The Importance of Diabetes Mellitus in the Global Epidemic of Cardiovascular Disease: The Case of the State of Qatar

Abstract

As a manifestation of the epidemiologic transition being experienced throughout the developing world, the prevalence of diabetes mellitus (DM) is increasing. However, whether an individual's risk of cardiovascular diseases as a consequence of DM is also higher in these countries is unknown. We conducted a case-control study at the medical center in the state of Qatar comparing the prevalence of DM in 512 patients who were admitted with acute myocardial infarctions (MI) and 262 cases of cerebrovascular accidents (CVA) to 382 hospital and outpatient controls to calculate the odds ratios (OR) associated with DM for MI and CVA. The OR for MI was estimated to be 4.01 compared to 2.92 for other countries in the Middle East and 1.75 for North America. The OR was even higher for Qatari natives. Understanding the reasons for this increase, including genetic differences, lifestyle, and medical management issues, is critical for the design and prioritization of effective interventions.

INTRODUCTION

Cardiovascular diseases (CVD) are a leading cause of death worldwide. The World Health Organization (WHO) estimated CVD mortality to be 17.1 million people in 2004 with 42% of these deaths attributed to coronary artery disease and 33% to stroke (1). Mortality from CVD is predicted to reach 23.6 million people worldwide in 2030, with the largest increase occurring in the Eastern Mediterranean region (1).

This is, in part, a manifestation of the accelerated epidemiological transition taking place in the Middle East and North Africa (MENA), driven by the high rates of urbanization and lifestyle changes, and by the rapidly aging population (2, 3). High-income countries, particularly in the Arabian Gulf region, have seen major shifts in their demographic and lifestyle profile with reduced physical activity and increased consumption of calories and fat rich diets contributing to this trend (4, 5). Consequently, hypertension, insulin resistance, diabetes, hyperlipidemia, smoking, and metabolic syndrome, have all increased (2, 5, 6). For example, the prevalence of diabetes is estimated to exceed 15% in Qatar as compared to 11% in the United States (1). And the trend is predicted to continue: by 2030, the number of people in Qatar with diabetes is expected to increase by 130% (1). These changes are also accompanied by rapid population growth and an influx of low-income multi-ethnic workers, where in some countries such as Qatar, migrant workers constitute at least 73% of the total population (7). As a result, these countries are challenged by a major and increasing burden of cardiovascular diseases (8).

The number of hospital admission for coronary heart disease (9), acute myocardial infarction (10), stroke (11–13), and heart failure (14) in Qatar is substantial. Although several studies have attempted to examine the characteristics of patients admitted with CVD in this nation (9–14), these were based on convenience samples, and none of them involved a comparison to healthy controls. We conducted a systematic case-control study aimed at quantifying the strength and importance of diabetes as a risk factor for CVD in Qatar relative to other known risk factors associated with CVD and relative to other countries in the region and elsewhere in the world. In an attempt to understand the epidemiologic transition more completely, we asked whether, in addition to its higher prevalence, diabetes was also a more virulent risk factor for CVD. That is, whether a person with diabetes in Qatar is at a higher risk for developing CVD than a diabetic elsewhere.

METHODS

The primary aim of this case-control study was to estimate the strength of the association between preventable risk factors and the development of myocardial infarction (MI) or cerebrovascular accidents (CVA). We were interested in the odds ratios (ORs) for MI and CVA for diabetes, hypertension, dyslipidemia, smoking, and obesity, with a special emphasis on their association with diabetes; controlling for age, gender, and smoking.

Selection of Cases and Controls

Cases were recruited from Hamad Medical Corporation (HMC) in Qatar and included both Qatari nationals and non-Qatari expatriates admitted with incident MI or CVA during the period of the study. Controls were selected from inpatient and outpatient departments at HMC including the staff, dermatology, and orthopedic clinics. Controls were also from the outpatient departments at Rumailah Hospital, located close to the main campus of HMC. The choice of sites for the selection of cases and controls was dictated by the fact that HMC is the main public hospital that serves the population of Qatar, and it is the sole health care facility that is fully equipped to address MI and CVA in that region. Ideally, in case-control studies, controls should be selected from the base of people who would have been admitted to HMC had they been diagnosed with the disease of interest, i.e., MI/CVA. Because HMC is the sole hospital providing care for acute MI and CVA in Qatar, it is reasonable to assume that inpatient and outpatient controls would have been admitted to HMC had they been diagnosed with MI or CVA (and thus captured in the case groups). Therefore, our controls should be representative of the population of Qatar who may have developed MI or CVA and been admitted to HMC during the study period. This advantage allowed for the use of hospital inpatients and outpatients (HMC inpatient and outpatient departments; Rumailah Hospital outpatient departments) as a source for controls. Furthermore, many of the conditions/diagnoses for which patients are seen at each of these sites, such as orthopedics and dermatology, are unlikely to be related to known or potential risk factors for MI or CVA and, therefore, they constitute appropriate diagnoses for the selection of the control group for this study. Cases and controls younger than 30 years of age were excluded from the study. Controls with a history of heart disease or CVA events were similarly excluded.

Sampling Procedures

All patients admitted to HMC for MI or CVA between June 2006 and June 2008 were selected to potentially participate in the study. Identification of MI and CVA cases was performed by a review of the HMC hospital admission logbook, which was maintained in the HMC admission's office. The diagnosis of MI was confirmed by reviewing medical records for the peak troponin T and CK-MB (creatine kinase-muscle) levels for all enrolled MI cases, whereas CVA cases were ascertained using magnetic resonance imaging (MRI), computed tomography (CT) scans, and neurological examinations. All MI cases had detectable elevated peak troponin T values (median, 3.90 ng/mL; range, 0.03 to 63.88 ng/mL) and elevated peak CK-MB values (median, 106.45 ng/mL; range, 2.54 to 893.10 ng/mL). Enrolled CVA cases were admitted with a recent ischemic or hemorrhagic infarct as identified by MRI, CT scans, or symptoms characteristic of a neurological deficit (i.e., cranial nerve deficit, upper/lower extremity weakness, upper/lower extremity sensory loss, speech and/or swallowing deficit, etc.).

Inpatients eligible for control recruitment included those who were admitted for gastrointestinal, urological, or orthopedic conditions. During each study day, a random sample of 10 eligible control patients was drawn from the HMC hospital's admission logbook using a random number generator. No further attempts were made to contact control patients who were randomly selected but were not available in their rooms for the interview.

Controls recruited from the outpatient staff clinic at HMC included those who were at the clinic for sick-leave requests, pre-employment screenings, follow-up visits, vaccinations, and prescription renewals. The appointment lists were collected 1 day in advance, and 40 subjects were randomly selected for interview. Subjects who were selected to participate in the study but did not show up for their appointment were not pursued further for recruitment.

Data Collection

Data collection took place between June 2006 and June 2008 for cases. Data collection was extended until October 2008 for controls. Tools for gathering information included a structured interview, a review of the patient's medical record, and blood and urine specimens.

The patient interviews and the reviews of the medical charts were performed by research assistants who were employees of Weill Cornell Medical College - Qatar (WCMC-Q), the institution conducting the study. All of the research assistants had clinical research training and a good command of Arabic and English, and some spoke Hindi. Specialized nurses at HMC and trained research assistants with nursing backgrounds were responsible for drawing blood and obtaining urine samples if the required tests were not already available.

When possible, research assistants and patients were paired for data collection in a gender-appropriate manner, consistent with Qatari cultural norms; that is, female patients being interviewed by female research assistants. Proxy respondents were interviewed for instances in which patients had severe medical illnesses or profound neurological deficits (including stroke-related disabilities such as aphasia or dependence on a ventilator). Proxies were usually adult relatives designated by the study participant to represent him/her in the study.

The interviews required approximately 20 minutes to complete. Controls participating in the study received a minimal sum as compensation for their time.

Ethical Considerations

The study was approved by the Institutional Review Board of HMC and Weill Cornell Medical College in New York. Written informed consent was collected from all study participants, including proxy respondents. In some instances, a close relative such as the head of the family (husband, father, brother, or son) signed the informed consent jointly with a female participant to avoid cultural sensitivities. Written informed consent was also obtained from the attending physicians of eligible patients. Physicians could view any additional study-related laboratory results performed for their patients.

Study Instruments

All questionnaires and consent forms were available in Arabic, English, and Hindi to minimize language barriers. The structured interview assessed the participant's socio-demographic background, family history, medical history, symptoms that prompted hospitalization (for cases), lifestyle characteristics (smoking, diet, and physical activity), quality of life, and depression. The medical record review provided patients' height, weight, heart rate, blood pressure, waist circumference, diagnosis, prescribed medications, results of blood and urine tests, symptoms before admission (for cases), results of electrocardiograms and other procedures (including cardiac catheterization, MRI scans, and CT scans; for cases), as well as results of neurological examinations (for CVA cases). Height and weight information were used to calculate the body mass index (BMI).

Sample Size

The sample size was based on the estimated odds ratios (ORs) for the association between the five hypothesized risk factors (i.e., diabetes, hypertension, dyslipidemia, smoking, and obesity) and the occurrence of MI or CVA. A sample size of 400 patients with MI, 200 patients with CVA, and 400 inpatient/outpatient controls would allow for the detection of ORs ≥ 2.4 and 2.8 for MI and CVA, respectively, assuming an alpha level of 1% (two-sided test) and a power of 80%. An alpha level of 1% was selected to be more conservative because of the multiplicity of hypothesized risk factors under investigation. In addition, 400 MI and 200 CVA cases would allow for the detection of an expected risk factor prevalence of 30% (among cases) with a 95% confidence interval (95% CI) width of 9% (i.e., ± 4.5%) for MI cases and 12.8% (i.e., ± 6.4%) for CVA cases.

Statistical Analysis

Descriptive statistics were performed to characterize study participants. Unadjusted ORs and 95% CIs were calculated to estimate the association between the hypothesized risk factors of interest (i.e., diabetes, hypertension, dyslipidemia, smoking, obesity, and metabolic syndrome) and the development of MI or CVA. Metabolic syndrome was defined using variables available in our study as three or more of the following four factors: history of diabetes, history of hypertension, history of high cholesterol, and waist circumference (≥ 35 inches for females or ≥ 40 inches for males). The manuscript describing the full analysis, including all lifestyle and socio-demographic variables, is currently under review and is available from the author. Multivariate logistic regression analyses (performed separately for MI and CVA outcomes) were performed to examine the association between the corresponding risk factors and MI or CVA events after adjusting for age, sex, and smoking (i.e., models controlling for three covariates). Waist circumference was used instead of BMI to assess obesity in the final multivariate models because of the high degree of collinearity between waist circumference and BMI (Pearson correlation coefficient 0.68 among MI cases/controls and 0.62 among CVA cases/controls). All analyses were completed using SAS Version 9.2 (SAS Institute Inc., Cary, NC), SPSS Version 19.0 (SPSS Inc., Chicago, IL), and Stata Version 11.0 (StataCorp, College Station, TX).

RESULTS

From June 2006 to June 2008, a total of 1,044 MI and 634 CVA cases were admitted to HMC that were eligible for participation in this study. Of those cases, 512 (49.0%) with MI and 262 (41.3%) with CVA were enrolled in the study. The main reasons for non-participation of eligible MI and CVA cases were the inability of the research team to contact the patient (19.6% and 13.7%, respectively), or the unavailability of a proxy-respondent for patients with severe medical conditions (10.5% and 17.8%, respectively). Only 12.7% of MI cases and 10.7% of CVA cases refused to participate in the study. Eligible MI and CVA cases that were enrolled tended to include younger, non-Qatari males as compared to non-enrolled MI and CVA cases. Comparative analyses between inpatient and outpatient controls also revealed a different socio-demographic profile. Inpatients comprised 53% of the total controls, and tended to be slightly older and more likely to be male as compared to outpatient controls. The majority of outpatient controls were selected from the HMC staff clinic (67%), whereas the rest were recruited from the dermatology and orthopedic clinics, or outpatient departments at Rumailah Hospital.

More than two thirds of MI and half of CVA cases were younger than 55 years of age (Table 1). Approximately 90% of the controls were younger than 55 years of age. Males accounted for a large proportion of MI (92%) and CVA (81%) cases, whereas approximately half of the controls were male (56%). MI and CVA cases were, respectively, 8.8 and 3.4 times more likely to include males compared to controls. Approximately one quarter of the MI and CVA cases had parents who were related.

TABLE 1.

Characteristics of patients with MI and CVA

Characteristic	MI Cases (N = 512) %	CVA Cases (N = 262) %
Age (y)
≤ 55	69	54
> 65	9	23
Gender
Male	92	81
Parents
Related	25	26

MI and CVA cases were much more likely than controls to report being diagnosed with diabetes and to have high plasma glucose levels upon hospital admission (or at time of the outpatient visit for controls) (Table 2). Hypertension was the second most important risk factor in the univariate analysis. MI cases were also more likely to report a history of hypertension and to have elevated blood pressure upon hospital admission compared to controls. MI and CVA cases also more frequently had a history of high cholesterol, elevated plasma lipids on admission, and a history of smoking as compared to controls. Approximately a third of cases and controls had elevated waist circumferences. Sixteen percent and 28% of the MI and CVA cases, respectively, met our criteria for metabolic syndrome as defined by the variables available to us as opposed to controls, of which only 4.5% fulfilled these criteria.

TABLE 2.

Number and Percentage of Risk Factors in MI and CVA Cases and Controls

Primary Risk Factor	MI Cases (N = 512) %	CVA Cases (N = 262) %	Controls (N = 382) %
History of diabetes	40	52	9
Plasma glucose level ≥ 200 mg/dL (11.0 mmol/L)	42	37	6
History of hypertension	44	64	19
History of high cholesterol	13	19	7
Elevated plasma lipids	41	52	29
Smoking (ever)	60	40	29
Elevated waist circumference (≥ 35 inches for females or ≥ 40 inches for males)	30	39	38
Metabolic syndrome	16	28	4.5

To assess the relative importance of the underlying risk factors for MI and CVA, we performed multivariate analyses controlling for age, sex, and smoking (Table 3). This revealed that a history of diabetes (adjusted OR, 4.01) was the strongest predictor of MI, followed by a history of high cholesterol (adjusted OR, 2.81), a history of hypertension (adjusted OR, 2.34), and past or current smoking (adjusted OR, 2.37). The strength of the association of metabolic syndrome with acute MI and CVA was roughly similar to that for diabetes alone. Analogously, diagnosed diabetes (adjusted OR, 5.12), followed by a history of hypertension (adjusted OR, 3.59), and a history of high cholesterol (adjusted OR, 2.45), were the main risk factors for CVA (Table 3).

TABLE 3.

Odds Ratios of Risk Factors for MI and CVA (Adjusted for Age, Sex, Smoking)

Primary Risk Factor	Adjusted OR
	MI	CVA
History of diabetes	4.01 (2.51–6.42)	5.12 (3.11–8.60)
History of hypertension	2.34 (1.54, 3.59)	3.56 (2.29, 5.66)
History of high cholesterol	2.81 (1.48, 5.55)	2.45 (1.27, 4.83)
Smoking (ever)	2.37 (1.65. 3.43)	1.34 (0.86, 2.09)
Elevated waist circumference (≥ 35 inches for females or ≥ 40 inches for males)	1.40 (0.92, 2.12)	1.05 (0.64, 1.72)
Metabolic syndrome	4.04 (2.02, 8.10)	4.74 (2.32, 9.69)

Because of the small sample size, only exploratory analyses on the Qatari nationals were performed to evaluate potential risk factors for MI (n = 66 cases) and CVA (n = 65 cases) specific to the native Qatari population (relative to Qatari controls [N = 65]). Overall, the medical profile of Qatari MI and CVA cases revealed that close to 70% were diagnosed with diabetes and hypertension, and more than 30% had a history of elevated cholesterol, a history of smoking, and an enlarged waist circumference. Diagnosed diabetes and hypertension prevailed as very strong risk factors in native Qataris, with MI being approximately six times more likely in Qatari national diabetics than controls (Table 4). The OR associated with a history of hypertension was 3.62. A history of high cholesterol levels was also a strong predictor of MI among Qataris (OR, 1.91). A history of smoking did not appear to be an independent risk factor, although the confidence limits were wide. The risk associated with an enlarged waist circumference in the Qatari nationals sub-population was modest; however, the magnitude of the risk for the full metabolic syndrome (defined as having any three of the four conditions) was high at OR = 4.85. When contrasted with the overall population, the ORs for diabetes were higher in the Qatari nationals. However, because of the small sample sizes, the 95% confidence limits were wide and overlapping. The same is true for hypertension and for metabolic syndrome (Table 4). The patterns and comparisons are very similar for the associations of risk factors with CVA (Table 5), but the differences between Qatari nationals and the overall population was less than that for MI (Table 5).

TABLE 4.

Odds Ratios of Risk Factors for MI in Qatari Nationals Compared to Overall Population

Risk Factor (ORs Adjusted for Age, Sex, Smoking)	Overall Population	Qatari
History of diabetes	4.01 (2.51–6.42)	6.03 (1.72–21.14)
History of hypertension	2.34 (1.54, 3.59)	3.62 (1.01, 13.04)
History of high cholesterol	2.81 (1.48, 5.55)	1.91 (0.49, 7.35)
Smoking (ever)	2.37 (1.65, 3.43)	0.47 (0.10, 2.27)
Elevated waist circumference (≥ 35 inches for females or ≥ 40 inches for males)	1.40 (0.92, 2.12)	1.27 (0.35, 4.66)
Metabolic syndrome	4.04 (2.02, 8.10)	4.85 (1.20, 19.35)

TABLE 5.

Odds Ratios of Risk Factors for CVA in Qatari Natives Compared to Overall Population

Risk Factor (ORs Adjusted for Age, Sex, Smoking)	Overall	Qatari
History of diabetes	5.12 (3.11–8.60)	5.90 (1.62–21.61)
History of hypertension	3.59 (2.29, 5.66)	3.03 (0.89, 10.37)
History of high cholesterol	2.45 (1.27, 4.83)	1.90 (0.58, 6.95)
Smoking (ever)	1.34 (0.86, 2.09)	0.98 (0.21, 4.56)
Elevated waist circumference (≥ 35 inches for females or ≥ 40 inches for males)	1.05 (0.64, 1.72)	0.56 (0.15, 2.08)
Metabolic syndrome	4.74 (2.32, 9.69)	4.77 (1.17, 19.41)

Table 6 shows a comparison of the ORs found in our study (first column) to those in other populations. The importance of diabetes as a risk factor for MI appears to be higher for the overall population of Qatar than that found in the INTERHEART study for the counties of the Middle East Crescent (MEC) and much higher than North America (NA) (15). The risk implications of hypertension also appear to be higher in Qatar whereas elevated cholesterol and enlarged abdominal girth alone may be less important than elsewhere. As with other countries in the region, smoking is an important risk factor for MI in Qatar.

TABLE 6.

Comparison of Odds Ratios for MI in Qatar to Elsewhere

Risk Factor	ORs
	WCMC-Q/HMC Study	Middle East Crescent	North America
History of diabetes	4.01 (2.51–6.42)	2.92 (2.26–3.79)	1.75 (0.94–3.29)
History of hypertension	2.34 (1.54, 3.59)	1.84 (1.45, 2.33)	1.58 (1.01, 2.47)
History of high cholesterol	2.81 (1.48, 5.55)	5.33 (3.48, 8.18)	4.75 (.134, 16.86)
Smoking (ever)	2.37 (1.65, 3.43)	2.64 (2.19, 3.19)	1.82 (1.14, 2.88)
Elevated waist circumference (≥ 35 inches for females or ≥ 40 inches for males)	1.40 (0.92, 2.12)	1.85 (1.43, 2.47)	4.68 (2.57, 8.50)

DISCUSSION

Diabetes is the largest preventable risk factor for MI and CVA among the highly diverse population of Qatar. Diabetic individuals were estimated to be four times more likely than non-diabetics to develop MI adjusted for age, sex, and smoking (Table 3). Hypertension was the second highest preventable risk factor for CVA and was also an important predictor of MI. Our exploratory analyses of risk factors in the sub-group of Qatari nationals suggested similar trends, but there were higher odds ratios for diabetes and hypertension. Our findings are, in many respects, comparable to the results of the INTERHEART study, a standardized case-control study assessing the determinants of acute MI across 52 different countries around the globe (15). These results are also consistent with a descriptive study conducted among MI patients in the United Arab Emirates (UAE) (16). Qatar and UAE share comparable demographic characteristics and are classified among the most diversified populations in the world, with expatriates and migrant workers accounting for more than 70% of the population (7, 16). Specifically, in our study, a substantial proportion of MI cases included men from South East Asia, employed as workers, with minimal education and a low monthly income (53%), whereas Qataris constituted only 13% of MI cases. Moreover, most patients with MI in our study and from other countries in the region in the INTERHEART study were younger than 55 years of age, with a large fraction being younger than 40 years of age (12.0% and 12.6%, respectively) (10, 15). This high percentage of young people less than 40 years of age among those with MI is striking and is one indication of the urgency for establishing population-based awareness campaigns and prevention programs.

Although the risk factor patterns we found are similar, the OR for MI in Qatar were higher than the age/sex/smoking-adjusted ORs reported for MEC in the INTERHEART study in which diabetic individuals had approximately three times the risk of developing MI (Table 6) (15) and from the UAE (16). Importantly, the ORs estimated in Qatar for diabetics was more than twice those reported for developed countries, such as NA (Table 6) (15).

Unlike diabetes and hypertension, elevated cholesterol levels did not appear as strong of a risk factor for MI in Qatar as elsewhere in the INTERHEART study (15). This might have been because our assessment was based on a reported or documented history of high total cholesterol rather than measured plasma lipids. Similarly, our estimate of the prevalence of elevated total cholesterol among CVA patients (19.2%) was also less than that reported in the other Arab Gulf States that ranged between 29% in Bahrain and 61% in Kuwait (6).

Smokers had more than twice the risk of developing MI as compared to non-smokers (Table 3). Smoking was an even more pronounced risk factor for MI in MEC (Table 6), whereas in European regions, reported ORs were in line with our estimate (15). Our findings also agreed with results of another descriptive study based on the review of medical records for MI patients in Qatar (10). The prevalence of smoking among patients admitted with MI in our study was also less than that measured among MI patients in UAE (59.5% versus 73.7%, respectively) (16).

Interestingly, although abdominal obesity increased the risk of MI by almost five times in NA, and two times in MEC, abdominal obesity was not associated with MI or CVA in our study (Table 6). The reason may be the high overall prevalence of people who are overweight or obese in Qatar. Indeed, a recent cross-sectional study among Qatari adults revealed that 32% of Qatari adults are overweight and another 47% are obese (17). Other studies conducted among Qatari adults suggested a prevalence of obesity in the range of 29.8% to 65.8% (19, 18–21). In our study, the prevalence of obesity was approximately 30% for cases and 34% for controls (Table 2), and was even higher among Qatari nationals in whom it was 34% to 38% for cases and 54% for controls. These observations provide a plausible explanation for the absence of statistically significant associations between obesity and MI or CVA; however, they do not reduce the importance of obesity as a risk factor to be addressed at the national level. This is also suggested by the magnitude of the ORs for metabolic syndrome in which obesity combined with hypertension, hyperlipidemia, or diabetes may be accounting for this finding (Table 3). The somewhat lower ORs for metabolic syndrome than for diabetes alone may also be a reflection of the high prevalence of obesity in the controls, thereby decreasing the OR for metabolic syndrome somewhat. Apparently, the prevalence of obesity in other Arab Gulf countries, such as SA and the UAE, is similar to Qatar (22).

Of course, some of the differences in the ORs (for diabetes and MI or CVA) across settings may be due to variations in study designs or research methodology used to estimate these parameters. However, they may also be explained by population genetics and demographics, the phases of the epidemiological transition across countries, the adequacy of medical management, and the extent of public health interventions and prevention strategies used (8). Understanding the reasons for these differences is critically important and should be a priority for future research (23). The design and selection of interventions most likely to be effective depends on it.

Our study was not without limitations. As in all case-control studies, selection bias might be an issue if the eligible MI/CVA patients who chose not to participate in the study had a lower prevalence of risk factors compared to those who did. This could lead to an overestimation of the association between the risk factors of interest and MI/CVA occurrence. Especially important for this estimation is the representativeness of our control group as a reflection of the overall population of Qatar.

We assessed this by comparing the prevalence of the five primary risk factors for MI and CVA events in our Qatari control group (N = 65) to the population of Qatari nationals (as retrieved from published cross-sectional studies). The prevalence estimates for the majority of the primary preventable risk factors among Qatari controls fell in the range of estimates retrieved from other studies in the literature (17, 19–22, 24–28). The general similarity between most of the preventable primary risk factors in our Qatari control group and those reported from other studies supports the validity of our findings. Furthermore, although case-control studies are not optimal for establishing causality, the inclusion of only incident cases of MI and CVA ensures the temporal sequence of events.

In conclusion, the risk factors for CVD in the population of Qatar are similar to other countries undergoing the epidemiological transition in the rest of the Middle East as well as in other parts of the world. However, the higher risk associated with diabetes stands out as of special importance in Qatar. There is an urgent need for population-based public health strategies to address and alleviate the incidence and burden of diabetes as well as the risk factors leading to it in Qatar. Designing population-level prevention interventions with wide-ranging awareness campaigns and supporting a culture of preventive health in Qatar are critical for both Qatari nationals and the expatriate population.

DISCUSSION

Cohen, Washington: Al, thank you very much. That's really a very exciting result from, I'm sure, the first of a long series of very interesting insights that are going to come from the research work that is being done in Qatar. I am as enthusiastic as you are about the potential outcomes of this collaboration because I think it does demonstrate what can be done by bringing Western US science into a part of the world that really has not had access to that kind of expertise, and what it does not only for that country in terms of understanding the epidemiology and the management of the very common disease, but also insights which will also, I think, inform us about the general nature of these problems. I wanted to ask you a stupid question. You know that the majority of the population in Qatar is ex-patriots. It's only a minority that are actually native Qatari, and that the subset that are native Qatari have this exceptional observation in terms of diabetes control. What explains the fact that the ex-patriots are also experiencing a greater risk of these complications from diabetes?

Mushlin, New York: First of all, Jordy, I want to thank you for your question and your comments. Your own involvement in Qatar is much appreciated. I know you have a lot of insights and knowledge about the country. With regard to the dual or bimodal population, I think there may be two separate explanations. We may well find lifestyle and management issues in the worker population and a genetic profile that's of real interest in the Qatar natives. Hopefully, as you suggest, and I know Ron Crystal is looking forward to this as well, insights from a genetic profile will be applicable not only to the Qatari population but of importance in terms of understanding the biology of the disease and its complications.

Zeidel, Boston: Wonderful talk. A comment and question. One hopes, and I think you're doing this, which is a great plus that you're developing an indigenous capability in the region to do these kinds of studies and, I think it's crucial, to guide policies and other things. The second thing is I noticed you started at the central hospital, does Qatar, like Kuwait and the UAE, have a system of articulated clinics that are free clinics for people to go to? Because it would seem to me that to the extent that you could link those clinics in and develop over time the ability to look at what's going on in those clinics and evaluate patients at that stage, that's where I think the difference will be made.

Mushlin, New York: There is an extensive primary care network of health centers in Qatar and, in fact, for our subsequent studies, we plan on sampling patients from those health centers.

Crowley, Boston: Two questions, which may be related a little bit to the genetics of this. Number one, could you be a little more specific about the control group? Because you referred to the control group but didn't give their characteristics and then refer to the overall population. I am specifically looking for the degree of the endogamy in those matched. And the second related question is: what did you make of the fact that the footprint of lipids and centripetal obesity seem to be less in the Qataris than anybody else? So, they may be related questions. You know, half of research is getting those controls to come out right, so the key is, who are those controls and how well are they matched.

Mushlin, New York: Absolutely. That's a very insightful question and one that has been a subject of a lot of our thinking about the data and its robustness, if you will. I will try to short-cut my answer for all of this. All the information that we have suggests that the data in our control group looks similar to what we know about the overall population, which should enhance the robustness of these results. We sampled patients for controls from people coming into the walk-in clinics and from people admitted to services where we did not feel that there would be any bias with regard to the prevalence of cardiovascular or vascular disease in general. So, our inclusion and exclusion criteria for the controls, I think, were correct and the data that we have indicates that the control group is reflective of the overall population. But as you know, I think the main threat to the case control design is the question that you asked. The other point is that the comparative data that I showed you, particularly from the Middle East, came from a similar study design, so it's also from a case control study. Therefore, even if the absolute odds ratios are off a bit, the comparative data should be similar.