They were the main mode of hyperglycemic decompensation. Infections are one of the most frequent and serious of recurrent complications of diabetes. [12] . On the one hand diabetes favours immunodepression for infections and on the other hand, infections favour the decompensation of diabetes. It results in a vicious circle hard to separate [13] . The infectious sites remain the same with a distribution, reported in various ways in the study [10] . Infections of the skin and soft tissues (leg erysipelas, boil, diabetes at the foot, over-infected bedsore) mostly drew our attention as well as in [14] series opposed to some studies which consider them in the second place [4] [8] [15] . This difference of opinion could be explained by the particularity of the bacterial area, climate or lifestyle (poor corporal hygiene) with regard to the regions where the works are carried out. Regarding the 5 cases of pulmonary infections, lung tuberculosis has been noted with three patients. Indeed, the incidence of tuberculosis remains two to three times greater in diabetes patients than in non-diabetics. Radi reported two cases in his sets of study (6.6%) of pleuropneumonia infections including one Fendi tuberculosis [16] and Mbadinga-Mupangu [17] got respectively in their series 15% and 5.2% of respiratory infections. Urogenital infections caused the ketoacidosis decompensation in 7 patients. Indeed, most of the time, urinary infection in the diabetes patient is asymptomatic and therefore develop silently and can be complicated by pyelonephritis in patients already at a risk, some of nephropathy diabetes.

It is preceded by infection in second position. The reasons for this treatment termination were: weariness of treatment, financial issues and religious beliefs. Indeed, the quality of treatment observance remains a major problem in the caring of chronic diseases, and the financial problem as well as religious beliefs constitute the major obstacles to access to treatment in developing countries [17] [18] [19] . A national policy of free antidiabetic drugs must be considered in our countries. Moreover, ethno-cultural approach of the disease has to be taken into account in therapy education so as to optimize the caring of patients.

We dealt with four of our patients. In his set Sarr [10] also reported myocardial ischemia as a factor of ketoacidosis decompensation. Painless ischemia occurrence and an acute myocardial infarct in a diabetes patient is much higher compared to non-diabetics [20] . These patients with myocardial infarction have a poor prognosis and hyperglycemia on admission and is as a result a predictive hospital mortality factor.

Two of our patients have decompensated because of corticosteroid treatment. Corticosteroids induce insulin resistance. They reduce the effect of insulin by increasing gluconeogenesis and lowering the collecting and use of possibilities of glucose by peripheral tissues [21] . However, diabetes imbalance is higher if the level of glyceric hemoglobin exceeds 8% and when age and body mass index are high [21] .

Pregnancy originated the inaugural ketoacidosis decompensation in one of our patients. It was a question of an extra-uterine pregnancy. One must note that the retention of dead egg has been reported as a cause of ketoacidosis decompensation by Sarr [10] . So one must insist more on diabetes screening of our people and on treatment awareness because a pregnancy does with metabolic preparation and the conception must be conceived by controlling glycaemia and consider a perfect normalized glycemic hemoglobin.

In our sets, it was not possible to find the active factor in 16.87%. In the Fendi sets [16] , ketosis remains unclear in two cases. In Sarr’s [10] , no etiological factor was found in 6.84% cases. As for Mbadinga-Mupangu [17] , he reported 12.3% of undefined factors.