A simple clinical score for type 2 diabetes mellitus screening in the Canary Islands

https://doi.org/10.1016/j.diabres.2007.10.022Get rights and content

Abstract

Aim

To develop a straightforward risk score for type 2 diabetes (DM2) screening to use in clinical practice.

Methods

A sample of 6237 adult inhabitants of the Canary Islands (Spain) was randomly divided into two subgroups: one yielded data used to develop the instrument, and the other yielded data used for validation testing. Performance of the instrument was compared in persons with clinically diagnosed DM2 and undiagnosed diabetes. The risk score, calculated by multivariate logistic regression, included the potential risk variables that yielded the highest odds ratio in the univariate analysis. A cut-off point for screening purposes was established at a 99% negative predictive value.

Results

In men, variables included in the risk score were age, waist/height ratio, familial antecedents of diabetes, and systolic blood pressure (ROC curve 0.837, 95% CI: 0.803–0.871). In women, the risk score contained the same variables plus gestational diabetes history (ROC curve 0.874, 95% CI: 0.847–0901). Excluding systolic blood pressure from the score had no significant effect on the area under the curve. This instrument resulted valid only for people aged less than 55 years.

Conclusions

This simple risk score for DM2 would be easy to apply in clinical practice.

Introduction

In recent decades the prevalence of obesity and type 2 diabetes mellitus (DM2) in developed countries has increased alarmingly [1] and the Canary Islands are no exception [2].

The American Diabetes Association currently recommends fasting plasma glucemia as a screening method for DM2 in adults older than 45 years, mainly in conjunction with BMI  25 kg/m2. In younger persons who are overweight and have other risk factors for diabetes, it has been proposed that screening should take place every 3 years [3]. Many physicians claim lack of time as one of the main obstacles to providing care for patients with diabetes [4]. Moreover, screening programs carried out in primary care services have been shown to require large amounts of staff time, and need to be targeted at persons over a certain age and with multiple risk factors [5].

A simple, valid, reliable screening method able to rapidly identify asymptomatic persons in whom fasting blood glucose should be measured to rule out DM2 would help make diagnosis and management of the disease in primary care more effective. Recent years have seen the publication of a number of studies that searched for a risk score able to simplify and facilitate the detection of diabetes in patients who are unaware they may have the disease [6], [7], [8], [9], [10], or able to predict who will develop DM2 [11], [12], [13].

The aim of this study was to develop a risk score for DM2 screening that is easy to use in clinical practice.

Section snippets

Methods

The Canary Islands are located off the northeast coast of Africa, and the current population of about 2 million inhabitants is mainly Caucasian [14]. The study sample consisted of 2648 men and 3589 women aged 18–75 years selected randomly from the general population. Two-step sampling (by county of residence and person) with stratification by island and by gender was used [2], [15]. The recruitment of the participants was conducted between 2000 and 2005. We excluded from the sample people with

Results

The mean age was higher in women with diabetes (n = 137) than women without diabetes (n = 1650) (55.4 ± 0.8 years vs. 42.0 ± 0.3 years, respectively; p < 0.001). Men with diabetes (n = 122) were also older than those without the disease (n = 1210) (54.0 ± 0.8 years vs. 42.0 ± 0.3 years, respectively; p < 0.001). Correlation analysis confirmed that some component variables within in each dimension were redundant (correlation coefficients 0.35–0.96; p < 0.05), the exception being the obstetric antecedents domain, in

Discussion

The risk score for DM2 described here can be calculated from the responses to fewer questionnaire items than most measures published to date. This means that less time is needed to screen individuals with this risk score, a factor that facilitates its use in clinical settings.

It was opted to develop separate scores for men and women because of the well-known association between gestational diabetes and the later appearance of DM2 [18], and because high birth weight of newborns has been reported

Conflict of interest

The authors state that they have no conflict of interest.

Acknowledgements

We thank K. Shashok for translating the original manuscript into English. This study was supported by the Fundación Ciencia y Salud (FUNCIS 45/98) and by Grant from the Fondo de Investigaciones Sanitarias (FIS 021189).

References (27)

  • C. Glumer et al.

    A Danish diabetes risk score for targeted screening: the Inter99 study

    Diabetes Care

    (2004)
  • J. Lindstrom et al.

    The diabetes risk score: a practical tool to predict type 2 diabetes risk

    Diabetes Care

    (2003)
  • W. Aekplakorn et al.

    A risk score for predicting incident diabetes in the Thai population

    Diabetes Care

    (2006)
  • Cited by (27)

    • Clinical applicability and cost-effectiveness of DIABSCORE in screening for type 2 diabetes in primary care

      2017, Diabetes Research and Clinical Practice
      Citation Excerpt :

      The study was piloted previously in five primary care centers in each region to detect possible limitations in managing the computer application, the time consumed or other aspects related to the completion of the protocol. With the data obtained, DIABSCORE was automatically computed as described in the report of its previous validation [11], giving each variable the following score: age (scored as the number of years), waist-to-height ratio (scored as the value of the waist-to-height ratio multiplied by 100), family history of type 2 diabetes (scored as Yes = 10 points, No = 0 points), and history of gestational diabetes (score as Yes = 25 points; No = 0 points). In addition to the above, each patient gave a fasting (≥10 h) venous blood sample for measurement of baseline glucose (mg/dL) and HbA1c (%) following the usual procedure of the health center.

    • Application and applicability of non-invasive risk models for predicting undiagnosed prevalent diabetes in Africa: A systematic literature search

      2015, Primary Care Diabetes
      Citation Excerpt :

      The risk models showed an apparent discriminatory power ranging from 0.65 [45] to 0.88 [23,36] in the derivation cohorts. For those studies reporting the performances by gender, point estimates for the discrimination were higher among women than men in three studies [12,38,45]. The discriminatory in the split-sample internal validation ranged from 0.63 [13] to 0.82 [40] and from 0.66 [44] to 0.88 [36] in external validation.

    • How diabetes risk assessment tools are implemented in practice: A systematic review

      2014, Diabetes Research and Clinical Practice
      Citation Excerpt :

      The limited number of risk assessment tools that have been implemented might be due to the need to restrict use to certain population groups with similar characteristics with those in which they were derived, although there is limited evidence to support such conjecture [56]. In addition, diabetes risk assessment tools that are currently available cover most race/ethnicity in all continents [45,57–62], which means there are many tools from which to choose to suit any particular population group. Nevertheless, tools always need to be validated for an individual population before implementation [63,64].

    • Inverse association between serum resistin and insulin resistance in humans

      2008, Diabetes Research and Clinical Practice
      Citation Excerpt :

      Analysis were also made of relationships between resistin and factors associated with the three conditions, anthropometric parameters, and physical activity. Participants in this cross-sectional, descriptive study were the first 713 individuals enrolled in the “CDC – Cáncer, Diabetes, Cardiovascular – of the Canary Islands” study, some findings from which have been reported earlier [11,12]. Briefly, participants were recruited from the general population of the Canary Islands (Spain).

    View all citing articles on Scopus
    View full text