Predictors Of Mortality

The Big Diabetes Lie

Diabetes can be Reversed

Get Instant Access

Many factors are documented to increase the risk of cardiovascular mortality in young diabetic patients (Rosengren et al 1989; Rossing et al 1996), but very few studies have addressed this topic in older patients (Ford and De Stefano 1991). Since the strength of predictors of mortality can change with age (Frost et al 1996), further studies are needed.

Recognition of the predictors of mortality is the first step in planning an intervention aimed at reducing mortality. Predictors can be divided into unmodifiable and modifiable factors. The former include age, gender and family history; the latter include cigarette smoking, high blood pressure, high blood glucose, elevated total and low-density lipoprotein cholesterol, obesity, diabetes treatment, model of diabetes care and so on. For the purpose of this review it is more useful to classify these predictors into the classic, which are shared by both the diabetic and non-diabetic population, and the diabetes-specific, which are specifically correlated with the disease, its natural history, treatment and complications (Table 8.3).

Classic Cardiovascular Risk Factors

When evaluating cardiovascular risk factors, age represents the strongest predictor of mortality: ''the older

Table 8.3 Predictors of cardiovascular mortality in Type 2 diabetes

Classic Age and sex

Lipid, blood pressure, BMI Smoking, lifestyle etc. Diabetes-specific

Duration, age of onset, severity of the disease Treatment and level of care Long-term glucose control (i.e. FPG, HbAjc Long-term glucose instability (FPG, post-prandial peaks, recurrent hypoglycemia)

the subject the higher the absolute risk of dying''. Moreover, age often behaves as a confounder or as an effect modifier when assessing the importance of other risk factors.

The leading cause of death in the Verona diabetes study was cardiovascular disease, which accounted for 42% of the overall mortality. As previously observed for all-cause mortality, the overall cardiovascular mortality was higher in men than in women at all ages, both in the diabetes cohort and in the general population. However, when the SMR was considered, the impact of diabetes on mortality was higher in women than in men, especially in the 65-74 age group (De Marco et al 1999). The 16-year follow-up of the Fra-mingham study demonstrated an equal risk among diabetic men and women in terms of cardiovascular morbidity, but mortality was more pronounced in diabetic women (Kannel and McGee 1979). The higher mortality rate in women from all causes and from ischaemic heart diseases was also found in other studies (Barrett-Connor 1997). All these data suggest that diabetes results in partial or complete loss of the 'female survival advantage' (Pyorala, Laakso and Uusitupa 1987). However, the interplay among sex, diabetes and cardiovascular mortality has not been fully explained, but many factors, such as overweight, hypertension and compliance to treatment, may account for it.

The role of smoking, high blood pressure, low HDL cholesterol, high total cholesterol and obesity have not been studied extensively in elderly subjects. Among the surveys of elderly subjects, only a small number were prospective and the results controversial. Some studies suggest that the risk factors remain the same, while others found a smaller association or even no association at all (Castelli et al 1989; Beaglehole 1991; Krumholz et al 1994; Rossing et al 1996). Surveys showing positive associations consisted largely of subjects belonging to the 'young-old' age group (60-70 years), while negative studies recruited subjects aged 70 and over. In a Finnish study, smoking, high systolic blood pressure and low HDL cholesterol predicted cardiovascular events among elderly non-diabetic subjects, while total cholesterol did not (Kuusisto et al 1994). The risk factors for cardiovascular events remained substantially similar when non-diabetic subjects with previous myocardial infarction were excluded from the analysis. On the contrary, the same study found that in elderly Type 2 diabetic patients none of the classic cardiovascular risk factors, including smoking, hypertension, low HDL cholesterol and high total cholesterol, predicted cardiovascular mortality even when the parameters of glycaemic control were removed from the multivariate model. Although these results were obtained after a relatively short follow-up period, it is interesting to note that the levels of most cardiovascular risk factors were significantly elevated in Type 2 diabetic patients aged 75 years and over. Similar results were reached in the Verona diabetes study: neither smoking nor hypertension predicted cardiovascular mortality in Type 2 diabetic patients aged 75 years and over (Muggeo et al 1997). Similarly, in this study the stronger predictor of cardiovascular events was related to long-term metabolic control. This loss of strength of classic cardiovascular risk factors in predicting mortality in elderly subjects could be due to a selection phenomenon related to the fact that elderly people must be regarded as a cohort of 'survivors', since those patients with both diabetes and high level of cardiovascular risk factors presumably do not reach an old age. Moreover, regarding smoking, it should be acknowledged that in the Verona study the category of ex-smokers was not separately considered, and was included in the non-smoker group. This can lead to an underestimation of the real impact of smoking, because the category of ex-smokers can include patients affected by severe conditions for which smoking was stopped. Paradoxically the actual smokers can be 'healthier' than the other categories.

In the Verona study the effect of body mass index (BMI) on mortality was addressed in a cohort of 5920 Type 2 diabetic patients. BMI was categorized in quintiles and the multivariate analysis was carried out using the first quintile as the reference category. It was found that the relation between all-cause mortality and BMI tended to be U-shaped. This pattern was more pronounced and statistically significant in women. The U-shaped relationship could be observed also for cardiovascular mortality. No relationship between mortality from malignancy and BMI was found in men, while in women a J-shaped pattern was observed.

In most studies of cardiovascular risk factors, a single value of the parameters, measured at the baseline, is usually used. This approach could blunt the effect of the parameter under study, as single determinations present greater random variability than parameters derived from repeated measurements at different periods. Moreover, the analysis could be biased by the 'regression-to-the-mean' effect: owing to the interplay between inter-individual and intra-in-dividual variability, the actual level of a risk factor is usually overestimated in those subjects with the highest values and underestimated in those with the lowest values. In this regard, more information could be obtained from studies in which serial determinations over time of each parameter were used in the analysis, instead of using a single value collected at the baseline of the study. In the Verona diabetes study, data were collected with this second approach for BMI and for glycaemic control. Survival probability was higher in those patients whose weight and metabolic control remained approximately stable.

Diabetes-specific Predictors

The excess mortality in Type 2 diabetes is only partially explained by the classic risk factors. Other fac tors specifically related to the natural history of diabetes strongly affect survival. Age of onset, duration of the disease, long-term glucose control, severity of the disease and presence of chronic complications, mode of therapy and the chosen model of care all contribute to patient survival.

Duration and Severity of the Disease

Duration of diabetes is computed from the time of diagnosis, which does not coincide with the biological onset of the disease. There is a latency period of 4-7 years between the biological onset of Type 2 diabetes and its clinical recognition (Harris et al 1993). During this period the mortality risk is similar or even higher than the mortality risk of known diabetic patients, as hyperglycaemia and other risk factors remain untreated (Wingard and Barrett-Connor 1995). This can explain why in the Verona diabetes study patients with duration of disease 0-4 years already showed a 23% increase of mortality risk (SMR = 1.23; 95%CI 1.191.27) (Brun et al 2000). In univariate analysis, mortality risk increases with the duration of diabetes, but when the changes in therapy over time are taken into account, the effect of diabetes duration loses its predictive value in relation to mortality. This suggests that the progression of diabetes and its severity are better described by the changes of therapy rather than by the time elapsed since diagnosis. Figure 8.4 clearly shows

Figure 8.4 Relation between duration of diabetes and mortality in multivariate analysis with and without adjustment for treatment

Verona Diabetes Study

Tertiles of duration

Figure 8.5 In the Verona Diabetes Study the percentage of patients treated with diet decreased with duration of diabetes; conversely the proportion of insulin-treated patients increased that when the risk of mortality is computed accounting only for age and duration, this latter is a predictor of mortality. However, when other confounders are accounted for, the effect of duration is no longer significant. With increasing duration of diabetes the proportion of patients treated with diet progressively decreases, as observed both in the Verona study (Figure 8.5) and in the UKPDS (1998). Conversely the proportion of patients requiring insulin treatment increases four-fold from the first to the third tertile of duration. Figure 8.6 shows that mortality is significantly associated with therapy: use of oral agents and, even more strikingly, use of insulin is associated with a significant increase in mortality as compared

Figure 8.6 Standardized mortality ratios (SMRs) according to diabetes treatment

with diet, and this further supports the concept that mode of treatment is a potent marker of severity of the disease. Severity of the disease includes not only the degree of hyperglycaemia, but also the clustering of other risk factors (Stamler et al 1993), as well as the presence of chronic complications. Figure 8.7 shows that the prevalence of diabetic complications at baseline was significantly higher in patients who died in the following 5 years as compared with diabetic patients still alive after 5 years.

Duration of diabetes in Finnish subjects aged 65-74 years was found to be a strong predictor of cardiovascular events (Kuusisto et al 1994). This study, owing to the high (40.2%) prevalence of newly detected Type 2 diabetic patients at the baseline, allowed a better evaluation of the importance of the duration of the disease with respect to cardiovascular risk. It is reasonable to think that mechanisms associated with long-lasting hyperglycaemia underlay this relation. These findings on cardiovascular events along with the above reported evidence strongly suggest that diabetic patients should be identified and treated at the earliest stages of the disease in order to reduce the incidence of complications and, hence, mortality.

Metabolic Control

In studies on general populations including also diabetic patients (Haffner et al 1990; Stamler et al 1993;

Figure 8.7 Prevalence of diabetic complications at baseline was significantly higher in diabetic patients who died in the subsequent 5 years; *p < 0.001

Balkau et al 1998), it has been clearly shown that glycaemia and, of course, diabetes are associated with increased risk of mortality, primarily from cardiovascular diseases. There is a strong relationship between the degree of metabolic control, as measured by a single determination of fasting plasma glucose or HbA1c at baseline, and the incidence of microvascular complications. A poor metabolic control also amplifies the effect of other diabetes-specific risk factors, such as duration of diabetes (Kuusisto et al 1994) and microalbuminuria (Gall et al 1995). However, intensive treatment of hyperglycemia, resulting in a reduction of HbA1c from 7.9% to 7%, was associated with an impressive decrease in microvascular complications with a moderate reduction in cardiovascular events, such as fatal and non-fatal myocardial infarction (UKPDS 1998).

The weaker association between HbA1c and cardiovascular mortality could be due to the fact that the assessment of long-term glucose control by FPG or HbA1c does not fully reflect the complex interrelation between everyday glucose control and outcomes. For instance, a frequent recurrence of hyperglycaemic spikes and hypoglycaemic episodes could disproportionately increase the overall glycaemic risk. This additional risk, related to the 'valley and peak phenomenon', is not detected by a single determination of fasting plasma glucose or HbA1c. The latter correlates with the mean glucose level of a given patient and does not reveal the excursions of plasma glucose over time.

To detect 'glycaemic variability' we have suggested computation of the coefficient of variation (CV = (SD/mean) x 100) of a time series of fasting plasma glucose (FPG) determinations (Muggeo et al 1995b, 1997, 2000). This parameter offers additional information on the impact of long-term glucose control on mortality. In fact, in a cohort of 566 elderly Type 2 diabetic patients, grouped in tertiles of coefficient of variation of FPG over a 3 year-period (1984-86), the subsequent 5-year mortality was greater in patients of the third tertile (Muggeo et al 1995b). This association was stronger than that between the mean of fasting plasma glucose (M-FPG) and mortality (see the right panel of Figure 8.8). Interestingly, the higher mortality experienced by the patients of the third tertile of CV-FPG was explained by an excess in cardiovascular mortality. These patients showed a longer duration of diabetes, a more frequent use of insulin and tolbut-amide, a higher M-FPG, and a higher number of hypoglycaemic events than did patients belonging to the first and the second tertiles (Muggeo et al 1997).

The association between CV-FPG and mortality was confirmed in a larger cohort of Type 2 diabetic patients aged 56-74 years (left panel of Figure 8.8). In these patients, CV-FPG during 1984-86 was a stronger

Figure 8.8 Relative risk of all-cause mortality, by tertiles of M-FPG and CV-FPG


Figure 8.8 Relative risk of all-cause mortality, by tertiles of M-FPG and CV-FPG

predictor of 10-year mortality (1987-96) than M-FPG, which appeared as an independent predictor of mortality only when CV-FPG was not included in multivariate survival analysis (Muggeo et al 2000) (Figure 8.9). Of course, these results do not necessarily mean that the severity of hyperglycemia is not important in determining the outcome in Type 2 diabetes; but they do indicate that the prognostic value of M-FPG is lower than that of CV-FPG. Indeed, these data suggest that CV-FPG might be more reliable than M-FPG in assessing the relationship between long-term glucose control and survival. On the other hand, CV-FPG might be a feature of glucose control 'variability' distinct from hyperglycemia. The cardiovascular risk associated with glucose variability includes both the known poor outcome related with recurrent hypoglycaemia, and the deleterious effects associated with recurrent acute hyperglycaemia. The former predisposes to several adverse events such as trauma, myocardial infarction and arrhythmia (Frier 1993); the latter induces several changes in coagulation (Ceriello 1998), en-dothelial function (Giuliano et al 1997), activation of circulating adhesion molecules (Marfella et al 2000), and electrocardiographic QTc abnormalities (Marfella et al 2000), which all contribute to increase the cardiovascular risk of these patients, mainly in the postprandial state (Ceriello 2000).

All these data underline the importance of gly-caemic peaks occurring in diabetic patients especially after meals. In the Diabetes Intervention Study, carried out in 1139 Type 2 diabetic patients by Hanefeld et al (1996); the cumulative incidence of cardiovascular events during 11 years of follow-up was significantly correlated with post-prandial blood glucose rather than with fasting blood glucose at baseline, suggesting that post-prandial glucose levels could better describe the glycaemic risk for cardiovascular diseases. Recent population studies carried out in Europe (DECODE 1999) and in the USA (Sievers, Bennett and Nelson 1999) have demonstrated that both in non-diabetic and in Type 2 diabetic patients 2-hour OGTT plasma glucose, and not fasting plasma glucose, is a strong predictor of mortality. On the contrary, in a 10-year prospective study of patients with newly diagnosed Type 2 diabetes, cardiovascular mortality increased three-fold in patients included in the highest blood glucose tertile at baseline when compared with patients included in the lowest blood glucose tertile (Uusitpa et al 1993). The 8- to 10-year follow-up study of the Wisconsin cohort reported that high glycated haemoglobin (HbA1c) was associated with increased mortality, mainly due to vascular causes, both in younger-onset and in older-onset diabetic people after controlling for other risk factors (Moss et al 1994).

Relative risk

Relative risk

Was this article helpful?

0 0
Spiritual Weight Loss Mentality

Spiritual Weight Loss Mentality

Awesome Ways To Get Over Your Mentality That Keeps you Overweight! This Book Is One Of The Most Valuable Resources In The World When It Comes To Results In Your Slim-down and Health Efforts! Day in day out we keep ourselves absorbed with those matters that matter the most to us. A lot of times, it might be just to survive and bring in some money. In doing so we at times disregard or forget about the extra matters that are essential to balance our lives. They’re even more essential to supply real meaning to our world. You have to pay attention to your wellness.

Get My Free Ebook

Post a comment