Influences on Primary Care Provider Imaging for a Hypothetical Patient With Low Back Pain .
Hai H. Le: Matthew DeCamp: Amanda Bertram; Minal Kale; Zackary Berger. South Med J. 2018;111(12):758-762.
Abstract and Introduction
Objective: How outside factors affect physician decision making remains an open question of vital importance. We sought to investigate the importance of various influences on physician decision making when clinical guidelines differ from patient preference.
Methods: An online survey asking 469 primary care providers (PCPs) across four practice sites whether they would order magnetic resonance imaging for a patient with uncomplicated back pain. Participants were randomized to one of four scenarios: a patient’s preference for imaging (control), a patient’s preference plus a colleague’s opinion against imaging (colleague), a patient’s preference plus a professional society’s recommendation against imaging (profession), or a patient’s preference plus an accountable care organization’s quality metric that measures physician use of imaging (ACO). Demographic information and the reasoning behind participants’ decisions also were obtained.
Results: A total of 168 PCPs completed the survey, yielding a 36% completion rate. A majority chose not to pursue imaging: control 68%, colleague 85%, profession 87%, and ACO 78%. Multivariate logistic regression revealed that participants were more likely not to order advanced imaging only when reminded of a professional society recommendation (P = 0.017). Regression also suggested that practice site exerted an effect on the primary outcome. Evidence-based medicine and clinical judgment were the most cited reasons for the decision.
Conclusions: Our results reinforce the potential to leverage professional societies to advance evidence-based medicine and reduce unnecessary testing. At the same time, practice site appeared to exert influence, suggesting that these recommendations must be part of local institutional culture to be effective.
Up to 80% of individuals experience one or more episodes of low back pain in their lifetime, one of the most common presenting symptoms to physicians’ offices. The prevalence of patients with chronic back pain has steadily increased, leading to significant disease burden and socioeconomic impact; back pain is the second leading cause of disability in the United States. Despite clear clinical guidelines recommending against advanced imaging in uncomplicated back pain, patients may request such imaging as reassurance.
On average, advanced imaging such as magnetic resonance imaging (MRI) for uncomplicated back pain does not reassure patients or improve outcomes; however, a study of primary care providers (PCPs) demonstrates significant variation in practice. As many as 53% of providers continue to pursue imaging in uncomplicated back pain, despite the Choosing Wisely recommendations and “do-not-do” lists devised by multiple professional societies to improve care and cost-effectiveness.
Although physician preferences and specialty are associated with frequency of imaging,[9,10] the degree to which other factors contribute to practice variation, such as geographical location of practice, length of practice, individual physician style, professional guidelines, colleagues’ input, or patient preferences, remains unclear. We sought to examine factors that may be associated with the practices of PCPs related to ordering imaging in uncomplicated low back pain, reasoning that understanding these factors can advance evidence-based medicine and reduce unnecessary testing. We hypothesize that most PCPs would not order advanced imaging and that younger physicians, having less experience and thus less confidence in their clinical decisions compared with senior providers, would rely more on diagnostic tests compared with their older counterparts. Furthermore, we anticipate that geographical location, patient preference, and recommendations by various entities would exert a detectable difference as well.
We fielded an online cross-sectional survey of PCPs at four practice sites via Qualtrics (Provo, UT). Of these, site 1 and site 3 were primary care clinics at academic centers (Johns Hopkins Medical Institutions and Mount Sinai Hospital, respectively), site 2 was a community practice affiliated with an academic center (Johns Hopkins Medical Institutions), and site 4 was part of a concierge private practice group (One Medical Group). Recruitment e-mails were sent to 469 internal medicine PCPs, which represented all of the PCPs at the four sites and included providers with MDs (doctor of medicine degree), DOs (doctor of osteopathic medicine), PAs (physician assistants), and NPs (nurse practitioners). Participation was voluntary, and a $5 Amazon gift card was used to encourage participation. Three reminder e-mails were sent (one original and two reminders) in a 1-month period in 2016. The institutional review board for each site approved the study.
The survey underwent several stages of iterative development among the coauthors. An initial draft outlined the conceptual framework on the basis of the primary research question regarding practice and practitioner factors that may be associated with the likelihood of ordering advanced imaging for uncomplicated lower back pain. The question stem was based on common presentations of uncomplicated back pain and informed by clinical expertise. Response options, reason for the decision, and different contexts that may influence such a decision were decided after extensive literature review and a pilot test with a convenience sample of PCPs for face and content validity. The final version was based on iterative discussion and redrafting by the study team until consensus was reached.
The survey collected the following demographic information: age, sex, number of years since completion of medical training, professional role or degree (MD/DO, PA, NP), and proportion of time spent per week in direct patient contact. Our hypothesis was that these characteristics would be associated with the primary outcome, the PCP’s decision regarding imaging.
Participants, acting as the PCP in the survey, were randomized to receive one of four versions of a clinical vignette in which a patient presents with uncomplicated back pain and seeks advanced imaging (ie, MRI). Three possible decisions were order the MRI, not order the MRI, or not sure. The four scenarios were a patient’s preference for imaging (control), a patient’s preference for imaging plus a colleague’s opinion against imaging (colleague), a patient’s preference for imaging plus a professional society’s recommendation against imaging (profession), or a patient’s preference for imaging plus an accountable care organization’s quality metric that measures physician use of imaging in uncomplicated back pain (ACO) (Supplemental Figure, http://links.lww.com/SMJ/A124). Finally, the survey also asked participants to provide a reason for their decision: evidence-based medicine, patient-centered care, clinical judgment, outside recommendation or guidelines, or other.
Our primary outcome was the proportion of PCPs who would not order the MRI for the patient with uncomplicated lower back pain. To focus on whether the respondent would decide to order the MRI, we grouped “not sure” responses with “not order MRI” to clearly divide those who would order the MRI from those who would not. Comparisons between randomization groups on the outcome were calculated with Fisher exact tests. We also estimated associations among the outcome, the variation of the clinical vignette, and the characteristics of the PCPs using multivariate analysis of variance. Given a type I error of 0.05 and power of 0.8, with a detectable effect size of 0.25, we estimated a required sample size of 176; in other words, 44 PCPs per randomization group.
The χ 2 tests were conducted to determine association between the outcome (whether a PCP chooses to perform the MRI) and potential covariates. Next, a logistic regression model assessing the relation between the outcome and the exposure groups, adjusting for practice site, years since medical school graduation, sex, and percent of direct patient contact, was run and odds ratios were calculated. Given the multiple comparisons, we adopted a criterion for statistical significance of P = 0.05. A secondary χ 2 analysis also was performed to assess the association between the primary outcome (decision to image) and the reasoning for such a decision.
Of the 469 PCPs invited to participate, 181 started the survey, for a response rate of 38.6%. Of those 181, 168 completed the survey, yielding a final completion rate of 35.8%. The mean age of the study’s participants was 44.8 years, with a range of 27 to 79 (standard deviation 11.1) and the mean number of years since medical school graduation was 17.1, with a range of 2 to 53 (standard deviation 11.5). More than half of the participants were female (58.5%), and 88.9% held either an MD or DO degree (Table 1). Overall, the largest fraction (41.5%) of participants spent ≥75% of their time in direct patient care (Table 1).
|Total||Site 1||Site 2||Site 3||Site 4|
|Age, y, mean (SD)||44.8 (11.1)||50.2 (11.4)||44.8 (9.1)||40.5 (10)||33.4 (3.5)|
|Male sex, n (%)||71 (41.5)||41 (63.1)||17 (30.4)||8 (24.2)||5 (29.4)|
|Years since medical school graduation, mean (SD)||17.1 (11.5)||22.8 (11.5)||16.5 (9.8)||12.1 (10.8)||6.9 (2.7)|
|Primary degree, n (%)|
|MD/DO||152 (88.9)||64 (98.5)||48 (85.7)||31 (93.9)||9 (52.9)|
|NP||13 (7.6)||0 (0)||8 (14.3)||2 (6.2)||3 (17.6)|
|PA||6 (3.5)||1 (1.5)||0 (0)||0 (0)||5 (29.4)|
|Time spent in direct patient contact, %, n (%)|
|0||2 (1.1)||2 (3.1)||0 (0)||0 (0)||0 (0)|
|1–25||34 (19.9)||26 (40)||2 (3.6)||6 (18.2)||0 (0)|
|26–50||26 (15.2)||14 (21.5)||6 (10.7)||5 (15.2)||1 (5.9)|
|51–75||38 (22.2)||14 (21.5)||10 (17.9)||11 (33.3)||3 (17.6)|
|76–100||71 (41.5)||9 (13.8)||38 (67.9)||11 (33.3)||13 (76.5)|
DO, Doctor of Osteopathic Medicine; MD, Doctor of Medicine; NP, nurse practitioner; PA, physician assistant; PCPs, primary care providers; SD, standard deviation.
In all four groups, the majority of participants chose not to order an MRI in patients with uncomplicated lower back pain: control 68% (26/38), colleague 85% (40/47), profession 87% (40/46), and ACO 78% (29/37). No statistically significant differences in this primary outcome were detected among the four groups. Bivariate analyses revealed a statistically significant association between the PCP practice site and MRI ordering (P = 0.006; Table 2). In the multivariate model, we only found that participants were more likely not to order an MRI in the profession group compared with the control group (odds ratio 4.9, 95% confidence interval 1.3–18), indicating that the odds of not ordering the MRI were approximately 5 times more likely in the profession group than the control group (Table 3). We did not observe statistically significant differences in the colleague or ACO group when compared with the control (P > 0.05 for both comparisons; Table 3).
|Yes MRI||No MRI||Neither||P|
|Age, y, mean (SD)||45.2 (11.0)||44.3 (11.2)||45.0 (8.8)||0.945|
|Male||5 (7.1)||53 (75.7 )||12 (17.1)|
|Female||6 (6.2)||82 (84.5)||9 (9.3)|
|Years since medical school (average)||15.9 (9.7)||16.6 (11.6)||18.2 (9.5)||0.80|
|Primary degree (%)||0.301 (Fisher exact test)|
|MD/DO||8 (5.4)||120 (81.1)||20 (13.5)|
|NP||2 (15.4)||10 (76.9)||1 (7.7)|
|PA||1 (16.7)||5 (83.3)||0 (0.0)|
|Percentage of direct patient contact (%)||0.05|
|0–25||3 (9.1)||21 (63.6)||9 (27.3)|
|26–50||1 (3.8)||20 (76.9)||5 (19.2)|
|51–75||3 (8.1)||32 (86.5)||2 (5.4)|
|76–100||4 (5.6)||62 (87.3)||5 (7.0)|
|Practice site (%)||0.006*|
|1||3 (5.5)||49 (89.1)||3 (5.5)|
|2||7 (11.1)||41 (65.1)||15 (23.8)|
|3||0 (0)||31 (93.9)||2 (6.1)|
|4||2 (11.8)||14 (82.4)||1 (5.9)|
*P ≤ 0.05.
DO, Doctor of Osteopathic Medicine; MD, Doctor of Medicine; MRI, magnetic resonance imaging; NP, nurse practitioner; PA, physician assistant; PCPs, primary care providers.
|Years since medical school||1.05||0.997–1.11||0.065|
|Percentage of direct patient contact|
*P ≤ 0.05.
ACO, accountable care organization; CI, confidence interval; OR, odds ratio; PCPs, primary care providers.
Evidence-based medicine was most often cited as the primary reason for the decision (Table 4). A statistically significant difference (P < 0.001) was detected via a χ2analysis, suggesting that evidence-based medicine and clinical judgment were more likely to be cited as the reason by PCPs who did not order advanced imaging (Table 4). Of note, among those who chose to order the MRI, patient-centered care was the more common reason (Table 4).
|Reason for decision||Order MRI||Not order MRI||Not sure|
aP < 0.00 via χ2 analysis.
MRI, magnetic resonance imaging; PCPs, primary care providers.
Decision making by clinicians can be influenced by multiple external factors. Although lacking the nuance of a real clinical encounter, hypothetical case studies presented in a research context can help elucidate specific factors that may play a role in decision making. In this study, given a hypothetical patient who presents with uncomplicated back pain but nevertheless seeks advanced imaging (ie, MRI) for reassurance, we found that PCPs, in general, would not pursue advanced imaging. More than 80% of the 168 primary care providers surveyed in this study chose not to order the MRI for uncomplicated, “garden-variety” back pain. An ACO’s quality metric that measures physician use of imaging in uncomplicated back pain did not significantly affect this decision. One possible reason for this is the mixed views from providers regarding the effectiveness of ACOs as a whole, as opposed to a professional society’s recommendations, which we found significantly affected the decision to pursue advanced imaging.
Recent findings have shown that recommendations against the use of diagnostic imaging for uncomplicated back pain by multiple professional organizations in other contexts (eg, the Choosing Wisely initiative) have not significantly altered clinical decisions, with back pain imaging present in 53% of observed cases. One possible reason for the apparent inconsistency between PCPs’ responses to a hypothetical scenario and these findings is the gap between theory and practice. Compared with a theoretical encounter, actual clinical practice is characterized by barriers to following such recommendations. A survey among nonresident clinicians in a Veterans Administration practice found that such barriers include lack of time to discuss risks and benefits, fear of litigation, desire to not upset the patient, or inability to refer to a specialist in the future. Of note, providers find that recommendations against testing for symptomatic patients compared with asymptomatic patients were particularly difficult to follow in practice and for patients to accept. Our finding that those ordering imaging were more likely to cite patient-centered care as the rationale for the decision may support this observation.
We suggest that one barrier to the success of Choosing Wisely may be the common conflict between evidence-based medicine, as understood by the clinician, and patient-centered care, as represented by patient preferences. The conflict between acknowledging a low chance of benefit from imaging and patients’ reactions to lack of imaging, may be worth exploring in understanding how PCPs’ preferences are affected by their perception of patients’ preferences. A second potential barrier is application of targeted reminders and clinical decision support, especially via the increasing utilization of electronic medical records, which has been shown to effectively reduce the rate of imaging among providers for lower back pain.[15,16] In this context, the variation in practice among different practice sites is an important consideration.
We found some indication in the bivariate analyses that academic and concierge practice sites were less likely to order an MRI in such a case than an academic-affiliated community practice. Reasons for this association may include difference in geographical location, perception of institutional policies and culture, the characteristics of the patient population, or the medical training backgrounds of employees. Specifically, the location of a physician’s residency training program has been shown to exert an effect on the healthcare spending of his or her practice later on in his or her career. We did not have the location of training from our participants, and our sample size did not allow us to compare possible differences among holders of different degrees (MD vs DO vs PA vs NP). Prior work, however, has shown that healthcare utilization and outcome do not significantly differ between NPs and physicians in a primary care setting.
We did not find differences in ordering patterns based on years since completion of medical training, age of the provider, or percentage of time in patient care contact. Previous studies have shown that expert differed from novice with regard to generating diagnostic hypotheses; therefore, our hypothesis was that younger providers, having less experience and thus less confidence in their clinical decisions compared with senior providers, would rely more on diagnostics tests and imaging in their practice and as such would order more tests and imaging. Our study, however, did not reveal this to be the case.
The limitations of our study include a low number of respondents, resultant lack of power, and limited generalizability to other practice sites. We also were unable to perform a nonresponse analysis because of the lack of data from nonresponders. To limit the burden of the survey, we kept the detail of the vignette to a minimum; therefore, participants may have unconsciously filled in details to the vignette that were not explicitly stated. We also were unable to explore other factors that may influence decision making such as experience with similar clinical scenarios, perceived barriers to implementing the recommendation, or fear of litigation. Furthermore, we must recognize the potential for social desirability bias in our study—a tendency for participants to answer questions the way they believe they should instead of what they would do in practice.
Despite such limitations, our study elucidated how PCPs approach advanced imaging for uncomplicated back pain across several practice locations and which factors may affect such a decision. Most PCPs would not order advanced imaging for uncomplicated back pain. Making PCPs explicitly aware of a professional society recommendation reduced ordering frequency; the presence of ACO quality metrics or advice from colleagues had no such effect. Practice site was associated with the likelihood of ordering imaging, and evidence-based medicine and clinical judgment were the most common justifications. Future studies should investigate additional influencing factors and explore this decision further in real-life patient encounters.
- Most primary care providers would not order advanced imaging for uncomplicated back pain.
- Only making providers explicitly aware of a professional society recommendation reduced ordering frequency.
- Evidence-based medicine and clinical judgment were the most common justifications.
- Rubin DI. Epidemiology and risk factors for spine pain. Neurol Clin 2007;25:353–371.
- Physician office visits for low back pain. Frequency, clinical evaluation, and treatment patterns from a U.S. national survey. Spine (Phila Pa 1976) 1995;20:11–19.
- Freburger JK, Holmes GM, Agans RP, et al. The rising prevalence of chronic low back pain. Arch Intern Med 2009;169:251–258.
- Centers for Disease Control and Prevention. Prevalence of disabilities and associated health conditions among adults—United States, 1999. JAMA 2001;285:1571–1572.
- Srinivas SV, Deyo RA, Berger ZD. Application of “less is more” to low back pain. Arch Intern Med 2012;172:1016–1020.
- Tan A, Zhou J, Kuo YF, et al. Variation among primary care physicians in the use of imaging for older patientswith acute low back pain. J Gen Intern Med 2015;31:156–163.
- Rosenberg A, Agiro A, Gottlieb M, et al. Early trends among seven recommendations from the Choosing Wisely campaign. JAMA Intern Med 2015;175:1913–1920.
- Cassel CK, Guest JA. Choosing wisely: helping physicians and patients make smart decisions about their care. JAMA 2012;307:1801–1802.
- Cherkin DC, Deyo RA, Wheeler K, et al. Physician variation in diagnostic testing for low back pain. Who you see is what you get. Arthritis Rheum 1994;37:15–22.
- Patterns of ordering diagnostic tests for patients with acute low back pain. The North Carolina Back Pain Project. Ann Intern Med 1996;125:807–814.
- Ryan J, Doty MM, Hamel L, et al. Primary care providers’ views of recent trends in health care delivery and payment findings from the Commonwealth Fund/Kaiser Family Foundation 2015 National Survey of Primary Care Providers. https://www.commonwealthfund.org/publications/issuebriefs/2015/aug/primary-care-providers-views-recent-trends-health-caredelivery. Published August 5, 2015. Accessed September 23, 2018.
- Cabana MD, Rand CS, Powe NR, et al. Why don’t physicians follow clinical practice guidelines? A framework for improvement. JAMA 1999;282: 1458–1465.
- Sears ED, Caverly TJ, Kullgren JT, et al. Clinician’s perception of barriers to avoiding inappropriate imaging for low back pain—knowing is not enough. JAMA Intern Med2016;176:1866–1868.
- Zikmund-Fisher BJ, Kullgren JT, Fagerlin A, et al. Perceived barriers to implementing individual Choosing Wisely® recommendations in two national surveys of primary care providers. J Gen Intern Med 2017;32: 210–217.
- Jenkins HJ, Hancock MJ, French SD, et al. Effectiveness of interventions designed to reduce the use of imaging for low-back pain: a systematic review. CMAJ 2015;187:401–408.
- Forseen SE, Corey AS. Clinical decision support and acute low back pain: evidence-based order sets. J Am Coll Radiol 2012;9:704–712.e4.
- Fisher ES, Wennberg DE, Stukel TA, et al. The implications of regional variations in Medicare spending. Part 1: the content, quality, and accessibility of care. Ann Intern Med2003;138:273–287.
- Huang DT, Clermont G, Kong L, et al. Intensive care unit safety culture and outcomes: a USmulticenter study. Int J Qual Health Care 2010;22:151–161.
- Asch DA, Nicholson S, Srinivas S, et al. Evaluating obstetrical residency programs using patient outcomes. JAMA 2009;302:1277–1283.
- Chen C, Petterson S, Phillips R, et al. Spending patterns in region of residency training and subsequent expenditures for care provided by practicing physicians for Medicare beneficiaries. JAMA 2014;312: 2385–2393.
- Mundinger MO, Kane RL, Lenz ER, et al. Primary care outcomes in patients treated by nurse practitioners or physicians: a randomized trial. JAMA 2000;283:59–68.
- Hobus PP, Schmidt HG, Boshuizen HP, et al. Contextual factors in the activation of first diagnostic hypotheses: expert-novice differences. Med Educ 1987;21:471–476.