Intensive Care Unit Acquired Hypernatremia after Major Surgery is Associated with Urine Concentrating Defect: An Observational Study

Background: Although intensive care acquired hypernatremia is a common event, limited knowledge exists about the pathogenesis of this disorder. The present study attempts to show that patients undergoing major surgery develop hypernatremia in the presence of both high salt and volume load and concentration disorder of the kidney with insucient sodium excretion. Methods: In a retrospective study, all patients who were admitted to a 40-bed tertiary surgical intensive care unit of a university hospital from July 2019 to December 2019 with major surgery were examined. Hypernatremia was dened as a sodium value exceeding 145 mmol/l. In addition to the analysis of all patients, complete water and salt balances were performed in a smaller subgroup with 142 patients. Results: 23.9% of patients undergoing major surgery developed hypernatremia, whereby hypernatremia was associated with increased mortality. Patients with hypernatremia showed a renal concentration defect with decreased urine sodium concentration (65 (IQR: 44.8-90) mmol/l vs 78 (IQR: 46-107) mmol/l, p = 0.007) and decreased urine osmolality (514 (IQR: 465-605) mmol/l vs 602 (IQR: 467-740) mmol/l, p < 0.001). In the subgroup of patients with complete sodium and water balance, a positive salt and water balance was observed. After propensity score matching, we found a signicantly increased electrolyte free water clearance (1020 ±1740 ml vs -560 ±1620 ml, p <0.001) in the hypernatremia group, together with an inadequately lower total sodium urine excretion (401 ±303 mmol vs 593 ±400 mmol, p = 0.02). Conclusion: The present study shows that postoperative hypernatremia is associated with an imbalance between perioperative salt and water load and renal sodium and water handling with inadequately low renal sodium excretion and inadequately high renal water excretion. The underlying renal concentration disorder may be explained by a defect in a natriuretic-ureotelic response a recently described renal urea-mediated water conservation mechanism after salt exposure. are mean values collected within four postoperative days. For patients with propensity score matching, the following matching variables were applied: total water intake, total sodium intake, maximum argipressin dose per day, maximum loop diuretic dose per day, maximum hydrocortisone dose per day, maximum norepinephrine dose per day. Results are presented as median [interquartile range] or as means ± standard deviations. In case of non-signicance, the results are presented as “ns”. Abbreviations: SAPS II, Simplied Acute Physiology Score; EFWC, electrolyte free water clearance SAPS II, Simplied

This regulatory response to increased dietary salt consumption can possibly be transferred to a scenario in which salt exposure is caused by the administration of saline solutions. Perhaps similar mechanisms are triggered here to ensure that salt and water are excreted in an appropriate ratio. And probably ICU acquired hypernatremia could be explained by a combination of salt exposure during volume loading and impaired renal urine-concentrating ability due to missing natriuretic-ureotelic response.
Perioperatively during major surgery, patients are exposed to high salt loads within a short time, making this period suitable for studying the renal response. The present retrospective study aims to characterize the development of hypernatremia in postoperative patients. The main objective is to nd evidence for a disturbed urine concentration in the sense of an impaired natriuretic-ureotelic response in this group of patients. To the authors' knowledge, it is the rst time that this new concept is applied to perioperative patients under volume and salt load.

Methods:
This is a retrospective study that we conducted in a 40-bed tertiary surgical intensive care unit of the University Medical Center Augsburg. We screened a database of all patients (1375) admitted between July 2019 and December 2019. Data was extracted from the patient data management system (Orbis Agfa, Bonn Germany). The following patient characteristics were used: age, gender, Simpli ed Acute Physiology Score (SAPS) II, type of surgery, in-hospital mortality, and if available routine daily measurements of serum and urine electrolytes, urea, osmolality and creatinine. Due to our patient database management system water balances as well as sodium and potassium content of all supplied substances are automatically recorded in our electronic chart. It is part of our daily routine that 24 h urine analyses are regularly performed.
We de ned hypernatremia as a serum sodium value exceeding 145 mmol/l. In the analysis of all patients, we examined in the group with normonatremic patients the values available within the rst four days after surgery as long as the patients were still in the intensive care unit. The hypernatremia group comprised all patients who developed hypernatremia during their intensive care stay. The patients were only included, if the onset of hypernatremia occurred less than four days after surgery. The next four days were examined from that point on. We always used the rst serum values of the day.
From the existing values of the four days for serum sodium, urine sodium and urine osmolality we calculated average values. The maximum values and the duration of hypernatremia refer to the entire intensive care unit stay.
For the SAPS II values, we recorded the highest value that occurred within the rst 96 hours examined. In order to record a SAPS II score independently from hypernatremia, which is included in the score, one point was subtracted from the score on days the patients developed hypernatremia, as already described elsewhere [8].
Since we only wanted to select and examine patients who developed hypernatremia in the intensive care unit after major surgery, we excluded all patients without surgery, neurosurgical patients, patients with hyponatremia (< 135 mmol/l) or hypernatremia already present at admission and patients under the age of 18. Beyond that, we only included patients with major surgery (Fig. 1).
Of the 168 patients with hypernatremia, 71 patients had enough urine values, which enabled a complete balancing over the rst three or four postoperative days. These 71 hypernatremic patients could be compared with 71 normonatremic patients who did not differ signi cantly regarding water and sodium intake perioperatively on day one and water intake over the entire three or four days. For further analysis, two subgroups were formed within the normonatremic patients. These subgroups were divided into patients (n = 29) who had no urine sodium values above 100 mmol/l and patients (n = 42) who reached values above 100 mmol/l (Fig. 1).
We calculated an electrolyte free water balance (EFWC) using the following formula [6]: Additionally, to show the difference between the sodium concentration of the input side and the urine sodium concentration, we calculated a non-isotonic sodium balance determining the sodium load which is not isotonic: In the non-isotonic sodium balance as well as in the sodium and potassium balance, we did not include any not clearly determinable quantities like perspiratio, or other body uids like stool or re ux, because either volume or sodium content could only be estimated. Nethertheless, these uid losses were included in the water balance.
For calculating the tonicity of the added solutions, we used only the sodium concentration with the following formula: Tonicity of Total Intake = Total Sodium Intake / Total Water Intake For comparing the drugs applied, we used the maximum dose per day of loop diuretics, hydrocortisone, argipressin and norepinephrine within the 4 days studied.

Statistical analysis
Data were analysed using R version 4.0.0 with the following external packages: Table1, sjPlot, sjmisc, rms, MatchIt, coin [9]. Means were reported with standard deviations and medians with their interquartile ranges. The nonparametric Fisher's Exact Test was used with categorical data and the nonparametric Mann-Whitney U test with numerical data because data was not normally distributed. Statistical signi cance was set at a p value of less than 0.05 for all tests. Multivariable logistic regression analysis was used to explore the association between sodium as exposure and hospital mortality as outcome.
Logistic regression models were also used to study the association of urine sodium concentration, nonisotonic sodium balance and various factors such as diuretic use, hydrocortisone and catecholamines with the development of hypernatremia. We checked for non-linear relationships between continuous covariates and the log-odds using restricted cubic splines. Hypernatremia was used as the dependent variable. We applied propensity score matching in the fully balanced hypernatremic and normonatremic patients by using as matching variable total water intake, total sodium intake, cumulative argipressin dose, cumulative loop diuretic dose, cumulative hydrocortisone dose and cumulative norepinephrine dose ( Fig. 1).  Table S1).
Results are presented as median [interquartile range], as means ± standard deviations or as absolute number (%). In the group with major surgery patients, all patients were excluded without major surgery, neurosurgical patients, patients with hyponatremia (< 135 mmol/l) or hypernatremia already at admission. In case of non-signi cance, the results are presented as "ns". Abbreviations: SAPS II, Simpli ed Acute Physiology Score A similar pattern was seen in the group of 142 patients with complete water and sodium balance ( Table 3). Sodium and water intake on day of surgery and total water intake over the rst four postoperative days did not differ signi cantly between patients with hypernatremia and patients with  (Table 3 and Fig. 2).  (Table 3 and Fig. 2). In contrast to the lower total sodium urine excretion (381 ± 270 mmol versus 547 ± 364 mmol, p = 0.005), the average tonicity of volume intake was higher (131 ± 15.9 mmol/l versus 115 ± 19.1 mmol/l, p < 0.001). Total water balance, total urea balance did not differ. The water balance was positive, whereas the urea balance was negative.
The potassium balance was negative in both cases without signi cant difference.
In order to verify whether the lower urine sodium concentration and the higher tonicity of the added infusion solutions are indeed signi cantly associated with the development of hypernatremia, we applied a multivariate logistic regression model (Additional le, Table S2). After adjusting for loop diuretics, hydrocortisone, argipressin, norepinephrine, total water intake and corrected SAPS II, logistic regression analysis revealed only a signi cant association with the average urine sodium concentration (OR 0.98 (95% CI 0.96-0.99), p = 0.001) and the tonicity of the added volume (OR 1.05 (95% CI 1.02-1.08), p < 0.001) (Additional le, Table S2A). In an additional model we could con rm a signi cant association between the non-isotonic sodium balance (OR 1.004 (95% CI 1.002-1.007), p < 0.001) and the development of hypernatremia as well (Additional le, Table S2B).
In another approach, we used propensity score matching to create groups of hypernatremic and normonatremic patients with adjusted water and salt intake and drug therapy (  Table 3, lower part). The SAPS II value in the hypernatremia group was lower than before due to selection of patients within the propensity score matching.
A subgroup of the fully balanced patients with normonatremia had no increased urine sodium concentration despite sodium exposure (Table 4). Urine urea and urine osmolality in this group did not differ from the values of other patients with normonatremia. However, the total water balance was more positive (2610 ± 2770 ml vs. 1180 ± 2300 ml, p = 0.01) due to reduced total urine volume (4870 ± 1820 ml vs. 7110 ± 2190 ml, p < 0.001) ( Table 4 and Fig. 3). After propensity score matching (Table 4, lower part) in order to examine patients with comparable sodium and water load and to avoid differences in tonicity, the reduced urine excretion (4870 ± 1820 ml vs. 7180 ± 2240 ml, p < 0.001) and thus the more positive water balance (2610 ± 2770 ml vs. 425 ± 1960 ml, p < 0.001) could be con rmed signi cant. A gradual increase in natriuresis was observed in some normonatriemic patients. The course of this pattern of renal sodium excretion in some patients (n = 27) as opposed to patients developing hypernatremia (n = 58) is shown in Fig. 4. Discussion: The present study shows that in surgical intensive care patients undergoing major surgery hypernatremia was a frequent nding with an incidence of 23.9%. We were able to demonstrate that hypernatremia developed within the rst days after surgery and was associated with an increased mortality rate.
In regard to the underlying mechanism of hypernatremia in our study, we see evidence for urine concentration defect with lowered urine sodium values and lowered urine osmolality values in all patients developing hypernatremia after major surgery. In patients with complete salt and water balance, an association between the development of hypernatremia and a combination of sodium load and concomitant renal inability to excrete sodium su ciently was detected. With logistic regression analysis and propensity score matching, we con rm that this association was signi cant.
The incidence of ICU-acquired hypernatremia in our study is comparable to other studies conducted in surgical and medical intensive care units [14,16]. Compared to two studies with exclusively cardiac surgical patients, the incidence and mortality is higher [2,4], which may be partly explained by the more severe degree of disease in our study re ected in higher SAPS II values [2].
Patients undergoing major surgery are exposed to a considerable salt and volume load within a short period of time. Therefore, perioperative patients are very suitable to study how the body handles salt exposure. In our study, we saw a positive water and salt balance on the day of surgery, which continues until the 4th postoperative day in both hypernatremic and normonatremic patients (Table 3).
Normonatremic patients were able to increase their natriuresis as shown by the higher urine sodium levels ( Fig. 2 and Table 3). At the same time the urine concentrating capacity was maintained without any additional loss of water re ected by the signi cantly higher osmolality ( Fig. 2 and Table 3). In contrast to this, patients with hypernatremia were not able to increase natriuresis. Simultaneously they lost too much water in relationship to sodium. Hypernatremia developed due to inadequate excretion of free water in relation to the supply of free water. In the balance, this was re ected in a signi cantly increased amount of non-isotonic sodium in the hypernatremia group and the increased EFWC ( Fig. 2 and Table 3). The relevance of the lowered urine sodium concentration and the increased non-isotonic sodium balance was con rmed in the logistic regression analysis (Additional le, S2). The corresponding changes remained signi cant after propensity score matching (Table 3).
In some previous studies, a urine concentration defect in combination with salt exposure already was discussed [1,6]. A recently published study proposed a natriuretic-ureotelic adaption in response to salt load introducing a new aspect of renal salt handling [7]. In brief, this model comprises three steps. First, aldosterone must be suppressed to ensure increased salt excretion in the collection tube. Secondly, in order to prevent salt-induced water excretion, the medullary urea concentration is increased via urea transporter, mainly UTA-1, in the distal collecting tube. The increased medullary urea concentration provides the necessary osmotic driving force to avoid inadequate water loss. Thirdly, there is a complex catabolic metabolic conversion with urea production and metabolic water production (Fig. 5). These changes, affecting the whole organism besides the kidneys, are described in detail in the abovementioned work on mice that receive salt but do not have su cient access to water. Evidence for a similar pattern supposed to protect against dehydration can also be found in humans [7].
Certainly, it needs to be discussed to what extent this concept can be applied to the postoperative intensive care situation. In contrast to the probands of the study by Kitada [7], postoperative patients are not only exposed to salt load but also to volume load, which possibly provokes different physiological reactions. Moreover, postoperative patients are exposed to considerable circulatory stress with catecholamine support. Factors such as diuretics and the administration of hydrocortisone may have an additional impact.
Nevertheless, previous studies on volunteers after saline infusions showed how patients were able to excrete sodium and water in adequate ratio [1] with simultaneously suppressed aldosterone [11], consistent with the described concept. Similarly in our study, normonatremic patients reacted with an increase in natriuresis without signs of inadequate water diuresis, despite considerable salt and volume load. Figure 4 shows a subgroup of normonatremic patients with stepwise increase in urine sodium concentration possibly demonstrating the interplay between declining aldosterone and urea driven water conservation mechanism in the collecting duct. Correspondingly in the sodium and water balance, the non-isotonic sodium reaches negative values. In the hypernatremia group, this response does not occur.
Are there other mechanisms explaining the decreased urine sodium concentration and the urine concentrating defect with decreased urine osmolality in patients with hypernatremia? Compared to patients with normal sodium values, hypernatremic patients were sicker supported by the higher corrected and not corrected SAPS II values. They had higher doses of norepinephrine and empressin and received hydrocortisone more frequently (Table 3). However, excessive mineral corticoid stimulation resulting from this seems unlikely to be the only reason, since the inadequate water excretion cannot be explained in this way. Moreover, there was no difference in the potassium balance (Table 3). Supporting this, in logistic regression analysis, no association could be seen between the development of hypernatremia and the administration of loop diuretics, hydrocortisone, noradrenalin or argipressin (Additional le, S2). Interestingly, the potassium balance in both groups was negative as observed in a study with heart surgery patients [12].
In some studies osmotic diuresis is discussed [13]. However, the patients in the present study did not ful l the criteria of a polyuria with increased urine quantity, did not receive osmotically active substances such as mannitol and did not suffer from excessive blood sugar levels. Criteria for urea-induced osmotic diuresis were also not met [13].
The study by Kita demonstrates how a urea gradient is built up simultaneously to increased natriuresis [7]. The authors can show that UTA-1 urea transporters are increasingly expressed. Conversely, it is known from knockout mice that, in absence of the UTA-1 transporter, a concentration defect occurs, which causes urine to be produced with low osmolality, urea and sodium and inadequate water losses [14,15].
A pattern with a certain similarity to the patients examined here. Possibly, similar pathophysiological mechanisms are present here in the sense of a defect urea-mediated urine concentration (Fig. 5).
In addition to the renal concentration disorder, other extrarenal factors may be involved causing hypernatremia after surgery. We have to question to what extent balances are at all able to explain changes in sodium concentration [16]. There is evidence that sodium is stored non-osmotically in critically ill patients [17]. Moreover, there are always inaccuracies in the balance due to factors with unknown quantity or unknown sodium concentration like perspiration, gastric juice, stool or internal bleeding. In our study, these factors are not included in the calculation of non-isotonic sodium which does not mean that the non-isotonic sodium is not subject to quantitative error. In terms of quality, however, the difference is very clear and signi cant, so that the balance of non-isotonic sodium certainly covers the main components contributing to the development of hypernatremia. The signi cance of the association between hypernatremia and changes in the non-isotonic sodium balance is con rmed by logistic regression analysis (Additional le, S2). Apart from the lack of sodium excretion with simultaneous inadequate water diuresis, it is noticeable that patients with hypernatremia have a higher tonicity of added uids. However, in both groups the tonicity is below the tonicity of the plasma. The higher tonicity in hypernatremic patients can be explained by the fact that these patients are sicker, receive higher doses of catecholamines and therefore receive isotonic uids as volume therapy for longer, in contrast to normonatremic patients who get access to free water earlier. With propensity score matching we were able to eliminate the in uence of the tonicity of the added volume and reveal that hypernatremic patients have above all a renal urine concentration disturbance (Table 3).
In the normonatremic group, there is a subgroup of patients who showed no increase in urine sodium despite similar perioperative volume load (Table 4). This can be explained by the more positive water balance due to lower urine volumes. Apparently this group saves more volume during the examined postoperative 4 days and therefore does not need to increase the sodium concentration in the urine. Hence the renal response to the supply of water and salt is adequate. After propensity score matching, the signi cantly more positive balance remains in these patients (Table 4, lower part).
A weakness of the present study is its retrospective nature, so a considerable number of patients do not have su cient data for complete balances, despite our clinical practice of performing regular urine analyses. Indirect evidence for the mechanisms that take place during volume and salt loading can be found with the help of the balances. Whether a natriuretic ureotelic response is indeed the basis of the changes in urine concentrating processes, and whether the development of hypernatremia actually has its origin in a lack of response, can only be shown indirectly on the basis of the described developments.

Conclusion:
In summary, the study shows that hypernatremia can develop postoperatively after major surgery. We demonstrate that the development of hypernatremia was comprehensible with sodium and water balances. We were able to show for the rst time in a larger patient cohort that a defect in urine concentration was signi cantly associated with hypernatremia. We demonstrate that the change in urinary sodium excretion due to a large volume and salt load corresponded to a pattern that could be in line with the natriuretic ureotelic concept and that this response could be impaired in the pathogenesis of hypernatremia. Regarding therapy of hypernatremia, for which hardly any data exists [18], the lower tonicity of the infusion volume seen in our normonatremic patients could be an interesting factor.
List Of Abbreviations SAPS II, Simplified Acute Physiology Score II; IQR, interquartile range; OR, odds ratio; CI, con dence interval;

Declarations
Ethics approval and consent to participate   Serum and urine values of completely balanced patients (A) as well as electrolyte free water clearance (EFWC) and non-isotonic sodium balance (B) with normal sodium (n=71) and hypernatremia (n=71).

Figure 3
Subgroup of normonatremic patients divided into patients (low urine sodium, n=29) who had no urine sodium values above 100 mmol/l and patients (high urine sodium, n=42) who reached values above 100 mmol/l. The differences in total urine volume and total water balance are shown.

Figure 5
The concept of natriuretic-ureotelic adaption in response to salt load is supposed to protect against dehydration [7]. (A) Aldosterone must be suppressed to ensure increased salt excretion in the collection tube. (B) In order to prevent salt-induced water excretion, the medullary urea concentration is increased via urea transporter, mainly UTA-1, in the distal collecting tube. (C) The increased medullary urea concentration provides the necessary osmotic driving force to avoid inadequate water loss. (D) Metabolic changes with urea production and metabolic water production as further measures to conserve bodywater.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. AdditionalFile.doc