Temporal changes in unadjusted prevalence of IFG and diabetes among Kuwaiti adults in the 8 years since 2002 show a significant decrease in both males and females. When adjusted for age, BMI, HC and education level, logistic regression analysis showed that, compared with 2002–03, both males and females remained significantly less likely to be diabetic and/or to have IFG in 2008–09 and thus confirmed the declining trends in both IFG and Diabetes among Kuwaiti adults. Time trend data for the region that can be used for comparison: a study among Iranians showed an increased trend for diabetes from 2005 to 2007 , as did data from Oman for 1991–2000, Saudi Arabia for 1993–2000, and UAE for 1995–2000 . However, these studies represented earlier data collection period comparable to the earlier part of our study except Iranian study (2005–07).
The overall crude prevalences of IFG and diabetes were significantly higher in males than females. When the data were stratified by study periods, the prevalence of IFG in males was significantly higher than females only in 2004–05, and for diabetes, it was higher in males than in females in 2002–03 and 2008–09. The results of linear regression showed a greater decrease over time in males and closer values between the genders in 2008–2009. When logistic regression was carried out with combined genders, after adjusting for various factors, females were less likely to have IFG (OR: 0.65; 95% CI: 0.55-0.77) and/or diabetes (OR: 0.63; 95% CI: 0.53-0.75) than males. A similar gender difference in diabetes has recently been reported in Kuwaitis , though a lower prevalence of diabetes was reported in Kuwaiti males in an earlier study . Studies from other countries in the region also reported mixed results. Men had higher prevalences of diabetes in Saudi Arabia , Oman , Yemen  and Jordan , while women had higher prevalences in Iran , UAE  and Bahrain . These variabilities between countries in diabetes with gender may reflect the variation in other factors that can influence diabetes including patient identification. Further, in the present study we found a significant interaction of education level and gender with the prevalence of IFG and diabetes. Thus, it is also possible that some of the “confounders” that we adjusted for in our analysis altered the apparent gender differences.
In the present study we also explored the association of selected socio-demographic (age, education), biological (BMI and serum TC) and lifestyle (exercise and smoking) factors. As observed in earlier studies among Kuwaitis [12, 17], and as would be expected, age was found to be an important predictor of IFG and diabetes prevalence for both genders in the present study. The odds of IFG and diabetes increased with age until the 6th decade in both genders. A similar age related change in diabetes has also been reported in other studies in the Arab Gulf region [16, 18, 22, 24].
In the present study, both overweight and obese males and females were more likely to have IFG and diabetes which are in accordance with the findings from other countries in the region [9, 18, 25–28] where a consistently positive association of overweight and obesity with diabetes is demonstrated. It is known that adipocytes (fat cells) secrete a number of adipocyte hormones and adipokines, which may in turn increase the risk of diabetes via several pathways such as increasing insulin resistance .
Unadjusted prevalence of diabetes was significantly higher in both genders with moderate-HC and high-HC than in individuals with normal cholesterol level. After adjusting for confounding factors in logistic regression, the OR for diabetes in males (OR: 1.32; 95% CI: 0.95-1.83) and females (OR: 1.22; 95% CI: 0.89-1.66) were higher in those with high-HC than the subjects with normal cholesterol level, however they did not reach the level of significance.
More highly educated females were significantly less likely to have IFG or diabetes in our study, but only highly educated (undergraduate or above) males were less likely to be diabetes. An earlier study in Kuwaiti adults also showed that higher education level was associated with a significant reduction in blood glucose levels . Furthermore, a lower education level has been found to be associated increased likelihood of diabetes among Bahrainis , Omanis  and Iranians . It is possible that those who were more highly educated may have greater awareness of the risk factors of hyperglycemia and diabetes, and thus more likely to practice more effective preventing measures. However, higher education had no significant impact on IFG prevalence in Kuwaiti males implying that they are either less aware of the risk factors of hyperglycemia.
The prevalence rates indicate that both males and females who were exercising were possibly less likely to be diabetic than those who were not exercising. However, the results (odds ratio) were not statistically significant when adjusted for other confounders including BMI supporting excess body weight as the primary risk factor. Kuwaitis who participate in even moderate exercise have been shown to have delayed weight gain with age . In a recent study we have also reported that the Kuwaiti males who exercised were significantly less likely to be obese . Since we do not have any information on the duration and intensity of the exercise it is difficult interpret the present findings.
We found that both males and females who were smokers were significantly less likely to have IGF than non-smokers even when adjusted for age and BMI, though no significant association between smoking and diabetes was observed. Studies from UAE  and Qatar  showed that smokers were more likely to be diabetic than non-smokers, while a longitudinal study in Iranian adults found no association with smoking and incidence of diabetes . It is important to note that most prospective studies have shown higher risk of diabetes for smokers, especially those who smoke more than 1 packet per day . Unfortunately, in the present study we do not have information on the frequency/intensity and duration of the smoking. Further, it should be noted that smoking may not be a protective for IFG. Since the study is cross sectional, no cause and effect could be determined.
This study has a number of limitations. The prevalences of IFG and diabetes were estimated by single measurement of blood glucose using Accutrend GCT, which could introduce some errors. However, given that the survey included a relatively large population-based sample it is unlikely that the true prevalence will be different than that we have reported here. Further, in the present study those with known diabetes (those taking oral agents or insulin) were excluded. It is possible that this number has been increasing over time with more attention to diabetes and its treatment by Physicians, and thus our estimation of diabetes might have underestimated the prevalence in recent years, but the prevalence is still alarmingly high. Given its cross-sectional nature no causal relationship of socio-demographic and lifestyle factors with IFG and diabetes prevalence can be established. There was no information on the frequency/intensity and duration of the smoking and exercise pattern. More carefully designed studies that capture lifetime exercise, smoking and food intakes are needed to explore these potential associations between lifestyle factors and hyperglycemia. While efforts were made to obtain representative samples, the KNSS uses convenience sampling and thus the results of the present may not be representative of the wider population. Nevertheless, the main strength of this study is that it includes a relatively large population-based sample so it is likely that the true situation will not be different, on the whole, than reported here.