Impact of Serum Phosphate in Mechanically Ventilated Patients With Severe Sepsis and Septic Shock

Abstract

Background:

Hypo- and hyperphosphatemia are common in severe sepsis and septic shock. Published outcome data in patients with phosphate derangements primarily focus on hypophosphatemia and the general critically ill population. This study aimed to determine the impact of serum phosphate on clinical outcomes in patients with severe sepsis and septic shock.

Methods:

A retrospective cohort analysis of adult mechanically ventilated patients with severe sepsis or septic shock was performed. Patients were randomly selected from an internal intensive care unit (ICU) database at an academic medical center in the United States and screened for inclusion and exclusion criteria. Time-weighted phosphate was calculated using all phosphate measurements obtained during ICU admission. The associations between time-weighted phosphate and duration of mechanical ventilation, 28-day mortality, and ICU and hospital length of stay were evaluated using linear or logistic regression as appropriate.

Results:

One-hundred ninety-seven patients were evaluated: 33 were categorized as hypophosphatemia, 123 as normophosphatemia, and 41 as hyperphosphatemia. Patients with time-weighted hyperphosphatemia had a higher Simplified Acute Physiology Score III score and incidence of septic shock. Significantly higher rates of 28-day mortality were observed among those with time-weighted phosphate levels above 3.5 mg/dL. However, both time-weighted hypo- and hyperphosphatemia were associated with decreased duration of mechanical ventilation. For every 0.5 mg/dL increase in time-weighted phosphate referent values from 4.0 to 6.0, the duration of mechanical ventilation decreased by 8% to 26%. For every 0.5 mg/dL decrease in time-weighted phosphate referent values from 3.0 to 1.0, significant decreases in duration of mechanical ventilation ranged from 14% to 41%.

Conclusion:

Time-weighted hyperphosphatemia may be associated with increased mortality in mechanically ventilated patients with severe sepsis or septic shock. However, time-weighted hypo- and hyperphosphatemia were associated with decreased duration of mechanical ventilation. Future studies should further describe the impact of hypo- and hyperphosphatemia on clinical outcomes among critically ill patients with severe sepsis or septic shock.

Keywords

hypophosphatemia hyperphosphatemia severe sepsis septic shock mechanical ventilation

Introduction

Severe sepsis and septic shock carry a high burden of morbidity and mortality. Host response to a pathogen invokes numerous immunologic and metabolic changes that can hinder survival. Serum phosphate derangements (hypophosphatemia and hyperphosphatemia) are common metabolic findings associated with severe sepsis and septic shock. Phosphate, in the form of adenosine triphosphate, is required for cellular activity and is vital in improving physiology in critically ill patients. A vast majority of the current literature on the effects of phosphate in critically ill populations highlights the negative impact of hypophosphatemia. In mechanically ventilated patients, phosphate supplementation has been shown to increase diaphragmatic contractility by a mean of 70%.¹ Other studies have shown hypophosphatemia to have a deleterious effect on outcomes in patients on mechanical ventilation^2,3 and continuous renal replacement therapy (CRRT)^3
–5 and intensive care unit (ICU) patients as a whole.⁶ However, these studies had varying definitions of hypophosphatemia and none focused solely on patients with sepsis.

Two studies suggest that both less phosphate supplementation and hypophosphatemia that occur early during sepsis treatment are associated with a higher incidence of cardiac arrhythmias.^7,8 The mechanism for this interaction is not well defined in the existing literature but may be related to decreased myocardial contractility caused by hypophosphatemia.⁸ Another small study found that severe hypophosphatemia (<1.0 mg/dL) worsened mortality in a group of patients with sepsis.⁹ Only 1 study has described the impact of hyperphosphatemia on outcomes in the critically ill patients.⁶ Patients with hyperphosphatemia had significantly worse ICU and hospital mortality, duration of mechanical ventilation, and ICU and hospital length of stay (LOS) in univariate analysis. Hyperphosphatemia has never been studied strictly in a population of patients with sepsis.

Comparing patients based on their phosphate levels over the entire ICU stay rather than the level upon admission or a single deranged level during their stay would likely better assess the impact phosphate status has on outcomes. In a general ICU population, Suzuki et al showed that ICU and hospital nonsurvivors had a lower time-weighted hypophosphatemia compared to survivors.⁶ This strategy of assessing phosphate status takes into account all phosphate levels obtained in the ICU over time. Due to the frequency of hypophosphatemia in sepsis, a more robust outcome analysis of this population is warranted. The purpose of our study was to determine the impact of serum phosphate on clinical outcomes in ICU patients with severe sepsis and septic shock. We hypothesized that lower and higher phosphate levels are independently associated with worse clinical outcomes in patients with severe sepsis and septic shock admitted to the ICU.

Materials and Methods

Study Design and Patient Population

This 2-center, retrospective cohort study was conducted at a tertiary academic medical center and its affiliated community hospital in the United States. Mechanically ventilated patients with severe sepsis or septic shock during ICU admission were evaluated. Septic shock was defined as any vasopressor use during ICU admission. Severe sepsis was defined using criteria derived from the Surviving Sepsis Campaign 2012 guidelines. Patients aged 18 to 89 years admitted to an ICU between August 2013 and July 2015 with severe sepsis or septic shock who required mechanical ventilation for at least 6 hours were eligible for inclusion. Exclusion criteria were admission to the burn service, past medical history of hypoparathyroidism or chronic kidney disease, admission from a different hospital, chronic tracheostomy, admission to the ICU for less than 24 hours, less than 2 serum phosphate levels obtained in the ICU, incarceration, and pregnancy. For patients admitted to the ICU with severe sepsis or septic shock more than once during the same hospital stay, all data obtained in the ICU were combined and they were classified according to the severity of sepsis during their first ICU stay. This study received approval from the institutional review board of The Ohio State University (Protocol #2015E0638). Requirement of informed consent was waived, given the retrospective nature of this study.

Data Collection

Patients were identified from an internal database of ICU patients. Patients were assigned a reference number using a random number generator and were then sorted by their randomly assigned reference number and reviewed them in order to review for inclusion and exclusion criteria. Baseline characteristics, including age, gender, height, weight, Charlson comorbidity index, Simplified Acute Physiology Score III (SAPS III) within 24 hours of ICU admission, location prior to ICU admission (emergency department, operating room, hospital ward, outpatient clinic), type of ICU admitted to, and the most recent lactate level prior to ICU admission (if available), were collected from the electronic medical record. During admission to the ICU, the following variables were collected: maximum lactate and all phosphate levels while in the ICU, total mmol of both intravenous (IV) and enteral phosphate supplementation in the ICU, requirement of intermittent hemodialysis and/or CRRT while in the ICU, and presence of another type of shock. Cardiogenic shock was defined as the use of an inotrope (milrinone or dobutamine). Neurogenic shock was defined as the use of vasopressor support in the setting of recent spinal cord injury, suffered immediately prior to hospital admission or iatrogenically via surgery, with documentation of neurogenic shock in the history and physical or progress notes. Anaphylactic shock was determined based on documentation in the history and physical or progress notes. Hemorrhagic shock was defined as activation of massive transfusion protocol or hemorrhagic shock as documented in the history and physical or progress notes. Clinical outcome variables including ICU and hospital LOS, all-cause 28-day mortality, tracheostomies, and duration of mechanical ventilation were also collected.

Study Outcomes

The primary outcome measure was 28-day all-cause mortality. Patients discharged from the hospital prior to 28 days were presumed to be alive at 28 days. Secondary outcomes included duration of mechanical ventilation, tracheostomies, and ICU and hospital LOS. Mechanical ventilation was counted from the time of intubation until an extubation that was successful for at least 24 hours or passing a breathing trial and ultimately continuing with unassisted breathing (including tracheostomy mask, T-piece or continuous positive airway pressure, and pressure support of ≤5 cm water for at least 48 hours).¹⁰ For patients who were intubated for <24 hours, intubations lasting at least 6 hours were counted toward total intubation time.

Phosphate Status

Using all phosphate values obtained in the first ICU stay, the area under the curve was found using cubic splines. Time-weighted phosphate was then calculated by taking the area under the curve and dividing by the total time between the patient’s first and last phosphate values. This method has been utilized in other studies to measure laboratory values such as blood glucose.¹¹ We felt that this method is the most accurate representation of serum phosphate’s impact on outcomes since it describes a patient’s phosphate exposure during their entire ICU stay. Categorical time-weighted phosphate levels were then defined such that hypophosphatemia is <2.5 mg/dL, normophosphatemia is between 2.5 and 4.2 mg/dL, and hyperphosphatemia is >4.2 mg/dL. This is the normal reference range at our institution, and thus, the values protocolized and nonprotocolized phosphate supplementation and limitation were based off.

Statistical Analysis

A χ² test was used to compare the unadjusted 28-day mortality across time-weighted phosphate categories and logistic regression to find the relationship between 28-day mortality and continuous time-weighted phosphate level controlling for sepsis status and patient characteristics. The Kruskal-Wallis rank test was used to compare the differences in the duration of mechanical ventilation and categorical phosphate. Linear regression was used to find the association between duration of mechanical ventilation and continuous time-weighted phosphate, controlling for patient demographics and clinical characteristics. These controlling variables were entered into the regression model as either a confounder or an effect modifier. Starting with all confounders in the model, we removed them one at a time if they were no longer confounders of time-weighted hypophosphatemia or any other covariate in the model. This procedure was repeated until a confounder could no longer be dropped (if it changed at least one other coefficient in the model by more than 15% in either direction). This method leaves a parsimonious multivariable model. None of the controlling variables were determined to be an effect modifier in our analyses.

The natural log of mechanical ventilation on hypophosphatemia status was regressed in order to meet normality and homoscedasticity assumptions needed in linear regression. When the regression results were transformed back to the original units of days, the coefficient for hypophosphatemia variable was interpreted as a ratio of the geometric mean of the hypophosphatemia value relative to the mean hypophosphatemia referent level. The geometric mean is similar in value to the median value. In addition, the model contained a term for sepsis status (severe sepsis vs septic shock). Method of fractional polynomials was used to determine whether the relationship between duration of mechanical ventilation and continuous phosphate is other than linear.¹²

Linear regression was also used to find the relationship between hospital (and ICU) LOS and continuous time-weighted phosphate level in survivors, again controlling for sepsis status. Hospital LOS and ICU LOS were natural-log-transformed for the same reasons duration of mechanical ventilation was transformed. Our sample size was developed for our original study design, which was to compare time-weight hypophosphatemic and time-weighted nonhypophosphatemic groups. Incorporating existing literature, we estimated that mean duration of mechanical ventilation would be 3 days in the nonhypophosphatemic group and 5 days in the hypophosphatemic group. The standard deviation (SD) was also set at 3 and 5 days, respectively, as the number of days of mechanical ventilation tends to follow an overdispersed Poisson distribution. We also estimated the ratio of patients who are hypophosphatemic to nonhypophosphatemic would be 3:7. A total of 197 patients, 59 patients with hypophosphatemia and 138 patients without hypophosphatemia, would have at least 80% power to detect this 2-day difference based on a likelihood ratio test with an alpha level of .05.¹³ However, upon determining the number of patients with time-weighted hyperphosphatemia, we opted to analyze the data using the aforementioned regression models instead. This adjustment was necessary to account for the unexpectedly high number of patients with hyperphosphatemia. Patient characteristics were presented across the time-weighted phosphate categories using counts and percentages for categorical variables, while continuous variables used either means ± SD or medians with interquartile ranges (IQRs, 25% to 75%) depending on their distribution. All analyses were performed using Stata 14.1 (StataCorp, College Station, Texas).

Results

A total of 1679 patients were screened for inclusion and 197 met the inclusion and exclusion criteria (Figure 1). A majority (n = 1268) of patients were excluded for baseline chronic kidney disease, admission from another hospital, or an intubation duration of less than 6 hours. After obtaining the time-weighted phosphate values, 33 were assigned to the hypophosphatemia group, 123 to the normophosphatemia group, and 41 to the hyperphosphatemia group. The median number of ICU phosphate values available to calculate the time-weighted phosphate was 7, 9, and 9 in the hypophosphatemia, normophosphatemia, and hyperphosphatemia groups, respectively.

When compared to the normophosphatemia group, the hyperphosphatemia group had significantly higher SAPS III, incidence of septic shock, pre-ICU and peak ICU lactate levels, and incidence of genitourinary and intra-abdominal infections. Finally, this study included a predominantly medical ICU population, with at least 73.2% medical ICU patients in each phosphate group (Table 1).

Table 1.

Baseline Patient Characteristics and Laboratory Values.^a

Variable	Hypophosphatemia (n = 33)	Normophosphatemia (n = 123)	Hyperphosphatemia (n = 41)
Age	60.0 (12.7)	59.3 (14.2)	62.5 (13.6)
Male (%)	42.4	48.8	48.8
BMI	28.8 (8.3)	29.7 (12.0)	33.4 (12.5)
Charlson comorbidity index	4.0 [2.5-6.0]	4.0 [2.0-8.0]	5.0 [3.0-9.0]
SAPS-III^b	69 [65-86]	70 [62-81]	78 [68.0-94.5]
Location prior to ICU (%)
ED	48.5	40.7	39
Hospital ward	45.5	52.0	53.7
Operating room	6.0	6.5	7.3
Outpatient clinic	0	0.8	0
Number of phosphate levels drawn during ICU	7.0 [4.0-9.5]	9.0 [5.0-18.0]	9 [5.0-12.0]
Type of ICU
Medical (%)	90.9	79.7	73.2
Surgical (%)	3.0	8.1	7.3
Mixed (%)	6.0	12.2	19.5
Infection source (%)
Lung	60.6	54.4	29.3
Genitourinary^b	21.2	13.8	26.8
Intra-abdominal^b	18.2	10.6	22.0
Bloodstream	24.2	22.0	51.2
Severe sepsis (%)	21.2	20.3	4.9
Septic shock (%)^b	78.8	79.7	95.1
Pre-ICU lactate, n = 176 (mmol/L)^b	3.5 [1.6-5.9]	2.3 [1.4-4.3]	7.6 [2.8-14.9]
Peak ICU lactate, n = 181 (mmol/L)^b	2.3 [1.7-3.9]	2.2 [1.4-5.0]	9.0 [5.0-12.0]

Abbreviations: BMI, body mass index; ED, emergency department; ICU, intensive care unit; SAPS-III, Simplified Acute Physiology Score III.

^aData are reported as median [interquartile range] or mean (standard deviation).

^bP < .05 between normo- and hyperphosphatemia groups.

Twenty-Eight-Day In-Hospital Mortality

A total of 76 (38.6%) patients died in the hospital within 28 days. The rates were 36.3%, 32.5%, and 58.5% in the hypophosphatemia, normophosphatemia, and hyperphosphatemia groups, respectively. These differences were statistically significant (P = .012). The method of fractional polynomials suggested that the relationship between 28-day mortality and continuous time-weighted phosphate was quadratic, with higher odds of mortality at the extremes of time-weighted phosphate values. When adjusted for site of infection, the risk of 28-day in-hospital mortality displays a higher risk of mortality among patients with lower and higher time-weighted phosphate levels (Figure 2). Sepsis severity (severe sepsis vs. shock) did not change the odds ratios displayed in Figure 2 since there was not a significant interaction between time-weighted phosphate an dsepsis severity. There were higher rates of mortality among those with lower and higher time-weighted phosphate levels; however, only comparisons among patients with phosphate levels above 3.5 mg/dL were associated with statistically significant increases in risk of mortality.

Duration of Mechanical Ventilation

The unadjusted median [IQR] duration of mechanical ventilation was 3.0 [1.7-5.9] days, 4.8 [2.3-10.5] days, and 3.0 [1.8-6.8] days for the hypophosphatemia, normophosphatemia, and hyperphosphatemia groups, respectively. This difference was statistically significant (P = .005). A palliative extubation was performed in 13 (39.3%), 41 (33.3%), and 25 (60.9%) patients in each of these groups, respectively. There were more palliative extubations than deaths because 9 patients underwent palliative extubation but did not die during their hospitalization. Twenty-two (11.1%) patients underwent a tracheostomy; 19 (86.4%) of these were in the normophosphatemia group.

When adjusted for location prior to the ICU and site of infection, duration of mechanical ventilation was lowest for patients with lower and higher time-weighted phosphate values (Figure 3). Sepsis severity (severe sepsis vs. shock) did not change the odds ratios displayed in the Figure 3 since there was not a significant interaction between time-weighted phosphate and sepsis severity. In this analysis, 28-day mortality was not significantly associated with the outcome (P-value = .969), thus was dropped from the model. The method of fractional polynomials suggested that the relationship between duration of mechanical ventilation and continuous time-weighted phosphate was quadratic, with shorter durations of mechanical ventilation at the extremes of time-weighted phosphate values. Consequently, the ratio of the geometric mean of the duration of mechanical ventilation is dependent on the phosphate value of interest and the phosphate referent value. Thus, for every 0.5 mg/dL increase in time-weighted phosphate in the referent range 1.0 to 3.0, there is an increase in geometric mean duration of mechanical ventilation that ranged from 14% to 41% (Figure 3). Conversely, for every 0.5 mg/dL increase in time-weighted phosphate in the referent range from 4.0 to 6.0, there was a decrease in mean duration of mechanical ventilation between 8% and 26%. These ratios were not impacted by severity of sepsis, and thus, each ratio of geometric means applies to all patients. After removing patients who died during their hospital stay, there was still a statistically significant increase in duration of mechanical ventilation for each 0.5 mg/dL increase in time-weighted phosphate in the referent range from 1.0 to 3.0, but no statistically significant difference was present for any other 0.5 mg/dL increase (Figure 4).

The ICU and Hospital LOS

Among survivors, the median [IQR] ICU LOS was 6.3 [4.3-8.3] days, 7.8 [4.8-13.5] days, and 5.0 [2.3-9.5] days in the hypophosphatemia, normophosphatemia, and hyperphosphatemia groups, respectively. These differences were statistically significant (P = .004). Median [IQR] hospital LOS was 10.0 [8.3-17.0] days, 15.5 [10.0-22.8] days, and 10.0 [5.0-16.6] days for survivors in these 3 groups. There were no statistically significant differences between any 2 consecutive 0.5 mg/dL time-weighted phosphate levels for either ICU or hospital LOS among survivors (n = 121) regardless of classification as severe sepsis or septic shock (Figure 5A and B).

Other ICU Characteristics

The CRRT was used in 4 (12.1%), 21 (17.1%), and 19 (46.3%) patients in the hypophosphatemia, normophosphatemia, and hyperphosphatemia groups, respectively. Intermittent hemodialysis was less common, as it was utilized in 1 (3.0%), 11 (8.9%), and 5 (12.2%) patients in the 3 respective groups. Phosphate supplementation was most common in the hypophosphatemia group, with 28 (84.8%) patients receiving IV and/or enteral supplementation. Three (7.3%) patients in the hyperphosphatemia group received phosphate supplementation.

Discussion

In this retrospective cohort study of 197 mechanically ventilated patients with severe sepsis or septic shock, we found that patients with time-weighted hyperphosphatemia in our cohort were more likely to die within 28 days of ICU admission. However, 28-day mortality was not statistically significant in those participants with time-weighted hypophosphatemia. Patients had shorter durations of mechanical ventilation on the extremes of time-weighted phosphate values based on measurements obtained during their ICU admission. This is contradictory to our hypothesis that extremes of phosphate, in particular hypophosphatemia, would hinder patients’ ability to wean off ventilatory support.

The impact of hypophosphatemia on outcomes in ICU patients has yielded mixed results that are further confounded by a variety of definitions of hypophosphatemia. These include, but are not limited to, initial value upon ICU admission,^7,8 highest or lowest value during ICU admission or within a certain time from ICU admission,^3,4,6 ratio of days with hypophosphatemia,⁵ and time-weighted value.⁶ To date, there have been 2 large studies that focused on the impact of hypophosphatemia in ICU patients.^5,6 Suzuki et al focused on those who had at least 1 episode of hypophosphatemia, defined as a phosphate of <1.86 mg/dL.⁶ Hypophosphatemia (both cross-sectional and time-weighted) was associated with increased mortality and duration of mechanical ventilation with univariate analysis, but this association did not persist with multivariable analysis of cross-sectional hypophosphatemia. No such analysis was done using time-weighted phosphate. The authors therefore concluded that hypophosphatemia acts as a surrogate for severity of disease but does not predict worse clinical outcomes in a mixed ICU population. In the present study, we chose to use the time-weighted phosphate value because this allows for an assessment of a patient’s total phosphate exposure rather than a cross-sectional value taken during their ICU admission. Time-weighted hypophosphatemia was not associated with a difference in mortality but was associated with a decrease in duration of mechanical ventilation. These conflicting results may be explained by a smaller sample size in our study (n = 197) compared to the study by Suzuki et al (n = 2730). Additionally, the current study focused on patients with severe sepsis and septic shock, so there may be an inherent difference in this cohort compared to a general ICU population. Another large study by Yang et al compared ICU patients receiving CRRT with a high ratio (≥0.58) to those with a low ratio (<0.58) of days with hypophosphatemia (<2.5 mg/dL).⁵ A high ratio of hypophosphatemic days was associated with a 1.45-fold increase in mortality. As previously mentioned, we did not find a significant increase in 28-day mortality with time-weighted hypophosphatemia in our study. Yang et al demonstrated a 50% 28-day mortality rate in their cohort, compared to 38.5% in our study cohort. It should also be noted that their study evaluated a mixed ICU population that necessitated CRRT. In our study of patients with severe sepsis and septic shock, CRRT was utilized in only 12.1%, 17.1%, and 46.3% of patients in the hypophosphatemia, normophosphatemia, and hyperphosphatemia groups, respectively. The large studies by Suzuki et al and Yang et al highlight the frequency of hypophosphatemia in ICU patients and its impact on clinical outcomes but do not specifically isolate its impact on patients with severe sepsis and septic shock. Our study is the largest to focus on clinical outcomes with hypophosphatemia in this patient population.

The impact of hyperphosphatemia in ICU patients is not as well studied as hypophosphatemia. Suzuki et al performed a subgroup analysis that assessed patients who developed hyperphosphatemia (>4.3 mg/dL) at least once.⁶ In this cohort, hyperphosphatemia was associated with increased ICU mortality, duration of mechanical ventilation, and ICU LOS compared to those without hyperphosphatemia. Furthermore, Jung et al showed a significant increase in 28- and 90-day mortality in patients with hyperphosphatemia (>4.5 mg/dL) receiving CRRT for sepsis-induced acute kidney injury (AKI), when compared to patients without hyperphosphatemia.¹⁴ Another study assessed the impact of hyperphosphatemia on all patients presenting to a Swiss emergency department.¹⁵ They found that hyperphosphatemia (>4.5 mg/dL) was associated with an increase in 28-day mortality. However, the authors did not include any assessment of baseline severity of illness and conducted an analysis limited to data solely obtained in the emergency department. Our study is the first to describe the impact of hyperphosphatemia in ICU patients using a definition that describes overall phosphate exposure rather than a cross-sectional representation via a single phosphate value obtained during ICU admission. Patients with hyperphosphatemia in our study had significantly higher SAPS III compared to normophosphatemic patients, which translated to significantly higher 28-day mortality on multivariate analysis. However, these patients also had shorter durations of mechanical ventilation, which is contrary to the results from the study by Suzuki et al. Excessive phosphate may simply be a surrogate for severe cell death in the setting of ongoing septic shock, but direct toxicity from hyperphosphatemia is also possible. Although the data in critically ill humans are limited, possible mechanisms include vascular inflammation,¹⁶ reactive oxygen species caused by disrupted mitochondrial function,¹⁷ anemia,¹⁸ and exacerbation of ischemia and cardiovascular disease caused by a high calcium:phosphate product.^19
–21

The current study does possess some limitations. It was conducted before the new sepsis and septic shock criteria were published in February 2016, and therefore, our inclusion criteria will differ from future studies in this area.²² However, 82.7% of our patients required vasopressor therapy, and many patients had a lactate of >2 mmol/L, which indicates that this population would frequently meet the updated septic shock criteria. Patients with chronic kidney disease were excluded in an attempt to have a balanced distribution of time-weighted phosphate values. We acknowledge that this is a common comorbidity in ICU patients which impacts the ability to excrete phosphate, and thus, these results may not pertain to this patient population. Patients with AKI were included, but AKI was not specifically adjusted for in our models. However, the SAPS-III includes serum creatinine and was a variable in our statistical models. Although our study population was randomly selected, we did not screen all patients in our internal database. Therefore, there is a possibility that the assessed sample does not represent all cases at our sites. However, we feel that the inclusion and exclusion criteria and random selection of patients were able to develop a cohort that is similar to a typical critically ill population at our sites. Since the reference range for phosphate at our institution is 2.5 to 4.2 mg/dL, clinical decisions surrounding supplementation and restriction of phosphate are made using these parameters and thus were practical ranges to use to categorize our patients. In our regression analysis for 28-day in-hospital mortality, the first significant change occurred between 4.0 and 3.5 mg/dL. While describing a time-weighted phosphate value of 4.0 mg/dL, as hyperphosphatemia is discordant with our categorization of patients into 1 of 3 phosphate status groups, this regression analysis used the entire study cohort and therefore does not change the results presented. Additionally, as our objective was to evaluate the impact of overall phosphate exposure, this study does not add to the available data surrounding the use of cross-sectional phosphate levels for prognostication. Although this may be a simpler tool to use in practice, a time-weighted phosphate value likely better represents total phosphate available for cellular metabolism over an entire ICU stay. Finally, the current study was limited to critically ill medical and surgical patients. Cardiac, cardiothoracic surgery, and neurocritical care patients were not included, and therefore, these results may not be applicable to these populations.

Conclusion

In patients with severe sepsis and septic shock, time-weighted hyperphosphatemia was associated with an increase in 28-day in-hospital mortality. Time-weighted hypo- and hyperphosphatemia were associated with a decrease in duration of mechanical ventilation. Future studies are warranted to further describe the impact of hypo- and hyperphosphatemia on clinical outcomes among critically ill patients.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflict of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Christopher J. Miller, PharmD, BCCCP

References

Aubier

Murciano

Lecocguic

, et al. Effect of hypophosphatemia on diaphragmatic contractility in patients with acute respiratory failure. N Engl J Med. 1985;313(7):420–424.

Alsumrain

Jawad

Imran

Riar

DeBari

Adelman

. Association of hypophosphatemia with failure-to-wean from mechanical ventilation. Ann Clin Lab Sci. 2010;40(2):144–148.

Demirjian

Teo

Guzman

, et al. Hypophosphatemia during continuous hemodialysis is associated with prolonged respiratory failure in patients with acute kidney injury. Nephrol Dial Transplant. 2011;26(11):3508–3514.

Schiffl

Lang

. Severe acute hypophosphatemia during renal replacement therapy adversely affects outcome of critically ill patients with acute kidney injury. Int Urol Nephrol. 2013;45(1):191–197.

Yang

Zhang

Cui

, et al. Hypophosphatemia during continuous veno-venous hemofiltration is associated with mortality in critically ill patients with acute kidney injury. Crit Care. 2013;17(5):R205.

Suzuki

Egi

Schneider

Bellomo

Hart

Hegarty

. Hypophosphatemia in critically ill patients. J Crit Care. 2013;28(4):536.e9-e19.

Schwartz

Gurman

Cohen

, et al. Association between hypophosphatemia and cardiac arrhythmias in the early stages of sepsis. Eur J Intern Med. 2002;13(7):434.

Schwartz

Brotfain

Koyfman

, et al.

Association between hypophosphatemia and cardiac arrhythmias in the early stage of sepsis: could phosphorus replacement treatment reduce the incidence of arrhythmias?

Electrolyte Blood Press. 2014;12(1):19–25.

Shor

Halabe

Rishver

, et al. Severe hypophosphatemia in sepsis as a mortality predictor. Ann Clin Lab Sci. 2006;36(1):67–72.

10.

Contentin

Ehrmann

Giraudeau

. Heterogeneity in the definition of mechanical ventilation duration and ventilator-free days. Am J Respir Crit Care Med. 2014;189(8):998–1002.

11.

Finney

Zekveld

Elia

Evans

. Glucose control and mortality in critically ill patients. JAMA. 2003;290(15):2041–2047.

12.

Royston

Altman

. Regression using fractional polynomials of continuous covariates: parsimonious parametric modelling. J R Stat Soc Ser C Appl Stat. 1994;43(3):429–467.

13.

Chow

Shao

Wang

. Sample Size Calculations in Clinical Research. New York, NY: Marcel Dekker; 2003.

14.

Jung

Kim

Park

, et al. Electrolyte and mineral disturbances in septic acute kidney injury patients undergoing continuous renal replacement therapy. Medicine (Baltimore). 2016;95(36):e4542.

15.

Haider

Lindner

Wolzt

, et al. Hyperphosphatemia is an independent risk factor for mortality in critically ill patients: results from a cross-sectional study. PLoS One. 2015;10(8):e0133426.

16.

Lau

Pai

Moe

, et al. Direct effects of phosphate on vascular cell function. Adv Chronic Kidney Dis. 2011;18(2):105–112.

17.

Papa

Skulachev

. Reactive oxygen species, mitochondria, apoptosis and aging. Mol Cell Biochem. 1997:174(1-2):305–319.

18.

Otero

TMN

Canales

Yeh

, et al. Association of serum phosphate levels and anemia in critically ill surgical patients. JPEN J Parenter Enteral Nutr. 2017;41(6):986–992.

19.

Dhingra

Sullivan

Fox

, et al. Relations of serum phosphorus and calcium levels to the incidence of cardiovascular disease in the community. Arch Intern Med. 2007;167(9):879–885.

20.

Hioki

Miyashita

Shiraki

, et al. Impact of deteriorated calcium-phosphate homeostasis on amputation-free survival after endovascular revascularization in patients with critical limb ischemia on hemodialysis. Vasc Med. 2016;21(2):137–143.

21.

Zettervall

Soden

Ultee

, et al. Elevated serum phosphate levels are associated with decreased amputation-free survival after interventions for critical limb ischemia. J Vasc Surg. 2017;65(2):431–437.

22.

Singer

Deutschman

Seymour

, et al. The third international consensus definitions for sepsis and septic shock (sepsis-3). JAMA. 2016;315(8):801–810.