computer-aided risk management— a software tooi for the
TRANSCRIPT
»li^Dentistr
Computer-aided risk management—A software tooi for the Hidep modelUno G, H. Fors, DDS, PhDVHans C, H. Sandberg, DDS'
Objective: The Hid9p risk manageiriGnt model has been developed and tested in clinical settings withpromising results, but a tool facilitating the work has been suggested. The aim of the present study was tooréate and evaluate a computerized tool capable ot creating overviews of the oral health situation as wellas identitying risk factors and at-risk patients. The system developed shouid also facilitate the clinicai work,for example, by assisting the user with automatic oalouialion of suitabie Hidep groups and selection andprinting of relevant patient information letters. Method and materials: The system developed was basedon the Hidep model, combining a number of available examination methods, risk estimation systems, andtreatment suggestions. The deveiopment strategy included stepwise improvements and functionalityincrease based on continuous clinical applicabiiity tests in a iarge international test bed. Results: Theresults indicated that the software created was user friendly enough to be used in a common dental clinicand capabie of handling the basic data of both patients and their oral heaith situation. The system oouldpresent useful statistics and graphs describing the overali oral health situation and identifying relevant riskgroups and risk factors, based on virtually unlimited parameter combinations. Conclusion: The computersystem developed seems to be an important step toward the possibility of creating a close-tc-the-clinicmodel for oral heaith care management based on actual and locally derived patient data and risk factors.The results of this project encourage further studies of the Hidep model and its computer support.(Quintessence Int 2001:32:309-320)
Key words: computer support, dental risk management, Hidep model, oral health care planning
CLINICAL RELEVANCE: Computer support might be agood way for obtaining an overview of the oral heaith situ-ation in a clinic and for determining risk levels in thepatient population.
A number of different methods and models for assess-/jLment of dental health and dental risks are avail-able. Many are aimed for research or epidemiology, likedecayed missed filled surfaces (DMFS) and decayedmissed filled teeth (DMFT).' Some others are moreadapted to the clinical situation, like the communityperiodontai index of treatment needs (CPITN, CPP},simplified periodontal examination (FPU^''), commu-nity caries index of treatment needs (CCITN'), perio-dontai screening and recording (PSR ), and the Cario-
'Oeparlment ot Hurnanilies, Informatics and Social Sciences, KaiolinskaInsi Ilute. Stocklioim, Sweden.
Reprint requests: Dr Uno G. H. Fors, Department of Humanities,Infomalics and Social Sciences, Karolinska Institute, PO Box t7913, S-1 t895 Stcckholm, Sweden, E-mail: Uno.ForsShis.ki.se
gram.' However, many of these more clinically orientedmethods or indices still have some problems with link-ing examination results and index valties to individualtreatment procedures and preventive measures. Asomewhat different approach is used in the screeningand management method called the Hidep model(health improvement in dental practice model), whichuses predefined risk groups for selecting and managingindividual treatment and prevention schemes*" "(Forsand Sartdberg, unpublished data, 2000}.
This model has been clinically tested with promis-ing results in Sweden and elsewhere and consists ofidentifying risk factors and their individual impact ona patient's risk of acquiring caries or periodontitis.Most of the initial model development work was per-formed in cooperation with international domainexperts,^- '(Fors and Sandberg, unpublished data, 2000)who formed a scientific reference group used to definethe strategies behind the Hidep model. The model,specifically aimed toward clinical use in general den-tistry, utilizes common clinical examination methodsto help determine which of 9 different disease (caries)or risk (periodontitis) groups a certain patient shouldbe placed in.
Ouintp -Q International 309
• Fors/Sandberg
TABLE t The 14 examination parameters used inthe Hidep model
No.
12
3456
789, 10
11121314
Parameter
Total number of teethTotal number of intact teeth (teetti without restora-tions, caries, or crowns)Number of caries lesions (initiai iesions exciuded)Caries experienceFiuoride expositionSaiiva diagnostics (including secretion, bufferingcapacity, laotobacilii criteria, and Streptococcusmutans). In most patients, only secretion is used.Sugar frequencyOral hygiene screeningProfessionai risk estimation for caries andperiodontitisGingivai bleedingProbing of periodontal pocketsRadiograptiic examinationRegistration of tartar and/or overhangs
After performing this grouping, the Hidep groupdata are used for clinic-based bealth care outcomemeasurements, selecting appropriate treatment andprevention plans, determining individual patient infor-mation schemes, managing the clinic's human andtechnical resources, and more*-' (Fors and Sandberg,unpublisbed data, 2000), However, the data neededfor both the calculation of the different risk groupsand, especially, for the management procedures of theclinic are rather extensive. Because of this, the need ofa facilitating support and analysis tool was evident.Early in the development process of the Hidep model,a computerized support tool was proposed and alsotested in a premature form." The results of those testsindicated that the support tool should preferably becomputerized and should be very user friendly to ful-fill the user's requirements.
Compttter systems for tbe support of dental preven-tion have not been widely used in the past, but somestudies bave been performed with interestingresults.''"^" However, due to the still limited accep-tance of computers in dental clinics and perhaps alsodue to a fear of losing control,^' not many systemshave been aimed for this type of clinical purpose.Another reason for the slow introduction of dentalinformation systems might be tbe limited market thedental profession represents, as well as the complexityof the daily work within a dental practice.
The first computer system supporting tbe Hidepmodel was developed from 1990 to 1991"''' and couldthen only handle manually entered risk codes andsome basic statistical and graphic data output.Altbough the tested system was simple and rough, theresults encouraged development of a more refined and
well laid-out system and a pilot test of tbe applicabilityof tbe final software system.
The present project dealt with two different goals:{1} to develop a refined computer-based support toolto facilitate the work according to the Hidep modeland (2) to perform a pilot test of the clinical applica-bility of the computer system in general dentistry.
METHOD AND MATERIALS
The Hidep model
Tbe Hidep model does not use any new risk estima-tion techniques, but it combines a number of alreadyavailable examination metbods, risk estimation sys-tems, and treatment suggestions into a new entity"(Fors and Sandberg, unpublisbed data, 2000). Themodel is based on 3 steps: (1) clinical examination; [2)analysis of parameters examined; and (3) selection ofan adequate Hidep group for caries and periodontitis,respectively. The examination is proposed to containup to 14 different parameters (Table 1) and is based onscreening techniques wbere possible, with the mainobjective to "sort out" bealtby and low-risk patients,
Tbe strength of the different parameters examinedis estimated according to a set of rules, and the corre-sponding Hidep groups are then determined for eachpatient. The Hidep model contains 5 risk and 4 dis-ease levels for caries and periodontitis, respectively.The different values of these parameters defines wberea patient is situated on the "health-sick" scale, a scalefrom a very low risk of attracting disease to verysevere symptoms of existing disease (Fig 1),
As seen by the Hidep model, a person can either besick or healthy; there are no in betweens. Tbe patientcan be less or more sick (1 to 4} or bave less or tnorerisk of attracting disease (OS to 4S}, where the S indi-cates need of support. For patients without any teeth,the group codes are set to CD (complete dentures),Thus, the model suggests that the groups for caries (C)should be any of the 9 groups COS to C4 (COS, CIS,C2S, C3S, C4S, Cl, C2, C3, and C4), and any of thePOS to P4 groups should be used for periodontitis (P)"(Fors and Sandberg, unpublished data, 2000). For amore detailed explanation of the risk groups for cariesand periodontitis, please refer to the technical adden-dum (see page 315} describing the automated Hidepgroup function of the software system.
System development
The initial computer software was, as mentioned, verylimited in its functionality and consisted more or lessof a center for manual data entry and a rough calcula-
310 Voiume 32, Number 4, 2001
Fors/Sandberg •
Severe symptoms
Sick
Mild symptoms
i-iigh risk
Healthy
Low risit
4
3
2
1
4S
3S
2S
IS
OS
Treatment need
5upport need
Fig 1 The heaith-sick scaie used by the Hidep modei
TABLE 2
Country
SwedenSweden
GermanyEnglandandSweden
Sweden
Germany
Sweden
Test base for the software system used in the Hidep model
Mo. ofclinics
50
25
15
9
1
50
1
Approximatetime used
12 moiamo
3 mo6 mo
1+1+1 mo
12 mo
l O y
Period
19911993-94
19951995-96
1995-97
1996-97
1989-99
Type of setting
Private practitioners
Function tested
Gênerai functionsCommunity dental centers Gênerai, import
Private practitionersPrivate practitioners
Dentai school
Private practitioners
Private practitioners
functions, statisticaifunctionsGênerai functionsGênerai functions,patient informationietters and statistics,reliability.reproducibilityGênerai and education a ifu nef ion sGênerai, patient informa-tion ietters and file transferLong-term management
lion of risk group statistics. However, a number ofsuggestions for further development had been gath-ered during the pilot test of the old system. Based onthese suggestions, it was decided that to be usable in ageneral dental clinic (where computer expertise sel-dom is available), the user interface must be verystraightforward and a minimum of computer literacyshould be required. Also clear was tbe need for agraphic presentation of most data to allow generaldental health care personnel with little or no interestin statistics to use and take advantage of the system.Other suggested improvements were facilitated dataentry, import functions from computerized patientrecord systems, automatic Hidep risk group calcula-tion, refitied and expanded risk group analyses, patientinformation letters, and a possibility for the user toadjust calculations and presentations.
During 1991 to 1999, 10 preliminary versions of thecotnputer system were developed, all with subsequentclitiical user tests where suggestions from users were
gathered and analyzed to serve as a base for tbe nextversion developed. In tbis way, tbe computer programwas developed stepwise, witb continuous improve-ments and functionality increase between all chnicaltrials. Tbis user-driven strategy was cbosen to improvethe clinical acceptance and genera! applicability of tbesystem, Tbe tests were performed in clinical settings inSweden, the United Kingdom, and Germany over dif-ferent periods [Table 2). All test clinics were cbosenaccording to 3 criteria: (1) the clinic should have atleast 1 year of experience witb computerized dentalsystems; (2) the clinic should have at least one dentistand one dental hygienist; and (3) tbe dentai teamssbould bave participated in a course describing tbeHidep model and its use. All clinics were instructed toregularly report possible problems associated witb tbesoftware and/or suggestions for improvements.
Before a new version of tbe system was developed,data from the prior test clinics were analyzed, and a listof priorities was created to find out which changes
311
• Fors/Sanciberg
should be included in the following version. As a ruleof thumb, any suggestion for change that was presentedby more than 50% of the test clinics was prioritized.
Appiicabiiity test of the computer supportfor the Hidep modei
The Hidep model itself was also continuously devel-oped during the test periods between 1991 and 1999.All tests were made for refining and improving hoththe model and the computer support, and therefore,most of them were rather short in time and used dif-ferent versions of the computer system. Due to thisfact, a consistent multiclinic-hased long-term test ofthe usefulness of the system was not possible to per-form. However, a limited apphcability test was per-formed during all clinical trials. This was based on 4different criteria of applicability that included;
1. All clinics must be able to piace all patients thatwere clinically examined during the test period intorisk groups using the software system.
2. They shouid be able to produce detailed diagramsof the C and P groups displaying at least SO /o of thetotal patient population of the clinics.
3. At least 90% of tbe users should find the systemeasy to use (measured by choosing between "easy"or "not easy").
4. At least 80% of the automatically calculated Hidepgroups should be correctly determined as comparedwith manually determined risk groups.
The data needed for this study were collected bymeans of team member surveys, interviews of teammembers, and also by data forms and diagramsprinted out from the software system. The comparisonof the automatic grouping function and the manualchoice of suitable Hidep groups was assisted by a cal-culation form with a description of the general princi-ples for the Hidep group code system"'^ {Fors andSandberg, unpublished data, 2000); the users weretold to enter their own code choice as well as theautomatically calculated Hidep code presented by thesystem. Data from all clinical test beds in Sweden,England, and Germany available between 1995 and1999 were used in the study (Table 2).
RESULTS
Development of computer supportof the Hidep modei
A special computer program, DentiGroup, was devel-oped to handle all data and calculations needed for
the Hidep model. The system was initially developedusing a database programming system namedSuperbase (Software Publishing). To furnish necessaiygraphic functions, graph data functions were calledfrom the Graphics Server (Bits Per Second) system,The current version of the DentiGroup system is basedupon MS-Access 2000 (Microsoft) and GraphicsServer 5.0, enabling a system that is compatible withMS-Windows 95, 98, NT, and 2000. Both the earlyand the recent system versions were made available inSwedish, German, and English and also in both stand-alone and network versions.
The main function of the DentiGroup system is tocollect patient data, calculate automatic proposals ofHidep risk groupings, and provide the user with nec-essary data to get an overview of the cariologic andperiodontal situation in the clinic. The patient recorddata and the relevant clinical findings can be eitherimported from existing administrative or clinicalpatient record systems or manually entered by theuser. The main patient record form, on which all basicdata for the patient are displayed, is presented in Fig2. The system was developed to be able to share andimport data from a large number of administrative sys-tems from Sweden, Great Britain, and Germany aswell as to use normal American Standard Code forInformation Interchange (ASCII) files. In the presentsystem, patient data from more than 15 different sys-tems might be imported, including some of tbe mostpopular electronic patient record systems in Swedenand Germany.
To assist the user in finding accurate Hidep riskgroups, an automatic risk group calculation functionwas created. This so-called AutoGroup form (Fig 3) iswhere clinical data should be entered to serve as thebasis for tbe Hidep group calculation. Note that themore examination data entered, the better the base forcalculation of the Hidep risk groups. The AutoGroupfunction was one of the most demanded functionalilyimprovements from the initial prototype. This iuncdonis based on a set of rules and individual parameterweights. Please refer to the technical addendum (onpage 315) for more information regarding this function.
Apart from calculating a variety of statistical andgraphic overviews of the risk groups of a clinic, the sys-tem is also capable of printouts of patient informationletters and suggestions of treatment plans that corre-spond to the respective Hidep group code, and it canalso compile clinic-wide or group-wide risk factors (Fig4). These are all important features to give the clinicianan overview concerning the overall health situation ofthe patients in a clinic. These and other DentiGroupfunctions may be used for analyzing variations in ocalhealth, risk factors, etc; and variations over time,between groups, or between new and old patients. For
312 Voiume 32, Number 4. 20D1
Fors/Sandberg •
Fig 2 The DentiGfoup system main entry form dis-playirig oommon data ttiat might be entered into (tiesystem.
lea Elit
PS-91IID
] K S I B ,
[Œnn
OhüaN
Rcluid
1Í3«
JHídsp
• • . -
ts
Tí
O:Í
Cgro
CiS
C2S
C2S
iruU'l 1 Nissí.
p P.giouF
P2S
P3S
P2
Pi
^ ^ - ^ ^ ^
ilnEJ
analiíí- Q,
3 Calí
|m« r
bisii>
Ho, orieethlNo23
22
"^»""»«'-^ 1 iwlTf
Smakefl 2D4tel«| 0íiímij 0 1^£niil| Û
ofM,leelhD2
iihsmi D
DIIKISII D
ll>>»3il 0
Oi
So
•I IWE
Giinip
MI
M|
'Ç]
m
lepdüe
.|S|x
1
1JJ
Fig 3 Ttie DentiGroup system AutoGroup form dis-playing examination data that might be entered into thesysiem Ttie user does not haue to enter aii ciinicaidala to use the AutoGroup function. For example, thesaiiva tactois are noi necessary to enter: ttie systemcar stili caicuiate a suggestion of a suitable ÍHidepgroup. At the upper right, the automatioaliy calcuiatedHidep groups tor caries and periodontitis are displayed(C group and P group, respectiveiy). Note that the pro-posed Hidep group oode can be manuaily adjusted.
f l DenliB'Dup >I
CariesType/no. Ql caire
Caíesenpaience
SaSva secreJionB Jfe< capÄily
Laembscift cgunl
S « u u n ! oï in l
Dialay ndftcFkjoiide e>Tiosi(nir
PJsoiie sÍLasTion.Piof. CaiB! na
L3 .
dsi;
Me
,Aul<iG
tee* 1 23
feïTons. Inibd
rrrra(^rr^
PeriodontiltsPlaaue oliHliBi.Gingival ilalur
PoaiíT d iWíSí
SoneliK!.
Talanl iM:
Piol Pciiod nsc:
rr(T
(T
r
r u»'^ .10í" .6'' 1,000f" i.ooo^ 0,1r" High
r None
r ü
r Nonel~ OOKr Noner Nomf Honsr 0
Dup Iwm
j Pat
Wr,o(^a=ltrT-
! 2 M'f in/su
'^ Aveiager o7,¡
f 10.000
i " lO.OOOí"2-3r' NDimd
f~ ^30Ür 1
<" <2flS
r lo^.ajs35 C [lOÄ 3.5-5
r nosiv:?f SimaC 1
• .Mi l .
rrrIT
rrrfT
rr
í-
r5 r
rrr
|55l.-|017S
upITi—
M/ioc f Ö "
High
lOOflW
lOÚ.OOO3.4
ätß.5t«
ÎQ\,5OÏ
ÍCIÜ.50S
ilOÄ 3.5-5,5
>lOSí1/3
Suprai'Stiigr»
2
P-S
U
rrrrcrr-(T
f
s(T
rfí í
r
' ^ ResülraioraäL.
,'alHicäl \ T
Veiíhish
íOStJ
500,00G
i.ooaooo>Ahotie
150%
3 rñT
>6OÏ -
)5f)ï
. 1 ( | : Ï ) 5 5
>15ïftifC/veilBsiaíltubg>t
3 1 273 -
,;8/M)|
!1
, 1 jjj i 11 1 1
k l 1
Ujldo Prml
alienl ^ ^ ^ j
.^ \— 11 |(= 1 -1 11 _ l J
. 1
Divided Carfes Diagram1a4A aolMJiuii ei*«ñt Palivil nlKTw aimw
* !,J * 'iV 19* "'5, : f '35.39°"Vi.™ 'sV ¡t ^¡5- BS* '
B e i
flcisD CIS
76-
TuUlwiill
UpiUMn
Ulliny
Du« h«
HUE« On
Figs "la and 4b Two DentiGroup system presentations displaying the distribution of the caries riskgroups for the patients of a hypothetioai clinic (lett) and an exampie ot ttie statistical functions of thatdrspiay data over Ihe ditterent Hidep groups clinic (right). Any combination of risk factors might bechecked and determined.
313
• Fors/Sandberc
Risk Group Diagrcim
GrndFr
Altlina
S Irak«,
Metlcir»
Ciabelei
r
''c
Huns
hcms
heiilt
Neing
r
r
' •
' •
Mal;
Aiflina
Smol
M M 1
Dial
(•-•
Sn
' •
r
<
l-en«*
u f f | ^
s™
MM
Un,;
r Cl
' s
<' M
r- D
Snufl Qllie3|f^
0 | I . B 4 ^
0 l l » l 6 ^
kar>
" " -
Nene
None
Um^
Nun«
(T
r
rr
c
UliithZofi
r?r3
r 2r3r 2r 3
#
TcHArsajrcle
Ujxise rïs
O» f 1'
C0ËyP(ISC1S(PlSC2SJPÎSC3BIP3BClS^lS C l lp l C2IP2 C3fl=3 C4JPÍ
Fig 5 Example oí ttie input system tor selecting different criteria for calculationof statistics or graptis. Here is ttie BisK Diagram displayed tor female patientswith oerlain nsk factors, including smoking and diabetes.
example, in Fig 4a, 3 patients with very high cariesactivity (code C4) may be identified in the age groups25 to 29, 30 to 34, and 50 to 54; it might be importantto clinicaily follow these patients. Furthermore, in agegroups 35 to 39 and 45 to 49, there are 2 patients viithvery high caries risk (C4S), who also might be impor-tant to clinically follow. However, most patients in theclinic have a low risk for caries, indicated by the CISgroup. Compilations like in Fig 4b might be useful indetermining underlying factors of oral health problems.As can be seen, 49 out of 97 patients have gingivitis(P2), and subsequent risk factor analysis might reveal ifany factor is the main reason for this, as well as iden-tify the actual patients. It is possibie to get specificdetails, like identifying the actual patients and theirrespective examination results and risk factors, in all ofthe graphs and tables presented.
The computer system might also be used to deter-mine and analyze subgroups of patients with certainrisk factors and their impact on the oral situation. Anexample of the input system for these functions isshown in Fig 5. Also, individual patients might beidentified and listed by the system for follow-up stud-ies or similar. The system contains several other statis-tical functions usable for work according to the Hidepmodel, such as health economics, analyses of generalrisk factors with importance to the oral health (smok-ing, diabetes, medicines, etc), and also numerotis func-tions for identifying individual patients with certainrisk groups or intraindividual follow-ups. As a matterof fact, the system is capable of calculating virtuallyany type of statistical or graphic presentation of any ofthe variables of the system in any combination. In thisway, for example, a periodontal diagram of all male
smokers of the clinic treated with a certain procedure,a graph of all female patients with a certain medica-tion and low salivary secretion, or statistics for all newpatients of the last year can be presented. Accordingto the Hidep model, all these functions may be usedfor managing preventive measures and treatments "'"'(Fors and Sandberg, unpublished data, 2000). Otheranalyses or functions that are instantly available fromthe menu system include individual treatment plansfor caries and periodontitis, automatic suggestiotis forrelevant preventive measures, policy letters, patientinformation letters, automatic suggestions of recallintervals and recall letters, health economy, statisticsover tooth loss causes, risk group statistics, age groupstatistics, caries and periodontal diagrams divided byoverview or age group, and risk group diagrams.
An interesting function that has been found usefulin the clinical tests is the capability of the system toautomatically suggest and print patient informationletters, individualized to suit the specific needs of apatient. More than 15 different letters are available,including fluoride, caries, gingivitis, periodontitis, oralbacteria, and tartar. This so-called print-on-demandfunction can also produce suggestions of individualpreventive measure plans and treatment plans forcaries and periodontitis. These functions are automaticand are steered by the actual oral examination data forthe patient and the corresponding Hidep groups (Figs6a and 6b), The system also enables the user to freelyedit all information letters and treatment plans to fitthe needs and possibilities of the local clinic.
The computer system developed seemed to executeall requested functions well for facilitating the clinicalwork according to the Hidep model.
314 Volume 32, Number 4, 2001
Fors/Sandberc
Crudfll inform
fimnEiDun diËD«
GkitfvUlicanbrilcPslDátndUf If d«fi
^ - - - - ^
' • ' i ;
_ J L—I •! •
1
Haliti; laxh
•tíai
nidiltd u
l l L t l ?
IhCD
Th
Itki
IhIhiluiptiff
ih
SoAlduID
m
roriho
«nmBunm
whar dHi4
EnnalheT
utrucKd bu
aiui The aihr caLIunoth TfLibpngivaL ppock it doeearner, B
e of jou who
uoiL cfihf ^ins
lOfi
wanl to ktep 1 gSB W QM U
v;tL,chfl.pp.rt.«.dr!Li,n(heE«a.
hnlthy mouth look Uke?
will n n d TbisiiDSiFvl of cmdlcHidlendc) n enn^vd n nol I E ^
fa polo neck iwghp bílwcín Ehe
dtrl When ihcnot ncecd 3 nui
0 ihEiE 11 no brilhy sidF, Ihcbxighiaf diEjuid [fach a
nlBïcau»>VUr OWD D
tODil f l l Q
UL, i n CDDi
C h i m a n . WL
« UDtll ID lootmxk c i ab r duly nbacEoLB Lf Ehe»
ícl wilh (he «IdnoDce rhii Aid
QIDlluIO how Ùa ]«WbD]IC11
L mefe tn tcHh m die hoinvfiTd wol a aim layei calledhed ic &e teelh, iT acls morr —Uf 1 ind EUrroundi Ehe nedi of
fLgiVB IS hralthy ihc iLeplh ofm i ibe gEn iTa EI U ^ I ird m
iedmB I LTD louEhrd k Uuiw boat (the drplh oEibE hole)
cm d i cnnb of thi l « t h T h ït i« i« ia HE ED (he aedi atibrwt, thfû tbe body^ iLel«i4:p
irai la piüipci tinif A heallbv "
- - 1 -rFig 6a Example ol a palient information ieiter.
MutivaDOD iiLd Btfoi
QfL^jJ nldr» BOP RiECitiOn minivEiPothpt dipth probing Taaih niobihly
Oral Hyffmt InitnicXoni
Plaque ••durinn ;' 50% X
1 vxnact, tndodoKics F ^
ProfttsionBl Toalh
-I .-r:Fig 6b Exampie ol a Irealmenl pian fo' P grouf
Applicabiiity of the computer support system
Seventy-six different dental clinics were included inthe applicability test of the system. The resultsobtained showed that all chnics were able to place allpatients that were clinically examined during the testperiod into risk groups. The users were also capable ofworking with the corresponding Hidep group data inthe software system to create demanded statistics andgraphs. The data comprised both new patients andrecall patients, resulting in a base of more than 50,000patients. All 76 clinics were also able to create age-grouped diagrams of the C and P groups {similar toFig 4a) using the software for all examined patients.This graphic overview of tbe actual oral health situa-tion was discussed with all clinicians, who in turncould accurately point out the patient groups thatmost needed preventive measures and care, indicatinga clinical relevance.
A general finding was that the computer systemgave a good overview of the oral bealth situation ofthe respective clinic, something all clinicians foundvery positive. A common opinion was that this couldnot have been done without tbe computer-supporttool. A few users even mentioned that the computer-ized system sometimes revealed a different actual oralhealth situation of the clinic than what was believedby the clinician.
Even though some users reported minor problemswhen using the software, 91"/o of the users found thesystem easy to use. Furthermore, the automatic calcu-lation of the Hidep groups was rated positively, and inmost cases, 90% agreed with the manual calculations.
The possible use of the system for management of
oral health risks and preventive measures according tothe Hidep model was also measured by means of thepossibility to use tbe data obtained from theDentiGroup program to display tbe oral bealth situa-tion and to steer the preventive measures to tbepatients who need them most. This was performed inall test clinics; data from one typical clinic, wbere tbeoverall oral bealth situation could be improved overtime for many of the patients, are shown in Fig 7, Notethe possibility to determine increased periodontalhealth as reflected in lowered P-index values between1994 and 1998, especially noticeable in tbe age groups45 and older. Also, the mean number of teeth perpatient increased in some age groups from 1994 to1998, Similar results were obtained in most other clin-ics, indicating that the Hidep model and its computersupport might be a good tool in assisting the manage-ment of preventive measures in a local dental clinic,
Altbough these applicability tests were limited, theresults indicate that the computer system might be ofgood use for the general clinician in finding out the oralbealth situation of the local patient population, andalso for educating the patients about their oral health.
Technical addendum
The automatic calculation of the Hidep groups isbased on a numeric synthesis of tbe diflerent opinionsof several domain experts in the cariologic and perio-dontai field. However, since there are no objectiverules to use, the DentiGroup system is only proposinga certain value. The user can always change this value.Also, the base of the AutoGroup function might bechanged by the user in accordance with experience
315
• Fors/Sandberg
Figs 7a to 7e Exampie of diagrams displaying tf>e ciianges in a periodontai situation in one dimefrom Í994to 1998.
• • No, of teeth
o o p-irdeii
• • No of intact taetli
0 15 20 25 30 35 40 45 50 55 60 65 70 75
Age (yl
• • No, of teeth
!• O P-irfle«
• n No, of intact teeth
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75
b *ge (V)
• • No. of teeth
o o p-jrdex
E • No. of intact teetli
15 20 25 30 35 40 45 50 55 50 65 70 75
Age |y¡
• • No. of teeth
=• o p^^äex
u • No, of intact teetli
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75
• • No, of teetli° o P-indexn n No, ot irtaot teeth
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75
Age (y)
316 Vciume32, Number 4, 2001
Poís/Sandberg •
and the progress of science. For the interested reader,the weights and calculations used by default in thecurrent system are displayed in Table 3, and the logicused when automatically calculating tbe risk groups isshown in Table 4,
DISCUSSION
The present study was aimed toward two differentgoals. The first goal was to try to develop a refinedcomputer-based support tool to facilitate the workaccording to the Hidep model, and the second goalwas to perform a limited test of the applicability of thecomputer system in general clinical dentistry.
The computer system developed was aimed to beboth user-friendly enough to be used by the evet fdayclinician and at the same time advanced enough to beused as an aid for oral health management for generaldental clinics. The basis of the system is the recentlydescribed Hidep model, a close-to-the-clinic risk man-agement system aimed for the general dental clinic"-'^(Fors and Sandberg, unpublished data, 2000), Eventhougb there might be other suitable systems for mea-suring and managing oral heahh, the Hidep modelseems to have a clinical potential, as indicated in theclinical studies performed. Whether a computerized sys-tem is the best way of facilitating the work according tothe Hidep model migbt be discussed; however, otherstudies have reported interesting results when usingsoftware systems in dental prevention,'^"-" and it wastherefore decided to be tested. The computer systemused in this study has been continuously improved dur-ing the developmental process, and although severalversions were tested, it seems to accurately perform themain tasks demanded, like managing basic patientrecord data, importing such data from other systems,calculating Hidep risk groups automatically, presentingand printing patient information, and performing a vari-ed of risk factor and risk group statistics and graphs.
A common drawback of computerized systems isthat the data input procedures might be complicatedand time consuming. In the DentiGroup system, theadministration of the basic data needed to work withthe Hidep model is improved by means of tbe auto-tnatic patient data import functions and also by theautomatic calculation of the Hidep groups. However, itmust again be stressed that all calculations of groupdata are only proposals, and the user might at any timechange the groups proposed by the system to better suitthe actual case. It must also be noted that the systemallows the user to change the strength, or weight, of ailvariables in the automatic grouping function, resultingin the possibility to adapt the system to the clinicalknowledge or belief of the local user (Tables 3 and 4).
The calculations of statistics, graphs, treatmentplans, and information letters are also functionsintended to facilitate the clinical work. The programautomatically carries out all these functions withinseconds, grouping patients in suitable risk groups, andgiving the dentist or dental hygienist a completelynew way to deal with dental risk factors and the pos-sibility to steer preventive measures. It also includespossibilities for local follow-ups and quality assur-ance. Whether this means that the everyday manage-ment of preventive measures for caries and periodon-titis really is facilitated for the general dental clinicremains to be studied. However, the easy and promptavailability of clinical overviews and the capability toidentify individual pafients and risk factors indicatesthis possibility.
Interestingly, when displaying the age-grouped dia-grams over the C and P groups for the first time, mostciinicians reacted witb astonishment; they had neverthought that the actual oral health situation of theirclinic was that good (or that bad). This is really thekey feature of this type of software system-to give anoverview, to reveal hidden oral heath risks, and toidentify the groups of pafients with the most promi-nent need for care. These features and findings indi-cate where the use of computers could be of the great-est importance in dentistry, namely as a complementto our usual manual procedures and human memoryfunctions. Computers can support us witb difficulttasks like storing and calculating large amounts ofdata tbat a buman being normally does not want tospend time with or would have problems with.
The capability of tbe system to calculate virtuallyany kind of grapb or statistical presentation using anyvariable combination furthermore strengthens ourinterpretation of the system being ciinically applicableand useful to the general clinician. A system with onlypredetermined calculation and reporting capacities isusually deemed not so useful by the general cliniciandue to the limitations and the lack of freedom; a sys-tems developer can never tell what parameters theactual clinician will need to analyze.
Due to the stepwise development method, the pres-ent study could not be based on standardized testswith the same software version. Sfill, according to thelarge clinical test in 3 countries covering more than50,000 pafients, the system seems to be applicable forboth smaller and larger clinics in both private andcommunity settings. This was shown by the capabilityof all users to enter basic data, create automatedHidep groupings, and calculate various stafistical andgraphic presentafions. Although some users reportedminor problems when using the system, most of thesewere related to either general computer skill problemsor were solved in updated versions of the software.
317
Fors/Sandberg
TABLE 3 Weights by the AutoGroup system of DentiGroupby default
Parameter Default value'
Saiiva secretionCaries experienceButler capacityLaclobacilti counXS mutans oountDietary indexPlaque indexFluoride expositionProfessional caries riskestimationGingivai indexBone levei reductionPocket depthTartarProfessional periodontairisk estimation
4955
19
22
19
19
19
40
19
29
30
19
19
19
" Valje used it one is rot erlerefl by (lie userweigtits.
Level 1
0
0
0
0
0
0
0
0
0
0
0
0
00
Level 2
50
20
20
20
20
20
20
1
20
30
40
20
20
20
Level 3
80
60
40
30
30
30
40
60
50
50
70
40
50
50
Level 4
120
140
80
SO
50
50
60
100
120
701007080
120
Note: The individual usei might cliange all
TABLE 4 Calculations by the AutoGroup system fordetermining different group levels of caries (C group) andperiodontitis (P group)
CariesIf number of atypical caries lesions > 2If number of aii caries lesions = 1If number ol aii caries lesions = 2 or 3If number of ali caries lesions - 4If number of ali caries lesions > 5If sum of ali caries nsk parameters = 0-180If sum of ali caries nsk parameters = 181-330If sum of ali caries risk parameters = 331—130If sum of ali caries risk parameters = 431-569if sum of a[i caries risk parameters > 570if number of teeth = 0
PeriodontitisIf sum of ail periodonfai fisk parameters = 0If bone levei = 0 and GI - 0 and sum of otherperiodontai parameters > 0If bone levei < 40 and GI = 0If bone level < 70 and GI = 0If bone level < 100 and GI = 0If bone level = 0 and GI > 0if bone ievel < 40 and GI > 0if bone ievel < 70 and GI > 0If bone ievel < 100 and GI > 0If number of leeth - 0
GI = gingival inden.
C groupC4
C1
C2
C3
C4
COS
C I S
C2S
C3SC4SCD
P groupPOSPIS
P2SP3SP4SPIP2P3P4CD
As mentioned before, the data available only indi-
cate that the DentiGroup system might be of real
clinical value in managing preventive measures; nev-
ertheless, the results are pointing out that this might
be an important step toward the possibility of creat-
ing a real close-to-the-clinic model for risk manage-ment in general dentistry. Future scientific clinicalstudies are planned to investigate the possibilities ofboth the Hidep model and the computer supporttool.
318 Volume 32, Number 4, 2001
Fors/Sandberg •
The applicability of the computer system in generalilinical dentistry was further investigated in a long-erm test spanning 10 years in one specific clinic,j/here it was concluded that the Hidep system wasible to support the oral health management in that:iinic during the time allotted. Interestingly, that spe-cific clinic has been using the DentiGroup softwaresystem for S of those 10 years for all of their workaccording to the Hidep model (the first 2 years servedas a startup period), which also supports our beliefthat this computer system is of practical use in themanagement of general dentistry.
The programming tools and operating system cho-sen might also be discussed, but seeing the MS-Windows environment as the platform of choice is nothard; most dental systems are written today for MS-Windows or are under conversion from DOS to MS-Windows, leaving DOS, Apple OS, OS/2, or UNIXout of the question. Tbe change from Superbase toMS-Access 2000 was decided when Superbase tem-porarily went off the market and simultaneously theuser requirements of enabling totally free editing ofthe information leaflets and treatment plans camethrough. Other systems like Oracle, 4th Dimension, orothers would have been possible to use, but since themainstream seemed to go to MS-Access 2000 with itscapability of creating both secure and reliable systems,that program was used. A very interesting opportunitywith MS-Access is also the possibility to create Web-based interactive systems, something that might beinteresting for the future of dentistry. As a matter offact, a very limited version of the cariologic part of theDentiGroup system was developed for the Web in1996 and is still running as a demonstration examplefor the International Health Care Foundation.
When comparing the DentiGroup system withother possibilities, there are, to our knowledge, noother systems available supporting the Hidep model;however, there are a number of other computerizedsystems for risk assessment like the CCITN^ and theCariogram,' The Cariogram system, which initially wasonly a model describing risks, is now available as asoftware system that has gained increasing popularity.This system is good in assessing the cariologic risk fac-tors, and the DentiGroup system might in future ver-sions allow the user to choose whether to use the cur-rent built-in function or the Cariogram model forassessing the cariologic risks in the Hidep grouping.Possibly, an advanced electronic dental record systemwould also be capable of calculating disease groups fortaries and periodontitis. Even if this is not so com-mon, at least one example exists in the Apollonia sys-tem (SEMA Group) used in some Swedish communityclinics. DMFS for caries and periodontal disease lev-els can be automatically calculated based upon the
clinical findings and used in DentiGroup as a basis inthe AutoGroup function for determining the Hidepgroups. However, these calculations are not based onrisks or risk factors, but on the present level of disease.
Internationally, a number of other electronic dentalpatient record systems have been introduced, butalthough some of these systems are capable of record-ing both present and historic cariologic and periodon-tal status, they have not been proven to be able to dealwith risk groupings and management. As we see it, theDentiGroup system, developed in its initial form asearly as 1990, is one of the first systems ever to be ableto handle this. Although some other systems havebeen described, the combination of a elose-to-the-clinic risk grouping system (the Hidep model) and adedicated computer-support system (DentiGroup)seems to be rather unique and promising. Althoughsome raise eoneerns regarding the possibility for gov-ernmental authorities or insurance companies to findan easy way to control us through computer systems, 'we instead think of this solution as a way toward realfreedom to choose the best level of treatment and pre-vention for all of our patients as judged in both ahealth as well as an economic aspect.
CONCLUSIONS
1. A computer-based tool supporting oral health caremanagement according to the Hidep model hasbeen developed and tested in different clinical set-tings, with good results in both private clinics andin community dentistry.
2. The study covered a large international clinical testbed incorporating more than 50,000 patients.
3. The results indicated that the software system isuser friendly enough to be used in a general dentalclinic and is capable of handling the oral healthdata of the patients and of presenting numeroustypes of important statistics and graphs on demand.
4. The computer system developed seems to be animportant step toward the possibility of creating areal, close-to-the-clinic model for risk management.
5. The results of this project encourage hirther devel-opment of both the Hidep model and its computersupport.
ACKNOWLEDGMENTS
Thi,î studs' wa.s iuppurted by Karolinska Institute (Sweden),Zahnarztekammcr Nordrhein (Gci-many), Iniernational Health CareFoundation (Liechtenstein], and Hidep (Sweden).
319
• Fors/Sandberç
REFERENCES
1. Klein H, Palmer CE, liramer M. Studies on denial caries: Useof normal probability i urve for expressing age distribution oferuption of permanent teeth. Growth 1937 ;1 385-394.
2. Ainamo J, Barmes D, Beagrie G, Cutress T, Martin J, Sardo-lntirri |, Development of the World Health Organization(WHO] community periodontal index of treatment needs(CPITN), Int Dent | 1982;32:281-291.
3. Jacobsson L, Jaeobsson L, How to discover periodontitis intime [in Swedish]. Tandlakartidningen 1979;23:1412-1421.
4. Ericsson L, Simplified periodontal treatment-A treatmentmodel |in Swedish]. Tandlakartidningen 1983;75(3]:80-8ö.
5. Axelsson P, Paulander J, Tollskog G. A new computer aidedanalytical oral epidemiological system ¡abstract]. Presentedat the Second International Conference on Preventive Den-tistry and Epidemiology, 1990.
6. Nasi ]H, Background ta, and implementation of, the perio-dontal screening and recording (PSR) procedure in theUSA. Int Dent ] 1994;44(siippl lj.585-588.
7. Brattliall D, Dental caries: intervened-interrupted-inter-preted. Concluding remarks and eariography. Eur | Oral Sei1996;104(4[pt 21):486-491,
8. Sandberg H. Total quality management by professional den-tal prevention. Eur Dent 1994;3:41-42.
9. Sandberg H, Fors U. Total quality management by profes-sional dental prev en ti on-The patient as a team member,Eur Dent 1994;3:55-58.
10. Sandberg H, Fors U. Management of preventive treatmentin a modem dental practice. Eur Dent 1994;3:32-35.
11. Sandberg H. Fors U. The HIDEF model: A tool for effectivemanagement of a modern dental practice. In: Ming Tay W,Lyncb E (eds). General Dental Treatment. Edinburgh:Churchill Livingstone, 1995:5.1.2.01-5.1.2.11.
12. Sandberg H, Fors U. The Risk-Level Group Model |inSwedish]. Stockholm: LIC Forlag/Investodont, 1990.
13. Sandberg H, Fors U. Dental health care planning. A modelused to group patients according to the incidence and sever-ity of decay and periodontitis-A simple aid [in Swedish]Tandlakartidningen 1991;83:684, 686-691.
14. Fors U, Sandberg H. Computer support for management ofpreventive dentistry. Eur Dent 1994;3;18-20.
15. Eisner J. The electronic patient record; New opportunities fordecision support and quality assurance. In: Preston ] (ed).Computers in Clinical Dentistry. Chicago: Quintessence,1991:186-198,
16. Marthaler TM, Bandi A, Steiner M. Computers in preven-tive dentistry. Int Dent J 1980;30:20!-213.
17 Meadow D, Rosenthal M A corporation-based computer-ized preventive dentistry program. J Am Dent Assoc 1983;106:467-470.
18. Gbirlanda C, Tiberti S, Ottolenghi L, Siciliani G, New soft-ware for a prevention program in orodental pathology |inItalian]. Prev Assist Dent 1990;16(6]:4-9.
19. Crall J, Informatics and issues related to assessing, assuringand improving quality of care. In: Abbey LM, ZimmermanJL (eds). Dental Informatics. Berlin: Springer, 1992:249-258.
20. Fetersen PE, Johansen KS, ORATEL: Telematic system forquality assurance in oral health care (AIM proieet A2029).Comp Meth Prog Biomed 1994;45(l-2];14l-143.
21. Sally ]], Zimmerman JL, Ball MJ. Dental informatics:Strategic issues for the dental profession. In: Rienhoff 0,Lindberg DAB (eds). Lecture Notes in Medical Informatics.Berlin: Springer, 1990.
UCLA Center for Esthetic DentistryUCLA CENTER FOR ESTHETIC DENTISTRY announces theCENTER FOR ESTHETIC DENTAL DESIGN. Next programstarts July 1, 2001. This laboratory training facility, (openedJuly 1, 2000), has 1 - and 2-year programs featuring:
• Specific training in esthetic laboratory procedures• Hands-on patient experience with all-ceramic
systems• Esthetic metal-ceramic systems• New polymer fiber-based restorative systems• Courses taught by Dr. Ed McLaren, DDS,
Joe Weisz, CDT, and adjunct facultyMark Anderson, CDT, and Vincent DeVaud, MDT
MastershipTraining
for LaboratoryTechniciansFor more information,
contact UCLA ContinuingEducation at
310-206-8388 or Dr. EdMcLaren, Director, UCLA
Center for Esthetic Dentistry,at 310-794-4858
Additional Courses for Technicians throughout the year,next available course;
Advanced Anterior Ceramic WorkshopMarch 1 -4 Dr. Ed McLaren Joe Weisz CDT
Call UCLA Continuing Education at:(310)206-8388