The effect of anaemia and abnormalities of erythrocyte indices ...

Page 1

SYSTEMATIC REVIEW

The effect of anaemia and abnormalities of erythrocyte indiceson HbA1c analysis: a systematic review

Emma English1& Iskandar Idris1 & Georgina Smith1

& Ketan Dhatariya2 &

Eric S. Kilpatrick3& W. Garry John4

Received: 7 January 2015 /Accepted: 2 April 2015 /Published online: 21 May 2015# Springer-Verlag Berlin Heidelberg 2015

AbstractAims/hypothesis The use of HbA1c for the diagnosis of diabe-tes is now widely advocated despite caveats to its use.Anaemia is cited as a major confounder to this use; however,the effect of erythrocyte indices and to what degree anaemiainfluences HbA1c levels is not known.Methods A systematic electronic database search ofMEDLINE, EMBASE, the Cumulative Index to Nursing &Allied Health Literature (CINAHL) and the Cochrane Librarywas conducted for relevant articles published between January1990 and May 2014. Included studies had at least one mea-surement of HbA1c and glucose, and a least one index ofhaematinic deficiency, involving non-pregnant adults, notknown to have diabetes.Results A total of 12 articles from 544 were included. Themajority of studies focused on iron deficiency anaemia(IDA) and, in general, demonstrated that the presence of irondeficiency with or without anaemia led to an increase inHbA1c values compared with controls, with no concomitant

rise in glucose indices. Data on the effects of other indices oferythrocyte abnormalities on HbA1c are limited but show apossible decrease in HbA1c values with non-iron deficiencyforms of anaemia.Conclusions/interpretation HbA1c is likely to be affected byiron deficiency and IDA with a spurious increase in HbA1c

values; conversely, non-IDA may lead to a decreased HbA1c

value. This may lead to confusion when diagnosing diabetesusing HbA1c. This review clearly identifies the need for moreevidence, especially in identifying the types and degrees ofanaemia likely to have significant impact on the reliability ofHbA1c.

Keywords Anaemia . Diabetes . HbA1c. Iron deficiency .

Systematic review

AbbreviationsCINAHL Cumulative Index to Nursing & Allied Health

LiteratureFPG Fasting plasma glucoseIDA Iron deficiency anaemiaIFCC International Federation for Clinical Chemistry

and Laboratory MedicineMCH Mean corpuscular haemoglobinMCHC Mean corpuscular haemoglobin concentrationMCV Mean corpuscular volumeNHANES National Health and Nutrition Examination

SurveyPCV Packed cell volumeRDW Red cell distribution widthSIGN Scottish Intercollegiate Guidelines NetworkTSAT Transferrin saturation

Electronic supplementary material The online version of this article(doi:10.1007/s00125-015-3599-3) contains peer-reviewed but uneditedsupplementary material, which is available to authorised users.

* Emma [email protected]

1 School of Medicine, University of Nottingham,Derby DE22 3DT, UK

2 Elsie Bertram Diabetes Centre, Norfolk and Norwich UniversityHospitals NHS Foundation Trust, Norwich, UK

3 Department of Clinical Biochemistry, Hull Royal Infirmary,Hull, UK

4 Department of Clinical Biochemistry, Norfolk and NorwichUniversity Hospitals NHS Foundation Trust, Norwich, UK

Diabetologia (2015) 58:1409–1421DOI 10.1007/s00125-015-3599-3

Page 2

Introduction

The traditional role of HbA1c analysis has been for assessingglycaemic control in patients with diabetes. The results of sem-inal studies [1, 2] demonstrated that early, intensive glycaemiccontrol could significantly reduce the risk of a range ofdiabetes-related complications, and permitted the establish-ment of precise HbA1c target values for treatment goals [3].

More recently, there has been a move towards the use ofHbA1c for the diagnosis of type 2 diabetes. The WHO and theADA have both advocated the use of HbA1c for diagnosingtype 2 diabetes, at a value of 6.5% (48 mmol/mol) [4, 5].

Further to the recommendations of the WHO, the UK is-sued an expert position statement on the application of theserecommendations in clinical practice in the UK [6]. One keyfactor thought to be a confounder in the use of HbA1c is analtered erythrocyte lifespan, in particular due to anaemia. TheWHO defines anaemia in adults as 120 g/l Hb in non-pregnantwomen and 130 g/l in men [7]. It is widely purported thathaemolytic anaemia can lead to decreased HbA1c values dueto reduced erythrocyte lifespan, and iron deficiency anaemia(IDA) may result in increased HbA1c values due to an elon-gation of the erythrocyte lifespan. However, it is not known towhat degree alterations in erythrocyte indices affect HbA1c

values especially around the diagnostic cut point of 6.5%(48 mmol/mol) or the degree of abnormality severity requiredto result in a significant change. With approximately 29% ofnon-pregnant women worldwide having anaemia in 2011 [8],this translates to a significant number of people where the useof HbA1c for diagnosis of diabetes may be precluded.

Since the publication of the recommendations, there hasbeen a demand for clarity on this topic. The key questionsasked are: ‘At what level of anaemia should I not use HbA1c

for diagnosis?’ and ‘Should I routinely screen patients foranaemia when using HbA1c for diagnosis and if so, what testshould I use?’

This systematic review aims to address the above ques-tions by assessing the available evidence on the impact ofabnormalities of erythrocyte indices and anaemia, onHbA1c levels around the diagnostic cut off point of 6.5%(48 mmol/mol).

Methods

This systematic review is registered on Prospero (registrationno. CRD42013005251). Approval of an ethics committee wasnot required.

Study identification

An electronic database search for relevant articles publishedbetween January 1990 and May 2014 was conducted using

the following databases: MEDLINE, EMBASE, CumulativeIndex to Nursing & Allied Health Literature (CINAHL) andThe Cochrane Library.

The search was carried out using a combination of key-words and MESH terms or EMtree terms depending on theparticular database (see Electronic Supplementary Material[ESM] Methods for details).

In general, the search strategy consisted of: (1) HbA1c/glycated haemoglobin; (2) iron deficiency; (3) folate, B12deficiency; and (4) anaemia.

Study selection

Results from all searches were combined and duplicateswere removed. Two investigators (EE and GS) evaluatedthe title and abstract of each reference identified by thesearch. Inclusion criteria are fully detailed in Table 1;for inclusion, all studies required at least one HbA1c

value with fasting plasma glucose (FPG) or OGTTvalues and erythrocyte and/or iron indices measured.This was to ensure that any changes in HbA1c valuewere due to changes in the erythrocytes and not dueto changes in glycaemia. Any that were clearly irrele-vant were removed and the full texts of remaining arti-cles were retrieved. Upon further scrutiny of the fulltexts, those that did not meet the inclusion criteria weresubsequently excluded. The reference lists of includedarticles were reviewed in order to identify any furtherarticles of relevance to the subject area, and that metthe inclusion criteria.

Data extraction

Two investigators extracted the results from each articleusing a data extraction form based on the pre-defined studyinclusion criteria (Table 1). Any disagreement regardingstudy selection was resolved by means of consensus, involv-ing a third investigator, according to a-priori agreed criteria.The main categories extracted included: author, year, title,study overview, patient characteristics, definition of diabetes,definition of anaemia and key conclusions, as detailed inTable 2 and ESM Table 2.

Quality assessment of included articles

The methodological quality of each study was indepen-dently rated by two investigators (EE and GS—who werenot blinded to author, journal or institution) according tothe Scottish Intercollegiate Guidelines Network (SIGN)criteria [9]. Quality assessment was not used as an exclu-sion criterion, though articles conducted to high standardsin minimising the risk of confounding were noted (ESMTable 1).

1410 Diabetologia (2015) 58:1409–1421

Page 3

Data analysis

Variation within study designs, and heterogeneity of results,meant that the data was unsuitable for meta-analysis. Instead, the existing analyses described in the includedarticles were extracted and reported in a systematic format,as a narrative synthesis of the main outcomes and results ofeach study. In line with current recommendations [10, 11], theHbA1c units in the main text have been converted to SI unitsusing the master equation.

Results

The electronic database searches identified 451 potentiallyrelevant, unique articles. Titles/abstracts of the 451 articleswere reviewed and 408 were excluded based on evaluationof the title and/or abstract. The remaining 43 full text articleswere reviewed, 31 were excluded upon further reading. Theresidual 12 were approved and quality assessed using theSIGN criteria. The main reasons for exclusion were irrelevantor incomplete data when compared with the inclusion criteria.No additional articles were identified through review of thereference lists (see Fig. 1 for flow chart).

Narrative analysis of included studies

Cross-sectional studies in patients with and withoutdiabetes Kim et al [12] investigated the influence of irondeficiency on HbA1c distribution among adults who werenot known to have diabetes, over 7 years of the NationalHealth and Nutrition Examination Survey (NHANES). Ofthe 6,666 female participants, 13.7% had iron deficiency and

30% of these had IDA. A much lower proportion of males(1.6%) had iron deficiency and 33% of these had IDA. WhenHbA1c values in women were adjusted for age and ethnicity,the difference between iron deficient and non-iron deficientbecame significant (5.33% vs 5.27% [35 mmol/mol vs34 mmol/mol], p=0.002). The authors found that iron defi-ciency in women of reproductive age was associated with ashift in HbA1c from <5.5% to 5.5–6.0% (<37 mmol/mol to37–42 mmol/mol), although no association was found athigher levels, possibly owing to the lower number of partici-pants in those groups. After adjusting for age and ethnicity, theauthors concluded that HbA1c was higher in iron-deficientindividuals and was likely to result in an upward shift ofHbA1c distributions.

Ford et al [13] evaluated 1999–2002 NHANES data sets butincluded analysis of both IDA and non-IDA in participants bothwith and without diabetes. They found that non-diabetic partic-ipants with low Hb but normal iron levels had lower HbA1c

values (5.16% [33 mmol/mol]) than those with normal Hb andnormal iron levels (5.31% [35 mmol/mol], p<0.001). In con-trast, those with normal Hb but low iron values had borderlinehigher HbA1c values than normal participants (5.39% [35mmol/mol], p=0.061). In addition, in all participants without diabetes,the HbA1c values increased with increasing Hb levels rangingfrom a mean HbA1c of 5.18% (33 mmol/mol) at Hb <100 g/lthrough to 5.50% (37mmol/mol) with Hb >170 g/l. The authorsadvocate caution when interpreting HbA1c values near diagnos-tic cut points when Hb levels are high or low. The study waslimited by low numbers of iron deficiency and/or anaemia casesand few severe cases but the authors suggest that the likelyimpact of haemoglobin concentration on HbA1c values was anapproximate change of HbA1c 0.2% (2.2 mmol/mol), betweenthe extremes of Hb levels.

Table 1 Inclusion criteria for the systematic review

Inclusion criteria Description

Population All non-pregnant adults (≥18 years), without known diabetes (HbA1c should not be used for diagnosis of diabetes in childrenor pregnancy). Participants should have at least one measurement of HbA1c and measure of glucose, as well as at least oneindex of anaemia or haematinic deficiency. A range of definitions of anaemia were used in the studies detailed in thisreview, with some basing the diagnosis on Hb levels and others refining this further based on indices such as MCV, MCH,ferritin and TSAT levels.

Intervention The use of HbA1c for the diagnosis of diabetes or comparison of HbA1c in participants with and without anaemia or otherabnormalities of erythrocyte indices.

Comparisons HbA1c values ideally will be compared with FPG or OGTTs and values compared between participants with and without anaemiaor erythrocyte indices abnormalities. Whilst some studies may include glucose values, others only report that all participantshad values below a set point.

Outcomes Changes in HbA1c values, compared with glucose indices, with decreased haemoglobin concentration. Changes in HbA1c,compared with glucose indices, with other erythrocyte indices.

Study design Cohort studies evaluating the association between HbA1c levels and anaemia or other erythrocyte indices with or without anaemiawere included. Case report, case–control, case series studies, letters and commentaries were excluded.

Search limits Adults (aged ≥18 years); English language; humans; publication year 1990–current. HbA1c was not widely used in routine clinicalpractice and methods showed very high inter and intra-laboratory variation prior to 1990. Identification of grey literatureincluding unpublished data, conference reports and presentations was not attempted.

Diabetologia (2015) 58:1409–1421 1411

Page 4

Tab

le2

Dataextractio

nform

andkeydatafrom

included

studies

Author,

year

andtitle

n(m

ale)

Study

Diabetes,anaemia

orerythrocyteindices

MeanHb

MeanHbA

1c

Other

keyfindings

Conclusion

Behan,2006[21]

169(0)

Com

parisonof

HbA

1cin

prem

enopausal

(n=100)

andpostmenopausal(n=69)

wom

enwithoutd

iabetes.

AuthorscorrelateHbA

1cto

FPG

inboth

groups.T

heyalso

introduceabsolute

A1c,(totalH

b×HbA

1c)andtestits

correlationto

FPGin

both

groups.

Diabetesdeterm

ined

byknow

ndiabetes

orFP

G≥7

.0mmol/l

(126

mg/dl).

Prem

enopausal

136g/l

Postmenopausal

139g/l

(p=0.009)

Prem

enopausal

5.4%

Postmenopausal

5.4%

,(p=0.081).

FPGshow

edweak

correlationwith

HbA

1cvalue

overallr

0.251

(p=0.001),w

hich

was

higherin

postmenopausal

wom

enindicating

thereis

haem

odynam

icvariability

that

rendersHbA

1c

values

less

reliablein

term

sof

representing

pastglycaemiain

prem

enopausal

wom

en.

Thisstudysuggeststhat

menstruationduring

middle

agecanhave

asignificant

effecton

theaccuracy

ofHbA

1canalysisandthat

effectisnullified

after

cessationof

menstruation.

Thisisasignificantfinding

becausehalfof

wom

enwho

developtype

2diabetes

doso

during

theperi-m

enopause.

The

authorssuggestthe

useof

absoluteA1cin

orderto

accountfor

haem

atinic

variation.

Coban

etal,2004[17]

100(40)

Studyconsistsof

50patientswith

IDA

and50

healthy,age-andsex-matched

controls.

Hb,PC

V,M

CV,M

CH,ferritin,serum

fasting,andpostprandialglucoseand

HbA

1clevelsmeasured.

Patientswith

IDAweretreatedwith

oral

ferroussulphate100mgadayfor

3months.

IDAdeterm

ined

bylowferritin

levels(<9ng/m

lfor

wom

enand<15

ng/m

lfor

men)

associated

with

microcytic

hypochromicanaemia.

Pretreatm

ent

108±12

g/l

Posttreatm

ent

127±9.6g/l

Controls

136±9g/l

Pretreatm

ent

7.4±0.8%

Posttreatm

ent

6.2±0.6%

Controls

5.2±0.2%

Inpatientswith

IDA,

HbA

1cdecreased

significantly

after

iron

treatm

ent

(p<0.001).

Nosignificant

difference

across

thethreegroups

inFP

Gor

postprandial

glucose.

Thisstudysuggestscorrecting

iron

deficiency

beforemaking

adiagnosticor

therapeutic

decision

basedon

HbA

1c.

El-Agouzaetal,2002[18]

81(N

K)

University

students(n=81)identifiedwith

microcytic,hypochrom

icanaemiaof

which

47wereiron

deficient,26

were

β-thalassaemiacarriersandfour

had

both

conditions.The

participantswith

anaemiaweretreatedwith

oralferrous

sulphate(325

mg/day)

for20

weeks

andrepeatanalyses

performed.

IDAdeterm

ined

asmicrocytic

(<80

flMCV),hypochromic

(<26

pgMCH)andlow

ferritin(undefined).

Pretreatm

ent

109.6±11.2g/l

Post20

weeks

treatment

132±6.5g/l

(p<0.001)

Pretreatm

ent

6.15

±0.62%

Post20

weeks

treatm

ent

5.25

±0.45%

(p<0.001).

HbA

2%

Pretreatm

ent

1.89

±0.45%

Post20

weeks

treatm

ent:

2.19

±0.53%

(p<0.001)

HbF

%Pre

treatm

ent:

0.94

±0.18%

Post20

weeks

treatm

ent:0.95

±0.17%

(p>0.05)

7%of

university

studentsin

Gazahadiron

deficiency.

HbA

1cfalls

significantly

with

iron

replacem

enttherapy

butresultsdidnotp

lateau

at20

weeks.

The

authorsalso

demonstrate

anincrease

inHbA

2post

treatm

entindicatingthat

diagnosesof

β-thalassaemia

carriersmay

bemissedin

iron-deficient

anaemia.

Ford

etal,2011[13]

8,296(N

K)

Datafrom

theNHANES,

1999–2002.

HbA

1candcompletebloodcountw

ere

used

toexam

inewhethertherewas

anassociationbetweenanaemiaand

HbA

1candwhetherassociations

betweenconcentrations

ofglucose

andHbA

1cdiffered

accordingto

Hbandiron

status.

IDAdeterm

ined

by:

Low

Hb(<120g/linwom

enaged

20-69,<118g/lin

wom

enaged

≥70,<137g/l

inmen

aged

20-49,<133g/l

inmen

aged

50-69and

<124g/linmen

≥70)

and

twoor

moreabnorm

alities:

lowTSA

T(<15%),low

serum

ferritin(<26.96

pmol/l)

andelevated

erythrocyteprotoporpyrin

(>1.24

μmol/l).

Prevalence

ofanaemiain

totalcohort:

IDA2.3±0.2%

Non-IDA3.2%

±0.3%

NormalHbbut

iron

deficient

4.4±0.3%

NormalHband

norm

aliron

90.1±0.4%

Totalp

opulation

meanHbA

1c

5.28%

with

Hb

<100g/lto

5.72%

with

Hb>170g/l

The

adjusted

mean

concentrations

ofHbA

1cwere5.56%

and5.46%

among

participantswith

andwithoutiron

deficiency

(p=0.095).

Low

Hbbutn

ormal

iron

vsnorm

alHb

andnorm

aliron

hadsignificantly

lowerHbA

1c

(p=0.001)

NormalHbbutlow

iron

vsnorm

alHbandnorm

aliron

hadborderline

higherHbA

1c

(p=0.061)

Greatesteffectseen

inHbA

1c<5.0%

group;

also

highest

numberof

low

Hbnorm

aliron

seen

inthisgroup.

Hbconcentrationappearsto

bepositivelycorrelated

with

HbA

1cvalue.

HbA

1cappearshigherin

patientswith

iron

deficiency

andsimilarto

values

inpatientswith

norm

aliron

andnorm

alHb.

The

authorssuggestcaution

should

betakenwhen

diagnosing

diabetes

and

prediabetesin

peoplewith

ahigh

orlowHbwhenthe

HbA

1cleveliscloseto

the

6.5%

or5.7%

cuto

ff.

1412 Diabetologia (2015) 58:1409–1421

Page 5

Tab

le2

(contin

ued)

Author,

year

andtitle

n(m

ale)

Study

Diabetes,anaemia

orerythrocyteindices

MeanHb

MeanHbA

1c

Other

keyfindings

Conclusion

Gram-H

ansenetal,1990[19]

30(N

K)

Tenpatientswith

IDA,ten

with

vitamin

B12

deficiency

anaemiaandtenhealthy

controlswerestudied.HbA

1c,Hband

erythrocytevariablesweremeasured

at0,3,6and9weeks

afterinitiation

oftreatm

ent.

Effecto

firon

andB12

therapyon

HbA

1c

over9weeks

was

assessed.

Novariablesgivenfordiabetes,

IDAor

vitamin

B12

deficiency.

Haematinicdataforcontrols

also

notp

resented.

MedianHb

IDApre

treatm

ent:

5.1mmol/l

B12Apre

treatm

ent:

6.1mmol/l

Controlsnot

presented;

assumed

within

quoted

ref

rangeof

7.5-10.4

mmol/l

IDAat3weeks:

7.2mmol/l

B12Aat3weeks:

8.3mmol/l

IDAandB12A

at9weeks

notp

resented.

MedianHbA

1c

IDApretreatm

ent

4.9%

B12Apretreatment

5.1%

Controls

5.0%

IDAat9weeks

4.6%

B12Aat9weeks

4.8%

.

Noinitialsignificant

difference

between

theID

Agroup4.9%

andvitamin

B12

deficiency

group

5.1%

(p<0.4).

After3weeks

oftherapy,HbA

1c

inID

A4.3%

and

vitamin

B12

deficiency

anaemia

4.4%

decreased

significantly

(p<0.01).

Significance

not

stated

at9weeks.

The

authorsconclude

that,in

patientswith

iron

and

vitamin

B12

deficiency,

HbA

1cisan

important

markerof

thechangesin

theerythrocytepopulation

thatoccurwhentherapy

isinitiated.

Thispaperiswidelycitedby

otherarticleson

thesubject

ofanaemiaandHbA

1c

measurement.

Hardikaretal,2012[14]

243(136)

Studyincluded

youngadultsfrom

thePu

neChildren’sStudycohortin

India.The

authorsinvestigated

thediagnostic

performance

ofHbA

1cagainsta

standard

OGTT,

andlooked

atthe

haem

atological,nutritionaland

other

factorsinfluencingHbA

1cconcentration.

Diabetesdeterm

ined

byOGTTWHOcriteria.

HbA

1cADAcriteria;

prediabetes5.7–6.4%

);diabetes

≥6.5%.

Anaem

iadeterm

ined

byHb<120g/linfemales

and<130g/linmales.

MeanHb130±20

g/l

Prevalence

ofanaemia:

Approx.onethirdhad

anaemia,43.6%

had

microcytosis,2.5%

hadmacrocytosis.

66.7%

wereiron

deficient,30.8%

wereB12

deficient

and15.4%

were

folatedeficient.

MeanHbA

1c5.4%

,range4.4-6.7%

.OGTTshow

edthat

7.8%

ofparticipants

wereprediabetic

and

2.6%

werediabetic.

The

HbA

1cvalues

show

edthat23.3%

wereprediabetic

and

2.6%

werediabetic.

Thisincreasedto

33%

having

diabetes

orprediabetesin

the

anaemicgroup.

Therewas

higherprevalence

ofiron

deficiency

inthe

participantsclassified

byHbA

1cas

prediabetic

ordiabetic.S

erum

ferritin

concentrations

were

significantly

lowerin

prediabetesanddiabetes

comparedwith

the

norm

algroup.

The

authorssuggestthat

diagnosing

diabetes

and

prediabetesin

iron-deficient

populatio

nsmay

lead

tospuriously

high

HbA

1c

concentrations

andpotential

forincreasedmis-diagnosis

ofdiabetes.

Kim

etal,2010[12]

10,535

(3869)

Datafrom

theNHANES1999–2006surveys.

The

distributionof

HbA

1cvalues

were

exam

ined

atacutp

ointof

<5.5%

vs≥5

.5%

andbecauseof

therecommended

cutp

oint

of<6.5%

vs≥6

.5%.S

tudy

looksattheinfluenceof

iron

deficiency

onHbA

1cdistributionam

ongadults

withoutd

iabetes.

Iron

deficiency

determ

ined

byatleasttwoabnorm

alities

includingfree

erythrocyte

protoporphyrin

>1.24

μmol/lerythrocytes,T

SAT

<16%,orserum

ferritin

≤15μg/l.

IDAdeterm

ined

by:asabove

with

Hb<135g/linmen

and

<120g/linwom

en.

MeanHbdatanot

provided.

Prevalence

data:

13.7%

(n=1150)

wom

enhadiron

deficiency,and

ofthese30%

had

anaemia(n=345).

1.6%

(n=75)men

hadiron

deficiency,

andof

these33%

hadanaemia(n=33).

2.3%

(n=127)

wom

enand3.3%

(n=125)

men

hadanaemia

butn

otiron

deficiency.

MeanHbA

1cin

wom

enwith

iron

deficiency

was

5.31

±0.02%

andwithoutiron

deficiency

5.27

±0.01%

(p=0.127)

MeanHbA

1cin

men

with

iron

deficiency

was

5.43

±0.06%

andwithoutiron

deficiency

5.29

±0.02%

(p=0.035).

Although

inwom

enthemeanHbA

1c

didnotd

ifferbetween

theiron

deficientand

iron

sufficient

there

was

anupwardshift

indistributionof

HbA

1cfrom

≤5.4%

to5.5-6.0%

.316wom

enhadiron

deficiency

and

HbA

1c≥5

.5%

and

32hadID

Aand

HbA

1c≥6

.5%.13

men

hadiron

deficiency

and

HbA

1c≥5

.5%

and

only

12hadID

AandHbA

1c≥6

.5%.

The

authorsconcludedthat

iron

deficiency

was

common

amongwom

en;thisiron

deficiency

was

notn

ecessarily

accompanied

byanaemia.

Iron

deficiency

inthefemale

cohortledto

are-distribution

ofHbA

1cvalues.

The

shiftinlevelsatthehigher

endof

HbA

1c(6.0–6.5%)was

notsignificant

buto

nlyasm

all

numberof

participantshad

highervalues

ofHbA

1c(n=13).

Diabetologia (2015) 58:1409–1421 1413

Page 6

Tab

le2

(contin

ued)

Author,

year

andtitle

n(m

ale)

Study

Diabetes,anaemia

orerythrocyteindices

MeanHb

MeanHbA

1c

Other

keyfindings

Conclusion

Kogaetal,2007[22]

423(0)

HbA

1c,erythrocytecount,PCV,H

b,MCV

andMCHin

423wom

enwith

norm

alglucosetoleranceweretaken.180of

thepatientswereprem

enopausaland

theother243werepostmenopausal.

Theyexam

ined

therelationship

betweenHbA

1canderythrocyte

indicesin

preandpostmenopausal

participants.

Anaem

iadeterm

ined

byHb<114g/l.

Impaired

glucosemetabolism

anddiabetes

diagnosed

usingOGTTand

WHOcriteria.

Hbin

prem

enopausal

wom

en:1

26g/l

vspostmenopausal:

130g/l(p<0.0001).

FPGin

pre-vs

postmenopausal

wom

en:

5.1±0.3mmol/land

5.3±0.3mmol/l

(p<0.0001).

HbA

1cin

pre-vs

postmenopausal

wom

en:

4.9±0.2%

and

5.1±0.3%

(p<0.0001).

Erythrocytecounts

(4.30±0.30

×10

6 /μl)

ofprem

enopausal

wom

enwerepositively

associated

with

HbA

1c

(4.9±0.2%

),whereas

Hb(126

±12

g/l),

MCV(89.9±6.1fl)

andMCH(29.4±2.7

pg/cell)show

ednegativeassociation.

Incontrast,no

significantassociation

ofanyindiceswith

HbA

1c(5.1±0.3%

)was

detected

inthe

postmenopausal

wom

en.

The

authorsconcludedthat

erythrocyteindicesare

associated

with

HbA

1c,

independently

ofplasma

glucoselevels,in

prem

enopausalw

omen

even

whenthey

arenotanaem

ic.

The

authorshighlight

thatMCH

andMCVareearlyindicators

ofiron

deficiency

andmay

change

priorto

afallin

Hblevels.

Theyrecommendthisbe

considered

wheninterpretingHbA

1cin

prem

enopausalp

atientswith

diabetes.

Kogaetal,2010[23]

104(0)

Studyincluded

57patientswith

norm

aliron

state(N

IS),30

with

IDSand17

with

IDA.M

easurementsof

FPG,

erythrocytecount,GA,H

b,PCV

MCV,serum

iron,S

TATandserum

ferritinwereincluded.

IDAdeterm

ined

byHb<114g/land

serum

ferritin

<15

ng/m

l.ID

Sdeterm

ined

byHb≥114

g/land

serum

ferritin

<15

ng/m

l.NIS

determ

ined

byHb≥114

g/land

serum

ferritin

≥15ng/m

l.

NIS

Hb131±8g/l

IDS124±8g/l

IDA103±6g/l

p<0.001forNIS

vsID

Sandalso

forID

Avs

NIS

andvs

IDS.

NIS

4.8±0.2%

IDS5.0±0.2%

IDA5.1±0.2%

p<0.001forNIS

vsID

Aandp<0.05

forNIS

vsID

S.

HbA

1cshow

edsignificantinverse

associationwith

serum

iron,T

SAT

andserum

ferritin.

HbA

1clevelsin

IDA

andID

Swereslightly

butsignificantly

higherthan

NIS

participants.

Therewereno

significantly

differentv

aluesof

GAbetweengroups.

Serum

ferritin(decreased)

was

identifiedas

asignificantriskfactor

forincreasedHbA

1c.

The

authorsconcludedthatiron

metabolism

indicesinfluence

HbA

1clevels,but

notserum

GAlevels,inprem

enopausal

wom

en.

The

authorsstatethatthey

found

noevidence

ofID

Acausing

increasedlifespanof

erythrocytes.

Shanthietal,2013

[16]

100(38)

50non-diabeticID

Apatientsand50

healthyagematched

participants.

Hb,PC

VMCV,M

CH,ferritin,F

PGandHbA

1cwereanalysed,to

determ

inetheeffectsof

IDAon

HbA

1clevelsin

nondiabetics.

Anaem

iadeterm

ined

by:H

b<110

g/l,ferritinlevels<9ng/m

lforwom

enand<15

ng/m

lformen

andon

theirperipheral

bloodsm

earsmostly

microcytic

hypochromic,

which

suggestsID

A.

The

meanHb

(106

±14

g/l)

levelinthepatients

with

IDAwas

lowerthan

those

inthecontrol

group(134

±9.6g/l)

(p<0.05).

MeanHbA

1c(7.6±0.5%

)levelinpatientswith

IDAwas

higherthan

thosein

thecontrol

group(5.5±0.8%

)(p<0.001).

Alliron

indiceswere

lowerin

IDA

comparedwith

controls;h

owever,

FPG

was

not

significantly

different.

The

authorsconclude

thatHbA

1c

isnotaffectedby

bloodsugar

levelsalone,andthatiron

deficiency

hasasubstantial

effecton

HbA

1c.

The

authorssuggestthatanaem

iamay

lead

toashortenedhalf-life

oferythrocytes

ratherthan

elongation.Theyalso

suggest

thatID

Achangesthequaternary

structureof

Hbleadingto

increasedglycation.

1414 Diabetologia (2015) 58:1409–1421

Page 7

Tab

le2

(contin

ued)

Author,

year

andtitle

n(m

ale)

Study

Diabetes,anaemia

orerythrocyteindices

MeanHb

MeanHbA

1c

Other

keyfindings

Conclusion

Sinhaetal,2012[20]

50(16)

Controls50

(29)

Thisstudyinvestigated

theeffectsof

IDAon

HbA

1clevelsandwhether

treatm

ento

fID

Ainfluenced

HbA

1c

levels.50patientsconfirmed

tohave

IDAweretreatedwith

iron

supplements.

Hb,MCH,P

CV,M

CVandMCHC

atbaseline,at1month

andat2months

wereanalysed;the

controlswereonly

analysed

once.

Diabetesdeterm

ined

by-FP

G>5.6mmol/l

Mild

anaemiadeterm

ined

byHb120-129g/linmales

and

110-119g/linfemales

Moderateanaemiadeterm

ined

byHb90-119

g/linmales

and80-109

g/linfemales.

Severeanaemiadeterm

ined

byHb<90

g/linmales

and

<80

g/linfemales.

IDdeterm

ined

bypredom

inantly

microcytic

indices(M

CV<80

fl)

andhypochromicindices

(MCH<26

pg/cell),confirm

edby

lowserum

ferritin(<10

ng/m

lin

females

and<29

ng/m

linmales).

The

meanHbatbaseline

inanaemicpatients

(62±21

g/l)was

significantly

lower

than

thatin

the

controlg

roup

(134

±6g/l)

(p<0.01).

The

meanHbafter2

months(125

±10

g/l)

was

significantly

higherthan

atbaseline

(p<0.01).How

ever,

thiswas

still

lower

than

controls

(p<0.01).

MeanHbA

1catbaseline

inanaemicpatients

(4.6%)was

significantly

lowerthan

thatin

the

controlg

roup

(5.5%)

(p<0.05).

MeanHbA

1cafter2

months(5.9±0.6%

)was

significantly

higherthan

atbaseline(p<0.01).

Asignificantincrease

was

observed

inthe

absoluteHbA

1c

levelsat2months

aftertreatm

ent(0.29

g/dl

vs0.73

g/dl,

p<0.01).

ThisstudyfounddecreasedHbA

1c

levelsatbaselineandarise

inHbA

1cwith

iron

supplementation;

theseresults

arein

completecontrastto

the

majority

ofotherstudies.

Noexplanationof

theresults

was

presented.

Sonetal,2013[15]

329(111)

Thisstudyexam

ined

anaemicindividuals

who

weredrug

naiveandsuspectedof

having

diabetes

andcomparedwith

age-

andsex-matched

controls.P

articipants

underw

entanOGTTandHbA

1cvalues

weresimultaneouslyassessed.

Diabetesdeterm

ined

byFPG

≥7.0mmol/lor

≥11.1mmol/l,

2hafterglucoseload

Prediabetes:

FPG

5.6-69

mmol/lor

2hglucose

7.8-11.0mmol/l

Anaem

iadeterm

ined

byHb<130g/l

inmen

and<120g/linwom

en.

Anaem

icgroupHb:

113±11

g/l

Control

group

142±13

g/l

(p<0.001)

Inthenorm

oglycaem

icgroups

(byFPG

)therewas

nodifference

inHbA

1cvalues

betweenanaemic

andcontrols.

Inprediabetesgroups,

HbA

1cwas

6.4±1.0%

vs6.1±0.7%

inanaemic

vscontrols(p=0.05).

The

comparisons

ofHbA

1cin

anaemicand

controlsfor2h

glucoselevels

weresimilarbut

show

edborderline

significance.

Glucose-based

distributions

werenot

significantly

differentb

etween

anaemicand

non-anaemic

groups.

Stratifying

patientsby

HbA

1c

show

edan

increasedsensitivity

inanaemicgroupbutd

ecreased

specificity.

The

authorssuggestthata

diagnosis

ofDM

byHbA

1cin

anaemia

may

have

decreaseddiagnostic

significance

butacknowledge

thatstudynumbersaresm

all

andthetypesof

anaemiawere

notd

efined.

ESM

Table2provides

adetailedversionof

thistable,with

inclusionandexclusioncriteriaanddualreportingof

HbA

1cvalues

(%andmmol/m

ol)

CRFchronicrenalfailure,D

Mdiabetes

mellitus,G

Aglycated

albumin,IDiron

deficiency,IDSiron-deficient

state,IG

Tim

paired

glucosetolerance,NIS

norm

aliron

state,NKnotk

nown

Diabetologia (2015) 58:1409–1421 1415

Page 8

Studies comparing prevalence of diabetes and prediabetesdiagnosed by glucose or HbA1c Two studies [14, 15] com-pared the prevalence of diabetes/prediabetes determinedby glucose-based criteria and HbA1c-based criteria.Hardikar et al [14] compared HbA1c and OGTT deter-mined diabetes rates in a cohort of young adults inIndia. The authors found that the prevalence of diabetesor prediabetes was higher with HbA1c criteria than bystandard OGTT (25.9% vs 10.4%). Within a subgroup ofanaemic (IDA, B12 and folate deficiency) patients, thediscordance was even greater with 33% classed as predi-abetes or diabetes by HbA1c compared with 12% byOGTT.

Son et al [15] grouped 329 Korean participants, not pre-viously known to have diabetes, by diabetes status accord-ing to OGTT results. Participants were grouped as normal,prediabetes or diabetes, HbA1c levels were then comparedin each group for both anaemic and non-anaemic partici-pants. In the normoglycaemic group, HbA1c values werefound to be the same in both anaemic and non-anaemicgroups. In the prediabetes group, HbA1c values were foundto be higher in the anaemic group compared with controls(p=0.05) and borderline significant in the diabetes group.The authors acknowledge that small sample sizes and a lackof definition of types of anaemia are confounding factors intheir study.

Studies comparing HbA1c and glucose values in patientswithout diabetes, with and without IDA One study [16]compared the HbA1c values in 50 individuals with IDA, notpreviously known to have diabetes, with non-anaemic healthy

controls. This was a short article with limited description ordiscussion but essentially both FPG and postprandial glucosevalues were not significantly different between the IDA andcontrol groups. HbA1c values however were significantlyhigher in the IDA group (mean 7.6±0.5% [60±5.5 mmol/mol]) compared with the control group (5.5±0.8% [37±8.7 mmol/mol]; p<0.001).

Studies evaluating the effects of treatment, to resolve anae-mia, on HbA1c Four studies [17–20] evaluated HbA1c levelsin non-diabetic patients, pre and post treatment for anaemia.Each study included patients treated with oral iron replacementtherapy, three studies [17–19] reported a significant fall inHbA1c values after treatment with iron replacement over pe-riods of 9–20 weeks.

Coban et al [17] saw a fall from an HbA1c of 7.2±0.8%(55±8.7 mmol/mol) pre treatment to 6.2±0.6% (44±6.6 mmol/mol) post 3 months treatment but this was still notas low as the non-anaemic control group 5.2±0.2% (33±2.2 mmol/mol); however, the iron indices had not fully normal-ised compared with controls at that point in time either. Thesame pattern was demonstrated by El-Agouza et al [18] wherepatients were followed for 20 weeks and HbA1c fell steadi-ly over that period from 6.15±0.62% to 5.25±0.45% (44±6.8 mmol/mol to 34±4.9 mmol/mol). However, there wereno control participants included for comparison and itwould be of interest to know the point at which HbA1c

values plateaued post treatment. Similarly, a small study(n=10) [19], showed a significant decline (p<0.01) inHbA1c values by 3 weeks of treatment and the valuesremained below baseline values after 9 weeks, despite anupward rebound.

Interestingly, a study by Sinha et al [20] showed the con-verse pattern of results in 50 patients with moderate to severeIDA (mean±SD Hb, 62±1 g/l). The mean HbA1c was signif-icantly lower in the anaemic group compared with the non-anaemic controls (4.6±0.6% vs 5.5±0.6%; 27±6.6 mmol/molvs 37±6.6 mmol/mol). After 2 months of treatment forIDA, the HbA1c values were significantly higher than thecontrols raising the question of whether severe iron defi-ciency has the same effect on glycation as more mildanaemia.

Studies investigating the effect of menstruation on HbA1c

levels Three studies [21–23] investigated the impact of thepremenopausal state on HbA1c values. Behan [21] com-pared Hb values in non-diabetic, non-anaemic premeno-pausal women with postmenopausal white women. Thestudy found that Hb levels were significantly lower inpremenopausal women, despite no reported history ofanaemia. HbA1c values were not significantly differentbetween the two groups but the correlation betweenHbA1c and FPG was weaker in the premenopausal group.

EMBASE n=386 MEDLINE n=135 CINAHL n=12 Cochrane n=11

Total n=544

Total, after duplicates

removed n=451

Full text articles reviewed n=43

Exclusions based on title and/or

abstract n=408

Full text articles included n=12

Exclusions based on full text n=31

Fig. 1 Flow chart detailing the search strategy employed

1416 Diabetologia (2015) 58:1409–1421

Page 9

The author suggests that the varying effects of menstrua-tion on iron metabolism could lead to less reliable HbA1c

and FPG correlations.Koga [22, 23] and colleagues also looked at the effect of the

menopause on HbA1c values. The initial study identified a sig-nificant difference in HbA1c levels between pre- and postmen-opausal women (4.9±0.2% vs 5.1±0.3% [30±2.2 mmol/molvs 32±2.2 mmol/mol] p<0.0001). In premenopausal women,mean corpuscular haemoglobin (MCH) and mean corpuscularvolume (MCV) were inversely correlated with HbA1c but thisassociation was not significant in postmenopausal women. Theauthors suggest that every 1 pg decrease in MCH correlatedwith a 0.03% (0.3 mmol/mol) increase in HbA1c value and thaterythrocyte indices influence HbA1c values in premenopausalwomen in the absence of overt anaemia. Koga et al [23] alsolooked at the effect of iron deficiency and IDA compared withiron sufficiency on HbA1c and glycated albumin values in pre-menopausal women. Iron deficiency and IDAwere both asso-ciated with a significant increase in HbA1c levels with ironmetabolism indices being significantly negatively associatedwith HbA1c. This association was not present in glycated albu-min levels across the three iron status groups, essentially indi-cating that iron metabolism rather than glycaemic changes in-fluence changes in HbA1c in premenopausal women.

Studies evaluating erythrocyte indices andmarkers of ironstatus All studies were reviewed for data on erythrocyte indi-ces and markers of iron status as these provide further insightinto which adjunct tests may support the use of HbA1c fordiagnosis. Data extracted are detailed in Table 3.Combinations of the following indices were measured: Hb,packed cell volume (PCV), MCV, MCH, MCH concentration(MCHC), ferritin, transferrin saturation (TSAT), reticulocytes,red cell distribution width (RDW) and erythrocyte protopor-phyrin. Some studies only used the values to identify or excludepatients with iron deficiency or anaemia whereas others corre-lated changes in erythrocyte indices with changes in HbA1c

value. Of the studies that focused on IDA, three studies com-pared values pre and post treatment with iron replacement andtwo compared values in patients with anaemia against controlparticipants. All studies demonstrated an increase in MCVandMCHwith treatment or higher levels in controls compared withanaemia patients. Two studies [14, 22] demonstrated an inversecorrelation between HbA1c and Hb, MCV and MCH levels.Overall, the data indicate that iron deficiency, demonstratedby low Hb, low MCV and low MCH, is associated with in-creased HbA1c levels both with and without overt anaemia. Inaddition, normal MCV and MCH with low Hb would not beassociated with increased HbA1c levels but rather a decrease invalues.

Ferritin was measured in nine studies [12–14, 16–18, 20,21, 23]. Of these, most showed an increase in ferritin levelspost treatment for anaemia and also showed that ferritin levels

were lower in iron-deficient participants compared with con-trols. Three studies [14, 16, 23] showed that ferritin was anegative predictor of HbA1c and one study [13] showed asmall but significant positive correlation between HbA1c andferritin. Generally, ferritin was utilised as a marker of irondeficiency rather than analysed as an independent predictorof HbA1c values; where the latter was done, the results weremixed.

Discussion

Does anaemia affect HbA1c?

Generally, the studies investigating the effect of anaemia andabnormalities of erythrocyte indices on HbA1c analysis arelimited to small sample groups and two studies that bothanalysed the same large cross-sectional population study. Itis apparent that IDA can have a significant impact on HbA1c

values with most studies suggesting a spuriously high HbA1c

in IDA compared with other markers of glycaemia. The ex-ception to this is the study by Sinha et al [20], which foundlow values of HbA1c in severe cases of IDA that increasedwith iron replacement therapy. The authors did not discusswhy their data conflicts with other studies but it may be dueto the severity of anaemia as the participants in this study hadlow mean Hb levels (62 g/l); the duration of anaemia was notgiven. Together, these data support the hypothesis that irondeficiency per se may cause elevated HbA1c values, irrespec-tive of anaemia.

Limited data indicate that non-IDAs also affect HbA1c

values to a varying degree [13, 14, 19]. Ford et al [13] showedthat in patients without diabetes, with low Hb but normal ironlevels had significantly lower HbA1c values than those withnormal iron and normal Hb (see Table 2). The difference be-tween the non-IDA and control values was greater than wasobserved between IDA and iron deficiency values and con-trols, suggesting that HbA1c may be spuriously elevated iniron deficiency and spuriously depressed in non-IDAs.Ideally, any further studies would separate out iron deficiencyand non-iron deficiency cases prior to analysis as there is thepotential to null the data by combing the two pathologies inone evaluation.

Hardikar et al [14] investigated a population where of thosewith anaemia 30.8% had vitamin B12 deficiency, 15% hadfolate deficiency and 30% had multiple nutrient deficiencies.In multivariate analysis, B12 and folate were not significantlyrelated to HbA1c levels but this may be due to the small sam-ple size analysed. Where mixed nutrient deficiencies are ob-served, the combination of pathologies will lead to variableand unpredictable effects on HbA1c levels depending on therelative degree of each nutrient deficiency.

Diabetologia (2015) 58:1409–1421 1417

Page 10

Tab

le3

Correlatio

nsof

erythrocyteindicesandmarkersof

iron

status

with

HbA

1c

Study

Erythrocyteindicesmeasured

Mainobservations

Conclusions

Notes

Behan,2006[21]

Hb,MCVandRDW

Hbsignificantly

lower

inprem

enopausalw

omen,

nodifference

forMCVandRDW

Noconclusionsdraw

nregarding

erythrocyteindices.

WhilstH

bwas

lower,noneof

thepatients

hadlowferritin,toruleoutirondeficiency.

Coban

etal,2004[17]

Hb,PCV,M

CV,M

CH,ferritin

Hb,PCV,M

CV,M

CHandferritin↓intheID

Agroupcomparedwith

controls.P

osttreatment

nosignificantd

ifferences

betweengroups.

Treatmentw

ithiron

appearsto

norm

alisetheerythrocyteindices

alongwith

areductionin

HbA

1c

values

at3months.

Erythrocyteindicesused

todemonstrate

norm

alisationin

patientswith

anaemia.

RDW

values

may

have

assisted

inassessingtheheterogeneity

ofthe

erythrocytepopulatio

n.

El-Agouzaetal,2002[18]

Hb,ferritin,PCV,M

CV,M

CH,

MCHC,R

DW

Participantshadmicrocytic,m

icrochromic

indices—

notallhad↓ferritinlevels—notallID

.Postironreplacem

entin51

iron-deficient

participants—allindices

significantly

increased

except

MCHC(smalld

ecrease).

Treatmentw

ithiron

appearsto

norm

alisetheerythrocyteindices

alongwith

areductionin

HbA

1c

values.

Erythrocyteindicesused

todemonstratenorm

alisationof

iron

status.

RDW

meanwas

significantly

higher

than

thenorm

alrangepretreatm

entand

even

higher

posttreatm

entshowingavery

mixed

celltype

profile

at20

weeks

oftreatm

ent.Thiswould

indicatethat

erythrocyteindicesmay

notstabiliseeven

at20

weeks.

Ford

etal,2011[13]

Hb,PCV,M

CV,M

CH,M

CHC,

ferritin,TSA

TPearsoncorrelations:

HbA

1c↑asferritin↑

HbA

1c↑as

MCHandMCV↓

Not

asexpected

intheface

ofID

A:

↓Ferritin

,MCHandMCVwould

beexpected

with

↑HbA

1c.

MCVandMCHmay

bebettermarkers

forID

andID

Athan

ferritin.

Gram-Hansenetal,1990[19]

Hb,MCV,reticulocytes

Hb↑andMCV↑,HbA

1c↓w

ithiron

replacem

ent

Hb↑,MCV↓andHbA

1c↓with

B12

replacem

ent

MCVused

todifferentiatebetween

IDAandnon-ID

A.

Verylim

itedstudyandduratio

nof

follo

w-up.

Hardikaretal,2012[14]

Hb,PCV,M

CV,M

CH,M

CHC,

ferritin

Hb,MCV,M

CH,M

CHCnegativ

elycorrelated

with

HbA

1c

Markersof

microchromicanaemia

wereassociated

with

increased

HbA

1cvalues

inpeoplewho

were

norm

alby

OGTT.

MCVdidnotshowasignificantchange

inpreandposttreatm

entg

roups,

despite

negativ

ecorrelations

toHbA

1c.

Kim

etal,2010[12]

Erythrocyteprotoporphyrin,

ferritin,TSA

TIndicesmeasuredbuto

nlyused

toidentifyID

ANonecanbe

draw

n.

Kogaetal,2007[22]

RBC,H

b,PC

V,M

CH,M

CV

Hb,MCH,M

CVhadastatistically

significant

negativ

ecorrelationwith

HbA

1cin

prem

enopausalw

omen

only

↓Hb,MCH,M

CVassociated

with

↑HbA

1c,indicatesID

assessed

byMCHandMCV,correlates

to↑H

bA1cwith

nochange

inglucoselevels.

Nohistoryof

diabetes

oranaemiaincohort.

Kogaetal,2010[23]

Iron,ferritin

,TSAT

Serum

iron,T

SATandlogferritininversely

associated

with

HbA

1c

Low

iron

indicesassociated

with

↑HbA

1c.

MCHandPC

Vmeasuredanddecreased

with

IDAbutn

otcorrelated

toHbA

1c.

Shanthietal,2013

[16]

Hb,PCV,M

CV,M

CH,F

erritin

All↓in

IDAgroup,HbA

1c↑inthisgroup

Statisticalsignificance

isunclear.

Verypoorly

describedresults.

Sinhaetal,2012[20]

Ferritin

↓Hband↓ferritin

associated

with

↓HbA

1c,

allincreased

with

iron

replacem

enttherapy

HbA

1cappear

tocorrelatewith

Hb

andferritin.

MCV,M

CH,H

bused

todifferentiatetype

ofanaemiabutv

aluesnotreported.

Sonetal,2013[15]

Hb,MCV,M

CH,

Hb,MCVandMCHall↓

in‘anaem

ic’group

definedonly

byHb

Nocorrelations

madeto

HbA

1c.

Nodifferentiatio

nbetweentypesof

anaemiain

thestudyalthough

mean

MCVandMCHsuggestp

redominantly

IDA.

ID,irondeficiency

1418 Diabetologia (2015) 58:1409–1421

Page 11

To what degree do abnormalities of erythrocyte indicesaffect HbA1c values?

The two evaluations of the cross-sectional NHANES datashowed that having iron deficiency or IDA increased the oddsof having an HbA1c value shift upwards from <5.5%(<37 mmol/mol) to 5.5–5.9% (37–41 mmol/mol) [12, 13].Hardikar et al [14] demonstrated that markers ofmicrochromic anaemia were associated with increasedHbA1c values and also more patients diagnosed with predia-betes or diabetes by HbA1c compared with glucose indices. Inpatients with non-IDA, there was a marked shift downwardsin apparent distribution of HbA1c results [13]. Son et al [15]demonstrated that IDA in the patients with prediabetes definedby glucose levels had a mean HbA1c of 6.4% vs 6.1%(46 mmol/mol vs 43 mmol/mol) in controls; this is sufficientto re-categorise some patients from prediabetes to diabetes inthe anaemic group.

Other studies which looked more specifically at HbA1c

levels in patients with or without anaemia showed differencesof up to 2.1% (23 mmol/mol) HbA1c increase with IDA [16]and studies measuring HbA1c pre and post treatment for anae-mia showed changes as large as −1.2% (−13 mmol/mol)HbA1c post treatment [17].

What does this mean on a wider scale?

Recent data has shown a surge in the incidence and prevalenceof young onset obesity, many of whom are premenopausalfemales, at increased risk of iron deficiency. There has beena concomitant increase in HbA1c values in these participantsconsidered at high risk of diabetes, owing to factors such asobesity [24]. Data from England shows that the prevalence ofprediabetes rose from 11.6% in 2003 to 35.3% in 2011 [25]and in view of the rising prevalence of obesity, we wouldanticipate to identify a larger proportion of patients, withhigher HbA1c values. If the estimated upward shift in HbA1c

values seen with iron deficiency is combined with the appar-ent increased prevalence of prediabetes, it may result in asignificant number of patients where the combination of thetwo will be sufficient to shift HbA1c values to move from adiagnosis of prediabetes to diabetes.

Clinical use and relevance

One of the main questions still to answer is how to apply thisinformation to everyday clinical practice. In addition to pub-lished guidance [4–6], we suggest the following:

1. During monitoring of people with diabetes, when glucoseand HbA1c are discordant, consider abnormalities oferythrocyte indices.

2. When HbA1c is normal/elevated but Hb is low, do notassume that HbA1c is falsely elevated—check eryth-rocyte indices, in particular MCV and MCH; if low,consider iron deficiency by TSAT or ferritin. If MCVand MCH are not low then consider other forms ofanaemia—HbA1c may be falsely decreased in thesecases.

3. Iron deficiency, as well as IDA, may be sufficient to causea change in HbA1c values; this is highly relevant in wom-en of childbearing age.

4. If abnormalities of erythrocyte indices or anaemia areidentified, consider correction of the abnormality beforeusing HbA1c for diagnosis or monitoring. The studies in-cluded in this review suggest that it may take up to6 months after treatment is initiated to normalise erythro-cyte indices. RDW will provide an additional indicator ofnormalisation of the erythrocyte population and erythro-cyte lifespan.

Ferritin as a marker of iron status is useful if it is low butwould not rule out iron deficiency if it was found to be normalor elevated as ferritin is an acute phase reactant and inflam-mation may mask a true low ferritin. Raj and Rajan [26] in-vestigated 86 patients with type 2 diabetes mellitus and dem-onstrated that serum ferritin positively correlated with HbA1c

and was increased with increasing duration of disease, indi-cating that poor glycaemic control can contribute to elevatedferritin levels independent of iron status. In situations ofincreased inflammation, alternative measures of iron statussuch as TSAT or total iron binding capacity should beconsidered.

Changes in erythrocyte lifespan even with normalhaematological indices can affect HbA1c values [27],therefore, it is important to show that the erythrocytepopulation is stable after treatment for abnormalities be-fore using HbA1c. Erythrocyte indices including RDWmay assist with this.

Further research questions

Although it is clear that anaemia may influence HbA1c

results, further studies to identify the roles of erythrocyteindices as appropriate adjunct analyses to identify patientswhere this is an issue, are needed. Future studies shouldconsider the effect of erythrocyte indices rather than anae-mia alone. In particular, studies should aim to includesufficient participants to differentiate between the effectsof the type and severity of erythrocyte abnormalities/anaemia on HbA1c values. This may be achieved througha combination of population based studies and interventionstudies where changes in HbA1c values are assessed inrelation to therapy for erythrocyte abnormalities andanaemias.

Diabetologia (2015) 58:1409–1421 1419

Page 12

Strengths and limitations of this study

This review is limited to a selection of small study groups andtwo separate analyses of the NHANES survey data from over-lapping time periods. The conclusions drawn are limited by alack of robust evidence from a significant number of the in-cluded studies. The quality analysis of the included studies hashighlighted shortfalls and confounders in most of the cases,which would need to be addressed in future studies—in par-ticular, a lack of power to confer significance on many of thefindings.

Notes on standardisation of HbA1c methods

International standardisation of HbA1c was achieved in 2002[28–30]. An International HbA1c Consensus Committee wasformed which agreed that HbA1c values should be reportedin both SI units (mmol/mol) and converted to % units viathe master equations established by the InternationalFederation for Clinical Chemistry and Laboratory Medicine(IFCC) and designated comparison methods [10, 11]. Allstudies reported in this review either stated that they werealigned to the NGSP or did not state any performance infor-mation for the HbA1c analysis. In light of this, even if therewas sufficient data for meta-analysis, it is likely that a lackof standardisation of results would be a major confounder.Any further studies should endeavour to ensure that allHbA1c measurements are performed in alignment with theIFCC and clear quality data should be provided in thereports.

Conclusion

It is clear from the limited number of studies, many with lowparticipant numbers, that the subject of anaemia and HbA1c

warrants further investigation. Generally, the studies describeddemonstrate that abnormalities of erythrocyte indices are aconsiderable confounder in the analysis of HbA1c and thereis currently insufficient data to fully inform clinicians andscientists on how to address this in clinical practice.However, we have made some suggestions to facilitate itsuse whilst clarity is awaited.

The key questions that are still to be answered are whetheranaemia and erythrocyte abnormalities will have a significantimpact on the diagnosis of diabetes using HbA1c in the generalpopulation—something that is now widely performed.

Acknowledgements We would like to thank R. Curtis (LibraryServices, University of Nottingham, Nottingham, UK) for her helpand guidance in constructing the database searches.

Funding This research received no specific grant from any fundingagency in the public, commercial or not-for-profit sectors.

Duality of interest statement The authors declare that there is no du-ality of interest associated with this manuscript.

Contribution statement All of the authors listed have made a substan-tial contribution to the production of this article. EE, II, KD, EK andWGJall contributed to the concept of the work; EE, GS and II designed the datacollection; EE andGS collected and analysed the data. All authors revisedthe manuscript critically for important intellectual content and approvedthe final version to be submitted/published. EE drafted the manuscript.EE is the guarantor of this work.

References

1. The Diabetes Control and Complications Trial Research Group(1993) The effect of intensive treatment of diabetes on the devel-opment and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med 329:977–986

2. UK Prospective Diabetes Study (UKPDS) Group (1998) Intensiveblood-glucose control with sulphonylureas or insulin comparedwith conventional treatment and risk of complications in patientswith type 2 diabetes (UKPDS 33). Lancet 352:837–853

3. Inzucchi SE, Bergenstal RM, Buse JB et al (2012) Management ofhyperglycemia in type 2 diabetes: a patient-centered approach: po-sition statement of the American Diabetes Association (ADA) andthe European Association for the Study of Diabetes (EASD).Diabetes Care 35:1364–1379

4. World Health Organization (2011) Use of glycated haemoglobin(HbA1c) in the diagnosis of diabetes mellitus (abbreviated reportof a WHO consultation)

5. AmericanDiabetes Association (2013) Standards ofmedical care indiabetes–2013. Diabetes Care 36(Suppl 1):S11–S66

6. John WG (2012) Use of HbA1c in the diagnosis of diabetesmellitus in the UK. The implementation of World HealthOrganization guidance 2011. Diabet Med 29:1350–1357

7. United Nations Children’s Fund, United Nations University, WorldHealth Organization (2001) Iron deficiency anaemia assessment,prevention, and control: a guide for programme managers.Available from www.who.int/nutrition/publications/en/ida_assessment_prevention_control.pdf

8. Stevens GA, Finucane MM, De-Regil LM et al (2013) Global,regional, and national trends in haemoglobin concentration andprevalence of total and severe anaemia in children and pregnantand non-pregnant women for 1995-2011: a systematic analysis ofpopulation-representative data. Lancet Glob Health 1:e16–25

9. Lowe G, Twaddle S (2005) The Scottish Intercollegiate GuidelinesNetwork (SIGN): an update. Scot Med J 50:51–52

10. Consensus Committee (2007) Consensus statement on the world-wide standardization of the hemoglobin A1C measurement: theAmerican Diabetes Association, European Association for theStudy of Diabetes, International Federation of Clinical Chemistryand Laboratory Medicine, and the International DiabetesFederation. Diabetes Care 30:2399–2400

11. Hanas R, John G, International Hb ACC (2010) 2010 consensusstatement on the worldwide standardization of the hemoglobin A1cmeasurement. Clin Chem 56:1362–1364

12. Kim C, Bullard KM, HermanWH, Beckles GL (2010) Associationbetween iron deficiency and A1C Levels among adults withoutdiabetes in the National Health and Nutrition ExaminationSurvey, 1999–2006. Diabetes Care 33:780–785

13. Ford ES, Cowie CC, Li C, Handelsman Y, Bloomgarden ZT (2011)Iron-deficiency anemia, non-iron-deficiency anemia and HbA1camong adults in the US. J Diabetes 3:67–73

1420 Diabetologia (2015) 58:1409–1421

Page 13

14. Hardikar PS, Joshi SM, Bhat DS et al (2012) Spuriously high prev-alence of prediabetes diagnosed by HbA(1c) in young Indians part-ly explained by hematological factors and iron deficiency anemia.Diabetes Care 35:797–802

15. Son JI, Rhee SY, Woo JT et al (2013) Hemoglobin A1c may be aninadequate diagnostic tool for diabetes mellitus in anemic subjects.Diabetes Metab J 37:343–348

16. Shanthi B, Revathy C, Manjula Devi AJ, Subhashree (2013) Effectof iron deficiency on glycation of haemoglobin in nondiabetics.J Clin Diagn Res 7:15–17

17. Coban E, Ozdogan M, Timuragaoglu A (2004) Effect of iron defi-ciency anemia on the levels of hemoglobin A1c in nondiabeticpatients. Acta Haematol 112:126–128

18. El-Agouza I, Abu Shahla A, Sirdah M (2002) The effect of irondeficiency anaemia on the levels of haemoglobin subtypes: possibleconsequences for clinical diagnosis. Clin Lab Haematol 24:285–289

19. Gram-Hansen P, Eriksen J, Mourits-Andersen T, Olesen L (1990)Glycosylated haemoglobin (HbA1c) in iron- and vitaminB12 deficiency. J Intern Med 227:133–136

20. Sinha N, Mishra TK, Singh T, Gupta N (2012) Effect of iron defi-ciency anemia on hemoglobin A1c levels. Ann Lab Med 32:17–22

21. Behan KJ (2006) Cessation of menstruation improves the correla-tion of FPG to hemoglobin A1c in Caucasian women. Clin Lab Sci19:225–230

22. Koga M, Morita S, Saito H, Mukai M, Kasayama S (2007)Association of erythrocyte indices with glycated haemoglobin inpre-menopausal women. Diabet Med 24:843–847

23. Koga M, Saito H, Mukai M, Matsumoto S, Kasayama S (2010)Influence of iron metabolism indices on glycated haemoglobinbut not glycated albumin levels in premenopausal women.Acta Diabetol 47(Suppl 1):S65–S69

24. Selvin E, Parrinello CM, Sacks DB, Coresh J (2014) Trends inprevalence and control of diabetes in the United States, 1988–1994 and 1999–2010. Ann Intern Med 160:517–525

25. Mainous AG 3rd, Tanner RJ, Baker R, Zayas CE, Harle CA (2014)Prevalence of prediabetes in England from 2003 to 2011: popula-tion-based, cross-sectional study. BMJ Open 4:e005002

26. Raj S, Rajan GV (2013) Correlation between elevated serumferritin and HbA1c in type 2 diabetes mellitus. Int J Res Med Sci 1:12–15

27. Cohen RM, Franco RS, Khera PK et al (2008) Red cell life spanheterogeneity in hematologically normal people is sufficient to alterHbA1c. Blood 112:4284–4291

28. Hoelzel W, Weykamp C, Jeppsson JO et al (2004) IFCC refer-ence system for measurement of hemoglobin A1c in humanblood and the national standardization schemes in the UnitedStates, Japan, and Sweden: a method-comparison study. Clin Chem50:166–174

29. Jeppsson JO, Kobold U, Barr J et al (2002) Approved IFCCreference method for the measurement of HbA(1c) in humanblood. Clin Chem Lab Med 40:78–89

30. Weykamp C, John WG, Mosca A et al (2008) The IFCCReference Measurement System for HbA1c: a 6-year prog-ress report. Clin Chem 54:240–248

Diabetologia (2015) 58:1409–1421 1421