BACKGROUND AND OBJECTIVE: Time to opioid administration (TTO) has been suggested as a quality of care measure for sickle cell disease patients with vaso-occlusive crisis (VOC). We sought to determine whether TTO was associated with outcomes of emergency department (ED) visits for VOC.
METHODS: We conducted a single-center retrospective cohort study of ED visits for VOC. The primary outcome was hospital admission, with secondary outcomes of change between the first 2 pain scores, area under the curve (AUC) for pain scores at 4 hours (pain score AUC), total ED length of stay, and total intravenous opioids. In both univariate and multivariate analyses, mixed regression (logistic for admission, linear for secondary outcome variables) was used to evaluate association of TTO with outcome.
RESULTS: In 177 subjects, 414 ED visits for VOC were identified. Inpatient admission occurred in 53% of visits. The median TTO for admitted patients was 86 minutes vs 87 minutes for those not admitted. TTO was not associated with inpatient admission in either univariate or multivariate analyses. In multivariate analyses with secondary outcomes, decreased TTO was associated with greater improvement between the first 2 pain scores, decreased pain score AUC, decreased total ED length of stay, and increased total opioids.
CONCLUSIONS: Although TTO was not associated with admission, it was independently associated with 4 important secondary outcomes: change in initial pain scores, pain score AUC, total ED length of stay, and total intravenous opioids. The association of a process measure, TTO, with these outcomes encourages the institution of TTO reduction efforts in the ED.
What’s Known on This Subject:
Patients with sickle cell disease frequently express dissatisfaction with emergency department treatment of painful crises. Time to opioid administration has been suggested as a quality of care measure for painful crises.
What This Study Adds:
Although not associated with hospital admission, time to opioid administration in sickle cell disease painful crises was associated with secondary outcomes including improvement between the first 2 pain scores, decreased pain score area under the curve at 4 hours, decreased emergency department length of stay, and increased total opioids.
The acute painful episode or vaso-occlusive crisis (VOC) is the clinical hallmark of sickle cell disease (SCD). VOC accounts for many of the 113 000 annual hospitalizations in the US and is the most common reason for hospitalization in both children and adults with SCD.1–3 These admissions result in high costs for the medical system and the individual patient.4 Children with SCD frequently have pain after discharge and accordingly miss multiple days of school, contributing to diminished academic achievement.4 Additionally, caregivers of SCD patients with VOC miss days of work, which may adversely affect the socioeconomic status of patient families. Unfortunately, the management of VOC, consisting primarily of opioids, anti-inflammatory medication, and intravenous hydration, has not changed in decades.5–8 Recent studies of novel agents for VOC treatment, including inhaled nitric oxide and purified poloxamer-188, demonstrated little or no clinical benefit.9,10
Although some centers in larger cities have established “day hospitals” for the specialized care of SCD patients with VOC,11 the vast majority of initial medical care occurs in hospital emergency departments (EDs). In both the ED and the inpatient setting, both patients and parents report a high level of dissatisfaction with VOC treatment, including delays in receiving treatment and undertreatment of pain.12,13 Patients with SCD and their families also rate their in-hospital care as of lower quality than race- and age-matched controls.14,15 In an effort to define and ultimately improve the quality of care (QOC) for children with SCD, novel QOC measures were recently proposed.16 These measures were evaluated for both the quality of supporting evidence and their perceived clinical importance. Among 41 proposed quality indicators, 4 received the highest importance rating of 9 on a 10-point scale. Two of these 4 indicators pertained to the treatment of VOC: (1) receipt of parenteral analgesic within 60 minutes of registration or 30 minutes of triage, and (2) initial pain assessment performed using an age-appropriate scale and repeated within 30 minutes of initial analgesic dosing. These measures are consistent with other measures in the ED wherein the timeliness of interventions is associated with improved QOC and outcomes.17 Unfortunately, both of these measures had the lowest level of supporting evidence (descriptive studies and/or expert opinion). Therefore, we aimed to study a single center’s SCD population with VOC to determine whether the time to opioid administration (TTO) was associated with outcomes of care in the ED. We hypothesized that increased TTO would be associated with worse patient outcomes, including increased likelihood of inpatient admission.
This retrospective cohort study was of TTO in children with VOC who received parenteral opioids in the ED of Children’s Medical Center Dallas. The Institutional Review Board of the University of Texas Southwestern Medical Center approved the study and waived the requirement for informed consent.
Subjects were identified for inclusion in the cohort by query of administrative records from our center. First, all ED encounters for SCD patients between January 1, 2008, and December 31, 2010, were identified by using International Classification of Diseases, Ninth Revision, Clinical Modification codes for SCD (282.41, 282.42, and 282.6x) as either primary or secondary diagnosis. To this list of encounters, the eligibility criteria were applied to create the final cohort. Inclusion criteria were (1) previously established diagnosis of SCD (any genotype); (2) presence of VOC defined as the new onset of pain for which there is no explanation other than vaso-occlusion1; (3) age ≥5 and ≤18 years; and (4) treatment with parenteral opioids. Children under 5 years of age were not included owing to concern that the differentiation of VOC from other sources of pain can be difficult in this age group. Exclusion criteria were (1) confounding sources of pain (eg, headache and cholelithiasis); (2) transfer to the ED from other medical centers; (3) any surgical procedure within 2 weeks of presentation; (4) simple transfusion in the preceding 30 days; and (5) participation in a chronic transfusion program within the preceding 6 months.
The primary outcome variable was hospital admission. Secondary outcomes included (1) change in first 2 recorded pain scores (defined as pain score 1 minus pain score 2); (2) area under the curve (AUC) for all pain scores at 4 hours (pain score AUC)18; (3) total ED length of stay in minutes; and (4) total intravenous opioid dose received in milligrams per kilogram of morphine equivalent.19 Our center routinely uses either a 5-point patient-reported numeric pain scale or the 5-point Faces pain scale for assessment of pain intensity. For pain score AUC, pain scores were graphed as a function of time for the first 4 hours of the ED visit. The pain score AUC serves as a quantitative representation of the total amount of pain experienced by the patient.18
TTO was defined as the time in minutes from presentation to ED (either triage or registration, whichever was first) to receipt of the first dose of parenteral opioids. TTO was analyzed in minutes, as both a continuous variable and categorized in quartiles.
Relevant covariates were collected, including patient age (analyzed as both a continuous and categorical variable); initial dose of opioid and total opioids received (both converted to morphine equivalents19); gender; presence of fever; baseline hemoglobin and baseline reticulocyte count (baseline defined as a rolling average of the past 3 values from sickle cell clinic visits as documented in our center’s sickle cell database); hemoglobin and reticulocyte count at presentation to the ED; primary payer; arrival by ambulance; number of previous admissions for VOC in the preceding 12 months; number of missed clinic visits in the preceding 12 months; pain location; number of pain locations; weekend presentation; and year of presentation. Additionally, our ED instituted the routine use of an electronic medical record (EMR) on May 29, 2009, nearly in the middle of our study frame, so we included a variable to indicate before or after EMR initiation. Because of the retrospective nature of this study, data summarizing the provision of therapies before the hospital visit were unavailable. Recommended outpatient VOC care in our center included oral hydration, local analgesia with applied heat (eg, heating pad), and a medication regimen consisting of alternating ibuprofen with an oral opioid (typically a combination of acetaminophen with codeine or hydrocodone).
Data were manually extracted from the EMR onto case report forms before entry into a Redcap database.20 To assess the accuracy of data entry by the first author, the senior author reviewed 20 charts independently. The agreement between authors was 97%, so no further double chart review was performed. At study completion, the senior author examined the data for extreme or missing values and inconsistencies in static variables (ie, gender) for patients with multiple hospitalizations.
The cohort was described with summary statistics. Univariate mixed regression models were used to test for association between the predictors and primary and secondary outcome variables. Variables that were statistically significant in the univariate regression analyses were subsequently evaluated in a multivariate mixed model (logistic for admission; linear for all secondary outcome variables). The final multivariate model was determined by using backward stepwise variable selection in addition to clinical consideration. The mixed models included patient random effects to account for multiple ED visits from certain subjects. Missing data were excluded from the analysis. All analyses were performed by using SAS 9.2 (SAS Institute, Cary, NC).
Descriptive Characteristics of Cohort
SCD patients made 2864 ED visits within the 3-year study frame. After ineligible visits were excluded, the final cohort consisted of 414 visits from 177 patients (Fig 1). The median number of ED visits was 2 (range 1–18). The median age for patients was 13.4 years, and 47% were female (Table 1). The majority of patients had hemoglobin SS disease (sickle cell anemia, 63%) and health insurance coverage by Medicaid (67%). No data were missing for the primary outcome or predictor variables. Three patients were missing the secondary outcomes of pain score AUC and change in pain scores 1 and 2. The covariates missing data were hemoglobin (n = 3) and reticulocytes (n = 7) in the ED.
Beginning with patient registration, the process of opioid administration was widely variable in duration and included a rapid triage (median duration 2.5 minutes) and rooming process (median duration 12 minutes) (Table 1). After patients were given rooms, the median duration to opioid order entry by a provider was 42 minutes. Opioid administration occurred at a median of 30 minutes after order entry. The median total TTO from registration was 86 minutes (range 10–405). The median total time in the ED was 390 minutes (range 80–1545).
Inpatient admission occurred in 53% (n = 219) of ED visits for VOC. The median TTO for admitted patients was 86 minutes vs 87 minutes for those not admitted. In univariate regression, TTO was not associated with admission (TTO quartile 1 [≤55 minutes] versus quartile 4 [≥131 minutes, referent], odds ratio 1.46, 95% confidence interval [CI] 0.82 to 2.59). In the multivariate analysis, the only predictor variables associated with increasing likelihood of admission were older age, larger number of VOC admissions for the previous year, and pain location in the chest (Table 2).
Change in First 2 Pain Scores
The mean improvement in the first 2 pain scores was 0.65 (5-point scale, SD 1.47). Three subjects had a first pain score without a second documented pain score and were excluded from the analysis. The median duration between the first 2 pain scores was 61 minutes. In univariate regression, decreased TTO was associated with greater improvements between the first 2 pain scores (TTO quartile 1 [≤55 minutes] versus quartile 4 [≥131 minutes, referent], β coefficient 0.77, 95% CI 0.33 to 1.21). In multivariate analysis, decreased TTO was again associated with greater improvements in pain scores 1 and 2, as were older age and higher first pain score (Table 3). These findings remained after adjusting for admission status.
Pain Score AUC
The median pain score AUC was 795 (interquartile range [IQR] 559–967). In univariate regression, decreased TTO was associated with decreased pain score AUC (TTO quartile 1 [≤55 minutes] versus quartile 4 [≥131 minutes, referent], β coefficient −87.0, 95% CI −161.0 to −13.0). In multivariate analysis, decreased TTO was again associated with decreased pain score AUC, as were lower age and lower first pain score (Table 3). The findings remained after adjusting for admission status.
Total ED Length of Stay
The median total ED length of stay was 390 minutes (IQR 302–525). In univariate regression, decreased TTO was associated with decreased total ED length of stay (TTO quartile 1 [≤55 minutes] versus quartile 4 [≥131 minutes, referent], β coefficient −121.2, 95% CI −181.6 to −60.7). In multivariate analysis, decreased TTO was again associated decreased total ED length of stay, as were discharge from the ED (as opposed to inpatient admission), decreased interval between pain scores 1 and 2, and presentation after EMR implementation (Table 4).
Total Dose of Parenteral Opioid
The median total dose of parenteral opioid was 0.18 mg/kg morphine equivalents (IQR 0.10–0.29). In univariate regression, decreased TTO was associated with increased total dose of parenteral opioid (TTO quartile 1 [≤55 minutes] versus quartile 4 [≥131 minutes, referent], β coefficient 0.042, 95% CI 0.009 to 0.076). In multivariate analysis, decreased TTO was again associated with an increased total dose of parenteral opioids, as were larger number of VOC admissions for the previous year, larger number of pain locations, and admission status (Table 4).
In a pediatric SCD population with VOC, we observed wide variability in the time between arrival and administration of standard treatment, intravenous opioids. Decreased TTO was not associated with admission but was associated with greater improvements in the first 2 pain scores, improved total pain over the first 4 hours of an ED visit, and decreased total ED length of stay. TTO was also associated with the receipt of increased total intravenous opioids, which may be attributable to the observation that those in the lowest quartile of TTO had the highest initial pain scores.
TTO has been described in SCD populations in previous studies, but whether it is associated with outcomes has not been addressed. One previous study has described TTO in adult SCD patients and found a similar median of 90 minutes.7 In that study, predictors of prolonged TTO included lower triage priority level and female gender. Previous reports have also examined disparities in time to treatment in the ED for SCD patients compared with other painful conditions. Interestingly, discrepant observations were made between adults and children. In pediatric patients, a recent case-control study of 152 cases of VOC demonstrated no differences in TTO (defined as time from first pain score in triage to administration of first analgesia) compared with race-matched patients with long-bone fracture.21 In that study, the mean TTO was 68 minutes, which is comparable to our observations. By contrast, a retrospective cohort study comparing adults with VOC to those with renal colic found a mean TTO of 80 minutes for SCD compared with 50 minutes for renal colic.22 Additionally, a cross-sectional analysis of the National Hospital Ambulatory Medical Care Survey found that adult SCD patients (mean age 27.6 years) with VOC experienced longer wait times to see a physician compared with both the general population and patients with long-bone fracture, although TTO was not compared between groups.23 Treatment delays for VOC may be deleterious due to underlying pain biology. When noxious stimuli are presented to peripheral sensory neurons, expression of c-fos is increased in neurons in the dorsal horn of the spinal cord, which can lead to enhanced pain responses to subsequent or ongoing stimuli.24 Hence, delays in provision of analgesia may worsen a VOC episode.7 The likelihood of the development of chronic pain may increase as a result.
Two prevailing theoretical frameworks for QOC include Donabedian’s model of structure-process-outcome25 and the Institute of Medicine’s 6 domains of quality including safety, effectiveness, timeliness, patient-centeredness, efficiency, and equitability.26 TTO is classified as a process measure in the Donabedian model and reflects the timeliness, patient-centeredness, and efficiency domains from the Institute of Medicine model. The appeal of many broadly applied process measures (such as the time to intervention in sepsis) is the association of process with outcomes. To that end, this study sought to broaden the obvious appeal of TTO by identifying associations of TTO with outcomes of care for VOC patients in the ED. Although TTO was not associated with this study’s primary outcome (hospital admission), the association of TTO with our secondary outcomes bolsters its validity as a process measure. In addition, TTO’s association with outcomes of ED visits suggests that TTO reflects the effectiveness of care for VOC in addition to the previously mentioned domains of QOC.
The predictors for VOC admission identified in this study (increased number of previous admissions, increasing age, and pain location in the chest) have not been previously reported. It has been reported that VOCs become more numerous and more intense as children age, so the finding of increasing age is not surprising.8 Similarly, a history of frequent VOCs is associated with subsequent VOCs, which likely explains the finding of increased admissions in the past 12 months in association with admission.27 In addition, chest pain may increase the risk for admission due to the concern for impending acute chest syndrome. In 1995, Frusch et al28 examined children with SCD presenting to the Duke University pediatric ED to identify predictors of admission. They found that a shorter duration of pain before presentation in afebrile patients was associated with increased likelihood of admission and that there was no correlation between pain location and admission.28 In this retrospective study, we were unable to reliably assess the duration of pain and, therefore, cannot evaluate its importance relative to our identified predictors.
Implementation of an EMR has been associated with variable impacts on the timeliness of care in previous studies.29,30 Frequently, the timeliness of care is transiently increased before returning to pre-implementation levels.31 In this study, we found an independent association of post-EMR implementation with decreased total ED length of stay. Implementation of the EMR was concomitant with ongoing efforts in our ED to improve throughput and decrease total length of stay, which may serve as a confounder to EMR implementation and may better explain the observation of decreased length of stay.
Our study is limited by the single-center design and the inherent imperfections in the medical record, including potential inaccurate recordings. In addition, retrospective cohorts are prone to selection bias. To circumvent this problem, our subject identification intentionally lacked specificity for VOC and included all ED visits for SCD patients. Another limitation is the potential for error in retrospectively identifying patients with VOC, which is a clinically defined syndrome and involves the subjective patient experience of pain. Additionally, our study frame spanned the introduction of an EMR. We observed that charted times throughout the record were more precise with electronic charting. Although pain scores are inherently subjective as measures of pain intensity, they are the best method currently available. Another potential limitation is the overrepresentation of a small subset of clinically severe patients in the data set. Our statistical modeling approach attempts to mitigate this problem but is imperfect.
TTO is independently associated with important outcomes of ED visits for children with SCD, including improved pain scores, shorter length of stay, and increased opioid delivery. Due to the severe nature of the pain in VOC, efforts to reduce TTO are of paramount importance. To that end, we now track TTO in our center and are planning a quality improvement initiative to aid in early identification of patients with VOC to achieve a goal TTO of <60 minutes. In addition, research is needed to (1) evaluate TTO in other populations, (2) further assess reasons for prolonged TTO, and (3) identify effective practices to reduce TTO. Finally, we propose that TTO should be included in the quality “dashboard” for both EDs and day hospitals involved in urgent care for children with SCD.
The authors thank Dr George R. Buchanan for his thoughtful review of the manuscript, Dr Song Zhang for his review of the statistical plan and analyses, and Ms Hailey Hodgkiss for her help in data collection.
- Accepted December 17, 2014.
- Address correspondence to Timothy L. McCavit, MD, U.T. Southwestern Medical Center, 5323 Harry Hines Blvd, Dallas, TX 75390-9063. E-mail:
Dr Mathias designed the data collection instrument, collected the data, and drafted the initial manuscript; Dr McCavit conceptualized and designed the study, collected and analyzed the data, and drafted and revised the manuscript; and both authors approved the final manuscript as submitted.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: Funded by CTSA grant UL1-RR024982.
POTENTIAL CONFLICT OF INTEREST: Dr McCavit has consulted for Glycomimetics and Pfizer in the developmental of a novel therapy for sickle cell pain; Dr Mathias has indicated she has no potential conflicts of interest to disclose.
- ↵Steiner C, Miller J. Sickle Cell Disease Patients in U.S. Hospitals, 2004: Statistical Brief #21. Healthcare Cost and Utilization Project, Rockville, MD. Agency for Healthcare Policy and Research; 2006:1–9
- Solomon LR
- ↵Tanabe P, Myers R, Zosel A, et al. Emergency department management of acute pain episodes in sickle cell disease. Acad Emerg Med. 2007;14(5):419–425
- Benjamin LJ,
- Swinson GI,
- Nagel RL
- Tanabe P,
- Dias N,
- Gorman L
- ↵Glassberg JA, Tanabe P, Chow A, et al. Emergency provider analgesic practices and attitudes toward patients with sickle cell disease. Ann Emerg Med. 2013;62(4):293–302.e210
- Murray N,
- May A
- Maxwell K,
- Streetly A,
- Bevan D
- Wang CJ,
- Kavanagh PL,
- Little AA,
- Holliman JB,
- Sprinz PG
- Smith JE,
- Rockett M,
- Squire R,
- et al
- ↵Institute of Medicine (U.S.); Committee on Quality of Health Care in America. Crossing the Quality Chasm: a New Health System for the 21st Century. Washington, D.C.: National Academy Press; 2001
- Quinn CT,
- Shull EP,
- Ahmad N,
- Lee NJ,
- Rogers ZR,
- Buchanan GR
- Connelly DP,
- Park YT,
- Du J,
- et al
- Copyright © 2015 by the American Academy of Pediatrics