Abstract
Background.
In Hong Kong, fasting plasma glucose (FPG) is the most popular screening test for diabetes mellitus (DM) in primary care. Individuals with impaired fasting glucose (IFG) are commonly encountered.
Objectives.
To explore the determinants of regression to normoglycaemia among primary care patients with IFG based on non-invasive variables and to establish a nomogram for the prediction of regression from IFG.
Methods.
This cohort study consisted of 1197 primary care patients with IFG. These subjects were invited to repeat a FPG test and 75-g 2-hour oral glucose tolerance test (2h-OGTT) to determine the glycaemia change. Normoglycaemia was defined as FPG <5.6 mmol/L and 2h-OGTT <7.8 mmol/L. Stepwise logistic regression model was developed to predict the regression to normoglycaemia with non-invasive variables, using a randomly selected training dataset (810 subjects). The model was validated on the remaining testing dataset (387 subjects). Area under the receiver operating characteristic curve (AUC) and Hosmer–Lemeshow test were used to evaluate discrimination and calibration of the model. A nomogram was constructed based on the model.
Results.
After a mean follow-up period of 6.1 months, 180 subjects (15.0%) had normoglycaemia based on the repeated FPG and 2h-OGTT results at follow-up. Subjects without central obesity or hypertension, with moderate-to-high-level physical activity and a lower baseline FPG level, were more likely to regress to normoglycaemia. The prediction model had acceptable discrimination (AUC = 0.705) and calibration (P = 0.840).
Conclusion.
The simple-to-use nomogram could facilitate identification of subjects with low risk of progression to DM and thus aid in clinical decision making and resource prioritization in the primary care setting.
Key words. Chinese, impaired fasting glucose, nomogram, normoglycaemia, primary care, regression.
Introduction
Impaired fasting glucose (IFG) refers to fasting plasma glucose (FPG) concentration below the diagnostic threshold for diabetes mellitus (DM) but higher than the normal levels. It is not a disease entity but reflects abnormal glucose regulation, in particular, hepatic insulin resistance (1). Individuals with IFG have a greater risk of developing DM compared to those with normal glucose level: the relative risk of DM was 4.66 (95% CI: 2.47–6.85) in individuals with IFG compared to those with normal glucose level (2), and the annualized incidence of DM was estimated between 1.6% and 31% (2). Among IFG individuals, the risk factors for progression to DM are similar to those demonstrated in the general population, including hypertension, smoking, obesity and dyslipidemia (3).
On the other hand, some subjects with IFG may regress back to normoglycaemia with lifestyle intervention or drug treatment (4,5). The regression is accompanied by a reduced risk of DM; in the Diabetes Prevention Program Outcome Study (DPPOS), individuals with pre-diabetes [defined as IFG and/or impaired glucose tolerance (IGT)] who achieved normoglycaemia were demonstrated to have a 56% lower risk of developing DM in 10 years compared to those still having pre-diabetes (6). Furthermore, another DPPOS article showed that the restoration of normoglycaemia through lifestyle or medication intervention could significantly reduce the 10-year cardiovascular risk calculated from Framingham scores in people with IFG, IGT or both (7).
In Hong Kong, FPG test is the recommended and the most popular screening test for DM in the primary care settings currently (8). Therefore, individuals with IFG can be more readily identified compared to other pre-diabetic subjects with IGT or elevated HbA1c, because IGT state can only be confirmed with an oral glucose tolerance test (OGTT) that is labour intensive and inconvenient, and HbA1c testing is still not widely utilized due to cost. However, we should bear in mind that this group of individuals may have concomitant IGT, elevated HbA1c or even DM. In view of the heterogeneity of the group and the significance of achieving normoglycaemia, risk stratification to identify those IFG individuals with highest likelihood of progression to DM or regression to normoglycaemia can be a potentially time- and cost-saving strategy in allocating appropriate resources to them accordingly, and thereby controlling development of type 2 DM and cardiovascular events.
To meet this goal, many risk prediction models were developed to identify individuals at risk of developing DM (9). Despite its usefulness in clinical prevention practice, none of the tools can be generalized to other populations due to the ethnic difference of DM risk (9). Furthermore, most of these studies focused on healthy population, rather than individuals with IFG who are at increased risk of developing DM. More importantly, none to the date focused on the regression to normoglycaemia. Regression from IFG to normoglycaemia should be an important goal for DM prevention. Because subjects with IFG can be more readily identified in the primary care setting with the simple FPG testing as aforementioned, focused intervention programme should be established for these IFG subjects in order to delay or prevent the development of DM. However, a nationally representative sample of 98658 Chinese adults in 2010 showed that the prevalence of IFG was 27.2% (95% CI: 26.8–27.6%) (10). Thus, it may not be possible to give programme-based interventions to this large high-risk population in consideration of the cost.
Therefore, in this study, we aimed to evaluate the determinants of regression from IFG to normoglycaemia based on the FPG levels and other non-invasive variables, and to develop and validate a nomogram that can be used to predict the regression in primary care clinical settings. The nomogram is a 2-dimensional diagram designed to allow fast calculation of complicated formulas to a practical precision. In our study, it may act as a simple-to-use tool to aid clinical decision making in primary care, such as prioritizing patients for further screening testing or intervention. Information sharing process through the usage of the nomogram will inform patients on their individual DM risk and facilitate discussion on risk factor management in order to promote regression to normoglycaemia.
Methods
Study design and subjects
The study design and subject recruitment at baseline had been reported previously (11). In brief, convenience sampling was used to recruit subjects from three public primary care clinics in Hong Kong from May 2013 to February 2015, where FPG testing is performed routinely among at risk patients, including those with hypertension, obesity and a positive family history of DM, to screen for the presence of DM. Eligible subjects were Chinese aged over 18 years, without known history of DM, who had been diagnosed to have IFG according to the American Diabetes Association (ADA) criteria with a FPG level between 5.6 – 6.9 mmol/L (12) regardless of their results of 75-g OGTT and HbA1c test. These subjects were identified by frontline doctors and nurses at the participating clinics during their follow-up visits and written inform consents were obtained from all eligible subjects who agreed to participate in the study. The participants then completed a self-reported questionnaire and physical examination by clinic nurses according to standardized protocol and a follow-up FPG test and 2-hour post-challenge plasma glucose level after a 2h-OGTT to determine the change in glycemic status within an 18-month period. A total of 1200 participants were recruited. Participants with previously diagnosed DM at baseline (n = 3) were excluded from the subsequent analysis, thus a complete-case analysis was conducted. All participants have signed a written consent form.
This study protocol was registered at the Hong Kong Clinical Trial Centre (Ref: HKCTR-1684) and ClinicalTrials.gov, the U.S. National Institutes of Health (Ref: NCT02439684).
Variables measured
A standardized questionnaire was used to collect the socio-demographic information, smoking status, family history of DM and physical activity of the participants. Physical activity was collected using the self-reported short version of the International Physical Activity Questionnaire (IPAQ), which covered questions related to the frequency and duration of walking, moderate- and vigorous-intensity activity (13). To measure the volume of activity, a metabolic equivalent task (MET) minute was computed by multiplying the duration of a specific type of activity (minutes) in a week and the corresponding constant value of that specific type of activity, which was 3.3 METs for walking, 4 METs for moderate-intensity activity and 8 METs for vigorous-intensity activity. Both the total volume of activity and the number of days/session of activity were used to calculate the activity level of an individual, which were finally classified into low, moderate and high level of activity according to the IPAQ guidelines. Medication information, including usage of β-blockers, diuretics and statin, was extracted by trained research assistant from the Clinical Management System (CMS) of the Hospital Authority, a validated research platform with high level of data completeness prescription in details (99.8%) (14).
A physical examination including body weight, height, waist circumference and blood pressure were conducted by registered nurses according to standardized procedures in the participating clinics. Body mass index (BMI) was calculated as the ratio of body weight to the square of body height. Overweight was defined as 23 < BMI < 25kg/m2 and obesity was defined as BMI ≥25kg/m2 in these Asian subjects. Waist circumference was measured at the level midway between iliac crest and lower rib margin while standing. Central obesity was defined as waist circumference ≥80cm in women and ≥90cm in men. Blood pressure (BP) was measured after 10 minutes of rest with an automated blood pressure monitoring device. Hypertension was defined as systolic BP/diastolic BP ≥140/90 mmHg and/or previously diagnosed hypertension.
At follow-up, blood samples were collected after overnight fasting for at least 8 hours to assess the level of FPG and 2h-OGTT. Regression to normoglycaemia was defined as FPG <5.6 mmol/L and 2h-OGTT <7.8 mmol/L.
Statistical analysis
The baseline characteristics between those with regression to normoglycaemia and those without were compared using independent t-test for continuous variables and χ 2 test for categorical variables. Around two-third of the total participants were randomly selected as the training dataset and the remaining as the testing dataset. Similarly, the differences between training dataset and testing dataset were compared using independent t-test and χ 2 test, where appropriate. All data were expressed as mean ± [standard deviation (SD)] or percentages, as appropriate.
Stepwise binary logistic regression model was established to determine the non-invasive variables that were associated with regression from IFG to normoglycaemia on the training dataset and were further validated on the remaining testing dataset. Age, gender, smoking status, levels of physical activity, family history of DM, hypertension, BMI status, central obesity, follow-up duration, FPG at recruitment and medication usage, including β-blockers, diuretics and statin medication, were initially entered into the stepwise logistic regression model. Among the covariates included, physical activity status was coded as low, moderate and high level of activity according to the IPAQ analysis algorithms. Follow-up duration was grouped into <3 months, 3–6 months, 6–9 months, 9–12 months and ≥12 months. FPG at recruitment was divided into three groups in the model: 5.6–6.0 mmol/L, 6.1–6.5 mmol/L and 6.5–6.9 mmol/L. Variables with a P-value less than 0.1 in the stepwise binary logistic regression analysis with normoglycaemia as the outcome were retained in the final model.
Each variable in the final model was assigned a weighting score using the respective β-coefficients multiplied by 10 and rounded to the nearest integer, which was subsequently constructed as the variable axes in the nomogram. The total score for each subject was the sum of points contributed by each variable identified by the final risk assessment model. A detailed example was attached to the figure legend of Figure 1.
Figure 1.
Nomogram for predicting the regression from impaired fasting glucose (IFG) to normoglycaemia using non-invasive variables. Instruction for using the nomogram is as follows. First, draw a straight upward line to the ‘Points’ axis and determine the points contributed by central obesity. In this case, no central obesity contributes to 6 points, while having central obesity contributes to 0 points. Second, repeat the same process for hypertension, physical activity level and fasting plasma glucose (FPG) at recruitment. Third, adding up all the points from the four domains to get total points. Last, draw a straight downward line from the “Total Points” axis to the ‘Estimated probability of regression from IFG to normoglycaemia’ axis and get the estimated probability. An example of using the non-invasive variables-based model: an individual has a FPG of 6.2 mmol/L (10 points), without central obesity (6 points), hypertension (12 points) and with moderate level of physical activity per week (1 point). The total points for this subject therefore are 10 + 6 + 12 + 1 = 29. The corresponding probability of regression is around 0.20.
Area under the receiver operating characteristic curve (AUC) and Hosmer–Lemeshow test were used to evaluate model discrimination and calibration, respectively, to detect regression to normoglycaemia. The optimal cut-off points in detecting regression to normoglycaemia were determined by Youden’s index (15). All statistical analyses were performed using Stata/SE 13.0 (Stata-Corp, College Station, TX). Statistical significance was set at P-value <0.05 unless otherwise indicated.
Results
A total of 1197 participants were included in the final analysis after exclusion of three patients with previously diagnosed DM at baseline. The mean age of the study population at baseline was 64.2 years (SD: 8.9) and 48.8% were males (Table 1). The average FPG level at recruitment was 6.1 mmol/L (SD: 0.3). After a mean follow-up period of 6.1 month (SD: 4.7), 180 (15.0%) IFG participants regressed to normoglycaemia and 406 (33.9%) progressed to DM according to the ADA criteria of FPG ≥7.0 mmol/L and/or 2h-OGTT ≥11.1 mmol/L (12). The average FPG level of the included participants was 5.8 mmol/L (SD: 0.6) and the mean 2h post-challenge plasma glucose level was 9.9 mmol/L (SD: 3.2) at follow-up. Compared to people who still had abnormal glycaemia status (i.e. pre-diabetes or progressed to DM), those with normoglycaemia were younger and thinner with both smaller BMI and waist circumference at baseline (Table 1). A significantly greater proportion of individuals with regression to normoglycaemia engaged in moderate or high level of physical activity. Furthermore, people who regressed from IFG to normoglycaemia were less likely to have hypertension and on average had lower FPG at baseline. Random selection divided 810 participants into the training dataset and remaining 387 participants into the testing dataset. There were no statistically significant difference between training and testing datasets of all the variables measured at both baseline and follow-up (Table 2).
Table 1.
Non-invasive clinical variables and FPG measured at baseline and FPG/2h-OGTT results at follow-up of IFG subjects who regressed to normoglycaemia and those who had persistent abnormal glycaemia status
| Total (N = 1197) | Normoglycaemia (N = 180) | Abnormal glycaemia status (N = 1017) | P | |
|---|---|---|---|---|
| Variables measured at baseline | ||||
| Demographic and lifestyle factors | ||||
| Mean age (years) | 64.2±8.9 | 62.8±8.1 | 64.4±9.0 | 0.026 |
| Male sex, n (%) | 584 (48.8%) | 92 (51.1%) | 492 (48.4%) | 0.499 |
| Family history of DM, n (%) | 395 (33.0%) | 58 (32.2%) | 337 (33.1%) | 0.810 |
| Smoking status, n (%) | ||||
| Never smoker | 953 (79.7%) | 143 (79.9%) | 810 (79.6%) | 0.948 |
| Ex-smoker | 170 (14.2%) | 26 (14.5%) | 144 (14.2%) | |
| Current smoker | 73 (6.1%) | 10 (5.6%) | 63 (6.2%) | |
| Physical activity level, n (%) | ||||
| High | 220 (18.4%) | 53 (29.4%) | 167 (16.4%) | <0.001 |
| Moderate | 791 (66.1%) | 107 (59.4%) | 684 (67.2%) | |
| Low | 186 (15.5%) | 20 (11.1%) | 166 (16.3%) | |
| Clinical measurements | ||||
| Mean BMI (kg/m2) | 25.9±4.0 | 25.2±3.8 | 26.1±4.0 | 0.005 |
| BMI group, n (%) | ||||
| Overweight | 264 (22.1%) | 42 (23.3%) | 222 (21.9%) | 0.070 |
| Obesity | 657 (54.9%) | 86 (47.8%) | 571 (56.2%) | |
| Waist circumference (cm) | ||||
| Female | 87.7±9.9 | 85.7±10.0 | 88.0±9.8 | 0.044 |
| Male | 92.0±9.1 | 89.7±8.7 | 92.5±9.1 | 0.008 |
| Central obesity, n (%) | 832 (69.7%) | 107 (59.8%) | 725 (71.4%) | <0.001 |
| Systolic BP (mmHg) | 141.6±15.7 | 141.0±14.7 | 141.7±15.9 | 0.572 |
| Diastolic BP (mmHg) | 83.6±9.5 | 84.2±9.2 | 83.5±9.6 | 0.350 |
| Hypertension, n (%) | 1149 (96.0%) | 168 (93.3%) | 981 (96.5%) | 0.049 |
| Baseline FPG | ||||
| Mean FPG at recruitment (mmol/L) | 6.1±0.3 | 5.8±0.3 | 6.1±0.3 | <0.001 |
| FPG group at recruitment, n (%) | ||||
| 5.6–6.0 mmol/L | 655 (54.7%) | 147 (81.7%) | 508 (50.0%) | <0.001 |
| 6.1–6.5 mmol/L | 409 (34.2%) | 29 (16.1%) | 380 (37.4%) | |
| 6.6–6.9 mmol/L | 133 (11.1%) | 4 (2.2%) | 129 (12.7%) | |
| Prescribed medication | ||||
| β-blockers, n (%) | 68 (5.7%) | 7 (3.9%) | 61 (6.0%) | 0.260 |
| Diuretics, n (%) | 297 (24.8%) | 35 (19.4%) | 262 (25.8%) | 0.070 |
| Statin, n (%) | 171 (14.3%) | 19 (10.6%) | 152 (14.9%) | 0.121 |
| Variables measured at follow-up | ||||
| Mean FPG at follow-up (mmol/L) | 5.8±0.6 | 5.2±0.2 | 6.0±0.6 | <0.001 |
| 2h-OGTT (mmol/L) | 9.9±3.2 | 6.3±1.0 | 10.5±3.0 | <0.001 |
| Follow-up duration (months) | 6.1±4.7 | 5.5±4.2 | 6.2±4.8 | 0.078 |
BMI: body mass index; BP: blood pressure; FPG: fasting plasma glucose; IFG: impaired fasting glucose; OGTT: 75-g 2-hour oral glucose tolerance test.
Table 2.
Comparison of non-invasive clinical variables and FPG measured at baseline and FPG/2h-OGTT results at follow-up between training dataset and testing dataset
| Training dataset (n = 810) | Testing dataset (n = 387) | P | |
|---|---|---|---|
| Variables measured at baseline | |||
| Demographic and lifestyle factors | |||
| Mean age (years) | 64.2±9.1 | 64.2±8.6 | 0.878 |
| Male sex, n (%) | 398 (49.1%) | 186 (48.1%) | 0.728 |
| Family history of DM, n (%) | 261 (32.2%) | 134 (34.6%) | 0.408 |
| Smoking status, n (%) | |||
| Never smoker | 649 (80.2%) | 304 (78.6%) | 0.154 |
| Ex-smoker | 118 (14.6%) | 52 (13.4%) | |
| Current smoker | 42 (5.2%) | 31 (8.0%) | |
| Physical activity level, n (%) | |||
| High | 144 (17.8%) | 76 (19.6%) | 0.450 |
| Moderate | 577 (71.2%) | 262 (67.7%) | |
| Low | 89 (11.0%) | 49 (12.7%) | |
| Clinical/biochemical measures | |||
| Mean BMI, (kg/m2) | 25.9±3.8 | 26.0±4.2 | 0.689 |
| BMI group, n (%) | |||
| Overweight | 182 (22.5%) | 82 (21.2%) | 0.727 |
| Obesity | 446 (55.1%) | 211 (54.5%) | |
| Waist circumference, (cm) | |||
| Female | 87.9±9.4 | 87.3±10.8 | 0.545 |
| Male | 92.0±9.1 | 92.1±9.2 | 0.987 |
| Central obesity, n (%) | 571 (70.8%) | 261 (67.4%) | 0.244 |
| Systolic BP, (mmHg) | 140.6±43.0 | 137.9±60.2 | 0.381 |
| Diastolic BP, (mmHg) | 82.7±39.2 | 80.0±55.8 | 0.339 |
| Hypertension, n (%) | 777 (95.9%) | 372 (96.1%) | 0.870 |
| Baseline FPG | |||
| Mean FPG at recruitment, (mmol/L) | 6.1±0.3 | 6.1±0.4 | 0.857 |
| FPG group at recruitment, n (%) | |||
| 5.6–6.0 mmol/L | 447 (55.2%) | 208 (53.7%) | 0.614 |
| 6.1–6.5 mmol/L | 278 (34.3%) | 131 (33.9%) | |
| 6.6–6.9 mmol/L | 85 (10.5%) | 48 (12.4%) | |
| Prescribed medication | |||
| β-blockers, n (%) | 47 (5.8%) | 21 (5.4%) | 0.793 |
| Diuretics, n (%) | 206 (25.4%) | 91 (23.5%) | 0.472 |
| Statin, n (%) | 117 (14.4%) | 54 (14.0%) | 0.820 |
| Variables measured at follow-up | |||
| FPG (mmol/L) | 5.9±0.6 | 5.8±0.6 | 0.493 |
| 2h-OGTT (mmol/L) | 9.9±3.2 | 9.8±3.2 | 0.684 |
| Follow-up duration (months) | 6.0±4.7 | 6.3±4.8 | 0.354 |
BMI: body mass index; BP: blood pressure; FPG: fasting plasma glucose; OGTT: 75-gram 2-hour oral glucose tolerance test.
Table 3 shows the results of stepwise logistic model. Individuals without central obesity or hypertension, with lower FPG and adequate physical activity, were more likely to regress from IFG to normoglycaemia. The model had a total score ranging from 0 to 50, with an optimal cut-off point of 25 yielding a sensitivity of 0.833 and specificity of 0.505. In the internal validation of model using testing dataset, the AUC of this model was 0.705 (95% CI: 0.635–0.775) (Table 4), which showed acceptable discrimination. The P-value of Hosmer–Lemeshow test was 0.840, indicating adequate calibration of the non-invasive variable-based model.
Table 3.
Factors associated with regression to normoglycaemia using non-invasive variables upon diagnosis of IFG
| Factors | Model: Non-invasive variable-based factors | ||||
|---|---|---|---|---|---|
| β-Coefficient | OR | 95% CI | P-value | Score | |
| Central obesity | |||||
| Yes | – | 1 | – | – | 0 |
| No | 0.584 | 1.793 | 1.178–2.730 | 0.006 | 6 |
| Hypertension | |||||
| Yes | – | 1 | – | – | 0 |
| No | 1.185 | 3.270 | 1.298–8.239 | 0.012 | 12 |
| FPG at recruitment | |||||
| ≥6.5 | – | 1 | – | – | 0 |
| 6.1–6.5 | 1.042 | 2.834 | 0.638–12.583 | 0.171 | 10 |
| 5.6–6.0 | 2.491 | 12.078 | 2.881–50.628 | 0.001 | 25 |
| Physical activity level | |||||
| Low | – | 1 | – | – | 0 |
| Moderate | 0.057 | 1.058 | 0.512–2.189 | 0.879 | 1 |
| High | 0.734 | 2.083 | 0.939–4.618 | 0.071 | 7 |
FPG: fasting plasma glucose; OR: odds ratio; CI: confidence interval; IFG: impaired fasting glucose.
Table 4.
Discrimination and calibration of the non-invasive model in the training and testing dataset, respectively
| AUC (95% CI) | Hosmer–Lemeshow test | |
|---|---|---|
| Training dataset | 0.732 (0.688–0.776) | 0.399 |
| Testing dataset | 0.705 (0.635–0.775) | 0.840 |
AUC: Area under the receiver operating characteristic curve; CI: confidence interval.
A nomogram was further established (Fig. 1). The results showed that the probability of regression from IFG to normoglycaemia increased gradually from 0.01 at a total score of 0 to 0.46 at a total score of 50. A detailed example of using nomogram was listed in the figure legend.
Discussion
Among a cohort study from three public primary care clinics across Hong Kong, 15.0% IFG subjects regressed to normoglycaemia over a mean follow-up period of 6.1 months. We demonstrated that the determinants of regression from IFG to normoglycaemia were absence of central obesity or hypertension, having a moderate-to-high physical activity level and lower FPG at recruitment. The model showed acceptable discriminations during internal validation. We further developed a simple-to-use nomogram based on the model to estimate the probability of regression from IFG to normoglycaemia.
The determinants of regression from IFG to normoglycaemia identified in our study were opposite to the known risk factors of progression to DM, not surprisingly. Absence of central obesity, defined by waist circumference that more accurately reflects visceral fatness and the associated risk for type 2 DM, was a key factor in determining the regression from IFG to normoglycaemia (16) rather than BMI. On the other hand, hypertension was associated with a lower probability of IFG regression in consistence with previous research that demonstrated hypertension increased DM incidence (17). The possible mechanism might be the connection between hypertension with obesity (18) and insulin resistance (19). Participating in at least 150 minutes of moderate-to-vigorous-intensity physical activity per week has been recommended by the American Diabetes Association (20), the American College of Sports Medicine and the American Heart Association (21), in order to reduce the risk of cardiovascular disease and type 2 DM. In Diabetes Prevention Program trial, intervention on increasing physical activity can effectively prevent or delay the onset of DM (4). Our findings further added fuel to the conclusion that in primary care settings, emphasizing on physical activity can increase the probability of regression from IFG to normoglycaemia.
In recent years, the responsibility for DM care and management has shifted from hospitals to primary care (22). Of the 29 million DM patients in the USA, at least 90% of them were treated by primary care physicians (23). Primary care providers thus are the key players to delay or prevent DM development by identifying individuals at increased risk and providing them with strategies to reduce the risk. Based on the model and nomogram developed in our study, primary care professionals may identify individuals with low probability of regression to normoglycaemia for further testing or interventions. The nomogram serves as an easy visual guide for primary care professionals, as well as for the subjects with IFG. Nevertheless, the cost-effectiveness of this strategy should be evaluated in the future study.
The strengths of our studies included development of a nomogram using non-invasive variables, which would be both useful and convenient for primary care professionals to provide individualized advices to control DM risk for IFG individuals. The visual guide could be easily understood by layman with IFG and thus enhance compliance. Secondly, structured and identical questionnaires were used among a large primary care sample, which minimized the measurement errors. Third, medication information was extracted from CMS through detailed chart review, which guaranteed the correctness of the information.
This study had several limitations. First, our study was conducted among Chinese primary care patients with risk factors for DM, including hypertension, obesity and positive family history of DM. The generalizability to other populations warranted further validation on external datasets. Secondly, most risk factors were continuous variables, but we categorized them as binary or categorical variables with arbitrary cut-off points to facilitate risk score calculation and clinical use. Lastly, we could not rule out the influence of other unmeasured risk factors and confounders. For example, the level of serum triglycerides was a significant determinant of DM incidence (24). However, as the aim of our study was to establish a simple-to-use tool for primary care clinical decision making, which would be applicable to patients without blood taking examination, models using laboratory-based variables may not resolve the issue. Nevertheless, the accuracy and performance of the model using laboratory-based variables should be compared with our model in the future.
In conclusion, among individuals with IFG, controlling the modifiable risk factors, including central obesity, hypertension, glucose level and lack of physical activity, increased the likelihood of regression to normoglycaemia. Although IFG is a commonly seen condition in primary care, no structured intervention or management recommendation is currently available to this group of people who are at high risk of DM in Hong Kong and many other parts of the world. Our results provided a simple tool for primary care providers and IFG subjects to estimate the probability of IFG regression and guide risk factors management. The usage of nomograms in primary care can be a time- and resources-saving strategy to tackle the increasing DM epidemic.
Declaration
Funding: Hong Kong College of Family Physicians research fellowship 2012. This study protocol was registered at the Hong Kong Clinical Trial Centre (Ref: HKCTR-1684) and ClinicalTrials.gov, the U.S. National Institutes of Health (Ref: NCT02439684).
Ethical approval: This study had received ethical approval by the institutional review board of the University of Hong Kong—the Hospital Authority Hong Kong West Cluster, Reference number: UW 13–299 and the institutional review board of Joint Chinese University of Hong Kong—New Territories East Cluster CRE, Reference number: 2013.585.
Conflict of interest: CLKL serves on the Editorial Board of the Family Practice. The authors declare no conflict of interest.
Acknowledgments
Authors would like to express sincere thanks to the medical and nursing staff of ALC GOPC, TYH RAMP clinic and LY GOPC, especially Dr Wendy Tsui, Dr WK Ko, Dr Alfred Kwong, Ms Joanna Yang and Ms KK Yeung from the Department of Family Medicine, Hong Kong West Cluster and Prof. Samuel Wong, Dr HW Li, Ms Lucia Tam and Ms Maggie Wong from the Division of Family Medicine and Primary Health Care, The Jockey Club School of Public Health and Primary Care, Faculty of Medicine, The Chinese University of Hong Kong for their generous support and facilitation in setting up and conducting this study, registered nurses Ms Herminia Tang and Ms Ida Leung for conducting participants’ assessment and venipuncture at the participating clinics and Ms Frances Kan for data entry.
References
- 1. Abdul-Ghani MA, Tripathy D, DeFronzo RA. Contributions of beta-cell dysfunction and insulin resistance to the pathogenesis of impaired glucose tolerance and impaired fasting glucose. Diabetes Care 2006; 29: 1130–9. [DOI] [PubMed] [Google Scholar]
- 2. Gerstein HC, Santaguida P, Raina P, et al. Annual incidence and relative risk of diabetes in people with various categories of dysglycemia: a systematic overview and meta-analysis of prospective studies. Diabetes Res Clin Pract 2007; 78: 305–12. [DOI] [PubMed] [Google Scholar]
- 3. de Abreu L, Holloway KL, Kotowicz MA, Pasco JA. Dysglycaemia and other predictors for progression or regression from impaired fasting glucose to diabetes or normoglycaemia. J Diabetes Res 2015; 2015: 373762. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Knowler WC, Fowler SE, Hamman RF, et al. ; Diabetes Prevention Program Research Group. 10-year follow-up of diabetes incidence and weight loss in the Diabetes Prevention Program Outcomes Study. Lancet 2009; 374: 1677–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Gerstein HC, Yusuf S, Bosch J, et al. Effect of rosiglitazone on the frequency of diabetes in patients with impaired glucose tolerance or impaired fasting glucose: a randomised controlled trial. Lancet 2006; 368: 1096–105. [DOI] [PubMed] [Google Scholar]
- 6. Perreault L, Pan Q, Mather KJ, et al. ; Diabetes Prevention Program Research Group. Effect of regression from prediabetes to normal glucose regulation on long-term reduction in diabetes risk: results from the Diabetes Prevention Program Outcomes Study. Lancet 2012; 379: 2243–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Perreault L, Temprosa M, Mather KJ, et al. ; Diabetes Prevention Program Research Group. Regression from prediabetes to normal glucose regulation is associated with reduction in cardiovascular risk: results from the Diabetes Prevention Program outcomes study. Diabetes Care 2014; 37: 2622–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Task Force on Conceptual Model and Preventive Protocols, Working Group on Primary Care and Food and Health Bureau. Hong Kong Reference Framework for Diabetes Care for Adults in Primary Care Settings; http://www.fhb.gov.hk/en/press_and_publications/otherinfo/101231_reference_framework/diabetes_care.html (accessed on January 2016). [Google Scholar]
- 9. Buijsse B, Simmons RK, Griffin SJ, Schulze MB. Risk assessment tools for identifying individuals at risk of developing type 2 diabetes. Epidemiol Rev 2011; 33: 46–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Xu Y, Wang L, He J, et al. ; 2010 China Noncommunicable Disease Surveillance Group. Prevalence and control of diabetes in Chinese adults. J Am Med Assoc 2013; 310: 948–59. [DOI] [PubMed] [Google Scholar]
- 11. Yu EY, Wong CK, Ho SY, Wong SY, Lam CL. Can HbA1c replace OGTT for the diagnosis of diabetes mellitus among Chinese patients with impaired fasting glucose? Fam Pract 2015; 32: 631–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Genuth S, Alberti KG, Bennett P, et al. ; Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Follow-up report on the diagnosis of diabetes mellitus. Diabetes Care 2003; 26: 3160–7. [DOI] [PubMed] [Google Scholar]
- 13. Craig CL, Marshall AL, Sjöström M, et al. International physical activity questionnaire: 12-country reliability and validity. Med Sci Sports Exerc 2003; 35: 1381–95. [DOI] [PubMed] [Google Scholar]
- 14. Wong MC, Jiang JY, Tang JL, et al. Health services research in the public healthcare system in Hong Kong: an analysis of over 1 million antihypertensive prescriptions between 2004-2007 as an example of the potential and pitfalls of using routinely collected electronic patient data. BMC Health Serv Res 2008; 8: 138. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Youden WJ. Index for rating diagnostic tests. Cancer 1950; 3: 32–5. [DOI] [PubMed] [Google Scholar]
- 16. Gautier A, Roussel R, Ducluzeau PH, et al. ; DESIR Study Group. Increases in waist circumference and weight as predictors of type 2 diabetes in individuals with impaired fasting glucose: influence of baseline BMI: data from the DESIR study. Diabetes Care 2010; 33: 1850–2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Conen D, Ridker PM, Mora S, Buring JE, Glynn RJ. Blood pressure and risk of developing type 2 diabetes mellitus: the Women’s Health Study. Eur Heart J 2007; 28: 2937–43. [DOI] [PubMed] [Google Scholar]
- 18. Landsberg L, Aronne LJ, Beilin LJ, et al. Obesity-related hypertension: pathogenesis, cardiovascular risk, and treatment: a position paper of The Obesity Society and the American Society of Hypertension. J Clin Hypertens (Greenwich) 2013; 15: 14–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Goff DC, Jr, Zaccaro DJ, Haffner SM, Saad MF; Insulin Resistance Atherosclerosis Study Insulin sensitivity and the risk of incident hypertension: insights from the Insulin Resistance Atherosclerosis Study. Diabetes Care 2003; 26: 805–9. [DOI] [PubMed] [Google Scholar]
- 20. Sigal RJ, Kenny GP, Wasserman DH, Castaneda-Sceppa C, White RD. Physical activity/exercise and type 2 diabetes: a consensus statement from the American Diabetes Association. Diabetes Care 2006; 29: 1433–8. [DOI] [PubMed] [Google Scholar]
- 21. Haskell WL, Lee IM, Pate RR, et al. ; American College of Sports Medicine; American Heart Association. Physical activity and public health: updated recommendation for adults from the American College of Sports Medicine and the American Heart Association. Circulation 2007; 116: 1081–93. [DOI] [PubMed] [Google Scholar]
- 22. Goyder EC, McNally PG, Drucquer M, Spiers N, Botha JL. Shifting of care for diabetes from secondary to primary care, 1990-5: review of general practices. Br Med J 1998; 316: 1505–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Davidson JA. The increasing role of primary care physicians in caring for patients with type 2 diabetes mellitus. Mayo Clin Proc 2010; 85 (suppl 12): S3–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Dotevall A, Johansson S, Wilhelmsen L, Rosengren A. Increased levels of triglycerides, BMI and blood pressure and low physical activity increase the risk of diabetes in Swedish women. A prospective 18-year follow-up of the BEDA study. Diabet Med 2004; 21: 615–22. [DOI] [PubMed] [Google Scholar]