Evaluation of medication regimen complexity as a predictor for mortality

Abstract

While medication regimen complexity, as measured by a novel medication regimen complexity-intensive care unit (MRC-ICU) score, correlates with baseline severity of illness and mortality, whether the MRC-ICU improves hospital mortality prediction is not known. After characterizing the association between MRC-ICU, severity of illness and hospital mortality we sought to evaluate the incremental benefit of adding MRC-ICU to illness severity-based hospital mortality prediction models. This was a single-center, observational cohort study of adult intensive care units (ICUs). A random sample of 991 adults admitted ≥ 24 h to the ICU from 10/2015 to 10/2020 were included. The logistic regression models for the primary outcome of mortality were assessed via area under the receiver operating characteristic (AUROC). Medication regimen complexity was evaluated daily using the MRC-ICU. This previously validated index is a weighted summation of medications prescribed in the first 24 h of ICU stay [e.g., a patient prescribed insulin (1 point) and vancomycin (3 points) has a MRC-ICU = 4 points]. Baseline demographic features (e.g., age, sex, ICU type) were collected and severity of illness (based on worst values within the first 24 h of ICU admission) was characterized using both the Acute Physiology and Chronic Health Evaluation (APACHE II) and the Sequential Organ Failure Assessment (SOFA) score. Univariate analysis of 991 patients revealed every one-point increase in the average 24-h MRC-ICU score was associated with a 5% increase in hospital mortality [Odds Ratio (OR) 1.05, 95% confidence interval 1.02–1.08, p = 0.002]. The model including MRC-ICU, APACHE II and SOFA had a AUROC for mortality of 0.81 whereas the model including only APACHE-II and SOFA had a AUROC for mortality of 0.76. Medication regimen complexity is associated with increased hospital mortality. A prediction model including medication regimen complexity only modestly improves hospital mortality prediction.

Full text

1
Vol.:(0123456789)
Scientic Reports | (2023) 13:10784 | https://doi.org/10.1038/s41598-023-37908-1
www.nature.com/scientificreports
Evaluation of medication regimen
complexity as a predictor
for mortality
Andrea Sikora
1*, John W. Devlin
2,3, Mengyun Yu
4, Tianyi Zhang
4, Xianyan Chen
4,
Susan E. Smith
1, Brian Murray
5, Mitchell S. Buckley
6, Sandra Rowe
7 & David J. Murphy
8
While medication regimen complexity, as measured by a novel medication regimen complexity-
intensive care unit (MRC-ICU) score, correlates with baseline severity of illness and mortality, whether
the MRC-ICU improves hospital mortality prediction is not known. After characterizing the association
between MRC-ICU, severity of illness and hospital mortality we sought to evaluate the incremental
benet of adding MRC-ICU to illness severity-based hospital mortality prediction models. This was
a single-center, observational cohort study of adult intensive care units (ICUs). A random sample of
991 adults admitted ≥ 24 h to the ICU from 10/2015 to 10/2020 were included. The logistic regression
models for the primary outcome of mortality were assessed via area under the receiver operating
characteristic (AUROC). Medication regimen complexity was evaluated daily using the MRC-ICU.
This previously validated index is a weighted summation of medications prescribed in the rst 24 h
of ICU stay [e.g., a patient prescribed insulin (1 point) and vancomycin (3 points) has a MRC-ICU = 4
points]. Baseline demographic features (e.g., age, sex, ICU type) were collected and severity of illness
(based on worst values within the rst 24 h of ICU admission) was characterized using both the Acute
Physiology and Chronic Health Evaluation (APACHE II) and the Sequential Organ Failure Assessment
(SOFA) score. Univariate analysis of 991 patients revealed every one-point increase in the average 24-h
MRC-ICU score was associated with a 5% increase in hospital mortality [Odds Ratio (OR) 1.05, 95%
condence interval 1.02–1.08, p = 0.002]. The model including MRC-ICU, APACHE II and SOFA had a
AUROC for mortality of 0.81 whereas the model including only APACHE-II and SOFA had a AUROC for
mortality of 0.76. Medication regimen complexity is associated with increased hospital mortality. A
prediction model including medication regimen complexity only modestly improves hospital mortality
prediction.
Robust mortality prediction models in the intensive care unit (ICU) facilitate clinical decision making, clinical
investigation, and quality improvement1,2. Severity of illness scores (Acute Physiology and Chronic Health Evalu-
ation [APACHE II], Sequential Organ Failure Assessment [SOFA]) have remained the standard for mortality
prediction but have potential limitations3,4. Medications are frequently prescribed in the ICU to improve patient
outcomes, but critically ill adults are also at high risk for experiencing adverse drug events, some of which are
associated with an increased risk for mortality5.
e medication regimen complexity-intensive care unit (MRC-ICU) score has been proposed to succinctly
characterize medication regimens in the ICU6,7. In small, preliminary studies the MRC-ICU demonstrated
correlation to illness severity scores (APACHE II and SOFA), hospital mortality, ICU complications (e.g., uid
overload, drug-drug interactions), and the number and intensity of medication interventions performed by
critical care pharmacists6–13. In a subsequent study of 3,908 ICU adults at 28 centers, every one-point increase
in the MRC-ICU score was associated with a 7% increase in the odds of mortality and a 0.25day increase in ICU
length of stay14; however, this study did not incorporate adjustments for severity of illness.
OPEN
1Department of Clinical and Administrative Pharmacy, University of Georgia College of Pharmacy, 1120 15th
Street, HM-118, Augusta, GA 30912, USA. 2Bouve College of Health Sciences, Northeastern University, Boston,
MA, USA. 3Division of Pulmonary and Critical Care Medicine, Brigham and Women’s Hospital, Boston, MA,
USA. 4Department of Statistics, University of Georgia Franklin College of Arts and Sciences, Athens, GA,
USA. 5Department of Pharmacy, University of North Carolina Medical Center, Chapel Hill, NC, USA. 6LaJolla
Pharmaceuticals, Waltham, USA. 7Oregon Health and Science University, Portland, OR, USA. 8Division of
Pulmonary, Allergy, Critical Care and Sleep Medicine, Emory University School of Medicine, Atlanta, GA,
USA. *email: sikora@uga.edu
Content courtesy of Springer Nature, terms of use apply. Rights reserved

2
Vol:.(1234567890)
Scientic Reports | (2023) 13:10784 | https://doi.org/10.1038/s41598-023-37908-1
www.nature.com/scientificreports/
Delineating how holistic medication use in the ICU relates to patient mortality in the context of illness sever-
ity is an important next step in optimizing pharmaceutical care during critical illness6. Aer characterizing the
association between MRC-ICU, severity of illness and hospital mortality, we sought to evaluate the incremental
benet of adding MRC-ICU to illness severity-based ICU mortality prediction models. We hypothesized that
incorporation of medication complexity into illness severity-based ICU mortality prediction models would
improve hospital mortality prediction.
Methods
Study population. is retrospective, observational study was reviewed by the University of Georgia
(UGA) Institutional Review Board (IRB) and deemed to be exempt from IRB oversight (Project00001541). All
methods were performed in accordance with the ethical standards of the of the UGA IRB and the Helsinki
Declaration of 1975. Patient data were obtained via the Carolina Data Warehouse, which houses Epic® electronic
health record (EHR) data from the University of North Carolina Health System (UNCHS), an integrated health-
care delivery system. Given the intensity of the data collection eort, particularly for MRC-ICU calculations, we
employed random number generation to identify a sample of 1000 adults aged ≥ 18years admitted to the ICU
for ≥ 24h between October 2015 and October 2020. Only data from the rst ICU admission for each patient
was included. Patients were excluded if they were placed on comfort care within the rst 24h of their ICU stay.
Variables. e primary outcome was hospital mortality. e EHR was queried to obtain baseline patient
characteristics, medication information, and patient outcomes. Baseline patient characteristics including age,
sex, race, admitting diagnosis and ICU type were collected. e admission APACHE-II and SOFA scores (based
on values from the rst 24h of ICU admission) were calculated when values for all domains were available15,16.
Medication information including drug, dose, route, duration, and timing of administration were recorded.
e MRC-ICU score was calculated at 24h and on each ICU day for up to one week. e MRC-ICU consists of 35
discrete (i.e., medication) items where each is assigned a weighted value and then summed to create a MRC-ICU
score for a patient’s medication regimen at any given time point9. For example, a patient receiving vancomycin
(3 points), a norepinephrine infusion (1 point), and topical chlorhexidine (1 point) on ICU day 2 would have
an ICU day 2 total MRC-ICU score of 5. In addition to the mean daily MRC-ICU score, the frequency of each
item contributing to the score was calculated.
Analysis. Descriptive statistics were computed for the relevant variables and a plot of the MRC-ICU score by
ICU day and mortality was made to visualize the relationship between mortality, ICU day and MRC-ICU score.
Additionally, a component analysis of MRC-ICU was performed to identify the frequency of medication use in
the cohort and how it related to hospital mortality.
Multiple logistic regression models with the inclusion of dierent combinations of MRC-ICU, SOFA,
APACHE II and their interactions were developed to evaluate the relationship of MRC-ICU and severity of ill-
ness scores on ICU day 1 and hospital mortality. e Variance Ination Factor (VIF) was calculated during this
model-building process to avoid potential concerns with multicollinearity and therefore ensure model reliability
and robustness. Odds ratios (OR) were reported alongside their corresponding 95% condence intervals (CI)
for the predictors of interest.
To further evaluate the models, the cohort (containing 521 patients with complete APACHE-II, SOFA, and
MRC-ICU data) was split into training and test sets, using a ratio of 4:1. To evaluate the predictive abilities of
each model on hospital mortality, area under the receiver operating characteristic curve (AUROC) was computed
in addition to sensitivity, specicity, negative predictive value (NPV), and positive predictive value (PPV) in the
test set. Results were subsequently compared between AUROCs using DeLong’s test where prediction thresh-
olds were chosen by maximizing Informedness, Matthew’s Correlation Coecient (MCC) and F1 scores in the
training set. A two-sided p-value less than 0.05 was used to determine statistical signicance for all variables.
All analyses were performed usingR(version 4.1.2).
Ethical approval. is was an observational study that was reviewed by the University of Georgia Institu-
tional Review Board (IRB) and determined to be exempt from IRB oversight (Project00001541).
Results
Of the 38,729 patients admitted to UNCHS during the evaluation period, 7515 (19.4%) met all study criteria.
A total of 1000 (13.3%) of these 7515 patients were randomly selected for the study cohort. Aer excluding an
additional 9 patients because the ICU admission did not represent their index ICU admission, a total of 991
patients were included in the nal cohort. ese patients were 61.2 (standard deviation [SD] 17.5) years old
and predominantly medical (40.7%), cardiac (30.8%), surgical (9.8%) and neurological (9.4%). e mean 24-h
APACHE II score (based on complete data from 963 patients) was 14.2 (6.3) and the mean 24-h SOFA score
(based on complete data from 533 patients) was 8 (6.4). e mean MRC-ICU score at 24h (based on the 991
patients) was 10.3 (7.7). e overall hospital mortality rate was 9.8%. Table1 provides a full summary of the study
cohort characteristics, and Fig.1 provides a plot of mortality in relation to MRC-ICU and severity of illness.
e frequency of each MRC-ICU score component is presented in Supplemental Digital Content (SDC)
Supplemental Table1. e association between each MRC-ICU component and hospital mortality is presented
in SDC Supplemental Table2. e change in MRC-ICU over each ICU day is presented in SDC Supplemental
Figure1, with the highest score observed on ICU day 1. Hospital mortality by MRC-ICU score (as well as baseline
APACHE-II and SOFA score) is presented in SDC Supplemental Figure2. e results of the multiple logistic
Content courtesy of Springer Nature, terms of use apply. Rights reserved

3
Vol.:(0123456789)
Scientic Reports | (2023) 13:10784 | https://doi.org/10.1038/s41598-023-37908-1
www.nature.com/scientificreports/
regression models with dierent combinations of MRC-ICU, APACHE-II, SOFA score and their interactions for
hospital mortality, are presented in SDC Supplemental Table3 and SDC Supplemental Figure3.
While the positive relationship between APACHE II or SOFA score and hospital mortality did not change aer
controlling for MRC-ICU, the relationship between MRC-ICU and hospital mortality changed aer controlling
for APACHE II or SOFA (Table2). While the univariate analysis revealed a one-point increase in the MRC-ICU
was associated with a 5% increase in the odds of hospital mortality, the multivariate analysis (that accounted for
Table 1. Study population characteristics. Data are presented as n (%) or mean (standard deviation (SD))
unless otherwise stated. AKI acute kidney injury; CRRT continuous renal replacement therapy; SOFA
sequential organ failure assessment, APACHE II Acute Physiology and Chronic Health Evaluation; ICU
intensive care unit.
Feature N = 991
Age 61.2 (17.5)
Female 428 (43.2)
Race
Caucasian 645 (65.1)
Black 235 (23.7)
Other 111 (11.2)
ICU type
Medical 404 (40.7)
Cardiac 305 (30.8)
Surgical 97 (9.8)
Neurosciences 93 (9.4)
Burn 70 (7.1)
Other 22 (2.2)
Admission diagnosis
Cardiovascular 244 (24.6)
Neurology 139 (14.0)
Pulmonary 124 (12.5)
Trauma 111 (11.2)
Infection including sepsis 97 (9.8)
Gastrointestinal 83 (8.4)
Neoplasm 62 (6.3)
Dermatology 59 (6.0)
Renal 47 (4.7)
Endocrine 24 (2.4)
Other
Use of mechanical ventilation 312 (31.5)
Use of vasopressors 287 (29.0)
Severity scores
APACHE II at 24h, mean (SD), n (%) 14.2 (6.3), 963 (97.2%)
SOFA at 24h, mean (SD), n (%) 8 (6.4), 533 (53.8%)
Patient outcomes
Mortality 97 (9.8)
ICU length of stay, days 5.1 (9.5)
Table 2. Univariate and multivariable analysis of MRC-ICU association with mortality. e multiple variable
model presented includes MRC-ICU, APACHE II, and SOFA scores at 24h. SOFA sequential organ failure
assessment, APACHE II Acute Physiology and Chronic Health Evaluation, ICU intensive care unit. p-values are
obtained from Wald tests for each variable using normal approximation.
Mortality
Univariate Multiple Variable
OR 95% CI p-value OR 95% CI p-value
MRC-ICU at 24h 1.05 1.02, 1.08 0.002 0.94 0.90, 0.98 0.007
SOFA at 24h 1.32 1.21, 1.44 < 0.001 1.16 1.04, 1.29 0.006
APACHE II at 24h 1.17 1.12, 1.23 < 0.001 1.19 1.12, 1.28 < 0.001
Content courtesy of Springer Nature, terms of use apply. Rights reserved

4
Vol:.(1234567890)
Scientic Reports | (2023) 13:10784 | https://doi.org/10.1038/s41598-023-37908-1
www.nature.com/scientificreports/
APACHE II and SOFA) found a 6% reduction in hospital mortality for each one-point MRC-ICU increase (OR
0.94, 95% CI 0.90–0.98, p = 0.007).
e VIF scores for APACHE-II, SOFA, and MRC-ICU were found to be below 5 (2.1, 1.4 and 1.9 respectively)
indicating low correlation between these variables and thus an absence of signicant multicollinearity within
the models. Comparative performance of the APACHE-II, SOFA, and MRC-ICU ICU mortality models is pre-
sented in Table3. AUROC curves were plotted (Fig.2) for the six models comprising combinations of severity
of illness scores and medication data. e addition of medication variables resulted in an AUROC change from
0.76 to 0.81. Negative predictive value exceeded 90% for all mortality models while positive predictive value
remained low. Sensitivity, specicity, positive predictive value, and negative predictive values are reported in
SDC Supplemental Table4.
e results of the DeLong testing (SDC Supplemental Table5) for the training and test sets revealed MRC-ICU
inclusion improved both the APACHE-II (p = 0.014) and combined APACHE-II and SOFA (p = 0.007) mortality
prediction models. e results for the threshold analyses (SDC Supplemental Table6) revealed the APACHE-II,
SOFA, and MRC-ICU ICU mortality model to be the most robust across various threshold selection strategies
and that it strikes the optimal balance between sensitivity and specicity (SDC Supplemental Figure4). e
calibration plot (SDC Supplemental Figure5) revealed strong model predictiveness when the predicted score
was < 0.25. However, the model tends to under-predict hospital mortality when the predicted score was between
0.25 and 0.50 and tends to over-predict mortality when the predicted score was > 0.5.
Figure1. Hospital mortality in relation to MRC-ICU and severity of illness. In the le panel, the blue line
indicates the tted regression line of MRC-ICU versus APACHE II score (i.e., the typical MRC-ICU score of a
patient with a certain level of APACHE II score). In the right panel, the blue line indicates the tted regression
line of MRC-ICU versus SOFA score. Colors are set to be 50% transparent, indicating that darker colors have
more overlap of patients. All deaths occurred in patients with APACHE II scores over 10, and a possibility exists
that those patients with lower MRC-ICU scores had a higher mortality than expected given their APACHE II
score.
Table 3. Comparison of predictive accuracy for logistic regression models of hospital mortality. AUROC area
under the receiver operating characteristic curve, SOFA sequential organ failure assessment, APACHE II Acute
Physiology and Chronic Health Evaluation, ICU intensive care unit. e multiple variable model includes
MRC-ICU, APACHE II, and SOFA scores at 24h. *Number of bins for Hosmer–Lemeshow chi square was 5;
APACHE II and the multiple variable model had models with p > 0.05 indicating goodness of t.
Model
Univariate Multiple variable
APACHE II SOFA MRC-ICU Model
AUROC 0.73 0.74 0.47 0.81
Brier Score 0.096 0.090 0.10 0.09
Hosmer–Lemeshow chi square 8.54* 10.51 10.57 0.66*
Content courtesy of Springer Nature, terms of use apply. Rights reserved

5
Vol.:(0123456789)
Scientic Reports | (2023) 13:10784 | https://doi.org/10.1038/s41598-023-37908-1
www.nature.com/scientificreports/
Discussion
Aer establishing MRC-ICU, APACHE II and SOFA each have a direct association with increased hospital mor-
tality, we found that a severity of illness-based prediction model that includes medication regimen complexity
only modestly improved hospital mortality prediction, with this combination resulting in an AUROC thresh-
old exceeding 0.8. While the MRC-ICU continued to show association with hospital mortality, the inclusion
of severity of illness resulted in a switch from an increased odds for mortality with increased MRC-ICU to a
reduced odds for mortality with increased MRC-ICU. ese ndings represent the rst characterization of the
relationship between medication regimen complexity to mortality in the context of severity of illness predictors.
ere are a number of possible explanations as to why the addition of the MRC-ICU to APACHE II and
SOFA-based model only incrementally improved our ability to predict hospital mortality. Both APACHE II and
SOFA are good at predicting mortality. Mortality in the ICU results from many factors that evolve over the course
of admission; incorporating medication complexity represents only one factor that may change in a critically ill
adult’s ICU trajectory. e MRC-ICU does not represent the ‘ground truth’ for appropriate ICU pharmacologic
intervention; we do not know if the medication(s) administered were right or wrong for the patient.
Calculating baseline mortality risk is essential for clinical trials and quality improvement evaluation because
it facilitates patient enrollment, evaluation of treatment groups, and comparison of cohorts at dierent time
points1. Moreover, it allows for meaningful benchmarking among and within institutions looking to improve
quality of care1,2. Historically, this mortality risk was inferred through the objective quantication of severity
of illness using gold standards like APACHE-II and SOFA scores derived from regression models. Although
still widely accepted and used given their relative simplicity of calculation, these approaches assume a degree of
linearity that may not be present in the patient with a uctuating degree of critical illness and can suer from
missing data17,18. Technological advancements may support incremental improvement of these metrics, particu-
larly through modeling that allows for additional information to be included and that allows for such non-linear
relationships. us, adding medication data, which certainly inuences patient outcomes, has the potential to
modestly improve prediction.
While prior MRC-ICU evaluations have shown a direct relationship between medication regimen complexity
and increased mortality, these studies did not account for baseline severity of illness6–13. Aer accounting for
severity of illness, medication regimen complexity was associated with decreased mortality. In our cohort, death
occurred in patients with the higher APACHE II scores, and mortality appeared to sharply increase in patients
with an MRC-ICU score over 10. A possibility exists that an “ideal” level of medication complexity to reduce
hospital mortality is present for a given severity of illness.
Our results have implications for clinical decision-making, healthcare workload designations, and future
prediction-based research. e inter-related nature of a patient’s baseline severity of illness and their treat-
ment requires further exploration: the possibility exists that the addition of medication variables to mortality
prediction models may yield particular relevance in patients whose severity of illness is not so extreme but in a
middle ground where medication therapy is most likely to play the most impactful role on outcomes. Although
medications are treatment for critical illness, they are also independent risk factors for ICU complications that
can adversely aect ICU outcomes. Even then, incremental benets between ‘a medication’ for a disease and
‘the optimal medication’ for that disease are notable. As such, medications warrant further investigation for use
as important predictor variables for ICU outcomes such as mortality, length of stay, and duration of mechanical
ventilation. Moreover, our results suggest that the non-linear interaction between illness severity and MRC-ICU
as it relates to outcomes is an important consideration that warrants further investigation. Notably, MRC-ICU
may be unique because it is solely based on medications being ordered thus precluding the potential for missing-
ness. us, the possibility exists that including medications may improve outcome prediction models.
Limitations of our study include a sample size that may be too small to capture all medication-related per-
mutations of heterogenous ICU patients. Additionally, the single center design (and resultant lack of validation
test set) limits denitive conclusions. Severity of illness was estimated only at baseline; consideration of daily
ICU SOFA and MRC-ICU scores in a time-dependent hospital mortality model might lead to dierent results.
Although it was expected that MRC-ICU and severity of illness would correlate given both previous studies
and the general construct (i.e., increasingly complex medication regimens are required for sicker patients) and
Figure2. AUROCs for hospital mortality prediction.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

6
Vol:.(1234567890)
Scientic Reports | (2023) 13:10784 | https://doi.org/10.1038/s41598-023-37908-1
www.nature.com/scientificreports/
that incremental value of the addition of medication variables to regression models would be observed, these
patterns may be better elucidated by methods other than traditional regression, including articial intelligence
methodologies that are better equipped to handle non-monotone and non-linear relationships and may pro-
vide superior calibration2. As such, important next steps from this rst evaluation of how medication regimen
complexity relates to mortality include further evaluation of temporal relationships of MRC-ICU and severity
of illness, particularly in the context of machine learning and articial intelligence methodologies.
Conclusion
In this study, MRC-ICU was shown to have a relationship with mortality as well as severity of illness. Moreover,
severity of illness and medication regimen complexity play an important role in mortality prediction. Further
analysis of the complex interplay of these variables is warranted.
Data availability
Following request and authorship team approval of the appropriateness of the request, datasets can be made
available.
Received: 26 April 2023; Accepted: 29 June 2023
References
1. Vincent, J. L., Ferreira, F. & Moreno, R. Scoring systems for assessing organ dysfunction and survival. Crit. Care Clin. 16(2), 353–366
(2000).
2. Cosgri, C. V. et al. Developing well-calibrated illness severity scores for decision support in the critically ill. NPJ. Digit. Med. 2,
76 (2019).
3. Keegan, M. T., Gajic, O. & Afessa, B. Severity of illness scoring systems in the intensive care unit. Crit. Care Med. 39(1), 163–169
(2011).
4. Kramer, A. A., Zimmerman, J. E. & Knaus, W. A. Severity of illness and predictive models in society of critical care medicine’s rst
50 years: A tale of concord and conict. Crit. Care Med. 49(5), 728–740 (2021).
5. Cullen, D. J. et al. Preventable adverse drug events in hospitalized patients: A comparative study of intensive care and general care
units. Crit. Care Med. 25(8), 1289–1297 (1997).
6. Newsome, A. S. et al. Optimization of critical care pharmacy clinical services: A gap analysis approach. Am. J. Health Syst. Pharm.
78(22), 2077–2085 (2021).
7. Smith, S. E., Shelley, R. & Newsome, A. S. Medication regimen complexity vs patient acuity for predicting critical care pharmacist
interventions. Am. J. Health Syst. Pharm. 79, 651 (2021).
8. Webb, A., Rowe, S., & Newsome, A. S. Automated MRC-ICU calculations in the electronic medical record of an academic medical
center: Applications and considerations for critical care pharmacist practice. Am. J. Health Syst. Pharm. (2021) [Under review].
9. Gwynn, M. E. et al. Development and validation of a medication regimen complexity scoring tool for critically ill patients. Am. J.
Health Syst. Pharm. 76(Supplement_2), S34-s40 (2019).
10. Al-Mamun, M. A., Brothers, T. & Newsome, A. S. Development of machine learning models to validate a medication regimen
complexity scoring tool for critically ill patients. Ann. Pharmacother. 55(4), 421–429 (2021).
11. Olney, W. J. et al. Medication regimen complexity score as an indicator of uid balance in critically ill patients. J. Pharm. Pract.
35, 573 (2021).
12. Webb, A. J., Rowe, S. & Sikora Newsome, A. A descriptive report of the rapid implementation of automated MRC-ICU calculations
in the EMR of an academic medical center. Am. J. Health Syst. Pharm. 79, 979 (2022).
13. Sikora Newsome, A. et al. Medication regimen complexity is associated with pharmacist interventions and drug–drug interactions:
A use of the novel MRC-ICU scoring tool. J. Am. Coll. Clin. Pharm. 3(1), 47–56 (2020).
14. Sikora, A. et al. Impact of pharmacists to improve patient care in the critically Ill: A large multicenter analysis using meaningful
metrics with the medication regimen complexity-ICU (MRC-ICU). Crit. Care Med. 50(9), 1318–1328 (2022).
15. Knaus, W. A., Draper, E. A., Wagner, D. P. & Zimmerman, J. E. APACHE II: A severity of disease classication system. Crit. Care
Med. 13(10), 818–829 (1985).
16. Vincent, J. L. et al. Use of the SOFA score to assess the incidence of organ dysfunction/failure in intensive care units: Results of a
multicenter, prospective study. Working group on “sepsis-related problems” of the European Society of Intensive Care Medicine.
Crit. Care Med. 26, 1793–1800 (1998).
17. Buchman, T. G. Nonlinear dynamics, complex systems, and the pathobiology of critical illness. Curr. Opin. Crit. Care 10(5), 378–382
(2004).
18. Seely, A. J. et al. Proceedings from the Montebello round table discussion. Second annual conference on complexity and variability
discusses research that brings innovation to the bedside. J. Crit. Care 26(3), 325–327 (2011).
Acknowledgements
Data acquisition were supported by NC TraCS, funded by Grant Number UL1TR002489 from the National
Center for Advancing Translations Sciences at the National Institutes of Health, and Data Analytics at the Uni-
versity of North Carolina Medical Center Department of Pharmacy.
Author contributions
A.S. was responsible for project execution, design, and initial manuscript writing. J.D. and D.M. provided critical
revisions of manuscript, data interpretation, and senior level oversight. M.Y., T.Z, and X.C. handled data pre-
processing and analysis (M.Y., T.Z.) and methodology support and data interpretation (X.C.). B.M. served as site
coordinator for all data validation and procurement as well as manuscript revisions and data interpretation. S.S.,
M.B., and S.R. provided clinical interpretation, results interpretation, and manuscript revision.
Funding
Funding through Agency of Healthcare Research and Quality for Drs. Devlin, Murphy, Sikora, Smith, and
Kamaleswaran was provided through R21HS028485 and R01HS029009.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

7
Vol.:(0123456789)
Scientic Reports | (2023) 13:10784 | https://doi.org/10.1038/s41598-023-37908-1
www.nature.com/scientificreports/
Competing interests
e authors declare no competing interests.
Additional information
Supplementary Information e online version contains supplementary material available at https:// doi. org/
10. 1038/ s41598- 023- 37908-1.
Correspondence and requests for materials should be addressed to A.S.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Open Access is article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
© e Author(s) 2023
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com