The major findings of this study are a hypernatreamia prevalence rate of 11.7% based on uncorrected serum sodium >145 mmol/L and 30.1% based on a serum sodium of >145 mmol/L corrected for the ambient blood glucose concentration and the independent association of hypernatreamia with advanced age, altered level of consciousness and a new diagnosis of diabetes. The prevalence rate of 11.7% that we observed for uncorrected hypernatreamia is much higher than the 1.2% reported in another study [2
] with similar definition of hypernatreamia (uncorrected serum sodium >145 mmol/L). Our admissions consisted of ketoacidosis, hyperosmolar non-ketotic state and hyperglycaemia with hypernatreamia rates of 17.4%, 37.9% and 0% respectively. Further analysis of our ketoacidosis patients revealed a hypernatreamia rate of 83.3% for hyperosmolar ketoacidosis and 1.2% for non-hyperosmolar ketoacidosis. Perhaps, the study [2
] with a hypernatreamia rate of 1.2% was predominantly patients with non-hyperosmolar ketoacidosis. Although hyperosmolar DKA is increasingly being reported in children [11
] and adults [13
], their hypernatreamia rates were not documented. The proportions of hyperglycaemic crisis admissions presenting with hyperosmolar DKA was 15.1% (n = 8/53) in one study [13
], and 45% (n = 288/613) in another study [14
As uncorrected serum sodium >145 mmol/L was found in none of 119 hyperglycaemic and only 1 of 97 non-hyperosmolar ketoacidosis admissions, the specificity of uncorrected serum sodium >145 mmol/L for the diagnosis of hyperosmolality was 99.5% based on calculated effective osmolality >320 mosmols/kg. The sensitivity of uncorrected serum sodium >145 mmol/L for the diagnosis of hyperosmolality was 62.6% as 20 of 24 admissions for hyperosmolar ketoacidosis and 11 of 29 admissions for hyperosmolar non-ketotic state had uncorrected serum sodium >145 mmol/L. Thus while almost all admissions with presenting uncorrected serum sodium >145 mmol/L were hyperosmolar, not all hyperosmolar admissions had initial uncorrected serum sodium level >145 mmol/L. This suggests that in our setting, hyperglycaemic patients with uncorrected serum sodium levels >145 mmol/L should be managed as hyperosmolar states.
A study [15
] that was conducted on patients admitted to an intensive care unit reported the mechanisms for hypernatreamia to include salt overload and fluid depletion with the use of sodium bicarbonate, mannitol, impaired urinary concentration and sepsis as the independent determinants of hypernatreamia. Unlike our patients who were already hypernatreamic at presentation, the patients in this study [15
] developed hypernatreamia in the course of hospitalization. Sodium bicarbonate, mannitol or hypertonic saline could not have been contributory to the hypernatreamia in our patients as there was no prior administration of these agents at their referring hospitals before presentation to us. In another study [16
], 50% of patients who developed hypernatreamia during hospitalization and 89% of patients presenting with hypernatreamia had urinary concentration defects primarily associated with diuretic therapy or solute diuresis. Although, our patients with hyperglycaemic crisis will expectedly have glycosuria induced diuresis, hypernatremia was mainly a problem in those who were elderly, had altered sensorium or were newly diagnosed with diabetes.
The independent association of age ≥ 60 years with hypernatreamia may be partly explained by an increased threshold for thirst and vasopressin deficiency that is associated with ageing [17
]. Therefore, patients with advanced age may become hypernatreamic due to inadequate compensatory increased oral fluid intake and renal fluid retention in the face of glycosuria induced osmotic diuresis. Altered level of consciousness, regardless of aetiology may result in hypernatreamic dehydration because the patient is unable to replenish renal fluids loss orally due to impairment of the mental state. As our study was not only retrospective but cross-sectional in design, we can only state that impaired mental state was associated with hypernatreamia. The study design does not permit for the exploration of a causal relationship between hypernatreamia and altered level of consciousness. A report [19
] which found all patients with hyperosmolar non-ketoacidotic state to have altered level of consciousness did not indicate the serum sodium levels or any association of hypernatreamia to coma. It is interesting that a new diagnosis of diabetes was significantly independently associated with hypernatreamia. Although admission blood glucose levels was non-statistically higher in the newly diagnosed than known diabetic patients (35.35 ± 19.0 mmol/L versus 31.5 ± 14.6 mmol/L, P = 0.079), the proportion of admissions with HbA1c level above 10% was more in the newly diagnosed than known diabetic patients (87.7% versus 74.8%, P = 0.049). This suggests that the majority of patients with hyperglycaemic crisis as the first manifestation of diabetes had more severe chronic hyperglycaemia than previously diagnosed diabetic patients. Perhaps, these newly diagnosed diabetic patients consequently had more prolonged osmotic diuresis with hypernatreamic dehydration.
The findings from this study suggest that particular attention should be given to serum sodium levels in admissions for hyperglycaemic crisis associated with advanced age, unconsciousness at presentation and a new diagnosis of diabetes. These patients will require more attention to fluids therapy and may require prophylactic anticoagulation as all hypernatreamic admissions met the criteria for hyperosmolality.
Limitations of study
The limitations of this study include its retrospective design and determination of serum osmolality by calculation rather than laboratory measurement. We may have underestimated serum osmolality where osmotically active substances other than glucose and sodium are present in the serum as we calculated rather than measured the serum osmolality. Another limitation is that results of urine osmolality, urine specific gravity and urine electrolytes concentrations were not provided as these are not routinely done in our practice.