Diagnostic Workup of Microcytic Anemia: An Evaluation of Underuse or Misuse of Laboratory Testing in a Hospital Setting Using the AlinIQ System Open Access

The study was conducted in the University Hospital Salzburg (Salzburg, Austria), a tertiary referral hospital serving about 77 000 inpatients and 561 000 outpatients per year. All patients older than 18 years in whom a microcytic anemia was detected within the year 2018 were included in the study. Definition of anemia was set according to the World Health Organization criteria (hemoglobin: men <13 g/dL; women <12 g/dL).¹⁰  Microcytosis was defined as mean corpuscular volume (MCV) values below 80 μm³. For definition of hypochromia, a cutoff of 28 pg per cell was used for the mean corpuscular hemoglobin value.¹¹  For more detailed analysis, 3 subgroups were defined for patients with MCV values ≥75 μm³ to <80 μm³ (group 1), ≥65 μm³ to <75 μm³ (group 2), and <65 μm³ (group 3). All laboratory data from the individual patient visits to our hospital from June 2007 until February 2020 potentially contributing to the diagnostic workup of the anemic state were collected from the LIS. The newly introduced clinical decision support software AlinIQ CDS (Abbott Laboratories, North Chicago, Illinois) was used for merging and evaluating these data. This clinician-driven expert knowledge system is a case-based rules engine, one that executes defined diagnostic algorithms based on input from any number of data sources (LIS, electronic health records, etc). For this study, however, it was used in a retrospective manner, importing data from various data sources (LIS and electronic health records) into the software in order to combine information that is not generally linked.

The data were sequentially screened according to the query-algorithm depicted in Figure 1, which was programmed into the AlinIQ system to categorize patients according to the specified cornerstones. Percentages were calculated in reference to the according baseline numbers. In anemic patients who were not followed up correctly, all available ICD-10 codes appropriate for causes of microcytic anemia were extracted from the electronic health record to determine whether patients had a preexisting diagnosis of anemia. These investigated ICD codes were D50.-* - Iron deficiency anemia, D56.-* - Thalassemia, D57.-* - sickle cell anemia, D63.-* - anemia of chronic disease and D64.-* - other anemia.

For statistical calculations of differences and trends among the subgroups, a χ² test was performed using the GraphPad Prism 9 Software (GraphPad Software Inc, San Diego, California). Differences with a P value less than .05 were considered significant. Figures were made using Microsoft Excel and Microsoft Visio Software (Microsoft, Redmond, Washington).

The study was approved by the local ethics committee (EC-No. 1009/2020).

In 2018, microcytic anemia was detected in 2244 patients older than 18 years (Table). Division into MCV subgroups yielded 1363 (60.7%), 727 (32.4%), and 154 (6.9%) patients in groups 1, 2, and 3, respectively. At least 1 parameter to test for iron deficiency was ordered in 1483 cases (66%), but only in 1104 (49%) cases were these tests performed within a time frame of 2 weeks before or after the assessment of a microcytic, hypochromic anemia (Figure 2, A). These numbers indicate an underuse of iron status testing in 34% and a potential misuse in 17% of all cases. Ferritin levels were tested in 1032 patients (93.5%) in whom iron status was investigated in a timely, correct manner. Applying a ferritin cutoff of 30 ng/mL,¹²  637 (61.7%) of the investigated patients would be classified as iron deficient. In patients with ferritin levels always above this cutoff (n = 395; 38.3%), concomitant C-reactive protein (CRP) testing, defined as testing 1 day before or after ferritin testing, was done in most cases (n = 380; 96.2%; Figure 2, B). In those cases where ferritin could not be interpreted conclusively because of elevated CRP results (n = 336), reticulocyte hemoglobin content (Ret-Hb) or soluble transferrin receptor testing could provide clarification. These tests were missing in 86 (26%) of these patients (Figure 2, C).

In patients with ferritin levels above 30 ng/mL and CRP levels within the reference range (≤0.5 mg/dL; n = 214) and no previous MCV results >80 μm³ (n = 41), subsequent evaluation for other causes like thalassemia/hemoglobinopathy should be performed. Overall, these tests were not performed in 11 (27%) of these cases (Figure 2, D).

Frequency of the use of other iron status laboratory parameters, including subgroup analysis of groups 1 to 3, is shown in Figure 2, E. The Mentzer index (MCV/red blood cell count) could provide a diagnostic lead toward thalassemia in patients with constantly low MCV results.¹³  Calculating the number of cases without thalassemia testing in the cohort of patients with a Mentzer index <13 and no previous MCV results >80 μm³ (n = 131) showed that 74 (57%) of these cases had no according follow-up diagnostic workup (Figure 2, F). Subtracting cases with preexisting anemia diagnoses (n = 4) yielded a total of 70 cases (55%).

By summing up the cases with missing or inadequate iron status and/or thalassemia testing, laboratory tests were underused in 940 cases (41.9%). A total of 36 of these (3.8%) had a preexisting anemia diagnosis, indicating a sufficient workup prior to the hospital visit, leaving 904 cases (40.3%) that appeared to not be followed up correctly (Figure 3).

In this study, we could demonstrate that in 904 patients (40%) with microcytic anemia and no previous documentation thereof, the laboratory diagnostic workup might have been performed incorrectly or not at all.

Anemia is a common condition worldwide, most prevalent in infants (5.7%), teenage girls (5.9%), young women (5.8%), and elderly men (4.4%).¹⁴  According to the World Health Organization, 22.7% of European women in the reproductive age are anemic (17.3% in Austria).¹⁵  These numbers may vary substantially, depending not only on the country but also on the setting and patient population, age, sex, and race. Numbers may be as high as 67.4% for male and 64.5% for female individuals in outpatient wards of a tertiary hospital¹⁶  or 58.4% at an internal inpatient ward.¹⁷  Koch et al¹⁸  found that of 188 447 hospitalized patients, 74% already had or developed anemia. As a consequence, anemia may contribute to prolonged hospital stay, decreased survival, and other adverse outcomes.^17,19  Therefore, it is important to follow up anemic patients in order to reveal the root cause thereof and thereby be able to counteract it appropriately.

There are 3 main causes of microcytic anemia: iron deficiency, thalassemia, and anemia in chronic disease. The first 2 conditions may be diagnosed by laboratory analyses alone, based on a large set of evidence.^8,9,20,21  However, evaluation of real-life routine test results indicate an underuse of laboratory tests in 40% of our patients. Presumably, a lack of knowledge or confidence of clinicians in test result interpretation is one of the major contributing factors.^22–24  Several authors have shown that educative intervention is able to overcome such issues.^25–27  Currently several freely accessible online platforms exist, helping educate in test selection and interpretation: Choosing wisely,^28,29  Re-Testing Program United Kingdom,³⁰  LabTests Online,³¹  the European Federation of Laboratory Medicine e-learning platform,³²  the National Institute for Health and Care Excellence pathways,³³  the Centers for Disease Control and Prevention Laboratory Medicine Best Practices initiative,³⁴  Klug entscheiden,³⁵  and the Quality Use of Pathology Program (Australia),³⁶  among others.

Inappropriate use of diagnostic testing may cause overuse or underuse. Studies have shown that overuse of laboratory resources may be as frequent as 70% of high-throughput laboratory testing.¹  However, few studies have aimed to quantify laboratory underuse. The most prominent reason surely is the inability of laboratories to detect underuse if the clinical question is not provided alongside the test order. Some authors therefore chose the approach of adding useful tests to the order based on the information the laboratory has from all patients. Salinas et al,³⁷  for instance, added serum calcium in all patients older than 45 years if it was not measured in the last 3 years, thereby detecting 34 new cases of primary hyperparathyroidism. Other authors evaluated retrospective data with known diagnosis and adverse patient outcome in order to quantify laboratory underuse. Gandhi et al,⁷  who examined closed malpractice claims, found inappropriate diagnostic test ordering in 100 of 181 cases (55%). Zhi et al³⁸  performed a multidatabase systematic review and calculated the rate of laboratory underuse of 44.8%. Because the population in these studies consisted of cases in which adverse events were documented and/or filed for lawsuit, a preselection bias is likely. Therefore, we chose a different approach by focusing on microcytic anemia because it is a symptom, demanding a root cause analysis. Additionally, the diagnosis of the 2 main causes thereof (iron deficiency and thalassemia) can be made solely by using laboratory testing. Thereby, we could include all patients with that indication without the need for preselection, resulting in a more accurate calculation of laboratory underuse.

The number we ultimately calculated (40.3%) was similar to those found by Gandhi et al⁷  and Zhi et al,³⁸  and it supports the finding that our observation might not only be true for microcytic anemia or the clinical questions other authors investigated, but for far more indications than currently assumed, gravely jeopardizing patient safety.

The magnitude of some of the suspected underuse scenarios was astounding. For example, 32.8% of all cases lacked any iron deficiency investigation. In patients with suspected thalassemia, 55.1% (n = 70; 3.1% of all cases) received no follow-up thalassemia investigation. Even though CRP was tested in almost all patients (n = 380; 96%) with elevated ferritin values, overall follow-up investigations depending on the CRP level were lacking in many cases.

The consequences of missed or wrong follow-up of microcytic anemia may include incorrect treatment (eg, iron supplementation in a thalassemia patient, which leads to severe patient harm and may reflect clinical malpractice), no treatment (eg, delayed recovery after acute illness when no iron is supplemented), and missed genetic counseling of parents with thalassemia, which may lead to infants born with homozygous or double heterozygous thalassemias or hemoglobinopathies, potentially resulting in severe illness and the need for lifelong transfusion therapy.

When evaluating the subgroups, stratified by MCV values, the frequency of follow-up diagnostic testing concerning iron studies as well as thalassemia/hemoglobinopathy was higher in groups with lower MCV levels (Figure 2, A, D, and F), a finding that could lead to the assumption that the focus of attention in clinicians depends on how far test results deviate from their reference interval. Brown et al³⁹  reported that when presenting physicians with all available data of a patient within an electronic health record system, physicians tend to spend the least amount of their time (8%) reviewing laboratory results. Additionally, the main task when reviewing test results most probably lies in detecting deranged values by focusing mostly on emphasized results. We hypothesize that results with more severe flagging (eg, “+++” or “− − −”) are more likely to lead to medical action than those with lesser flagging, a theory that would put more gravity on the long-overdue issue of reformatting laboratory reports.⁴⁰ 

Interestingly, the proportion of patients in whom ferritin, serum iron, transferrin, or transferrin saturation was determined showed a negative trend with decreasing MCV value, a finding standing in contrast with the general positive trend of iron status testing with decreasing MCV values. However, ferritin was measured in most patients with iron status testing in all 3 groups (n = 538 [97%]; n = 400 [92%]; and n = 94 [82.5%] of patients in groups 1, 2, and 3, respectively), whereas soluble transferrin receptor testing remained at a similar level across all groups. As a possible cause, one could speculate that once an iron deficiency is identified, clinicians would most probably initiate iron supplementation or pursue other tests for root cause analysis not evaluated in this study, such as fecal occult blood or urine analysis.

Ret-Hb as a parameter to test for iron-deficient states has been known for some years.⁴¹  It may even have diagnostic value in the detection of thalassemia or hemoglobin variants.⁴²  However, when interviewing clinicians about this parameter during routine patient care, their knowledge about it is very heterogeneous, with the most knowledge at internal wards and lesser awareness in other departments (personal observation in our local setting). Ret-Hb holds additional valuable information while not adding to the laboratory's expenses. Therefore, it is one of the few reflex testing schemes implemented without an expressed need from the clinics and is hence measured with every reticulocyte count in our laboratory. This fact most probably explains the positive trend in testing this parameter with declining MCV levels. Additionally, we assume that even though Ret-Hb results are provided in many of the evaluated cases in this study, the interpretation thereof is often lacking.

Apart from the studied underuse of laboratory testing, we also identified distinct overuse in cases in which several parameters for iron deficiency testing were ordered simultaneously, although 1 or 2 of these might have been sufficient (Figure 2, E). Because this was not the focus of this study, we did not calculate the exact numbers of this overuse.

Many nonlaboratory physicians struggle with correct test selection and result interpretation.²⁴  In most medical curricula, laboratory diagnostics play only a minor role, if any.⁴³  Because this task is the main expertise of laboratory professionals, support of clinicians and continuous education lies within their responsibility. A recent survey in 9 European countries demonstrated that such support would be appreciated by most clinicians.⁴⁴  Of course, it would be out of range to assist in the appropriate use of laboratory tests in each single case personally. Therefore, semiautomated clinical decision support software systems, programmed and maintained by laboratory specialists, could be most helpful in the future. Such systems may cover the range from simple reflex testing (eg, ordering CRP whenever ferritin is ordered) to reflective testing (eg, lupus antibody investigation in patients with activated prothrombin time prolongation and no clinical sign of bleeding) to automated commenting (eg, “iron deficiency anemia” in patients with microcytic anemia and a ferritin level <30 ng/mL), or ideally to actual laboratory diagnostic algorithms, including result interpretation. The combination of results from simple analyses such as the red blood cell (RBC) count and erythrocyte indices could be evaluated automatically, based on published indices.¹³  In our study the calculation of the Mentzer index (MCV/RBC) in patients with no preceding MCV results >80 μm³ and no documented thalassemia diagnosis (n = 127) would have potentially identified 70 thalassemia patients (55.1%).

We are convinced that in the near future artificial intelligence in the form of machine learning or even deep learning systems will be part of many laboratory processes. To date, only a few software solutions, able to communicate with different information systems and/or instruments, are available. Clinical decision support software solutions, such as the one used in this study, are very promising solutions that can provide a detailed evaluation of the available data set. Nevertheless, laboratory specialists need to engage proactively in clinical care as consultants because the described systems are only able to make suggestions, based on their programming. There is an extensive knowledge from which clinicians and patients could benefit, but it is too often smoldering within the laboratories, waiting for someone to tap into it.

This manuscript is intended also to provide an algorithm for the diagnostic workup of microcytic anemia from which other laboratories may benefit. Because the definition of such an algorithm surely accounts for most of the work needed in order to implement semiautomated follow-up diagnostics into routine patient care, we hope and believe that our efforts may aid laboratories in this endeavor.

One might think that automation of parts of the diagnostic process would undermine the work of laboratory specialists. In fact, the opposite is true, because all of these systems need to be developed, implemented, and continuously revised based on current evidence, a task of which laboratory experts are most capable. Therefore, we believe that all these advances will help free up time currently used to perform repetitive tasks, in order to refocus on the medical part of our profession.

One limitation of our study is based on the fact that we only used test results available in our LIS. We could not account for all testing done outside of our hospital. However, because diagnoses like thalassemia or other chronic RBC disorders from external sources would surely have been added to the patient's record, which we included in our evaluation, we may assume that the number of patients with diagnosed but undocumented hemoglobinopathies/thalassemia was low or even negligible. The same is true for other diseases triggering microcytic anemia. All patients with a preexisting anemia diagnosis were ruled out by treating them as cases with correct diagnostic workup. In the investigation of an acute microcytic anemia, a diagnostic workup in the hospital would be mandatory because many of the needed parameters will not be analyzed externally, because these parameters are not reimbursed.

Another limitation is the fact that anemia of chronic disease was not incorporated into this study because this diagnosis cannot be met solely by the use of laboratory testing. Additionally, there surely will be some single cases within the studied cases in which a full laboratory workup is not necessary for the diagnosis (eg, elderly man with chronic gastrointestinal bleeding, microcytic anemia with high red blood cell distribution width, and previously normal complete blood count under current iron supplementation). Because the needed clinical information was not accessible to us in a structured manner and due to the fact that incorporating this information into the already very detailed algorithm would have overcomplicated the study, we refrained from including this information into our calculations.

The combination of all of these factors might slightly decrease the calculated percentages of misuse and underuse.

Forty percent of the 2244 patients with nondocumented microcytic anemia investigated in this study appeared not to have been followed up correctly by the use of laboratory testing.

To our knowledge, this is the first study investigating laboratory underuse on a broad scale in hospital patients with the only preselection of having a complete blood count ordered. Because undiagnosed patients may suffer from missing or wrong treatment, it is necessary that laboratory specialists act by implementing semiautomated systems aiding in test selection and interpretation, defined in close collaboration with clinicians.

We want to thank Nico Osterauer and Michael Gartner, BSc, as well as the company Abbott GmbH for supplying the AlinIQ CDS software alongside the manpower for data evaluation.