The incorporation of best practice guidelines into one's institution is a challenging goal of utilization management, and the successful adoption of such guidelines depends on institutional context. Laboratorians who have access to key clinical data are well positioned to understand existing local practices and promote more appropriate laboratory testing.
To apply a novel approach to utilization management by reviewing international clinical guidelines and current institutional practices to create a reliable mechanism to improve detection and reduce unnecessary tests in our patient population.
We targeted a frequently ordered genetic test for HFE-related hereditary hemochromatosis, a disorder of low penetrance. After reviewing international practice guidelines, we evaluated 918 HFE tests and found that all patients with new diagnoses had transferrin saturation levels that were significantly higher than those of patients with nonrisk genotypes (72% versus 42%; P < .001).
Our “one-button” order that restricts HFE genetic tests to patients with transferrin saturation greater than 45% is consistent with published practice guidelines and detected 100% of new patients with HFE-related hereditary hemochromatosis.
Our proposed algorithm differs from previously published approaches in that it incorporates both clinical practice guidelines and local physician practices, yet requires no additional hands-on effort from pathologists or clinicians. This novel approach to utilization management embraces the role of pathologists as leaders in promoting high-quality patient care in local health care systems.
Clinical practice guidelines (CPGs) are intended to improve health care quality by translating scientific evidence into practice. Although CPGs are invaluable resources, the implementation of those guidelines is a challenge.1 Guideline characteristics (eg, user-friendliness), as well as context (eg, intended users and practice setting), factor into whether a CPG will be successfully implemented and incorporated into daily practice.1 Barriers to implementation may include practitioner concerns that the guidelines do not apply to their particular patients, which may be true in select instances.2,3 At health care institutions with informatics resources, computerized order entry provides an opportunity to apply CPGs within a unique health care system based on the target patient population and institutional culture and resources.4
Pathologists have specialized training in both clinical medicine and laboratory sciences and can drive multidisciplinary efforts to understand the limitations of CPGs and develop interventions that promote better patient care. Our team undertook a project to compare international practice guidelines with current ordering practices within our organization. We selected a relatively commonly ordered genetic test with practice guidelines issued by multiple professional organizations for proof-of-concept.5,6 Other institutions have also studied utilization of this test, which allowed us to compare our local ordering patterns to those of other practices.7 If our current practice was concordant with recommendations, no interventions would be necessary. If it was discordant, we would attempt to understand provider ordering patterns and leverage electronic decision support to improve compliance, which has been successfully used, as reported by Riley et al.8
Hereditary hemochromatosis due to homozygous C282Y mutations in the HFE gene is a well-known cause of inherited iron overload syndrome. The genetic polymorphism, H63D, is weakly associated with iron overload, and the clinical significance of S65C is debatable. Furthermore, HFE-related hereditary hemochromatosis (HFE-HH) has low disease penetrance, such that genetic test results in isolation have little clinical utility or predictive value. Even in patients with homozygous C282Y mutations, development of clinical manifestations of HFE-HH is multifactorial and may be less than 1%.9 Frequently used screening tests, such as serum ferritin level and transferrin saturation, are widely available and commonly used in a variety of clinical scenarios, but they have low specificity for HFE-HH.
In July 2015, as part of the American Board of Internal Medicine Choosing Wisely initiative, the American College of Medical Genetics and Genomics advised against ordering HFE genetic testing for patients without either iron overload or a family history of HFE-HH.10 The recommendations were consistent with published guidelines from the European Association for the Study of Liver, American Society for the Study of Liver Diseases, and European Molecular Quality Network 2015 guidelines.5,6,10,11 These guidelines denote clearly that genetic testing should only be performed after excluding other causes of increased ferritin,5 because ferritin is nonspecific and only elevations consistently above 1000 μg/L support iron overload.11 Increased ferritin alone is insufficient to warrant HFE genetic testing, but it should prompt testing for transferrin saturation.6 Measurements should include fasting transferrin saturation and serum ferritin; HFE testing should only be performed in patients with increased transferrin saturation (>45% in female patients, >50% in male patients).5 All of the guidelines advise against general population screening.6,10,11
The HFE genetic test performed in our molecular genetics pathology laboratory assesses for C282Y, H63D, and S65C variants by multiplex polymerase chain reaction and postamplification melt curve analysis. Our laboratory performs nearly 1000 HFE tests per year from main campus, regional hospitals and clinics, and outside providers. Our laboratory is part of an active laboratory utilization management program, which works to implement best practice testing guidelines. We have a full-time laboratory-based genetic counselor who performs daily genetic test review, and a Laboratory Stewardship Committee that recommends electronic clinical decision support tools for utilization management. Although not yet attempted at our institution, electronic reflexive testing strategies combined with computerized physician/provider order entry is an end goal of this initiative and has been shown to improve utilization patterns on a large scale in other health care systems.12 Herein, we sought to clarify the HFE ordering patterns of our physicians to determine whether it would be beneficial to integrate international practice guidelines into a proposed algorithmic “one-button” electronic order algorithm.
MATERIALS AND METHODS
This study was performed as a laboratory quality improvement project and did not require formal institutional review board approval. To evaluate local ordering practices and determine the potential use of algorithmic testing, we performed a retrospective review of 918 HFE clinical test results from January 6, 2015, to January 4, 2016, which included a detailed clinical chart review of 98 patients. Our laboratory functions as a reference laboratory, so exhaustive physician and patient data were not available for all patients. A detailed subset analysis was performed on all 43 C282Y/C282Y results from 2015 to assess the patient and provider characteristics that predicted positive results. Similar analyses were performed for 55 consecutive orders which served as a control group. Parameters collected included the ordering physician's specialty, patient's clinical history, family history, and selected results of laboratory tests performed before, during, and after the HFE test was ordered (ie, current and previous hemoglobin, transferrin saturation, ferritin, total iron-binding capacity, and serum iron levels). Patients with duplicate genetic tests were only included once. We limited our review to adult patients because hemochromatosis is extremely rare in childhood.
RESULTS
A total of 918 HFE genotyping assays were performed in 2015, with a new diagnosis rate of less than 3.38% (fewer than 31 of 918 assays). In total, there were 875 genotypes with minimal to no increased risk of HFE-HH (Figure 1). Compound and simple heterozygosity for H63D or S65C, and simple heterozygosity for C282Y were considered insufficient for causing clinically significant disease, as denoted in current guidelines.6
Detailed clinical data were available for 33 of 43 C282Y/C282Y homozygous results on 40 unique patients (76.4%; 1 patient was tested twice, and 1 patient was tested 3 times). A family history of HFE-HH is an accepted indication for genetic testing, and 10 of the 40 unique patients (25%) had an affected family member. However, our review showed that 12 of 33 orders (36.36%) were for 10 patients already receiving therapeutic phlebotomy or close clinical follow-up for known HFE-HH. A total of 6 of these 10 patients (60%) had a known positive family history. The reasons for these duplicate constitutional genetic tests were explored as available clinical data permitted.
There were 31 HFE-HH patients with ordering physician information available. Generalists (4 of 31; 12.9%) gave diagnoses for 4 patients. There were 2 dermatologists (2 of 31; 6.45%) who recognized HFE-HH based on dermatologic symptoms and abnormal iron studies in a 55-year-old man with hair loss and a 42 year-old woman with a blistering skin condition. Hematology (13 of 31; 41.94%) and gastroenterology (12 of 31; 38.71%) were responsible for the greatest number of positive results, including results for 11 patients who had known HFE-HH. Based on the number of repeat orders, there were likely fewer than 31 new diagnoses (43 positive test results minus the 12 previously known positive results from available clinical history); the new diagnosis rate for patients with the C282Y/C282Y genotype was less than 3.38% (fewer than 31 new diagnoses of 918 tests performed in 2015).
We also reviewed clinical and demographic data on patients from 55 consecutive HFE orders from July 20 to August 3, 2015, as a control group. No patients with C282Y homozygosity were identified during this time frame. There were 2 simple C282Y carriers and 3 C282Y/H63D compound heterozygous carriers in this group. Primary care physicians identified 3 of the carriers, including 2 patients with a family history of HFE-HH. Physician specialty information was available for 44 of the 55 consecutive orders (80%). Orders were placed by gastroenterologists (50%; 22 of 44), generalists (31.81%; 14 of 44), hematologists (13.64%; 6 of 44), and cardiologists (4.55%; 2 of 44). With the exception of 4 patients who had a family history of HFE-HH, the remaining patients had less compelling reasons to justify genetic testing. Although 29 patients had either elevated transferrin saturation (>45%) or ferritin (>300 μg/L) levels, many had alternative explanations for their abnormal iron studies, including 5 patients with active malignancy as a likely cause for elevated acute-phase reactants. Liver dysfunction was frequently seen in the presence of abnormal iron studies (n = 13). Altogether, 21 patients had liver disease or abnormal liver function test results, including at least 6 with documented cirrhosis.
Our findings indicated suboptimal utilization of the HFE test at our institution. We had a low positivity rate, and for unknown reasons some providers frequently performed genetic testing on patients receiving therapeutic phlebotomy treatment who had a known diagnosis of HFE-HH. Genetic testing for HFE-HH was also performed on patients with alternative explanations for liver dysfunction and elevated iron studies. Although several guidelines include elevated ferritin as a reason to further investigate for HFE-HH, transitory and nonspecific elevations of ferritin are well established in the literature.5,11 Patients at our institution often present in a state of acute physiologic distress, so ferritin elevations may be even less specific. We needed to identify a screening measure that would account for these observations at our institution.
We sought to identify parameters consistent with existing CPGs that would increase the diagnostic yield of the assay and improve the clinical utility of laboratory orders (Table). Patients with a new diagnosis of HFE-HH had the highest levels of ferritin (versus C282Y-negative patients; Welch unequal variances t test, 2-tailed P = .04), but the discriminating yield of ferritin decreased when comparing all C282Y/C282Y patients (ie, those under treatment and those with new diagnoses) with unaffected patients (Welch unequal variances t test, 2-tailed P = .18). In contrast, transferrin saturation levels of C282Y homozygous patients were consistently higher than levels in patients without C282Y (Welch unequal variances t test, 2-tailed P < .001).
When only patients with new diagnoses were analyzed, a transferrin saturation of less than 45% was 100% sensitive for excluding patients with the C282Y/C282Y genotype. Combining all C282Y/C282Y patients (ie, C282Y/C282Y new diagnoses and those who had previously received diagnosis and treatment), a transferrin saturation of less than 45% was 93% sensitive for excluding clinically significant HFE-HH. Based on this observation, a transferrin saturation cutoff of 45% would decrease unnecessary HFE testing while still detecting true positives. In other words, by using such a cutoff, we would appropriately provide diagnoses and care for our highest-risk patients while decreasing the time and money associated with unnecessary testing. Also, based on the inconsistent follow-up for HFE testing ordered on hospitalized patients who also had alternative explanations for their abnormal iron study results (eg, acute infection, metastatic cancer, and alcohol intoxication), we recommend against ordering HFE (and other genetic tests) on inpatients.
DISCUSSION
Our proposed intervention is a single “one-button order” (Figure 2) that activates appropriate algorithmic testing. Initial studies for transferrin saturation would be followed by genetic testing only when transferrin saturation levels were above 45%. Although clinicians are traditionally responsible for ordering a full panel of laboratory and genetic tests to evaluate for HFE-HH, a more intuitively named order, such as “rule out hereditary hemochromatosis” or “hereditary hemochromatosis workup,” may help promote more appropriate ordering. Ordering information would specify that the test is a transferrin saturation assay that reflexes to HFE genotyping if transferrin saturation is greater than 45%. Because of the nature of our referral population, patients with a positive family history of hereditary hemochromatosis could also undergo testing. This “one-button order” is a tool designed specifically for our institution, based on local ordering patterns and patient characteristics.7 Authors from a Canadian academic tertiary medical center observed a higher number of C282Y homozygotes (8.1%) than our positivity rate of less than 3.38%, which suggests practice differences in testing threshold and detection rate.7 At our institution, which has a large referral base, transferrin saturation appears to be superior to ferritin levels as the initial test when HFE-HH is suspected. These differences highlight institutional variability and the need for institution-specific interventions.
Helping clinicians deliver quality patient care in more efficient and affordable ways is the primary goal of any institution's test utilization management program. As laboratorians, we were able to review ordering practices and demonstrate a substantial number of unnecessary genetic tests for hereditary hemochromatosis, a departure from international practice guidelines. This discovery presented an opportunity to develop a testing algorithm to improve utilization. By applying published guidelines and limiting HFE testing in our patient population to those with either a positive family history or transferrin saturation greater than 45%, our system would reduce clinically unnecessary tests without missing any true HFE-HH diagnoses. In addition to recommending transferrin saturation as the initial test, the test utilization committee has now recommended a restriction on inpatient orders, which had a higher rate of inappropriate testing and inconsistent follow-up. We estimate that if used broadly, the algorithm should detect all patients at risk for developing true clinical disease (ie, end-organ damage secondary to excess iron storage) and result in a 43% reduction of unnecessary HFE tests and a lower cost per new diagnosis.
Once implemented, our approach requires no additional hands-on effort from pathologists or clinicians and differs from previously reported algorithmic approaches in that our laboratory screening cutoff has been vetted against published guidelines and our local institution's data. For example, the previously reported “one-click” solutions for the laboratory evaluation of lupus incorporate time-intensive pathology triage and interpretation.13 Our approach also differs from decision support that requires clinicians to document the indication for testing at order entry.4 Documentation requirements can lead to “pop-up button fatigue.” By design, no additional pathology intervention would be required because the laboratory cutoff for further testing has already been evaluated and established in our patients. In all instances, our providers have an opportunity to override these best practice interventions by calling the molecular laboratory, but they are subject to audit.
The findings of this review demonstrate that more expensive diagnostic testing could be averted using inexpensive screening methods designed with institution-specific data. Additionally, unless there are concerns about the validity of the existing test result, the Choosing Wisely campaign notes that constitutional genetic tests should not be repeated.10 In our experience, approximately one-third of the HFE genetic tests with positive results were ordered for patients who had previously received diagnosis and treatment. Additional genetic tests could have been averted through adherence to international guidelines. These laboratory tests were cost expenditures with no health care benefits. From our review, this ordering pattern may reflect a lack of understanding of the nature of constitutional genetic tests and the costs associated with testing, or simply difficulty locating previous genetic test results in the medical record. Our health care system has instituted an electronic clinical decision support tool that prevents duplicate genetic test orders.
CONCLUSIONS
As health care transitions from volume-based fee-for-service to value-based care, we will increasingly be held accountable for adhering to clinical practice guidelines. Our proof-of-concept utilization intervention could be expanded to other conditions for which there are multiple testing options and clearly established CPGs or diagnostic pathways.
The National Comprehensive Cancer Network has published guidelines in the form of downloadable documents, but an algorithmic approach would help streamline and standardize the diagnostic workup and monitoring of cancer patients. For example, there are myriad diagnostic test options for hematologic malignancies, which can cause confusion and unnecessary delay, but creating locally developed automated tests would not only save time and money but would also promote best evidence-based practices in the care of oncology patients.
We recognize that every institution has distinctive utilization patterns and challenges. When designing utilization management strategies, it is vital to consider organizational culture in developing customized interventions. We emphasize the importance of understanding how people (eg, patients and clinical teams) and their environment (eg, tools and technology) interact. Test menu and turnaround time vary from institution to institution, so how these guidelines are implemented is institution dependent. Only by examining local institutional practices and resources can CPGs be successfully leveraged to maximize benefits to our patients.
We have described a novel approach of integrating international practice guidelines into a proposed algorithmic “one-button” electronic order to improve the care of our patients. Because we have access to key laboratory data, pathologists are ideally positioned to lead efforts to promote high-quality patient care in local health care systems.
References
Author notes
From the Robert J. Tomsich Pathology and Laboratory Medicine Institute, Cleveland Clinic, Cleveland, Ohio. Dr Zhou is now with the Department of Pathology, University of Oklahoma Health Sciences Center, Oklahoma City.
The authors have no relevant financial interest in the products or companies described in this article.
Competing Interests
Presented in poster form at the 105th Annual Meeting of the United States & Canadian Academy of Pathology; March 12–18, 2016; Seattle, Washington.
