Diabetes & Metabolism 37 (2011) 144151

Original article

Validation of the Finnish diabetes risk score (FINDRISC) questionnaire for

screening for undiagnosed type 2 diabetes, dysglycaemia and the metabolic
syndrome in Greece
K. Makrilakis a, , S. Liatis a , S. Grammatikou a , D. Perrea b , C. Stathi a , P. Tsiligros a ,
N. Katsilambros a,b
a First Department of Propaedeutic Medicine, Athens University Medical School, Laiko General Hospital, 17, Ag. Thoma street, 11527 Athens, Greece
b Laboratory for Experimental Surgery and Surgical Research, Christeas Hall, University of Athens Medical School, Athens, Greece

Received 24 June 2010; received in revised form 6 September 2010; accepted 7 September 2010
Available online 7 December 2010

Abstract
Aim. The present study aimed to validate the Finnish Type 2 Diabetes Risk Score (FINDRISC) questionnaire for its ability to predict the
presence of any glucose homoeostasis abnormalities and the metabolic syndrome (MetS) in the Greek population.
Methods. Validation was performed on a sample of individuals who had agreed to participate in a screening program for type 2 diabetes
(T2D) prevention (the Greek part of the DEPLAN study), using both FINDRISC and oral glucose tolerance tests (OGTT). Impaired fasting
glucose (IFG) was defined as a fasting plasma glucose level of 6.16.9 mmol/L, and impaired glucose tolerance (IGT) as a 2-h plasma glucose of
7.811.0 mmol/L. The predictive value of the FINDRISC was cross-sectionally evaluated using the area under the receiver operating characteristic
(AUROC) curve method.
Results. A total of 869 individuals (379 men, aged 56.2 10.8 years) were screened from the general population living in the city and suburbs
of Athens. OGTT revealed the presence of unknown diabetes in 94 cases (10.8%), IFG in 85 (9.8%) and IGT in 109 (12.6%). The sensitivity of a
FINDRISC score greater or equal to 15 (45% of the population) to predict unknown diabetes was 81.9% and its specificity was 59.7%. The AUROC
curve for detecting unknown diabetes was 0.724 (95% CI: 0.6770.770). For any dysglycaemia, the AUROC curve was 0.716 (0.6800.752) while,
for detection of the MetS, it was 0.733 (0.6990.767).
Conclusion. The FINDRISC questionnaire performed well as a screening tool for the cross-sectional detection of unknown diabetes, IFG, IGT
and the MetS in the Greek population.
2010 Elsevier Masson SAS. All rights reserved.

Keywords: Diabetes screening; FINDRISC questionnaire; Diabetes prevention; Diagnosis; Type 2 diabetes; Impaired fasting glucose; Impaired glucose tolerance;
Greece

Rsum
Validation du questionnaire finlandais calculant un score de risque de diabte (FINDRISC) pour le dpistage du diabte de type 2, des anomalies
de la glycorgulation et du syndrome mtabolique en Grce.
But. La prsente tude avait pour but de valider le questionnaire finlandais FINDRISC en termes de prvision des anomalies de la glycorgulation
et du syndrome mtabolique dans la population grecque.
Mthodes. La validation a t ralise partir dun chantillon dindividus qui avaient accept de participer un programme de dpistage
dans le cadre de la prvention du diabte type 2 (DT2) (participation grecque ltude DEPLAN), en utilisant le score FINDRISC ainsi quune
hyperglycmie provoque par voie orale (HGPO). Lhyperglycmie modre jeun (IFG) a t dfinie par une glycmie jeun comprise entre
6,1 et 6,9 mmol/L et lintolrance au glucose (IGT) par une glycmie deux heures aprs glucose comprise entre 7,8 et 11,0 mmol/L. La valeur
prdictive du score FINDRISC a t value transversalement, en utilisant les caractristiques opratoires de la surface sous courbe (AUROC).

Corresponding author. Tel.: +30 213 2061061; fax: +30 210 7791839.
E-mail address: kmakrila@med.uoa.gr (K. Makrilakis).

1262-3636/$ see front matter 2010 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.diabet.2010.09.006
 K. Makrilakis et al. / Diabetes & Metabolism 37 (2011) 144151 145

Rsultats. Un total de 869 sujets (379 hommes, gs de 56,2 10,8 ans) a t slectionn parmi la population gnrale de la ville et des environs
dAthnes. LHGPO a rvl un diabte mconnu chez 94 sujets (10,8 %), une IFG chez 85 sujets (9,8 %) et une IGT chez 109 sujets (12,6 %).
La sensibilit du score FINDRISC suprieur ou gal 15 (45 % de la population) pour la prdiction dun diabte mconnu tait de 81,9 % et sa
spcificit de 59,7 %. LAUROC pour le dpistage dun diabte mconnu tait de 0,716 (0,6800,752), lAUROC pour la mise en vidence du
syndrome mtabolique de 0,733 (0,6990,767).
Conclusion. Cette tude montre que le questionnaire FINDRISC savre tre un outil performant pour le dpistage du diabte, de lhyperglycmie
modre jeun, de lintolrance au glucose et du syndrome mtabolique dans un chantillon de la population grecque.
2010 Elsevier Masson SAS. Tous droits rservs.

Mots cls : Dpistage du diabte ; Questionnaire FINDRISC ; Diagnostic ; Intolrance au glucose ; Hyperglycmie modre jeun ; Prvention du diabte ; Diabte
de type 2 ; Grce

1. Introduction been validated in different, but mostly Caucasian [12,2227],

The frequency of type 2 diabetes (T2D) is increasing world-
wide to nearly epidemic proportions as the general population
becomes older, less active and more obese. Its future burden is
expected to increase dramatically, with increases in both long-
term morbidity and mortality. Furthermore, it is estimated that
approximately one-third of all people with diabetes may be undi-
agnosed [1]. The economic burden of treating diabetes and its
complications is likewise enormous. For this reason, effective
primary prevention programmes for T2D are urgently needed to
reduce the clinical and economic healthcare burden.
The disease is characterized by a long prediabetic period,
during which impaired fasting glucose (IFG), impaired glucose
tolerance (IGT) and many of the components of the metabolic
syndrome (MetS) interact to finally end with overt clinical dia-
betes [2]. Many studies have shown that interventions at the
prediabetic state using either lifestyle modifications [3,4] or
medications [5,6] can prevent, or at least delay, progression of
the disease. This means that the identification of those at high
risk for T2D is warranted to allow for a more timely implemen-
tation of preventative action aimed at reducing their risk [7].
This has been recently shown to be cost-effective as well [8].
Measuring plasma glucose (either fasting or 2 h after an oral
glucose tolerance test [OGTT]) and glycosylated haemoglobin
(HbA1c ) levels are the recommended methods for screening
the general population [9]. However, these are invasive proce-
dures, costly and time-consuming (especially OGTT) and, thus,
not suitable for mass screening. Furthermore, as they are based
solely on measuring glycaemia, they may diagnose the disease
too late, when complications have already occurred [10].
The MetS is a cluster of metabolic risk factors for cardiovas-
cular disease (CVD) that are associated with insulin resistance.
The presence of the MetS predicts the future development of both
T2D and CVD. Diagnosing the syndrome, however, requires
measuring five clinical and biochemical parameters, whereas
its predictive ability does not appear to surpass other, simpler
risk-assessment tools, such as the Framingham risk score (for
predicting CVD) or simple plasma glucose measurement (for
predicting T2D) [11].
Recently, many attempts have been made to develop simple,
fast, non-invasive and practical screening tools for identify-
ing individuals at high risk of future development of T2D
[1221]. Of these tools, the Finnish Type 2 Diabetes Risk Score
(FINDRISC) questionnaire [12] is widely used, and has already
populations for detecting unknown diabetes, abnormal glucose
tolerance and the MetS.
It is questionable, however, to what extent screening proto-
cols, such as risk-score assessment followed by blood glucose
measurements, that have been developed in one population can
be applied in other countries populations [28]. As the effective-
ness of the FINDRISC questionnaire has not yet been studied in
Greece, the aim of the present study was to validate the question-
naire in a population-based, cross-sectional setting in Greece,
and to study its performance as a screening tool for undetected
T2D, other abnormalities of glucose homoeostasis and the MetS
in middle-aged individuals.
2. Subjects and methods
Validation was performed using the baseline data from a
cross-section of participants in a T2D prevention program in
Greece, using both the FINDRISC questionnaire and OGTT
[29]. The main aim of this EU-funded project (Diabetes in
EuropePrevention using Lifestyle, Physical Activity and Nutri-
tional Intervention, DEPLAN) was to establish a model for the
efficient identification of community-living individuals at high
risk of T2D in EU member countries, as well as to test the fea-
sibility and cost-effectiveness of translating the interventional
concepts learned from prevention trials to the currently used
healthcare systems [30].
The project was approved by the participating hospitals
ethics committee and the National Drug Organization of Greece.
All participants gave their informed consent according to the
general recommendations of the Declaration of Helsinki.
The study has been described in detail elsewhere [29]. In
brief, the FINDRISC questionnaire was distributed to around
7900 individuals, aged 3575 years, with no known diabetes
and residing in the metropolitan areas of Athens. The question-
naire comprises eight items: age; body mass index (BMI); waist
circumference; physical activity; dietary consumption of fruits,
vegetables and berries; use of antihypertensive medications;
history of high blood glucose; and family history of diabetes
(see www.diabetes.fi/english/risktest). The total test score (max-
imum: 26) provides a measure of the probability of developing
T2D, with a score greater or equal to 15 being indicative of a
high probability [12].
Of the 3240 completed questionnaires, 869 respondents
agreed to undergo an OGTT. On the day of the test, the partic-
146 K. Makrilakis et al. / Diabetes & Metabolism 37 (2011) 144151
ipants weight, height, waist circumference and blood pressure
were measured, and their medical histories recorded. Plasma
glucose, total and high-density lipoprotein (HDL) cholesterol,
and triglyceride levels were also measured from fasting blood
samples at a central, accredited university research laboratory,
using enzymatic assays. Low-density lipoprotein (LDL) choles-
terol was calculated using the Friedewald formula [31].
Based on the OGTT results, participants were categorized as
having either normal glucose tolerance (NGT, fasting plasma
glucose less than 6.1 mmol/L and 2-h plasma glucose less than
7.8 mmol/L), impaired fasting glucose (IFG, fasting plasma glu-
cose 6.16.9 mmol/L), impaired glucose tolerance (IGT, 2-h
plasma glucose 7.811.0 mmol/L) or diabetes (fasting plasma
glucose greater or equal to 7.0 mmol/L and/or 2-h plasma glu-
cose greater or equal to 11.1 mmol/L) [32]. The MetS was
defined according to National Cholesterol Education Program
Adult Treatment Panel III (NCEPATP III) criteria [33].
2.1. Statistical analysis
Continuous variables were presented as means 1 standard
deviation (SD), and qualitative variables as absolute and rela-
tive frequencies (%). Comparisons between normally distributed
continuous variables were performed with calculations of Stu-
dents t test. Associations between categorical variables were
tested with the use of contingency tables and calculations of
the Chi2 test. Patients characteristics according to their FIND-
RISC scores were compared by one-way analysis of variance
(ANOVA) for continuous variables. Those with IFG, IGT and
T2D were analyzed all together as a combined dysglycaemia
group. Associations and correlation coefficients between the
FINDRISC and clinical parameters were evaluated by Pearsons
correlation test. The predictive value of FINDRISC to detect
diabetes, dysglycaemia and the MetS was cross-sectionally eval-
uated using the area under the receiver operating characteristic
(AUROC) curve method. Sensitivity was plotted against the y
axis, and false-positive rates (1specificity) against the x axis,
followed by plotting of the AUROC curve. The optimal cutoff
levels used were the peaks of the curve, where the sum of sen-
sitivity and specificity is at maximum. The AUROC curve (for
example, for detection of diabetes) can be regarded as reflecting
the probability that a randomly chosen subject with undiag-
nosed diabetes is more likely to be classified as diabetic than
a randomly chosen subject without diabetes and, therefore, is
indicative of the overall ability of the risk score to correctly
identify those with undiagnosed diabetes. All reported P values
were derived from two-sided tests and compared with a signifi-
cance level of 5%. Data were analyzed using SPSS version 16.0
software (SPSS Inc, Chicago, IL, USA).
3. Results
The demographic, clinical and laboratory characteristics of
our study participants are presented in Table 1. The prevalence
of undiagnosed diabetes (based on OGTT results) was 10.8%,
while that of dysglycaemia (IFG + IGT + T2D) was 33.6%, with
significant differences between genders (higher in men). How-
ever, the prevalence of the MetS (38.4%) did not differ between
men and women.
Of the total 869-screened participants, 840 had complete
FINDRISC data, and a marked increase in the prevalence of
undiagnosed diabetes and dysglycaemia was observed as FIND-
RISC score values increased (Table 2). The prevalence of
unknown diabetes in those with a FINDRISC score greater than
15 (44.5% of the study population) was 19.3% whereas, for any
degree of dysglycaemia (IFG + IGT + T2D), the prevalence was
51.1%.
The AUROC curve for detecting unknown diabetes (Fig. 1A)
was 0.724 (95% CI: 0.6770.770) overall, with 0.731
(0.6670.794) for men and 0.726 (0.6590.793) for women (NS,
not significant). The AUROC curve for any degree of dysgly-
caemia was 0.716 (0.680.752), with 0.721 (0.6680.724) for
men and 0.720 (0.6710.770) for women (NS; Fig. 1B).
The optimal cutoff value for detecting unknown diabetes was
a FINDRISC greater or equal to 15, at which the sensitivity was
81.1% and specificity was 59.8% (Table 3). With a FINDRISC
greater or equal to 10 (73% of the population), the sensitivity
was 96.7% and the specificity was 29.5%, whereas with a score
greater or equal to 7 (93.5% of the population), the sensitivity
was 100%, but its specificity was only 10.7%.
The best cutoff value for detecting any dysglycaemia was
again, a FINDRISC greater or equal to 15, with sensitivity of
67.7% and specificity of 67.2%.
With the exception of total and LDL cholesterol, all other
measured CVD risk factors had strong, significant associ-
ations with FINDRISC values (Table 2). Specifically, the
FINDRISC was significantly correlated with systolic (r = 0.28,
P < 0.001) and diastolic (r = 0.12, P = 0.001) blood pres-
sure, HDL cholesterol (r = 0.12, P < 0.001) and triglycerides
(r = 0.16, P < 0.001). Also, there was an inverse association
between the FINDRISC and smoking (r = 0.26, P < 0.001).
The prevalence of the MetS was also positively correlated
with the FINDRISC (r = 0.39, P < 0.001). The AUROC curve for
detecting the MetS was 0.733 (95% CI: 0.6990.767) overall,
with 0.707 (0.6530.760) for men and 0.757 (0.7130.801) for
women. Yet again, the optimal cutoff value for detecting the
MetS was a FINDRISC greater or equal to 15, with sensitivity
of 67.3% and specificity of 70.6% (Table 3).
4. Discussion
The American Diabetes Association (ADA) recommends
screening for T2D at 3-year intervals, beginning at age 45 years,
especially in those with a BMI greater or equal to 25 kg/m2 [34].
However, these recommendations are not widely followed, as
indicated by the fact that one-third of those who have diabetes go
undiagnosed [1]. As the main reason for this problem is the cost
and inconvenience of diabetes testing [35], one way to address
the issue is to develop a simple, inexpensive tool that can identify
those at high risk of having either pre-diabetes or undiagnosed
diabetes [36]. However, although a number of investigators have
developed diabetes risk-assessment tools [1221], most of these
have not been tested in populations other than those in which
they were originally developed [28].
 K. Makrilakis et al. / Diabetes & Metabolism 37 (2011) 144151 147

Table 1
Clinical, demographic and laboratory characteristics of the study participants by gender.
Variable Men Women P* Total

Participants (n) 379 490 869

Age (years) 56.0 (10.7) 56.4 (10.9) NS 56.2 (10.8)
BMI (kg/m2 ) 29.4 (4.0) 29.8 (5.7) NS 29.6 (5.0)
Waist circumference (cm) 103.3 (10.8) 94.0 (12.4) < 0.001 98.0 (12.6)
Smoking (never/ex/current) (%) 28/37/35 58/6/36 < 0.001 45/20/35
Systolic BP (mmHg) 128.3 (17.2) 124.1 (19.9) 0.001 126.0 (18.8)
Diastolic BP (mmHg) 79.9 (11.2) 76.0 (11.7) < 0.001 77.7 (11.6)
Fasting glucose (mg/dL) 5.83 (1.18) 5.61 (1.05) 0.004 5.71 (1.12)
Glucose at 120 min (mg/dL) 6.28 (2.91) 6.57 (2.79) NS 6.44 (2.84)
Total cholesterol (mmol/L) 5.61 (1.01) 5.82 (1.08) 0.003 5.73 (1.06)
Triglycerides (mmol/L) 1.47 (0.78) 1.25 (0.69) < 0.001 1.34 (0.74)
HDL cholesterol (mmol/L) 1.20 (0.19) 1.36 (0.24) < 0.001 1.29 (0.24)
LDL cholesterol (mmol/L) 3.76 (0.92) 3.90 (0.96) 0.026 3.84 (0.95)
FINDRISC value 12.6 (4.9) 13.6 (4.9) 0.005 13.1 (4.9)
Undiagnosed diabetes (%) 11.9 10.0 < 0.001 10.8
Dysglycaemia (%) 36.0 31.7 < 0.001 33.6
MetS (%) 39.1 37.9 NS 38.4

Data are presented as means ( SD) unless otherwise specified; NS: not significant; BMI: body mass index; BP: blood pressure; HDL: high-density lipoprotein;
LDL: low-density lipoprotein; MetS: the metabolic syndrome; Dysglycaemia = combined impaired fasting glucose, impaired glucose tolerance and type 2 diabetes.
* Men vs women.

Table 2
Characteristics of the study participants by FINDRISC.
Variable Finnish Type 2 Diabetes Risk Score (FINDRISC) P*

03 46 710 1114 1519 2026 Total

n (%) 18 (2.1) 63 (7.4) 190 (22.4) 201 (23.6) 299 (35.2) 79 (9.3) 840
Age (years) 46.5 (5.2) 48.6 (9.9) 54.0 (8.9) 57.5 (10.5) 57.1 (11.2) 61.8 (10.1) 56.2 (10.8) < 0.001
BMI (kg/m2 ) 23.6 (2.6 25.7 (2.7) 27.3 (3.9) 29.6 (4.5) 31.7 (5.0) 32.8 (4.4) 29.6 (5.0) < 0.001
Waist circumference (cm) 79.3 (10.4) 89.1 (9.9) 93.9 (11.4) 98.7 (12.0) 102.0 (11.6) 104.7 (10.0) 98.0 (12.6) < 0.001
Smoking (never/ex/current) (%) 43/14/43 29/21/50 32/24/44 55/21/24 47/25/28 44/29/27 45/20/35 0.004
Systolic BP (mmHg) 106.4 (19.4) 119.0 (15.5) 121.5 (17.3) 124.5 (19.5) 130.0 (18.3) 134.6 (16.7) 125.9 (18.8) < 0.001
Diastolic BP (mmHg) 68.7 (13.5) 76.6 (10.4) 77.2 (11.5) 76.8 (11.8) 79.4 (11.4) 78.6 (11.6) 77.8 (11.6) 0.01
Fasting glucose (mmol/L) 5.08 (0.53) 5.16 (0.53) 5.41 (0.72) 5.44 (0.93) 6.04 (1.31) 6.38 (1.28) 5.71 (1.12) < 0.001
Glucose at 120 min (mmol/L) 4.49 (1.53) 4.85 (1.51) 5.40 (1.87) 5.92 (2.19) 7.37 (3.1) 8.29 (3.58) 6.44 (2.84)) < 0.001
Total cholesterol (mmol/L) 5.66 (1.22) 5.68 (1.17) 5.77 (1.01) 5.71 (1.13) 5.73 (0.99) 5.76 (1.16) 5.73 (1.06) NS
Triglycerides (mmol/L) 0.92 (0.41) 1.22 (0.80) 1.29 (0.74) 1.31 (0.65) 1.38 (0.65) 1.67 (1.12) 1.34 (0.74) < 0.001
HDL cholesterol (mmol/L) 1.41 (0.27) 1.33 (0.24) 1.29 (0.24) 1.31 (0.24) 1.27 (0.22) 1.24 (0.23) 1.29 (0.24) 0.022
LDL cholesterol (mmol/L) 3.83 (1.10) 3.83 (1.08) 3.90 (0.93) 3.81 (1.02) 3.83 (0.84) 3.82 (1.06 3.84 (0.95) NS
Undiagnosed diabetes (%) 0.0 0.0 3.2 5.5 18.4 22.8 10.8 < 0.001
Dysglycaemia (%) 5.6 9.5 20.1 23.4 47.5 64.6 33.6 < 0.001
MetS (%) 11.1 9.8 20.0 30.5 56.0 71.8 38.9 < 0.001

Data are presented as means ( SD) unless otherwise specified; BMI: body mass index; BP: blood pressure; HDL: high-density lipoprotein; LDL: low-density
lipoprotein; NS: not significant; MetS: the metabolic syndrome; Dysglycaemia = combined impaired fasting glucose, impaired glucose tolerance and type 2 diabetes.
* One-way ANOVA or Chi2 test.
148 K. Makrilakis et al. / Diabetes & Metabolism 37 (2011) 144151
Table 3
Characteristics of the Finnish Type 2 Diabetes Risk Score (FINDRISC) questionnaire, using different cutoff values for screen-detected unknown type 2 diabetes
(SDM), dysglycaemia (combined impaired fasting glucose, impaired glucose tolerance and type 2 diabetes) and the metabolic syndromea .
Sensitivity Specificity
SDM
Cutoff = 7 100 10.7
Cutoff = 10 96.7 29.5
Cutoff = 15 81.1 59.8
Dysglycaemia
Cutoff = 7 97.5 13.1
Cutoff = 10 87.4 33.9
Cutoff = 15 67.7 67.2
MetS
Cutoff = 7 97.6 13.8
Cutoff = 10 89.0 37.0
Cutoff = 15 67.3 70.6
PPV: positive predictive value; NPV: negative predictive value.
a National Cholesterol Education Program Adult Treatment Panel III criteria.
Nevertheless, the present study has demonstrated that the
FINDRISC questionnaire can perform reasonably well in pre-
dicting undiagnosed T2D, glucose homoeostasis abnormalities
and the MetS in cross-sections of the Greek population, and at
least as well as it has in the other Caucasian populations stud-
ied so far [12,2226]. Using an optimal cutoff value of greater
or equal to 15, the questionnaire was able to identify 81.1%
of those with unknown diabetes and 67.7% of those with any
dysglycaemic condition (pre-diabetes and unknown diabetes).
In the original FINRISK studies [12], the FINDRISC ques-
tionnaire was developed as a predictor of future drug-treated
diabetes in the Finnish population (in 1987), and was validated
5 years later (in 1992). The AUROC curve for a cross-sectional
evaluation of the questionnaire to predict undiagnosed diabetes
was 0.80 (compared with 0.72 in the present study). It is well
known that models generally perform best with the data on which
they were initially developed, and that they perform better on
training rather than test data [37]. However, as the original ques-
tionnaire comprised only seven items (excluding a family history
of diabetes) for a possible maximum score of 20 [12], direct
comparisons with the present study data are difficult.
Nevertheless, the FINDRISC questionnaire was subse-
quently cross-sectionally validated in 2002 by a similar
FINRISK survey in Finland, using the full questionnaire of eight
items and a possible maximum score of 26 [22], thereby allowing
direct comparison with the present study. The AUROC curves
for the prevalence of unknown diabetes (0.722 in men and 0.732
in women) in that study were almost identical to those found in
our present study. In addition, the prevalence of screen-detected
diabetes was also similar (11.6% in men and 6.4% in women).
The AUROC curve for the detection of abnormal glucose tol-
erance (with IFG defined as an FPG greater than 6.1 mmol/L)
was lower than that of the present study (0.648 in men and 0.659
in women compared with 0.721 for men and 0.720 for women,
respectively). Furthermore, the AUROC curve for the prevalence
of the MetS (NCEPATP III definition) was also similar to the
present results (0.724 in men and 0.753 in women). Moreover,
the FINDRISC was associated with other CVD risk factors, such
PPV NPV Study sample %
11.7 100 90.5
14.0 98.7 73.3
19.3 96.4 44.5
36.2 91.4 90.5
40.0 84.1 73.3
51.1 80.5 44.5
41.9 89.9 90.5
47.4 84.1 73.3
59.3 77.2 44.5
as hypertension and lipid abnormalities, which is again similar
to the present findings (Table 2).
The accuracy of the FINDRISC for predicting unknown dia-
betes and CVD risk was further investigated by Italys Impaired
Glucose Tolerance and Long-Term Outcomes Observational
(IGLOO) study [23], which recruited 1377 individuals, aged
5577 years, to assess an opportunistic screening strategy based
on the FINDRISC. Those with a score greater or equal to 9 and
a subsequent FPG of 5.66.9 mmol/L underwent OGTT, and
these risk scores were found to represent a useful alternative
to indiscriminate FPG measurements, with an AUROC curve
for predicting unknown diabetes of 0.72 (95% CI: 0.680.76),
identical to that of our present study. The sensitivity and speci-
ficity for detecting unknown diabetes with a cutoff score greater
or equal to 9 were 86% and 41%, respectively (which is lower
than the 96.7% sensitivity, but higher than the 29.5% specificity,
found in the present study for a similar cutoff value greater or
equal to 10).
A simplified version of the FINDRISC questionnaire, based
on six questions (omitting diet and physical activity, and yield-
ing a maximum possible score of 23), was used to validate the
questionnaire in a middle-aged and older German population
[24]. In that case, the AUROC curve for detecting unknown dia-
betes was again similar to our present findings (0.745, 95% CI:
0.6770.813).
Two other studies evaluated the FINDRISC questionnaire and
then compared it with other available screening tools for the
detection of unknown diabetes [25,26]. The Cooperative Health
Research in the Region of Augsburg (KORA) study used the
tool in a German population (aged 5574 years; n = 1353), and
found that its sensitivity (82%) and positive predictive value
(12%) were similar to those reported in the FINRISK survey
(using a cutoff value greater or equal to 9), although the speci-
ficity (4%) and AUROC curve (65%) were somewhat lower [25].
The FINDRISC also performed equally as well as the Rotter-
dam Diabetes Study tool [14] and the San Antonio Heart Study
diabetes-prediction model [15] in terms of AUROC curves, all
of which were lower than that of the present study. Validation
 K. Makrilakis et al. / Diabetes & Metabolism 37 (2011) 144151
Fig. 1. Receiver operating characteristic (ROC) curves for the prevalence of
unknown type 2 diabetes (a), Dysglycaemia (b), and the metabolic syndrome
(c), in Greek men and women.
149
of the FINDRISC questionnaire in the Whitehall II cohort [26]
showed an AUROC curve of 0.67, which was considerably lower
than that originally presented in the FINRISK surveys and also
lower than ours.
A higher FINDRISC was also associated with a poorer CVD
risk-factor profile (Table 2), as shown in other studies as well
[22,23]. Smoking was the only risk factor negatively associated
with the FINDRISC. Although the prevalence of smoking was
rather high in our present population (35%) which is consis-
tent with the rest of Greece (around 40%) [38]smoking is not
part of the FINDRISC or any other questionnaire so far used to
predict dysglycaemia. However, this observation should perhaps
be further examined in larger studies, as smoking has also been
implicated in the development of T2D [39].
One limitation of the present study is that participants were
drawn from a limited population survey carried out in health-
care centers and occupational settings around Athens [29] and,
thus, the results may not be applicable to the rest of Greece.
Also, the number of participants was relatively small (n = 869),
although the results were similar to those of other studies in
different populations and larger cohorts [12,2226], which sup-
ports the validity of our present findings. However, in our study,
the diagnosis of diabetes and dysglycaemia was based on only
one OGTT value, not two as recommended. On the other hand,
the strengths of our study were that the diagnosis of diabetes
was not self-reported, and that the age distribution of the partic-
ipants was wide and included the vast majority of the high-risk
population (aged 3575 years).
5. Conclusion
The present study has validated the FINDRISC questionnaire
as a useful screening tool to identify unknown diabetes, abnor-
mal glucose homoeostasis and the MetS in a cross-section of
the Greek population. The questionnaire was originally devel-
oped in a prospective setting to identify those at high risk of
developing T2D in the future. The performance of the ques-
tionnaire in the present study was good, and comparable to its
performance in the other Caucasian populations tested [2226].
This confirms that the FINDRISC tool is indeed also valid for
Southern European populations, despite having lifestyles that
are very different from Northern European ones, and supports
the fact that risk factors for T2D are similar in different Euro-
pean populations. The FINDRISC tool is simple, inexpensive
and non-invasive [36], and has been specifically recommended
as a European and global screening tool by the European IMAGE
project, a programme for the Development and Implementation
of a European Guideline and Training Standards for Diabetes
Prevention, which aims to improve the management and reduce
the impact of T2D across Europe [7,40]. Another advantage of
the FINDRISC is that it can also identify those considered to
be at high risk for the development of T2D, but who currently
have normal blood glucose levels. This means that implement-
ing lifestyle interventional programmes in such cases will be a
true attempt at primary prevention of hyperglycaemia.
150 K. Makrilakis et al. / Diabetes & Metabolism 37 (2011) 144151
Conict of interest statement
None of the authors have any conflict of interest to declare.
Acknowledgements
The present study was supported by the Commission of
the European Communities, Directorate CPublic Health and
Risk Assessment, Health & Consumer Protection, Grant Agree-
ment number 2004310, and the DEPLAN project. It was
also co-funded by the private sector in this case, an unre-
stricted educational grant from Bristol-Myers Squibb in Greece.
We would also like to thank the following health centers and
individuals for their help in implementing the present study:
the medical staff of the Health Centre of Alimos (especially
Dr Ourania Zacharopoulou); the staff of the Centre for the
Elderly of Agioi Anargyroi; the medical and nursing staff of the
Health Centre of Markopoulo (especially Mrs R. Salonikioti);
the medical and nursing staff of the Hellenic Telecommunica-
tions Company (especially Drs C. Pietris and C. Alexopoulos);
the medical and nursing staff of the Hellenic Broadcasting Cor-
poration (especially Dr M. Katsorida); the medical and nursing
staff of the Bank of Greece (especially Drs V. Spandagos and
P. Konstantopoulou); the medical staff of the Olympic Village
complex (especially Dr S. Tigas); the staff of the electrical equip-
ment manufacturer Pitsos-Bosch; the medical and nursing staff
of the Health Centre of Vari (especially Dr M. Dandoulakis); and
the medical and nursing staff of the Health Centre of Vyronas
(especially Dr K. Kyriakopoulos).
References
[1] Cowie CC, Rust KF, Ford ES, Eberhardt MS, Byrd-Holt DD, Li C, et al.
Full accounting of diabetes and pre-diabetes in the U.S. Population in
19881994 and 20052006. Diabetes Care 2009;32:28794.
[2] Kahn SE. The relative contributions of insulin resistance and beta-cell
dysfunction to the pathophysiology of type 2 diabetes. Diabetologia
2003;46:319.
[3] Tuomilehto J, Lindstrom J, Eriksson JG, Valle TT, Hamalainen H, Ilanne-
Parikka P, et al. Prevention of type 2 diabetes mellitus by changes in
lifestyle among subjects with impaired glucose tolerance. N Engl J Med
2001;344:134350.
[4] Knowler WC, Barrett-Connor E, Fowler SE, Hamman RF, Lachin JM,
Walker EA, et al. Reduction in the incidence of type 2 diabetes with lifestyle
intervention or metformin. N Engl J Med 2002;346:393403.
[5] Chiasson JL, Josse RG, Gomis R, Hanefeld M, Karasik A, Laakso M.
Acarbose for prevention of type 2 diabetes mellitus: the stop-NIDDM
randomized trial. Lancet 2002;359:20727.
[6] Gerstein HC, Yusuf S, Bosch J, Pogue J, Sheridan P, Dinccag N, et al. Effect
of rosiglitazone on the frequency of diabetes in patients with impaired
glucose tolerance or impaired fasting glucose: a randomized controlled
trial. Lancet 2006;368:1096105.
[7] Paulweber B, Valensi P, Lindstrom J, Lalic NM, Greaves CJ, McKee M,
et al. A European evidence-based guideline for the prevention of type 2
diabetes. Horm Metab Res 2010;42(Suppl 1):S336.
[8] Waugh N, Scotland G, McNamee P, Gillett M, Brennan A, Goyder E, et al.
Screening for type 2 diabetes: literature review and economic modelling.
Health Technol Assess 2007;11(iii-iv, ix-xi):1125.
[9] American Diabetes Association. Diagnosis and classification of diabetes
mellitus. Diabetes Care 2010;33(Suppl 1):S629.
[10] Haffner SM, Stern MP, Hazuda HP, Mitchell BD, Patterson JK. Cardiovas-
cular risk factors in confirmed prediabetic individuals. Does the clock for
coronary heart disease start ticking before the onset of clinical diabetes?
JAMA 1990;263:28938.
[11] Kohli P, Greenland P. Role of the metabolic syndrome in risk assessment
for coronary heart disease. JAMA 2006;295:81921.
[12] Lindstrom J, Tuomilehto J. The diabetes risk score: a practical tool to
predict type 2 diabetes risk. Diabetes Care 2003;26:72531.
[13] Glumer C, Carstensen B, Sandbaek A, Lauritzen T, Jorgensen T, Borch-
Johnsen K. A Danish diabetes risk score for targeted screening: the inter99
study. Diabetes Care 2004;27:72733.
[14] Baan CA, Ruige JB, Stolk RP, Witteman JC, Dekker JM, Heine RJ, et al.
Performance of a predictive model to identify undiagnosed diabetes in a
health care setting. Diabetes Care 1999;22:2139.
[15] Stern MP, Williams K, Haffner SM. Identification of persons at high risk
for type 2 diabetes mellitus: do we need the oral glucose tolerance test?
Ann Intern Med 2002;136:57581.
[16] Griffin SJ, Little PS, Hales CN, Kinmonth AL, Wareham NJ. Diabetes
risk score: towards earlier detection of type 2 diabetes in general practice.
Diabetes Metab Res Rev 2000;16:16471.
[17] Colagiuri S, Hussain Z, Zimmet P, Cameron A, Shaw J. Screening for type
2 diabetes and impaired glucose metabolism: the Australian experience.
Diabetes Care 2004;27:36771.
[18] Herman WH, Smith PJ, Thompson TJ, Engelgau MM, Aubert RE. A new
and simple questionnaire to identify people at increased risk for undiag-
nosed diabetes. Diabetes Care 1995;18:3823847.
[19] Kanaya AM, Wassel Fyr CL, de Rekeneire N, Shorr RI, Schwartz AV,
Goodpaster BH, et al. Predicting the development of diabetes in older
adults: the derivation and validation of a prediction rule. Diabetes Care
2005;28:4048.
[20] Heikes KE, Eddy DM, Arondekar B, Schlessinger L. Diabetes risk calcu-
lator: a simple tool for detecting undiagnosed diabetes and pre-diabetes.
Diabetes Care 2008;31:10405.
[21] Balkau B, Lange C, Fezeu L, Tichet J, de Lauzon-Guillain B, Czernichow
S, et al. Predicting diabetes: clinical, biological, and genetic approaches:
data from the epidemiological study on the insulin resistance syndrome
(Desir). Diabetes Care 2008;31:205661.
[22] Saaristo T, Peltonen M, Lindstrom J, Saarikoski L, Sundvall J, Eriksson
JG, et al. Cross-sectional evaluation of the Finnish diabetes risk score: a
tool to identify undetected type 2 diabetes, abnormal glucose tolerance and
metabolic syndrome. Diab Vasc Dis Res 2005;2:6772.
[23] Franciosi M, De Berardis G, Rossi MC, Sacco M, Belfiglio M, Pellegrini
F, et al. Use of the diabetes risk score for opportunistic screening of undi-
agnosed diabetes and impaired glucose tolerance: the IGLOO (impaired
glucose tolerance and long-term outcomes observational) study. Diabetes
Care 2005;28:118794.
[24] Bergmann A, Li J, Wang L, Schulze J, Bornstein SR, Schwarz PE. A sim-
plified Finnish diabetes risk score to predict type 2 diabetes risk and disease
evolution in a German population. Horm Metab Res 2007;39:67782.
[25] Rathmann W, Martin S, Haastert B, Icks A, Holle R, Lowel H, et al.
Performance of screening questionnaires and risk scores for undiagnosed
diabetes: the KORA survey 2000. Arch Intern Med 2005;165:43641.
[26] Witte DR, Shipley MJ, Marmot MG, Brunner EJ. Performance of existing
risk scores in screening for undiagnosed diabetes: an external validation
study. Diabet Med 2010;27:4653.
[27] Lin JW, Chang YC, Li HY, Chien YF, Wu MY, Tsai RY, et al.
Cross-sectional validation of diabetes risk scores for predicting diabetes,
metabolic syndrome, and chronic kidney disease in Taiwanese. Diabetes
Care 2009;32:22946.
[28] Glumer C, Borch-Johnsen K, Colagiuri S. Can a screening programme for
diabetes be applied to another population? Diabet Med 2005;22:12348.
[29] Makrilakis K, Liatis S, Grammatikou S, Perrea D, Katsilambros N. Imple-
mentation and effectiveness of the first community lifestyle intervention
programme to prevent Type 2 diabetes in Greece. The Deplan study. Diabet
Med 2010;27:45965.
[30] Schwarz PE, Lindstrom J, Kissimova-Scarbeck K, Szybinski Z, Barengo
NC, Peltonen M, et al. The European perspective of type 2 diabetes preven-
tion: diabetes in Europeprevention using lifestyle, physical activity and
nutritional intervention (DE-PLAN) project. Exp Clin Endocrinol Diabetes
2008;116:16772.
 K. Makrilakis et al. / Diabetes & Metabolism 37 (2011) 144151
[31] Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentra-
tion of low-density lipoprotein cholesterol in plasma, without use of the
preparative ultracentrifuge. Clin Chem 1972;18:499502.
[32] WHO: Definition, diagnosis and classification of diabetes mellitus and
its complications. Report of a WHO Consultation. Part 1: Diagnosis and
classification of diabetes mellitus. Geneva: World Health Organization.
Department of Noncommunicable Disease Surveillance.1999.
[33] Third report of the national cholesterol education program (NCEP) expert
panel on detection, evaluation, and treatment of high blood cholesterol
in adults (Adult Treatment Panel III) final report. Circulation 2002;
106:3143421.
[34] American Diabetes Association. Screening for type 2 diabetes. Diabetes
Care 2003;26(Suppl 1):S214.
[35] Wareham NJ, Griffin SJ. Should we screen for type 2 diabetes? Evaluation
against national screening committee criteria. BMJ 2001;322:9868.
151
[36] Schwarz PE, Li J, Lindstrom J, Tuomilehto J. Tools for predicting the
risk of type 2 diabetes in daily practice. Horm Metab Res 2009;41:
8697.
[37] Bleeker SE, Moll HA, Steyerberg EW, Donders AR, Derksen-Lubsen G,
Grobbee DE, et al. External validation is necessary in prediction research:
a clinical example. J Clin Epidemiol 2003;56:82632.
[38] Vardavas CI, Kafatos AG. Smoking policy and prevalence in Greece: an
overview. Eur J Public Health 2007;17:2113.
[39] Haire-Joshu D, Glasgow RE, Tibbs TL. Smoking and diabetes. Diabetes
Care 1999;22:188798.
[40] Lindstrom J, Neumann A, Sheppard KE, Gilis-Januszewska A, Greaves CJ,
Handke U, et al. Take action to prevent diabetes the IMAGE toolkit for the
prevention of type 2 diabetes in Europe. Horm Metab Res 2010;42(Suppl
1):S3755.