2023 年 46 巻 10 号 p. 1365-1370
Several cases of severe hyponatremia induced by linezolid (LZD) were reported. However, severe infections could also cause hyponatremia by increasing vasopressin secretion. To prove that hyponatremia is associated with LZD rather than infection, we compared the incidence and risk of developing hyponatremia between patients receiving LZD and those receiving vancomycin (VCM). A retrospective, single-center, observational cohort study was conducted in patients aged 18 years or older who received intravenous LZD or VCM for 7 d or longer. Hyponatremia was defined as serum sodium level lower than 134 mEq/L and more than 5% decrease from baseline after treatment initiation. The incidence and risk of developing hyponatremia were analyzed between LZD and VCM groups using chi-square test. Four hundred and fifty patients who satisfied the selection criteria were divided into LZD (n = 97) and VCM groups (n = 353). Significant differences in patient characteristics between LZD and VCM groups were observed before propensity score matching, but no significant differences were found after matching. LZD group showed a significantly higher incidence and risk of developing hyponatremia compared to VCM group both before (LZD: 16.5%, VCM: 5.4%; p < 0.001, odds ratio 3.472 [95% confidence interval (CI) 1.711–7.048]) and after (LZD: 17.8%, VCM: 5.5%; p = 0.020, odds ratio 3.738 [95% CI 1.157–12.076]) propensity score matching. In conclusion, propensity score analyses suggest that the risk of hyponatremia associated with LZD is approximately 3.7-fold higher than that associated with VCM, regardless of patient background.
Linezolid (LZD), an oxazolidinone antibacterial agent, exerts bacteriostatic activity against Gram-positive resistant bacteria such as methicillin-resistant Staphylococcus aureus (MRSA), vancomycin-resistant Staphylococcus aureus and vancomycin-resistant Enterococcus.1) Both oral and injectable formulations can be used depending on the condition of the patient, as oral bioavailability is high at approximately 100%.2) This drug distributes abundantly to tissues such as epithelial lining fluid, spinal fluid and osseous tissue, relative to other anti-MRSA agents.3–5) Because of these benefits, the drug is often used as empirical therapy for various infections.
On the other hand, LZD causes numerous specific adverse effects such as thrombocytopenia, anemia, serotonin syndrome, optic neuropathy, and metabolic acidosis,6–11) which often leads to treatment discontinuation. In addition to these adverse effects, reports of severe hyponatremia while receiving LZD have recently increased. Four case reports have been published,12–15) two of which suggest that LZD-induced hyponatremia may be caused by syndrome of inappropriate antidiuretic hormone secretion (SIADH).13,14) Nishi et al.16) reported that among subjects who received LZD and had measured plasma LZD concentrations, 23.6% developed hyponatremia (defined as a nadir serum sodium level ≤130 mmol/L). They also revealed the association of hyponatremia development with LZD exposure (area under the curve (AUC)0–12 and accumulated AUC), baseline serum sodium levels and age. Moreover, previous retrospective research indicates that some patients receiving LZD develop hyponatremia and thrombocytopenia.17)
Hyponatremia is a condition of excessive total body water relative to total body sodium, commonly defined as a status of reduced serum sodium level lower than 134 mEq/L. Hyponatremia could lead to clinical symptoms ranging from minor symptoms such as headaches, nausea and anorexia to severe and even life-threatening conditions.18) Hyponatremia is classified into three types; hypertonic, isotonic and hypotonic, depending on the difference in osmotic pressure.19) These three types have different etiologies, and the onset of SIADH evokes hypotonic hyponatremia. SIADH is a condition in which vasopressin, an antidiuretic hormone, is secreted excessively and inappropriately. Some medications such as antidepressants have been reported to cause SIADH as an adverse effect,19) while infections can also cause it.20) In severe infection, inflammatory cytokines such as tumor necrosis factor-α (TNF-α) and interleukin-6 (IL-6) are secreted abundantly into the circulation. A previous review suggests that these cytokines, especially IL-6, promote vasopressin secretion under physiological conditions.21) Therefore, the possibility that hyponatremia is caused by changes in physiological conditions due to infection and not by linezolid cannot be ruled out. This hypothesis may be partially supported by the previous finding that C-reactive protein (CRP) level before LZD initiation is a risk factor for hyponatremia.22)
Given the above background, this study aimed to prove that hyponatremia is associated with LZD use rather than infection by comparing the incidence and risk of developing hyponatremia between patients receiving LZD and patients receiving vancomycin (VCM), another anti-MRSA drug. However, various factors other than severity and type of infection may increase the risk of hyponatremia development. Hence, propensity score matching was used to minimize the impact of biases generated from the differences in patient characteristics between LZD and VCM groups on the incidence of hyponatremia.
This retrospective, single-center, observational cohort study reviewed electronic medical records to extract the subjects’ clinical information. This study included patients aged 18 years or older who received LZD or VCM intravenously for 7 d or more between April 2015 and March 2021 at Oita University Hospital. Exclusion criteria were as follows: admission to intensive care unit or emergency center during the period of LZD or VCM administration; no routine measurement of serum sodium level at least twice before and during LZD or VCM administration; serum sodium level below 130.0 mEq/L before LZD or VCM initiation; and switch between LZD and VCM. VCM was selected as control because it is most commonly and widely used for MRSA infections in Japan. We estimated that there would be a sufficient number of cases for analysis even after performing propensity score matching to minimize biases due to differences in background factors. The study was conducted in accordance with the ethical standards of our institute and with the Helsinki Declaration of 1975, as revised in 2013. The protocol for this study was approved by the Oita University Faculty of Medicine Ethics Committee (review reference number: 1420) before study initiation.
Data Collection and Definition of Hyponatremia and Acute Kidney InjuryThe following demographic and laboratory parameters before LZD or VCM treatment were collected: sex, age, body weight, white blood cell count, platelet count, serum albumin, serum creatinine, CRP, and serum sodium. Creatinine clearance was calculated by the Cockcroft-Gault equation.23) Daily calorie intake was estimated based on the data obtained from the hospital meals, and was calculated from the food items in each meal and the percentage of each item consumed. Concomitant drugs used continuously for more than 3 d from LZD or VCM initiation were recorded. Diuretics (thiazides, loop diuretics), antidepressants (tricyclic antidepressants, selective serotonin reuptake inhibitors, venlafaxine), antipsychotic drugs (phenothiazines, butyrophenones), antiepileptic drugs (carbamazepine, sodium valproate, lamotrigine), antidiabetic drugs (sulfonylurea hypoglycemics), nonsteroidal anti-inflammatory drugs, and anticancer drugs (vinca alkaloids, platinum compounds, alkylating agents, miscellaneous) were defined as concomitant drugs that may cause hyponatremia, based on a review article.24) Anticancer drugs administered as a single dose were included in concomitant drugs, even though they were not used on consecutive days. Tolvaptan was also defined as a concomitant drug because it may cause hypernatremia. The types and volumes of all infusions and solutions used on the day of LZD or VCM initiation were extracted to calculate daily sodium intake for each patient. Information of infectious disease was obtained from medical records before and during treatment and classified into the following nine types: sepsis, catheter-related infection, febrile neutropenia, pneumonia, osteomyelitis or infectious arthritis, infectious meningitis, wound infection or cellulitis or abscess, others, and unknown. Both definitive and suspected infectious diseases were included. Results of cultures conducted immediately before or during treatment for infection were collected. Detection of MRSA in blood, catheter, sputum, urine, spinal fluid, joint fluid, drain, pleural effusion, peritoneal effusion, secretion, tissue aspirate, or wound tissue was recorded. The drug administration period was also extracted from electronic medical records. Serum sodium levels were followed from drug initiation to one day after discontinuation. Hyponatremia was defined as serum sodium level lower than 134 mEq/L and more than 5% decrease from baseline after initiation of LZD or VCM, as reported previously.17) As such, patients with serum sodium levels of 130 to 134 mEq/L at baseline were deemed as hyponatremia when decreasing 5% from baseline. Acute kidney injury was defined as 0.5 mg/dL or more than 50% increase in serum creatinine level compared to that before initiation of each drug.25)
Statistical AnalysisPredictive Analysis Software (PASW) Statistics version 27 (SPSS Inc., Chicago, IL, U.S.A.) was used for statistical analysis. Data are expressed as number (%) for categorical variables and median [interquartile range] for continuous variables. All patients were divided into LZD and VCM groups. Continuous variables were analyzed by Mann–Whitney U test. Categorical variables were analyzed by chi-square test, or by Fisher’s exact test when more than 20% of the cells have expected frequencies <5. Statistical significance was set at p < 0.05.
One-to-one matching of propensity scores was used to adjust for differences in patient characteristics between two groups. A logistic regression model was adapted to estimate propensity scores, and c-index was calculated to assess the goodness of fit. Using the nearest-neighbor matching method, one patient in one group was matched to a patient in the other group based on estimated propensity (0.2 standard deviations) nearest within the caliper without replacement strategy. The demographic and laboratory variables incorporated into the propensity score one-to-one matching comprised sex, age, body weight, white blood cell count, platelet count, serum albumin, CRP, creatinine clearance, serum sodium, administration period, dietary intake, sodium intake from infusions and solutions, MRSA detection, infection type, and concomitant drugs. After propensity score matching, adjusted p-values and odds ratios were estimated using the same statistical methods as for the data before propensity score matching.
Patient characteristics prior to initiation of VCM or LZD administration are listed in Table 1. A total of 450 patients satisfied the selection criteria, and were divided into LZD group (n = 97) and VCM group (n = 353). Compared to VCM group, LZD group showed significantly higher body weight (p = 0.002), platelet count (p < 0.001), serum albumin level (p < 0.001), serum creatinine level (p = 0.017), and daily dietary intake (p = 0.003), as well as lower CRP level (p = 0.046) and daily sodium intake from infusions and solutions (p < 0.001). Compared to VCM, LZD was more frequently used for osteomyelitis or infectious arthritis (p < 0.001) and wound infection, cellulitis or abscess (p < 0.001), but less frequently for catheter-related infection (p = 0.004), febrile neutropenia (p = 0.001), and pneumonia (p = 0.007). As concomitant drugs, LZD group received diuretics (p < 0.001) and nonsteroidal anti-inflammatory drugs (p = 0.017) more frequently than VCM group, while VCM group received anticancer drugs (p < 0.001) more often than LZD group. There were no significant differences in the other clinical and laboratory parameters between LZD and VCM groups.
| Characteristics | Linezolid | Vancomycin | p-Value |
|---|---|---|---|
| Total patients; n | 97 | 353 | |
| Sex (male/female); n (%) | 61 (62.9)/36 (37.1) | 248 (70.3)/105 (29.7) | 0.166a) |
| Age (years) | 67.6 [55.4–80.7] | 68.0 [57.0–76.0] | 0.609c) |
| Body weight (kg) | 61.3 [51.6–70.8] | 53.6 [46.4–62.4] | 0.002c) |
| White blood cell count (×103 cells/mm3) | 7.28 [5.27–10.37] | 7.18 [4.17–10.68] | 0.940c) |
| Platelet count (×103 cells/mm3) | 268.0 [217.0–336.0] | 215.0 [100.0–294.0] | <0.001c) |
| Serum albumin (g/dL) | 3.03 [2.47–3.46] | 2.69 [2.24–3.25] | <0.001c) |
| Serum creatinine (mg/dL) | 0.83 [0.62–1.24] | 0.71 [0.58–0.95] | 0.017c) |
| C-reactive protein (mg/dL) | 4.71 [1.42–9.90] | 7.34 [2.05–12.7] | 0.046c) |
| Serum sodium (mEq/L) | 138.1 [135.9–140.2] | 137.9 [135.1–140.4] | 0.228c) |
| Creatinine clearance (mL/min) | 69.4 [39.2–109.1] | 76.0 [59.3–97.9] | 0.671c) |
| Administration period (d) | 12.0 [9.0–14.0] | 11.0 [8.0–14.0] | 0.300c) |
| Dietary intake (kcal/d) | 633.3 [16.7–1266.7] | 450.0 [0.0–1066.7] | 0.003c) |
| Sodium intake from infusions and solutions (mEq/d) | 80.9 [15.4–149.7] | 130.8 [65.9–209.9] | <0.001c) |
| Methicillin-resistant Staphylococcus aureus detection; n (%) | 37 (38.1) | 65 (38.0) | 0.448a) |
| Infection; n (%) | |||
| Sepsis | 2 (2.1) | 25 (7.1) | 0.065a) |
| Catheter-related infection | 0 (0.0) | 29 (8.2) | 0.004a) |
| Febrile neutropenia | 0 (0.0) | 36 (10.2) | 0.001a) |
| Pneumonia | 6 (6.2) | 61 (17.3) | 0.007a) |
| Osteomyelitis or infectious arthritis | 21 (21.6) | 7 (2.0) | <0.001a) |
| Infectious meningitis | 2 (2.1) | 18 (5.1) | 0.156b) |
| Wound infection, cellulitis or abscess | 46 (47.4) | 79 (22.4) | <0.001a) |
| Others | 10 (10.3) | 55 (15.6) | 0.191a) |
| Unknown | 10 (10.3) | 43 (12.2) | 0.612a) |
| Concomitant drugs; n (%) | |||
| Diuretics | 21 (21.6) | 30 (8.5) | <0.001a) |
| Antidepressants | 0 (0.0) | 11 (3.1) | 0.067b) |
| Antipsychotic drugs | 2 (2.1) | 16 (4.5) | 0.216b) |
| Antiepileptic drugs | 4 (4.1) | 15 (4.2) | 0.609b) |
| Antidiabetic drugs | 0 (0.0) | 4 (1.1) | 0.377b) |
| Nonsteroidal anti-inflammatory drugs | 30 (30.9) | 69 (19.5) | 0.017a) |
| Anticancer drugs | 7 (7.2) | 91 (25.8) | <0.001a) |
| Tolvaptan | 2 (2.1) | 9 (2.5) | 0.565b) |
Data are expressed as number (%) for categorical variables and median [interquartile range] for continuous variables. p-Values were obtained by univariate analyses comparing linezolid group and vancomycin group. Categorical variables were analyzed by a) chi-square test or b) Fisher's exact test. Continuous variables were analyzed by c) Mann–Whitney U test.
C-index calculated as the area under the curve of propensity scores was 0.868, suggesting adequate goodness of fit. A total of 73 patients in each group were matched by propensity score. The patient characteristics in each group after propensity score matching are listed in Table 2. In contrast to before matching, there were no significant differences in all the demographic and laboratory parameters between LZD and VCM groups.
| Characteristics | Linezolid | Vancomycin | p-Value |
|---|---|---|---|
| Total patients; n | 73 | 73 | |
| Sex (male/female); n (%) | 23 (31.5)/50 (68.5) | 24 (32.9)/49 (67.1) | 0.859a) |
| Age (years) | 67.6 [55.4–81.0] | 67.7 [58.3–76.1] | 0.713c) |
| Body weight (kg) | 57.8 [50.5–69.5] | 57.0 [47.3–65.7] | 0.693c) |
| White blood cell count (×103 cells/mm3) | 7.90 [5.27–10.48] | 7.54 [5.39–10.54] | 0.868c) |
| Platelet count (×103 cells/mm3) | 273.0 [218.0–333.0] | 254.0 [193.0–359.0] | 0.595c) |
| Serum albumin (g/dL) | 2.94 [2.42–3.45] | 2.86 [2.52–3.40] | 0.693c) |
| Serum creatinine (mg/dL) | 0.79 [0.61–1.14] | 0.70 [0.59–0.91] | 0.148c) |
| C-reactive protein (mg/dL) | 4.15 [1.23–10.02] | 4.12 [0.90–11.40] | 0.879c) |
| Serum sodium (mEq/L) | 137.8 [136.0–140.1] | 138.5 [135.6–141.1] | 0.385c) |
| Creatinine clearance (mL/min) | 73.1 [39.9–109.1] | 71.6 [53.7–102.0] | 0.781c) |
| Administration period (d) | 11.0 [9.0–14.0] | 11.0 [9.0–14.0] | 0.852c) |
| Dietary intake (kcal/d) | 600.0 [16.7–1256.7] | 850.0 [120.0–1360.0] | 0.563c) |
| Sodium intake from infusions and solutions (mEq/d) | 92.4 [30.8–177.2] | 117.6 [32.9–177.2] | 0.218c) |
| Methicillin-resistant Staphylococcus aureus detection; n (%) | 30 (41.1) | 28 (38.4) | 0.735a) |
| Infection; n (%) | |||
| Sepsis | 2 (2.7) | 3 (4.1) | 0.500b) |
| Catheter-related infection | 0 (0.0) | 0 (0.0) | — |
| Febrile neutropenia | 0 (0.0) | 0 (0.0) | — |
| Pneumonia | 6 (8.2) | 2 (2.7) | 0.138b) |
| Osteomyelitis or infectious arthritis | 4 (5.5) | 4 (5.5) | 0.641b) |
| Infectious meningitis | 2 (2.7) | 2 (2.7) | 0.690b) |
| Wound infection, cellulitis or abscess | 39 (53.4) | 38 (52.1) | 0.868a) |
| Others | 10 (13.7) | 12 (16.4) | 0.644a) |
| Unknown | 10 (13.7) | 12 (16.4) | 0.644a) |
| Concomitant drugs; n (%) | |||
| Diuretics | 15 (20.5) | 12 (16.4) | 0.522a) |
| Antidepressants | 0 (0.0) | 0 (0.0) | — |
| Antipsychotic drugs | 2 (2.7) | 3 (4.1) | 0.500b) |
| Antiepileptic drugs | 3 (4.1) | 4 (5.5) | 0.500b) |
| Antidiabetic drugs | 0 (0.0) | 0 (0.0) | — |
| Nonsteroidal anti-inflammatory drugs | 22 (30.1) | 23 (31.5) | 0.858a) |
| Anticancer drugs | 7 (9.6) | 10 (13.7) | 0.439a) |
| Tolvaptan | 2 (2.7) | 1 (1.4) | 0.500b) |
Data are expressed as number (%) for categorical variables and median [interquartile range] for continuous variables. p-Values were obtained by univariate analyses comparing linezolid group and vancomycin group. Categorical variables were analyzed by a) chi-square test or b) Fisher's exact test. Continuous variables were analyzed by c) Mann–Whitney U test.
Table 3 presents the incidence and risk of developing hyponatremia before and after propensity score matching. LZD group showed a significantly higher incidence of developing hyponatremia compared to VCM group not only before (LZD: 16.5% [16 of 97 patients], VCM: 5.4% [19 of 353 patients]; p < 0.001) but also after (LZD: 17.8% [13 of 73 patients], VCM: 5.5% [4 of 73 patients]; p = 0.020) propensity score matching. The odds ratios for hyponatremia before and after matching were 3.472 [95% confidence interval (CI) 1.711–7.048] and 3.738 [95% CI 1.157–12.076], respectively, indicating an approximately 3.7-fold higher risk of hyponatremia associated with LZD than with VCM.
| Before propensity score matching | After propensity score matching | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| LZD | VCM | p-Value | Odds ratio | 95%CI | LZD | VCM | p-Value | Odds ratio | 95%CI | |
| Hyponatremia (+) | 16 (16.5) | 19 (5.4) | <0.001 | 3.472 | 1.711–7.048 | 13 (17.8) | 4 (5.5) | 0.020 | 3.738 | 1.157–12.076 |
| Hyponatremia (−) | 81 (83.5) | 334 (94.6) | 60 (82.2) | 69 (94.5) | ||||||
Data are expressed as number (%). p-Values, odds ratio, and 95%CI were obtained by chi-square test comparing linezolid group and vancomycin group, before and after propensity score matching. LZD, linezolid; VCM, vancomycin; CI, confidence interval.
Table 4 shows the characteristics of 16 and 19 patients with hyponatremia in LZD and VCM groups, respectively. Median minimum serum sodium level was 125.5 in LZD group and 129.2 mmol/L in VCM group, with no statistically significant difference (p = 0.082). VCM frequently causes acute kidney injury via direct cytotoxicity to renal tubular epithelial cells,26) suppressing sodium reuptake from urine and leading to hyponatremia. The incidence of acute kidney injury in patients with hyponatremia after the initiation of VCM was 36.8%, which was approximately three times higher than LZD (12.5%), but no statistically significant difference was observed (p = 0.135).
| Drug | Serum sodium concentration (mmol/L) | Period from start of treatment to onset (d) | Incidence of acute kidney injury | ||
|---|---|---|---|---|---|
| Before treatment | Minimum during treatment | Rate of decrease (%) | |||
| Linezolid | 137.5 [134.8–140.5] | 125.5 [123.4–129.4] | 7.6 [6.8–9.5] | 10.0 [8.5–14.5] | 2 (12.5) |
| Vancomycin | 140.0 [135.8–141.7] | 129.2 [127.1–131.1] | 7.3 [6.0–9.2] | 10.0 [6.0–13.0] | 7 (36.8) |
| p-Value | 0.301 | 0.082 | 0.243 | 0.909 | 0.135 |
There were 16 patients in linezolid group and 19 patients in vancomycin group. Data are expressed as number (%) for categorical variables and median [interquartile range] for continuous variables. p-Values were obtained by univariate analyses comparing linezolid group and vancomycin group. Categorical variables were analyzed by Fisher's exact test. Continuous variables were analyzed by Mann–Whitney U test.
Although the major adverse effect of LZD is thrombocytopenia, reports on the occurrence of hyponatremia have increased recently. However, infection per se is also a potential cause of hyponatremia. To verify that hyponatremia is associated with LZD use rather than infection, we compared the incidence and risk of developing hyponatremia in patients administered LZD with patients treated with VCM as control, because VCM is not known to primarily cause hyponatremia. Our analysis suggested a significantly higher incidence and risk of developing hyponatremia in LZD group than in VCM group both before and after propensity score matching.
LZD exhibits superior tissue penetration into bone, skin blister fluid, muscle, fat, alveolar cells, lung extracellular lining fluid, and cerebral spinal fluid.2) A randomized controlled trial using modified intent-to-treat analysis demonstrated higher therapeutic efficacy of LZD compared to VCM for complicated skin and soft tissue infections by MRSA, including cellulitis, abscess, and burn.27) A meta-analysis synthesizing nine randomized controlled trials suggests that LZD exerts significantly superior clinical and bacteriological efficacy for MRSA-induced complicated skin and soft tissue infections compared to VCM.28) This would account for the higher frequency of using LZD than VCM for wound infection, cellulitis, or abscess (p < 0.001) in the present study. Clinical practice guidelines prepared by the Infectious Diseases Society of America for the treatment of MRSA infections recommend VCM as the first choice for bacteremia,29) which may explain the higher frequencies of VCM use for catheter-related infection (p = 0.004) and febrile neutropenia (p = 0.001) than LZD in the present study. LZD exhibits excellent distribution to lung extracellular lining fluid. A randomized controlled trial showed superiority of LZD to VCM in clinical and bacteriologic efficacy for MRSA pneumonia.30) However, in our hospital, LZD tends to be reserved for severe pneumonia or for switch from VCM for pneumonia not responding to VCM. As a result, LZD was less frequently used than VCM for pneumonia in the present study.
When demographic and laboratory parameters were compared between the two groups, VCM group showed lower serum albumin level (p < 0.001), higher CRP level (p = 0.046), and higher frequency for treating pneumonia (p = 0.007). Oncotic pressure generally decreases in hypoalbuminemia, reducing the effective plasma volume in blood vessel. Pressoreceptors sense the decrease in pressure and activate the antidiuretic hormone secretion and renin-angiotensin-aldosterone systems as compensatory response, allowing water reabsorption by the kidney tubules. Hence, patients with hypoalbuminemia tend to have lower serum sodium levels. Moreover, IL-6 levels are elevated in an inflammatory state accompanied by high CRP levels. As noted in Introduction, IL-6 enhances the secretion of vasopressin under physiological conditions,21) which increases the risk of hyponatremia development. Further, the incidence of hyponatremia increases when patients have viral and bacterial pneumonia.31) Based on these findings, there is a possibility that patients receiving VCM in the present study had underlying conditions that made them more susceptible to develop hyponatremia compared with patients treated with LZD. Nevertheless, LZD group showed a significantly higher incidence and risk of developing hyponatremia both before and after propensity score matching. This finding suggests that hyponatremia that occurs while using LZD may be an adverse effect associated with LZD regardless of patient background. However, the bias remains after adjusting the difference in the background with propensity score matching since the infection severity was not incorporated into the matching variables. Therefore, we caution that the results should be carefully interpreted.
When acute kidney injury is caused by renal tubular epithelial disorder or tubular necrosis, reuptake of sodium from urine is suppressed, resulting in decrease in serum sodium level. VCM frequently causes acute kidney injury via direct cytotoxicity to renal tubular epithelial cells.26) Thus, hyponatremia in VCM group is possibly due to salt-losing nephritis accompanying renal tubular necrosis. However, there was no significant difference in the incidence of acute kidney injury between LZD (12.5%) and VCM (36.8%) groups. This lack of statistical significance may be attributed to the small numbers of patients who developed hyponatremia in both groups (LZD: n = 16, VCM: n = 19), which is one of the limitations of our study.
This retrospective, single-center, observational cohort study has some limitations. First, as the number of patients in the LZD group (n = 97) was small, the number of subjects after propensity score matching was even smaller (n = 73). Second, although the dietary intake and infusion fluid compositions at baseline were incorporated as variables for propensity score matching, these were changed during LZD or VCM treatment in some patients in both groups, resulting in difference in sodium intake between LZD and VCM groups. Moreover, physicians usually adjust the type and amount of infusion fluids and oral electrolyte solutions, or add/discontinue/decrease/increase drugs that regulate electrolyte balance (such as diuretics) while monitoring the electrolyte levels of inpatients. Third, although two case reports13,14) and previous retrospective studies17,22) suggest the possibility of hyponatremia caused by SIADH, the mechanism of LZD-induced hyponatremia remains unclear due to the lack of basic research. Therefore, we cannot deny the possibility that biases related to the mechanism of hyponatremia development exist in both groups. Fourth, the severity of infection was assessed based on routine laboratory tests such as white blood cell count and CRP, although SOFA score and APACHE II score would be more desirable as indicators of severity of infection. However, this retrospective study excluded patients admitted to the intensive care unit or emergency center during LZD or VCM treatment. Hence, organ dysfunction severity, arterial blood gas, and Glasgow Coma Scale were not measured in some patients, and it was not possible to assess infection severity using SOFA score and APACHE II in all patients. Fifth, factors that potentially influence serum sodium level were subjectively selected (based on the literature as far as possible) as variables for calculating propensity scores. Since this selection method cannot include all the relevant factors in the model, some confounding factors would remain.
In conclusion, to our best knowledge, this is the first report comparing the incidence of hyponatremia between LZD and VCM. The results of propensity score analyses suggest that the risk of hyponatremia associated with LZD is approximately 3.7-fold higher than that associated with VCM, regardless of patient background characteristics. Regular monitoring of serum sodium level and infusing sodium chloride promptly when the level decreases are important when administering LZD.
The authors declare no conflict of interest.
