Drugs. Journal of Clinical Hypertension 2012, 14(7):435–446. pmid:22747616
  14. 14. Lieb K, Scheurich A: Contact between doctors and the pharmaceutical industry, their perceptions, and the effects on prescribing habits. PLoS One 2014, 9(10):e110130. pmid:25330392
  15. 15. DeJong C, Aguilar T, Tseng CW, Lin GA, Boscardin WJ, Dudley RA: Pharmaceutical Industry-Sponsored Meals and Physician Prescribing Patterns for Medicare Beneficiaries. JAMA Intern Med 2016, 176(8):1114–1110. pmid:27322350
  16. 16. Yeh JS, Franklin JM, Avorn J, Landon J, Kesselheim AS: Association of Industry Payments to Physicians With the Prescribing of Brand-name Statins in Massachusetts. JAMA Intern Med 2016, 176(6):763–768. pmid:27159336
  17. 17. Sondergaard J, Vach K, Kragstrup J, Andersen M: Impact of pharmaceutical representative visits on GPs' drug preferences. Fam Pract 2009, 26(3):204–209. pmid:19273463
  18. 18. Hurley MP, Stafford RS, Lane AT: Characterizing the relationship between free drug samples and prescription patterns for acne vulgaris and rosacea. JAMA dermatology 2014, 150(5):487–493. pmid:24740450
  19. 19. Pinckney RG, Helminski AS, Kennedy AG, Maclean CD, Hurowitz L, Cote E: The effect of medication samples on self-reported prescribing practices: a statewide, cross-sectional survey. J Gen Intern Med 2011, 26(1):40–44. pmid:20809157
  20. 20. Alkhaled L, Kahale L, Nass H, Brax H, Fadlallah R, Badr K et al: Legislative, educational, policy and other interventions targeting physicians' interaction with pharmaceutical companies: a systematic review. BMJ open 2014, 4(7):e004880. pmid:24989618
  21. 21. Guyatt GH, Oxman AD, Vist G, Kunz R, Brozek J, Alonso-Coello Pet al: GRADE guidelines: 4. Rating the quality of evidence—study limitations (risk of bias). J Clin Epidemiol 2011, 64(4):407–415. pmid:21247734
  22. 22. Fincham JE: Response rates and responsiveness for surveys, standards, and the Journal. Am J Pharm Educ 2008, 72(2):43. pmid:18483608
  23. 23. Sterne JA, Sutton AJ, Ioannidis JP, Terrin N, Jones DR, Lau J et al: Recommendations for examining and interpreting funnel plot asymmetry in meta-analyses of randomised controlled trials. BMJ 2011, 343:d4002. pmid:21784880
  24. 24. Iorio A, Spencer FA, Falavigna M, Alba C, Lang E, Burnand Bet al: Use of GRADE for assessment of evidence about prognosis: rating confidence in estimates of event rates in broad categories of patients. BMJ 2015, 350:h870. pmid:25775931
  25. 25. Caamano F, Figueiras A, Gestal-Otero JJ: Influence of commercial information on prescription quantity in primary care. Eur J Public Health 2002, 12(3):187–191. pmid:12232957
  26. 26. Figueiras A, Caamano F, Gestal-Otero JJ: Influence of physician's education, drug information and medical-care settings on the quality of drugs prescribed. Eur J Clin Pharmacol 2000, 56(9–10):747–753. pmid:11214787
  27. 27. Bowman MA, Pearle DL: Changes in drug prescribing patterns related to commercial company funding of continuing medical education. J Contin Educ Health Prof 1988, 8(1):13–20. pmid:10294441
  28. 28. Miller DP, Mansfield RJ, Woods JB, Wofford JL, Moran WP: The impact of drug samples on prescribing to the uninsured. South Med J 2008, 101(9):888–893. pmid:18708971
  29. 29. Orlowski JP, Wateska L: The effects of pharmaceutical firm enticements on physician prescribing patterns. There's no such thing as a free lunch. Chest 1992, 102(1):270–273. pmid:1623766
  30. 30. Chren MM, Landefeld CS: Physicians' behavior and their interactions with drug companies. A controlled study of physicians who requested additions to a hospital drug formulary. Jama 1994, 271(9):684–689. pmid:8309031
  31. 31. Haayer F: Rational prescribing and sources of information. Social science & medicine 1982, 16(23):2017–2023.
  32. 32. Symm B AM, Forjuob SN, Preece : Effects of using free sample medications on the prescribing practices of family physicians. J Am Board Fam Med 2006, 19:443–449. pmid:16951293
  33. 33. Anderson BL, Silverman GK, Loewenstein GF, Zinberg S, Schulkin J: Factors associated with physicians' reliance on pharmaceutical sales representatives. Acad Med 2009, 84(8):994–1002. pmid:19638762
  34. 34. Becker MH, Stolley PD, Lasagna L, McEvilla JD, Sloane LM: Correlates of physicians' prescribing behavior. Inquiry 1972, 9(3):30–42. pmid:4262972
  35. 35. Mizik N JR: Are Physicians "Easy Marks"? Quantifying the Effects of Detailing and Sampling on New Prescriptions. Management Science 2004, 50(12):1704–1715.
  36. 36. Muijrers PE, Grol RP, Sijbrandij J, Janknegt R, Knottnerus JA: Differences in prescribing between GPs: impact of the cooperation with pharmacists and impact of visits from pharmaceutical industry representatives. Fam Pract 2005, 22(6):624–630. pmid:16055474
  37. 37. Peay MY, Peay ER: The role of commercial sources in the adoption of a new drug. Social science & medicine 1988, 26(12):1183–1189.
  38. 38. Grande D: Limiting the influence of pharmaceutical industry gifts on physicians: self-regulation or government intervention? J Gen Intern Med 2010, 25(1):79–83. pmid:19756874
  39. 39. Rosenthal MB, Mello MM: Sunlight as disinfectant—new rules on disclosure of industry payments to physicians. N Engl J Med 2013, 368(22):2052–2054. pmid:23718162
  40. 40. Othman N, Vitry AI, Roughead EE, Ismail SB, Omar K: Medicines information provided by pharmaceutical representatives: a comparative study in Australia and Malaysia. BMC Public Health 2010, 10:743. pmid:21118551
  41. 41. Mintzes B, Lexchin J, Sutherland JM, Beaulieu MD, Wilkes MS, Durrieu G et al: Pharmaceutical sales representatives and patient safety: a comparative prospective study of information quality in Canada, France and the United States. J Gen Intern Med 2013, 28(10):1368–1375. pmid:23558775
  42. 42. Silversides A: Tight regulation of French drug reps mean French doctors get more balanced information than doctors in the US. BMJ 2010, 341:c6964. pmid:21131344
  43. 43. 2016 AMAMAnCoC-wimfmp.
  44. 44. Spithoff S: Industry involvement in continuing medical education: time to say no. Can Fam Physician 2014, 60(8):694–696, 700–693. pmid:25122806
  45. 45. Zetterqvist AV, Mulinari S: Misleading advertising for antidepressants in Sweden: a failure of pharmaceutical industry self-regulation. PLoS One 2013, 8(5):e62609. pmid:23650519
  46. 46. Ostini R, Hegney D, Jackson C, Williamson M, Mackson JM, Gurman K et al: Systematic review of interventions to improve prescribing. Ann Pharmacother 2009, 43(3):502–513. pmid:19261953
  47. 47. O'Brien MA, Rogers S, Jamtvedt G, Oxman AD, Odgaard-Jensen J, Kristoffersen DT et al: Educational outreach visits: effects on professional practice and health care outcomes. Cochrane Database Syst Rev 2007(4):CD000409. pmid:17943742
  48. 48. The Hilltop Institute. Academic Detailing: A Review of the Literature and States’Approaches. 2009. Accessed from: http://www.hilltopinstitute.org/publications/AcademicDetailingReviewOfTheLiteratureAndStates'Approaches-December2009.pdf.
  49. 49. Norris P, Herxheimer A., Lexchin J., & Mansfield P.: Drug promotion. What we know, what we have yet to learn Reviews of materials in the WHO/HAI database on drug promotion. 2005.
  50. 50. Carroll AE, Vreeman RC, Buddenbaum J, Inui TS: To what extent do educational interventions impact medical trainees' attitudes and behaviors regarding industry-trainee and industry-physician relationships? Pediatrics 2007, 120(6):e1528–1535. pmid:18055668
  51. 51. Zipkin DA, Steinman MA: Interactions between pharmaceutical representatives and doctors in training. A thematic review. J Gen Intern Med 2005, 20(8):777–786. pmid:16050893