key: cord-0727823-j86uc8l6
authors: Amin, Alpesh; Moon, Rena; Agiro, Abiy; Rosenthal, Ning; Brown, Harold; Legg, Randall; Pottorf, William
title: In-hospital mortality, length of stay, and hospitalization cost of COVID-19 patients with and without hyperkalemia
date: 2022-04-29
journal: Am J Med Sci
DOI: 10.1016/j.amjms.2022.04.029
sha: 89fd59e7e1a37fa621914c7663206b1601505a78
doc_id: 727823
cord_uid: j86uc8l6

BACKGROUND: : Hyperkalemia (HK) may be associated with poor clinical outcomes among COVID-19 patients. This study aimed to describe the prevalence of HK and evaluate the associations between HK and in-hospital mortality, intensive care unit (ICU) admission, length of hospital stay (LOS), and hospitalization cost among COVID-19 inpatients. METHODS: : A retrospective cohort study was conducted using a large hospital discharge database (PINC AI Healthcare Database) for COVID-19 inpatients discharged between April 1 and August 31, 2020. HK was defined with discharge diagnosis and potassium binder use. RESULTS: : Of 192,182 COVID-19 inpatients, 12% (n=22,702) had HK. HK patients were more likely to be older (median age 67 vs 63 years), male (63% vs 50%), black (30% vs 22%), and have a history of chronic kidney disease (45% vs 16%) or diabetes mellitus (55% vs 35%) than non-HK patients (all p<.001). A significantly higher proportion of patients with HK had in-hospital mortality (42% vs 11%, p<.001) than those without HK; this was persistent after adjusting for confounders (adjusted odds ratio [(a)OR] 1.69, 95% confidence interval [CI]1.62-1.77). Patients with HK were also more likely to be admitted to ICU ((a)OR 1.05, 95% CI 1.01-1.09), incur higher cost of care (adjusted mean difference $5,389) and have longer LOS (adjusted mean difference 1.3 days) than non-HK patients. CONCLUSIONS: : Presence of HK wass independently associated with higher in-hospital mortality, LOS, and cost of care among COVID-19 inpatients. Detecting and closely monitoring HK are recommended to improve clinical outcomes and reduce LOS and healthcare cost among COVID-19 patients.

Newly emerged coronavirus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is responsible for almost one million deaths in the United States (US) alone. 1 Mortality rate was especially high among adults with advanced age and pre-existing conditions such as cardiovascular disease, diabetes mellitus (DM), hypertension, chronic lung disease, cancer, chronic kidney disease (CKD), and obesity. [2] [3] [4] [5] Renal complications of the disease were also closely associated with the risks of severe illness and death among COVID-19 inpatients. [6] [7] [8] [9] In a systemic review, acute kidney injury (AKI) was present in 17% of the hospitalized COVID-19 patients, of which 77% had a severe illness and 52% died. 9 For patients with AKI, the primary indication of renal replacement therapy was hyperkalemiahigh serum potassium level (>5.0 mmol/L or >5.5mmol/L). [10] [11] [12] [13] [14] Not surprisingly, hyperkalemia was one of the most frequent electrolyte imbalances with an incidence of 12.5% among COVID-19 inpatients. 15 In a large cohort study of US patients with COVID-19, hyperkalemia was more frequently observed among the deceased compared to recovered patients. 16, 17 In another recent study, serum potassium level of ≥5.0 mmol/L was associated with significantly increased 30-day mortality among COVID-19 patients independent of age, sex, history of CKD, pulse oxygen saturation, and serum creatinine. 18 COVID-19 inpatients are already at an increased risk of developing hyperkalemia, due to the high prevalence of congestive heart failure (CHF), DM, and CKD. 17 Furthermore, hypertension is the most common comorbidity among COVID-19 inpatients and use of renin-angiotensinaldosterone system (RAAS) inhibitors in these patients could lead to disturbed potassium homeostasis and elevated serum potassium levels. 9, 17, 19 However, data are limited on the independent risk conferred by hyperkalemia and how RAAS inhibitors are used among hospitalized COVID-19 patients. Furthermore, no study to date has assessed the impact of hyperkalemia on healthcare resource utilization (HRU) and cost among COVID-19 inpatients. Using the largest hospital discharge database in the US, the aims of this study were: 1) to describe the prevalence of hyperkalemia, 2) to evaluate the association between hyperkalemia and in-hospital mortality, and 3) to assess the impact of hyperkalemia on HRU and cost among COVID-19 patients.

A retrospective cohort study was performed using PINC AI Healthcare Database (PHD, formerly known as Premier Healthcare Database). The PHD is a hospital-based, service-level, all-payer discharge database for geographically diverse inpatient and outpatient visits. Inpatient discharges in PHD represents approximately 20-25% of all inpatient admissions in the US since 2000. 17, 20 PHD has been used by the National Institute of Health and the Centers for Disease Prevention and Control for evaluating the impact of COVID-19 on patients across the US. 21, 22 The standard hospital discharge files included demographic characteristics, disease states, and time-stamped log of billed items (e.g., procedures, medications, laboratory services, and diagnostic services) of patient visits and geographic location, urbanicity of served population, teaching status, and bed capacity of the hospitals. 20 All data were statistically deidentified and compliant with the Health Insurance Portability and Accountability Act. The study was exempted from institutional board review based on US Title 45 Code of Federal Regulations, Part 46. We did not pursue informed consent from the study participants because individuals could not be identified directly or through linked identifiers. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) 23 reporting guideline.

All COVID-19-related inpatient visits were identified using the principal or secondary discharge diagnosis of COVID-19 (International Classification of Diseases, 10 th Revision, Clinical

Modification [ICD-10-CM] diagnosis code U07.1) between April 1 and August 31, 2020. 24 If the patient had multiple inpatient visits with the discharge diagnosis of COVID-19 during the study period, only the first inpatient visit (index hospitalization) was included in the analysis to represent patient-level data. Patients were only included in the study if they were 18 years old or older upon admission and did not acquire COVID-19 at the hospital (presence on admission is not equal to 'No').

Hyperkalemia was identified using the principal or secondary discharge ICD-10-CM diagnosis code for hyperkalemia (E87.5) or hospital chargemaster descriptions of potassium binder use (sodium zirconium cyclosilicate, sodium polystyrene sulfonate, or patiromer).

Main outcomes of interest were in-hospital mortality, intensive care unit (ICU) admission, hospital length of stay (LOS), and hospitalization cost during index hospitalization. Secondary outcomes included acute complications of COVID-19 during index hospitalization. In-hospital mortality was identified when the patient's discharge status was 'expired'; ICU admission was identified when the patient incurred any room and board charges related to ICU (observations after surgery and step-down ICU were not included). Total cost for the index hospitalization included the sum of all costs incurred by the hospital including room and board, pharmacy, laboratory, imaging, and central supply.

For sensitivity analysis, we examined the composite measure of in-hospital mortality + referral to hospice at the time of hospitalization. Furthermore, we examined the association between inhospital mortality and hyperkalemia in a subgroup of patients who developed AKI during their inpatient stay.

Baseline patient characteristics including age, sex, patients' self-reported race and ethnicity, and hospital characteristics including geographical region (i.e., Midwest, Northeast, South, or West), hospital size (i.e., number of beds), urbanicity of served population (rural vs. urban) and teaching status were provided by the participating hospitals. For baseline comorbidities, Charlson-Deyo comorbidities were identified using ICD-10-CM diagnosis codes (eTable 1 in Supplement) and

Charlson Comorbidity Index (CCI) score was calculated using a previously validated method. 25, 26 In addition to Charlson-Deyo comorbidities, morbid obesity was identified using COVID-19-related acute complications assessed included acute ischemic heart disease, AKI, acute liver injury, acute respiratory failure, acute respiratory distress syndrome (ARDS), hypokalemia, hyponatremia, metabolic or respiratory acidosis, epileptic seizures, rhabdomyolysis, sepsis, shock, and venous thromboembolism using ICD-10-CM diagnosis codes (eTable 2 in Supplement).

Descriptive statistics were used to compare baseline patient and hospital characteristics of COVID-19 patients with and without hyperkalemia. Continuous variables were reported as mean (standard deviation) or median (1 st quartile, 3 rd quartile) and categorical variables were reported as counts and percentages. For statistical difference, we used pooled t-test or Mann-Whitney test for continuous variables and χ 2 test for categorical variables.

For in-hospital mortality and ICU admission, we first examined the association between hyperkalemia and mortality using a multivariable logistic regression model, adjusting for sex, age group, race, and ethnicity (data not shown, available upon request). We then adjusted for patient characteristics (sex, age group, race, ethnicity, comorbidities), treatment medications and supplements, and complications. A priori covariates were factors that varied by greater than 10% across exposure groups at baseline were all considered: sex, age group, race, ethnicity, CCI category, APR-DRG category, AKI, in-hospital dialysis, ARDS, hyponatremia, acidosis, sepsis, shock, medication use (albumin, antiarrhythmic, beta blocker, blood growth factor, bronchodilator, calcium channel blocker, corticosteroid, antibiotics other than azithromycin, statin, and intravenous administration of insulin, diuretics, bicarbonates, and calcium gluconates). Final covariates were selected using backward elimination method, with significance level of p < .05 to stay in the model (ethnicity, corticosteroids, and sepsis were eliminated). AKI and CHF did not significantly modify the association between hyperkalemia and in-hospital mortality (interaction terms). Multicollinearity between covariates in the final model was tested using variance inflation factor.

For LOS and hospitalization cost, we used multivariable generalized linear regression models with gamma (for cost) and Poisson (for LOS) variances and log-link functions using the same covariates described above. Mean values for both variables were estimated using recycled prediction methods and bootstrapping for 95% confidence interval (95% CI) estimations.

All analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, North Carolina), and the figure was generated using R version 3.6.3 (R Foundation for Statistical Computing, Vienna, Austria) with forestplot package. 9.0%) were slightly lower in the hyperkalemia group compared to non-hyperkalemia group (all p < .01), but the difference was small. Intravenous administration of insulin, diuretics, bicarbonates, and calcium gluconates were significantly higher among hyperkalemic COVID-19 patients compared to patients without hyperkalemia (85.1% vs. 50.2%, p < .001).

In-hospital mortality was significantly higher among COVID-19 patients experiencing hyperkalemia (42.3%) compared to patients without hyperkalemia (11.1%, p < .001) ( Table 3) .

The association between hyperkalemia and in-hospital mortality attenuated after adjusting for significant predictors of mortality including age, sex, race, ethnicity, CCI score, APR-DRG severity of illness, treatment medications and complications. However, hyperkalemia was still associated with significantly increased odds of in-hospital mortality by 69% among COVID-19 patients ( a OR 1.69, 95% CI 1.62−1.77) ( Table 3 , Fig 1) .

Other factors associated with increased in-hospital mortality among COVID-19 patients were male sex, older age, 1 or greater CCI score, extreme APR-DRG severity of illness, intravenous administration of insulin, diuretics, bicarbonates, or calcium gluconates, dialysis, AKI, ARDS, hyponatremia, metabolic or respiratory acidosis, shock, albumin use, antiarrhythmic use, bronchodilator use, and antibiotics other than azithromycin use (eTable 3 in Supplement).

Using the composite of 'in-hospital mortality + discharged to hospice' as the outcome variable did not change the association between hyperkalemia and mortality. Hyperkalemia was still associated with significantly increased odds of in-hospital mortality + discharged to hospice by 61% among COVID-19 patients ( a OR 1.61, 95% CI 1.54-1.68)( Table 3 ).

In a subgroup of COVID-19 patients with AKI, those who developed hyperkalemia were 84% more likely to die during hospitalization ( a OR 1.84, 95% CI 1.75-1.93)(eTable 4 in Supplement).

Significantly higher proportion of patients in the hyperkalemia group were admitted to the ICU (50.1% vs. 19.5%) than patients without hyperkalemia (Table 3) . Median hospital LOS was significantly longer among patients who developed hyperkalemia compared to those who didn't (10 days vs. 5 days). Total median cost of hospitalization was also significantly higher for hyperkalemic patients ($34,233 vs. $11,168) than non-hyperkalemia patients. The differences attenuated after adjusting for age, sex, race, CCI score, APR-DRG severity of illness, and treatment medications and acute complications, but patients with hyperkalemia were still more likely to be admitted to the ICU ( a OR 1.05, 95% CI 1.01-1.07), incur higher hospitalization cost (adjusted mean difference $5,389, 95% CI $5,367-$5,410), and have longer LOS (adjusted mean difference 1.26 days, 95% CI 1.26-1.26).

This study shows that hyperkalemia was relatively common among hospitalized COVID-19 patients (~12%). More importantly, hyperkalemia was associated with significantly increased likelihood of death as well as increased cost and LOS during hospitalization. While hyperkalemia was more common among patients experiencing acute complications of COVID-19 such as AKI, ARDS, and shock, this study showed that increased odds of in-hospital mortality was independent of these complications.

The prevalence of hyperkalemia among COVID-19 inpatients may be explained by two main reasons. First, risk factors for developing severe illness of COVID-19 were closely related to the patient's pre-existing conditions that also put COVID-19 inpatients at a higher risk of experiencing hyperkalemia. Using the same database (but a different cohort), Rosenthal et al 17 .

reported that among COVID-19 inpatients, 15.0% had CHF, 40.5% had DM, and 22.9% had CKD while only 2.5% had CHF, 12.9% had DM, and 3.3% had CKD among COVID-19 outpatients during early months of the pandemic. We observed a similar distribution of comorbidities among COVID-19 inpatients. Regardless of COVID-19 status, patients with CHF and CKD were already at a higher risk of impaired urinary potassium excretion and were likely to develop hyperkalemia. 34, 35 For patients with DM, uncontrolled hyperglycemia was one of the most common reasons for redistributive hyperkalemia among non-COVID-19 patients. 35 Second, hyperkalemia was closely related to AKI, one of the most prevalent and serious renal manifestations of COVID-19. Fisher et al. 36 reported that AKI incidence was higher among COVID-19 inpatients (56.9%) compared to non-COVID-19 inpatients (25.1%). Furthermore, compared with non-COVID-19 ICU patients, critically ill COVID-19 patients had a higher incidence of AKI. 37 inpatients and in 75% of those admitted to ICU. 39 In our study cohort, 31.2% of COVID-19 inpatients developed AKI and the proportion was much higher (70.1%) among those with hyperkalemia. In addition, hyperkalemia was also associated with higher likelihoods of other acute complications of COVID-19 (i.e., sepsis, shock, and ARDS).

After accounting for the severity of COVID-19 illness, hyperkalemia-related pre-existing conditions and acute complications, hyperkalemia was still associated with 69% increased odds of in-hospital mortality ( a OR 1.69). Liu et al. 18 also demonstrated that increased serum potassium level was a predictor of mortality among COVID-19 patients independent of other risk factors.

Furthermore, we showed that among hospitalized patients with COVID-19 and AKI, patients with hyperkalemia were 71% more likely to die than those without hyperkalemia. Our findings highlight the importance of monitoring potassium level among COVID-19 inpatients with or without the presence of previously reported conditions and acute complications.

We also demonstrated the negative impact of hyperkalemia on HRU and cost. Betts et al. 40 reported that hyperkalemia inpatients had 1.51 days longer hospital stay than patients without hyperkalemia in the pre-COVID-19 era. Our results showed a similar difference among COVID-19 inpatients, as hyperkalemia patients had an adjusted mean of 1.26 days longer LOS than those without. Betts et al. 40 also showed that hyperkalemia patients incurred $4,128 higher 30-day total healthcare costs than non-hyperkalemia patients. In this study, we showed that a single hospitalization for COVID-19 patients with hyperkalemia can incur $5,389 additional hospitalization cost compared to COVID-19 patients without hyperkalemia.

This study has several limitations. First, this was a secondary data analysis using hospital administrative database. Hyperkalemia status was captured by ICD-10-CM diagnosis code and potassium binder use instead of by laboratory values, which is more likely to capture more prominent hyperkalemia cases. Mild hyperkalemia patients might have been misclassified as non-hyperkalemia patients and bias the results towards the null. We did not differentiate CKD and AKI stages nor specify the dosages for RAAS inhibitors. In addition, other clinical conditions and medications were also identified using hospital-reported diagnosis and procedure codes and chargemaster descriptions; misclassification bias is possible. However, the potential misclassification is expected to be non-differential between hyperkalemia and non-hyperkalemia patients and may bias the association towards the null. Second, due to the nature of observational studies, we were not able to draw any causal association between hyperkalemia and in-hospital mortality due to unmeasured confounding variables. Although we know that all other conditions including acute complications preceded death, we were not able to identify the temporal association between hyperkalemia, ICU admission, and acute complications. Third, the race and ethnicity variables were self-reported by patients at the time of hospitalization and 19% of patients had unknown ethnicity and 25% of patients had other or unknown race. This could have resulted in underestimating the actual percentage of Hispanic or black patients with COVID-19.

Despite these limitations, this study described the association between hyperkalemia and inhospital mortality, ICU admission, LOS, and hospitalization cost using a large COVID-19 cohort in the US. Hyperkalemia was a relatively common electrolyte imbalance among hospitalized COVID-19 patients; patients were more likely to die and get admitted to the ICU during hospitalization, be hospitalized longer, and incur higher hospitalization cost when they experienced hyperkalemia. The associations were significant after accounting for the severity of illness and hyperkalemia-related conditions and acute complications such as AKI. Our findings warrant a close monitoring of potassium levels among hospitalized COVID-19 patients. Future randomized trials are needed to confirm and better understand the impact of hyperkalemia on mortality and HRU among COVID-19 patients and to identify effective methods to predict and control hyperkalemia.

All authors conceived and designed the study; R.M., N.R., and H.B. acquired, analysed, and interpreted the data; R.M. and H.B. performed statistical analysis; R.M., N.R., and A. Agiro drafted the paper; R.M., A. Agiro, N.R., A. Amin, R.L., and W.P. critically revised the paper; N.R., A. Agiro, and A. Amin provided administrative, technical, and material support; N.R. and A. Agiro supervised the study. 

This study was funded by AstraZeneca.

The funder had a role in the design and conduct of the study; management, analysis, and interpretation of the data; preparation, review, and approval of the manuscript as well as decision to submit the manuscript for publication; however, the funder had no role in collection of the data.

Deyo 

Adjusted Odds Ratio 1.69 1.00 -

COVID-19 Dashboard by the

Factors associated with COVID-19-related death using OpenSAFELY

Characteristics of and Important Lessons From the Coronavirus Disease 2019 (COVID-19) Outbreak in China: Summary of a Report of 72314 Cases From the Chinese Center for Disease Control and Prevention

Factors associated with hospital admission and critical illness among 5279 people with coronavirus disease 2019 in New York City: prospective cohort study

Preliminary Estimates of the Prevalence of Selected Underlying Health Conditions Among Patients with Coronavirus Disease 2019 -United States

Acute Kidney Injury in a National Cohort of Hospitalized US Veterans with COVID-19

Kidney disease is associated with in-hospital death of patients with COVID-19

The Incidence, Risk Factors, and Prognosis of Acute Kidney Injury in Adult Patients with Coronavirus Disease

Outcomes for Patients With

Acute Kidney Injury: A Systematic Review and Meta-Analysis

Treatment of electrolyte disorders in adult patients in the intensive care unit

Hyperkalemia: recognition and management of a critical electrolyte disturbance

Association between Serum Potassium and Outcomes in Patients with Reduced Kidney Function

Oral potassium binders: increasing flexibility in times of crisis

Renal complications in COVID-19: a systematic review and meta-analysis

Clinical characteristics of 113 deceased patients with coronavirus disease 2019: retrospective study

Risk Factors Associated With In-Hospital Mortality in a US National Sample of Patients With COVID-19

Association Between Average Plasma Potassium Levels and 30-day Mortality During Hospitalization in Patients with COVID-19 in Wuhan, China

Potassium Homeostasis and Renin-Angiotensin-Aldosterone System Inhibitors

Premier Healthcare Database (COVID-19): Data that informs and performs

Patients Discharged and Experiencing Same-Hospital Readmission -United States

Premier Healthcare Database Being Used by National Institutes of Health to Evaluate Impact of COVID-19 on Patients Across the

being-used-by-national-institutes-of-health-to-evaluate-impact-of-covid-19-on-patientsacross-the-u-s

The Strengthening the Reporting of Patients With SARS-CoV-2 Infection: The COVID A to Z Randomized Clinical Trial

Convalescent plasma treatment is associated with lower mortality and better outcomes in high risk COVID-19 patientspropensity score matched case-control study

Controversies in Management of Hyperkalemia

Long-Term Mortality and Trajectory of Potassium Measurements Following an Episode of Acute Severe Hyperkalemia

AKI in Hospitalized Patients with and without COVID-19: A Comparison Study

Epidemiology of acute kidney injury in intensive care units in Beijing: the multi-center BAKIT study

Severe Acute Kidney Injury in Patients with COVID-19 and Acute Respiratory Distress Syndrome

AKI in Hospitalized Patients with COVID-19

The Cost of Hyperkalemia in the United States

*All adjusted models were adjusted for sex, age category, race, CCI score category, APR category, dialysis, AKI, ARDS, hyponatremia, acidosis, shock, and medication use (albumin, antiarrhythmics, beta blocker, blood growth factor, bronchodilator, calcium channel blocker, antibiotics other than azithromycin, statin, and IV non-binder treatment for HK).*Cost model was adjusted using GLM regression with gamma variance and log-link function.*LOS model was adjusted using GLM regression with Poisson variance and log-link function.*Both cost and LOS models were estimated using recycled prediction method and bootstrapping for the 95% CI.