Critical care primary services are associated with reduced midazolam use in the intensive care unit

Abstract

BACKGROUND

Analgesia and sedation are commonly prescribed therapies within the intensive care unit (ICU) for patients receiving mechanical ventilation. Current guidelines recommend utilizing an analgesia-first approach to initially reach appropriate pain control, while potentially achieving sedation goals concurrently. Our system employs a guideline-based ICU sedation order-set that features an electronic medical record (EMR) integrated ICU checklist that combines analgesia and sedation.

AIM

To identify systems-based factors that are associated with the use of continuous midazolam infusion administration in mechanically ventilated patients.

METHODS

We extracted EMR data from patients who received mechanical ventilation between January 1, 2021, and December 31, 2023. Subjects included were 18 years or older who received mechanical ventilation. “R” version 4.3.2 was used for data processing and statistical analysis. We performed a multivariable regression analysis to predict the administration of a continuous midazolam infusion with modified Sequential Organ Failure Assessment score, Charlson comorbidity index, and critical care medicine (CCM) primary service.

RESULTS

Of 3805 patients that underwent mechanical ventilation, 62% were male, with a mean age of 66.9 years. 3429 patients were treated by a provider team with a CCM attending, and 376 patients were managed by a non-CCM primary team with CCM consultative services. A midazolam infusion was used in 187 of 3429 (5%) patients with CCM as primary and in 166 of 376 (56%) patients with non-CCM primary (χ² 598.23, P < 0.001). Of the patients who received continuous midazolam, 117 (21%) died vs 236 (7%) survived hospitalization. Continuous midazolam was associated with more days with coma and more days with delirium (P < 0.0001).

CONCLUSION

Continuous midazolam infusion was more likely in patients admitted to the ICU under an open unit with a non-CCM physician with an intensivist consult available, despite guided order-sets and checklists integrated into the EMR.

Key Words: Delirium; Analgosedation; Intensive care unit; Critical care medicine; Midazolam infusion

Core Tip: Advancements have been made to improve outcomes among critically ill patients admitted to the intensive care units. Our study outlines the importance of adhering to guideline-based therapy when it comes to sedation, which is more commonly done in an intensivist staffing model. The choice of sedation has a strong impact on the overall care of the patient during their intensive care unit stay as well as the recovery period after they are discharged from the hospital.

INTRODUCTION

Analgesia and sedation are two core components within the intensive care unit (ICU) that significantly impact patient outcomes. Patients often require analgesia and sedation in the ICU for treatment of pain, assistance in mechanical ventilation, and management of agitation. The choice of medications and target sedation level can affect how long patients remain on mechanical ventilation and their stay in the hospital. Recent guidelines have recommended using opioid based analgesia, both non-pharmacologic and pharmacologic options, to treat patients’ pain in the ICU[1]. Additionally, intravenous opioids are recommended as the preferred agents for analgesia in mechanically ventilated patients with adjunctive agents such as acetaminophen or neuropathic agents, depending on the patient scenario[1]. As for sedation, propofol and dexmedetomidine are preferred over benzodiazepines to target a light level of sedation[1]. Recognizing that pain can go untreated, which can lead to agitation or delirium, the guidelines recommend a standard practice of analgesia first, then sedation, coined “analgosedation”[1].

Delirium is an independent predictor of morbidity and mortality in the ICU, and for each additional day a patient in the ICU spends in delirium, there is a 10% increased risk of death[2]. Delirium that persists for 2 or more days increases absolute mortality[3]. In addition to delirium being detrimental to patient outcomes, it carries a high-cost burden on the United States health system. Benzodiazepines are a known risk factor for ICU delirium[4]. Benzodiazepines provide deeper levels of sedation while increasing mechanical ventilation days, hospital, and ICU length of stay, and delirium in the landmark trials for ICU patients[5-7]. When comparing benzodiazepines to other sedatives, the MENDS trial indicated that dexmedetomidine had statistically significantly more days alive without delirium or coma when compared to lorazepam[6]. In that trial, providers were more successful in achieving their goal of sedation level with dexmedetomidine when compared to lorazepam, while also seeing more coma-free and delirium-free days[6]. In another study, lorazepam was identified to be an independent risk factor for daily transition to delirium[4]. On further review, the SEDCOM study showed that dexmedetomidine had fewer days on the ventilator, had less delirium and as well as less tachycardia and hypertension when compared to midazolam[8].

The ICU triad of delirium, pain, and agitation increases the morbidity of patients in the ICU[9]. Physicians should strive to address all three in a coordinated effort to achieve better outcomes and to minimize delirium[9]. Since benzodiazepines have been linked to increasing delirium and have been associated with increased mortality, they are no longer recommended to be used as first line therapy for critically ill patients[1]. Our institution provides our ICU teams with several evidence-backed tools that are embedded in the electronic medical record system in the form of an evidence-based “ICU pain and sedation orders for mechanically ventilated adult patients” order set. Our goal was to identify systems-based factors that are associated with guideline-inconsistent ICU sedation practice, defined as the use of continuous sedation without a trial of analgesia and continuous midazolam infusion use in mechanically ventilated patients. We hypothesized that midazolam infusion was not commonly prescribed in units with a critical care primary attending.

MATERIALS AND METHODS

Study design

We performed a retrospective observational study aimed to identify risk factors for guideline-inconsistent continuous sedation in patients admitted to our hospital system’s ICUs between January 1, 2021 and December 31, 2023. Our institution comprises seven hospitals, eight ICUs, five of which are teaching institutions. We queried the patients who were admitted to any of the seven hospitals in our Central Pennsylvania hospital system that underwent mechanical ventilation. We assessed receipt of the following continuous infusions during the ICU admission: Fentanyl, propofol, midazolam, dexmedetomidine, ketamine, and hydromorphone.

We obtained order details, including the date and time of the initial order and the ordering provider’s specialty. Data was extracted from order history, flowsheets, vital signs, demographics, and laboratory results. Sedative infusion administration was compared against the current 2018 Society for Critical Care Medicine practice guidelines[1] and our institution’s ICU sedation protocol. We classified patients as having received either guideline-consistent or inconsistent sedation according to the 2018 Society for Critical Care Medicine guidelines and the order of sedative infusion initiation. We identified patient factors [e.g., age, Sequential Organ Failure Assessment (SOFA) score, admission diagnosis] and system factors (e.g., teaching vs non-teaching, critical care vs non-critical care ordering provider) that are associated with guideline-inconsistent sedative use.

Participants

Inclusion criteria: Subjects 18 years old or older who were admitted to the ICU between January 1, 2021 and December 31, 2023, and who were administered any of the following continuous sedatives during mechanical ventilation (fentanyl, propofol, midazolam, dexmedetomidine, ketamine, and hydromorphone). We included patients who were documented in a comatose state using the Riker Sedation-Agitation Scale (RSAS), with a score of 1 or 2. The RSAS score was obtained from the nursing flowsheets, and we used the first score after sedation was started.

We excluded patients who were younger than 18 years old, patients who were placed on extracorporeal membrane oxygenation, and patients who were not receiving mechanical ventilation.

Variables

Primary outcome: Our primary outcome was the percentage of ICU patients who received continuous midazolam infusion between critical care and non-critical care providers, and between teaching and non-teaching ICU settings. We hypothesized that patients under the care of an intensivist were less likely to receive midazolam infusion.

Secondary outcome: Our secondary outcome was to compare predetermined baseline characteristics and to identify systems-based factors that are associated with guideline-inconsistent ICU sedation practices between our ICUs, defined as the use of continuous midazolam infusion without a trial of analgesia in mechanically ventilated patients.

Statistical analysis

Normality tests of continuous variables were performed with the Shapiro-Wilk test. Variables with normal distribution were reported as mean, standard deviation, and range. Variables with non-normal distribution were presented as median (interquartile range). The student’s t-test or two-sample Wilcoxon test was used to analyze group differences for the continuous variables. Categorical variables, expressed as numbers and percentages, were compared using the χ² test or Fisher’s exact test. A multiple logistic regression model was developed to identify significant factors associated with the use of continuous midazolam infusion. Odds ratios with 95% confidence intervals were reported for each independent variable. A P-value less than 0.05 was considered statistically significant. Missing data was imputed with multiple imputations of chained equations. All the analyses were done by R (version 4.3.2) and SAS 9.4 (SAS Institute, Cary, NC, United States).

Bias

This study is at risk for selection bias and other confounding variables, given the retrospective nature of the study.

RESULTS

We identified 3805 patients during the treatment period, with 3429 patients (90%) treated under a critical care attending physician (intensivist staffing model or CCM), and 376 patients (10%) were treated under a non-critical care attending physician with the availability of consultation from an intensivist. Of the patient sample, 62.9% were male gender in the intensivist staffing model, with a mean age of 65.8 years (Standard deviation 12.9); the non-intensivist staffing model had 55.6% male patients with a mean age of 68.1 years (Standard deviation 13.9) (Table 1). There was a difference in the Charlson Comorbidity index between the two groups, with the Charlson Comorbidity Index being slightly higher in the non-intensivist staffing group (t-test = 4.03, P < 0.001). Likewise, the SOFA score was marginally higher in the non-intensivist staffing group (t-test = 3.85, P < 0.001). The absolute mortality was higher in the non-intensivist staffing model. Table 1 also demonstrates the list of common diagnoses admitted to the ICU in our cohort. This data is also demonstrated graphically in Figure 1.

Figure 1 Common diagnoses on admission to the intensive care unit.

Table 1 Demographics by critical care primary, n (%).

	Intensivist staffing model
	No	Yes
Number patients	376.0	3429.0
Number of males	209.0	2156.0
Percent males	55.6	62.9
Mean age (years)	68.1	65.8
Age (SD, deviation)	13.9	12.9
Mean weight (kg)	90.5	88.8
Weight (SD, deviation)	27.3	24.1
Charlson comorbidity index (mean ± SD)	6.0	5.6
Charlson comorbidity index (SD, deviation)	2.8	2.8
Sofa score (mean ± SD)	7.9	6.8
Sofa score (SD, deviation)	2.5	2.6
Hospital mortality	26.1	13.2
Hospital diagnoses
Acute respiratory failure	217 (58)	883 (26)
Sepsis	194 (52)	660 (19)
Septic shock	137 (36)	471 (14)
Pneumonia	130 (35)	549 (16)
Cardiac arrest	44 (12)	322 (9)
Unspecified shock	42 (11)	214 (6)
COVID-19 pneumonia	40 (11)	130 (4)
Viral pneumonia	38 (10)	109 (3)
Alcohol withdrawal	35 (9)	136 (4)
Cardiogenic shock	22 (6)	233 (7)
Hypovolemic shock	21 (6)	70 (2)
ARDS	18 (5)	110 (3)
Seizure	18 (5)	88 (3)
Status epilepticus	2 (1)	47 (1)

COVID-19: Coronavirus disease 2019; ARDS: Acute respiratory distress syndrome.

Figure 2 demonstrates the distribution of continuous infusions between critical care primary services and non-critical care primary services. Figure 3 demonstrated the final disposition of patients who received midazolam infusion (t-test = 2.14, 95% confidence intervals: 4.95-167.99, P = 0.038). Table 2 showed 117 (21%) patients discharged as deceased who received midazolam infusion (χ² = 108.73, P < 0.001). Amongst the providers who ordered midazolam infusion, CCM ordered the infusion on 187 (5%) patients when compared to 166 (44%) patients amongst the non-CCM providers (χ² = 598.23, P < 0.001) (Table 3). Amongst the teaching ICUs, teaching institutions ordered midazolam infusion on 128 (4%) patients vs 225 (27%) patients in non-teaching institutions (χ² = 390.78, P < 0.001) (Table 4), which is also demonstrated in Figure 4. Figure 5 showed the distribution of patients who received midazolam infusion across our ICU locations (χ² = 835.51, P < 0.001). 271 (77.87%) patients were identified to be in a comatose state (RSAS score ≤ 2) vs 1340 (38.76%) patients who received continuous midazolam infusion and who did not, respectively (P < 0.001). The median number of days with coma was 4 vs 2, respectively (P < 0.001). The 104 (29.89%) patients were identified to have delirium (confusion assessment method-ICU positive) vs 586 (16.95%) patients who received continuous midazolam infusion and who did not, respectively (P < 0.001). The median number of days with delirium was 3 vs 2, respectively (P = 0.0006). This is represented in Figure 6 on a logarithmic scale.

Figure 2 Proportion of sedatives ordered by intensivists vs non-intensivists.

Figure 3 Discharge disposition with midazolam use (n = 3805).

Figure 4 Association of midazolam use and teaching intensive care unit. ICU: Intensive care unit.

Figure 5 Association of midazolam use and intensive care unit locations. ICU: Intensive care unit.

Figure 6 Association of coma and delirium with midazolam infusion.

Table 2 Discharged deceased and midazolam use, n (%).

	Midazolam use1,a
Deceased	No	Yes
No	3020 (93)	236 (7)
Yes	432 (79)	117 (21)

¹χ² = 108.73.

^aP < 0.001.

Table 3 Critical care medicine primary and midazolam use.

	Midazolam use1,a
Intensivist staffing model	No	Yes
No	210 (56)	166 (44)
Yes	3242 (95)	187 (5)

¹χ² = 598.23.

^aP < 0.001.

Table 4 Teaching intensive care unit and midazolam use.

	Midazolam use1,a
Teaching intensive care unit	No	Yes
No	614 (73)	225 (27)
Yes	2838 (96)	128 (4)

¹χ² = 390.78.

^aP < 0.001.

We performed a multiple logistic regression, shown in Table 5, which showed that older adults (age > 65), teaching ICU, and intensivist staffing modeled ICUs are less likely to order continuous midazolam infusions [odds ratios (OR) = 0.43, OR = 0.31, OR = 0.11, respectively, P < 0.001]. There was no difference in patients who received continuous midazolam infusion with a Charlson comorbidity index of more than 5 (OR = 0.94, P = 0.72). Patients who had a greater illness severity, indicated by SOFA score more than 7, were more likely to receive continuous midazolam infusion (OR = 1.74, P = 0.0002). Table 5 also shows that among the diagnoses that are admitted to the ICU, the ones that are more commonly associated with continuous midazolam infusion are: Acute respiratory failure, acute respiratory distress syndrome, cardiogenic shock, pneumonia (bacterial and viral) (P < 0.0001).

Table 5 Multivariate analysis on demographics and risk factors associated with the use of continuous midazolam infusion.

Predictors	OR	95%CI (lower limit, upper limit)	P value
Age > 65 years	0.43	0.31, 0.59	< 0.0001
Teaching ICU	0.31	0.23, 0.42	< 0.0001
CCM primary	0.11	0.08, 0.16	< 0.0001
Charlson score > 5	0.94	0.69, 1.29	0.7151
SOFA score > 7	1.74	1.30, 2.33	0.0002
Acute respiratory failure	2.11	1.51, 2.94	< 0.0001
ARDS	6.95	3.82, 12.64	< 0.0001
Cardiogenic shock	3.01	1.95, 4.64	< 0.0001
Septic shock	0.80	0.51, 1.25	0.3324
Sepsis	1.22	0.79, 1.88	0.3714
Hypovolemic shock	1.05	0.52, 2.12	0.8912
Unspecified shock	1.51	1.00, 2.28	0.0504
Cardiac arrest	1.38	0.94, 2.02	0.1034
Pneumonia (bacterial)	1.56	1.14, 2.14	0.005
Viral pneumonia	7.80	2.11, 28.80	0.0021
COVID PNA	0.41	0.11, 1.49	0.1742
Seizure	1.77	0.92, 3.42	0.0895
Status epilepticus	13.17	6.56, 26.45	< 0.0001
Alcohol withdrawal	2.30	1.41, 3.74	0.0008

OR: Odds ratios; CI: Confidence interval; ICU: Intensive care unit; CCM: Critical care medicine; SOFA: Sequential Organ Failure Assessment; ARDS: Acute respiratory distress syndrome; COVID PNA: Coronavirus disease peptide nucleic acid.

DISCUSSION

Analgosedation means the use of analgesia first, in the form of an opioid, before the addition of a sedative to achieve a sedation goal[1]. Our retrospective observational study has shown that continuous midazolam infusions were more likely to be administered in open, non-intensivist staffed units. We also identified that older adults, teaching ICU, and closed ICU were consistent with less midazolam use, whereas there was no difference between the Charlson comorbidity index. Patients who had a greater illness severity (SOFA score > 7) were more likely to receive continuous midazolam infusion. We observed that mortality was higher among patients who received continuous midazolam infusion compared to those who did not (21% vs 7%) (Table 2). This finding is consistent with the literature regarding benzodiazepines. Additionally, it is known that benzodiazepines are an independent risk factor for delirium. Our study showed that there were longer coma days and delirium days in patients who received continuous midazolam infusion. Patients who were admitted with acute respiratory failure, acute respiratory distress syndrome, cardiogenic shock, and pneumonia (bacterial and viral) were more likely to receive continuous midazolam. Given the findings of our study, we can postulate that there is the possibility of a lack of familiarity with the 2018 and 2025 Society of Critical Care Medicine guidelines, along with unfamiliarity with available EMR tools in our ICUs that do not have an intensivist staffing model. Our findings highlight the importance of ongoing education and, in addition to institutional quality improvement initiatives, delivering the best possible care to mechanically ventilated patients.

The guidelines recommend targeting a lighter sedation goal by using either propofol or dexmedetomidine instead of benzodiazepines[1], and more recently, they have updated their guidelines, and a conditional recommendation was made to use dexmedetomidine over propofol for light sedation[2]. Deep sedation has been associated with decreased in-hospital and two-year survival[10]. In addition, deep sedation goals have been found to be associated with delirium, increased duration on mechanical ventilation, and longer ICU length of stay[11]. Our study was limited by the retrospective and observational design. When identified, we imputed missing data using multiple imputation chained equations. Our findings may also have been affected by changes in practice patterns that may have occurred during the global coronavirus disease 2019 pandemic. Nonetheless, they demonstrate the importance of fundamental ICU care in the choice of sedative infusions. Additionally, our retrospective study did not directly evaluate the use of continuous neuromuscular blocking agents, which may have impacted the decision on the use of continuous midazolam infusion.

Surviving an ICU stay does not equate to full recovery. A growing body of evidence has drawn attention to Post-Intensive Care Syndrome (PICS), which affects up to 80% of ICU survivors[12]. PICS encompasses a range of long-term physical, cognitive, and psychological impairments, including depression, anxiety, post-traumatic stress disorder, memory loss, and deficits in executive function[12]. These complications can persist well after discharge, significantly diminishing quality of life. The burden of PICS extends beyond the patient. Families often face emotional strain, financial hardship from medical debt and lost income, and increased caregiver responsibilities. The ripple effects of an ICU admission can be profound and enduring[12].

Given the serious implications of PICS, ICU providers must proactively address modifiable risk factors[12]. Evidence-based strategies - such as minimizing sedation, promoting early mobility, and actively managing delirium - are essential. Among these, the choice of sedative is particularly impactful. Benzodiazepines, especially when used continuously, have been strongly associated with worse cognitive and psychological outcomes. By reducing their use and following current sedation guidelines, we can take meaningful steps toward reducing the incidence and severity of PICS.

CONCLUSION

Continuous midazolam infusion was more likely in patients admitted to the ICU under an open unit with a non-CCM physician with an intensivist consult available, despite guided order-sets and checklists integrated into the electronic medical record.