- Email: [email protected]
Failure to diagnose anemia in medical inpatients
Failure to Diagnose Anemia in Medical Inpatients Is the Traditional Diagnosis of Anemia a Dying Art?
KRISTI 0. SELF, B.S. MARY M. CONRADY, B.S. EDWARD R. EICHNER, M.D. Oklahoma
The diagnosis and treatment of anemia in medical inpatients were studied. Anemia was detected from the admitting complete blood cell count, the medical charts were perused for management of anemia by the physicians, and the cases of anemia were classified by the investigators, using the complete blood cell count and peripheral blood smear. Anemia was seldom dlagnosed in the traditional manner. In about 25 percent of cases, it was not even recognized. In about 20 percent, it was recognized but not evaluated or treated. In the 55 percent of cases adequately diagnosed and/or treated, the diagnosis was often more “situational” than analytic, the treatment more empiric than specific. Physicians ignored or misused information from the complete blood cell count and smear description and examined the smear themselves in fewer than one in 10 anemic patients. Not all of the missed cases of anemla were mild, expected from the patient’s illness, or unimportant to the patient’s care.
City, Oklahoma
Anemia is not a disease but a nonspecific sign of diverse underlying diseases. Therefore, internists should be sensitive to anemia. Others have sliown, however, that physicians tend to ignore mild cases of anemia and improperly evaluate anemia [l-3]. To understand better the approach to anemia in our medical center, we conducted a prospective study in newly admitted m&W patients.
SUBJECTS AND METHODS
From the Departments of Medicine and Pathology, University of Oklahoma Health Sciences Center and Oklahoma City Veterans Administration Medical Center, Oklahoma City, Oklatioma. Requests for reprints should be addressed to Dr. Edward R. Eichner, Hematology Section, University of Oklahoma Health Sciences Center, P.O. Box 26901, Oklahoma City, Oklahoma 73190. Manuscript submitted September 18, 1985, and accepted November 11, 1985.
786
Ndvember
1986
The American
Journal
The study group comprised all patients admitted to the medical service of a university hospital for four three-week periods: in June and July 1984, and in June and July 1985. June and July were used to compare new with experienced house staff. Anemia was defined as a hemoglobin value of 13 g/dl or less for men, and 12 g/dl or less for women. Anemia was identified upon admission by screening the routine complete blood cell count obtained for all new patients via an Ortho ELT-8 particle counter. The hemoglobin values were obtained again within 24 hours of admission in most patients, s6 we recorded the second value, too, to control for rehydration. The lower hemoglobin value of the two was used for definition. After a case of anemia was identified, that patient’s peripheral biood smear was obtained from the main hematology laboratory within 12 to 24 hours of admission (smears are routinely used as quality control for the complete blood cell count). Thus, if a physician or student wanted to see an anemic patient’s smear, he or she had to come to the special hematology section, where a record of the visit could be kept. The smear was reviewed by a hematologist, with only the complete blood cell count as an aid for diagnosis. The hematologist’s classification of the anemia was recorded for comparison with the patient’s discharge diagnosis and with the routine hematology morphologic summary.
of Medicine
Volume
81
FAILURE
TABLE
I
Recognition
Rate by Category
16 8 6 22 7 17 8 a4
ANEMIA-SELF
ET AL
of Anemia
June to July 1984 Number Recognized Number Microcytic/hypochromic Hemorrhagic/hemolytic Macrocytic Chronic renal disease Marrow damage Chronic inflammation Mixed/miscellaneous Total
TO DIAGNOSE
June 10 July 1985 Number Number Recognized
13 7 4 16 7 12 6 65
23 23 14 25 3 18 9 115
The patients’ charts were perused within 24 hours of admission and again after discharge. The house staff and attending physicians were unaware of our purpose. Recognition of anemia was defined as any mention of anemia or action to evaluate or treat anemia by medical students, interns, residents, or attending physicians. Initial assessments, all progress reports, all orders, laboratory test results, and discharge summaries were checked for management of anemia, although the simple recording of hemoglobin/hematocrit values in the admitting write-up was not counted as recognition. The physicians’ names were recorded for each case. Anemia was categorized by the hematologist as follows: (1) anemia of chronic inflammation, with normocytic/normochromic or slightly microcytic/hypochromic red cells and no increase in polychromasia; (2) hemorrhagic or hemolytic anemia, with major polychromasia, usually normocytic/normochromic, and sometimes specific morphologic clues such as sickle cells, schistocytes, and spherocytes; (3) microcytic/hypochromic anemia, with some polychromasia, advanced microcytic and hypochromic cells, anisocytosis, and, occasionally, target cells; (4) macrocytic anemia, with macro-ovalocytes and hypersegmented neutrophils; (5) marrow damage anemia, with a leukoerythroblastic smear and/or major anisopoikilocytosis and polychromasia, and seen here mainly in patients with hematologic malignancies (four with leukemia, two with lymphoma, two with myelodysplasia); (6) chronic renal disease, normocytic/normochromic, with variable degrees of burr cells and little or no polychromasia; and (7) miscellaneous and mixed types of anemia, including those in which possible iron-deficiency anemia could not be distinguished from possible anemia of chronic inflammation. All cases of anemia in which the smear and/or mean corpuscular volume strongly suggested iron deficiency were placed in the microcytic/hypochromic category. RESULTS
Twenty-six percent of the cases of anemia were in patients undergoing cancer chemotherapy. Because such patients might logically be assumed to be anemic from the chemotherapy, they were excluded from further analysis. In the first study year (1984), 84 (21.8 percent) of the
November
20 17 10 16 3 12 7 a5
Number 39 31 20 47 IO 35 17 199
Total Percent of Total
Percent Recognized
19.6 15.6 10.1 23.6 5.0 17.6 8.5
84.6 77.4 70.0 68.1 100 68.6 76.5
remaining 386 patients were anemic, versus 115 (38 percent) of 303 in the second year. The anemia ranged from mild to severe; 19 percent of the oases were mild (hemoglobin value of 11.1 to 12.0 g/d1 in women, 12.1 to 13.0 g/dl in men), and 16 percent were severe (hemoglobin value below 8 g/dl). Most cases were at least moderate; in 60 percent, values were IO g/d1 or less in women and 11 g/dl or less in men. In the hematologist’s categorization of the cases of anemia (Table I), the most common, respectively, were: chronic renal disease, microcytic/hypochromic, chronic inflammation, hemorrhagic/hemolytic and macrocytic. This distribution is generally consistent with prior studies of anemia in medical inpatients [4,5]. This categorization, based on only the complete blood cell count and peripheral smear, was provisional and could not be proved in the cases not evaluated by the patients’ physicians. For the thoroughly evaluated cases, however, this categorization generally proved correct; in the second study year, only 11 (9.6 percent) of the 115 cases were recategorized at our final review of all clinicai and laboratory data. This categorization was not to prove that the traditional diagnosis of anemia is effective, but to understand the kinds of anemia we were dealing with, to see if management differed by type of anemia, and to detect early any serious hematologic problems that would call for our communicating with patients’ physicians. Table I also shows the rates of recognition of the presence of anemia by the patients’ physicians. In the first study year, 65 (77.4 percent) of 84 cases of anemia were recognized, versus 85 (73.9 percent) of 115 in the second study year. There was no difference in recognition rate in June versus July. There was no significant difference in recognition rate by category of anemia. The 100 percent recognition of marrow damage anemia was attributed to prior knowiedge of the diagnosis of these hematologic malignancies. The 85 percent recognition rate for microcytic/hypochromic anemia is somewhat inflated by the cases in which the type of anemia was identified solely or mainly on the basis of the clinical
1996
The
American
Journal
of Medicine
Volume
61
767
FAILURE
TO DIAGNOSE
1
ANEMIA-SELF
’
5
I
1
7 HEMOGLOBIN
ET AL
1
’
9
I
1
11
failure to recognize the anemia may have had little clinical importance; four had end-stage renal disease with unremarkable complete blood cell counts and peripheral smears, and six had the anemia of chronic inflammation with either very mild anemia or an underlying disease sufficient to account for the anemia. Twenty patients, however, could have benefited from appraisal of the anemia: 12 appeared to have iron-deficiency anemia (six in the coronary care unit, two with diabetes! and one each with cancer, poiyarteritis, pneumonia, and renal disease); three cases of macrocytic anemia in alcoholic patients were missed; three patients with end-stage renal disease had unusual blood smears (schistocytes, spherocytes, and target cells, respectively); and two other patients had many target cells. Roughly 20 percent of the cases of anemia were recognized but either not evaluated or misidentified. Seven of these 24 cases of anemia may have had little clinical importance; they occurred in inpatients with endstage renal disease who had unremarkable complete blood cell counts and blood smears. Eight other cases should have been evaluated, however, including four in patients with end-stage renal disease who had abnormal smears. In five cases, the diagnosis was “situational,” not analytic, (in two patients with sickle ceil hemoglobinopathy, two with alcoholism, and one with gastrointestinal bleeding). Four cases were misdiagnosed: two macrocytic anemias were misclassified (one as microcytic, one as “rehydration” anemia), a clearcut anemia of chronic inflammation was mistaken for hemorrhagic anemia, and a clear&t iron-deficiency anemia was mistaken for the anemfa of renal disease. Even in the roughly 55 percent of the cases adequately managed, some of the diagnoses seemed to be more “situational” than analytic. In addition, treatment of the anemia tended to be empiric; blood and iron were used for patients with bleeding and folic acid was given to alcoholic patients.
I
13
(g/di)
igure 1. Recognition of anem&. Recognition rate (percenQ by severjty of arieri7ia (hemoglobin concentration): 1984 (circles), 1985 (squares), and combined data (friang/es).
setting rather than on the basis of the hematologic data in conjunction with the clinical setting. That is, some of the diagnoses of iron-deficiency anemia were based on the “situation” or clinical setting when a patient was admitted with prominent gastrointestinal bleeding. The correct diagnosis of iron-deficiency anemia was made, but with no recorded analysis of routine complete blood cell count or blood smear data. The recognition rates for the other five categories of anemia are ajl in the same lower range of 68 to 77 percent. Fig&e 1 shows recognition rate by severity of anemia. Recognition rate increased with severity of anemia but did not reach 100 percent until the hemoglobin value was below 9 g/dl. There was a trend toward greater recogni2 tion of anemia in women; 78 percent of the cases of anemia in women, versus 70 percent in men, were recognized (not significant). There was a difference in recognition rate by location of the patient; only 38.5 percent (five of 13) of the cases of anemia were recognized in the coronary care unit, versus 78.4 percent (80 of 102) in all the other medical wards combined (chisquare i 9.6, p KO.005). Table II shows the physicians’ management of all 115 cases of anemia in the second study year. In 30 patients (26.1 percent), the anemia was not even recognized. in general, these cases tended to be less severe; 64 percent of the mildest cases were in this category. Some of these cases, however, were not mild. In 10 of these patients,
788
November
1988
The American
Journal
of Medicine
COqMENTS
In this study, about 25 percent of the cases of anemia were not recognized, about 20 percent were recognized but not evaluated or treated, and diagnosis was sometimes “situational” and treatment sometimes “empiric” in the 55 percent that were adequately managed. The physicians examined the blood smear for only 8 percent of their anemic patients. Cjthers have argued that physicians need not examine the smears if routine evaluations of those smears are accurate [5]. We found such e$aluations accurate, but physicians did not always heed them. Some physicians also ignored or misjudged the mean corpuscular volume. in short, our physicians seldom identify anemia in the traditional manner. Others have shown that the diagnosis of anemia can
Volume
81
FAILURE
TABLE
II
Physician
Category No recognition
Recognition
only
Misdiagnosis Recognition/ diagnosis Recognition/ treatment
Percent
Number of Patients*
is given
of 115 Cases of Anemia
in Second
Number of Patients by Severity of Anemia (gidl) Men 113 112 111 Women I12 511 210 7
7
20 (17.4)
3
4
13
1 0
1 1
2 4
16 (13.9)
1
3
12
18 (15.7)
2
1
15
(7.8)
0
4
5
Anemia of chronic damage anemia
13 (11.3)
2
3
8
Blood loss and/or iron-deficiency alcoholic patients (3); renal ic inflammation (2)
9
ET AL
Comments on Management
16
(3.5) (4.3)
ANEMIA-SELF
Study Year
30 (26.1)
4 5
Recognition/ diagnosis/ treatment Recognition/ diagnosis with smear examination/ treatment Work-up on previous admission
l
Management
TO DIAGNOSE
20 cases suggested or showed: iron deficiency anemia (12); megaloblastic anemia (3); many target ceils (3); schistocytes (1); spherocyfes (1) In 5 cases, the diagnosis was situational; 8 other cases suggested or showed: iron deficiency anemia (5); megaloblastic anemia (1); schistocytes (1); thalassemia (1) Macrocytic anemia misclassified (2); see text In 2 cases, the diagnosis was situational; 3 cases were assumed (correctly) to be anemia of chronic inflammation In 12 cases, therapy was empiric: patients with bleeding were given blood or iron (10); alcoholic patients were given folic acid (2); see text Good management of blood loss and/or iron-deficiency anemia (10); macrocytic anemia (3); renal disease anemia (2); others (3) inflammation (4); blood (2); SC hemoglobinopathy
loss anemia (1)
(2); marrow
anemia (6); macrocytic anemia in disease anemia (2); anemia of chron-
in parentheses.
vary with the sex of the patient, the location of the patient in the hospital, and the house staff assigned to the patient [6]. We, too, found greater recognition of anemia in women, but this trend could have occurred by chance. We found a lower recognition rate for anemia in the coronary care unit. As for house staff, one person, a thirdyear student, accounted for one third (three of nine) of the blood smears examined. We also found that some teams frequently identified the anemia, whereas others rarely did. Why are hematologic data unheeded? Physicians may forget the correct textbook definition of anemia [2], or they may be “information-overloaded” [7,8]. Our computer printouts of laboratory data are not easy to decipher quickly. Physicians, however, peruse routine chest radiographic films and electrocardiographic recordings, and in this study, we were impressed that they usually noted mild abnormalities in renal and liver function on the computer printouts. Similar attention to routine hematologic data would improve patient care. The nondiagnosis of anemia may be more apparent than real in some cases. Almost 40 percent (19 of 50) of the cases of anemia not recognized or evaluated were in patients with end-stage renal disease. Physicians may recognize this anemia but not mention or evaluate it because they assume it is implicit in, and explained by,
November
the diagnosis of end-stage renal disease. If this is the case, it reflects an oversimplified view of anemia. In this study, 40 percent of the patients with end-stage renal disease had clues in the blood smear to mechanisms of anemia besides erythropoietin lack: iron deficiency, folate deficiency, and hemolysis. The “situational” diagnoses-that patients with bleeding have “iron deficiency,” alcoholic patients have “anemia of alcoholism,” and patients with inflammatory diseases have “anemia of chronic disease”-also imply that too little thought is given to mixed types of anemia. In our study, several patients thought to have the anemia of chronic disease had blood smear clues inconsistent with that diagnosis: striking polychramasia, schistocytes, or clearcut irondeficiency anemia. Several men were sent home with undiagnosed iron-deficiency anemia. All these oversights have implications for patient care. What can be done to improve the management of anemia? We need to re-emphasize its definition, implications, and proper diagnosis. We need to simplify our computer printouts to highlight important features of the complete blood cell count and blood smear results. Chart reminders about anemia had no impact in one study [B], but a well-designed, informative chart reminder should improve the recognition, diagnosis, and therapy of anemia.
1986
The American
Journal
of Medicine
Volume
91
789
FAILURE
TO DIAGNOSE
ANEMIA-SELF
ET AL
REFERENCES 1.
2. 3. 4. 5.
790
Carmel R: Macrocytosis, mild ariemia, and delay in the diagnosis of pernicious anemia. Arch Intern Med 1979; 139: 47-50. darniel R, Denson TI), Mussel1 6: Anemia. Textbook vs practice: JAMA 1979; 242: 2295-2297. spivtik JL: Masked megdl6blastic anemia. Arch Intern Med 1982; 142: 2111:2114. Paine CJ, Poik A; Eichner ER: Analysii of anemia in medical inpatients. Am J Med Sci 1974; 268: 37-44. Jen P, Woo 6, Rosenthal PE, Bunn HF, Loscalzo A, Goldman
November
1986
The
American
Journal
of Medicine
6.
7. 8.
Volume
81
L: The value of the peripheral blood smear in anemia inpatients. The laboratory’s reading v. a physician’s rgading. Arch Intern Med 1983; 143: 1120-l 125. Woo B, Jen P, Rosenthal i)E, !&nn HF, Goldman L: Anemic inpatients. Correlates of house dfficer performanre. Arch intern Med 1981; 141: 1199-1202. McDonald CJ, Wilson GA, McCabe GP: Response to computer reminders. JAMA 1980; 244: 1579-1581. Wigton RS, Zimmer JL, Wigton JH, Patil KD: Chart rdminderk in the diagnosis of anemia. JAMA 1981; 245: 1745-1747.
KRISTI 0. SELF, B.S. MARY M. CONRADY, B.S. EDWARD R. EICHNER, M.D. Oklahoma
The diagnosis and treatment of anemia in medical inpatients were studied. Anemia was detected from the admitting complete blood cell count, the medical charts were perused for management of anemia by the physicians, and the cases of anemia were classified by the investigators, using the complete blood cell count and peripheral blood smear. Anemia was seldom dlagnosed in the traditional manner. In about 25 percent of cases, it was not even recognized. In about 20 percent, it was recognized but not evaluated or treated. In the 55 percent of cases adequately diagnosed and/or treated, the diagnosis was often more “situational” than analytic, the treatment more empiric than specific. Physicians ignored or misused information from the complete blood cell count and smear description and examined the smear themselves in fewer than one in 10 anemic patients. Not all of the missed cases of anemla were mild, expected from the patient’s illness, or unimportant to the patient’s care.
City, Oklahoma
Anemia is not a disease but a nonspecific sign of diverse underlying diseases. Therefore, internists should be sensitive to anemia. Others have sliown, however, that physicians tend to ignore mild cases of anemia and improperly evaluate anemia [l-3]. To understand better the approach to anemia in our medical center, we conducted a prospective study in newly admitted m&W patients.
SUBJECTS AND METHODS
From the Departments of Medicine and Pathology, University of Oklahoma Health Sciences Center and Oklahoma City Veterans Administration Medical Center, Oklahoma City, Oklatioma. Requests for reprints should be addressed to Dr. Edward R. Eichner, Hematology Section, University of Oklahoma Health Sciences Center, P.O. Box 26901, Oklahoma City, Oklahoma 73190. Manuscript submitted September 18, 1985, and accepted November 11, 1985.
786
Ndvember
1986
The American
Journal
The study group comprised all patients admitted to the medical service of a university hospital for four three-week periods: in June and July 1984, and in June and July 1985. June and July were used to compare new with experienced house staff. Anemia was defined as a hemoglobin value of 13 g/dl or less for men, and 12 g/dl or less for women. Anemia was identified upon admission by screening the routine complete blood cell count obtained for all new patients via an Ortho ELT-8 particle counter. The hemoglobin values were obtained again within 24 hours of admission in most patients, s6 we recorded the second value, too, to control for rehydration. The lower hemoglobin value of the two was used for definition. After a case of anemia was identified, that patient’s peripheral biood smear was obtained from the main hematology laboratory within 12 to 24 hours of admission (smears are routinely used as quality control for the complete blood cell count). Thus, if a physician or student wanted to see an anemic patient’s smear, he or she had to come to the special hematology section, where a record of the visit could be kept. The smear was reviewed by a hematologist, with only the complete blood cell count as an aid for diagnosis. The hematologist’s classification of the anemia was recorded for comparison with the patient’s discharge diagnosis and with the routine hematology morphologic summary.
of Medicine
Volume
81
FAILURE
TABLE
I
Recognition
Rate by Category
16 8 6 22 7 17 8 a4
ANEMIA-SELF
ET AL
of Anemia
June to July 1984 Number Recognized Number Microcytic/hypochromic Hemorrhagic/hemolytic Macrocytic Chronic renal disease Marrow damage Chronic inflammation Mixed/miscellaneous Total
TO DIAGNOSE
June 10 July 1985 Number Number Recognized
13 7 4 16 7 12 6 65
23 23 14 25 3 18 9 115
The patients’ charts were perused within 24 hours of admission and again after discharge. The house staff and attending physicians were unaware of our purpose. Recognition of anemia was defined as any mention of anemia or action to evaluate or treat anemia by medical students, interns, residents, or attending physicians. Initial assessments, all progress reports, all orders, laboratory test results, and discharge summaries were checked for management of anemia, although the simple recording of hemoglobin/hematocrit values in the admitting write-up was not counted as recognition. The physicians’ names were recorded for each case. Anemia was categorized by the hematologist as follows: (1) anemia of chronic inflammation, with normocytic/normochromic or slightly microcytic/hypochromic red cells and no increase in polychromasia; (2) hemorrhagic or hemolytic anemia, with major polychromasia, usually normocytic/normochromic, and sometimes specific morphologic clues such as sickle cells, schistocytes, and spherocytes; (3) microcytic/hypochromic anemia, with some polychromasia, advanced microcytic and hypochromic cells, anisocytosis, and, occasionally, target cells; (4) macrocytic anemia, with macro-ovalocytes and hypersegmented neutrophils; (5) marrow damage anemia, with a leukoerythroblastic smear and/or major anisopoikilocytosis and polychromasia, and seen here mainly in patients with hematologic malignancies (four with leukemia, two with lymphoma, two with myelodysplasia); (6) chronic renal disease, normocytic/normochromic, with variable degrees of burr cells and little or no polychromasia; and (7) miscellaneous and mixed types of anemia, including those in which possible iron-deficiency anemia could not be distinguished from possible anemia of chronic inflammation. All cases of anemia in which the smear and/or mean corpuscular volume strongly suggested iron deficiency were placed in the microcytic/hypochromic category. RESULTS
Twenty-six percent of the cases of anemia were in patients undergoing cancer chemotherapy. Because such patients might logically be assumed to be anemic from the chemotherapy, they were excluded from further analysis. In the first study year (1984), 84 (21.8 percent) of the
November
20 17 10 16 3 12 7 a5
Number 39 31 20 47 IO 35 17 199
Total Percent of Total
Percent Recognized
19.6 15.6 10.1 23.6 5.0 17.6 8.5
84.6 77.4 70.0 68.1 100 68.6 76.5
remaining 386 patients were anemic, versus 115 (38 percent) of 303 in the second year. The anemia ranged from mild to severe; 19 percent of the oases were mild (hemoglobin value of 11.1 to 12.0 g/d1 in women, 12.1 to 13.0 g/dl in men), and 16 percent were severe (hemoglobin value below 8 g/dl). Most cases were at least moderate; in 60 percent, values were IO g/d1 or less in women and 11 g/dl or less in men. In the hematologist’s categorization of the cases of anemia (Table I), the most common, respectively, were: chronic renal disease, microcytic/hypochromic, chronic inflammation, hemorrhagic/hemolytic and macrocytic. This distribution is generally consistent with prior studies of anemia in medical inpatients [4,5]. This categorization, based on only the complete blood cell count and peripheral smear, was provisional and could not be proved in the cases not evaluated by the patients’ physicians. For the thoroughly evaluated cases, however, this categorization generally proved correct; in the second study year, only 11 (9.6 percent) of the 115 cases were recategorized at our final review of all clinicai and laboratory data. This categorization was not to prove that the traditional diagnosis of anemia is effective, but to understand the kinds of anemia we were dealing with, to see if management differed by type of anemia, and to detect early any serious hematologic problems that would call for our communicating with patients’ physicians. Table I also shows the rates of recognition of the presence of anemia by the patients’ physicians. In the first study year, 65 (77.4 percent) of 84 cases of anemia were recognized, versus 85 (73.9 percent) of 115 in the second study year. There was no difference in recognition rate in June versus July. There was no significant difference in recognition rate by category of anemia. The 100 percent recognition of marrow damage anemia was attributed to prior knowiedge of the diagnosis of these hematologic malignancies. The 85 percent recognition rate for microcytic/hypochromic anemia is somewhat inflated by the cases in which the type of anemia was identified solely or mainly on the basis of the clinical
1996
The
American
Journal
of Medicine
Volume
61
767
FAILURE
TO DIAGNOSE
1
ANEMIA-SELF
’
5
I
1
7 HEMOGLOBIN
ET AL
1
’
9
I
1
11
failure to recognize the anemia may have had little clinical importance; four had end-stage renal disease with unremarkable complete blood cell counts and peripheral smears, and six had the anemia of chronic inflammation with either very mild anemia or an underlying disease sufficient to account for the anemia. Twenty patients, however, could have benefited from appraisal of the anemia: 12 appeared to have iron-deficiency anemia (six in the coronary care unit, two with diabetes! and one each with cancer, poiyarteritis, pneumonia, and renal disease); three cases of macrocytic anemia in alcoholic patients were missed; three patients with end-stage renal disease had unusual blood smears (schistocytes, spherocytes, and target cells, respectively); and two other patients had many target cells. Roughly 20 percent of the cases of anemia were recognized but either not evaluated or misidentified. Seven of these 24 cases of anemia may have had little clinical importance; they occurred in inpatients with endstage renal disease who had unremarkable complete blood cell counts and blood smears. Eight other cases should have been evaluated, however, including four in patients with end-stage renal disease who had abnormal smears. In five cases, the diagnosis was “situational,” not analytic, (in two patients with sickle ceil hemoglobinopathy, two with alcoholism, and one with gastrointestinal bleeding). Four cases were misdiagnosed: two macrocytic anemias were misclassified (one as microcytic, one as “rehydration” anemia), a clearcut anemia of chronic inflammation was mistaken for hemorrhagic anemia, and a clear&t iron-deficiency anemia was mistaken for the anemfa of renal disease. Even in the roughly 55 percent of the cases adequately managed, some of the diagnoses seemed to be more “situational” than analytic. In addition, treatment of the anemia tended to be empiric; blood and iron were used for patients with bleeding and folic acid was given to alcoholic patients.
I
13
(g/di)
igure 1. Recognition of anem&. Recognition rate (percenQ by severjty of arieri7ia (hemoglobin concentration): 1984 (circles), 1985 (squares), and combined data (friang/es).
setting rather than on the basis of the hematologic data in conjunction with the clinical setting. That is, some of the diagnoses of iron-deficiency anemia were based on the “situation” or clinical setting when a patient was admitted with prominent gastrointestinal bleeding. The correct diagnosis of iron-deficiency anemia was made, but with no recorded analysis of routine complete blood cell count or blood smear data. The recognition rates for the other five categories of anemia are ajl in the same lower range of 68 to 77 percent. Fig&e 1 shows recognition rate by severity of anemia. Recognition rate increased with severity of anemia but did not reach 100 percent until the hemoglobin value was below 9 g/dl. There was a trend toward greater recogni2 tion of anemia in women; 78 percent of the cases of anemia in women, versus 70 percent in men, were recognized (not significant). There was a difference in recognition rate by location of the patient; only 38.5 percent (five of 13) of the cases of anemia were recognized in the coronary care unit, versus 78.4 percent (80 of 102) in all the other medical wards combined (chisquare i 9.6, p KO.005). Table II shows the physicians’ management of all 115 cases of anemia in the second study year. In 30 patients (26.1 percent), the anemia was not even recognized. in general, these cases tended to be less severe; 64 percent of the mildest cases were in this category. Some of these cases, however, were not mild. In 10 of these patients,
788
November
1988
The American
Journal
of Medicine
COqMENTS
In this study, about 25 percent of the cases of anemia were not recognized, about 20 percent were recognized but not evaluated or treated, and diagnosis was sometimes “situational” and treatment sometimes “empiric” in the 55 percent that were adequately managed. The physicians examined the blood smear for only 8 percent of their anemic patients. Cjthers have argued that physicians need not examine the smears if routine evaluations of those smears are accurate [5]. We found such e$aluations accurate, but physicians did not always heed them. Some physicians also ignored or misjudged the mean corpuscular volume. in short, our physicians seldom identify anemia in the traditional manner. Others have shown that the diagnosis of anemia can
Volume
81
FAILURE
TABLE
II
Physician
Category No recognition
Recognition
only
Misdiagnosis Recognition/ diagnosis Recognition/ treatment
Percent
Number of Patients*
is given
of 115 Cases of Anemia
in Second
Number of Patients by Severity of Anemia (gidl) Men 113 112 111 Women I12 511 210 7
7
20 (17.4)
3
4
13
1 0
1 1
2 4
16 (13.9)
1
3
12
18 (15.7)
2
1
15
(7.8)
0
4
5
Anemia of chronic damage anemia
13 (11.3)
2
3
8
Blood loss and/or iron-deficiency alcoholic patients (3); renal ic inflammation (2)
9
ET AL
Comments on Management
16
(3.5) (4.3)
ANEMIA-SELF
Study Year
30 (26.1)
4 5
Recognition/ diagnosis/ treatment Recognition/ diagnosis with smear examination/ treatment Work-up on previous admission
l
Management
TO DIAGNOSE
20 cases suggested or showed: iron deficiency anemia (12); megaloblastic anemia (3); many target ceils (3); schistocytes (1); spherocyfes (1) In 5 cases, the diagnosis was situational; 8 other cases suggested or showed: iron deficiency anemia (5); megaloblastic anemia (1); schistocytes (1); thalassemia (1) Macrocytic anemia misclassified (2); see text In 2 cases, the diagnosis was situational; 3 cases were assumed (correctly) to be anemia of chronic inflammation In 12 cases, therapy was empiric: patients with bleeding were given blood or iron (10); alcoholic patients were given folic acid (2); see text Good management of blood loss and/or iron-deficiency anemia (10); macrocytic anemia (3); renal disease anemia (2); others (3) inflammation (4); blood (2); SC hemoglobinopathy
loss anemia (1)
(2); marrow
anemia (6); macrocytic anemia in disease anemia (2); anemia of chron-
in parentheses.
vary with the sex of the patient, the location of the patient in the hospital, and the house staff assigned to the patient [6]. We, too, found greater recognition of anemia in women, but this trend could have occurred by chance. We found a lower recognition rate for anemia in the coronary care unit. As for house staff, one person, a thirdyear student, accounted for one third (three of nine) of the blood smears examined. We also found that some teams frequently identified the anemia, whereas others rarely did. Why are hematologic data unheeded? Physicians may forget the correct textbook definition of anemia [2], or they may be “information-overloaded” [7,8]. Our computer printouts of laboratory data are not easy to decipher quickly. Physicians, however, peruse routine chest radiographic films and electrocardiographic recordings, and in this study, we were impressed that they usually noted mild abnormalities in renal and liver function on the computer printouts. Similar attention to routine hematologic data would improve patient care. The nondiagnosis of anemia may be more apparent than real in some cases. Almost 40 percent (19 of 50) of the cases of anemia not recognized or evaluated were in patients with end-stage renal disease. Physicians may recognize this anemia but not mention or evaluate it because they assume it is implicit in, and explained by,
November
the diagnosis of end-stage renal disease. If this is the case, it reflects an oversimplified view of anemia. In this study, 40 percent of the patients with end-stage renal disease had clues in the blood smear to mechanisms of anemia besides erythropoietin lack: iron deficiency, folate deficiency, and hemolysis. The “situational” diagnoses-that patients with bleeding have “iron deficiency,” alcoholic patients have “anemia of alcoholism,” and patients with inflammatory diseases have “anemia of chronic disease”-also imply that too little thought is given to mixed types of anemia. In our study, several patients thought to have the anemia of chronic disease had blood smear clues inconsistent with that diagnosis: striking polychramasia, schistocytes, or clearcut irondeficiency anemia. Several men were sent home with undiagnosed iron-deficiency anemia. All these oversights have implications for patient care. What can be done to improve the management of anemia? We need to re-emphasize its definition, implications, and proper diagnosis. We need to simplify our computer printouts to highlight important features of the complete blood cell count and blood smear results. Chart reminders about anemia had no impact in one study [B], but a well-designed, informative chart reminder should improve the recognition, diagnosis, and therapy of anemia.
1986
The American
Journal
of Medicine
Volume
91
789
FAILURE
TO DIAGNOSE
ANEMIA-SELF
ET AL
REFERENCES 1.
2. 3. 4. 5.
790
Carmel R: Macrocytosis, mild ariemia, and delay in the diagnosis of pernicious anemia. Arch Intern Med 1979; 139: 47-50. darniel R, Denson TI), Mussel1 6: Anemia. Textbook vs practice: JAMA 1979; 242: 2295-2297. spivtik JL: Masked megdl6blastic anemia. Arch Intern Med 1982; 142: 2111:2114. Paine CJ, Poik A; Eichner ER: Analysii of anemia in medical inpatients. Am J Med Sci 1974; 268: 37-44. Jen P, Woo 6, Rosenthal PE, Bunn HF, Loscalzo A, Goldman
November
1986
The
American
Journal
of Medicine
6.
7. 8.
Volume
81
L: The value of the peripheral blood smear in anemia inpatients. The laboratory’s reading v. a physician’s rgading. Arch Intern Med 1983; 143: 1120-l 125. Woo B, Jen P, Rosenthal i)E, !&nn HF, Goldman L: Anemic inpatients. Correlates of house dfficer performanre. Arch intern Med 1981; 141: 1199-1202. McDonald CJ, Wilson GA, McCabe GP: Response to computer reminders. JAMA 1980; 244: 1579-1581. Wigton RS, Zimmer JL, Wigton JH, Patil KD: Chart rdminderk in the diagnosis of anemia. JAMA 1981; 245: 1745-1747.