Association of admission serum levels of vitamin D, calcium, Phosphate, magnesium and parathormone with clinical outcomes in neurosurgical ICU patients

To evaluate the association of admission serum levels of 25(OH)D, parathormone and the related electrolytes with severity of illness and clinical outcomes in neurosurgical critically ill patients, serum levels of 25(OH)D, parathormone, calcium, magnesium, and phosphate, along with APACHE II score were measured for 210 patients upon admission. Mean serum 25(OH)D was 21.1 ± 7.4 ng/mL. 25(OH)D deficiency (less than 20 ng/dL) and elevated serum parathormone level were found in 47.6% and 38% of patients respectively. Hypocalcaemia, hypophosphatemia, hypomagnesaemia and hypermagnesaemia were found in 29.5%, %63.8, 41.9% and 27.6% of patients respectively. The APACHE II score was significantly correlated with serum levels of 25(OH)D, parathormone, calcium, and phosphate. Multivariate regression analysis adjusted by other risk factors showed that among all clinical outcomes, admission hypovitaminosis D was associated with longer duration of ICU stay and a high admission of parathormone was associated with in ICU mortality. We concluded that disorders of admission serum levels of 25(OH)D, parathormone, calcium, magnesium, and phosphate are related to the presence of multiple causal factors such as severity of disease and are not independently associated with clinical outcomes. Most often they are normalize spontaneously with resolution of the disease process.


Data Collection
Serum 25(OH)D, PTH, total calcium, phosphate, magnesium, albumin and total protein, were measured within 24 hours of ICU admission. Necessary variables were recorded to calculate the acute physiology and chronic health evaluation II (APACHE II) score on the first day of admission. Occurrence of organ failure was monitored during hospitalization. Each patient was evaluated daily for cardiovascular failure, central nervous system failure, coagulation failure, hepatic failure and renal failure. Mean sequential organ failure assessment (SOFA) score was used to determine the extent of a person's organ function. Other clinical outcomes were rate of mortality, use of mechanical ventilation, and hospital length of stay.
Definition of Vitamin D deficiency: 25 (OH) D is considered the best available parameter for the definition of vitamin D status, since it reflects vitamin D body stores better than other vitamin D metabolites. To date, there is no consensus regarding normal vitamin D levels and the definition of vitamin D deficiency in critically ill patients. We categorized 25 (OH) D levels as follows: severe deficiency with serum levels less than 10 ng/dL, mild deficiency with serum levels as 10-20 ng/dL, insufficiency with serum levels as 20-30 ng/dL and normality with serum levels more than 30 ng/dL 34,35 .
Serum calcium, magnesium, phosphate and PTH were measured on the day of admission. Hypocalcaemia and hyperkalemia were defined as serum calcium levels less than 8.5 mg/dL and more than 10.6 mg/dL respectively. Corrected calcium was calculated by measured total calcium + (0.8 × (4.0 − (albumin))). Hypophosphatemia and hypophosphatemia were defined as serum phosphate levels less than 3.5 mg/dL and more than 4.5 mg/dL respectively. Hypomagnesaemia and hypermagnesaemia were defined as serum magnesium levels less than 1.7 mg/dL and more than 2.1 mg/dL respectively. Elevated PTH level was defined as a serum PTH level more than 65 pg/mL 36 .
Statistical analysis: Data were analyzed using the SPSS Version 21 software. Subjects' characteristics are reported as mean ± SD and median, interquartile range for continuous variables and in frequencies and percentages for categorical ones.
Differences between variables in categorical groups of vitamin D were evaluated with the one-way analysis of variance (ANOVA), Kruskal-Wallis, the chi-square or Fisher Exact test. Regression analysis was used to estimate the relationship between serum vitamin D, PTH, calcium, magnesium and phosphate levels with clinical outcomes (mortality, length of stay, duration of mechanical ventilation and organ failure). In this model length of stay (LOS) in the ICU, SOFA score for organ failure and duration of mechanical ventilation were categorized by their median to two groups. Any covariate associated with the response variables (P < 0.05) in univariate analysis, retained in the final model or multivariate logistic regression. Furthermore variables that could have a clinical significance, retained in the final model. P-values less than 0.05 were considered significant.
Median LOS in the ICU was 6 days in total population with minimum and maximum of 3 and 56 days respectively. Organ failure was assessed by the SOFA score. Median SOFA score was 5 in total population with minimum and maximum of 2 and 12 score respectively. One hundred eight (51.4%) patients were under mechanical ventilation. Median duration of mechanical ventilation was 6 days with minimum and maximum of 2 and 51 days respectively. Rate of mortality was 27.6% in total population (n = 58). Median length of stay in the ICU was significantly higher in the severe deficiency and mild deficiency groups compared to the insufficiency and normal groups ( Table 1). Result of univariate logistic regression analysis with the LOS as dependent variable and serum 25(OH)D as independent variable showed that with each point increase in serum 25(OH)D, risk of hospitalization more than 6 days is decreased by 20% ( Table 2).

Serum level of 25 (OH) vitamin D and outcomes.
The mean serum level of 25 (OH) D was 23.08 ± 6.9 ng/dL in dead patients and 16.15 ± 6.1 ng/dL in alive patients (P = 0.001). Rate of mortality was significantly higher in the severe deficiency group compared to the insufficiency and normal groups, as well as in the mild deficiency group compared to the normal group ( Table 1). Result of univariate logistic regression analysis with mortality as dependant variable and serum 25(OH)D as independent variable showed that with each point increase in serum 25(OH)D, risk of mortality decreased by 15% ( Table 2).
The mean serum level of 25 (OH) D was 17.98 ± 7.1 ng/dL in ventilated patients and 24.53 ± 6.2 ng/dL in non-ventilated patients (P = 0.001). All patients in the severe 25(OH) D group were under mechanical ventilation and the mean duration of mechanical ventilation was significantly higher in the severe deficiency and mild deficiency groups compared to the insufficiency and normal groups ( Table 1). Result of univariate logistic regression analysis with duration of mechanical ventilation as the dependant variable and serum 25(OH)D as the independent variable showed that with each point increase in serum 25(OH)D, risk of ventilation more than 6 days decreased by 12% (Table 2).
Median SOFA score was significantly higher in the severe deficiency group compared to the mild deficiency, insufficiency and normal groups ( Table 1). Result of univariate logistic regression analysis with the SOFA score as the dependant variable and serum 25(OH)D as the independent variable showed that with each point increase in serum 25(OH)D, risk of SOFA score more than 5 decreased by 8% (Table 2).
A multivariate logistic regression analysis with serum 25(OH)D as the regressor and each clinical outcome as the dependant variable was separately built (adjusted by APACHE II score, serum albumin, PTH, calcium, phosphate, magnesium, energy intake, age and sex). In this regard, results showed that with each point increase in serum 25(OH) level, risk of hospitalization more than 6 days decreased by 11% whereas serum level of 25(OH)D was not an independent variable for mortality, mechanical ventilation and organ failure ( Table 2).
Results of Pearson correlation showed that serum vitamin D was significantly correlated with PTH, albumin, total protein and APACHE II (r = −0.37, P < 0.001; r = 0.23, P = 0.01; r = 0.2, P = 0.03; r = −0.37, P < 0.001 respectively). Serum level of calcium and outcomes. The mean total serum calcium corrected for albumin was 8.85 ± 0.73 mg/dL with minimum value of 7.20 and maximum value of 10.50. Sixty two patients (29.5%) had low serum calcium (lower than 8.5 mg/dL). Median length of stay in the ICU and median SOFA score were not different between the calcium deficient group and the normal group (P = 0.2, P = 0.8 respectively).
The mean serum level of calcium was 8.67 ± 0.67 mg/dL in ventilated patients and 9.03 ± 0.75 mg/dL in non-ventilated patients (P = 0.5). Results of univariate logistic regression analysis with each clinical outcome separately (LOS, SOFA score and duration of mechanical ventilation) as the dependant variable and serum calcium level as the independent variable showed that there was no relation between serum calcium level and clinical outcomes ( Table 2). The mean serum level of calcium was 8.95 ± 0.74 mg/dL in dead patients and 8.59 ± 0.63 mg/dL in alive patients (P = 0.4).
Result of univariate logistic regression analysis with mortality as the dependant variable and serum calcium level as the independent variable showed that with each point increase in serum calcium level, risk of mortality decreased by 13% (Table 2), but in multivariate logistic regression model adjusted by APACHE II score, serum albumin, PTH, 25(OH) D, phosphate, magnesium, energy intake, age and sex, this relation was not observed ( Table 2). Serum level of phosphate and outcomes. The mean serum phosphate level was 3.26 ± 0.81 mg/dL with minimum value of 1.40 and maximum value of 5.5. Hypophosphatemia with serum phosphate level lower than <3.5 mg/dL was seen in 63.8% (n = 134) of patients. Hyperphosphatemia was seen only in one patient with serum level of 5.5 mg/dL. We did not consider this patient in the statistical analysis. Median length of stay in the ICU and median SOFA score was not different between the phosphate deficient group and the normal group (P = 0.8, P = 0.8 respectively). The mean serum level of phosphate was 3.19 ± 0.77 mg/dL in dead patients and 3.28 ± 0.83 mg/dL in alive patients (P = 0.8). The mean serum level of phosphate was 3.2 ± 0.87 mg/dL in ventilated patients and 3.32 ± 0.75 mg/dL in nonventilated patients (P = 0.4).
Result of univariate logistic regression analysis with each clinical outcome (LOS, SOFA score, duration of mechanical ventilation and mortality) as the dependant variable and serum phosphate level as the independent variable showed that there was no relation between serum phosphate level and clinical outcomes ( (serum magnesium level <1.7 mg/dL). Median length of stay in the ICU and median SOFA score was not different between the hypomagnesemic, hypermagnesemic and normal groups (P = 0.69, P = 0.08 respectively). The mean serum level of magnesium was 1.86 ± 0.38 mg/dL in dead patients and 1.89 ± 0.38 mg/dL in alive patients (P = 0.46). The mean serum level of magnesium was 1.85 ± 0.37 mg/dL in ventilated patients and 1.92 ± 0.33 mg/ dL in nonventilated patients (P = 0.71).
Result of univariate logistic regression analysis with categorized serum magnesium level as the independent variable and each clinical outcome (LOS, SOFA score, duration of mechanical ventilation and mortality) as the dependent variable showed that there was no relation between serum magnesium level and clinical outcomes ( Table 2). Serum level of parathormone and outcomes. The mean serum PTH level was 68.5 ± 43.9 pg/mL with minimum value of 17.9 pg/mL and maximum value of 178.7 pg/mL. A total of 84 (38%) patients had elevated PTH level (serum PTH level >65 pg/mL). Result of univariate logistic regression analysis with serum PTH level as the independent variable and each clinical outcome (LOS, SOFA score, duration of mechanical ventilation and mortality) as the dependant variable showed that with increasing serum PTH level risk of mortality, LOS in the ICU and SOFA score increased significantly. In multivariate logistic regression model (adjusted by APACHE II score, serum albumin, calcium, 25(OH)D, phosphate, magnesium, age, sex, and calorie intake), only serum PTH was an independent variable for mortality ( Table 2).
Secondary hyperparathyroidism was defined as elevated serum PTH level associated with serum 25(OH) level lower than 30 ng/mL and/or serum calcium level lower than 8.5 mg/dL. Secondary hyperparathyroidism was seen in 51.6% of hypocalcemic and in 45.4% of vitamin D insufficient/deficient patients. Table 3 shows that PTH responders have higher APACHE II score, SOFA score, LOS in ICU, ventilator dependency, and mortality compared to non-responders ( Table 3).

Discussion
Our study showed that among all clinical outcomes, admission hypovitaminosis D was associated with longer duration of stay in the ICU and a high admission level of parathormone was associated with mortality in the ICU. Furthermore, PTH responders had higher APACHE II score, SOFA score, LOS in the ICU, ventilator dependency, and mortality compared to non-responder patients. We did not find any correlation between admission serum calcium, magnesium and phosphate level with clinical outcomes. There was a negative correlation between APACHE II score, serum calcium, phosphate, and vitamin D and a positive correlation with serum PTH.

Serum level of vitamin D and outcomes.
We found that 47.7% of patients had vitamin D deficiency and hypovitaminosis D was an independent risk factor for longer duration of stay in the ICU. Many retrospective and prospective studies have evaluated the effect of admission serum vitamin D level on clinical outcomes of medical, surgical, cardiac, mixed and traumatic critically ill patients. Some reported that hypovitaminosis D was associated with increased mortality 2,4,5,8 , longer ICU stay 3-5 and greater need for mechanical ventilation 6 , whereas some results did not show this association concerning mortality in the ICU 9 , duration of stay in ICU 2,8,9 , organ failure 7 and duration of mechanical ventilation 2,8,9 . In one study conducted in the neurocritical care patients, vitamin D deficiency was associated with higher in-hospital mortality in the subset of patients admitted on an emergency basis and this association was not seen in the total population (9). Several factors may contribute to vitamin D deficiency in critically ill patients with prolonged hospitalization including lack of exposure to sunlight, malnutrition, decreased renal hydroxylation and increased tissue conversion of 25(OH)D 3 to 1,25(OH) 2 D 3 however the reasons for the reduced serum level of vitamin D in the first day of hospitalization may be the reduced serum level of albumin or vitamin D binding protein, or receiving intravenous volumes in order to correct hypovolemia or hypotension 37 . The poor clinical outcomes seen in critically ill patients with admission hypovitaminosis D may be related to severity of the underlying disease. The extent of hypoalbinemia is related to the degree of inflammation and severity of acute illness and also the extent of volume replacement is related to the severity of disease 37 as we found an inverse correlation between admission serum vitamin D and APACHE II score as well as a direct correlation between admission serum vitamin D and serum albumin. Also it has been reported that novel vitamin D3 hydroxyderivatives resulting from the action of cytochrome P450 side chain cleavage, have anti-proliferative, prodifferentiation and anti-inflammatory activities and measuring their serum levels may be necessary to fully assess vitamin defciency or suffciency, as opposed to a single measurement of 25(OH)D3 38-40 . Serum level of calcium and outcomes. The prevalence of hypocalcaemia was 29.5%. We did not find any correlation between admission serum calcium level and clinical outcomes, however, we found an inverse correlation between admission hypocalcaemia and APACHE II score. Various studies have evaluated the effect of admission serum calcium level on clinical outcomes of medical, surgical, mixed and traumatic critically ill patients with diverse results. Results of some studies reported by comparing means, have shown that severe and moderate hypocalcaemia (either total or ionized) is associated with increased mortality and worse clinical outcomes in the critically ill patients [12][13][14] . In our study most of the patients had mild hypocalcaemia. It has been reported that mild hypocalcaemia has protective effects in critically ill patients and its correction is not recommended 41 . Results of some other studies using multivariate methods are consistent with our findings as they did not find any association between hypocalcaemia and clinical outcomes 15,16 . Abnormal values of admission serum calcium level in the ICU are common which are not the result of an underlying disease in most cases. Hypocalcaemia is a marker of disease severity and inflammation in the ICU as we found an inverse correlation between admission serum calcium and APACHE II score with resolution of the primary disease, hypocalcaemia will normalize 11 .
Serum level of phosphate and outcomes. The prevalence of hypophosphatemia was 63.8%. We did not find any correlation between admission serum phosphate level and clinical outcomes. Some studies have shown that hypophosphatemia in critically ill patients is associated with increased mortality [18][19][20] , while others did not find this association 21,22 . There are many factors responsible for the occurrence of hypophosphatemia in the ICU including sepsis, trauma, volume expansion, refeeding syndrome, acid-base disorders, medications, major surgery, head trauma and renal replacement therapyand it is unclear whether correcting admission hypophosphatemia improves outcomes in critically ill patients.
Serum level of magnesium and outcomes. The prevalence of hypomagnesaemia and hypermagnesaemia were 27.6% and 41.9% respectively. We did not find any association between admission serum magnesium and clinical outcomes. Several studies have evaluated the effect of admission serum magnesium level on clinical outcomes in medical, surgical, traumatic and mixed critically ill patients. Some have shown that hypomagnesaemia (either total or ionized) is associated with increased mortality [23][24][25][26][27][28][29][30] , longer ICU stay 25,26 , greater need for mechanical ventilation 24,26,30 , higher SOFA scores 28 while others did not find this association [31][32][33] . Results of some were reported by comparing means and without performing regression analysis [23][24][25][26]30 . Several factors are responsible for changes of admission serum magnesium. Some alter magnesium transport in the loop of Henle including PTH, calcitonin, glucagon, arginine vasopressin, and the beta-adrenergic agonists. Hypoalbinemia, nutrition, use of diuretics, metabolic acidosis, hypokalemia, phosphate depletion, volume expansion and aminoglycosides are other etiologies of hypomagnesaemia in critically ill patients 42 .
Moreover there is no clear evidence that magnesium supplementation improves outcomes in critically ill patients 43 . Serum level of parathormone and outcomes. Our study indicated that with increasing PTH level, risk of stay in the ICU increased by 2%. PTH is a marker of inflammation and disease severity. It has been reported that high PTH levels are associated with higher mortality rate 44 . With increasing disease severity and inflammation, serum PTH levels will increase more 45 . It is the reason for the existing of a significant correlation between admission serum PTH and APACHE II score in our study.
On the other hand, we found that PTH responders had higher LOS in the ICU, ventilator dependency, and mortality compared to non-responder patients. This might be due to higher APACHE II score and therefore, severity of disease in responders compared to non-responders. However, previous studies had conflicting results. Some showed that there was no difference between the PTH responders and non-responders in regard to clinical outcomes 46 , while others showed higher mortality rate in the PTH responders 10 .

Conclusion
It seems that changes in serum vitamin D, calcium and PTH on admission are the result of inflammation related to severity of diseases. Values of serum PTH, calcium and vitamin D will return to actual values with remission of inflammation. On the other hand, admission electrolyte disorders due to the presence of multiple causal factors are among the most common clinical problems in the critically ill patients. The vast majority of these patients do not have an underlying related electrolyte disease and most often the electrolyte disorders normalize spontaneously with resolution of the primary disease process. Conducting studies with measuring these parameters at other times of hospitalization is recommended. Moreover further studies to assay the effect of correcting these disorders on clinical outcomes are required.