Malaria And Insulin Dependent Diabetes Mellitus An Ecological Study

There is scientific merit in studying the association between insulin dependent diabetes mellitus and malaria, since they are both associated with the human leukocyte antigen system. The human leukocyte antigen system is involved in controlling immunological responses, and the association between this system and insulin dependent diabetes mellitus has long been established.12

Malaria is the most important natural selective factor on human populations that has been discovered to date.13 In areas of high endemicity, malaria operates the genetic selection responsible for the influence on the susceptibility to autoimmune diseases.14 In Sardinia, malaria is known to have selected for some serious hereditary diseases such as /3-thalassaemia, Cool-ey's disease and favism; the latter is caused by a deficiency of glucose-6-phospate dehydrogenase enzyme.10 Sardinia is therefore particularly suitable for investigating the association between insulin dependent diabetes mellitus and malaria. The incidence of insulin dependent diabetes mellitus in Sardinia is quite atypical of other Mediterranean countries. Sardinia has the second highest incidence in Europe at 33.2 per 100 000 person years; Finland has the highest incidence at 40 per 100 000 person years. A study of 18-year-old military conscripts born between 1936 and 1971 showed that the risk for insulin dependent diabetes mellitus began increasing with the male birth cohort of 1950 and that the increasing trend was much higher than observed in the remainder of Europe.15 Population genetic studies suggest that, in the plains of Sardinia where malaria had been endemic, some genetic traits were selected to provide greater resistance to the haemolysing action of the Plasmodium vector. In the hilly and mountainous areas, where malaria is almost absent, this adaptation did not occur.16

In another study17 the incidence of insulin dependent diabetes mellitus was obtained from a case registry which had operated in Sardinia since 1989. The incidence data referred to the period 1989-92 and covered the population aged between birth and 29 years. The number of insulin dependent diabetes mellitus cases was available for the 366 communes of Sardinia. Also considered was the number of malaria cases in the communes for the period 1938-40, and the 1936 census populations were available for each commune. The prevalence of malaria between 1938 and 1940 was considered as a covariate, in the model for the calculation of the incidence of insulin dependent diabetes mellitus.

In their modelling approach, the researchers17 assumed a Poisson likelihood regression model for the counts of insulin dependent diabetes mellitus. But they also found extra-Poisson variation and included a random effect term to allow for this variation. This leads to wider standard errors in the parameter estimates of the regression fit. In addition they found that the malaria prevalence may also include extra noise or error and they modelled also for that effect. They note that: 'in practice, ecological covariates can rarely be observed directly'. Available data may be either imperfect measurements of, or proxies for the true covariate. Sometimes epidemiological data concerning another disease may be used as a proxy variable. For example, to study the geographical variation of heart disease mortality, an important covariate would be the proportion of smokers living in each area. Such specific data on smoking would generally not be available, so the prevalence of lung cancer recorded by the cancer registry for each area might be a useful proxy. The simplest approach to this problem would be to estimate the true covariate from the proxy for each area independently, using the proxy estimate in the ecological regression. When the proxy variable is an accurate measure of the true covariate, this approach would be reasonable. However, when the correspondence between the two is not close, this approach has several disadvantages as not accounting for measurement error causes the point estimate of the regression coefficient to be underestimated and its precision overestimated.

The results of the geographical study of the lagged effect of malaria prevalence and insulin dependent diabetes mellitus suggested a significant negative association between long-term malaria endemicity and diabetes. This suggests that people who live in areas where malaria has been particularly frequent, have a lower risk of insulin dependent diabetes mellitus than those who lived in a low prevalence area in 1938. For instance, the risk of diabetes is considerably lower in the low-lying regions than in the hills and mountains of Sardinia. Malaria endemicity in the low-lying areas could have prevented the onset of insulin dependent diabetes mellitus via stronger selection processes. The 95% credible interval (confidence interval) for the correlation between malaria and insulin dependent diabetes mellitus is —0.812 to —0.182 with a point estimate of nearly —0.6. This interval is wide, but there is some support for a negative relationship.

How To Quit Smoking

How To Quit Smoking

Did You Ever Thought You Could Quit Smoking And Live A Healthy Life? Here Are Some Life Saving Tips On How To Do It. Have you ever thought about quitting smoking, but either thought it was impossible or just simply wasn’t that important? Research shows that most smokers do want to quit smoking and they are waiting for that auspicious day eagerly.

Get My Free Ebook


Post a comment