Testing Practices, Interpretation, and Diagnostic Evaluation of Iron Deficiency Anemia by US Primary Care Physicians

This survey study uses hypothetical patient and clinical scenarios to assess how US primary care physicians (PCPs) approach screening for iron deficiency anemia, interpret iron laboratory studies, and refer patients with iron deficiency anemia for gastrointestinal endoscopy.


Introduction
Iron deficiency anemia (IDA) is a classic early diagnostic sign of gastrointestinal (GI) tract malignant neoplasm. 1,2 Between 1% and 10% of adult patients with IDA may have undiagnosed GI tract cancer. 3,4 Prompt and thorough evaluation of IDA is essential. However, retrospective studies have reported that delays in diagnostic evaluation of IDA are common 5,6 and can lead to a delayed diagnosis of colorectal cancer. 7,8 This problem is of increasing importance given the increasing incidence of colorectal cancer among younger patients outside the age range for which routine colorectal cancer screening is recommended, for whom IDA may be the first diagnostic sign of malignant neoplasm. 9,10 Clinical practices related to identification and evaluation of IDA likely vary substantially, with at least 30 different guidelines from 10 specialty societies. [11][12][13][14] However, variation related to testing practices, interpretation of laboratory test results, and diagnostic evaluation of IDA among physicians is not well studied even though these factors may be associated with diagnostic delays, missed cancer diagnoses, and low-value care. For instance, because of a low diagnostic yield, anemia screening is not recommended for adults except during pregnancy, and recommendations about screening during pregnancy are mixed. [15][16][17] Inappropriate screening may be associated with health care waste and potential unintended harm (eg, additional downstream diagnostic tests). Inaccurate test interpretation of iron studies (eg, serum ferritin level and iron saturation) can also lead to medical error, with misclassification of patients' iron level. Interpretation can be complicated because the ferritin level may be elevated in the context of an inflammatory process, potentially masking underlying iron deficiency. 2,18 Guidelines consistently recommend that men with IDA and postmenopausal women, in the absence of obvious causes, should be evaluated for potential occult GI tract blood loss with colonoscopy and esophagogastroduodenoscopy (EGD). 11,12 However, older guidelines, which predated the increasing incidence of colorectal cancer among younger patients observed over the past decade, recommended against initial endoscopic evaluation for premenopausal women with IDA given the more common cause of iron deficiency secondary to menstrual blood loss. 11 Together, these issues introduce ambiguity and complexity to the evaluation of IDA.
Our objective was to understand how primary care physicians (PCPs) approach testing and evaluation of patients for IDA. To do this, we conducted a survey study consisting of a vignette-based assessment of PCPs to examine their practices related to IDA testing and evaluation for male and female patients of different ages. A better understanding of current PCP evaluation and referral patterns for GI endoscopy in the context of IDA may reveal potential opportunities to intervene and promote earlier diagnosis of GI tract cancers.

Study Population and Survey Distribution
For this survey study, we conducted a national online survey of internal medicine PCPs in August 2019 using the American College of Physicians (ACP) Internal Medicine Insiders Panel. The ACP Insiders Panel is a nationally representative group of internal medicine physicians selected through stratified random sampling as representative of the ACP's membership as a whole. 19 The ACP is the largest internal medicine specialty organization, with 163 000 members. Potentially eligible participants (identified as PCPs who had completed training within the ACP Insiders Panel demographics profile) received an email invitation to complete the electronic survey, and those who completed the survey received points that could be redeemed for Amazon.com gift cards. Two email reminders were sent during a 2-week eligibility period. All respondents had to meet the following inclusion criteria: (1) practice outpatient primary care medicine, (2) practice primarily general internal medicine and not a subspecialty, and (3)

Questionnaire Design
We developed a survey to assess current PCP practices for the testing and evaluation of IDA. Clinical vignettes provided with the survey questions were used to address 3 sequential domains to elicit physicians' clinical practices and decision-making processes 20

Statistical Analysis
Survey responses are summarized using proportions, and summary statistics are reported for continuous variables. A 2-sided P < .05 was considered statistically significant. We used a χ 2 test for analysis of the categorical variables. Statistical analysis was performed using SAS, version 9.4 (SAS Institute). Graphs were produced using GraphPad Prism, version 8.0 (GraphPad Software).

Respondent Characteristics
The survey was distributed electronically via email invitation to 633 individuals on the ACP Insiders

JAMA Network Open | Health Policy
Diagnostic Testing and Evaluation of Iron Deficiency Anemia by US PCPs Demographic information from the ACP Insiders Panel was analyzed to compare potential differences between respondents and nonrespondents. Demographic data from August 2019 were available for 454 of the 633 participants (71.7%) who were invited to complete the survey, including 240 who responded to the survey and 214 who did not respond. There were no statistically significant differences between respondents and nonrespondents by geographic region (West,

Screening for Anemia
Respondents were asked about their practice of testing asymptomatic healthy patients for anemia by  respondents (13.5%) misidentified the cause as something other than iron deficiency even though the likelihood ratio for iron deficiency with a ferritin level less than 15 ng/mL is 51.9. 18 In a scenario with a low transferrin saturation (2%) but borderline low ferritin level (40 ng/mL), 86 of 325 respondents (26.5%) did not consider IDA as the cause even though the likelihood ratio for iron deficiency with a transferrin saturation less than 5% is 10.5. 18 In that scenario, 239 of 325 respondents (73.5%) correctly identified the scenario as IDA. There were no statistically significant differences in correct interpretation of these scenarios by academic affiliation status, gender, stage in clinical practice, or practice of ordering ferritin level measurement for anemia evaluation.

Diagnostic Evaluation of IDA
Of 325 respondents, 316 (97.2%) reported that PCPs, not specialists, were primarily responsible for management of IDA within their practice settings. Respondents were asked about which tests they would obtain in a series of scenarios with new-onset IDA and negative celiac serologic test results in  Although respondents reported managing IDA in 65-year-old men and women similarly, they year-old woman.

Discussion
In this nationally representative survey study of internal medicine PCPs, we found substantial variation in the management of anemia. Specifically, we found self-reported evidence of both overuse of screening CBCs (76.9% of respondents) and underuse of bidirectional endoscopy in patients with IDA. Certain knowledge gaps also existed in the interpretation of discordant laboratory  test results, which may contribute to diagnostic errors. Specifically, 26.5% of respondents misdiagnosed a scenario of IDA as anemia with an alternate cause despite a low transferrin saturation (2%) and a borderline low ferritin level (40 ng/mL). Several subtle but important nuances in interpretation of these studies may contribute to errors, including the high specificity of a low ferritin level (98% specific for iron deficiency with a ferritin level Յ12 ng/mL), 24 the potential for the ferritin level to be normal or elevated in inflammatory states when a patient is iron deficient, 2,18 and the importance of measuring serum iron level in a fasting state to avoid potential confounders of dietary or supplemental iron. 25,26 Given the relatively high baseline prevalence of anemia of 5.6% in the US, 13 incorrect interpretation of these laboratory results may lead to a substantial number of missed or delayed diagnoses of IDA and, by extension, potential delays in diagnosis of GI tract malignant neoplasm.

JAMA Network Open | Health Policy
Our findings are in line with those of older studies on this topic. Prior work has found limited clinical benefit associated with the common practice of obtaining routine CBCs for hospitalized patients 27,28 and outpatients. [29][30][31] But in a repeated survey study of family medicine PCPs, the frequency of routine CBCs remained persistently high between 1978 and 2004, with 67% of respondents reporting that they would obtain a CBC for a 55-year-old woman in the most recent survey. 32 Although overuse is a complex phenomenon, these data suggest that routine laboratory assessments are ingrained within medical practice. In this context, it is notable that, to our knowledge, there are no guidelines (eg, from Choosing Wisely or other organizations) explicitly recommending against screening for anemia in outpatients. 17,33,34 In contrast, Choosing Wisely explicitly recommends against routine or repeated CBCs for hospitalized patients. 33 displaying cost of the laboratory testing alone is unlikely to be sufficient, 40 but a multilevel intervention incorporating education, price information, feedback, and financial incentives has demonstrated success in an inpatient setting. 41

Limitations
This study has limitations. Respondents may have answered questions based on what they believed was the correct evidence-based practice rather than what was reflective of their actual clinical practice, potentially introducing response bias. Other limitations include the possibility that participants in the ACP Insiders Panel were more engaged with the organization and thus more aware of current evidence-based practices. Both of these biases would suggest that the knowledge gaps identified in our study are underestimates of those seen in actual practice. The results may not apply to other professionals who provide primary care, such as family medicine physicians or advanced practice physicians. A potential for nonresponse bias exists, and those who did not respond may have been less knowledgeable about the topic or may have differed in other ways related to the outcomes of interest. However, we attempted to mitigate this bias through systematic efforts to increase responses. In addition, we analyzed available demographic data for respondents and nonrespondents and found no significant differences in age, geographic location, or career stage; however, respondents were more likely to be White and male than were nonrespondents. Survey methods may not be as well suited for the study of actual practice behaviors as other methods.
However, our survey allowed us to indirectly assess physician knowledge and behavior with uniform, ideal clinical scenarios, which is not possible when secondary data are used.

Conclusions
In this survey study, PCPs' self-reported testing practices for anemia suggest overuse of screening laboratory tests, misinterpretation of iron studies, and underuse of bidirectional endoscopy in evaluation of new-onset IDA. These findings suggest that IDA, a common clinical condition and early sign of GI malignant neoplasm, is not uniformly understood, approached, and managed by practicing physicians in the US. Given the potential association with GI cancers and the increasing incidence of colorectal cancer in younger adults, thorough evaluation of IDA is important for preventing delays in diagnosis. 3,4,9 Further work should inform the design and implementation of tools to improve diagnostic evaluation of IDA and reduce the risk of potential diagnostic errors.