Abstract
Objectives:
Providing analgesia and sedation while allowing for neurological assessment is important in the neurocritical care unit (NCCU), yet data are limited about the effects of protocolised analgesia and sedation. We developed an analgesia-based sedation protocol and evaluated its effect on medication utilisation and costs in the NCCU.
Methods:
We conducted a retrospective cohort study of patients who are mechanically ventilated and admitted to a 12-bed NCCU over four years. To compare outcomes, we used gamma and negative binomial regression models, and interrupted time-series sensitivity analyses.
Results:
The study cohort consisted of 1197 patients: 576 pre-protocol and 621 post-protocol. The protocol resulted in an increase in fentanyl use [incidence rate ratio (IRR) = 2.8, (95% confidence limits (CLs) 1.9, 4.2)] and a decrease in propofol use (IRR = 0.8, CLs 0.6, 1.0). There was a decrease in fentanyl (cost ratio = 0.8, CLs 0.5, 1.1) and propofol costs (cost ratio = 0.6, CLs 0.5, 0.8). The sensitivity analyses results were similar. There was no effect on healthcare utilisation, healthcare costs, and in-hospital mortality.
Conclusion:
Protocolised analgesia and sedation increased analgesia use, decreased sedative use, and reduced medication-associated costs in the NCCU. Our results suggest that similar NCCUs should consider use of population-specific protocols to manage analgesia and sedation.
Keywords: Analgesia, hypnotics and sedatives, critical care, intensive care units, clinical protocols, drug costs
Introduction
Analgesia and sedation are often needed in the intensive care unit (ICU) to treat pain, anxiety and agitation (1). Proper sedation is especially important in patients with neurocritical illness who require frequent neurological assessment and exams (2,3). However, the use of analgesia and sedation can have a negative impact on patients, as it may result in increased days of mechanical ventilation, longer weaning times and increased lengths of stay (4). Analgesia and sedation can affect the clinician’s ability to perform neurological exams and may cause hemodynamic instability (5,6).
The Society of Critical Care Medicine recommends an analgesia-based sedation approach in the general critical care population (5). Studies suggest that the use of analgesia-based sedation can help achieve sedation goals, decrease the amount of sedatives used, shorten the length of ICU stays and reduce ventilator days (7–10). Providing periods of interrupted sedation or light sedation through a protocol can improve clinical outcomes in patients who are critically ill. Studies have shown that it decreases the length of ICU days, improves ICU outcomes, prevents accumulation of sedatives, and decreases the need for mechanical ventilation (1,5,11). There is little data to document the effects of protocolised analgesia and sedation on patients with neurological injury. The effects of protocolised analgesia and sedation may differ in patients with neurological injury as this population requires frequent awakenings to allow for neurological exams (12). Furthermore, in the absence of a protocol, there is often significant heterogeneity in how providers address pain or respond to agitation due to the complexity of this population.
Therefore, the first objective of this study was to develop a nurse-driven analgesia-based sedation protocol for the neurocritical care unit (NCCU) and evaluate its effect on medication utilisation and costs. The second objective was to examine the effect of the protocol on length of hospital and NCCU stays, ventilator days, hospital and NCCU costs, respiratory care costs and in-hospital mortality. We hypothesised that the protocol would decrease the utilisation and cost of sedation in the NCCU and subsequently decrease length of hospital and NCCU stays, ventilator days, hospital and NCCU costs, respiratory care costs and in-hospital mortality.
Methods
The study protocol was approved by the Institutional Review Board.
Study Population and Data Source
We conducted a retrospective cohort study of all adult patients age ≥18 years admitted to a 12-bed NCCU in a tertiary care academic medical centre over four years from 2/1/2011 to 1/31/2015. All intubated patients were included except for those with a diagnosis of status epilepticus or refractory intracranial hypertension. These patients were excluded because the usual sedative or analgesic doses and assessment tools might not be applicable or appropriate. Demographic data (age and sex), admission year, and diagnosis data were obtained from the electronic medical record during each patient’s hospitalisation. Information on utilisation outcomes like length of stay (in days) was also obtained from the electronic medical record. Pharmacy dispensing and billing records were used to ascertain medications dispensed and costs per patient. Hospital billing records permitted identification of hospital, ICU and respiratory care costs.
Protocol for Analgesia and Sedation
The study period before the nurse-driven protocol was implemented started on 2/1/2011 and ended 1/31/2013. The study period during which the protocol was in place started on 2/1/2013 and ended on 1/31/2015. There was not a washout period in this study, however the protocol was not initiated until all nursing competencies had been completed. Education was conducted extensively prior to the initiation of the protocol and routinely thereafter. Pain was assessed using the Face, Legs, Activity, Cry, Consolability (FLACC) Behavioral Scale (13), which was the standard non-verbal pain assessment tool at our hospital. The goal for analgesia was FLACC 0–3, indicating mild pain or pain not present (13). Pain was treated with fentanyl. Pain was assessed every 5 minutes at protocol initiation, and every 30 minutes to an hour once placed on the maintenance protocol based on the FLACC value. Agitation was assessed using the Richmond Agitation and Sedation Scale (RASS) (14). The goal for sedation was RASS 0 to −2, indicating patient is calm, alert or lightly sedated and treated with propofol (first-line) or midazolam (second-line) (14). Agitation was assessed every 5–15 minutes at protocol initiation, then every 15 minutes to an hour thereafter based on RASS value. Our target goals for FLACC and RASS were monitored daily on our multidisciplinary rounds. The detailed protocol is available in the supplementary materials.
Outcome Measures
Medication utilisation was measured in units, according to available package size. Fentanyl and midazolam units were intravenous vials or continuous infusion bags. Propofol units were intravenous continuous infusion bottles. Outcomes were ascertained during the hospital stay.
Statistical Analyses
We used descriptive statistics to describe the characteristics of our study population. All dollars were adjusted for inflation and represent fiscal year 2015 US dollars. To estimate the effect of the protocol on outcomes, we used gamma regression models to estimate cost ratios (CRs) with 95% confidence limits (CLs) for cost outcomes, negative binomial regression models to estimate incidence rate ratios (IRRs) for length of stay and drug utilisation outcomes, and log-binomial regression models to estimate risk ratios (RRs) for mortality (15). We selected gamma regression models to account for the skewed distribution of cost and because we observed the variance of our cost outcomes to increase proportionally with the mean (15). Negative binomial regression was used for length of stay and units of drug dispensed because it was count data and we observed that the variance was often greater than the mean in our data (15). Since the implementation of the protocol was at the unit level and unrelated to individual patient characteristics, concern about potential confounding bias was reduced (16). However, estimates were adjusted for age and sex to account for differences in these demographics over time. There was no missing data or loss to follow up in our study.
As a sensitivity analysis to account for any potential underlying secular trends (e.g. increases in patients’ clinical complexity over time) we used segmented time-series regression models (i.e. interrupted time series analyses) (17–19). We analysed outcomes by dividing time into periods of 90 days with 8 periods before the intervention and 8 periods after the intervention (2 years in each period). Segmented regression models fit a least squares regression line to each segment of the independent variable, quarter of calendar time, and thus assume a linear relationship between time and the outcome within each segment (19). Regression models included terms for baseline level and trend, as well as terms to estimate the changes in level (i.e. ‘level change’, or the immediate increase or decrease in the first quarter of calendar time after protocol implementation) and trend beginning with the first post-protocol period (19). To remove the influence of high-leverage outliers (x-axis outliers with the potential to exert undue influence on regression coefficient estimates) on healthcare utilisation and cost outcomes, we excluded two observed outliers in the first and last observation periods (15).
Statistical significance was based on a two-sided type 1 error of 0.05. Statistical analyses were conducted using Stata SE version 14 (Stata, College Station, TX, USA).
Results
Study population
A total of 1197 patients were studied: 576 patients before the analgesia and sedation protocol began and 621 patients after the protocol was initiated in the NCCU. The mean age of the overall study cohort was 60 years old and 57% were male. Most patients were admitted to the NCCU for traumatic brain injury (22.3%), intracerebral haemorrhage (19.7%), and subarachnoid haemorrhage (14.9%).
After stratifying by pre-protocol versus post-protocol, the distributions of age, sex, and admission diagnoses were similar between the pre-protocol and post-protocol groups: traumatic brain injury (21.4% vs. 23.3%), intracerebral haemorrhage (20.5% vs. 18.9%), subarachnoid haemorrhage (14% vs. 15.8%), respectively. (Table 1).
Table 1.
Baseline characteristics of neurocritical care patients, 2011–2015.
| Overall N = 1,197 | No Protocol n = 576 | Protocol n = 621 | |
|---|---|---|---|
|
| |||
| Age in years, mean (SD) | 60 (18) | 59 (18) | 61 (17) |
| Male, n (%) | 682 (57) | 315 (55) | 367 (59) |
| Admission diagnosis, n (%) | |||
| Traumatic brain injury | 267 (22.3) | 134 (23.3) | 133 (21.4) |
| Intracerebral haemorrhage | 236 (19.7) | 109 (18.9) | 127 (20.5) |
| Subarachnoid haemorrhage | 178 (14.9) | 91 (15.8) | 87 (14.0) |
| Ischemic stroke | 121 (10.1) | 54 (9.4) | 67 (10.8) |
| Brain tumour | 98 (8.2) | 52 (9.0) | 46 (7.4) |
| Seizure | 74 (6.2) | 26 (4.5) | 48 (7.7) |
| Nontraumatic subdural hematoma | 44 (3.7) | 23 (4.0) | 21 (3.4) |
| Traumatic spinal cord injury or cord compression | 33 (2.8) | 17 (3.0) | 16 (2.6) |
| Myelopathy | 31 (2.6) | 17 (3.0) | 14 (2.3) |
| Hydrocephalus | 17 (1.4) | 7 (1.2) | 10 (1.6) |
| Intracranial abscess | 15 (1.3) | 7 (1.2) | 8 (1.3) |
| Myasthenia Gravis/Lambert Eaton Syndrome | 9 (0.7) | 6 (1.0) | 3 (0.5) |
| Unruptured or elective aneurysm/vascular | 11 (0.9) | 5 (0.9) | 6 (1.0) |
| Malformation | |||
| Anoxia/Hypoxia | 5 (0.4) | 3 (0.5) | 2 (0.3) |
| Intraspinal abscess | 10 (0.8) | 2 (0.4) | 8 (1.3) |
| Amyotrophic lateral sclerosis | 1 (0.1) | 1 (0.2) | 0 (0.0) |
| Admission year, n (%) | |||
| 2011 | 251 (21.0) | 251 (43.6) | 0 (0.0) |
| 2012 | 298 (24.9) | 298 (51.7) | 0 (0.0) |
| 2013 | 315 (26.3) | 25 (4.3) | 290 (46.7) |
| 2014 | 311 (26.0) | 2 (0.4) | 309 (49.6) |
| 2015 | 23 (1.9) | 0 (0.0) | 23 (3.7) |
Medication utilisation and costs
In a primary analysis (Table 2) for the effect of the protocol on medication utilisation and cost, more fentanyl was used after protocol implementation than before (adjusted CR = 2.8, 95% CLs 1.9, 4.2). Conversely, the cost associated with fentanyl was reduced after protocol implementation (adjusted CR = 0.8, 95% CLs 0.5, 1.1), though this reduction was not statistically significant. Less propofol was used after the protocol was implemented (adjusted IRR = 0.8, 95% CLs 0.6, 1.0), but a similar amount of midazolam was used (adjusted IRR = 1.0, 95% CLs 0.8, 1.2). Likewise, we found that there was a decrease in propofol costs after implementation of the protocol (adjusted CR = 0.6, 95% CLs, 0.5, 0.8), but little difference in midazolam costs (adjusted RR = 0.9, 95% CLs 0.7, 1.1).
Table 2.
Primary analysis for the effect of a neurocritical care analgesia and sedation protocol on medication utilisation and costs, 2011–2015 (N = 1,197).
| Outcomes | Measure | No Protocol n = 576 | Protocol n = 621 | Unadjusted Estimate (95% CLs)a | P | Adjusted Estimate (95% CLs)a,b | P |
|---|---|---|---|---|---|---|---|
|
| |||||||
| Fentanyl unitsc used per patient | Median (IQR) | 66 (38–127) | 279 (240–325) | 2.4 (1.6, 3.6) | <0.001 | 2.8 (1.9, 4.2) | <0.001 |
| Mean (SD) | 122 (154) | 293 (72) | |||||
| Fentanyl cost per patient, USDd | Median (IQR) | 450 (161–834) | 272 (180–578) | 0.8 (0.5, 1.1) | 0.16 | 0.8 (0.5, 1.1) | 0.18 |
| Mean (SD) | 496 (353) | 372 (268) | |||||
| Propofol unitsc used per patient | Median (IQR) | 190 (145–240) | 148 (111–189) | 0.8 (0.7, 1.0) | 0.04 | 0.8 (0.6, 1.0) | 0.03 |
| Mean (SD) | 200 (73) | 160 (71) | |||||
| Propofol cost per patient, USDd | Median (IQR) | 2,306 (1,721 –2,955) | 1,585 (1,078–1,980) | 0.6 (0.5, 0.8) | <0.001 | 0.6 (0.5, 0.8) | <0.001 |
| Mean (SD) | 2,419 (887) | 1,541 (522) | |||||
| Midazolam unitsc used per patient | Median (IQR) | 96 (62–124) | 88 (62–100) | 0.9 (0.8, 1.2) | 0.53 | 1.0 (0.8, 1.2) | 0.80 |
| Mean (SD) | 91 (38) | 85 (23) | |||||
| Midazolam cost per patient, USDd | Median (IQR) | 45 (29–59) | 39 (28–45) | 0.9 (0.7, 1.1) | 0.20 | 0.9 (0.7, 1.1) | 0.41 |
| Mean (SD) | 44 (18) | 38 (10) | |||||
Abbreviation: CLs, Confidence Limits; IQR, interquartile range; SD, standard deviation; USD, U.S. Dollars.
Estimates are cost ratios for cost outcomes and incidence rate ratios for medication utilisation outcomes (units).
Adjusted for age at the time of admission and sex.
Medication utilisation was measured in units, according to available package size. Fentanyl and midazolam units were intravenous vials or continuous infusion bags. Propofol units were intravenous continuous infusion bottles.
Dollars are represented as 2015 fiscal year dollars.
In the segmented time-series regression sensitivity analysis (Table 3), results were similar to the primary analysis for fentanyl utilisation and costs, with a significant increase in fentanyl use in the first post-protocol time period (344.0 units per person per hospital stay, 95% CLs 238.0, 450.0) and nonsignificant decrease in cost (−251 USD per person per hospital stay, 95% CLs −750.1, 248.1). We also observed a significant decreasing time trend for fentanyl costs after protocol implementation—about 98 dollars (95% CLs −192.4, −4.5) on average per person per hospital stay for each additional quarter of calendar time. The sensitivity analysis results were similar to those from the primary analysis for propofol costs; there was a nonsignificant immediate decrease (−458.6 USD per patient, 95% CLs −2286.5, 1369.5) and a non-significant decreasing post-protocol time trend (−162.8 USD per patient, 95% CLs −519.6, 194.1). The results for midazolam were not markedly different from the primary analysis.
Table 3.
Segmented time-series regression sensitivity analysis for the effect of a neurocritical care analgesia and sedation protocol on medication utilisation and costs, 2011–2015 (N = 1,197).
| Outcomes | Model Variable (95% CLs) |
|||
|---|---|---|---|---|
| Intercepta | Baseline Linear Trendb | Level Changec | Post-protocol Linear Trendd | |
|
| ||||
| Fentanyl unitse used per patient | 130.0 (78.9, 181.1) | −14.9 (−24.5, −5.31) | 344.0 (238.0, 450.0) | 5.2 (−15.4, 25.9) |
| P-value | <0.001 | <0.01 | <0.001 | 0.59 |
| Fentanyl cost per patient, USDf | 229.0 (−144.7, 602.7) | 67.2 (−23.5, 157.9) | −251.0 (−750.1, 248.1) | −98.4 (−192.4, −4.5) |
| P-value | 0.21 | 0.13 | 0.29 | 0.04 |
| Propofol unitse used per patient | 174.5(74.9, 274.1) | 5.4 (−23.3, 34.0) | 1.0 (−151.7, 153.6) | −17.9 (−45.3, 11.5) |
| P-value | <0.001 | 0.69 | 0.99 | 0.21 |
| Propofol cost per patient, USDf | 2,194.8 (959.7, 3429.9) | 49.7 (−296.7, 396.0) | −458.6 (−2,286.5, 1,369.5) | −162.8 (−519.6, 194.1) |
| P-value | <0.001 | 0.76 | 0.59 | 0.34 |
| Midazolam unitse used per patient | 49.0 (−7.6, 105.6) | 9.2 (−0.7, 19.0) | −40.5 (−88.8, 7.8) | −7.2 (−18.4, 4.0) |
| P-value | 0.08 | 0.07 | 0.09 | 0.19 |
| Midazolam cost per patient, USDf | 24.5 (−3.55, 52.5) | 4.13 (−0.75, 9.02) | −19.16 (−42.48, 4.17) | −3.61 (−9.04, 1.82) |
| P-value | 0.08 | 0.09 | 0.10 | 0.18 |
Abbreviation: CLs, Confidence limits.
The intercept is the model estimate for the outcomes in the first quarter of the study period.
Baseline linear trend describes the change in the outcome per quarter in the pre-protocol change period (2/1/2011 to 1/31/2013).
Level change is the estimated instantaneous change in the outcome in the first post-protocol period (2/1/2013 to 4/30/2013) compared with the expected rate based on the baseline trend.
Post-protocol linear trend describes the change in trend in the post-protocol change period (5/1/2013 to 1/31/2015); the trend in the post-protocol change period is the difference between the pre-protocol and post-protocol slopes of the outcome.
Medication utilisation was measured in units, according to available package size. Fentanyl and midazolam units were intravenous vials or continuous infusion bags. Propofol units were intravenous continuous infusion bottles.
Dollars are represented as 2015 fiscal year dollars.
Healthcare utilisation and cost
In a primary analysis (Table 4) for the effect of the protocol on healthcare utilisation and cost, hospital days significantly decreased (adjusted IRR = 0.8, 95% CLs 0.7, 0.9) after implementation of the protocol. We did not observe any significant changes in NCCU days and ventilator days. Similarly, total hospital cost, NCCU cost and respiratory care cost did not change significantly after the implementation of the protocol.
Table 4.
Primary analysis for the effect of a neurocritical care analgesia and sedation protocol on healthcare utilisation and costs, 2011–2015 (N = 1,197).
| Outcomes | Measure | No Protocol n = 576 | Protocol n = 621 | Unadjusted Estimate (95% CLs)a | P | Adjusted Estimate (95% CLs)a,b | P |
|---|---|---|---|---|---|---|---|
|
| |||||||
| Total hospital days | Median (IQR) | 10 (5–20) | 10 (4–20) | 0.7 (0.7, 0.8) | <0.001 | 0.8 (0.7, 0.9) | <0.001 |
| Mean (SD) | 21 (54) | 15 (18) | |||||
| NCCU days | Median (IQR) | 6 (2–13) | 5 (2–12) | 0.9 (0.8, 1.0) | 0.11 | 0.9 (0.8, 1.1) | 0.32 |
| Mean (SD) | 10 (16) | 9 (11) | |||||
| Ventilator days | Median (IQR) | 2 (2–7) | 2 (2–7) | 0.9 (0.8, 1.0) | 0.08 | 0.9 (0.8, 1.0) | 0.19 |
| Mean (SD) | 6 (14) | 6 (8) | |||||
| Total hospital costs, USDc | Median (IQR) | 47,247 (24,337–88,487) | 50.935 (24,597–90,265) | 1.0 (0.8, 1.1) | 0.42 | 1.0 (0.9, 1.1) | 0.80 |
| Mean (SD) | 76,082 (109,430) | 66,142 (58,889) | |||||
| NCCU costs, USDc | Median (IQR) | 16,279 (5,426–35,272) | 14,833 (5,502–35,547) | 0.9 (0.8, 1.1) | 0.33 | 1.0 (0.8, 1.1) | 0.64 |
| Mean (SD) | 27,632 (40,832) | 25,667 (28,881) | |||||
| Respiratory care costs, USDc | Median (IQR) | 1,441 (731–4,733) | 1,699 (870–5,138) | 1.0 (0.8, 1.2) | 0.59 | 1.0 (0.8, 1.2) | 0.79 |
| Mean (SD) | 4,185 (8,311) | 3,966 (5,491) | |||||
Abbreviation: CLs, Confidence limits; IQR, interquartile range; SD, standard deviation; NCCU, neurocritical care unit; USD, US Dollars.
Estimates are cost ratios for cost outcomes and incidence rate ratios for healthcare utilisation outcomes (hospital days, NCCU days, ventilator days).
Adjusted for age at the time of admission and sex.
Dollars are represented as 2015 fiscal year dollars.
In the segmented time-series regression sensitivity analysis (Table 5), in contrast to the primary analysis, hospital length of stay was not significantly different in the first 90-day post-protocol time period (−0.5 days per person per hospital stay, 95% CLs −11.3, 10.3). We observed a similar effect on NCCU days and ventilator days with an immediate increase (2.7 NCCU days per person per hospital stay, 95% CLs −0.4, 5.7 and 2.2 ventilator days per person per hospital stay, 95% CLs 0.02, 4.2). Contrary to the primary analysis, in the sensitivity analysis, healthcare cost seemed to significantly increase in the first 90-day post-protocol period; total hospital cost (10,026.5 USD per person per hospital stay 95% CLs 5673.8, 25 726.8), NCCU cost (7640.8 USD per person per hospital stay, 95% CLs −395.3, 15 677.0), and respiratory care cost (1759.5 USD per person per hospital stay 95% CLs 179.8, 3339.2). When we re-included outliers in the dataset, this alternate version of the sensitivity analysis produced results that were similar to those from the primary segmented time-series regression sensitivity analysis that excluded outliers (see supplementary materials).
Table 5.
Segmented time-series regression sensitivity analysis for the effect of a neurocritical care analgesia and sedation protocol on healthcare utilisation and costs after removing high-leverage outliers, 2011–2015 (N = 1,197).
| Outcomes | Model Variable (95% CLs) |
|||
|---|---|---|---|---|
| Intercepta | Baseline Linear Trendb | Level Changec | Post-protocol Linear Trendd | |
|
| ||||
| Total hospital days | 23.0 (15.9, 30.0) | −0.7 (−2.8, 1.4) | −0.5 (−11.3, 10.3) | 0.2 (−2.0, 2.3) |
| P-value | <0.001 | 0.49 | 0.93 | 0.88 |
| NCCU days | 11.7 (8.9, 14.5) | −0.5 (−1.2, 0.2) | 2.7 (−0.4, 5.7) | 0.2 (−0.6, 1.0) |
| P-value | <0.001 | 0.13 | 0.08 | 0.58 |
| Ventilator days | 7.4 (5.0, 9.8) | −0.5 (−1.0, 0.0) | 2.2 (0.02, 4.2) | 0.3 (−0.3, 0.9) |
| P-value | <0.001 | 0.07 | 0.03 | 0.25 |
| Total hospital costs, USDe | 83,882.8 (64,503.7, 103,262.0) | −2,557.9 (−6,706.0, 1,590.3) | 10,026.5 (−5,673.8, 25,726.8) | 541.7 (−3,923.9, 5,007.4) |
| P-value | <0.001 | 0.20 | 0.19 | 0.80 |
| NCCU costs, USDe | 31,524.7 (25,027.7, 38,021.7) | −1,385.7 (−2,989.7, 218.3) | 7,640.8 (−395.3, 15,677.0) | 687.3 (−1,196.4, 2,571.0) |
| P-value | <0.001 | 0.08 | 0.06 | 0.44 |
| Respiratory care costs, USDe | 4,841.1 (3,432.4, 6,249.8) | −285.8 (−600.9, 29.4) | 1,759.5 (179.8, 3,339.2) | 184.8 (−228.5, 598.1) |
| P-value | <0.001 | 0.07 | 0.03 | 0.35 |
Abbreviation: CLs, Confidence limits; NCCU, neurocritical care unit.
The intercept is the model estimate for the outcomes in the first quarter of the study period.
Baseline linear trend describes the change in the outcome per quarter in the pre-protocol change period (2/1/2011 to 1/31/2013).
Level change is the estimated instantaneous change in the outcome in the first post-protocol period (2/1/2013 to 4/30/2013) compared with the expected rate based on the baseline trend.
Post-protocol linear trend describes the change in trend in the post-protocol change period (5/1/2013 to 1/31/2015); the trend in the post-protocol change period is the difference between the pre-protocol and post-protocol slopes of the outcome.
Dollars are represented as 2015 fiscal year dollars.
The protocol did not appear to strongly influence mortality in the unadjusted (RR = 1.0, 95% CLs 0.7, 1.3) or adjusted (RR = 0.9, 95% CLs 0.7, 1.3) analyses. Before implementation of the protocol, baseline in-hospital mortality was 23.9% (95% CLs 20.7, 27.0) with an increase of 0.7% (95% CLs 0.1, 1.4) per quarter and after implementation of the protocol, there was a decrease in in-hospital mortality of 1.2% (95% CLs 7.5, 5.1) per quarter.
Discussion
In this retrospective cohort study of patients in the NCCU, we found that the implementation of a nurse-driven NCCU analgesia and sedation protocol resulted in increased use of analgesia, decreased use of sedation and decreased medication-associated costs, specifically propofol. There was an increase in the use of fentanyl in both primary and segmented time-series regression analyses after protocol implementation, which was expected since the protocol assesses pain prior to initiating sedation. Surprisingly, there was a decrease in associated fentanyl costs, which might be due to the protocol’s preferential use of intermittent intravenous injections over the use of continuous infusion fentanyl. The utilisation and cost associated with propofol decreased after the implementation of the protocol in the primary analysis, but that effect was not observed in the segmented time-series analysis. The use of midazolam did not change after the implementation of the protocol, which was expected since midazolam remained our second line sedative in the NCCU after the implementation of the protocol. In a primary analysis for health care utilisation and cost, we found that the protocol significantly decreased hospital days, and led to non-significant decreases in ventilator days and NCCU days. The protocol did not appear to influence total hospital costs, NCCU costs, or respiratory care costs, however, the segmented time-series regression sensitivity analysis did suggest that there was an increase in these cost outcomes, possibly unrelated to the protocol.
There are limited previous studies available that examined the effect of an analgesia and sedation protocol in the NCCU. The most comparable study to our own is a pilot study of an analgesia and sedation protocol done in patient with neurocritical illness in Denmark (20). That study demonstrated the feasibility of a similar protocol. After implementation of the protocol, they found an increase in the use of analgesia and a reduction in the use of sedation (20). The findings from our study were qualitatively similar for medication use and healthcare utilisation. However, our study differed from the one in Denmark and contributes new information in several important ways: (1) we examined medication and healthcare utilisation costs, (2) our study was done over a longer period of time, (3) we used RASS instead of Ramsey and FLACC instead of Pain Intensity Scale in which pain estimation was determined by changes in vital signs, facial expressions and behaviour, (4) we utilised mainly intermittent opioid dosing and only used continuous infusion opioid dosing if the patient required multiple intermittent doses, (5) our study population was six-fold larger, and (6) our study was conducted in the USA.
Research from the general critical care literature has shown that there is a decrease in mechanical ventilation and a decrease in ICU and hospital length of stay, improved sedation weaning and decreased time to extubation associated with the use of a sedation protocol (21–23). In one study, the use of a sedation protocol was also found to result in lower rates of ventilator associated pneumonia (24). A recent Cochrane review to assess the effect of protocol-directed sedation, found no difference in duration of mechanical ventilation, ICU length of stay or hospital length of stay which is similar to our findings (25).
Finding the optimal dose for sedation in patient who are neurologically ill while allowing frequent neurological exams is challenging (6,12). Using an analgesia-based approach via protocol can decrease the amount of sedation used and sedation wean time and improve outcomes, such as ICU and mechanical ventilation days (7–10). The results of our study imply that addressing pain first in the NCCU leads to decreased use and subsequent cost of sedation. Even though our study did not show an effect on mortality, total length of stay, or total healthcare costs, this is unsurprising because many factors other than the protocol could exert a strong influence on these outcomes. Nonetheless, our results suggest that this protocol may be useful to other NCCUs. It appears that our protocol didn’t cause any gross harm.
Our findings must be interpreted in light of several limitations. First, this was a single-centre study, which could potentially limit generalisability. Second, we used the FLACC scale instead of the Critical Care Pain Observation tool (CPOT) because it was standard at our institution during the study period. The Society of Critical Care Medicine recommends using a behavioural scale like CPOT to assess pain in patients with critical illness who are nonverbal (5). The CPOT scale was recently validated in patients in the NCCU after the study concluded and our protocol now incorporates the use of the CPOT (26). Third, the retrospective design of our study limited our ability to collect information on several covariates, including severity of illness, FLACC score and RASS score prior to and after implementation of the protocol. Such information would have been useful for better understanding the efficacy of the protocol. The retrospective design of our study also prevented us from identifying the specific vial and bag sizes of the medications used in the protocol. If the default institutional product size changed over time during the study period, that may have influenced our results. Fourth, secular time trends in our outcomes were potentially important consideration. Although our use of the interrupted time series accounts for such time trends, some residual bias could still exist (19). Fifth, our study does not examine the burden of the protocol on nursing labor and resources, however, many nurses anecdotally reported that they were satisfied with the protocol. Future work should aim to formally capture the effects of this protocol on nursing resources and satisfaction, as well as other clinical outcomes that our study was not designed to measure, such as cognition and quality of life.
Conclusion
Protocolised analgesia and sedation increased use of analgesia, decreased use of sedatives, and reduced medication-associated costs in the NCCU. Our results suggest that health systems should consider use of a population-specific protocol to manage sedation and analgesia in patients in the NCCU.
Supplementary Material
Footnotes
Declaration of interest
Authors of this manuscript have the following to disclose concerning possible financial or personal relationships with commercial entities that may have a direct or indirect interest in the subject matter of this presentation: Dr Zullo is funded in part by an award from the Agency for Healthcare Research and Quality (5K12HS022998).
Supplemental data for this article can be accessed here.
References
- 1.Jacobi J, Fraser GL, Coursin DB, Riker RR, Fontaine D, Wittbrodt ET, Chalfin DB, Masica MF, Bjerke HS, Coplin WM, et al. Clinical practice guidelines for the sustained use of sedatives and analgesics in the critically ill adult. Crit Care Med. 2002;30 (1):119–41. doi: 10.1097/00003246-200201000-00020. [DOI] [PubMed] [Google Scholar]
- 2.Makii JM, Mirski MA, Lewin JJ 3rd. Sedation and analgesia in critically ill neurologic patients. J Pharm Pract. 2010;23(5):455–69. doi: 10.1177/0897190010372339. [DOI] [PubMed] [Google Scholar]
- 3.Mirski MA, Hemstreet MK. Critical care sedation for neuroscience patients. J Neurol Sci. 2007;261(1–2):16–34. doi: 10.1016/j.jns.2007.04.028. [DOI] [PubMed] [Google Scholar]
- 4.Arroliga A, Frutos-Vivar F, Hall J, Esteban A, Apezteguia C, Soto L, Anzueto A. International mechanical ventilation study G. Use of sedatives and neuromuscular blockers in a cohort of patients receiving mechanical ventilation. Chest. 2005;128(2):496–506. doi: 10.1378/chest.128.2.496. [DOI] [PubMed] [Google Scholar]
- 5.Barr J, Fraser GL, Puntillo K, Ely EW, Gelinas C, Dasta JF, Davidson JE, Devlin JW, Kress JP, Joffe AM, et al. Clinical practice guidelines for the management of pain, agitation, and delirium in adult patients in the intensive care unit. Crit Care Med. 2013;41(1):263–306. doi: 10.1097/CCM.0b013e3182783b72. [DOI] [PubMed] [Google Scholar]
- 6.Oddo M, Crippa IA, Mehta S, Menon D, Payen JF, Taccone FS, Citerio G. Optimizing sedation in patients with acute brain injury. Crit Care. 2016;20(1):128. doi: 10.1186/s13054-016-1294-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Breen D, Karabinis A, Malbrain M, Morais R, Albrecht S, Jarnvig IL, Parkinson P, Kirkham AJ. Decreased duration of mechanical ventilation when comparing analgesia-based sedation using remifentanil with standard hypnotic-based sedation for up to 10 days in intensive care unit patients: a randomised trial [ISRCTN47583497]. Crit Care. 2005;9(3):R200–10. doi: 10.1186/cc3495. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Rozendaal FW, Spronk PE, Snellen FF, Schoen A, van Zanten AR, Foudraine NA, Mulder PG, Bakker J. Si Ulti. Remifentanil-propofol analgo-sedation shortens duration of ventilation and length of ICU stay compared to a conventional regimen: a centre randomised, cross-over, open-label study in the Netherlands. Intensive Care Med. 2009;35(2):291–98. doi: 10.1007/s00134-008-1328-9. [DOI] [PubMed] [Google Scholar]
- 9.Park G, Lane M, Rogers S, Bassett P. A comparison of hypnotic and analgesic based sedation in a general intensive care unit. Br J Anaesth. 2007;98(1):76–82. doi: 10.1093/bja/ael320. [DOI] [PubMed] [Google Scholar]
- 10.Strom T, Martinussen T, Toft P. A protocol of no sedation for critically ill patients receiving mechanical ventilation: a randomised trial. Lancet. 2010;375(9713):475–80. doi: 10.1016/S0140-6736(09)62072-9. [DOI] [PubMed] [Google Scholar]
- 11.Sessler CN, Pedram S. Protocolized and target-based sedation and analgesia in the ICU. Crit Care Clin. 2009;25(3):489–513, viii. doi: 10.1016/j.ccc.2009.03.001. [DOI] [PubMed] [Google Scholar]
- 12.Kurtz P, Fitts V, Sumer Z, Jalon H, Cooke J, Kvetan V, Mayer SA. How does care differ for neurological patients admitted to a neurocritical care unit versus a general ICU? Neurocrit Care. 2011;15(3):477–80. doi: 10.1007/s12028-011-9539-2. [DOI] [PubMed] [Google Scholar]
- 13.Voepel-Lewis T, Zanotti J, Dammeyer JA, Merkel S. Reliability and validity of the face, legs, activity, cry, consolability behavioral tool in assessing acute pain in critically ill patients. Am J Crit Care. 2010;19(1):55–61. quiz 2. doi: 10.4037/ajcc2010624. [DOI] [PubMed] [Google Scholar]
- 14.Ely EW, Truman B, Shintani A, Thomason JW, Wheeler AP, Gordon S, Francis J, Speroff T, Gautam S, Margolin R, et al. Monitoring sedation status over time in ICU patients: reliability and validity of the Richmond Agitation-Sedation Scale (RASS). JAMA: J Am Med Assoc. 2003;289(22):2983–91. doi: 10.1001/jama.289.22.2983. [DOI] [PubMed] [Google Scholar]
- 15.Vittinghoff E Regression methods in biostatistics: linear, logistic, survival, and repeated measures models. Vol. xx, 2nd. New York: Springer; 2012. 509 p. [Google Scholar]
- 16.Hernan MA, Hernandez-Diaz S, Werler MM, Mitchell AA. Causal knowledge as a prerequisite for confounding evaluation: an application to birth defects epidemiology. Am J Epidemiol. 2002;155 (2):176–84. doi: 10.1093/aje/155.2.176. [DOI] [PubMed] [Google Scholar]
- 17.Ramsay CR, Matowe L, Grilli R, Grimshaw JM, Thomas RE. Interrupted time series designs in health technology assessment: lessons from two systematic reviews of behavior change strategies. Int J Technol Assess Health Care. 2003;19(4):613–23. doi: 10.1017/S0266462303000576. [DOI] [PubMed] [Google Scholar]
- 18.Shojania KG, Grimshaw JM. Evidence-based quality improvement: the state of the science. Health Aff. 2005;24(1):138–50. doi: 10.1377/hlthaff.24.1.138. [DOI] [PubMed] [Google Scholar]
- 19.Wagner AK, Soumerai SB, Zhang F, Ross-Degnan D. Segmented regression analysis of interrupted time series studies in medication use research. J Clin Pharm Ther. 2002;27(4):299–309. doi: 10.1046/j.1365-2710.2002.00430.x. [DOI] [PubMed] [Google Scholar]
- 20.Egerod I, Jensen MB, Herling SF, Welling KL. Effect of an analog-sedation protocol for neurointensive patients: a two-phase interventional non-randomized pilot study. Crit Care. 2010;14(2):R71. doi: 10.1186/cc8978. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Arias-Rivera S, Sanchez-Sanchez Mdel M, Santos-Diaz R, Gallardo-Murillo J, Sanchez-Izquierdo R, Frutos-Vivar F, Ferguson ND, Esteban A. Effect of a nursing-implemented sedation protocol on weaning outcome. Crit Care Med. 2008;36(7):2054–60. doi: 10.1097/CCM.0b013e31817bfd60. [DOI] [PubMed] [Google Scholar]
- 22.De Jonghe B, Bastuji-Garin S, Fangio P, Lacherade JC, Jabot J, Appere-De-Vecchi C, Rocha N, Outin H. Sedation algorithm in critically ill patients without acute brain injury. Crit Care Med. 2005;33(1):120–27. doi: 10.1097/01.CCM.0000150268.04228.68. [DOI] [PubMed] [Google Scholar]
- 23.Brook AD, Ahrens TS, Schaiff R, Prentice D, Sherman G, Shannon W, Kollef MH. Effect of a nursing-implemented sedation protocol on the duration of mechanical ventilation. Crit Care Med. 1999;27 (12):2609–15. doi: 10.1097/00003246-199912000-00001. [DOI] [PubMed] [Google Scholar]
- 24.Quenot JP, Ladoire S, Devoucoux F, Doise JM, Cailliod R, Cunin N, Aube H, Blettery B, Charles PE. Effect of a nurse-implemented sedation protocol on the incidence of ventilator-associated pneumonia. Crit Care Med. 2007;35(9):2031–36. doi: 10.1097/01.ccm.0000282733.83089.4d. [DOI] [PubMed] [Google Scholar]
- 25.Aitken LM, Bucknall T, Kent B, Mitchell M, Burmeister E, Keogh SJ. Protocol-directed sedation versus non-protocol-directed sedation to reduce duration of mechanical ventilation in mechanically ventilated intensive care patients. The Cochrane Database of Syst Rev. 2015;1:CD009771. doi: 10.1002/14651858.CD009771.pub2. [DOI] [PubMed] [Google Scholar]
- 26.Joffe AM, McNulty B, Boitor M, Marsh R, Gelinas C. Validation of the critical-care pain observation tool in brain-injured critically ill adults. J Crit Care. 2016;36:76–80. doi: 10.1016/j.jcrc.2016.05.011. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.