Introduction
Anemia, defined as a reduction of the hemoglobin concentration, hematocrit, or red
blood cell count, is an extremely common disorder affecting 1.62 billion people worldwide,
or nearly one-quarter of the world’s population. Iron deficiency anemia (IDA) accounts
for approximately one-half of this disease burden. The most common cause of IDA in
developed countries is blood loss. This can be either overt or occult [1]. Upper and lower gastrointestinal endoscopy for the investigation of anemia is therefore
common practice in public hospitals throughout Australia.
There is a significant body of literature justifying endoscopy in the setting of IDA
[2]. The indication for endoscopy to investigate anemia of causes other than iron deficiency
is less clear. Nonetheless, referral to exclude occult gastrointestinal bleeding in
patients with anemia of all causes remains a common occurrence [3]. Endoscopic investigation in this setting places further pressure on already strained
hospital systems, with escalating costs, increased waiting times for outpatient procedures,
and higher numbers of patients needing admission for bowel preparation.
Given these issues, a way of clearly stratifying the patients most likely to benefit
from endoscopic investigation vs. those who are not is required. We hypothesize that
the endoscopic investigation of non-IDA has less diagnostic utility and leads to less
changes in management than does the endoscopic investigation of IDA.
In this study, our primary aim was to determine the diagnostic yield of endoscopy
in patients referred to undergo investigation for anemia by comparing outcomes in
those with true IDA vs. outcomes in those with other types of anemia. Our secondary
aim was to identify and evaluate additional factors enabling the risk stratification
of patients likely to benefit from endoscopic investigation. In addition, we analyzed
the cost associated with endoscopic investigation for anemia within our institution.
Methods
We performed a retrospective analysis of all patients referred for endoscopy with
the indications of “anemia” and “IDA” at a major tertiary hospital in Melbourne, Australia,
over a 12-month period. Ethics approval was granted by the hospital’s human research
ethics committee. Patients with acute overt bleeding, positive fecal occult blood
test results, and known varices were excluded. Data were collected on patient demographics,
referring unit, stated indication for referral, inpatient or outpatient procedure,
type of endoscopy performed, hemoglobin level, and mean cell volume (MCV) at the time
of referral. The Charlson Comorbidity Index was calculated with methods previously
described in the literature [4]. Patients were separated into three categories: (i) true or early IDA (low hemoglobin
level, low or normal MCV, low ferritin); (ii) tissue iron deficiency without anemia
(TIDWA; normal hemoglobin, low ferritin, low or normal MCV); and (iii) anemia of other
cause (AOC; low hemoglobin, normal ferritin, normal or high MCV). Outcome measures
included the detection of a lesion on endoscopy explaining the anemia (defined as
a gastrointestinal malignancy, bleeding peptic ulcer, active/luminal bleeding of any
other cause, or celiac disease) and a significant change in management as a result
of the endoscopy (defined as the removal of a polyp ≥ 10 mm in size, the removal of
multiple polyps, argon photocoagulation treatment of nonbleeding angioectasia, or
other change in management deemed significant by the clinician). Per hospital endoscopic
protocol, when required, gastric and duodenal biopsy specimens were obtained to rule
out celiac disease, pernicious anemia, and Helicobacter pylori infection, even when the endoscopic appearance was normal. If a patient had two separate
pathologic conditions, the one most likely to result in anemia was taken as the reference.
To establish the cost associated with endoscopic procedures within our institution,
an activity-based costing method that is widely used within hospitals in Victoria,
Australia, was used [5]. This method allocates costs directly to individual patient episodes using various
cost drivers, such as transaction, duration, and intensity cost drivers (e. g., theater
duration), to obtain accurate costs. An up-to-date general ledger was extracted from
the finance system and loaded into health care-specific clinical costing software
to generate the cost of the admission being analyzed. The cost of patient care was
divided into the cost of direct patient care and other, remaining cost centers not
providing direct patient care, known as “overhead areas” or indirect costs. The total
cost was then calculated from an addition of direct and indirect costs.
Descriptive statistics used included median and interquartile range (IQR; 25th and
75th percentiles). Univariable analyses for continuous variables were carried out
using the Kruskal – Wallis one-way analysis of variance test. Binary variable analyses
were performed with Fisher’s exact test. A multiple logistic regression analysis was
performed with the Charlson Comorbidity Index, inpatient vs. outpatient procedure,
and type of endoscopy as independent variables and a referral for endoscopic investigation
of AOC as the dependent variable. For all statistical tests, a two-tailed P value of less than 0.05 was regarded as indicative of statistical significance. Stata/IC
Statistical Software Release 13 (StataCorp LP; College Station, Texas, USA) was used
for analysis.
Results
In total, 283 patients with complete data were identified. Of these, 150 (53 %) met
the criteria for IDA, 14 (4.9 %) for TIDWA, and 119 (42 %) for AOC. No significant
differences were found for the patients’ demographics, referring units, or types of
procedure. Patients with a higher Charlson Comorbidity Index were significantly more
likely to be referred for the investigation of non-IDA ([Fig. 1]), and patients with AOC were more likely to undergo an inpatient rather than an
outpatient investigation ([Table 1]). Specific lesions identified during endoscopy are listed in [Table 2]. A likely cause of anemia was found in 31 patients with IDA (21 %) and 0 patients
in the other two categories (P < 0.001). A change of management was observed in 35 of 150 patients with IDA (23 %),
1 of 14 patients with TIDWA, and 8 of 119 patients (6.7 %) with AOC (P < 0.001). A logistic regression analysis revealed that after adjustment for other
factors, patients with AOC were significantly less likely to have a change of management
than patients in with IDA (odds ratio [OR] = 0.2, confidence interval [CI] 0.08 – 0.0.47,
P < 0.001).
Fig. 1 Charlson Comorbidity Index and anemia by category. AOC, anemia of other cause; TIDWA,
tissue iron deficiency without anemia; IDA, iron deficiency anemia.
Table 1
Baseline characteristics of 283 patients enrolled in a study of endoscopic investigation
for non-iron deficiency anemia.
|
Variable
|
IDA, n = 150 (53 %)
|
TIDWA, n = 14 (4.9 %)
|
AOC, n = 119 (42.1 %)
|
P value
|
|
Age, median (IQR), years
|
72 (54 – 79.25)
|
69.5 (57.25 – 77)
|
70 (57 – 78.5)
|
0.9
|
|
Sex
|
|
|
|
0.186
|
|
Male, n (%)
|
71 (47.3)
|
7 (50)
|
68 (57.1)
|
|
|
Female, n (%)
|
79 (52.7)
|
7 (50)
|
51 (42.9)
|
|
|
Referring unit
|
|
|
|
0.148
|
|
General medicine, n (%)
|
38 (25.3)
|
0 (0)
|
30 (25.2)
|
|
|
Gastroenterology, n (%)
|
55 (36.7)
|
8 (57.1)
|
29 (24.4)
|
|
|
Surgery, n (%)
|
11 (7.3)
|
1 (7.1)
|
13 (10.9)
|
|
|
General practitioner, n (%)
|
12 (8)
|
1 (7.1)
|
5 (4.2)
|
|
|
Other, n (%)
|
34 (22.7)
|
4 (28.6)
|
42 (35.3)
|
|
|
Inpatient procedure, n (%)
|
54 (36)
|
1 (7.1)
|
62 (52.1)
|
0.001
|
|
Procedure
|
|
|
|
0.294
|
|
Gastroscopy
|
60 (40)
|
3 (21.4)
|
50 (42)
|
|
|
Colonoscopy
|
32 (21.3)
|
3 (21.4)
|
32 (27)
|
|
|
Both
|
58 (38.7)
|
8 (57.1)
|
37 (31)
|
|
|
Categorical CCI
|
|
|
|
0.001
|
|
0, n (%)
|
29 (19.3)
|
2 (14.3)
|
10 (8.4)
|
|
|
1, n (%)
|
71 (47.3)
|
8 (57)
|
39 (32.8)
|
|
|
2, n (%)
|
36 (24)
|
2 (14.3)
|
37 (31.1)
|
|
|
3, n (%)
|
14 (9.3)
|
2 (14.3)
|
33 (27.7)
|
|
IDA, iron deficiency anemia; TIDWA, tissue iron deficiency without anemia; AOC, anemia
of other cause; CCI, Charlson Comorbidity Index.
Table 2
Characteristics of 45 lesions found on endoscopy.
|
Type of lesion
|
All lesions
|
IDA
|
TIDWA
|
AOC
|
|
Malignancy, n (%)
|
13 (29)
|
13 (100)
|
0
|
0
|
|
Ulcer, n (%)
|
7 (16)
|
7 (100)
|
0
|
0
|
|
Active bleeding, n (%)[1]
|
10 (22)
|
10 (100)
|
0
|
0
|
|
Polyps, n (%)[2]
|
14 (31)
|
5 (36)
|
1 (7)
|
8 (57)
|
|
Nonbleeding angioectasia requiring argon plasma coagulation, n (%)
|
0
|
0
|
0
|
0
|
|
Other, n (%)[3]
|
1 (2)
|
1 (100)
|
0
|
0
|
IDA, iron deficiency anemia; TIDWA, tissue iron deficiency without anemia; AOC, anemia
of other cause.
1 Including blood in lumen, bleeding angioectasia, bleeding erosive gastropathy, and
bleeding vessels.
2 Removal of polyp 10 mm or larger in size or removal of multiple polyps.
3 One case of small-bowel adenocarcinoma found on capsule endoscopy after negative
findings on colonoscopy and gastroscopy.
The current costing of endoscopy procedures at our hospital includes $ 1585 in direct
costs, which include costs for direct medical care (e. g., medical and nursing staff
labor), and $ 625 in indirect costs, which include costs for administration, electricity,
and information technology. Therefore, the total cost for either a gastroscopy or
colonoscopy at our institution was approximated to be $ 2210.
Discussion
Our study has shown that a significant proportion of patients are referred to our
hospital for the endoscopic investigation of non-IDA. Interestingly, referrals for
non-IDA increased as the Charlson Comorbidity Index rose, perhaps reflecting the large
number of investigations performed in this group of patients. Within our cohort, no
patients with anemia of a cause other than iron deficiency obtained an explanation
for their anemia after endoscopic investigation, and a significantly smaller proportion
of these patients than of those with IDA had a change in management as a result of
endoscopy. We were unable to draw conclusions from the cohort of patients within the
TIDWA subgroup because of the small sample size and the fact that this cohort likely
represents a group of patients with early IDA. No other factors were identified that
could lead to better stratification in the referral of patients for endoscopy.
This study highlights the fact that in the absence of iron deficiency, the chance
of finding significant pathology on endoscopy to explain anemia is likely to be small.
Only 8 of the 119 patients referred for the investigation of AOC had a change in management
as a result of endoscopy. We believe that these findings are consistent with the background
incidence of the conditions that required treatment – for example, colonic polyps
– and therefore are not likely to justify endoscopic investigation in this group of
patients.
There is minimal information in the literature regarding endoscopic evaluation in
patients with non-IDA. One such retrospective analysis looked at 100 consecutive patients
who were investigated endoscopically for non-IDA. The findings were similar to ours
in that the prevalence of lesions found to be responsible for anemia was significantly
lower in the non-IDA group than in the IDA group (8 % vs. 22.9 %, P < 0.001). Interestingly, the patients who had non-IDA were also more likely to have
an alternative cause of their anemia identified later, with chronic kidney disease
and hematological disease among the most common causes [3]. These conclusions are reflected in clinical practice guidelines, which suggest
that endoscopic investigation is likely to result in little diagnostic yield in patients
who have iron deficiency without anemia, and that if performed, endoscopic investigation
should be limited to those older than 50 years of age, in whom routine colon cancer
screening is recommended [6].
The fact that the referring unit, procedure performed, and Charlson Comorbidity Index
were not significantly associated with specified outcomes likely reflects otherwise
appropriate referrals within our institution. Previous attempts to identify factors
contributing to a higher likelihood of finding a lesion responsible for IDA have shown
that only older age, male gender, and lower MCV are predictive of finding a lesion
responsible for IDA [7]
[8]. The rate of lesion detection in our study may be somewhat lower than previously
published rates [9]
[10]
[11]. This discrepancy may be explained by other studies’ use of inclusion criteria that
accepted patients with symptoms and positive fecal occult blood test results, as well
as their use of less stringent criteria for what was defined as a lesion responsible
for anemia (e. g., esophagitis or gastritis).
We have found that endoscopic investigation for non-IDA comes at a significant cost
to our institution. If the patients without IDA did not undergo endoscopy, the savings
for our institution would equate to approximately $ 293 797 per annum. This value
is even higher when patients who require hospital admission for bowel preparation
are taken into account, with the average cost of a day in the hospital on a general
medical ward being approximately $ 720.
There were several limitations to our study. First, the retrospective nature of the
study inherently introduces confounders. One potential source of confounding was that
patients may have received undocumented iron supplementation. Second, in order to
clearly delineate what was a significant “cause for anemia” and “change in management,”
strict parameters had to be used for the definition. It is possible that a minority
of patients had an endoscopic lesion associated with anemia that was not accounted
for in our criteria.
In conclusion, a large proportion of patients were referred for the endoscopic investigation
of non-IDA within our cohort. Patients with a higher Charlson Comorbidity Index were
more likely than those with a lower score to be referred for endoscopic investigation
in the context of non-IDA. Endoscopic investigation for non-IDA comes at a significant
cost to our institution and has provided minimal yield both in finding the cause of
the patients’ anemia and in altering their management. Referring unit, procedure performed,
and Charlson Comorbidity Index were not associated with specified outcomes. We believe
that the wide referral base within our hospital makes these results generalizable
to other organizations.
Given the escalating costs of our health system, finding ways to minimize spending
without compromising patient care is of paramount importance. The endoscopic investigation
of patients with non-IDA is one such area. Endoscopic investigation in these patients
should be minimized because it rarely yields useful information or results in changes
to patient management. Further investigation in this field should focus on determining
the causes of anemia in the cohort of patients with non-IDA in order to help guide
clinicians toward investigations that are more likely to yield results.
