A systematic review of predictive risk models for diabetes complications based on large scale clinical studies,☆☆,

https://doi.org/10.1016/j.jdiacomp.2012.11.003Get rights and content

Abstract

This work presents a systematic review of long-term risk assessment models for evaluating the probability of developing complications in diabetes patients. Diabetes mellitus can cause many complications if not adequately controlled; risk assessment models can help physicians and patients in identifying the complications most likely to arise and in taking the necessary countermeasures.

We identified six large medical studies related to diabetes mellitus upon which current available risk assessment models are built on; all these studies had duration over 5 years and most of them included some common demographic and clinical data strongly related to diabetic complications. The most common predictions for long term diabetes complications are related to cardiovascular diseases and diabetic retinopathy.

Our analysis of the literature led us to the conclusion that researchers and medical practitioners should take in account that some limitations undermine the applicability of risk assessment models; for example, it is hard to judge whether results obtained on a specific cohort can be effectively translated to other populations. Nevertheless, all these studies have significantly contributed to identify significant risk factors associated with the major diabetes complications.

Introduction

Risk factors for diabetic complications have been intensively studied during the last decades, and these studies greatly improved the current scientific knowledge about the biological processes underlying diabetes. Risk factors can be used for the definition of risk assessment models to be exploited in the clinical practice. These models are part of the computational core of clinical/medical applications able to stratify diabetes patients according to their probability of developing complications or experiencing adverse events.

A risk assessment model consists of any type of algorithm or mathematical formula (e.g., a set of rules, a decision tree, a weighted sum, etc.) for assessing the overall statistical probability of certain adverse outcomes to occur in the future. Medical risk assessment may provide probabilistic statements as the likelihood that certain complications may occur given the present and historical health status.

When risk assessment models are built upon data collected from large scale, longitudinal clinical studies, they are able to perform predictions in the long term, i.e. on a time horizon spanning up to a decade and beyond. These models are the backbone of risk assessment tools used in clinical practice.

Given the health and social burden caused by diabetes-related complications, it is not surprising that several scientific works have proposed risk assessment models able to evaluate the probability for diabetes patients of developing one or more complications on the long-term period.

The aim of the present review is to compare and summarize the most relevant risk assessment models for diabetes-related complications published in the literature. After an initial screening of the large-scale, longitudinal clinical studies performed for studying diabetes complications, the risk assessment models built on top of these studies are presented and compared with each other.

This review has been performed in the context of the European Union (EU) funded project REACTION (http://www.reactionproject.eu/). The scope of the REACTION project is to design and develop a platform of services that can provide professionals with remote monitoring and therapeutic management of diabetic patients in different healthcare regimes. Part of this effort consists in reviewing, developing and implementing tools able to provide long term risk assessment evaluations about diabetic complications based on patient's current health state and history. The tools will take advantage of integration between instantaneously measured data from sensors, historical data from the Electronic Patient Record (EPR), statistical data from stratification studies, statistical databases and evidence-based case management repositories. The methodological approach followed in the REACTION project for the realisation of the long term risk assessment models started from the evaluation of existing risk assessment studies and models in order to (a) identify available sources of data suitable for deriving long term risk assessment models and (b) exploit the experience gained from previous research. The present review reports and summarizes the results of this systematic literature search.

Section snippets

Methods

A selection of the studies related to diabetes mellitus complications was started. Only studies with at least 1000 subjects and 5 years of follow-up were considered. We included both prospective and retrospective, as well as Type I and Type II diabetes studies. Six major studies were identified, namely DCCT/EDIC(The Diabetes Control and Complications Trial Research Group, 2003, The Epidemiology of Diabetes Interventions and Complications (EDIC) Research Group, 1999), EuroDiab (The EURODIAB IDDM

Results

Large scale clinical studies typically last around a decade, involve thousands of patients in numerous health centres, and measure different aspects of patients' clinical medical profiles. Thus, not surprisingly the data collected in each study can be employed for deriving multiple risk assessment models, differing from each other for predicted outcomes, involved parameters or analytical techniques.

Table 1, Table 2, Table 3 give an overview of the long-term risk assessment studies and models.

Discussion

In the context of the REACTION project, this survey will provide the basis for the selection of the most relevant risk score to be implemented within the REACTION platform. Moreover, published risk models will provide insights and guidelines in a first phase for developing more advanced models based on the available studies, and in a second phase for devising and building new risk models based upon the data collected from the REACTION project.

A principal objective in the clinical management of

Conclusions

Risk factors for diabetic complications have been extensively studied during the last decades, and these studies greatly improved the current scientific knowledge about the pathophysiological processes underlying diabetes. The most common predictive models for diabetes complications are related to cardiovascular disease, coronary heart disease and diabetic retinopathy. However, such studies have important limitations, mainly due to their long temporal duration. Major differences in the medical

Acknowledgments

This work is supported by the European Commission's Seventh Framework Program in the area of Personal Health Systems under Grant Agreement no. 248590 (REACTION FP7-IP-No 248590).

References (20)

  • C. Green et al.

    Geographic analysis of diabetes prevalence in an urban area

    Social Science & Medicine

    (2003)
  • R.E. Carter et al.

    Intensive treatment of diabetes is associated with a reduced rate of peripheral arterial calcification in the diabetes control and complications trial

    Diabetes Care

    (2007)
  • J. Cederholm et al.

    Risk prediction of cardiovascular disease in type 2 diabetes: A risk equation from the Swedish National Diabetes Register

    Diabetes Care

    (2009)
  • P.M. Clarke et al.

    A model to estimate the lifetime health outcomes of patients with type 2 diabetes: The United Kingdom Prospective Diabetes Study (UKPDS) outcomes model

    Diabetologia

    (2004)
  • P.A. Cleary et al.

    The effect of intensive glycemic treatment on coronary artery calcification in type 1 diabetic participants of the Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) Study

    Diabetes

    (2006)
  • G.S. Collins et al.

    An independent external validation and evaluation of QRISK cardiovascular risk prediction: A prospective open cohort study

    BMJ

    (2009)
  • F.E. Harrel

    Regression modeling strategies, with applications to linear models, logistic regression, and survival analysis

    (2001)
  • J. Hippisley-Cox et al.

    Derivation and validation of QRISK, a new cardiovascular disease risk score for the United Kingdom: Prospective open cohort study

    BMJ

    (2007)
  • J. Hippisley-Cox et al.

    Predicting cardiovascular risk in England and Wales: Prospective derivation and validation of QRISK2

    BMJ

    (2008)
  • V. Kothari et al.

    UKPDS 60: Risk of stroke in type 2 diabetes estimated by the UK Prospective Diabetes Study risk engine

    Stroke

    (2002)
There are more references available in the full text version of this article.

Cited by (48)

  • Clinical outcome and determinants of amputation in a large cohort of Iranian patients with diabetic foot ulcers

    2020, Foot
    Citation Excerpt :

    The risk factors of DFU and LEA have been studies extensively during the recants years in various populations and ethnic groups [1,5,12,18,19,27]. The understanding of DFU and LEA risk factors can be used to compose risk stratification models and to predict the outcome measures [32]. The identified risk factors for DFU and LEA in patients with DM have been reported to be age [1,12,19], male gender and previous history of ischemic stroke [33,34], comorbidities such as ischemic heart disease and hypertension [35], chronic complications such as peripheral artery disease [1,12,18], nephropathy [5,18,36], duration of diabetes [14,27], and HbA1C [11,33].

  • A novel algorithm based on information diffusion and fuzzy MADM methods for analysis of damages caused by diabetes crisis

    2019, Applied Soft Computing Journal
    Citation Excerpt :

    In [7], available technological advancements have been utilized to develop prediction models for the prediction of a type II diabetic patient. One of the well-known studies in the field of long term risk of complications from diabetes is Diabetes Control and Complications Trial (DCCT), in this study, 1441 diabetic patients from 29 medical centers of America and Canada have been studied and the aim was to control the severity of the disease to prevent complications [8]. Another study is conducted specifically to control the disease’s severity of macro vascular complications which are not intended in the previous research.

  • Identifying people at risk of developing type 2 diabetes: A comparison of predictive analytics techniques and predictor variables

    2018, International Journal of Medical Informatics
    Citation Excerpt :

    Healthcare professionals have sought to overhaul electronic medical records (EMR), and to integrate data into large clinical data warehouses for using in auditing, continuous quality improvement, health service planning, epidemiological studies and evaluation research [9]. Predictive analytics techniques have the potential to recognize the complex patterns in electronic medical records and identify patients at risk of developing diseases [10]. For example, Cichosz et al. [11] reviews the literature and identify the predictive models for diabetes.

View all citing articles on Scopus

Grant Support: This work is supported by the European Commission's Seventh Framework Program in the area of Personal Health Systems under Grant Agreement no. 248590 (REACTION FP7-IP-No 248590).

☆☆

Authors' contributions: VL and LK were the main authors for the manuscript and conducted the article search, screening, selection and risk factors analysis. FC, EL and IT assisted in writing and reviewing the article.

Competing interests: The authors declare that they have no competing interests.

1

These authors contributed equally to this work.

View full text