Adult Emergency Department Patients with Sickle Cell Pain Crisis: A Learning Collaborative Model to Improve Analgesic Management
Presented at the Sickle Cell Disease Association of America meeting, New Orleans, LA, September 2008; American College of Emergency Physicians meeting, Chicago, IL, October 2008; and the American Pain Society, San Diego, CA, May 2009.
This project was supported by a research grant from the Mayday Fund. Dr. Tanabe also received support from the NIH, K23NR010940.
Address for correspondence and reprints: Paula Tanabe, PhD, MPH; e-mail: email@example.com.
Objectives: The objectives were to report the baseline (prior to quality improvement interventions) patient and visit characteristics and analgesic management practices for each site participating in an emergency department (ED) sickle cell learning collaborative.
Methods: A prospective, multisite longitudinal cohort study in the context of a learning-collaborative model was performed in three midwestern EDs. Each site formed a multidisciplinary team charged with improving analgesic management for patients with sickle cell disease (SCD). Each team developed a nurse-initiated analgesic protocol for SCD patients (implemented after a baseline data collection period of 3.5 months at one site and 10 months at the other two sites). All sites prospectively enrolled adults with an acute pain crisis and SCD. All medical records for patients meeting study criteria were reviewed. Demographic, health services, and analgesic management data were abstracted, including ED visit frequency data, ED disposition, arrival and discharge pain score, and name and route of initial analgesic administered. Ten interviews per quarter per site were conducted with patients within 14 days of their ED discharge, and subjects were queried about the highest level of pain acceptable at discharge. The primary outcome variable was the time to initial analgesic administration. Variable data were described as means and standard deviations (SDs) or medians and interquartile ranges (IQR) for nonnormal data.
Results: A total of 155 patients met study criteria (median age = 32 years, IQR = 24–40 years) with a total of 701 ED visits. Eighty-six interviews were conducted. Most patients (71.6%) had between one and three visits to the ED during the study period. However, after removing Site 3 from the analysis because of the short data enrollment period (3.5 months), which influenced the mean number of visits for the entire cohort, 52% of patients had between one and three ED visits over 10 months, 21% had four to nine visits, and 27% had between 10 and 67 visits. Fifty-nine percent of patients were discharged home. The median time to initial analgesic for the cohort was 74 minutes (IQR = 48–135 minutes). Differences between choice of analgesic agent and route selected were evident between sites. For the cohort, 680 initial analgesic doses were given (morphine sulfate, 42%; hydromorphone, 46%; meperidine, 4%; morphine sulfate and ibuprofen or ketorolac, 7%) using the following routes: oral (2%), intravenous (67%), subcutaneous (3%), and intramuscular (28%). Patients reported a significantly lower targeted discharge pain score (mean ± SD = 4.19 ± 1.18) compared to the actual documented discharge pain score within 45 minutes of discharge (mean ± SD = 5.77 ± 2.45; mean difference = 1.58, 95% confidence interval = .723 to 2.44, n = 43).
Conclusions: While half of the patients had one to three ED visits during the study period, many patients had more frequent visits. Delays to receiving an initial analgesic were common, and post-ED interviews reveal that sickle cell pain patients are discharged from the ED with higher pain scores than what they perceive as desirable.
ACADEMIC EMERGENCY MEDICINE 2010; 17:399–407 © 2010 by the Society for Academic Emergency Medicine
Approximately 70,000 Americans have sickle cell disease (SCD), which occurs in one of every 500 African American births.1 SCD and associated complications represent a serious disease process that often requires management in the emergency department (ED). Moreover, previous studies have demonstrated an association of acute pain syndromes with other complications of SCD, including death.2 National pain management guidelines and the need to rule out life-threatening comorbidities, including sepsis, stroke, and acute cardiac syndrome, suggest the need for rapid physician evaluation of patients who present to the ED with complaints related to SCD.3 Despite these recommendations, significant variability exists in the treatment of sickle cell pain crises in the ED, often resulting in a delay to pain medication administration.4
Data from sickle cell day hospitals demonstrate success in avoiding inpatient hospitalization for patients with sickle cell pain crisis when rapid, aggressive, analgesic management practices are implemented.5 Day hospitals are specialized units within a hospital that dedicate a small number of beds to caring for patients with SCD. The day hospital model allows patients with acute pain episodes to be rapidly evaluated and treated by specially trained staff, thus avoiding a visit to the ED. Often, treatment in the day hospital over a period of 4 to 12 hours avoids the need for subsequent hospital admission for pain control. This model has typically been implemented in sickle cell centers, and they are staffed with multidisciplinary professionals who work with experts in the management of SCD. Most units are open during the day hours; relatively few are open 24 hours/7 days. While these models are successful, currently there are a very small number of day hospitals operating in the United States.
We sought to replicate the success of the day hospital model by forming multidisciplinary quality improvement teams from three centers focused on improving pain management for adult patients who present to the ED with acute pain crisis. The overall goals of the larger project were to 1) develop and test an ED-based SCD surveillance system that will be used to track ED utilization and characteristics of visits; 2) implement a continuous quality improvement initiative and test the surveillance system for responsiveness to interventions; and 3) develop a tool box of assessment tools, analgesic protocols, and education materials designed to optimize quality care.
The specific goals of this article are to describe the learning collaborative methodology and to report baseline patient and visit characteristics, health services utilization (ED visits and hospital admissions), baseline and follow-up patient interview data, and analgesic management practices for each site.
A prospective, longitudinal cohort study using both research and quality improvement methods was used in the context of a learning collaborative model. Institutional review board approval was obtained at each participating center.
Learning Collaborative Model. The learning collaborative model is a health care quality improvement methodology developed by the Institute for Healthcare Improvement. It is a short-term learning system that brings together several teams of individuals from hospitals or clinics to solve a clinical topic area. The system is designed to close the clinical quality gap for a focused issue by creating a structure from which organizations can learn from one another and topic experts to produce local change. Team collaboration using the learning collaboration model has achieved results in reduced intensive care unit (ICU) costs, decreased congestive heart failure admissions, and reduced cesarean section rates.6 As far as we know, this is the first application of the learning collaboration model to acute sickle cell pain crises management in the ED. The structured medical record review was deemed exempt from informed consent, but subjects provided a one-time written consent for participation in all interviews for the duration of the 3-year study. Project oversight was reviewed three times during the first year by an advisory board. Members included a representative from the Sickle Cell Disease Association of Illinois, Illinois Emergency Nurses Association, Illinois College of Emergency Physicians, the Illinois Department of Public Health, and an internist and hematologist in the local area who provided care to the majority of adults with SCD. A patient with SCD provided input at the initial meeting to help determine important questions for patient interviews.
Study Setting and Population
Three sites enrolled subjects and included one rural (Peoria, IL) academic medical center with an annual ED census of 72,259 visits and two urban academic medical centers with annual ED censuses of 45,560 and 82,000 visits during the study period. Each site was affiliated with an emergency medicine residency program and employed full-time board-certified emergency physicians.
This study was composed of two components: 1) a structured medical record review to capture data on all ED sickle cell adult patient visits and 2) structured interviews with patients done shortly after ED care and at 30 days. The interviews were performed with a convenience sample of patients. The baseline study period (prior to implementation of analgesic protocols) reported in this article began October 1, 2007, at all sites, and data through July 2008 are reported for two of the three sites. Analgesic protocols were implemented at Sites 1 and 2 on August 1, 2008. Data for Site 3 are reported through mid-January 2008, at which time this site’s analgesic protocol was implemented. All ED patient visits with a chief complaint of an acute pain crisis and SCD as identified by discharge International Classification of Disease–Ninth Revision (ICD-9) codes were included in the medical record review, and all patients were also eligible for inclusion in patient interviews. Non–English-speaking patients, children less than 18 years of age, and patients who were unable to provide follow-up contact information were excluded from the interview portion of the study, and children were also excluded from the medical record review.
Structured Medical Record Review. All ED and hospital medical records for patients meeting study criteria were reviewed. A structured medical record review was conducted using the Gilbert criteria for review.7 Standardized data abstraction forms were used to collect data from original medical records. Reliability training was performed at the initiation of the study and prior to data abstraction. The principal investigator trained the coinvestigator at each site, who trained their individual data abstractors. Each investigator abstracted data from a single set of 20 randomly selected medical records from his or her institution using study inclusion criteria. The principal investigator (PT) and the two investigators then reviewed the records. Interrater disagreement was then resolved by the investigator at each site. To evaluate interrater reliability for the entire data set at the completion of the project, a random sample of 20% of visits (n = 141) was generated. Interrater reliability for three key medical record variables was calculated: time to initial analgesic (Cronbach α = 0.987), drug name of initial analgesic (κ = 0.88), and route of initial analgesic (κ = 0.98).
Characteristic study variables included patient demographics (i.e., age, sex, and race). Health services utilization and ED visit characteristics included the following: ED visit frequency data (total number of visits meeting study criteria and number of visits for each individual patient), arrival triage score, arrival pain score (0–10 cm) and pain locations, chief complaint (recorded from the triage note), and ED disposition (discharge home vs. admit to medicine, telemetry, the ICU, observation, transfer, death, or left without being seen or against medical advice). The Emergency Severity Index (ESI) triage system (a 5-level system) was used at all study sites. ESI Level 1 represents the highest acuity and ESI Level 5 represents the lowest acuity.8 ED analgesic management characteristics (time to administration of initial analgesic from time of arrival in the ED), analgesic agents (first and second dose), the route of administration, and discharge pain score (0–10) recorded within 45 minutes of disposition (only for patients discharged home from the ED) were abstracted from the medical record. The total number of analgesic doses, as well as whether or not analgesics were provided at ED triage, was also collected.
Patient Interviews. Each site conducted a total of 10 interviews per quarter (inclusive of a mix of initial and follow-up interviews). A convenience sample of patients was invited to participate. Patients were eligible for up to one interview during a 30-day time period. The overall purpose of the interviews was to provide preliminary data and gain an understanding of possible reasons for return ED visits, as well as gain insight into the patients’ perspectives of ED pain management. At the initiation of the study, additional interviews were conducted at Site 1 to gain an understanding of the difficulty of contacting patients and conducting the interviews. Two different interviews were conducted during two different time periods with two different purposes.
The first interview was conducted as close to the ED visit postdischarge as possible and no later than 14 days. This interview questioned the patient about interventions prior to the ED visit, relationship with a physician, and satisfaction with ED pain management and care. Several questions evaluated analgesic management. Of note, the question “Please rate how well you believe the ED doctors and nurses tried to manage your pain” (Likert scale, 1–10 with 1 being outstanding) was asked. This scale was modified to reflect one as the best score based on patient’s feedback that “1” is a better number, equivalent with a low pain rating. Patients also stated that even though their pain level may not have resolved, it was important to judge their opinion as to the “effort” physicians and nurses were making in attempting to decrease the pain.
Included in the 10 interviews per quarter was a second interview that was conducted between 21 and 37 days post-ED visit. The purpose of this interview was to determine if patients received and were able to fill an analgesic prescription, if they received and completed a follow-up appointment with a primary care physician, and had any return visits to an ED, physician office, or clinic. Each site had the option of conducting 10 initial or second interviews per quarter, based on the ability to contact follow-up patients. Selected data from these interviews are reported here.
Learning Collaborative Sessions. Study site learning collaborative meetings with members from each of the three study sites were held every 1 to 3 months to discuss enrollment issues and team initiatives aimed at improving the management of sickle cell patients in the ED. These meetings include site investigators and research staff. As the project evolved, site-specific data were shared. After the initial year, one face-to face meeting with all members from each site’s study team was conducted to discuss and review the project and discuss implementation challenges as well as next steps.
Site-specific Multidisciplinary Teams. Each participating center agreed to form a multidisciplinary team dedicated to implementing the project in its institution. Team members included a minimum of an emergency nurse, an emergency physician, and an internist or hematologist (two of three sites). Team members at the third site included the internist managing the adult sickle cell population, who met with the ED medical director and social worker. An emergency nurse was not routinely able to attend these meetings. Teams at each site met to disseminate institution specific data and to design strategies to meet the identified project measures. Individual centers were permitted to adopt different strategies to meet project measures. Each site elected to pursue a joint physician- and nurse-developed, nurse-initiated analgesic protocol. The three analgesic protocols were similar, yet slightly different. One site developed a method to implement individualized analgesic doses by creating a database of a suggested opioid and initial dose for patients well known to the adult sickle cell care team. A second site also developed educational materials and a nursing documentation flow sheet associated with the protocol that was disseminated for optional use at each study site.
All data were entered directly into a Web-based SNAP database (Snap Surveys, Ltd., London, UK). Data were exported into and analyzed using SPSS (v.17, SPSS Inc., Chicago, IL). Descriptive statistics were used to analyze the data. Means and standard deviations (SDs) are reported for normally distributed data, and medians and interquartile ranges (IQRs) are reported for variables with skewed distributions. Most variables were compared by study site. The number of total ED visits during the study period at each site was identified to determine the proportion of visits for SCD pain crisis among all ED visit causes. The mean number of visits per month for SCD pain crisis was determined by dividing the total number of visits for SCD pain crisis during the study period by the number of months in the study period (10 months for Site 1 and 2 and 3.5 months for Site 3). This allowed for a more direct comparison of the frequency of visits per site per month due to the different baseline data period for Site 3 when compared to Sites 1 and 2. The data for the baseline period reported in this paper represent 10 months for Sites 1 and 2, and 3 months for Site 3. Sites were compared on mean time to analgesic per ED visit and mean time to analgesic per patient (i.e., averaging multiple ED visits for each patient first and then averaging within site). To analyze the effect of multiple visits from individual patients, the time to initial analgesic was reanalyzed while controlling for the individual patient. For patients discharged home, a Wilcoxon signed rank test was used to analyze the difference between the highest tolerable pain score at discharge as reported by the patient during a follow-up interview, with the score documented in the medical record (when documented). The Kruskal-Wallis test with chi-square statistic was used to identify differences in pain scores at discharge for patients who were admitted to a medical bed versus to a telemetry bed versus discharged home from the ED.
A total of 155 patients are represented in the current study cohort. Table 1 reports sample characteristics per study site. The median age for the entire sample was 32 years (IQR = 24–40 years), with a range from age 18 to 66 years; 10 patients (6%) were age 50 years or greater. A total of 701 ED visits met criteria for inclusion in the baseline period. A total of 86 interviews were conducted with 40 different patients. Visit and health services utilization characteristics are reported in Table 2. Differences in the assignment of an initial triage category varied per site, with Site 1 more frequently assigning ESI Level 2.8 All patients reported pain as the chief complaint; however, 29% of the sample also complained of respiratory symptoms. Extremity and back pain were the most common sites of pain (73 and 59%, respectively); however, 42% of patients complained of chest pain, and 23% also reported abdominal pain. For the entire cohort, 71.6% of patients had between one and three visits to the ED during the study period, 14.8% had between four and nine visits, and 13.6% of patients had greater than nine visits. Each site had a number of patients with very frequent visits. Differences between sites were apparent (Table 2). While the time period for Site 3 was significantly shorter (3.5 months vs. 10 months), the same trend of relatively few patients with a larger number of visits was noted. We also analyzed number of visits per individual patient for Sites 1 and 2 only, because of the short time period for Site 3 (3.5 months), which influenced the mean number of visits for the entire cohort (Table 2). After removing Site 3 from this analysis, 52% of patients had between one and three visits to the ED during the study period, 21% had between four and nine visits, and 27% of patients had greater than nine visits. Fifty-eight percent of the patient visits were discharged home from the ED. During 17 of the 701 total visits, patients left against medical advice or before seeing the ED physician; all other ED visits resulted in a hospital admission.
Patient Sample Characteristics Per Site
|Study period||Sep 07–Jul 08||Sep 07–Jul 08||Sep 07–Jan 08|
|Number of individual patients||57||23||75|
|Age (yr), median (IQR)||36 (27–46)||27 (22–40)||30 (23–36)|
|Black or African American, %||98||100||100|
ED Visit and Health Services Utilization Characteristics per Site
|Study period||10-1-07 through 7-31-08||10-1-07 through 7-31-08||10-1-07 through 1-17-08|
|Number of visits||314||233||154|
|Number visits/per person, mean (±SD)||5.51 (±8.54)||10.13 (±14.61)||2.05 (±2.3)|
|Proportion of SCD visits/total ED visits (%)||314/ 68,717 (0.4%)||233/ 60,006 (0.4%)||154/13,867 (1%)|
|Mean number of visits/month ||31||23||44|
|Triage score (%)|
| ESI 2||58.5||0.4||15.2|
| ESI 3||41.1||96.1||84.1|
| ESI 4 or 5||0.3||3.4||0.7|
| Total n||287||233||151|
|Arrival pain score, mean (±SD)||9.17 (1.02)||8.41 (1.42)||8.85 (1.3)|
| Total n||276||229||81|
|Chief complaints (%)|
| GI symptoms||6.0||7.3||13.6|
| Dizzy, neurologic||4.6||2.2||2.6|
| Eye, ENT||0.3||0||2.6|
| Total n||307||232||154|
|Pain location (more than one may apply, n (%)|
| Head||12 (4)||31 (13)||6 (4)|
| Extremities||228 (76)||163 (70)||113 (74)|
| Chest||148 (49)||58 (25)||90 (60)|
| Abdomen||39 (13)||93 (40)||24 (16)|
| Back ||175 (58)||122 (53)||104 (68)|
|Number of visits/patient, n (%)*|
| 1–3||35 (61.4)||10 (43.5)||66 (88)|
| 4–9||5 (19.3)||5 (21.7)||7 (9.3)|
| 10–67||11 (19.3)||8 (34.8)||2 (2.7)|
| Total n of patients||57||23||75|
|ED disposition, n (%)|
| Home||148 (47)||178 (77)||81 (53)|
| Medicine admission||130 (41)||51 (22)||61 (40)|
| Telemetry admission||19 (6.1)||2 (0.9)||4 (2.6)|
| ICU admission||0||0||1 (0.6)|
| Observation admission||5 (1.6)||1 (0.4)||0|
| Transfer||0||0||1 (0.6)|
| AMA or LWBS||12 (3.8)||0||5 (3.3)|
Analgesic management practices are reported per site in Table 3. The median time to initial analgesic for the cohort was 74 minutes (IQR = 48–135 minutes). Differences between sites were evident (χ2 = 106, p < 0.01). Analgesics in general were not administered at triage with the exception of two cases. Differences between route of administration and choice of analgesic agents were also evident between sites (χ2 = 215, p < 0.01). The majority of patients at Site 1 received hydromorphone, and most subjects at Site 3 received analgesics via the intramuscular route. Patients who were discharged home had significantly lower pain scores documented at discharge (median = 4, IQR = 3–6) when compared with patients admitted to medicine (median = 8, IQR = 6–8) or telemetry beds (median = 8, IQR = 7–10, χ2 = 94, p < 0.01). Finally, patients reported a significantly lower targeted discharge pain score (mean ± SD= 4.12 ± 1.33) when compared with the actual documented discharge pain score within 45 minutes of discharge (mean ± SD = 5.14 ± 2.68, Z = −3.2, p = 0.001, n = 43). The clinical significance of this is unknown. Results from other individual questions from the interviews are reported in Table 4.
Analgesic Management per Site
|Time (min) to initial analgesic, median (IQR)||73 (45–135)||62 (44–89)||150 (70–278)|
|Average time (min) to initial analgesic per person, median (IQR)||88 (62–140)||70 (49–84)||166 (94–278)|
|Analgesic given at triage? n||n = 0||n = 0||n = 2|
|Total number of analgesic doses|
| Mean (±SD)||2.86 (±1.3)||3.05 (±0.89)||4.43 (±2.35)|
|Route of initial analgesic, n (%)|
| Oral||8 (2.7)||1||7 (4.6)|
| Intravenous||176 (60)||228 (98)||51 (33)|
| Subcutaneous||18 (6.1)||1||0|
| Intramuscular||93 (31.5)||2||95 (62)|
|First analgesic agent, n (%)|
| Morphine sulfate||70 (24)||123 (54)||90 (59)|
| Hydromorphone||211 (72)||45 (20)||55 (36)|
| Meperidine||0||0||6 (4)|
| Morphine sulfate and ibuprofen or ketorolac||4||48 (21)||0|
| Hydromorphone and ibuprofen or ketorolac||2||3||1|
| Fentanyl and ibuprofen or ketorolac||0||1||0|
| Meperidine and ibuprofen or ketorolac||0||1||0|
|Route of second analgesic, n (%)|
| Oral||4||2||74 (49)|
| Intravenous||177 (66)||217 (99)||47 (31)|
| Subcutaneous||18 (6.7)||0||0|
| Intramuscular||72 (27)||0||29 (19)|
|Second analgesic agent, n (%)|
| Morphine sulfate||58 (21)||155 (71)||15 (9.7)|
| Hydromorphone||211 (78)||48 (22)||129 (86)|
| Meperidine||0||0||6 (4)|
| Morphine sulfate and ibuprofen or ketorolac||0||10||0|
| Hydromorphone and ibuprofen or ketorolac||1||2||0|
|Medical record discharge pain score, mean (±SD)||6.4 (±2.5)||4.0 (±2.4)||5.0 (±2.67)|
| Total n||140||153||2|
|Patient reported desired discharge pain score mean (±SD)||4.1 (±1.4)||4.1 (±1.2)|| |
| Total n||52||30||0|
Selected Initial and Second Interview Results
|Initial interview: analgesic management*|
| What is the highest level of pain you could accept at discharge and tolerate at home (0–10)? mean (±SD)||4.12 (±1.33)|
| When you left the ED, did you feel you could you manage your pain at home? n (%) |
| Yes||21 (64)|
| No||12 (36)|
| Do you believe you received pain medicines within a reasonable time from your arrival? n (% yes)||56 (67)|
| Do you believe you received the correct medicine to manage your pain at home? n (% yes)||73 (87)|
| Patients were also asked “Please rate how well you believe the ED doctors and nurses tried to manage your pain” (Likert scale, 1–10, with one being outstanding), median (IQR)||2 (1–6)|
|Initial interview: interventions prior to arrival and physician relationship*|
| Before you came to the ED did you, n (%)|
| Call a physician||15 (18)|
| See a physician in the office||5 (6)|
| Take nonsteroidal anti-inflammatory drugs||27 (32)|
| Take oral opioids||56 (66)|
| Use heat||31 (37)|
| Use distraction||43 (51)|
| Use an oxygen tank||1 (1)|
| None of the above||7 (8)|
| If you talked to your physician, did he tell you to go to the ED? n (% yes)||13 (15)|
| Do you have a physician for long-term management of your SCD? n (% yes)||62 (73)|
| If yes, what type of physician is he? n (%)|
| Hematologist||43 (73)|
| Family practice||5 (8)|
| Internal medicine||11 (19)|
| When did you last see a physician in a clinic before the ED visit? n (%)|
| Within the past week||15 (19)|
| Within the past month||30 (37)|
| Within 3 months||16 (20)|
| Within 6 months||9 (11)|
| Within 12 months||4 (5)|
| More than 12 months||7 (8)|
|Second interview: selected questions†|
| Were you able to obtain a follow-up appointment with a primary care physician when you left the ED? n (% yes)||39 (70)|
| Did you keep the follow-up appointment? n (%) |
| Yes||32 (58)|
| No||2 (4)|
| A future appointment is scheduled||6 (11)|
| N/A (did not get a follow-up appointment)||15 (27)|
| When you were discharged from the ED or the hospital, did you get a prescription for pain medications? n (% yes)||38 (69)|
| Was the prescription for the medication you felt you needed? n (% yes)‡||34 (90)|
| Have you had trouble filling the prescriptions? n (% yes)‡||9 (24)|
To the best of our knowledge, this is the first multi-institution, prospective, cohort study of adults with SCD in an ED setting. In addition, the design of this study is unique in that it uses a learning collaborative model to design, implement, and measure outcomes of quality improvement initiatives at each site. The purpose of this article was to report patient, visit, and site characteristics for the preintervention period. The patient demographic characteristics of our cohort are worthy of further discussion. In our sample the median age was 32 years, and only 10 patients were greater than age of 50 years. Our data support previous work that reported the median age of death for males of 42 years and females of age 48 years; however, over 50% of this cohort survived beyond age 50 years.2 Thirty-three percent of the patients in the cohort of Platt et al.2 died during a pain episode and other predictors of death included acute chest syndrome, renal failure, seizures, a white blood cell count greater than 15,000, and a low fetal hemoglobin.2 These data suggest that careful evaluation for life-threatening comorbidities is warranted.
A striking finding is the large number of visits for a small number of patients that was found to be consistent across sites. In our sample, it is important to highlight that 72% of the patients had only one to three visits during the study period. Our cohort could be grouped into low (1–3 visits, 72%), medium (4–9 visits, 14%), and high utilizers (10 or more visits, 14%). After removing Site 3 from this analysis, over half of the patients had between one and three ED visits over 10 months, 21% were medium utilizers, and 27% were high utilizers. Very little is known about why some patients with SCD have many ED visits. Recently, Aisiku and colleagues9 compared high (more than three ED visits/year) with low utilizers and found that high utilizers were more likely to have lower hematocrit levels and more transfusions and report more pain days, more pain crises, higher mean pain and distress, and worse quality of life.9 If using the definition of high utilizers of Aisiku et al., 48% of our cohort would meet this definition. Providing care to patients with frequent ED visits can contribute to frustration levels experienced by both patients and ED staff.10–13 It is a common perception of emergency physicians and staff that SCD patients presenting to the ED are opioid dependent, and this belief may lead to inconsistent or inadequate treatment.14,15 Both physicians and nurses tend to overestimate the prevalence of addiction in the SCD population, which has been documented to range from 0% to 11%.12,15 Further exploration is required to fully understand the frequent use of the ED, which may represent a higher acuity of illness, lack of access to a sickle cell provider, or other complex psychosocial issues.
We also report similarities and differences in discharge-to-home rates between sites. Discharge rates were similar between Sites 1 and 3 (48% vs. 53%). Although no conclusions regarding cause and effect can be determined, it is tempting to speculate that the greater number of analgesic doses given at Site 3 may have allowed a comparable discharge rate. While Site 2 had the highest discharge rates (77%), the site had by far the smallest number of unique patients (23) and it is likely that this reflected ED practices with regard to one patient who was a high utilizer. One subject at this site had 67 visits and was admitted only five times. The optimal discharge-to-home rate for patients presenting to the ED with sickle cell pain is unknown; however, data from day hospitals designed specifically to treat this condition suggest that rates as high as 81% are achievable.5 While the day hospital model cannot be replicated 100% in an ED setting, especially given overcrowding challenges, it is possible that aggressive ED analgesic management and/or admitting patients with SCD to an ED observation area for a 4- to 12-hour trial of analgesic management may help decrease the number of patients requiring hospital admission for pain control alone.
We found significant variation in analgesic management and triage practices between sites. Significantly more patients at Site 1 were appropriately triaged as ESI Level 2, in accordance with the ESI. The ESI clearly defines patients with severe pain or distress that cannot be managed at triage or patients with a high-risk situation as ESI Level 2.8 Patients with SCD are at high risk of life-threatening complications of the disease and therefore often meet two ESI Level 2 criteria. This initial categorization is crucial, especially in the era of ED overcrowding. If the triage nurse recognizes the patient and does not “believe” the patient’s pain rating, the triage nurse may “downgrade” the acuity category and the patient may be forced to wait for hours. The ESI system triages patients with a pain score of ≥7 to a Level 2 category, but those patients complaining of a < 7 pain score may be triaged as ESI Level 3.8
Finally, analgesic management practices were explored. Utilization of guideline-based pain management for acute pain crises has been shown to result in less frequent ED visits and hospitalizations.16 The American Pain Society guidelines recommend initial assessment and analgesia within 30 minutes of arrival, which is a significant challenge in the current era of crowding.17,18 We found large variation in time to initial analgesic between sites as well as within each individual site, and none met the goal of initiation of analgesia within 30 minutes of arrival. Site 3 had much longer times to initial analgesia than the other two sites. This may be a result of the overall large ED volume at this site, as well as the assignment of ESI Level 3 to the majority of patients with acute pain of SCD. Even when controlled for individual patient, the median time to analgesia at Site 3 was double those of the other sites. We acknowledge that delays to administration of initial analgesic are not unique to patients with SCD because of the serious crowding problem facing EDs on a daily basis.
Yet another major practice variation between sites was in route of analgesia. Site 2 used the intravenous (IV) route almost exclusively, while Site 1 used IV most frequently followed by intramuscular (IM), and Site 3 used a combination of IM and IV. American Pain Society guidelines clearly recommend the IV route as the route of choice and subcutaneous as the preferable route when IV access is not possible.19 All sites have adopted the subcutaneous route into their protocols, and it is hoped that time series analysis will report a decrease in the number of analgesic agents administered using the IM route.
We were also able to obtain the patients’ evaluation of ED analgesic management, which is a critical perspective when attempting to analyze ED pain management practices. Thirty-six percent of those patients discharged from the ED expressed they would not be able to manage their pain at home. While this reflects a small sample size (n = 33), it is possible that discharging patients home when they do not feel they can manage their pain at home may contribute to ED revisits; further exploration is needed. Most patients felt that they received the correct analgesic in the ED, but approximately one-third of patients felt that they waited too long. This affirms data from the medical record review that found lengthy delays to delivery of an initial analgesic. On a positive note, patients reported that they were very pleased with the attempts made by ED providers at managing their pain. Patients also reported using multiple different strategies to manage their pain prior to coming to the ED; 66% of patients had already taken an oral opioid, and one-third had taken nonsteroidal anti-inflammatory drugs. Only seven patients came to the ED without using some intervention to avoid the visit. These data should be useful in guiding ED analgesic management. Most patients are already taking oral opioids, and this information can be useful in designing individual ED analgesic management plans.
We were also able to obtain information during the interviews that investigated the patients’ relationships with primary care physicians. During the interviews, 27% reported not having a physician. This is concerning, as all patients with SCD should have a close relationship with a physician. Of those patients with a physician, approximately 75% had seen their physician at least within the past 3 months. Lack of a primary care physician could contribute to poor disease management and frequent ED visits. Finally, we were also able to obtain preliminary data regarding difficulty making and keeping appointments and filling analgesic prescriptions. Our data demonstrate that a significant number of patients were not able to make or keep a follow-up appointment, and some patients expressed difficulty filling their analgesic prescriptions. Specific difficulties were not routinely specified; however, many subjects found that pharmacies did not carry the specific opioid or sufficient quantities. These difficulties may contribute to repeat ED visits and require further exploration.
One additional finding of particular interest at all three sites was the large number of patients who presented with complaints of respiratory symptoms as well as chest and abdominal pain. This is of concern because these may be signs of life-threatening complications of SCD, and they emphasize the need for assignment of the appropriate triage category to ensure rapid assessment and treatment.
Widespread applicability of these results may be difficult due to the modest sample size, although multisite enrollment affords better generalization. Patient care data were abstracted from ED medical records. Patients with an ICD-9 code indicating an SCD pain crisis were used to identify the study population, and it is possible that not all SCD patients were identified due to coding errors, although sites also cross-checked these lists with ED daily logs. Each of the sites used a different methodology for ED visit clinical documentation, with two of three sites using paper records. Although study personnel were trained in data abstraction techniques and definitions were set in advance, abstracting errors and misclassifications of data may have occurred. Analgesic administration times were based on information recorded in the chart and may not reflect a precise measure; however, typically medication administration times are accurate. Participants in the follow-up interviews were enrolled using a convenience sampling process. Most interviews were conducted with patients who were admitted to the hospital, because it was more difficult to successfully contact patients via follow-up phone calls. Although repeated interviews of the same patients were limited, some patients were interviewed more than once.
This report of baseline data does not detail any serious adverse events that may have been associated with delays to analgesia; however, this study was not designed to measure associated adverse outcomes. It is also possible that patients in this cohort visited other EDs and we have no way to capture these data. However, our data did capture a small number of patients who did have multiple ED repeat visits, and this cohort clearly requires further investigation. Finally, we did not compare the time to administration of analgesics for patients with SCD to other ED patients.
We found significant practice variations between the three sites and low adherence to recommendations for management of acute pain in sickle cell disease from national organizations such as the American Pain Society and the National Institutes of Health. Although it is well accepted that standardizing practices around existing evidence improves care delivery, no studies have demonstrated the impact of standardizing the ED approach to sickle cell acute pain management on measures such as effectiveness of pain control and health care resource utilization. We hope to address this in future reports of longitudinal data and quality improvement outcomes associated with efforts initiated at each of the three study sites.
The authors thank Alex Chiu, Christiana Chukwu, Matt Gilbert, Julie Stevens, Autumn Stevenson, and Kristen Turner for assistance with data abstraction.