Abstract
Background
Overuse of medical care is a pervasive problem. Studies using hypothetical scenarios suggest that physicians’ risk literacy influences medical decisions; real-world correlations, however, are lacking. We sought to determine the association between physicians’ risk literacy and their real-world prescriptions of potentially hazardous drugs, accounting for conflicts of interest and perceptions of benefit–harm ratios in low-value prescribing scenarios.
Setting and sample
Cross-sectional study—conducted online between June and October 2023 via field panels of Sermo (Hamburg, Germany)—with a convenience sample of 304 English general practitioners (GPs).
Methods
GPs’ survey responses on their treatment-related risk literacy, conflicts of interest and perceptions of the benefit–harm ratio in low-value prescribing scenarios were matched to their UK National Health Service records of prescribing volumes for antibiotics, opioids, gabapentin and benzodiazepines and analysed for differences.
Results
204 GPs (67.1%) worked in practices with ≥6 practising GPs and 226 (76.0%) reported 10–39 years of experience. Compared with GPs demonstrating low risk literacy, GPs with high literacy prescribed fewer opioids (mean (M): 60.60 vs 43.88 prescribed volumes/1000 patients/6 months, p=0.016), less gabapentin (M: 23.84 vs 18.34 prescribed volumes/1000 patients/6 months, p=0.023), and fewer benzodiazepines (M: 17.23 vs 13.58 prescribed volumes/1000 patients/6 months, p=0.037), but comparable volumes of antibiotics (M: 48.84 vs 40.61 prescribed volumes/1000 patients/6 months, p=0.076). High-risk literacy was associated with lower conflicts of interest (ϕ = 0.12, p=0.031) and higher perception of harms outweighing benefits in low-value prescribing scenarios (p=0.007). Conflicts of interest and benefit–harm perceptions were not independently associated with prescribing behaviour (all ps >0.05).
Conclusions and relevance
The observed association between GPs with higher risk literacy and the prescription of fewer hazardous drugs suggests the importance of risk literacy in enhancing patient safety and quality of care.
Keywords: Evidence-based medicine, Decision making, Healthcare quality improvement, Medical education, Patient safety
WHAT IS ALREADY KNOWN ON THIS TOPIC
The pervasive issue of overusing medical care, particularly in the context of potentially hazardous drugs, poses a threat to patients’ safety and quality of care. Previous studies using hypothetical scenarios indicate that physicians’ risk literacy can influence the propensity for low-value prescribing, but correlations with real-world prescriptions are lacking.
WHAT THIS STUDY ADDS
In our cross-sectional study involving 304 English general practitioners (GPs) and their National Health Service prescription data, we observed that GPs with lower risk literacy were considerably more likely to prescribe potentially hazardous drugs such as opioids or benzodiazepines compared with GPs with higher risk literacy. In addition, GPs with lower risk literacy reported more conflicts of interest and more often misevaluated the benefit–harm ratio of drugs in low-value prescribing scenarios.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Given the associations between GPs’ levels of risk literacy and overprescribing potentially hazardous drugs teaching programmes on risk in medical school and ongoing medical training should be made mandatory in order to foster safer patient care.
Background
The provision of too much medicine—which is likely to cause more harm than good—is a pervasive problem in high-income countries.1 Directly measuring medicine overuse is challenging due to the difficulty of defining appropriate care for patients with individual preferences and needs. Indirect approaches via examinations of variations in prevalence of procedures, prescriptions and intensity of care, however, suggest that high-income countries face high rates of overuse across a wide range of services and prescriptions.1 2 Overuse can detrimentally impact patients’ health, both physically and psychologically, and strain the healthcare system by squandering resources and funds that could be more effectively allocated elsewhere.
Past research indicates that physicians’ level of medical risk literacy,3,10 and, as a variant of risk literacy, their numeracy,11,13 can considerably influence their recommendations and decisions. Medical risk literacy refers to the cognitive ability to understand and interpret numerical statistical information (eg, relative vs absolute risk) related to medical interventions. While these studies provide important insights into the role of these cognitive abilities on physicians’ judgements and decisions, their significance is limited due to the fact that they usually employ hypothetical scenarios and do not investigate real-world behaviour. Little is known about how risk literacy impacts physicians’ real-world prescribing practices, especially in the context of potentially hazardous drugs like antibiotics, opioids, gabapentin and benzodiazepines. In Europe, a prescription from a physician, typically a general practitioner (GP), is mandatory for each of these drugs, as they carry significant risks to patient safety and health when not used appropriately.
Our study sought to determine the association between GPs’ level of risk literacy and their real-world prescribing of antibiotics, opioids, gabapentin and benzodiazepines in England. Acknowledging that factors such as conflicts of interest14 15 and perceptions of benefit–harm balance in low-value prescribing scenarios16 can influence prescription patterns as well, we also investigated whether these factors independently contribute to GPs’ prescriptions of these potentially hazardous drugs.
Methods
Study design
The study reported herein is of an explorative, cross-sectional design. We used the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline for our reporting. This study was approved by the Institutional Review Board of Charité – Universitätsmedizin Berlin (EA1/360/21). Written informed consent, granted by the Institutional Ethics Review Board of (name of institution), was obtained online from all participants at the study outset.
Participants and recruitment
A cross-sectional national convenience sample of 304 English GPs, which approximates the distribution in years of profession of the general population of GPs in England (online supplemental file 1), was drawn from an established internet physician panel maintained by Sermo (Hamburg, Germany) between June and October 2023. The panel contains about 7500 verified English GPs out of approximately 45 000 actively practising GPs and is representative in terms of age, gender and years in profession. Physicians are recruited to the panel via a multichannel recruitment strategy that includes email marketing, online campaigns, live events, recommendations from other physicians and classic partnership marketing. Before being accepted to the panel, physicians undergo a rigorous identification check that includes verification as a licensed physician and identity validation.
To accurately gauge the potential impact of risk literacy on the entirety of prescribing volume variation, we sought to balance the proportion of GPs falling into the categories of ‘low’, ‘medium’ and ‘high’ prescribers. We, therefore, applied a multiple-step procedure. In March 2023, we gathered National Health Service (NHS) data on the prescribing volumes of antibiotics, opioids, gabapentin and benzodiazepines for the period between July and December 2022 from OpenPrescribing.net, encompassing all active English practices during that timeframe. We then computed the monthly average prescribing volume, adjusted for patient numbers in each practice and z-standardised the data while addressing outliers. Z-standardised data were used to create a composite score that informed the creation of three prescription quotas (low, medium and high). Sermo was commissioned to sample about 100 GPs for each of the three prescription quotas. To match GPs with the quota and their prescribing data, Sermo required participants to enter their NHS practice code or their address as included in the OpenPrescribing.net data set. If participants matched to one of the quotas and consented to participate, they proceeded to the online questionnaire.
Survey questionnaire
To measure treatment-specific risk literacy, we asked GPs seven multiple-choice questions retrieved from two validated questionnaires by Caverly et al17 and Anderson and colleagues18 (survey, online supplemental file 1). First, we retrieved only the items from the two questionnaires that were generalisable across medical disciplines (eg, we excluded items in the questionnaire by Anderson and colleagues that were specific to obstetricians and gynaecologists). Next, we aligned the response options of the items by formatting the questions into a single-choice format that presented a choice between four options: the correct answer, two incorrect answers and one ‘don’t know’ option. The questions evaluated GPs’ interpretation of treatment effectiveness when expressed in various statistical formats such as number needed to treat, absolute risk and relative risk. To avoid order effects, questions and response options were presented in a randomised order among participants. Two further items initially intended to measure screening-related risk literacy were omitted due to lack of content proximity and low correlation with the treatment-related risk literacy score.To evaluate GPs’ conflicts of interest, participants answered five questions from a questionnaire by Lieb and colleagues14 that covered topics such as frequency of visits from pharmaceutical representatives, perceived influence of pharmaceutical representatives on prescribing behaviour and trustworthiness of received drug-related information. To investigate GPs’ perceptions of benefit–harm balance in low-value prescribing scenarios, participants were presented with three scenarios where evidence suggests an unfavourable benefit–harm ratio of prescribing: antibiotics for otitis media, long-term strong opioids for chronic noncancer pain)19 20 and benzodiazepines for insomnia. After each scenario, GPs were queried on their perception of the benefit–harm balance of prescribing using a 5-point scale (‘The benefits clearly outweigh the harms’, ‘The benefits somewhat outweigh the harms’, ‘Benefits and harms are balanced’, ‘The harms somewhat outweigh the benefits’, ‘The harms clearly outweigh the benefits’). Furthermore, Sermo inquired about individual prescribing volumes per drug.
Primary and secondary endpoint measures
The primary outcome was the absolute prescribing volume (details, online supplemental file 1) per drug (antibiotics, opioids, gabapentin, benzodiazepines) per practice over 6 months per 1000 patients, adjusted to the individual proportion of the drug-specific prescribing volume per drug for the period between July and December 2022. Secondary outcomes were associations between risk literacy, conflicts of interest and benefit–harm assessments of nonevidence-based prescription scenarios.
Data handling
Risk literacy, conflicts of interest and assessment of benefit–harm balance all yielded ordinal data. To better explore and illustrate the potential absolute effect between prescribing volumes and independent variables (eg, risk literacy), we binarised the range of the potential scores for risk literacy, conflicts of interest and the assessment of benefit–harm balance in the middle of their respective score distribution. That is, for risk literacy, with a score distribution of zero to seven correct responses, we categorised 0 to 3 correct answers as ‘low risk literacy’ and 4 to 7 correct answers as ‘high risk literacy’. For conflicts of interest, we binarised each of the five questions into ‘low’ and ‘high,’ created a sum score across the five questions (minimum: 0, maximum: 5) and split that score in the middle of the distribution by classifying values of 0 to 2 as ‘low conflicts of interest’ and of 3 to 5 as ‘high conflicts of interest’. For GPs’ perception of the benefit–harm balance in low-value prescribing scenarios, responses incorrectly assuming a benefit were coded 0 and correctly assuming a harm were coded 1. We created a sum score across the three questions (minimum: 0, maximum: 3) and split that score in the middle of the distribution by classifying 0 to 1 as ‘incorrectly assuming more benefits’ and 2 to 3 as ‘correctly assuming more harm’
Analysis
To ensure that effects are robust and independent of the cut-off criteria for the split scores, all associations for the primary analyses were analysed with Kendall’s Tau rank correlation (correlation coefficient τ) for the continuous, non-parametric data (details and analyses, online supplemental file 2). To better understand and illustrate how medical risk literacy affected absolute prescribing volumes, we tested differences in individual prescribing volumes per drug between low and high risk literacy groups with a two-sample t-test and used Cohen’s d (the difference between the pooled SD from both groups) as a measure of effect size. If a Levene’s test for homogeneity of variance was at least marginally significant (p≤0.10), we used the more robust Welch two-sample t-test instead. The same analysis strategy was pursued for testing the effects of low versus high conflicts of interest on prescribing behaviour. Drug-specific prescribing behaviour (eg, antibiotics) was tested against GPs’ drug-specific perception of the benefit–harm balance in low-value prescribing scenarios (eg, antibiotics for otitis media) with a t-test. χ2 tests were used to test the associations between the split scores of risk literacy, conflicts of interest and benefit–harm perception. Wilcoxon rank sum tests were used to determine whether the numbers of patients per practice and years of experience differed between practitioners with high and low scores in risk literacy, conflicts of interest, and benefit–harm assessment of low-value care. P values were two sided, with statistical significance set at p<0.05. All data were stored and analysed utilising R basic software and the packages effect size, DescTools, psych, and car.
Results
Table 1 shows demographic characteristics of our sample of 304 GPs. The largest group (n=119; 39.1%) reported working in a practice with 6–10 GPs and had been practising for 10–19 years (n=116; 38.2%).
Table 1. Summary of demographics of English general practitioners assessed in the study.
| Total | Lower risk literacy | Higher risk literacy | P value* | |
| Sample size (%) | 304 | 116 (38.2) | 188 (61.8) | |
| Years of experience, n (%)† | 0.386 | |||
| < 10 years | 63 (20.7) | 23 (19.8) | 40 (21.3) | |
| 10–19 years | 116 (38.2) | 42 (36.2) | 74 (39.4) | |
| 20–29 years | 75 (24.7) | 26 (22.4) | 49 (26.1) | |
| 30–39 years | 40 (13.2) | 19 (16.3) | 21 (11.2) | |
| ≥ 40 years | 10 (3.3) | 6 (5.2) | 4 (2.1) | |
| Size of practice, n (%)† | 0.093 | |||
| 1 practitioner | 3 (1.0) | 2 (1.7) | 1 (0.5) | |
| 2–3 practitioners | 17 (5.6) | 7 (6.0) | 10 (5.3) | |
| 4–5 practitioners | 80 (26.3) | 21 (18.1) | 59 (31.4) | |
| 6–10 practitioners | 119 (39.1) | 53 (45.7) | 66 (35.1) | |
| > 10 practitioners | 85 (28.0) | 33 (28.4) | 52 (27.7) | |
| Patient list size, M (SD) | 0.320 | |||
| 13 215 (9972) | 12 491 (8827) | 13 663 (10 615) | ||
| Conflicts of interests, n (%)† | 0.031 | |||
| Low | 249 (81.9) | 88 (75.9) | 161 (85.6) | |
| High | 55 (18.1) | 28 (24.1) | 27 (14.4) | |
| Perception of benefit–harm ratio across all scenarios, n (%)† | 0.007 | |||
| More benefits than harms | 125 (41.1) | 59 (50.9) | 66 (35.1) | |
| More harms than benefit | 179 (58.9) | 57 (49.1) | 122 (64.9) |
χ2 tests for differences between risk literacy groups. P values are two-sided, with satistical significance set at P < 0.05. Percentages are rounded and may not add up to 100.
Percentages are rounded and may not add up to 100.
Risk literacy and prescribing behaviour
GPs’ risk literacy was significantly associated with prescribing volumes for opioids (τ=–0.14, p<0.001), gabapentin (τ=–0.11, p<0.01) and benzodiazepine (τ=–0.14, p<0.001), but not for antibiotics (τ=–0.06, p=0.131). Binarising medical risk literacy across the sample, we observed that 38.8% of GPs (n=116) demonstrated low risk literacy and 61.8% (n=188) demonstrated high risk literacy (overall distribution of scores, onlinesupplemental files 1 2). Compared with GPs with low risk literacy (figure 1A), GPs with high risk literacy prescribed lower volumes of opioids (mean (M): 60.60 vs 43.88 per 1000 patients over 6 months; p=0.016, d=0.31), lower volumes of gabapentin (M: 23.84 vs 18.34/1000 patients/6 months; p=0.023, d=0.27), lower volumes of benzodiazepines (M: 17.23 vs 13.58/1000 patients/6 months; p=0.037, d=0.25) and comparable volumes of antibiotics (M: 48.84 vs 40.61/1000 patients/6 months; p=0.076, d=0.23).
Figure 1. Mean differences in GPs’ NHS-recorded prescribing volumes of antibiotics, opioids, gabapentin and benzodiazepines, measured over 6 months, adjusted by patient size of GPs’ practice and their individual proportion of drug-specific prescriptions in association to their (A) level of risk literacy, (B) level of conflicts of interest and (C) perception of harms in drug-specific low-value prescription scenarios (excluding gabapentin). Error bars show standard errors (SE) of the means. *Two-sided significance at p<0.05. GPs, general practitioners; NHS, National Health Service.
Conflicts of interest and prescribing behaviour
Most GPs (91.4%; n=278) reported taking no gifts from pharmaceutical representatives, not giving paid interviews (95.1%; n=289) and seeing a pharmaceutical representative less than once a month to never (76.3%; n=232). Just over half (54.9%; n=167), however, regarded themselves as ‘frequently to always’ receiving adequate and accurate information from their pharmaceutical representatives, and 35.9% (n=109) reported that their prescribing behaviour is ‘frequently to always’ influenced by pharmaceutical representatives’ advice. Differences in the level of conflicts of interest were not associated with differences in the prescribed volumes of any of the four drug types (antibiotics: p=0.489; opioids: p=0.873; gabapentin: p=0.942, benzodiazepines: p=0.197; figure 1B).
Perceptions of benefit–harm balance for low-value prescriptions and prescribing behaviour
For the low-value prescription of antibiotics to patients presenting with otitis media, most GPs (79.3%; n=241) believed that antibiotics would have more benefits than harms. Differences in that assessment were not associated with differences in GPs’ prescribed volumes of antibiotics (p=0.369). For prescribing strong opioids’ long term to patients presenting with chronic non-cancer pain, the majority of GPs (62.8%; n=191) knew that the harms outweighed the benefits. These differences were not associated with GPs’ prescribed volume of opioids (p=0.557). The low-value prescription of benzodiazepines for insomnia was also perceived as more harmful than beneficial by most GPs (71.4%; n=217), with no association with prescribing volumes of benzodiazepines (p=0.567; see figure 1C).
Other associations
GPs’ risk literacy was also associated with their conflicts of interest (τ=–0.11, p=0.013) and their perceptions of the benefit–harm balance for low-value prescriptions (τ=0.17, p<0.001). Compared with GPs with low risk literacy, GPs with high risk literacy more often had low conflicts of interest (χ2(1)=4.63, ϕ=0.12, p=0.031) and more often perceived that the harms outweighed the benefits in the low-value prescription scenarios (χ2(1)=7.36, ϕ=0.16, p=0.007). Similarly, compared with GPs with high conflicts of interest, GPs with low conflicts of interest were more likely to believe that the harms outweighed the benefits in the low-value prescription scenarios (χ2(1)=5.00, ϕ=0.13, p=0.025). GPs’ risk literacy and their benefit–harm perceptions were not associated with the number of colleagues working in their practice (PRiskLiteracy=0.261; PCOI=0.322; PAssessment=0.653) or their years of experience (PRiskLiteracy=0.260; PAssessment=0.566). However, there was an association between GPs’ conflicts of interest and their years of experience: GPs with ≤19 years of experience were significantly less likely to report conflicts of interest than those with ≥20 years of experience (χ2(1)=6.45, ϕ=0.15, p=0.011).
Discussion
Doing too much in medicine constitutes a significant concern in healthcare systems worldwide.1 21 22 Excessive prescriptions of drugs and administration of medical services can lead to unnecessary healthcare costs, adverse effects and harm to patients, particularly in the case of high-risk drugs. In this cross-sectional study of 304 English GPs, we observed that those with low risk literacy were significantly more inclined than those with high risk literacy to prescribe more opioids, gabapentin and benzodiazepines. These findings demonstrate the impact that physicians’ comprehension and integration of medical statistics can have on prescription practices and unwarranted variation of care.23 Our results are consistent with prior research highlighting inadequate levels of risk literacy among certain numbers of physicians.34 6,10 They also reinforce earlier studies, primarily employing hypothetical scenarios, which underscore the substantial influence of physicians’ risk literacy and numeracy on communication with patients,12 screening recommendations35,11 24 and treatment evaluation.4 18 25
We did not detect a similar effect of risk literacy on prescribing antibiotics, and we can only speculate on why this was the case. One reason might be that for over a decade now, the NHS has invested in numerous educational and awareness campaigns aimed at reducing unnecessary antibiotic prescriptions in response to the escalating concern over antibiotic resistance. These broad efforts, targeting both healthcare professionals and the public, have led to a widespread awareness of antibiotic resistance that may have fostered a collective effort by both physicians and patients to curtail unnecessary antibiotic prescriptions. While the NHS and other health organisations have recently also been promoting safe prescribing practices for opioids, these initiatives are less prominent and widespread as those for antibiotics. Another reason might be that antibiotics are commonly prescribed preemptively (‘delayed prescription’) to patients with mild symptoms as a precautionary measure due to challenges in accessing healthcare promptly, allowing patients to fill the prescription if their symptoms rapidly worsen rather than wait for another appointment. Healthcare systems that cannot ensure timely care may offset the potential impact of GPs’ risk literacy on their prescribing practices.
We did not find an independent association between GPs’ reported conflicts of interest and their prescriptions, which contrasts with the findings of some other studies.14 15 26 The study by Lieb and Scheurich14 that established such a relationship among German GPs found a considerably higher level of conflicts of interest than in our English sample: while 98% of German GPs reported seeing their pharmaceutical representatives at least once a week and 69% reported frequently accepting gifts from pharmaceutical representatives, rates among the GPs in our sample were much lower (14.8% and 8.6%, respectively). DeJong and colleagues15 also reported an association between pharmaceutical industry-sponsored meals and physicians’ prescribing patterns of statins, beta-blockers and other drugs for Medicare beneficiaries in the USA. Their analyses were based on payment data from the Open Payment Program, which provided explicit information on whether a meal promoted a specific brand-name drug. Because we assessed only general sponsored dinner participation, which does not allow for establishing a direct link between a dinner invitation and a particular drug, we may have underestimated the effect of conflicts of interest on prescribing behaviour. However, we found an association between GPs’ conflicts of interest and their risk literacy: GPs who were better at interpreting medical statistics appeared to be less likely to obtain information from and form relationships with pharmaceutical representatives.
Our study also suggests that GPs’ understanding of medical statistics heightens their awareness of potential harms associated with low-value prescribing practices. It is worth noting, however, that although unnecessary antibiotic use is one of the best-documented instances of medication overuse worldwide1 2 and has been the focus of numerous educational campaigns, a majority of GPs in our sample believed that the benefits of prescribing antibiotics in the low-value prescribing scenario outweighed the harms, which is in line with findings from other studies.27 28 The unwarranted positive assessment of low-value antibiotic prescription contrasts with the scenarios involving low-value prescribing for opioids and benzodiazepines, where the majority of GPs correctly perceived the harms of these drugs to outweigh the benefits.
Limitations
Our study has limitations. First, the generalisability of our results may be limited by our sample, which consisted of a convenience sample of GPs in England. Second, our study was explorative and thus does not establish causality between risk literacy and prescribing behaviour. Third, while we observed significant associations between prescriptions for opioids, gabapentin and benzodiazepines, it is noteworthy that correlations and effect sizes are small, implying the influence of additional factors on GPs’ prescribing behaviour. Fourth, while prescribing more rather than fewer drugs can be an indicator of overuse,1 it does not constitute definitive proof of overuse or non-evidence-based practice. Fifth, we had no NHS information for specific therapeutic group age–sex-related prescribing units (STAR-PU) for opioids, gabapentin and benzodiazepines, which may influence prescribing volumes. Primary analyses for antibiotic prescriptions that applied and did not apply the antibiotic-specific STAR-PU, however, left findings on the direction and size of effects unchanged.
Implications for policy and practice
Our findings have significant implications for policy and practice. In light of evidence demonstrating the effectiveness of brief and low-cost lessons in enhancing medical risk literacy,29,31 we advocate for efforts to incorporate instruction on understanding evidence, particularly health statistics, into medical training and continuing medical education. These easily implementable interventions have the potential to mitigate unnecessary prescribing. Additionally, ample evidence supports the use of transparent risk formats and visualisation decision aids727 31,33 to help physicians accurately judge the benefits and harms associated with drugs and other medical interventions. Adding visualisation aids to medical guidelines and educational materials would provide busy physicians with quick, comprehensive insights into a drug’s expected outcomes, thereby fostering a safer allocation of care.
Conclusion
Physicians worldwide provide low-value care for numerous reasons.22 Our study suggests a new and previously unexplored dimension to the problem of overuse and low-value care: physicians’ ability to correctly understand and deal with medical statistics. Given the devastating effects that unnecessary prescriptions of potentially hazardous drugs can have on patients’ health and safety, further studies are needed to investigate the generalisability of our findings in other healthcare settings and delve deeper into associations with other contributing factors (eg, barriers to timely healthcare access) in order to better understand what undermines the practice of evidence-based prescribing.
supplementary material
Footnotes
Funding: This study was funded by Deutsche Forschungsgemeinschaft (89726186).
Provenance and peer review: Not commissioned; externally peer-reviewed.
Patient consent for publication: Not applicable.
Ethics approval: This study involves human participants and was approved by Institutional Review Board (IRB) of Charité—Universitätsmedizin Berlin (EA1/360/21). Participants gave informed consent to participate in the study before taking part.
Data availability free text: Data are available upon reasonable request for replication or addition to a meta-analysis research project and only in accordance with the terms of ethics approval. The deidentified dataset is available from the corresponding author.
Correction notice: This article has been corrected since it was first published online. The licence was updated to CC-BY-NC on 31/05/24.
Contributor Information
Odette Wegwarth, Email: odette.wegwarth@charite.de.
Tammy C Hoffmann, Email: thoffman@bond.edu.au.
Ben Goldacre, Email: ben.goldacre@phc.ox.ac.uk.
Claudia Spies, Email: claudia.spies@charite.de.
Helge A Giese, Email: helge.giese@charite.de.
Data availability statement
Data are available upon reasonable request.
References
- 1.Brownlee S, Chalkidou K, Doust J, et al. Evidence for overuse of medical services around the world. Lancet. 2017;390:156–68. doi: 10.1016/S0140-6736(16)32585-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Saini V, Garcia-Armesto S, Klemperer D, et al. Drivers of poor medical care. Lancet. 2017;390:178–90. doi: 10.1016/S0140-6736(16)30947-3. [DOI] [PubMed] [Google Scholar]
- 3.Wegwarth O, Schwartz LM, Woloshin S, et al. Do physicians understand cancer screening statistics? A national survey of primary care physicians in the U.S. Ann Intern Med. 2012;156:340–9. doi: 10.7326/0003-4819-156-5-201203060-00005. [DOI] [PubMed] [Google Scholar]
- 4.Naylor CD, Chen E, Strauss B. Measured enthusiasm: does the method of reporting trial results alter perceptions of therapeutic effectiveness? Ann Intern Med. 1992;117:916–21. doi: 10.7326/0003-4819-117-11-916. [DOI] [PubMed] [Google Scholar]
- 5.Wegwarth O, Gaissmaier W, Gigerenzer G. Deceiving numbers: survival rates and their impact on doctors’ risk communication. Med Decis Making. 2011;31:386–94. doi: 10.1177/0272989X10391469. [DOI] [PubMed] [Google Scholar]
- 6.Prinz R, Feufel MA, Gigerenzer G, et al. What counselors tell low-risk clients about HIV test performance. Curr HIV Res. 2015;13:369–80. doi: 10.2174/1570162x13666150511125200. [DOI] [PubMed] [Google Scholar]
- 7.Wegwarth O, Gigerenzer G. US Gynecologists’ estimates and beliefs regarding ovarian cancer screening’s effectiveness 5 years after release of the PLCO evidence. Sci Rep. 2018;8:17181. doi: 10.1038/s41598-018-35585-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Eddy DM. In: Professional judgment A reader in clinical decision making. Dowie J, Elstein AS, editors. Cambridge University Press; 1988. Variations in physician practice: the role of uncertainty; pp. 45–59. [Google Scholar]
- 9.Wegwarth O, Pashayan N. When evidence says no: gynecologists’ reasons for (not) recommending ineffective ovarian cancer screening. BMJ Qual Saf. 2020;29:521–4. doi: 10.1136/bmjqs-2019-009854. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Caverly TJ, Prochazka AV, Binswanger IA, et al. Confusing relative risk with absolute risk is associated with more enthusiastic beliefs about the value of cancer screening. Med Decis Making. 2014;34:686–92. doi: 10.1177/0272989X14526641. [DOI] [PubMed] [Google Scholar]
- 11.Petrova D, Kostopoulou O, Delaney BC, et al. Strengths and gaps in physicians' risk communication: a scenario study of the influence of numeracy on cancer screening communication. Med Decis Making. 2018;38:355–65. doi: 10.1177/0272989X17729359. [DOI] [PubMed] [Google Scholar]
- 12.Anderson BL, Obrecht NA, Chapman GB, et al. Physicians' communication of down syndrome screening test results: the influence of physician numeracy. Genet Med. 2011;13:744–9. doi: 10.1097/GIM.0b013e31821a370f. [DOI] [PubMed] [Google Scholar]
- 13.Hanoch Y, Miron-Shatz T, Cole H, et al. Choice, numeracy, and physicians-in-training performance: the case of Medicare part D. Health Psychol. 2010;29:454–9. doi: 10.1037/a0019881. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Lieb K, Scheurich A. Contact between doctors and the pharmaceutical industry, their prescription and the effects on prescribing habits. PLoS One. 2014;9:e110130. doi: 10.1371/journal.pone.0110130. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.DeJong C, Aguilar T, Tseng C-W, et al. Pharmaceutical industry–sponsored meals and physician prescribing patterns for Medicare beneficiaries. JAMA Intern Med. 2016;176:1114–22. doi: 10.1001/jamainternmed.2016.2765. [DOI] [PubMed] [Google Scholar]
- 16.Hoffmann TC, Del Mar C. Clinicians’ expectations of the benefits and harms of treatments, screening, and tests. JAMA Intern Med. 2017;177:407–19. doi: 10.1001/jamainternmed.2016.8254. [DOI] [PubMed] [Google Scholar]
- 17.Caverly TJ, Prochazka AV, Combs BP, et al. Doctors and numbers:an assessment of the critical risk interpretation test. Med Decis Making. 2015;35:512–24. doi: 10.1177/0272989X14558423. [DOI] [PubMed] [Google Scholar]
- 18.Anderson BL, Gigerenzer G, Parker S, et al. Statistical literacy in obstetricians and gynecologists. J Healthc Qual. 2014;36:5–17. doi: 10.1111/j.1945-1474.2011.00194.x. [DOI] [PubMed] [Google Scholar]
- 19.Nury E, Schmucker C, Nagavci B, et al. Efficacy and safety of strong opioids for chronic noncancer pain and chronic low back pain: a systematic review and meta-analyses. PAIN. 2022;163:610–36. doi: 10.1097/j.pain.0000000000002423. [DOI] [PubMed] [Google Scholar]
- 20.Wegwarth O, Spies C, Ludwig W-D, et al. Educating physicians on strong opioids by descriptive versus simulated-experience formats: a randomized controlled trial. BMC Med Educ. 2022;22:741. doi: 10.1186/s12909-022-03797-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Redberg RF. Informed strategies for treating coronary disease: comment on ‘initial coronary stent implantation with medical therapy vs medical therapy alone for stable coronary artery disease’. Arch Intern Med. 2012;172:321. doi: 10.1001/archinternmed.2011.2313. [DOI] [PubMed] [Google Scholar]
- 22.Grady D, Redberg RF. Less is more: how less health care can result in better health. Arch Intern Med. 2010;170:749–50. doi: 10.1001/archinternmed.2010.90. [DOI] [PubMed] [Google Scholar]
- 23.Wennberg J. Oxford University Press; 2010. Tracking medicine: a researcher's quest to understand health care. [Google Scholar]
- 24.Schwartz LM, Woloshin S, Black WC, et al. The role of numeracy in understanding the benefit of screening mammography. Ann Intern Med. 1997;127:966–72. doi: 10.7326/0003-4819-127-11-199712010-00003. [DOI] [PubMed] [Google Scholar]
- 25.Garcia-Retamero R, Cokely ET, Wicki B, et al. Improving risk literacy in surgeons. Patient Educ Couns. 2016;99:1156–61. doi: 10.1016/j.pec.2016.01.013. [DOI] [PubMed] [Google Scholar]
- 26.Goldacre B, Reynolds C, Powell-Smith A, et al. Do doctors in dispensing practices with a financial conflict of interest prescribe more expensive drugs? A cross-sectional analysis of English primary care prescribing data. BMJ Open. 2019;9:e026886. doi: 10.1136/bmjopen-2018-026886. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Hoffmann TC, Jones M, Glasziou P, et al. A brief shared decision-making intervention for acute respiratory infections on antibiotic dispensing rates in primary care: a cluster randomized trial. Ann Fam Med. 2022;20:35–41. doi: 10.1370/afm.2755. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Grossman Z, del Torso S, Hadjipanayis A, et al. Antibiotic prescribing for upper respiratory infections: European primary Paediatricians' knowledge, attitudes and practice. Acta Paediatr. 2012;101:935–40. doi: 10.1111/j.1651-2227.2012.02754.x. [DOI] [PubMed] [Google Scholar]
- 29.Jenny MA, Keller N, Gigerenzer G. Assessing minimal medical statistical literacy using the quick risk test: a prospective observational study in Germany. BMJ Open. 2018;8:e020847. doi: 10.1136/bmjopen-2017-020847. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Buljan I, Tokalić R, Marušić M, et al. Health numeracy skills of medical students:cross-sectional and controlled before-and-after study. BMC Med Educ. 2019;19:467. doi: 10.1186/s12909-019-1902-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.McDowell M, Gigerenzer G, Wegwarth O, et al. Effect of Tabular and icon fact box formats on comprehension of benefits and harms of prostate cancer screening: a randomized trial. Med Decis Making. 2019;39:41–56. doi: 10.1177/0272989X18818166. [DOI] [PubMed] [Google Scholar]
- 32.Schwartz LM, Woloshin S, Welch HG. Using a drug facts box to communicate drug benefits and harms. Ann Intern Med. 2009;150:516–27. doi: 10.7326/0003-4819-150-8-200904210-00106. [DOI] [PubMed] [Google Scholar]
- 33.Akl EA, Oxman AD, Herrin J, et al. Using alternative statistical formats for presenting risks and risk reductions. Cochrane Database Syst Rev. 2011;2011:CD006776. doi: 10.1002/14651858.CD006776.pub2. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data are available upon reasonable request.