l,,rriirlri. ,pr,l l",r,r,.rl. Ll'rl.,trr(rt\q l'as 1:' st'Ptl 2 :;-i|' Tlie WHO Antenatal Care Randomised Controlied Trial: rationale and study desigu '. Y. Al-N'lazrou'l' rrolii, U Farnt)tq' . Berendesr for the ir GrouP al Care Trial Research are listed at the end of the PaPer os'ario' Arscrrti'ln' sHosl)ilnl Cirlcro- Ot)slfiriao'.4r'ic[i'n'Arlns'Hni'rlttn'Ctrlrrt'KJtcrrXncttUltii'crstlv' KLorr Xrtt't;, ,',',-1,,,,'i, ';'r'1 \'rlr()rrr;/ /':iliilr;l' oi Cirii'l H.'rtliL 'trr'l ;;:ffi;::. l.;;;;;;;'rLr r\irr/D\rii' trir:ri'':'; tir' u-( ' SumErary' The \Norld Heal tiorls in develoPing countrles controlled trial to eva consisting oi tests' clini demonstrated to be outcomes These actilities four visits during the cours or the traditiolal Protra .{rr,i ; css /,r, corrcsPortrrcr tcc: Dr, } ose .l i' Tfu tJ,?li.Y \il,l;: |^"fiT:,^o tl"iHfil Procrnmne of Research' DeteloPm( lili,i';:"il'",ilsation' 12r I Geneva 27' s'^ itzerl'\nd' r 19qS BlrckNell S(iencc Ltd'
36
Embed
The WHO Antenatal Care Randomised Controlled Trial: rationale and study design
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
f r n. lrtr td Drd.m rd. tu[l'fu!.om t [l tt' 6 'drErlr. fol'..cdllDDl.i.nuton
R..mDddrnoi tor.hcr!cn.'6
THIRD vlSITr lld DATEI
ir'!nurrorod (!..6d.io(,
R..6hh.nd.tm! fr l-rdDn rcntrarnion
FOITRTH VISIT: .dd DATE: ,
D.t..tion of btrc.ch pwnEno. & r.f.ml lor.r!.m.lr'Rd
CohDld. ANC !.r(r<omn.nd i' b. broughl tohotpn,
40 l. Villar et al.
medical records, and all women had a haemoglobin test at the fust visit which was
not recommended in the protocol ln Saudi Arabia, all Rh(-) tvomen were not
eligible tbr tlre basic comPonent of ANC, and the recommended routine vaBinal
"ti*ir,o,ion rvas not culturally acceptable ln Argentina' many phvsicians in the
inten'ention clinics requestcd one ultrasound assessment' even if there rvas no
medical indic^tion, ,.'hich was not recommended in the trial; if the woman had
,roi a"Ur"r"d by rveek 41 o[ g€station, the next visit was conducted at the ANC
clinic rather than in the hospital as required in the Proto€ol; physicians in one unit
randomised to the intervention ErouP imPlemented a glucose tolerance screening
test, although this was not included in the Protocol'A final
-explanation is needed in resPect of the imPlementation of the trial
in Cuba. Because of the universal availability of family doctors' it is possible
that, although rvomen had a reduced number of visits at the clinic' they rvould
;."piu."'tt""r" visits by Soing to the family doctor's of6ce' lve have instructed
these doctors to refer iU pregnant women to the A'NC clinic' but there is still the
possibilir,v of rvomen demanding care that cannot be refused Fortunately' there is
a routine data-collection system in B'hich family doctors have to record anv formal
ANC visit and such a visit is also registered in the \roman's 'd\C card All visits
recorded on this cJrd nie nbstracted iI the time of ,.1eliven bv the inl'estigators.
Tlrerefore'iftherearee\trauisitsrof.rmilv(ioctors,ihe\,rr.illbeincludedinthenumber of visits made to clinics other than the original antenatal clinic'
Moskitrg ol in te roen t ion
Considerable efforts are made to mask treatment status for deliverv and neonatal
care providers. The staff responsible ior data collection after birth at the hospiial
le,el are unarvare o[ the treatment status oi r\,omen ilr the Study' In Argentina, all
deliveries occurring in the ParticiPating hospitals' even those from other clinics in
the city, are reco.-ded, miking it very difficult for hospital clerk to identif'v
intervJntion status. The study site in Thailand is an excePtion' as delivery occurs
atthesameinstihrtionsrvherewomenattend(orANC.Therefore,itispossiblethatstaffinthedelivervroomkntrrvthetleatmentstatusandPerhaPsactdifferently based on their belief ol the effect of ANC' The Primary outcomes are
little influenced by intraParh'rnr care, horvever, and it is exPected tlut this effect
,eco,1d sta,r" (qual)titatiYc), aud rtill also L'e used kr itltcrPr('t tllc results of tlre
quantitatire cu'nPonents. The second plrase uses a standardised' close-ended
questio,r,raire to alt ANC Providers ParticiPating in the trial in every country' and
another questionnaire to a random sample of ra'omen attcnding the clinics'
Qualitatir"e data u'ill be anah'sed bv applving the'grounded theo4' approach'"
that e\ables an illductive illtcrPretation oi tllc illfortllatioll Quanlitative data rtillhe a:ralvsed using stalrdard statistical tcchrli(lues'
Basclinc surz,ev
A formal sun'ey describing ANC Practices uas inrplemented in all eligible clinics
immediatelv before randornisation to obtain a detailed description of clirric and
serr.ice clraracteristics Lrefore the interventiolls. lnfort:ratiolt is availatrle on thc'
clinicai uracticcs .rnd servicc. ()l tire clinics. Tlrere ar.' i\\'o c()l'r'lPo:lents oi this A\Csuncr':cliniclevclatrdPatienilevel.Atthecli:ricletcl,c't'teiornrr''asconrpleter'lfor each clinic particiPating in the study in each coutrtrl. bY the studv field director
and t5e clinic ftcal person. The s.,n,ev coordinators liaised with the clinic staff and
authorities to obtain the most uP-to-date description oi the clinic The visit was not
communicated to the clinic staff in advance and a 'tvpical day' of the clinic u'as
chosen to avoid selection bias of patients and procedures' The principal
investigator of each countrv prepared the agenda for tlle visits'
The-patient component of the survev rtas also conducted before randomisation
and after completing the 'clinic level' form. lt *'as carried out by the dinic focal
pe.son undei the iupen'ision of the shrdy field director and implemented
simultaneously in all ciinics at a given study site An index day n'as chosen by the
sh:dy field director to conduct the sun'ey. The clinic focal person used the dinic
Iogblok to determine the medical record numbers corresponding to the
co-nsecutive visits of women before the index day A minimum of 64 consecutive
records of pregnant women B'ere aimed at each clinic (cluster)' At the Patientlerel, 2913 meJical records u'ere revierr-ed. The desiSn and methodology of tltis
sun'ev is described elsctvhcre irr tl s Supplement ls
Tables 4 and 5 Present the characteristics of the 53 clinics randomised to tlre
Onc area of concern in a trial such as ours tllnl can seriousl], aficct its rEults isq hether or rrot clinicians are comolf in6 u,ith th('nor raodcl of ANC. Obviously,if the tu'o 6;roup; reccive more or lq;s tlre s,rnr(' tvpe of care the trial will be ofIimited valuc. This could r6ult from tlrc intcrtcntiul trouF being providcd u'ithcare additional to that recommcnded in trre tri,ll pr(')t(rcol because stalf see thestandard ANC nrore fatourabir'. Converselr', the stafi o[ thr'control group mayadopt the heu-' model ('therapeutic c()rrtamil,ttio!]') and reLiucc the number ofvisits. One trial on thc routine use of cpisiotorrrt h.rs incieed encountered thisproblem.sT The follorr,ing meclranisms arc in place in tlre prcsent trial to reduceand evaluate inten'ention compliance: exterEi\-e initial training with interventionprotocols; continuous monitorint of protocol conpliance by visits to the clinics atleast every fortnight by the study field director or local principal investigator; andhequent site visits to the study sites by trial coordinators. Compliance rvith theintervention was evalu.lted using the'ANC checklist' (Fig. 2) implemented in theinberlention eroup. This checkiist has comDl!'ted in a randorn sanrFlc oi Yisits bY
the clinician rsponsible tbr the visit, an e\ternal observer and bv using dataobtained from the medical records. Agreement anrong these three 'checklists'\r'as calculated.
Furthermore, u'e have informed care providers in the control clinics in detailthat they are part of a randomised controlled trial, and rve have explained itsnature and have encouraged them to adhere to their standard -ANC protocols andreferral practices. This is |en'important for a comrnr:nitY trial expectins to trul\,c()mpare t\l-o dj.iferent forms of care rather tilan a rigorous .{\C protocol(inien ention) r's. an ANC r\'ithoui a protocol.i' Clinics in both arms of the trialalso receivcd scheduled visits for monitoring patient recruitment and thecompleteness of medical records.
A trial rrith characteristics such as ours provides many opportunities forintroducing bias. We have rnade considerable efforts during the preparation of theprotocol and the implementation to follou, recommended guidelines forrandomised controlled trials in reproductive health3e and for the futue reportingof such trials.o An unpredictable allocation sequence was centrally prepardusing a computerZeneratd random sequence and kept at the cenEal unit untilthe time of treahnent implementation, thus eliminating the possibility oI allocationdeciphering which is more likely to occur in unmasked trials. Cl.inic staff were notaware of group stahrs (allocation concealment) until the time of initiation of thetraining workshops and patient recruitment, thus preventing selection bias, as
after this time there is no longer the possibility of refusing to accept anyparticipant into the sh:dy-
Ho$,evel, there are, in theory, at least t$,o instances in lr,hich clinic staff cansubvert the randomisation process. Firstl)', they can discourage high-risk patientsrthom they may know are considering attending an intervention clinic, iI theyperceive that these women will have poorer outcomes if enrolled in the new,
Tllc WHO nnt l!fll carc RCT: rntionnlc nnd strrr l dcslgrr 51
Tirerc is norr clear cvidcncr.'that h('alth L.'rioritl irltcrventiolls cnn b('evah'talt'dbt,developing countrt re;clrcltcrs ill the conte\t ol largc rancionriscd c()ntrolled
trials and rcstrlrs put'lished in learling iournals.:'{"-i" Dcspitc tlrese success stories,
conducting large-scalc, cutiins-ctl3e health serlicc rtscarch in developing
coL:ntrics is a lirrmitlable trsk. Donol agencics .lrc llot totltlll convincc.l t'i thc
lrcr..el ior thesc trials ftrr the r'\'alultioll of nrt'tiic.l] ilt tcr\'('n ti()l'ls. Partictrlarlv tht'
non-clrug forms oi cart' that s(--l.n 'lqgicalh' gotrd llris ilttitudt is Prescnt evell ifthe trials are relrtiveh' itrcxirensivc bl' industri.ilised coutltries' stand.rrcis ltr the
case of the current conrplex trial, several largc agcncies h.ld to be enroiled to
complete its total budget oi USS 2.5 million for 3 vears and includirrg more than
24 000 women in four countries. Furthermore, the cost of field 'a'ork, overheads
and researchers' salaries is often hard to explain to potential donors rthen thev use
the current logic of international research funding.5r lt is clear tirat the reu ard ofqroup menrbership, satisfling both iniellectuai curiositv atr..j comnlitnrent to
.{,lditional support rias pror-ided tor the imPlemenlatron oi thc srudY bl-: i\lunicipal
Govemment, Citi of Rosario, Argentina; Ministry of Health, Cuba; National Institute of
Public Health, Mexico; The Poputation Council - Regional Of6ce for Latin America and the
Caribbean; Ministry of Health, Saudi Arabia; Swedish Agencv ior Research Cooperation-
sith Developing iountries 6[DA/5AREC); Ministrv of Public Health and Faorlt-v of
lvledicine, Khon Kaen Universiqv, Thailand; DePartment for International DeveloPment(DflD) of the United Kingdom; Mother Care - Iohn Snow lnc.; National Institute for Child
Health & Human Development (NICHD), National Institutes of Health (NIH), USA; the
l{orld BankFor the preparatory phase: University o[ western Ontario, DePartment of EPidemiology
& Biostatistics, Canadi; National lnstitute of Public Health, Noruay; United Nations
Development Programme; University of Uppsala, DePartment of Obstetrics & Gynaecol'
ogy, Sweden.' We would like to thank esPecially the r'romen and their babies who participated in this
trial and the many doctors, nurses and other staff of the clinics and hosPitals that made the
implementation of this Proiect Possible.Special thanks are given to Dr M. Koblinskv lor her personal interest and suPPort for
this project, to Dr O. Meirik for his permanent encouragement and suPPort, to Dr D Khan
Neeiorui. for het eiforts in editing the trial's Ne\{sletter and to \4rs Christine Crav for her
help in the preparation of the manuscriPt.
r 1996 Blaclscll Science Ltd Pnnlintri. n .l P{ nhl El'itt ,iolosy 72, SUFPI 1 27-53
52 l. V illar et al'
Manbers of the WHO Antctlatll Care Trial R*eorch Grortp
Trial coordinating unit
Data coordinating unit
tisticianms Analyst, Statistiial Assistant (until July 1997)
en (from JulY 1997)
Steering committee
Health economics SrouP
Miranda Mugford, Health Economist' UK
iri" ii*,..,"nt"rtit Economics Researcher' uK
ffi fox-tusnbY, Health Economist' UK
QualitY of care grouP
Ana Lans.er, Reproductive Health Specialist' ltlerico
c"i,*o "ui*"n.iu, Public Ht"rlth Spccirlist' Nlexico
\lariana Ro-mero, Public Health EPidemiolotist' Argentrna
c, cnattingius 5. Tle scientiiic basis of nnlenatal cnre. Reporr fiom rr sr.te oi
the art conference. .{ctn Oisicrriri"''l Cv'tu'ritlolticrr Sconiimlico 1991;70: 105-'109'
z.- st".' p. Rituals in antenatal c.1re - do we need them? Brihs,, Medi.al ]ournal 1993; 307:
687-498.i firceffu f. Does Prenatal care imProve birth outcomes? A critical review ' Obslet'ics atd
cun,'coloe! 1995; 85: 468--179.g" fu.i""r t, Hall M, Horvie PlV' Reid !1, Barbour R, Florev CD, el nl Should obstetricians
see women with normal Pregnancils? A multicentre randomized conkolled trial of routine
"ii.ni*f .nr" 1'y generai Pr;ctitioners ar.l midrvives compared rvith shared care led by
obstetriciins. Briti;h tvladic'tl Jr'rrrrrnl l99tr; 312: 5il-559'iO- A"ff.","ig fS. Vcth(xloktqical l'robltrrrs and Possible entl['oints in thc elnlu'1tion oi
anten.rtal carJ h trnntionrl lol:'ntl ', T(,1'lr''\r/ 'lsstss'tt:ttl i Hcnhh Cott'1992; E (SuPPl l):
33-39.1l lteirert C. ivlulticctrtcr tri.ris .rnd irl[cl: luxurv or necessitv? Conlrollci Clitric'tl'frrnls
Khalid Al-Rube ran. MIr FRcrci Yacob Al-Mazrou. rtu. phD. Frr('Cp:
Omcr Al-Attas. phD: Nasscr Al-Daghari. [,tsc
Background: Thc pattcrn and factors which can bc assciated with the glyccmic control of Saudi adult diabeticpaLients were exanrincd in this study.Patienls alld llelhods: Confirmcd diabetic patrenls lionl all regions of Sardi Arahia conslituted thc stud]popularion. Random blood glucosc <10 mnrol/L and >l0lnnrol/L Nas used to catcgonze patieuts into Sood andpoor SI),cemic control paticnts. respectivel).Results: Thcre werc 613 confirnrcd non-insulin dependent diabelic patients (NIDDM).50% \['ith Sood gl!'ccmicconlrol. Patienls with poor gllcemic conrrol *ere sig ificanrly older than patients with good glycemic control(51 5 rs. .17 \ears. Pd) 0001). The insulin-rreared diabetic populrtion amountcd to l3ft. comp.tled rvith J3%and -l-1% for oral agenl and dier. respeclivel) The rite ol insulin users among poor gllcenric conlrol diab€ticpopulation uas Istt. compared with 509r for oral xErenls. There \\as:r significant relalionship bet$'een
Sl\cefiic conuol and age. and lrealmenl nloduljties ol Dl\'l Subjects who had good glycemic control of DMrvere yoonger and tbllowing J dicl regimen. 'rhilc lhose \rho had poor glFenlic control werc older and on
insulin trcatment- Multivariale analysis cornprising.lI5 indiYiduals $as conducted to fin(l out thc faclors that
can potentially influence. or ma)' be associaled \\ilh. lhe control of DM.Conclusion: The association ol insulin therap) with po()r glycemic control is not a cluse+ffect relationship-
lnsulin therapy in our study populat,on is underutilizfd. given the high rate of poor glycemic conrrol and high
rare of relauve occurrence of complication amon€l th. Scudi diabetic population. There is a nced to addrcss thc
imponrnce ol maintaining go<xl glyccmic conlrol. and the relson for the low rate of insulin users. Closeperiodic monitoring of g)ycemrc control. urilizins laboralories and homc glucose moniloring deviccs. is
required. Eflbctive impletncntation of rhese D)easures. in addirron ro diabetes education. will have an irnpJct on
rhe turure outcome of the Srudi dirbetic populalion.Ann Soutli ltled l99E: l8(2 t.109'l 12.
Ke1 Words: Dirl)eres mellltus. gl)-cemic control.
Drrhetes urellitus (DM) is a diseasi rvith a high prevalence
rvorldrvide. and in rhe last feu, lears ils prevalence has
become nrore widespread in the third \r'orld or developingcounrries.r This is especially the case in the MiddleEastern courrtrics. uhich havc expericnccd an upward
surge in the prevllcrrce of DI\1 over rhc lasl l0 yelrs.- -fhis
is rhe likclv result of economical dcvelopment and changcs
rn lifestyle. especially in nutritionul habits.r The long-ternr
nicrorasculur and nracrovascular coutplications of.liltere:' are rcsponsrblc for sigrific nt morbidity and
\Jdr.ss rcp r rcqucsrs Jnd corrcsB)ndcn.. rr' DI ^l-Nuairrr:DcprnnEnr ol Medrcinc (MBC-.16). KrnS Fri$l st-..eirh\r llosprrrl I
Rc!('rrch C.nrrc. P.O. Br)\ -1151. Rivrdh I ll I l. SJrdr ArrbirAcccpred lbr publicrlirnr l5 No\cmbcr l19? Rc.ttvcd l0 APril l9{)7
blindness in adulrs. the nrost common cause of end-stagte
renal disease, accounting for 3Ola-40% of lhis population.is responsible for.10c,/c of all non-traumatic ampulations ofthe lower exlremities in adults. and is a major risk factorfor cardiac and cerebrovascular diseases."
Saudi Arabia has an estinatcd population of 13.2
million. 70% of rvhom are under 30 years of age.6 DM'rclated complications arc a fiequent (rcurrence among
Suudi diabctic patients. A review of 1000 consecutive
Saudi diabetic patrents in a gencral hospital showed an
incidence rate ol 329i for rctinoPathy. 26% lorlrl,pertension. il.l9. fol ischenrrc hcart disease. and 6.9'lilu ren l insufticiencl'.' Diaberes-related hospituladmissions arc liequenl and tcnd to occupy hospital beds
[or a longer time. as comprred to orl]er di:,cases.s
Eprdenriological studies hrve establishcd a relalionshiptrcrween hyperglyccmi nd the de!elopment ol lhe long-rcrnr complications of diabcres.'' rr The importancc of tightglycenric control with respect to dl]lay or In the prevcntion
,turtlr o! Suth tlt,li itt lil lN. ,\r 1. lt)t)lt I09
.\l--\l il'.'
of compir.ation. \\a\ nol cslirhlrshed unril lecentlr'. 'flrc
Diaherei Compii,-ation Contr<tl Tri:rl (DCCT) has provcrl
lhe imponan:. o: tight glyccmic cotttrol lirr thc prcvcntiulof conrrol complicalion antong insulin-tlcpel]dcnr Dl\'lpatienrs. - The conclusions ol Ihe I)CCT stu(l)' rcprobabil apolicable to the non'insulin dcpcndent dilhctcsrnellitu< r\lDD\Ir pitricnts as u'cll. Scvcral faclors uercrccognize6 a. plly'ing a tole in glyccmic control. iulons*'hiCh ar: duration ol- DIll. obesilr and hr pcncnsion.r-' r'
This oap3: is part ol a populution-b sctl
cpirlemio)ogical studl of chronic ntctabolic dtstrrdcr untonlSludi aciul: subje:ts liotn all rcgions ol Sludi Arabil. Itlrruses o: lhe pattern and factors afl'ectin! glvccntic
control oi non-insulin dependent diabetic patients.
Patients and Methods
The \ational Epidemiological Household Survey forChronic \'letabolic Diseases, which included DM, was
conducted among Saudi subjects over the age of l-5 years
in different regions of Saudi Arabia between 1990 and
1993. This \ras a nal.ional study wiih several objectives,
among them to measurc the Prevalence of Dlvl,
h ypercholesterolem ia and obcsity at national. regional and
rrrbal lerels Based on this, a sample srze of 37.000 u'as
calculated in order to sludy the pattern o[ glycemic controlof Saudi diabetic p3trents- A multistage slratified cluster
r3r':dom sampling technique was used for the selection ofthc studi population. The assigned population sample oIrhc srudS *as distribured between the different regions in
accordanc,: rlrth the regional population distribution as
provided h1 the National Population Council (NPC). There
u'as an initial adjustment for t1'pe of area (urban or rural),and populatron distribution in each region. as per the NPC.
Ciries and rillages ofeach region were ltsted and a random
sclcction of .r certain number of cities and villages
conducled in accordance u'ith the allocated share of each
region rn the national sample. The administrarive nraps oiIhc selccted ctttes and villales wcre revieq'ed lnd a
random selection of a number oi dlstricls in these cities
and villasc\ !\ as made
Primarl' care physicians \\'ho worl in these localitiesu,crc assenrbled and givert tlrientation lecturcs and
rvorkshops on difl-erent aspecls of the studi'. such as fillingout thc forms that include pcrsonal. dtmographic. social
irnd medical historl'. spccificalll dara rclaling to DM. such
il\ prcsent historl oi scll--reportins. Thc nlcdicll rccords ol
llrc\c plucntr '...crc rcYic\r,cd tt, conlirnr the diagnosii ol'
DIr4. Thc primari cirrc nh-vsiciilns uere itlso lrarucd ttn the
propcr mcthod rrl blrrxl hirrdlinS. ilnd nleusurcmcDt ol
hui:hl and rrcighl. lvhich is done al thc Prrnrarv Carc
Clinic (l'CCr. usurlly t\\,o or thrcc drvs alicr the prttcnl'sittrtrul "isltEvcn'third purirllel strect in anv of thcsc ltrulitics u'as
sclccred- und thctr cvcl\ third llouse u'lrs includccl in thissrud) All subjccti ovct thc agc ol l5 1'curs in tlre\c lrouses
rvcrc askcrl to plrticilirlc in lhis slu(l!- All thc intcr\ icu,e(l
sul'riccls rvcre rcqucstcd t() xttcrd llrc PCC lirr rvciglrt.Ircilht and bltxxl rncnsurcrncnts. Rantkrn blcxxl samplcswerc dnrwn in nn IIDTA tutrc. ccntrilirgcd. and lhc scrunrstored liozcn unlrl conrpletion ol lhc t rget s mplc. 1'hcsumplc's u'clc senl li()zcn to the Cenlrul Laktratory. KingSaud Univclsity. Riyildh. from all ()vcr thc countr\,.
Slmples u,ere stu-cd ll -10'C until assayctl. Upolrcuchirs thc targct nurrber of 100 subiccts pcr physiciiln incach district or city. or 50 subiccts pcr physiciun in caclrvillase-. rhc records wcrc scnt ro thc ccntrll oflicc inRiyadh lirr data entry. A computer program was dcsigncd.usin!: DBasc IV soliware package for data enlry. Alicr lhccomplctron of data entry. thcre was a final adjuslmcnt lorgender. age. region, urban or rural population distributron.in accordance with the NPC. using Statpack Gold soltwarcpackage, of a number of records fronr different regions and
age groups.Serum samplcs were used for glucose analysis on a
glucose analyzer (Beckman Paragon, Fullcrton. CA. USA)The method depends on the rate of oxygen consunrption b]the enzymc glucose oxidase. The instrument rvas calibratetlprior to glucosc determination. using qualitl controlsamples provided with thc solutions. The mcan valucsobuined on the controls were within thc values quoted h1
the manufacturer. The results were expressed by SI units(mmol/L). The intra- and inlerassay coefficicnts ol'variation were 1.'7 7o and, 2.69o, respectivclv.
A cutoff of random blo<d glucose (RBS) of l0 mmol/l-was uscd to categorize patients rnro goo<J (RBS<10
mmol/L) and poor glycemic control ([tBS>10 mmol/L).The WHO criteria for defilition of obesity (BMI>30) v.as
adopred.r6There are many variables which are belicved to
rnfluence the glycemic control of diabetic paticnr!i. Thesernclude age, gender. body mass indcx, region. plirce r;lresidencc (urbau or rural), duration oi DM. and treatmcnLmodalities of DM (dret, oral agents and insulin).
Vanous types ol-univariate analyses rvere emploled t(lassess the distribution o[ observations. to providc' an
understanding of thc relationships between variables. unC
lo selecl rhe variables surtable for inclusion in multivarrareanallses. Chi-squared test. Student's l-test, and Fisher'sexact lesl wcre performed to assess the significance of thcrelationships bet,a'cen variables. The 57c lcvel was set s
thc ler cl ol significancc.\{ultiple logistic analysis was uscd lo cxanriDe thc
assoc'iation bctrvccn the odds ol gll'cenric control and anrpretlictrrr.belielc'd to inlluence rt. ildjusling il lbr thc othcrpretlicrors incluclcd in rhc Sclrlcssemln nrodel. I98?.r;
Lolit tp) = 115,*, + b.\, + .... + bt\l
A back\v rd elirnination rv s crnploycd to sclcct lhc bcst,tiuing rrrtrlcl. Thc initi l nrodel inclucled all uain cflcctsThc cll'ccl with thc srn llcst Z-ralro u,as rsnrolcd urrd thcmtxl,.,l rclitted witll rhe renlaiDing cfl'ccts. Thc- unal!sis hld
It0 Aa at..,l \ t.tt ihln ,,.l;'/ /ri. M,:. /q'Jl
proceeded by deleting effects onc at a time until the b€sl-firtin8 model was reached. The adequacl'of fit of a givenmcdel was assessed by both the likelihood ralio resr andthe goodness o[ fit chi-squarc. ln multivariatc analysis.only the obscrvalion that has information on all variablcsincluded in the multivariate equation would bc accountedlbr in the analysis. Hcncc. out of6l3 individuals. only 415$crc includcd in the multivariate analysis.
Srarisrical analyscs were performed bv using StatpackC<,1d.'' and subroutine of the BMDP LR softwareconrputcr package.re
Results
There were 613 sampled subjecrs who were confirmedas diabetic patients. 507c of whom had good glycemiccontrol (Table l). The mean (and standard deviarion) ofage for subjects who had good control of DM were 47.0(1,1.8) years. compared wirh 51.5 (13.8) years for those\{ i'ro had poor conuol (T=3.88. DF=61 l. P=0.0001 ).
There u'as normal age distriburion, hence there was no:l:ri ibr transformation or categorizarion. and the variablerr< u'as used as a continuous one. The dlsrribution of aeerras as follows: median, mean (SD) 49.48.92 (12.8) ),ears.nrode 45 years. standard error 0.48 years. variance 163.E9\ears squares. skewness and kunosis 0.06 and 0.2.
There was no specific difference of means of BMIbettreen good and poor glycemic control patienrs, 28.7(5.4) and 28.3 (5-5), respectively. The prevalence ofol)esrty sas high among the parients of the two groups.39c,t and 349a for good and poor glycemic control patients.respectively. There was. however. no significantdifTerence.
There rvas no specific pattern for mcans of RBS forcach aee eroup of male and female patients rvith gccd andpoor glycemic control There was no significant differencebet\\een means of RBS of male and female parients of eachage group. whether among good or poor glycemic groups.
There was a signiftcant relationship between glycemiccontrol and age (Table l.). and treatmenr modaliries of DM(Table 2). Subjects who had good glycemic control of DMwere younger and following a diet regimen, while thosewho had poor glycemic control were older and on insulintrcatment.
IUultiYariate analysis comprising 415 individuals wasconducted to find out the factors that can potentiallvinfluence or are associated wirh the control of DM. and-15 SrZ rvcrc found to have good glycemic control. Thcinl:ial model ircluded the follou'ing factors; agc. gender.BMl. region. placc ol rcsidence. urban vs rural. durationof DN'I. and trcatment modalities of DM. Thc treatmentmodality of DM was the only lcctor which had srgnificanrassociation with glycemic control. Compared to patients onthe diet rcgimen, those on insulin and oral agent treatmentwerc at 1647r and 657. more nsk o[ having poor glycemrccontrol. respectively (Table 3).
CLYCEMIC CONTROL OII DIAIJETIC PATIENTS
T^Br-r I Co,'tftri$ ttlo,,ornt hn..u,t t hlsi.tt dah t,l snot anttpoo. sl,'.n <o ftnldiul'uit l,otit, s
T^Btt- 2. Conpa.ison oI lifi}rc t luctot' \'hich c a[fcct sbc?Dti.
Clyc.mic conuol
Good control group Poor conlrol groupno ( ) no. (+) I. dF*
Rcgion
NonhSoudrCcnrralEast
Resid€nc)UrbanRural
Sex
MaleFcmal.
Treatmcntmodaliircs
lnsulinOral agcnrDicr
90 (52.3)21 110.',l)1-7 6l.O)55 (46.6)
93 (56.1t
lll (50.6)96 i50 0)
158 (49,1)l5t5l5)
23 (29.r)I l3 (42.8)112 (61.7)
6.2482 (47.?)35 (59,3)5l (53.0)63 (53 4)7l (43.31
0.0220E (49.a,
(50 0)
0.29r62 (50.6)I42 (48.5)
10.77
56 (70.9)r5l (57.2)% (35.8)
2 0.0(x)
018
*Chi-squared: "degree of frcedom
Discussion
This is an epidemiological population-based studyrvhich reflects the partern and factors which can potentially-affect glycemic control of Saudi diabetic patients at thenational leve!. Several obsen'ations can be made withrespect to the glycernic control: the high rate of parientswith poor glycenlic control. which amounts to 507c of thestudy diabetic population. is rather an alarnring rate whichprobably reflects the inadeguacy of services provided.including education. and lack of insight into theimportance of eood glvcemic control. whether from thephysicians. patients or both. lnsulin-rreared diabeticpirticnrs amounl ro l3?. comparcd rvith 43% ar,d 44cl tororal agents and diel. respectivell, ln spite of having a
significantly higlrcr ratc o[ insulin users among the poorglycenlic control group whcn compared rvith soodglycemic control diabetic patients (719i vs. 29clo). the rateof insulin users u'as lower than that of oral agents arnongpaticnls \.\,ith poor glycerric conrrol (187. vs. 507r). Poorglycemic diabcrrc patients were significantly oldcr thangood glycemic control diabetic patients.
Mukiple logistic regression analysis has shown thaitreatment modalities arc the only significanl factors whrchcan be associared with <legree of glycemic control. wherc
,\nnuk t'[ Soudi Lr.licttt. l'ol 13. No 2, l94t lll
't\r'lI. :l /r,!n,i(,1sr(rln'ri'xdl\ritl'rIl( ttrkt\\'lt h tt trtt
r\t.-Nt iAlM I I \i
D\l rrcurnrcnr I .l t--r.96 {r (x)5 hypcncnston.ln\ulin (l.t)7' l'(iil';1.", 0 soj r'(ri I oli-l'-ss
Acknowledgcnrents0-79 (l-57 l.ll 0 lr
(,'n\lx'rl
E-i =r',,rr fi acucc ; n rcn I tci=confidcncc;nrcn'rl' The authQrs acknowledge thc Ptolessional erccllencc ol