Associations Between Pain Scores and Opioid Doses With Emergency Department Disposition and Return Visit Rates in Children With Sickle Cell Disease

Abstract

Rapid treatment and frequent reassessment of pain are key components of treatment guidelines for acute sickle cell disease (SCD) pain. Few studies, however, report the associations between emergency department (ED) pain scores, number of ED opioid doses, receipt of an opioid prescription, ED visit disposition, or ED return visits. This seven-site retrospective cohort study analyzed 4983 ED visits by children with SCD pain using electronic health record data from the Pediatric Emergency Care Applied Research Network Registry. ED pain scores included initial, last, and change in scores (initial minus last), measured on a 0–10 scale. Dispositions of discharge and hospital admission were included. Modified Poisson regression and the Cochran–Armitage test of trend were used for analysis. The median (IQR) initial pain score was 8.0 (6–10); last pain score was 5.0 (2–8); and median decrease was 2.0 (0–5). In multivariable analysis, last pain score was the best predictor of disposition. For the return visit analyses, of the 2377 visits discharged at index ED visit, 29% returned within 14 days. Higher initial and last ED pain scores were associated with increased return visits. Children with no opioid discharge prescription and ≥3 ED opioid doses had a return visit rate of 36% compared to 22% if the child received an opioid prescription and only one ED opioid. Increasing discharge opioid prescriptions and targeting interventions for those who receive multiple ED opioid doses could decrease return visits.

1 |. Introduction

Approximately 100,000 people, of whom 36,000 are children, in the United States live with sickle cell disease (SCD) [1, 2]. SCD, an inherited group of hemoglobinopathies, is characterized by hemolysis and intermittent obstruction of vascular blood flow that can cause debilitating, severe vaso-occlusive acute pain events [3, 4], inflammation, and multiorgan damage all due to the polymerization of an abnormal hemoglobin [5, 6].

Children with SCD often present to the emergency department (ED) for severe pain treatment and visit the ED more frequently than patients with other chronic diseases [7, 8]. Among children with SCD, acute pain events are the most common reason for ED visits, hospitalization, and rehospitalization [8, 9]. National guidelines for moderate to severe acute SCD pain treatment recommend that patients receive prompt and timely administration of parenteral opioid pain medication less than 60 min from ED arrival [10, 11]. Further, serial pain reassessments should be completed every 15–30 min in the ED, with subsequent parenteral opioid doses administered until acute SCD pain is alleviated [10].

In the ED, pain assessments are often used to evaluate a child’s pain severity, document acute pain, guide further treatment, and monitor the effectiveness of the pain interventions [12, 13]. Although acute SCD pain scores play an important role in ED care, there are limited data on whether acute pain scores are associated with visit disposition and ED return visits. Our research team recently demonstrated a 29% ED return visit rate within 14 days of ED discharge following an uncomplicated SCD acute pain visit [14]. The relationship of pain scores to the number of opioid doses received prior to discharge, prescription at discharge, and return visit rates is unknown and vital to understand to improve SCD pain care and decrease the high return visit rates.

The objectives of our multicenter study of children with SCD acute pain were (i) to determine the relationship among initial pain scores, last pain score, change in pain scores, and disposition of ED visits; and (ii) to determine the association among pain scores, the number of opioid doses given in the ED, opioid prescription received at discharge, and ED return visits within 14 days.

2 |. Methods

2.1 |. Data Source

The Pediatric Emergency Care Applied Research Network (PECARN) Registry captured all ED visit electronic health record data from seven children’s hospital EDs (Ann & Robert H. Lurie Children’s of Chicago, Children’s Hospital of Philadelphia, Cincinnati Children’s Hospital Medical Center, Children’s National Medical Center, Children’s Colorado, Children’s Wisconsin, and Nationwide Children’s) [15].

2.2 |. Study Design

This is a multicenter retrospective cohort investigation of ED disposition and return visit rates for children with acute uncomplicated SCD pain events. Index visits included all uncomplicated acute pain events, defined using previously validated criteria [16], between January 2017 and November 2021 to allow for analysis of return visits by the end of the calendar year. Inclusion criteria included (i) administration of at least one parenteral opioid in the ED (intranasal fentanyl or any intravenous opioid); and (ii) at least one of the subsequent items: a primary billing diagnosis for SCD (ICD-10-CM = D57, except sickle cell trait), a chief complaint of sickle cell pain event, or a chief complaint of sickle cell fever. ED visits were excluded if secondary complications were present such as a diagnosis of acute chest, priapism, splenic sequestration, documented fever in ED greater than 38.5°C, the ED visit had less than two pain scores recorded, or the patient had a prior ED visit within the previous 14 days [14, 16]. Visits that were not discharged home from the ED were excluded from the return visit analyses. From the Registry, we also extracted pain scores (initial, last, and the change in scores), number of opioid doses given, and receipt of an opioid prescription at discharge.

2.3 |. Statistical Methods/Outcomes

For analyses related to initial disposition, we modeled the relationship between ED pain scores and hospital disposition, adjusting for age, sex, and the overall SCD hospitalization rate during the study period. Discharge prescription was not included in the model as it only applied to discharged patients. A variety of pain scales were used across the PECARN sites: 37% used the Numeric Rating Scale (NRS) [17], 25% used the Visual Analog Scale (VAS) [18], 15% used the Face, Legs, Activity, Cry, and Consolability (FLACC) [19], and 5% used the Bieri Faces [20]. Pain intensity scales were converted to 0–10 with consultation from PECARN principal investigators for statistical analysis. Multivariable analyses for disposition included either first pain score and change in pain score or last pain score and change in pain score, as including all three would violate model assumptions given that change in pain score is the linear combination of first and last pain scores. For return visit analyses, we included the number of parenteral opioid doses given and receipt of an opioid prescription, adjusting for the same covariates. For all analyses, adjustment for site was made using generalized estimating equations. Modified Poisson was used to model the hospital admission and return visit rates. Due to significant multicollinearity between ED pain scores and the number of parenteral opioid doses, multivariable modeling of return visits did not include ED pain scores. The Cochran–Armitage test for trend was performed to test for association between initial pain score category, discharge pain score category, and return visit rates. All statistical analyses were conducted using the Statistical Analysis Software (SAS)/STAT software version 9.4 (SAS Institute Inc., Cary, NC, USA).

This study was approved by the institutional review board at all the sites and the data coordinating center.

3 |. Results

A total of 4983 index ED visits for acute uncomplicated SCD pain were included. Of these index ED visits, 52% resulted in hospitalization. Approximately 61% of the ED visits were by children 12 to < 19 years old, 54% of the participants were female, and 96% were non-Hispanic Black. The number of eligible initial ED visits per site varied from 157 to 1853 during the study period. The overall initial ED median (interquartile range [IQR]) acute pain score was 8 (6–10). The overall last acute pain score was 5 (2–8), and the median decrease in ED acute pain scores was 2 (0–5). The study flow diagram is shown in Figure 1; 2377 (48%) were discharged home and eligible for return visit analyses.

FIGURE 1 |.

Study flow diagram. 1: Primary billing diagnosis for SCD (ICD-10 = D57,* except sickle cell trait) or a chief complaint of sickle cell pain event or sickle cell fever and received at least one parenteral opioid pain medication in the ED. Visits from January 2017 through November 2021. 2: Visits could be ineligible due to more than one exclusion criteria. 3: Left against medical advice, left without being seen, or missing disposition.

Table 1 shows the adjusted rate ratio of the association between hospital admission rate and age, sex, and pain scores. Ages 12 to < 19 years (1.15, 95% confidence interval [CI]: 1.07–1.23) and higher last ED pain score (1.15, 95% CI: 1.12–1.18) were associated with higher rates of hospital admission. In unadjusted analyses, higher initial and higher last pain scores and a smaller decrease in pain scores were all associated with higher hospitalization rates. When analyzing models that included combinations of pain score and decrease in pain score, the last pain score was the best predictor of disposition, with the decrease in pain score no longer significant. Hospitalization rate increased as the last ED acute pain score increased. Less than 30% of children with last ED acute pain scores 0–4 were admitted, in contrast to 50%–60% of children admitted if last ED pain score was 5–6, and greater than 80% admitted if last ED pain score was 7 or higher.

TABLE 1 |.

Hospital admission rate.

	Admitted		Unadjusted rate ratio (95% CI)	Adjusted rate ratio (95% CI)
	No (N = 2383)	Yes (N = 2600)
Age category
<12 years	1188 (61.4%)	746 (38.6%)	Reference	Reference
12–19 years	1195 (39.2%)	1854 (60.8%)	1.53 (1.43, 1.64)	1.15 (1.07, 1.23)
Sex
Female	1237 (45.7%)	1472 (54.3%)	Reference	Reference
Male	1146 (50.4%)	1128 (49.6%)	0.90 (0.83, 0.97)	1.01 (0.97, 1.05)
Initial ED pain score (0–10)
Median (IQR)	8.0 (6.0–9.0)	9.0 (7.0–10.0)	1.10 (1.07, 1.13)	a
Pain score at ED discharge (0–10)
Median (IQR)	2.0 (0.0–5.0)	7.0 (5.0–8.0)	1.17 (1.14, 1.21)	1.15 (1.12, 1.18)
Decrease in ED pain score (initial minus discharge pain score)
Median (IQR)	4.0 (2.0–6.0)	1.0 (0.0–3.0)	0.88 (0.86, 0.90)	0.98 (0.96, 1.00)
Initial SCD site hospitalization rate (change in return visit rate per 10% increase in SCD hospitalization rate)			1.20 (1.20, 1.21)	0.74 (0.61, 0.91)

Note: Correlation within site was accounted for using generalized estimating equation methods in each model. Subset to complete cases.

Abbreviations: CI, confidence interval; ED, emergency department; IQR, interquartile range; SCD, sickle cell disease.

^aNot included in the model.

In the return visit analysis, not receiving a discharge prescription for an opioid and receiving ≥ doses of parenteral opioid during the ED visit were independently associated with increased return visit rates (Table 2). Figure 2 shows the relationship between a 14-day ED return visit for SCD acute pain and number of ED parenteral opioid doses and receipt of an opioid prescription at discharge. The majority of the patients (52.2%) received three or more doses of parenteral opioid in the ED. Not receiving an opioid prescription at discharge and receiving three or more doses of parenteral opioid in the ED corresponded to a 36% ED return visit rate within 14 days, while an opioid prescription provided during discharge and one ED parenteral opioid dose corresponded to a 22% return visit rate (Figure 2). The return visit rate within 14 days was 25.2%, 26.6%, and 31.0% for children with initial ED pain scores of 0–4, 5–6, and 7–10, respectively (Cochran–Armitage test, p = 0.01). The return visit rate within 14 days was 27.8%, 33.3%, and 33.5% for children with last ED pain scores of 0–4, 5–6, and 7–10, respectively (Cochran–Armitage test, p = 0.01).

TABLE 2 |.

Return visit rate.

	Return visit within 14 days		Unadjusted rate ratio (95% CI)	Adjusted rate ratio (95% CI)
	No (N = 1681)	Yes (N = 696)
Age category
<12 years	867 (73.1%)	319 (26.9%)	Reference	Reference
12–19 years	814 (68.3)	377 (31.7%)	1.18 (1.07, 1.29)	1.12 (1.01, 1.24)
Sex
Female	875 (70.9%)	360 (29.1%)	Reference	Reference
Male	806 (70.6%)	336 (29.4%)	1.01 (0.92, 1.10)	1.02 (0.93, 1.11)
Number of parenteral opioid doses administered in the ED
1	678 (74.8%)	228 (25.2%)	Reference	Reference
2	548 (70.4%)	230 (29.6%)	1.17 (1.02, 1.35)	1.17 (1.02, 1.34)
3+	455 (65.7%)	238 (34.3%)	1.36 (1.22, 1.51)	1.35 (1.20, 1.51)
Received opioid prescription at ED discharge
No	787 (68.9%)	356 (31.1%)	Reference	Reference
Yes	894 (72.4%)	340 (27.6%)	0.88 (0.85, 0.92)	0.86 (0.82, 0.91)
Initial SCD site hospitalization rate (change in return visit rate per 10% increase in SCD hospitalization rate)			1.00 (0.95, 1.05)	1.03 (0.98, 1.09)

Note: Correlation within site was accounted for using generalized estimating equation methods in each model.

Abbreviations: CI, confidence interval; ED, emergency department; IQR, interquartile range; SCD, sickle cell disease.

FIGURE 2 |.

Association between ED opioid doses administered/receipt of an opioid prescription at discharge and 14-day ED return visit rates.

4 |. Discussion

Limited data are available about the relationship between acute pain scores and hospital disposition for children with SCD. Our study revealed that for children presenting to the ED for uncomplicated acute SCD pain, the last ED pain score is a better predictor of hospitalization than the change in pain score or initial pain score. Children with SCD presented with a median pain score of 8, had a median decrease in pain score of 2, and a median last pain score of 5. This initial pain score of 8, indicating severe initial pain, is consistent with previous studies of SCD acute pain events [21–24]. Children who had a lower average pain score or had a larger decrease in average pain scores in the ED are more likely to be discharged [25]. Portugal et al. [24] noted that the average pain score for discharged patients was 5.1. Whereas, Myrvik et al. reported a median pain score at discharge of 6 (IQR: 4–8.5) [22]. Our data are congruent with the pain scores reported by other researchers.

Pain scores are improved in children with SCD by the timeliness of the opioid doses [26]. Further analysis using multivariable models revealed that the last pain score is most associated with admission. More than 85% of children with a last pain score of 7 or higher were hospitalized, compared to only 52% of all visits analyzed. In contrast, in children with pain scores of 4 or lower, 23% were hospitalized in our study. While higher pain scores would logically be associated with higher hospitalization rates, the reasons why some children with high pain scores are not hospitalized and some with low pain scores are hospitalized are unknown. The pain score–hospitalization relationship could be influenced by issues such as social or environmental factors (e.g., transportation issues, lack of resources), patient/parent preferences, and differences in pain thresholds.

Our finding of increased admission rate with increased age is consistent with the previous literature [26, 27], and we showed that relationship continues to exist even after adjusting for pain scores.

Previous research by our study team, Coleman, et al., demonstrated a high ED return visit rate of 29% within 14 days of initial ED visit for children presenting with an uncomplicated SCD acute pain event [14].

In our current study, we found that return visit rates were higher among patients with higher initial and last ED pain scores. Rees et al., in an adjusted model, noted that children with lower mean pain scores and a greater decrease in pain scores in the ED were more likely to be discharged [25]. Furthermore, we report receiving more opioid doses in the initial ED visit and not receiving a prescription for an opioid at ED discharge were both associated with increased return visits. A single-center study aimed to determine if a prescription of scheduled opioids for children with SCD pain was associated with a lower 30-day revisit rate, and these data did not support an association between the overall 30-day return visit rate and taking scheduled opioids after discharge from the ED or hospital [28]. The investigators also noted that the ED and hospital have different opioid prescription practices, which, based on our study, may be a potential intervention to decrease return visits. Rees et al. reported that lower total morphine equivalence was associated with an increased likelihood of being discharged from the ED [25]. One possible strategy to decrease high return visit rates to the ED includes coordinating an outpatient follow-up visit [29] with a hematologist. Other potential considerations in SCD care include improvements in discharge instructions [30] and confirming outpatient medication availability. Our results suggest an opioid prescription at discharge may help prevent return visits for acute SCD pain.

Our findings may play a role in future care optimization to prevent ED return visits. We identified important factors (e.g., receipt of multiple doses of an opioid in the ED, older age, lack of an opioid prescription) that lead to an increased likelihood of ED return visits. These data assist in identifying high-risk groups and in the development of strategies and interventions to optimize pain treatment, champion clinical decision-making, and improve care, which could decrease ED return visits. Our work may be particularly impactful for the development and impact assessment of outpatient guidelines for acute SCD pain management following an ED visit for acute pain events. The lack of such evidence-based guidelines leads to varied prescriber practices. Further, improving SCD care and decreasing ED return visit rates for children with SCD is a priority, as evidenced by the National Association of Children’s Hospitals and Related Institutions (NACHRI) which set hospital readmission within-30 days as a quality-of-care standard for patients with SCD [31].

5 |. Limitations

There are some possible limitations to the study. The retrospective study uses Registry data collected from electronic health records across seven ED sites, and while certainly robust, there is potential for missing or incomplete data. ED visits with fewer than two pain scores were excluded from our analysis, as an outcome of interest included a change in pain score. Pain reassessment may present a challenge in the fast-paced ED setting. Further, unfortunately, historically, implicit bias has been well-described in patients with SCD and other ED patients may have been prioritized for ED care reassessment. We did not control for clustering within the individual in regression modeling. As a result, multiple visits from the same patient may have been included as independent observations. Data from repeated visits from the same patient may be correlated, which may affect parameter estimates.

6 |. Conclusion

In summary, for initial visit disposition, while there are clear associations between initial pain score and disposition rates, the last ED pain score is a better indicator of hospitalization following an SCD uncomplicated acute pain event. For return visits, the lack of an opioid prescription at discharge and receiving ≥ ED opioid doses were associated with increased ED return visits. Our data support that higher acute SCD pain scores in the ED are associated with increased rates of ED return visits within 14 days. More research is needed to determine why children return to the ED following treatment for SCD acute pain events to improve patient care, reduce burden on families, and optimize healthcare costs.

Abbreviations:

ED

emergency department

FLACC

Face, Legs, Activity, Cry, and Consolability

IQR

interquartile range

NACHRI

National Association of Children’s Hospitals and Related Institutions

NRS

Numeric Rating Scale

PECARN

Pediatric Emergency Care Applied Research Network

SAS

Statistical Analysis Software

SCD

sickle cell disease

VAS

Visual Analog Scale

Data Availability Statement

Data supporting this study are housed at the Pediatric Emergency Care Applied Research Network (PECARN) Emergency Medical Services for Children (EMSC) Data Center (EDC) at the University of Utah. Use of these data is governed by institutional review boards (IRBs) and data use agreements at each site. Data will be made available at the PECARN Public Use Data Set website following publication of the main paper from the parent study.