The Accuracy of Physical Examination to Diagnose Anemia Among Patients Five Years or Older: A Systematic Review

Abstract

Anemia remains a significant public health challenge, disproportionately impacting lower-income patients residing in areas of lesser healthcare resources. We sought to evaluate the accuracy of physical exam techniques to diagnose anemia among patients 5 years of age or older. A systematic review of 5 databases (MEDLINE via OVID, EMBASE, Scopus, Global Health and Global Health Archives, and WHO Global Index Medicus) was conducted. Studies that (1) compared non-invasive physical exam techniques with anemia diagnoses using standard laboratory measurements and (2) solely assessed or separately reported the diagnostic accuracy of physical exam techniques for patients 5 years or older were considered for inclusion. The diagnostic accuracies of individual and combinatorial physical exam techniques todiagnose anemia were documented. This systematic review was registered with PROSPERO. The systemic literature search yielded 6,457 unique studies after removal of duplicates. Fourteen studies were ultimately selected for inclusion. Eight studies solely assessed pregnant females, 4 solely assessed hospitalized patients, and 2 evaluated the general population. The diagnostic accuracy ranged widely for pallor assessments of conjunctivae (sensitivity: 19–97%, specificity: 65–100%), nailbed (sensitivity: 41–65%, specificity: 58–93%), and palms (sensitivity: 33–91%, specificity: 54–93%). Examining 9 or more sites leads to higher sensitivity (73.8–82.9%) and specificity (76.0–90.9%). No individual examination technique is superior to others for diagnosing anemia. Combinatorial approachs are associated with more acceptable accuracy measures, but improvements need to be balanced with time available for examination.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12288-022-01543-z.

Background

Anemia remains a significant public health concern internationally, affecting approximately one-fourth of the world’s population[1]. Anemia is clearly linked to worse outcomes among women, children, and patients with baseline coronary artery disease [2–4]. A large majority of the world’s anemic patients reside in developing countries, where access to standard laboratory testing is often limited [4]. Consequently, clinicians are frequently tasked with diagnosing anemia, empirically initiating iron, and monitoring improvements with treatment based upon physical examinations alone.

In the developing world, the ratio of clinicians to patients is markedly low and frequently below the threshold of “critical physician shortage” [5, 6]. General practitioners may only have a few minutes, if any, to complete routine examinations [7]. Within these few minutes, anemia is only one of the many potential medical conditions that need to be addressed, thereby preventing clinicians from performing comprehensive physical examinations. To conserve time, physical examination is often highly-focused and is weighed heavily towards findings that are clearly associated with underlying diseases. As a result, a global review and appraisal of the available data supporting commonly used physical examination techniques to diagnose anemia is warranted.

Chalco et al. previously conducted a systematic review and meta-analysis reporting the pooled associations between individual physical examination techniques and the diagnosis of anemia among children under 6 years of age [8]. The authors found substantial relationships between nailbed, palmar, and conjunctival pallor and severe anemia, which was defined as hemoglobin below 5 g/dL. However, the authors did not find strong relationships between pallor at these sites and hemoglobin measurments above 8 g/dL. The authors concluded that physical examination techniques are not excellent for diagnosing anemia among children, and clinicians and public health experts should consider initiating iron empirically among pediatric populations with widespread prevalence of iron deficiency anemia. Notably, this work was only among children, and all studies were based in Africa.

The primary aim of this current study was to systematically review the available data assessing the roles of singular and multilple physical examination techniques toward diagnosing anemia among patients 5 years of age or older. We hypothesized that evaluation of conjunctival pallor would serve as the optimal singular method to diagnose anemia, which could affect standard physical examination practices among overburdened clinicians managing lower-income patients in developing countries. We additionally hypothesized that a collection of examination findings, herein defined as “combinatorial” approaches, would provide acceptable sensitivity and specificity to serve as an adequate replacement for standard laboratory assessments of hemoglobin and hematocrit within areas of lesser healthcare resources.

Methods

Search Strategy

In concordance with a research librarian (author SH), a comprehensive search algorithm was designed to examine the efficacy of diagnosing anemia through physical examination technique(s). The following key words were used to construct this algorithm: anemia, physical examination, pallor, pallor site (conjunctivae, nailbed, tongue, and palms), and hemoglobin/ferritin. The detailed search algorithm is conveyed in Supplementary Table 1. The search algorithm was applied to the following five databases: Medline via OVID, Embase, Scopus, Global Health and Global Health Archives, and the WHO Global Index Medicus. Abstracts captured by the search strategy were extracted from all five databases and uploaded to the online Covidence platform [9]. Duplicate abstracts were removed prior to screening. This systematic review was registered with PROSPERO.

Study Selection Criteria

Two authors (NV and VV) initially screened titles and abstracts based upon relevance. Abstracts without associated full-text publications were immediately excluded. Following identification of relevant records, full text-publications were obtained for further review. The same two authors selected studies based on inclusion and exclusion criteria. Disagreements between the two authors regarding the potential inclusion of individual studies were addressed via discussion and consensus.

Selected publications were required to (1) exclusively evaluate patients five years of age or older or separately report the diagnostic accuracy of physical examination technique(s) for patients who were five years or older, (2) quantitatively assess the accuracy of at least one physical examination technique for diagnosing anemia, and (3) compare diagnostic accuracy measures against a gold-standard laboratory measurement of hemoglobin. Prospective studies and retrospective studies were eligible for inclusion.

Studies were excluded if (1) patients were all under the age of five years or if the published results did not selectively report the diagnostic accuracy of physical examination for those five years of age or older, (2) the patient population carried underlying diagnoses of hereditary hematologic conditions or active hematologic malignancies (i.e. sickle cell anemia, thalassemia, and/or acute leukemia), and/or (3) physical examination was not conducted by a physician or trained healthcare worker.

Data Extraction

For each selected full-text publication, one author (NV) extracted the following study information: geographical location of study, population demographics, physical examination methodology, and diagnostic accuracy measures. Population demographics included study size, age range, frequent baseline co-morbidities, and socioeconomic status. Each physical examination technique was characterized by anatomical site evaluated and method conducted to assess for anemia. Collected diagnostic accuracy measures included sensitivity, specificity, and likelihood ratios.

Data Synthesis

Selected studies were primarily categorized by geographical location, patient population, and anatomical site of evaluation. Thereafter, the findings from each were compared qualitatively. To identify the optimal singular examination finding associated with a diagnosis of anemia, diagnostic accuracy measures were compared within and across studies based upon individual examination techniques employed. To explore the role for combinatorial examination approaches, diagnostic accuracy measures were further compared within and across studies based upon collection of examination techniques employed. Emphasis was placed on studies that assessed statistical differences in accuracy measures between varying approaches. Bias in all included studies was assessed via the JBI’s critical appraisal tools, specifically the checklist for analytic cross sectional studies.

Results

Search Results

Translation of the search algorithm captured a total of 8620 abstracts, of which 2163 duplicates were excluded. During title and abstract screening of unique entries, 6360 records were excluded due to lack of relevance to study criteria. Ninety-seven full-text publications were selected for detailed assessment, of which 83 publications were excluded. The process to select studies is conveyed in Fig. 1. Fourteen publications, encompassing 6870 individuals, were ultimately included in this systematic review [10–23].

Fig. 1.

PRISMA diagram detailing screening process for selecting articles evaluating diagnostic accuracy of physical examination techniques in patients above the age of five

Key Features of Selected Studies

Studies were primarily conducted in South Asia (n = 5) and Africa (n = 4) with remaining studies completed in East Asia (n = 2), Latin America (n = 1), Europe (n = 1), and the Pacific Islands (n = 1). Twelve studies were conducted in developing nations, while two studies were conducted in developed nations. Patient populations that were evaluated included pregnant females (n = 8), hospitalized patients (n = 4), and the general population (n = 2). Study participants were clustered in the following age groups: early adulthood (n = 6) and late adulthood (n = 8). All studies evaluated patients for anemia by assessing clinical pallor. Descriptions regarding common pallor assessment techniques can be found in Table 1. Common anatomical sites assessed for pallor included: conjunctivae (n = 14) and nailbed (n = 6). Less commonly, palm (n = 5), tongue (n = 3), oral mucosa (n = 3), and face (n = 3) were examined. Additionally, all included studies satisfied the majority of requirements listed in the checklist. Each selected study’s objective(s), methodology, and findings are described in Table 2.

Table 1.

Overview of how commonly utilized physical exam techniques are conducted and how pallor is assessed with these techniques

Anatomical site being evaluated	Exam technique	Traditional grading scale*
Palm [17]	Pallor associated with anemia may be identified via examination of the palmar creases. Intensity of palmar crease pallor can then be ranked from absent (lack of pallor) to severe (extreme paleness)	Palmar pallor is traditionally graded as absent or present
Conjunctivae [24]	Paleness of the conjunctivae can be identified via examination of the anterior rim of the eye. Pale conjunctivae lack the distinct red coloration of normal conjunctivae. Anterior rim pallor can also lie between the two extremes, thereby permitting a “borderline” or “mediocre” classification	Pallor of the conjunctivae is traditionally graded as absent or present
Nailbed [25]	Nailbed pallor can be noted via classifying the degree of paleness of the nailbed	Nailbed pallor is traditionally graded as absent or present
Tongue [25]	Dorsum of the tongue may be examined for pallor. In addition, absence of papillae from the tongue may be indicative of anemia	Pallor of the tongue is also graded as absent or present
Oral Mucosa [26]	Oral membranes may also be examined for pallor. As with the similar pallor assessments, pallor can be graded from absent (lack of paleness) to severe (extreme paleness of the oral mucosa)	Pallor of the oral mucosa is graded as absent or present
Face [27]	Facial pallor is most often indicated or seen when the skin takes on a white hue. Severity of paleness is often associated with severity of anemia diagnoses	Facial pallor is commonly graded as present or absent

*Current research in the field has advocated for the development and utilization of a more nuanced pallor grading scale. These include scales classifying pallor as absent, mild, moderate, or severe¹, and/or utilizing a color tint scale (Chowdhury 2002.) to aid in classification of pallor

Table 2.

Summary of all studies included in the systematic review including information on sample, methods, results, and key findings

Study	Country; patient population	Methods	Quantitative outputted results	Main findings
Van Den Broek et al. [10] (1999)	Malawi; 729 Pregnant Females Mean age: not provided in the text	Conjunctival Examination performed by midwives. Pallor of eye was classified as pink (absence of anemia), pale (presence of anemia), or very pale (severe anemia)	Conjunctival Examination by Hemoglobin Concentration (Sensitivity, Specificity:	The diagnostic accuracy of anemia was found to increase as severity of anemia increases. Non-invasive examination of conjunctivae is valuable in low-income settings, however, has low diagnostic accuracy
			less than equal to 11 g/dL: 85.4%, 80.1%
			less than equal to 10: 94%, 79.1%
			less than equal to 8 g/dl: 96.6%, 79.6%
			less than equal to 6 g/dl: 80%, 99.5%
Butt et al. [11] (2010)	Pakistan; 305 Hospitalized Patients Mean Age: 48.3 years (SD: 19.3 years)	Pallor of conjunctiva, palms, and nailbed was assessed. Conjunctiva was designated “pale” if anterior rim color same as posterior pale rim. Palms were considered pale if hue of palmar crease is similar to skin. Nailbed pallor was assessed via noting color of nailbed after pressure release	Mild Anemia (Sensitivity, Specificity)	Nailbed pallor has moderate accuracy in detecting mild anemia. However, absence of nailbed pallor does not allow for ruling out of anemia. Absence of conjunctival and palmar pallor does allow for ruling out of anemia diagnosis. Pallor sensitivity at all sites increases as hemoglobin levels decrease. Overall, there was no difference in diagnostic accuracy of one site assessment versus multiple site pallor assessment
			Conjunctiva: 45.9%, 85.30%
			Nailbed: 41.3%, 92.70%
			Palm: 50.5%, 76%
			Any Site: 63.8%, 70.60%
			Severe Anemia (Sensitivity, Specificity)
			Conjunctiva: 91.3, 69.90%
			Nailbed: 65.2%, 73.80%
			Palm: 91.3%, 63%
			Any Site: 95.7%, 52.10%
Bergsjø et al. [12] (2008)	Tanzania; 904 Pregnant Females (369 low altitude, 535 high altitude) Mean Age: not provided in the text	Midwives assessed pallor in females at both high and low altitudes. Non-invasive measurements included general pallor, conjunctival pallor, palmar pallor, pallor of mucous membranes in mouth, and pallor of tongue. A survey assessing anemia symptoms was also administered to all females	Group 1: Lower Altitude (Sensitivity, Specificity):	Assessment of pallor at individual sites yielded statistically similar measures of diagnostic accuracy. Highest accuracy of non-invasive pallor assessment can be attained by looking at pallor at all four sites. Additionally, addition of survey information did not affect diagnostic accuracy of pallor assessment by midwives
			Pallor: 83%, 53%
			Conjunctivae: 83%, 65%
			Palms of hands: 83%, 54%
			Mouth/Tongue: 88%, 54%
			Total Impression: 100%, 56%
			Group 2: Higher Altitude: (Sensitivity, Specificity)
			Pallor: 33%, 91%
			Conjunctivae: 44%, 92%
			Palms of hands: 33%, 93%
			Mouth/Tongue: 33%, 93%
			Total Impression: 44%, 90%
Bala et al. [13] (2012)	India; 129 Pregnant Females Mean Age: not provided in the text	Medical officers examined pallor of skin, mucous membranes of mouth and tongue, and conjunctivae. Sensitivity and specificity of non-invasive exams were measured for both mild and severe anemia patients	Overall Sensitivity: 91.11%	Study found that assessment by non-invasive clinical signs were more accurate than the WHO’s hemoglobin color scale. Authors recommend non-invasive screening for anemia regardless of laboratory tests availability in order to help accurately diagnose anemia
			Overall Specificity: 12.82%
Chathurani et al. [14] (2012)	Sri Lanka; 900 Pregnant Females Mean Age: 27 years (SD: 5.5 years)	Conjunctival examination of women was conducted along with invasive hemoglobin measures in order to determine sensitivity and specificity	Sensitivity of conjunctival examination: 18.8%	Conjunctival screening should not be used as a singular measure for diagnosis of anemia due to its low sensitivity. There is a need for innovation in the field to develop better non-invasive measures to detect for anemia, especially in developing countries facing a disproportionate burden of anemia
			Specificity of conjunctival examination: 69.3%
Chowdhury et al. [15] (2002)	Bangladesh; 452 Pregnant Females Mean Age: 23.2 years (SD: not provided)	Nine anatomical sites were screened for pallor. These included conjunctivae, nailbeds, nailbed blanching, palm, palmar creases, face, lips, tongue, and oral mucosa. In addition, the color tint scale provided a range of 13 shades of pink ranging from white to red. Medical history was also taken for each patient, with special emphasis given to clinical signs of anemia (Ex: dizziness, fatigue, palpitation, etc.)	Sensitivity, Specificity for 3 Experimental Conditions:	Assessment of pallor at all nine clinical pallor sites had a higher sensitivity than any one site individually. Sensitivity was further increased with the addition of the color tint scale and the medical history of the patient. The highest sensitivity was obtained via pallor examination, color tint scale, medical history. Implementation of patient interviews and color tint scales can help increase the diagnostic accuracy of non-invasive diagnostic measures
			Clinical Pallor: 73.8%, 76%
			Clinical Pallor + Medical History: 78.3%, 84.7%
			Clinical Pallor + Medical History + Color Tint Scale: 82.9%, 90.9%
Collings et al. [16] (2016)	New Zealand; General Population (198 subjects recruited) Mean Age: 62 years (SD: 14 years)	Conjunctivae were evaluated for pallor via the use of an iPhone or digital camera via software developed by authors. Diagnostic accuracy of the algorithm was compared to clinician examination of conjunctivae images	Respective clinicians 1–3	Diagnostic accuracy of developed algorithm was greater than the diagnostic accuracy of conjunctivae examination by clinicians. Findings suggest that smartphone cameras may provide a cost-effective, non-invasive exam which can help diagnose anemia in low-income settings
			Sensitivitya: 0.58–0.88
			Specificitya: 0.24–0.71
Kalantri et al. [17] (2010)	India; 390 Hospitalized Patients Mean Age: 40.1 years (SD: 17.08 years)	Trained observers evaluated pallor at the conjunctivae, tongue, palm, and the nailbed. Pallor was categorized as either none, mild, moderate, or severe	Hemoglobin Level- 7 g/dl (LRb)	Study found tongue had the highest sensitivity, in contrast to widely utilized conjunctivae, nailbed, and palmar pallor. Additionally, pallor findings are more sensitive in patients with severe anemia. However, the absence or presence of pallor cannot diagnose or exclude diagnosis of anemia
			Conjunctiva (severe, moderate, mild, no pallor): 3.51, 2.32, 1.34, -
			Tongue (severe, moderate, mild, no pallor): 9.87, 4.98, 1.27, -
			Nailbed (severe, moderate, mild, no pallor): 3.51, 2.59, 1.35, 0.39,
			Palm (severe, moderate, mild, no pallor): 4.18, 2.20, 1.55, 0.25
			Hemoglobin Level- 9 g/dl (LR)
			Conjunctiva (severe, moderate, mild, no pallor): 4.42, 1.89, 1.23,0.38
			Tongue (severe, moderate, mild, no pallor): 7.60, 3.70, 1.21, 0.29
			Nailbed (severe, moderate, mild, no pallor): 4.42, 2.44, 1.29, 0.53
			Palm (severe, moderate, mild, no pallor): 5.94, 1.66, 1.43, 0.48
			Hemoglobin Level- 12 g/dl (LR)
			Conjunctiva (severe, moderate, mild, no pallor): 4.25, 1.86, 1.37, 0.40
			Tongue (severe, moderate, mild, no pallor): -, 3.19, 1.18, 0.53,
			Nailbed (severe, moderate, mild, no pallor): 4.25, 2.86, 1.48, 0.52
			Palm (severe, moderate, mild, no pallor): 3.7,1.93, 1.47, 0.60
Nardone et al. [18] (1999)	Thailand; 103 Hospitalized Patients Mean Age: 60 years (Range: 27–80 years)	Direct observation of conjunctivae, face, nailbeds, palms, and palmar creases were made by three internists. Each site was assigned “no pallor” or “pallor”	Sensitivity, Specificity By Hematocrit	Assessment of pallor of the conjunctivae, palms, and face is more powerful than assessment of only a single site. However, even analysis of only conjunctivae, palms, or the face is more accurate than nailbeds
			Cutoff Points: Hematocrit Cutoff: 0.3
			Conjunctivae: 58%, 74%
			Face: 48%, 81%
			Nailbeds: 52%, 58%
			Palms: 67%, 65%
			Palmar Creases: 12%, 98%
			CFP1 (pallor at one site [conjunctivae, face, palms]): 80%, 57%
			CFP2 (pallor at any two sites]: 60%, 76%
			CFP3 [ pallor at all three sites]: 37%, 87%
			Hematocrit Cutoff: 0.35
			Conjunctivae: 56%, 82%
			Face: 46%, 88%
			Nailbeds: 59%, 66%
			Palms: 64%, 74%
			Palmar Creases: 8%, 99%
			CFP1 (pallor at one site [conjunctivae, face, palms]): 77%, 66%
			CFP2 (pallor at any two sites]: 56%, 84%
			CFP3 [ pallor at all three sites]: 35%, 94%
			Hematocrit Cutoff: 0.4
			Conjunctivae: 45%, 85%
			Face: 34%, 90%
			Nailbeds: 49%, 65%
			Palms: 53%, 77%
			Palmar Creases: 7%, 100%
			CFP1 (pallor at one site [conjunctivae, face, palms]): 65%, 71%
			CFP2 (pallor at any two sites]: 43%, 85%
			CFP3 [ pallor at all three sites]: 26%, 95%
Sanchez-Carrillo et al. [19] (1989)	Mexico: General Population (219 subjects recruited) (no age range provided in the text)	Conjunctiva hue was assessed by trained field workers and patients were either found to be “anemic” or “not anemic”	Pink Conjunctival Hue (Sensitivity, Specificity): 50%, 81.8%	Subjects with pink conjunctival hue had a higher chance of being misclassified versus subjects with a red hue of the conjunctivae
			Red Conjunctival Hue (Sensitivity, Specificity): 36.8%, 76.9%
Shulman et al. [23] (1977)	Kenya;1787 Pregnant females Mean Age: not provided in text	Midwives assessed the conjunctivae, nailbeds, and tongue for pallor. In addition, pulse and respiratory rate were also measured, and a brief questionnaire was administered to understand the patients’ symptoms and clinical history	Model #1: designed to solely examine symptoms	Sensitivity of pallor examinations can be significantly increased via spending time obtaining a patient’s history via a questionnaire
			greater than or equal to 1 symptom present: 84%, 40%
			greater than or equal to 2 symptom present: 63%, 82%
			greater than or equal to 3 symptom present: 31%, 98%
			Model #2: designed to examine symptoms (from model 1) and physical examination findings
			greater than or equal to 1 symptom/exam technique present: 94%, 62%
			greater than or equal to 2 symptom/sign present: 48%, 99%
			Assessments Predicting Severe Anemia
			MW1c pallor test + ved: 62%, 87%
			MW2c pallor test + ve: 69%, 77%
			MW1 pallor test + ve or PRe > 99 or RRf > 99: 79%, 56%
			MW2 pallor test + ve or PR > 99 or RR > 99: 84%, 46%
			MW1 Pallor test + ve or PR > 99: 77%, 60%
			MW2 Pallor test + ve or PR > 99: 81%, 51%
			MW1 pallor test + ve or RR > 22: 66%, 77%
			MW2 pallor test + ve or RR > 22: 74%, 68%
Gies et al. [21] (2003)	Ethiopia; 403 Pregnant Females Mean Age: 23.4 years (SD: 5.18 years)	Nurse-Midwives performed conjunctival pallor examinations and attained patient history. Conjunctival pallor was cored as pale, slightly pale, and not pale	Conjunctival Pallor Sensitivity, Specificity by Hemoglobin Level: < 9 g/dl: 85.7%, 76.7% < 10 g/dl: 55.6%, 77.1% < 11 g/dl: 44.3%, 79.2%	Pallor of the conjunctivae is most noticeable in patients with severe anemia, leading to increasing sensitivity as hemoglobin levels decreased in women. The predictive value of conjunctival pallor examination is low for women with mild to moderate anemia
Wallace et al. [22] (2000)	Ireland; 101 Hospital Patients Mean Age: not provided in text	Conjunctival pallor was assessed by consultant surgeons, consultant ophthalmologist, registrar in general surgery, and registrar in ophthalmology. Assessments were used to determine the sensitivity and specificity of anemia diagnosis	Sensitivity, Specificity by Observer: Consultant Surgeon: 61%, 62% Consultant Ophthalmologist: 65%, 74% Registrar in General Surgery: 52%, 81% Registrar in Ophthalmology: 83%, 83%	Conjunctival pallor accuracy was greatest in groups with low concentration of hemoglobin. Authors suggest in countries where invasive diagnostic measures are available, those should be preferred over non-invasive clinical exams
			Overall Sensitivity: 52%
			Overall Specificity: 83%
Tan et al. [20] (2009)	Philippines; 250 Women Mean Age: 27.2 years (SD: 6.76)	Anemia Scoring Index was used to evaluate pallor in the nailbeds and conjunctivae. Additionally, heart rate, peripheral pulses, and BP were all measured for the patients. Scores of 0 were classified as no anemia, 1–3 were diagnosed with mild anemia, 4–6 were classified with moderate anemia, 7–10 were classified with severe anemia	False positive rate: 47.6%	The anemia scoring index was found to have a low sensitivity, and thus was found to be a poor predictor of anemia
			True Positive Rate: 52.4%

^aRange provided from three clinician observers

^b LR Likelihood ratio

^cMW1/2: Midwife 1 or 2

^d + Ve: Positive

^ePR Pulse rate

^f RR Respiratory rate

Individual Physical Examination Findings associated with Anemia

Ten studies reported at least one diagnostic accuracy measure correlating an individual physical examination technique with anemia[10–12, 14, 16–19, 21, 22]. Among the sites most commonly assessed for pallor, ranges in sensititivy were as follows: conjunctivae (19–97%), palms (33–91%), nailbeds (41–65%), face (34–48%), and tongue (33–88%). Ranges in specificity were as follows: face (81%-90%), conjunctivae (65–100%), nailbed (58–93%), palms (54–93%), and tongue (54–93%).

Only four studies specifically compared diagnostic accuracy measurements across pallor assessments of various anatomical sites[11, 12, 17, 18]. All four studies statistically compared the resulting diagnostic accuracy measures between varying individual sites, however only two reported signficiance of findings. Kalantri et al., which was the only study to compute likelihood ratios, observed a trend towards a greater association between tongue pallor and anemia compared to nailbed, palmar, and conjunctival pallor[17]. Nardone et al. compared the diagnostic accuracy of face, conjunctival, nailbed, and palm pallor[18]. The authors found that nailbed pallor was less sensititive than both individual and combinatorial assessments of pallor of the conjunctivae, face, and palm (p = 0.01).

Combinatorial Examination Approaches associated with Anemia

Seven studies assessed the accuracy of a composite of physical examination techniques [11–13, 15, 18, 20, 23]. In regard to sensitivity, Chowdhury et al. reviewed pallor of nine anatomical sites (conjunctivae, nailbed pallor, nailbed blanching, palm, palmar creases, face, lips, tongue, and oral mucosa) and combined this patient history and a color tint scale. The authors observed relatively high sensitivity (82.9%) When these nine anatomical sites were evaluated with only patient history and without a color tint scale, a sensitivity of 78.3% was observed. The authors further evaluated pallor at the same nine sites without including either additional patient history or color tint scale findings and observed a computed sensitivity of 73.8%. The remaining six studies studies by Bala et al., Butt et al., Bergsjø et al., Nardone et al., Shulman et al., and Tan et al. combined several physical exam technique, yet combined fewer than nine sites. Bala et al., Bergsjø et al., and Nardone et al. evaluated greater anatomical sites and yielded a higher range of sensitivity measures (65–100%) than Tan et al., Shulman et al., and Butt et al. (48–69%). Therefore, evaluation of greater anatomical sites may lead to an increase in specificity.

In regard to specificity, Chowdhury et al.’s evaluation of nine anatomical sites for pallor, observed high specificities with their three varying approaches (90.9, 84.7, and 76.0%, respectively). The six studies, Bala et al., Butt et al., Bergsjø et al., Nardone et al., Shulman et al., and Tan et al., evaluating the diagnostic accuracy of examining common sites for pallor observed a lower range of specificity (12.8–77%). None of these studies directly compared diagnostic accuracy measures between different approaches sharing a varying number of examination sites.

Discussion

To our knowledge, this is the first systematic review evaluating the accuracy of physical examination techniques toward diagnosing anemia for patients five years of age or older. Assessments of at-risk populations, namely pregnant females and hospitalized patients, were most frequently reported. Pallor examinations of the conjunctivae, nailbeds, and palms were the common individual examination techniques investigated. On one hand, this systematic review could not confirm that any individual physical exam technique was clearly superior to all others, This result is in general accord with the meta-analysis conducted by Chalco et al., in which the authors did not identify an optimal singular examination technique associated with anemia among pediatric patients under five years of age [8].On the other hand, this current systematic review did find that combinatorial approaches to physical examination are likely associated with a more acceptable diagnostic yield.

Of the individual sites examined for pallor, evaluations of the conjunctivae were by far most frequently investigated, thereby suggesting that pallor assessment of conjunctivae may be the least technically difficult, and perhaps least time consuming. Additionally, rates of sensitivity and specificity were similar to other commonly assessed sites, if not slightly better. Of the nine studies assessing conjunctival pallor individually, four studies reported a sensitivity greater than 70% and specifity greater than 80% [10, 11, 16, 21]. Of the seven studies assessing conjunctival pallor combinatorally, only one study evalued sensitivity and specificity of the conjunctivae alone in addition to in combination with other sites[11]. Furthermore, the study that observed the lowest sensitivity (19.9%) for conjunctival pallor examined a population of pregnant females who had high rates of mild anemia (Hg 10–10.9 g/dL) [14].

Of the three studies that directly compared the diagnostic accuracies of conjunctival, nailbed, and palmar pallor, only two reported differences between techniques. Nardone et al. found pallor assessment of the conjunctivae, face, and palms together to significantly outperform nailbeds (p = 0.01) [18]. Contrarily, Kalantri et al. found tongue pallor to be significantly more accurate than other pallor sites via analysis of ROC values [17]. At this time, an optimal simgular physical examination technique to diagnose anemia cannot be elucidated due to conflicting diagnostic accuracy measurements. Given the familiarity with examining conjunctival pallor across the selected studies, frequent high rates of reported sensitivity and specificity for conjunctival pallor, and mixed findings amongst the studies directly comparing the diagnostic accuracy of conjunctival pallor with other exam techniques, evaluation for conjunctival pallor is likely as good as any other individual technique to diagnose anemia, if not superior. Nevertheless, further well-designed prospective studies comparing conjunctival pallor to other singular examination techniques are warranted.

Seven studies assessed several anatomical sites collectively. With this combinatorial approach, ranges in sensitivity and specificity were less variable (65–100% and 12.8–77%, respectively) than the values observed among studies examining individual physical examination techniques[10–12, 14, 16–19, 21, 22]. Furthermore, in the work by Chowdhury et al., nine anatomical sites were assessed together for anemia, leading to an acceptable sensitivity and specificicity (sensitivity: 73.8%, specificity: 76%), which improved further with the addition of patient history and a color tint scale [15]. Though the accuracy with these comprehensive methods were excitedly high, these results do not clearly emphasize the need for such a wide-ranging combinatorial approach. Overall, examining more than one site likely improves the diagnosis of anemia., and addition of extra sites may increase sensitivity levels. However, the benefit in improved diagnostic accuracy is likely outweighed by the time required to complete several techniques, particularly for patients who are examined at nine anatomical sites. As an aside, within all of the studies assessing cominbitorial approaches, anemia was diagnosed if any individual examination technique was “positive”.

Thirteen of the 14 selected studies investigated at-risk populations, which included pregnant females (8), hospitalized patients (3), and elderly patients (2). Among pregnant women, all 8 studies assessed conjunctival pallor. Five of the selected studies evaluating pregnant females included a combinatorial approach. Again, wide ranges in sensitivity (18.8–96.6%) and specificity (65–99.5%) were observed. Sensitivity and specificity may have been negatively affected by the prevalence of women with mild anemia in the included studies. For the studies that did provide ranges of hemoglobin levels for included pregnant female subjects, sensitivity and specificity for conjunctival pallor increased as hemoglobin levels decreased [10, 21]. In addition to the examination of conjunctival pallor, palmor pallor and/or nailbed pallor were both examined by four studies [12, 15, 20, 23]. All displayed similar sensitivity and specificity ranges to conjunctival pallor. These findings suggest that no single individual examination technique provides a higher diagnostic yield than the others for pregnant women. Given the high risk for anemia experienced by pregnant females, the wide ranges in diagnostic accuracy measures with individual examination techniques, and the feasilibity of administering oral iron to patients in developing regions, further studies exploring the diagnostic accuracy of a combinatorial approach to pregnant women are warranted.

In regard to hospitalized patients, pallor assessments of the conjunctivae, nails, and palms were commonly conducted. The sites with the highest diagnostic accuracy were the palms (sensitivity:50.5–91.3%, specificity: 63–77%) and conjunctivae (sensitivity: 45–91.3%, specificity: 62–85.3%). Palmar pallor appears to carry a similar diagnostic acuracy to the conjunctivae among hospitalized patients, and may act as an alternative examination site to the conjunctivae or be evaluated along with the conjunctivae. It is important to note that hospitalized patients often share a higher incidence of severe anemia than other patient populations. Diagnostic accuracy of palmar or conjunctival pallor, however, may not differ with increasing anemia severity.

Limitations

A number of limitations of our comprehensive review of the literature must be noted. First, quantitative comparison of physical exam findings was not a viable option due to limited overlap of techniques assessed, with the clear exception of conjunctival pallor examinations. Second, the methodology of selected studies were not uniform. Despite all selected studies comparing physical examination techniques to a laboratory gold standard result, eleven studies specifically drew comparisons based upon varying categories of anemia severity. Third, none of the selected studies monitored changes in physical exaimation wth intervention, namely oral iron replacement therapy.

Conclusions

Our findings indicate that no singular individual examination technique is more clearly associated with anemia than others. Given familiarity with evaluating conjunctival pallor and its similar diagnostic accuracy to other commonly investigated techniques, sole evaluation of conjunctival pallor by clinicians serving a large volume of patients in areas of lesser healthcare resources is reasonable. Combinatorial approaches to physical examination are most likely associated with a higher diagnostic yield, but this benefit must be weighed against the time available to complete comprehensive exams. Future studies should investigate the baseline diagnostic accuracy of feasible combinatorial approaches involving 3–5 techniques for populations with widespread anemia prevelance and subsequent changes in examination findings following improvements in hemoglobin levels stemming from relevant interventions.

Supplementary Information

Below is the link to the electronic supplementary material.

Author’s Contribution

The authors confirm contribution to the paper as follows: study conception and design: NV, VV, and SH; data collection: NV and VV; analysis and interpretation of results: NV and VV; draft manuscript preparation: NV, VV, SH, TS, CM, RR, HC, AS. All authors reviewed the results and approved the final version of the manuscript.

Funding

The authors declare no financial interests.

Code Availability

All necessary data is provided in tables, figures, and supplementary materials.

Declarations

Conflict of interest

The authors declare no competing interests.

Ethical Approval

Not applicable for study.