An Evidence-Based Personal Finance Education Curriculum for Physicians and Medical Students

Many physicians complete medical school and graduate medical education (GME) burdened by high debt and nancial illiteracy. This places them at higher risk for ill-informed nancial decisions, which can result in increased stress and anxiety and a lower quality of life. In response, medical wellness programs have increasingly sought to offer personal nance education, but there is little guidance on optimal curricula. Our objective is to systematically review the existing literature examining physician nancial literacy curricula and to recommend a standardized curriculum. This review utilized the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) 2009 checklist to conduct literature searches in PubMed, ERIC, MedEdPortal, EBSCO, JSTOR, and Google Scholar. Three researchers used predetermined inclusion and exclusion criteria to select articles, including a focus on nancial concepts applicable in the United States. Articles were assessed using modied Côté-Turgeon and Kirkpatrick qualitative tools. 38 articles met all inclusion criteria. Six specically described personal nance literacy curricula for medical students or GME trainees, with varied criteria for selecting instructors, topics, and outcomes. All studies reported that audiences were ill-prepared for making nancial decisions but strongly desired nancial literacy education. Qualitative analysis revealed Strength of Findings summary scores ranging from 2-4, while applicable Kirkpatrick Model scores were all 3 or greater. Although students and GME trainees value nancial literacy, few publications report the impact of actual curricula. These efforts vary in depth, breadth, and measured impact. Future research should focus on development of valid testing instruments, content standardization, selection of credible instructors, and country-specic nancial concepts.

identify more relevant articles in each article's reference section by reviewing every reference in each article for additional articles that t our inclusion criteria, until saturation was achieved. This search spanned all UME and GME training programs with available publications.
All included article information was compiled in a Microsoft Excel (Microsoft Corporation, Redmond, WA) spreadsheet. Article acquisition was performed by all three co-authors with the help of an informationist, and full text analysis was performed by all authors.

Inclusion and Exclusion Criteria
The included articles satis ed the following requirements:

English language articles
Published between Jan 1, 2000 and September 1, 2020 Described curriculum or curricula Targeted UME or GME participants Since the included articles used a variety of words to describe nancial literacy topics, we categorized them using more general terms. For example, " nancial planning" referred to general nancial literacy content; "retirement accounts" was included in the "retirement planning" topic; "education debt," "credit cards," "home purchase," and "home mortgages" were all categorized into the "debt" topic; and "salary and bene ts" was included in "contract negotiation." Articles that were excluded did not t our inclusion criteria or displayed the following characteristics: Articles authored by nancial industry professionals, such as nancial advisors, insurance agents, brokers, or nancial consultants, (all of whom could have an inherent nancial interest in encouraging the sale of nancial instruments and thus harbor potential con icts of interest.) Articles whose authors appeared to have received signi cant compensation from the nancial services industry (which we determined through review of author disclosures and personal websites).
Articles that addressed retired physicians.
Articles that focused primarily on debt, a topic which has been well discussed in other articles. (14,(32)(33)(34)(35)(36)(37)(38)(39)(40)(41)(42) Modi ed Côté-Turgeon Tool and Kirkpatrick Model Descriptive statistics were employed for all included articles. Studies that implemented nancial literacy curricula were also assessed qualitatively using a BEME-guided Strength of Findings approach,(43) the Kirkpatrick Model which examined the reactions, learning, behavior, and results for each curriculum, (31) and a modi ed Côté-Turgeon tool, which critically assessed multiple characteristics for each qualitative medical education article. (30) Modi cations to these rating tools included splitting Item 7 on the Côté-Turgeon tool "Data analysis is credible," into three elds: triangulation, referring to whether researchers used multiple methods or data sources to support ndings (44); internal validity; and external validity. Additional elds assessed the depth and breadth of curricula.
Breadth considered the number of nancial literacy topics covered while depth re ected the total time allocated for the number of addressed topics.
All items in the modi ed Côté-Turgeon tool were independently ranked by each author on a scale of one to ve, with one being "poor," three being "neutral," and ve being "excellent." Strength of Findings scores were determined, with Grade 1 indicating non-signi cant ndings and Grade 5 signifying unequivocal results. For all assessment scores, inter-rater reliability was maintained by standardizing tool assessments prior to use. Reconciliation was performed if authors deviated by more than one point on any rating by discussions and score revisions. For each criterion, scores were averaged following reconciliation.
The number of con icts requiring reconciliation varied: Dhaliwal 1 con ict,(22) Boehnke 2 con icts,(2) Bar-Or 3 con icts,(28) Mizell 5 con icts,(23) Liebzeit 7 con icts, (45) and Meleca 10 con icts,(46). Of note, one of the papers was authored by two of the authors (YB and SZ).(28) This study was reviewed last, to help the authors anchor scores based on similar criteria as the other studies. To avoid conformity bias, the independent author (JI) provided rankings results rst.

Results
As summarized in Figure 1, the search strategy yielded 38 articles. In total, the initial search yielded 193 publications and of these, 155 were excluded. Articles were excluded because they focused on private practice management or business nance (50), had a primary focus on student debt (44), focused on nurses, pharmacists, chiropractors, or non-US physicians (30), primarily addressed moonlighting (8), were authored solely by nancial advisors, bankers, or others with potential con icts of interest (10), referred only tangentially to nance but focused on medical curricula (7), were only abstracts or posters (4), or addressed the needs of retired physicians (2). The remaining articles included six peer-reviewed reports of nancial literacy curricula (Table 1), 17 reports of crosssectional surveys, and 15 opinion pieces (Appendix Table A2). 18-hour curriculum spanning practice management and personal finance topics delivered by a physician who is also a certified financial planner, a surgeon with coding and reimbursement knowledge, a hospital attorney, and others. Practice management topics were delivered monthly during the daytime. Personal finance topics were delivered monthly in the evenings. Self-assessment surveys at onset and conclusion of the curriculum. Pre-and post-tests administered after each session. Level 4 Results -measures whether the material had a positive impact on participants' work and personal environments/organizations, e.g., a hospital, private practice, or household.
Most of the 38 included articles focused on GME trainees in a variety of specialty areas, although some addressed UME needs. All 38 articles (100%) reported that participants were ill-prepared to make nancial decisions and 31 articles (82%) recommended that UME or GME institutions provide nancial literacy education for their medical students and trainees.
As shown in Table 2, each of these articles addressed general nancial planning principles and/or a variable number of nance topics, such as debt/liabilities, savings/assets, investing, budgeting, money basics (e.g., time-value, discounting, and compounding), contract negotiation, selecting and interacting with nancial advisors, children's college savings plans, insurance, retirement planning, estate planning, and taxes. The amount of topics discussed in each article varied, with two articles addressing only general nancial planning,(25, 29) while four covered 10 or more topics. (16, 28, 47, 48) The most covered topic was debt (89%), followed by retirement planning (76%), budgeting (66%), savings (58%), investing (58%), and insurance (53%). Contract negotiation (11%), money basics (11%), children's college planning (11%), and estate planning (13%) were least popular.  Six studies discussed nancial literacy curricula, which were analyzed using the BEME-guided Strength of Findings approach.  Table A3 lists the reconciled scores for each criterion of the modi ed Côté-Turgeon assessment tool as well as the total score for each study. In general, individual scores ranged from 2-5, and were higher for components assessing research question de nition, study objectives, and participant selection. All studies had lower scores for components assessing validity and generalizability.

Discussion
Despite the importance attributed to personal nance literacy training by medical students and GME trainees, we found only six publications describing actual curricula. All six curricula were highly regarded by attendees, and most studies reported increased nancial knowledge and bene cial effects on nancial decision-making. While only two of the six studies assessed post-intervention nancial decision-making, there were tangible effects on retirement account choices and subjective improvements in nancial decision-making. However, none of the six studies assessed unequivocal and lasting long-term impacts. Also, the studies had issues with validation and generalizability, differences in instructor selection, and variations in curriculum selection, all of which are discussed in more detail below.

Validation and Generalizability
None of the studies utilized validated pre-or post-intervention surveys and knowledge tests. Due to this lack of validation, some questions measuring participant knowledge and attitudes could have been unclear or may have overestimated learning. In fact, one of the studies compared self-assessments with tests of nancial concepts and found a signi cant discrepancy between the two.(23) Reliable assessment of long-term knowledge was also uncertain, because the two studies that reported long-term outcomes lacked controls. (2,22) and taxes. In contrast, fewer addressed contract negotiation, children's college planning, or estate planning. The latter topics may re ect a relative lack of interest by medical students and trainees. For example, contract negotiation is only applicable to trainees who are poised to take their rst full-time jobs.
Similarly, children's college planning and estate planning are less relevant for younger trainees, who are less likely to have families or assets requiring protection. Nonetheless, more comprehensive nancial literacy curricula included these topics, since nancial needs can quickly change due to life events.

Development of Curriculum
Since there are strong bene ts for comprehensive, standardized nancial literacy curricula for physicians, we compiled a sample curriculum encompassing all the nancial themes identi ed in this review. The recommended curriculum is described in Appendix Table A4. Learning objectives are compiled in Appendix   Table A5. The curriculum topics are all derived from the studies summarized in Table 2, and are also considered core subjects in a standard personal nancial planning textbook. (49) This model curriculum may be delivered over the course of a semester or in a more concentrated fashion. If there are time constraints, the topics labeled as optional can be dropped as necessary. Of note, the curricula can be delivered in-person or virtually learning as recommended by Shappell,(8) In particular, virtual content can be delivered either synchronously or asynchronously. The latter may be particularly advantageous due to common scheduling challenges faced by medical students and trainees. Based on prior studies, we hypothesize that the comprehensive curriculum will improve Kirkpatrick model (31) Learning scores, and an exclusion of potentially biased instructors will likely lead to better Behavior scores. Ultimately, there will also be improved Results scores.

Limitations
This review highlights a number of challenges, including a paucity of validated measures to evaluate curricula and the lack of higher-level Kirkpatrick (Results) scores. In addition, it was di cult to compare the relative merits of the six studies due to varied depths of reporting, especially for two studies that were brief reports.(2, 45) Furthermore, participants varied from medical students to residents and fellows and spanned many specialties. Consequently, the resulting curricula were heterogenous, shaped by a variety of scheduling challenges, time constraints, and objectives. Also, we were unable to assess the quality or accuracy of the content offered to participants. We could not observe the researchers in action, so our Côté-Turgeon and Kirkpatrick Model assessments are based only on what we were able to glean from the written word.
Finally, our review may have excluded relevant articles by limiting our attention to physicians and medical trainees, even though dentists, nurses, and pharmacists, to name a few professions, have similar personal nance concerns. Also, we omitted articles from outside the US which may have useful recommendations, including two high-quality Canadian studies. Although some terminology and accounting rules differ across national jurisdictions, concepts such as nancial planning, saving, investing, budgeting, taxation, retirement planning, insurance coverage, estate planning, and dealing with nancial advisors are important for professionals worldwide.

Future Research
Further research is critical to determine the most effective modes of instruction and assessment. Quantifying knowledge retention is particularly important because the value of nancial literacy interventions dissipates over time. (52) Also, rigorous longitudinal studies are needed to quantify the impact of nancial literacy curricula on decision-making and wellness; and to explore whether alternative instructional modalities, such as webinars or asynchronous online presentations, are more effective than traditional in-person instruction. Identi cation of effective alternative delivery methods is especially important because novel modalities could help overcome scheduling challenges.

Conclusions
Our study is the rst review of existing research on personal nance literacy for medical students and physician trainees. It is also the rst study that proposes a curriculum guided by extensive review of published curricula. Our review con rms strong interest in nancial literacy among medical students and physician