Full blood count normal reference valuesfor adults in FranceXavier Troussard,1 Sylviane Vol,2 Edouard Cornet,1 Valérie Bardet,3

Jean-Paul Couaillac,4 Chantal Fossat,5 Jean-Charles Luce,2 Eric Maldonado,6

Virginie Siguret,7 Jean Tichet,2 Olivier Lantieri,2 Joël Corberand,8 for theFrench-Speaking Cellular Hematology Group (Groupe Francophone d’HématologieCellulaire, GFHC)

▸ Additional material ispublished online only. To viewplease visit the journal online(http://dx.doi.org/10.1136/jclinpath-2013-201687).1Laboratoire d’Hématologie,CHU Côte de Nacre, CaenCedex 9, France2IRSA, La Riche Cedex, France3Service d’HématologieBiologique, CHU Cochin-PortRoyal, Paris Cedex 14, France4Laboratoire Hémato-biologie,CH Cahors, Cahors, France5Laboratoire d’Hématologie,CHU Timone, Marseille Cedex5, France6Marketing Hématologie,Siemens HealthcareDiagnostics, Saint Denis,France7Service d’Hématologiebiologique, Hôpital EuropéenGeorges Pompidou, Paris,France8Laboratoire d’Hématologie,CHU Toulouse—Rangueil,Toulouse Cedex 9, France

Correspondence toDr Xavier Troussard,Laboratoire d’Hématologie,CHU Côte de Nacre, Caen14000, France;troussard-x@chu-caen.fr

Received 16 April 2013Revised 17 September 2013Accepted 23 September 2013

To cite: Troussard X, Vol S,Cornet E, et al. J Clin PatholPublished Online First:[please include Day MonthYear] doi:10.1136/jclinpath-2013-201687

ABSTRACTAims To determine full blood count (FBC) normalreference values for adults.Methods FBC normal values for healthy adults weredefined, after establishing preanalytical conditions, in apopulation of 33 258 subjects, 19 612 men and 13 646women. The values were established after excludingfrom this population all people having conditions liableto modify, directly or indirectly, FBC parameters.Results Results for values of standard parameters areprovided in detail for each parameter, by sex and by agegroup from 16 to 69 years of age. In addition, wepresent FBC values from a population of 339 subjectsaged over 69 years with no comorbidities.Conclusions These normal values are proposed for usein everyday practice. They make it possible to distinguish,without ambiguity, a normal situation from apathological situation. Moreover, they might be usedover all mainland France.

The full blood count (FBC) is the most frequentlyrequested laboratory test in France. The wealth ofinformation provided by this multiparameter ana-lysis can notably suggest specific diagnoses and canlead to discovery of a haematological disease in afortuitous and timely manner so that appropriateclinical management can be instigated.Requesting physicians lacking a solid haematology

background base their interpretation of an FBC onthe quantitative reference values provided by thelaboratory. In France, as a usual practice, patientsconsult or receive, in most cases, the results of theirFBC, read the results by themselves and thenattempt to interpret them and to compare themwith the reference values given on the results sheet.The quality of laboratory tests is constantly

improving. The technological development of ana-lysers and control over the analytical phase reducethe risk of having to question the accuracy of thequantitative data of the FBC. Nevertheless, patientspresent to haematologists after interpreting theirvalues as abnormal in comparison with the refer-ence values given on their results sheet. There is noneed to carry out structured surveys to demonstratethe very wide variability in the reference valuesprovided by laboratories in mainland France.Clinical pathology is currently undergoing

change and restructuring and a regulatory require-ment for all laboratories is to prove their

accreditation procedures and hence to have recog-nised and validated reference values.A group of experts from the French-Speaking

Cellular Hematology Group propose on the basisof established methodological principles FBCnormal reference values for use by all clinical path-ology laboratories in mainland France.

MATERIALS AND METHODSPopulationData from 142 110 adult subjects, 63 729 men(44.8%) and 78 381 women (55.2%), were ana-lysed. All subjects were aged 16–69 years and had aperiodic health assessment (PHA) at theInter-Regional Health Institute (IRSA) between 1January 2008 and 31 July 2010.IRSA consists of 11 Prevention and Public Health

Centers (CPSP) located in the Central-West area ofFrance. The number of subjects from each CPSPwere: La Riche: 21 307 (15%), Hérouville-St-Clair:20 981 (15%), St-Lô: 6747 (5%), Alençon: 11 605(8%), Le Mans: 21 806 (15%), Angers: 19 812(14%), Cholet: 7869 (5.5%), Saint Doulchard:6217 (4.5%), Châteauroux: 10 789 (8%), LeHavre: 6727 (5%) and Laval: 8250 (6%). Frenchlegislation does not allow knowing the ethnicity ofpatients. However, the regions covered by the studyare not known to be regions with a high percentageof non-Caucasians.In France, the PHA is a regulatory service gov-

erned by law and available to all people affiliated tothe general health insurance scheme, which covers85% of the French population. The content of thePHA is determined by a reference document underthe authority of the National Health InsuranceFund (Caisse Nationale d’Assurance Maladie) withinput from medical experts. It is updated periodic-ally and consists of three parts: (1) a self-questionnaire (146 items) for the collection ofsocioprofessional and family data, behavioursnotably diet habits, physical activity, habits ofsmoking and alcohol, and the subject’s past andpresent health status, (2) laboratory tests includingan FBC, lipids profile, a blood glucose measure-ment, liver function tests, and a urinary dipsticktest for glucose, proteins, blood and acetone. Othertests (ferritin, serology for hepatitis B and C) maybe added according to age, sex and context. HIVserology can also be carried out on the patient’srequest during the medical visit and (3) a medical

Troussard X, et al. J Clin Pathol 2013;0:1–4. doi:10.1136/jclinpath-2013-201687 1

Original article JCP Online First, published on October 29, 2013 as 10.1136/jclinpath-2013-201687

Copyright Article author (or their employer) 2013. Produced by BMJ Publishing Group Ltd under licence.

group.bmj.com on January 6, 2014 - Published by jcp.bmj.comDownloaded from

examination based on the preceding elements to establish asummary of the subject’s health.

A total of 142 110 individuals were initially analysed ofwhom 42 589 (30%) were in a situation of deprivation. Thetarget populations in precarious situations are all remote systemhealth people: lack of identification of referring physician,homelessness, unemployed, recipients of minimum income orassisted contracts. The score EPICES (Evaluation of insecurityand Health Inequalities in Health Examination Centers) is basedon 11 questions to determine if a person is in a precarious situ-ation.1 Given that the percentage of deprived individuals in theFrench population is estimated at 13%, the initial populationwas rectified in line with this criterion. A random number wasassigned to every precarious subject and 27 717 of them, theones with the n lowest random numbers, were excluded. Afterrectification, the population consisted of 114 393 subjects,including 51 636 men (45%), 62 757 women (55%), and14 872 precarious subjects (13%).

In order to determine normal FBC values in healthy subjects,81 135 subjects were excluded from our study because of cri-teria liable to directly or indirectly modify FBC parameters.Eighteen exclusion criteria were used (table 1), most commonly‘medication use’ in 58 009 subjects (52.5%) or ‘follow-up fordisease’ in 23 150 subjects (21%). In the end, 33 258 subjects,19 612 men (59%) and 13 646 women (41%), were used todetermine normal values in healthy adult subjects. Note thatour study does not allow excluding patients with iron deficiencyand/or thalassaemia/haemoglobinopathies.

Full blood countBlood samples from subjects in a fasting state were collectedfrom the elbow crease between 07:00 and 10:30 into Sarstedttubes with K3 EDTA. They reached the IRSA central laboratoryin Tours in less than 5 h. This laboratory is ISO 9001 certified.

The FBC was performed on an Advia 2120 analyser (SiemensFrance Holding SAS, 9 boulevard Finot—93 527 Saint-DenisCedex 2, France) that was quality control tested three times perweek by Eurocell Diagnostics (EUROCELL Diagnostics, 27, ruedu Village de la Métairie—35131 Chartres de Bretagne, France)using three samples (low, normal, high) with a monthly reportof results. Haematocrit on ADVIA 2120 is calculated. The resultcomes from mean cell volume (MCV) (measure) and red bloodcell (RBC) (measure) with a cytometry flow technology. MCVon ADVIA system is measured directly cell by cell after a specificreaction used RBCs reagent. No impact of potential residualplasma could be observed, as with a manual technology.International units were used.2

StatisticsAll analyses were carried out using Number CrunchingStatistical Software 2007. The age groups had a width of10 years, with the exception of one 4-year group (16–19 years)and four 5-year groups in women aged 40–59 years in order torefine FBC changes associated with the menopause. FBC vari-ables by age and sex were described with means and SDs as wellas by 23 percentiles, from the 1st to the 99th. The 2.5 and 97.5percentiles were used to describe normal values. When establish-ing the normal values, adjacent age groups were merged whenthe statistical difference between them was non-significant.Comparisons between age ranges and between sexes were madeusing Kruskal–Wallis test.

RESULTSThe selected population, with age and sex distribution, is shownin table 2. It consists of 32 919 subjects, including 19 393 menand 13 526 women, aged 16–69 years, and closely representa-tive of the population of mainland France. In addition, 339 sub-jects aged 70–79 years, 219 men and 120 women, were alsoanalysed, giving a total of 33 258 subjects.

Results for the values of standard parameters used in everydaypractice for the interpretation of the FBC are provided forevery parameter, by sex and by age group, in the appendixsection. The summary table (table 3) provides reference valuesby sex with upper and lower limits for subjects aged16–69 years. Given the fact that a high percentage (91.7%) ofsubjects aged 70–79 are not eligible in a reference population,we present reference values for that healthy limited population(339 subjects) with no known comorbidities in online supple-mentary appendix 1.

Variation between different age groups was observed.Red cell parameters (see online supplementary appendix 2):

For haemoglobin, in the age group 16–69 years, the mean

Table 1 Exclusion criteria applied to the initial population of114 393 subjects, of whom 14 872 were in a situation ofprecariousness, to obtain the analysis reference population of32 919 individuals

Exclusion criteria Men (%) Women (%) Total (n)

Taking medication 37.9 64.5 58 009Follow-up/treatment for disease 21.4 25.1 23 150Asthma 11.1 10.9 10 446Regular cough 11.0 6.3 8730Blood in stools 7.8 6.5 7268Personal cardiovascular history 7.6 6.5 7120Transfusion 3.9 7.5 5920Polyps 7.2 5.2 5761Blood donation less than 3 months ago 5.5 4.3 4958Pathological bleeding 6.1 3265Diabetes 3.1 2.6 2663Epilepsy 2.1 2.4 2119Tuberculosis 1.2 1.5 1257Follow-up for kidney failure 0.8 0.5 635Pregnancy 1.0 534Ulcerative colitis 0.7 0.6 590Crohn’s disease 0.2 0.3 255Colon cancer 0.3 0.2 232Population without exclusion criteria (%, n) 37.6%

n=19 39321.6%n=13 526

28.8%n=32 919

Table 2 Age distribution of analysis population

Age group (years) Men Women Total

16–19 1387 0901 228820–29 3054 1370 442430–39 4258 2711 696940–49 5146 1873 895150–54 3801 1932 724955–59 1870

60–64 1747 1578 303865–69 1291Total 19 393 13 526 32 919

2 Troussard X, et al. J Clin Pathol 2013;0:1–4. doi:10.1136/jclinpath-2013-201687

Original article

group.bmj.com on January 6, 2014 - Published by jcp.bmj.comDownloaded from

value is 150 g/L (134–167 g/L) for men and 134 (117–150) forwomen (p<10−6). For women, haemoglobin concentration ishigher in the age group 45–69 years than in the age group 16–44 years (135 vs 133 g/L, p<10−6). A slight variation in haemo-globin concentration, namely, a significant increase (p<10−6), isseen in women in the age groups (45–49 vs 50–54, p=10−3;50–54 vs 55–59, p=0.399 (NS); 55–59 vs 60–69, p=8×10−4)in comparison with lower age groups in which the level isstable. For the MCV, the mean value is 87.3 fL (80.2–95.0) inmen and 87.5 (78.4–95.3) in women (p=0.014). Regularincreases in the values with increasing age are seen in both sexes(significant increase at each age group and in both sexes,p<10−6). The variation is approximately 3 fL for the age limitsat the extremities. For the mean cell haemoglobin concentration,the mean value is 343 g/L (324–363) for men and 338 (319–358) for women (p<10−6). For the mean corpuscular haemo-globin, the mean value is 30.0 pg/cell (27.2–32.8) in men and29.7 (26.1–32.5) in women (p<10−6).

Platelet parameters (see online supplementary appendix 3):For the platelet count, mean values in women are higher thanthose in men. The values are, respectively, 269×109/L (171–397) in men and 394×109/L (186–440) in women (p<10−6).This difference is seen in all age groups analysed (p<10−6 foreach age group). For the mean platelet volume, the mean valueis 8.7 fL (7.4–10.8) in men and 8.8 fL (7.5–10.9) in women(p<10−6).

White cell parameters (see online supplementary appendix 4):The values of white cell parameters are expressed in absolutevalues (#) not percentages (%) (table 3). The mean value ofleucocytes is 6.6×109/L (4.1–10.8) in men and 6.6×109/L(3.9–10.9) in women (p=0.130). For neutrophils, the meanvalue is 3.6×109/L (1.8–6.8) in men and 3.7×109/L (1.7–7.1)in women (p<10−6). For eosinophils, the mean value is0.21×109/L (0.05–0.56) in men and 0.19×109/L (0.04–0.53) inwomen (p<10−6). For basophils, values are identical in menand women: 0.04×109/L (0.00–0.09) (p<10−6). For lympho-cytes, the mean value is 2.3×109/L (1.3–3.8) in men and2.2×109/L (1.3–3.6) in women (p<10−6). For monocytes, thevalue is 0.43×109/L (0.23–0.74) in men and 0.38×109/L(0.20–0.65) in women (p<10−6).

DISCUSSIONThe FBC is the most frequently requested laboratory test: itguides later additional investigations essential for diagnosingand monitoring the patient. It provides valuable information,provided that it is possible to distinguish between a normalsituation and a pathological situation in a very diverse rangeof conditions. Under French regulations, all public- andprivate-sector laboratories must give normal reference rangeson FBC results sheets. These values vary widely betweenlaboratories and are liable to lead to interpretation errors andalso unnecessary worry in patients. As other tests, the resultsof FBC analyses could be dependent on the analytical phase.Furthermore, technical advances in automation and the con-tinual improvement of quality programmes also contribute tothe high homogeneity of the analytical phase of the FBC. Theauto-analyser we used could have impacted on the valuesobtained, especially values from healthy subjects. If they doexist, they are probably minor and non-significant with theproviso that all requirements of the quality programme arerespected (preanalytical, internal and external qualitycontrols).

Mainland France, situated on the 45th North parallel, has atemperate climate even if mountain ranges are present. TheFrench population living at an altitude higher than 2500 m islimited and so variations due to environmental pressure are oflittle significance. Although individual variations, in particularethnic characteristics, cannot be ignored, we can consider thatin mainland France, individuals do not exhibit variation inher-ent to where they live or where the blood sample was col-lected. For this reason, the values that we report in this studycan be used by all laboratories in mainland France, regardlessof the technical analysis platform in use, and their adoption byall laboratories would eliminate the risk of interpretation errorsby physicians or patients.

Defining reference values is a delicate undertaking. Thestudies in the literature that include a sufficient number ofpatients investigated with current methods for performing anFBC3 4 are limited. The FBC normal values given in Frenchand foreign textbooks are also arbitrary, lacking referencesand usually old.5 In September 1997, the National HealthAccreditation and Evaluation Agency (Agence Nationale

Table 3 Full blood count normal reference values defined byvalues comprised between the 2.5 and 97.5 percentiles

Men(years) Min–max

Women(years) Min–max

Haematocrit (%) 16–69 39.2–48.6 16–44 34.4–43.945–54 34.7–44.655–69 35.9–44.6

Haemoglobin (g/L) 16–69 134–167 16–49 115–14950–54 118–15155–69 121–150

Red blood cells(1012/L)

16–29 4.53–5.79 16–29 4.01–5.1930–59 4.38–5.65 30–49 3.93–5.0960–69 4.28–5.57 50–69 3.99–5.12

MCV (fL) 16–19 78.0–91.9 16–19 75.7–92.720–39 79.6–94.0 20–29 74.7–94.240–49 81.0–94.9 30–39 77.9–95.350–59 82.2–96.3 40–69 79.9–95.660–69 82.1–97.0

MCH (pg/cell) 16–19 26.3–32.1 16–29 24.4–32.120–69 27.3–32.8 30–69 26.4–32.6

MCHC (g/L) 16–69 324–363 16–69 319–358Leucocytes (109/L) 16–49 4.09–11.00 16–44 4.02–11.42

50–59 4.06–10.46 45–49 4.01–11.0260–69 4.05–9.92 50–54 3.78–9.70

55–69 3.78–9.42Neutrophils (109/L) 16–49 1.780–6.946 16–44 1.750–7.500

50–59 1.915–6.634 45–49 1.812–7.15460–69 1.847–6.138 50–54 1.720–6.299

55–69 1.692–5.839Eosinophils (109/L) 16–19 0.046–0.630 16–19 0.040–0.576

20–29 0.048–0.593 20–49 0.041–0.54930–59 0.046–0.547 50–69 0.044–0.47460–69 0.052–0.576

Basophils (109/L) 16–39 0.000–0.097 16–19 0.000–0.08140–69 0.000–0.091 20–69 0.000–0.085

Lymphocytes (109/L)

16–39 1.340–3.919 16–29 1.370–3.96640–69 1.241–3.617 30–69 1.240–3.561

Monocytes (109/L) 16–39 0.228–0.773 16–29 0.201–0.71440–69 0.233–0.725 30–49 0.205–0.663

50–69 0.192–0.608Platelets (109/L) 16–59 172–398 16–54 185–445

60–69 161–393 55–69 187–420MPV (fL) 16–69 7.4–10.8 16–69 7.5–10.9

Min, minimum, 2.5 percentiles; Max, maximum, 97.5 percentiles.MCH, mean corpuscular haemoglobin; MCHC, mean cell haemoglobin concentration;MCV, mean cell volume; MPV, mean platelet volume.

Troussard X, et al. J Clin Pathol 2013;0:1–4. doi:10.1136/jclinpath-2013-201687 3

Original article

group.bmj.com on January 6, 2014 - Published by jcp.bmj.comDownloaded from

d’Accréditation et d’Evaluation en Santé) published a docu-ment giving FBC normal values based on a literature reviewcovering articles less than 10-years-old and involving morethan 200 patients.6 Laboratories are therefore faced with theproblem of providing normal values and not having nationalrecommendations. With accreditation programmes, thisproblem is becoming a veritable cause for concern for direc-tors of clinical pathology laboratories.

We have presented FBC normal data obtained between2008 and 2010 in a population living in mainland Francewith established preanalytical conditions and modern analyt-ical methods based on recent technologies. The populationstudied was large, consisting of 32 919 subjects selected afteranalysis of advanced clinical and biological criteria combinedwith a statistical/epidemiological approach to perform a finalsmoothing for societal criteria liable to interfere with thenutritional state (nutrient deficiencies, addiction, etc). Thepopulation investigated in this study can be considered to berepresentative of the current adult population of mainlandFrance. We have also presented normal reference values forthe population of healthy subjects aged 70–79 years, a popu-lation that often receives little attention in the literature.7

For methodological reasons, this study did not include chil-dren; it only included adult subjects. Paediatric normal valueshave been presented in other works.8–10 Given the French banon entering the ethnic origin of an individual into an electronicdatabase, it is impossible to present the biological variationrelated to ethnic origin, despite the multiethnic nature of theFrench population. Modifications in white cells and neutrophilscount in subjects of African ancestry are known.11 12 The ques-tionnaire used for the PHA does not take into considerationexposure to environmental factors, pesticides, radiation, electro-magnetic waves or genetic factors liable to modify certain haem-atological factors. Another limitation of our study is the lack ofability to eliminate in our population iron deficiency or anyhaemoglobinopathies.

Our study did not envisage normal values for reticulocytesas there is no reason to perform this investigation systematicallybeyond cases of suspected anaemia or in a PHA. Parametersprovided by the auto-analysers and dependent on the analyticaltechniques or the mode of result expression implemented bythe manufacturer, such as red cell distribution width or plateletparameters other than mean platelet volume, were notanalysed.

The normal reference values presented here, very close toother data,13 are proposed for use in everyday practice in orderto be able to affirm without ambiguity the abnormal nature ofany FBC parameter. Adoption of these normal values by alllaboratories in mainland France should make it possible to inter-pret correctly an FBC from any adult subject, regardless of thelaboratory and its geographic location.

Take home messages

▸ The full blood count is the most frequently requestedlaboratory test: it guides later additional investigationsessential for diagnosing and monitoring the patient.

▸ Laboratories are faced with the problem of providing normalvalues and not having national recommendations.

▸ We presented normal reference values in 32 919 subjectsconsidered to be representative of the French population.

▸ These normal values are proposed for use in everydaypractice. They make it possible to distinguish a normalsituation from a pathological situation.

Acknowledgements The authors thank all the collaborators, particularly ChristineGeslain, and the subjects participating in the study.

Contributors XT, EC and JC designed the study and wrote the manuscript. SVperformed the statistical analysis. VB, J-PC, CF, EM and VS discussed the results.J-CL analysed the biological samples. JT and OL enrolled the patients and providedthe clinical data.

Competing interests None.

Ethics approval Institut inter Régional pour la Santé (IRSA).

Provenance and peer review Not commissioned; externally peer reviewed.

REFERENCES1 Sass C, Moulin JJ, Guéguen R, et al. Le score EPICES : un score individuel de

précarité et évaluation du score dans une population de 197389 personnes. BullEpidemiol Hebdo 2006;14:93–6.

2 Kratz A, Ferraro M, Sluss PM, et al. Laboratory reference values. N Engl J Med2004;351:1548–63.

3 Nordin G, Mårtensson A, Swolin B, et al. A multicentre study of reference intervalsfor haemoglobin, basic blood cell counts and erythrocyte indices in the adultpopulation of the Nordic countries. Scand J Clin Lab Invest 2004;64:385–98.

4 Wakeman L, Al-Ismail S, Benton A, et al. Robust, routine haematology referenceranges for healthy adults. Int J Lab Hematol 2007;29:279–83.

5 Perkins SL. Normal blood and bone marrow values in humans. Wintrobe’s ClinicalHematology (Tenth Edition). Lee GR, Foerster J, Lukens J, Paraskevas F, Greer JP,Rodgers GM. Williams Wilkins, 1999:2738–48.

6 Lecture critique de l’hémogramme : valeurs seuils à reconnaître comme probablementpathologiques et principales variations non pathologiques. http://www.has-sante.fr

7 Tsang CW, Lazarus R, Smith W, et al. Hematological indices in an older populationsample: derivation of healthy reference values. Clin Chem 1998;44:96–101.

8 Bellamy GJ, Hinchliffe RF, Crawshaw KC, et al. Total and differential leucocytecounts in infants at 2, 5 and 13 months of age. Clin Lab Haematol 2000;22:81–7.

9 Geaghan SM. Hematologic values and appearances in the healthy fetus, neonate,and child. Clin Lab Med 1999;19:1–37.

10 Taylor MR, Holland CV, Spencer R, et al. Haematological reference ranges forschoolchildren. Clin Lab Haematol 1997;19:1–15.

11 Karita E, Ketter N, Price MA, et al. CLSI-derived hematology and biochemistry referenceintervals for healthy adults in eastern and southern Africa. PLoS One 2009;4:e4401.

12 Roshan TM, Rosline H, Ahmed SA, et al. Hematological reference values of healthyMalaysian population. Int J Lab Hematol 2009;31:505–12.

13 Barbara J. Bain. Blood cells: a practical guide. Normal ranges. Oxford, UK:Blackwell Publishing, 2006:198–216.

4 Troussard X, et al. J Clin Pathol 2013;0:1–4. doi:10.1136/jclinpath-2013-201687

Original article

group.bmj.com on January 6, 2014 - Published by jcp.bmj.comDownloaded from

doi: 10.1136/jclinpath-2013-201687 published online October 29, 2013J Clin Pathol

 Xavier Troussard, Sylviane Vol, Edouard Cornet, et al. adults in FranceFull blood count normal reference values for

http://jcp.bmj.com/content/early/2013/10/29/jclinpath-2013-201687.full.htmlUpdated information and services can be found at:

These include:

Data Supplement http://jcp.bmj.com/content/suppl/2013/10/29/jclinpath-2013-201687.DC1.html

"Supplementary Data"

References http://jcp.bmj.com/content/early/2013/10/29/jclinpath-2013-201687.full.html#ref-list-1

This article cites 10 articles, 1 of which can be accessed free at:

P<P Published online October 29, 2013 in advance of the print journal.

serviceEmail alerting

the box at the top right corner of the online article.Receive free email alerts when new articles cite this article. Sign up in

Notes

(DOIs) and date of initial publication. publication. Citations to Advance online articles must include the digital object identifier citable and establish publication priority; they are indexed by PubMed from initialtypeset, but have not not yet appeared in the paper journal. Advance online articles are Advance online articles have been peer reviewed, accepted for publication, edited and

http://group.bmj.com/group/rights-licensing/permissionsTo request permissions go to:

http://journals.bmj.com/cgi/reprintformTo order reprints go to:

http://group.bmj.com/subscribe/To subscribe to BMJ go to:

group.bmj.com on January 6, 2014 - Published by jcp.bmj.comDownloaded from