Emergency Medicine

Derivation and preliminary validation of a risk score to predict 30-day ED revisits for sickle cell pain

a b s t r a c t

Background: Emergency department (ED) revisits and 30-day readmissions have been proposed as markers for Quality of ED care for Sickle cell disease (SCD).

Objective: To create a scoring system that quantifies the risk of 30-day revisit after ED discharge for SCD vaso- occlusive pain

Methods: This was a dual-center retrospective derivation and validation cohort study. The derivation was performed at an academic, tertiary care center and the validation at an urban community hospital. The primary outcome was revisit to the ED within 30 days after an ED discharge for SCD pain. Recursive partitioning was used to derive a scor- ing system to predict 30-day revisits.

Results: Of a total of 1456 ED visits for SCD pain, there were 680 ED discharges (admission rate of 53%) in 193 unique individuals included in the derivation cohort. There were 240 (35.3%) 30-day revisits. Of a total of 126 ED visits for SCD, there were 79 ED discharges in 41 unique individuals in the validation cohort. The final risk score included 4 variables: (1) age, (2) insurance status, (3) triage pain score, and (4) amount of opioids admin- istered during the ED visit. Possible scores range from 0 to 6. The areas under the receiver operating characteristic curves were 0.746 (95% confidence interval, 0.71-0.78–derivation cohort) and 0.753 (95% confidence interval, 0.65-0.86–validation cohort). A cutoff of 4 or greater identified 60% of 30-day ED revisits in the derivation cohort and 80% of revisits in the validation cohort.

Conclusions: A risk score can identify ED visits for SCD pain with high risk of 30-day revisit.

(C) 2015

  1. Introduction

Sickle cell disease (SCD) is an inherited blood disorder that affects approximately 100000 individuals in the United States [1]. Although manifestations occur in every organ system, painful episodes are the most common complication of SCD [2,3] and acute care visits due to

? Funding: This work was supported by the National Heart Lung and Blood Institute, Mount Sinai Emergency Medicine Research Career Development Program5K12HL109005-02 as well as the Mentored Patient-Oriented Research Career Development Award5K23HL119351-02.

?? Gary Yu, DrPH, had access to the data and performed data analyses for the project in-

cluding data cleaning, preliminary data analyses, data cleaning, recursive partitioning, and multivariable analyses on the derivation cohort. Dr Yu did not participate in analyses of da- ta from the validation cohort.

* Corresponding author. Department of Emergency Medicine, Icahn School of Medicine at Mount Sinai, 1 Gustave L. Levy Pl. Box 1620, New York, NY 10029.

E-mail addresses: [email protected] (J. Glassberg), [email protected] (J. Simon), [email protected] (N. Patel), [email protected] (J.M. Jeong), [email protected] (J.J. McNamee), [email protected] (G. Yu).

pain account for most hospital admissions and health care expenditures [4-7]. Efforts to improve emergency care for sickle cell pain have fo- cused on hospital readmission rates and emergency department (ED) revisit rates as markers for quality of care [8-27].

Although there are no published investigations of factors associated with ED revisits, previous studies, limited mostly to large administrative data, have identified important factors associated with higher 30-day readmission rates for SCD pain including the following: socioeconomic deprivation, age, and use of corticosteroids [9,10]. Multicenter retro- spective analyses demonstrated that hospital and provider type are not significantly associated with readmission rates [10-12]. Factors as- sociated with lower rates of readmission included inpatient transfusion and successful outpatient follow-up [10,13]. Several reports of Quality improvement interventions for SCD pain management exist in the literature, with some showing a positive effect on readmission rates [14-20], however; these were generally implemented for all ED visits rather than focusing on high-risk individuals or visits with high-risk features.

Understanding factors that are associated with ED revisits for SCD pain is important as this can lead to targeted interventions of potentially

http://dx.doi.org/10.1016/j.ajem.2015.07.015

0735-6757/(C) 2015

modifiable risk factors. Currently, there are no established tools available to identify ED visits for SCD pain with high risk of return after discharge.

With this dual-center derivation and preliminary validation study, we sought to develop a simple scoring system to predict 30-day ED revisits for acute sickle cell pain. Our hypothesis was that factors associated with an index ED visit (the First visit) could be used to iden- tify visits with high risk for revisit within 30 days. Secondary aims of the study were to determine if admission to the hospital or successful out- patient follow-up had any effect on ED revisit rates in our population. The ability to identify High-risk situations can facilitate efficient re- source allocation and higher-intensity interventions to reduce revisits and readmissions.

  1. Methods
    1. Study design and setting

This was a dual-center derivation and validation cohort study of pa- tients with SCD who were discharged from the ED with a diagnosis of SCD pain episode. The derivation cohort was performed at the Mount Sinai Medical Center (MSMC), an academic, tertiary care, level 2 trauma center in New York City with approximately 106000 visits annually to the ED. The MSMC SCD clinics have a combined census (adult and pedi- atric) of 294 patients with an additional 89 individuals who use only the ED. The validation cohort was performed at St Joseph’s regional medical center in Paterson, New Jersey. St Josephs is a busy, urban community hospital with approximately 160000 annual ED visits. St Josephs does not have a dedicated SCD clinic; however, several hematologists in the area see patients with SCD. There is no overlap between providers who practice at the 2 medical centers. The protocol for this study was approved by the Mount Sinai School of Medicine Institutional Review Board and the St Josephs Regional Medical Center Institu- tional Review Board. A waiver of informed consent was granted for this observational study.

    1. Selection of participants
      1. Derivation cohort methods

This study was a retrospective medical record review of all visits to the MSMC ED related to SCD between January 1, 2007, and January 1, 2011. Medical record abstraction methods and data integrity proce- dures for the derivation cohort have been described previously and were in accordance with established guidelines for medical record re- views in emergency medicine [21,22]. Briefly, individuals who visited the ED during the sampling frame were identified via search of the elec- tronic medical record and a comprehensive review of each individual’s medical record (inpatient, outpatient, and ED) was performed. A list of 91 clinical and Sociodemographic variables (Appendix A) were extract- ed from each individual’s medical record and maintained in an SPSS da- tabase (SPSS, Chicago, IL; see Supplemental Appendix). Medical record abstractors were blinded to the study hypothesis. opioid doses were converted to intravenous (IV) Morphine equivalents according to a standard protocol [23]. Although the initial medical record extraction identified all ED visits for all people with SCD, the derivation used only visits for SCD pain (defined below) in which the patient was discharged from the ED.

      1. Validation cohort methods

The validation was a retrospective medical record review of all visits to the St Joseph’s ED between April 26, 2010, and November 1, 2010. Be- fore the start of the study, definitions were created for all study variables and kept in a manual of operations for the study protocol. Only variables necessary to calculate the derived risk score were extracted for the val- idation cohort. Medical record abstraction was performed by trained, blinded medical record abstractors [21]. Training included a review of

study variable definitions, medical record abstraction procedures, and supervised abstraction of 10 medical records. Data were abstracted di- rectly from the electronic medical record into paper case report forms. As with the derivation cohort, in addition to regular meetings with med- ical record abstractors to review study procedures, a 10% sample of medical records were double extracted in order to calculate interrater reliability with a required ? greater than 0.95.

    1. Definition of variables
      1. Sickle cell disease

Sickle cell disease was defined as HbSS, SC, S? thalassemia or S with other hemoglobin variant confirmed by hemoglobin electrophoresis. All electrophoreses were reviewed by a physician trained in transfusion medicine who was blind to the study hypothesis. For individuals with no electrophoresis on file, a review of the patient’s laboratory data (for signs of chronic hemolysis, anemia, and organ damage) was performed by the principal investigator to determine if SCD was likely. Equivocal cases were adjudicated by a second hematologist and excluded if a con- clusion could not be reached.

      1. Painful episode

A painful episode was defined as an ED visit that required treatment with opioids that could not be attributable to a cause other than SCD. For example, a visit for lower abdominal pain in which a urine infection is found would not be coded as a “pain episode.” A visit in which the pa- tient is diagnosed as having a urine infection, but also complains of lower back and leg pain similar to prior SCD crises would be coded as “painful episode.” In the event that the abstractor was not sure if a visit met the criteria for a painful episode, the medical record was reviewed by a second abstractor (an emergency medicine physician who was blinded to the study hypothesis).

      1. Thirty-day revisit

Revisits to the ED were defined as a return to the ED for acute sickle cell pain within 30 days of discharge. For patients discharged directly from the ED, 30 days was calculated from the time of ED departure. Al- though the derivation included only patients discharged from the ED, secondary analyses included admitted patients. For admitted patients, 30 days was calculated from the time of hospital discharge.

      1. Steady-state values

Steady-state values for hemoglobin, fetal hemoglobin, and leukocyte counts were recorded. The standard definition of “steady state” as de- fined by Ballas [24] was used.

  1. Statistical analyses

Statistical analyses were performed in SPSS and SAS 9.3 (SAS, Cary, NC). For the derivation cohort, univariate analyses were performed to determine distributional characteristics and assess for randomness of missing data (variables included in the final score had less than 5% miss- ing values so imputation was not performed). For the derivation, only ED visits for SCD pain in which the patient was discharged from the ED were included. Bivariate analyses with the primary outcome (30- day ED revisit) were performed on candidate variables (selected by the principal investigator) with a clinically or biologically plausible rela- tionship (based on known SCD pathophysiology and literature) with the outcome variable. For continuous variables, cutoffs were derived using the ?2 interaction detection followed by manual adjustment to ensure cutoffs make sense clinically. Recursive partitioning was used to derive a final scoring system to predict 30-day ED revisits, with the goal of minimizing the number of misclassified values in the final cell while maximizing simplicity of the score. Odds ratios at each decision node were rounded to the nearest integer to create the score. Operating characteristics of the derived risk score were calculated on both the

derivation and validation cohort. Although the initial derivation includ- ed all ED discharges for SCD pain, sensitivity analyses were performed to assess for bias from high-using patients including (a) an analysis that excluded individuals with more than 3 and 12 visits per year, and (b) an analysis that censured multiple repeat visits within a 30-day time frame such that a maximum of 2 visits within 30 days was included in the analysis.

To determine if admitting a patient to the hospital reduced the like- lihood of return visits to the ED for pain, a propensity-adjusted, log- binomial regression was performed using the method described by Lanehart et al [25]. A propensity score to quantify the likelihood of ad- mission was derived from the 91 variables in the database including vital signs at admission, vital signs at disposition, laboratory values (with change from baseline), and demographic values. Variables with a P value less than .1 were included in a preliminary multivariable model. This propensity score was used to adjust for the fact that visits where patients were admitted differed from those directly discharged. Log-binomial regression with clustering adjustment at the patient level (predictor variables included the derived risk score, propensity of admission score, and a binary disposition variable [admitted vs discharged]) estimated the adjusted effect of admitting a patient on their likelihood of revisiting the ED. Similar analyses were performed

to estimate the adjusted effect on revisit rate of successful follow-up with a physician within 2 weeks of visiting the ED for pain (defined as any outpatient visit recorded in the medical record).

  1. Results
    1. Characteristics of the derivation and validation cohorts

In the derivation cohort, there were 2076 ED visits related to SCD by 247 unique individuals during the 4-year sampling frame. There were 1456 visits for pain (1.40 per patient-year) of which 680 resulted in di- rect discharge from the ED (admission rate of 53%) with 240 (35.3%) 30- day revisits. The derivation was performed only on these 680 ED dis- charges (which occurred in 193 unique individuals) as the goal was to predict those who would revisit the ED after discharge (Fig. 1). In the validation cohort, there were 232 ED visits by 112 unique individuals during the 6-month sampling frame. There were 126 ED visits for pain (2.21 per patient-year), of which 79 were discharged from the ED (41 unique individuals with admission rate of 37.4%). Laboratory character- istics of the derivation cohort have been described previously [22]. De- mographic characteristics of both cohorts are listed in Table 1. Interrater agreement was greater than 95% for all extracted variables.

Derivation Cohort Validation Cohort

16 walkouts with no data

79 ED discharges for acute sickle cell pain included in risk score validation (41 individuals)

47 admitted, eloped or left without being seen

295 ED visits related to SCD among 159 unique individuals

~ 67,500 visits to the St.

Joseph’s ED between March 26, 2010 and Oct 28, 2010.

2,192 ED visits related to SCD among 281 unique individuals

~ 400,000 visits to the Mount Sinai ED between Jan 1, 2007

and Dec 31, 2010.

232 visits related to confirmed SCD among 112 individuals

1,456 ED visits for acute sickle cell pain

1 walkout with no data

7 patients with hemoglobin AA

46 patients with sickle cell trait

11 patients with sickle cell trait

680 ED discharges for acute sickle cell pain included in risk score derivation (193 individuals)

776 admitted, eloped or left without being seen

126 ED visits for acute sickle cell pain

2,076 visits related to confirmed SCD among 247 individuals

440 without 30-day revisit

240 with 30- day revisit

43 without 30-day revisit

36 with 30- day revisit

Fig. 1. CONSORT/STROBE diagram of patient flow. Flow diagram for the cohort.

Table 1

Demographic and clinical characteristics

Derivation cohort

Validation cohort

n

247

112

total ED visits for pain (discharges only)

1465 (680)

126 (79)

Pain visits/patient-year

1.40

2.21

Age (y), mean (SD)

25.3 (13.8)

23.6 (15.8)

Sex (%male)

50.6

47.3

On hydroxyurea (%)

Type of insurance

23.5

NA

Medicaid

71.0%

59.9%

Medicare

5.6%

8.2%

Private insurance

12.1%

16.4%

Uninsured

11.3%

15.5%

IV morphine equivalents/ED visit (IQR)

13.3 mg (4-40)

13.3 mg (5-40)

Triage pain score, mean (SD)

8.02 (2.63)

8.51 (2.01)

IQR, interquartile range.

    1. Risk score derivation

Using recursive partitioning, a risk score involving 4 of the 91 candi- date variables was derived (Table 2). Final variables included (1) age,

(2) insurance status, (3) triage pain score, and (4) amount of opioids ad- ministered during the ED visit. With respect to insurance, those who were uninsured and those whose insurance was not accepted at the SCD clinic had elevated risk of revisit. Possible scores range from 0 to 6, and the risk of 30-day revisit increases with higher scores (Fig. 2). The area under the receiver operating characteristic curve was 0.746 (95% confidence interval [CI], 0.708-0.784; Fig. 3). In the derivation co- hort, a cutoff of 4 or greater identified 60% of 30-day ED revisits (143/ 240 total revisits; misclassification tables are provided in the supple- ment). Super high-using patients (ie, those with N 12 ED visits in 1 year) were included in the derivation; however, in a post hoc analysis, removal of high users increased the calibration of the model (area under the curve [AUC], 0.780; 95% CI, 0.739-0.820). An additional anal- ysis was performed with removal of all patients with more than 3 visits in a 1-year period. This eliminated cases of clustered visits (ie, an indi- vidual visit represents both revisit and an index visit) and did not sub- stantially change the operating characteristics (AUC the curve, 0.743; 95% CI, 0.692-0.795). Removal of triage pain score from the scoring sys- tem (leaving only age, opioid dose, and insurance in the score) and ex- clusion of visits where the pain score was zero (as these visits were determine to involve medication refills) resulted in an AUC of 0.722 (95% CI, 0.679-0.764). We identified 3 visits which were for sickle cell pain, but opioids were not administered or prescribed. Inclusion of these additional visits did not change the AUC.

    1. Risk score validation

In the validation cohort, the risk score performed similarly with an area under the receiver operating characteristic curve of 0.753 (95% CI, 0.645-0.860). A cutoff of 4 or greater identified 80% (29/36) of ED re- visits for pain. Only 14 ED visits were followed by successful follow-up with an outpatient physician within 2 weeks. The rate of ED revisits for those who did and did not follow up outpatient was not significantly

Table 2

Risk score

Uninsured +1

Insurance not accepted by institution’s SCD clinic +1

Triage pain score >= 1 +1

Age >= 15 y +1

Opioid dose in IV morphine equivalents (mg)

0

0

0.1-8

+1

8.1-24

+2

N24

+3

different (38.4% vs 35.7%, P = .55). There were only 2 cases where pain score equaled zero at triage. Removal of triage pain score from the scor- ing system (and removal of the 2 patients with pain score of zero) did not substantially change the AUC (0.741; 95% CI, 0.630-0.852).

    1. Admission to the hospital is not associated with fewer 30-day ED revisits

A secondary analysis involving both admitted and discharged pa- tients (1456 visits in 247 unique individuals) was performed to assess the effect of admission to the hospital on the risk of 30-day revisit. A propensity score adjusted for clinical differences in visits where patients were admitted vs discharged (ie, higher-acuity patients were more like- ly to be admitted). In the propensity-adjusted, multivariable model, ad- mission to the hospital was not associated with a lower rate of 30-day ED revisits as compared with individuals who were discharged directly from the ED (rate ratio, 1.24; 95% CI, 0.64-2.40). Additional analyses stratified by risk score, age, and sex were performed to compare 30- day revisit rates for those admitted and discharged. Admission to the hospital was not associated with lower rates of 30-day revisit across any of the strata.

  1. Discussion

With this dual-center study, we report for the first time a derivation and preliminary validation of a tool to identify ED visits for SCD pain with high risk of 30-day return visit. The score is simple and easy to use with only 4 variables (age, insurance status, pain at triage, and ED opioid dose) and performed similarly well at an academic medical center with a dedicated comprehensive SCD program and at a commu- nity hospital without an SCD program. Further validation will be neces- sary before the tool would have clinical utility; however, the findings are still important as this is the first report of which clinical and sociodemographic factors are associated with ED revisits.

The 4 variables in this risk score provide important insights about the nature of acute care use in SCD. Prior studies have demonstrated that age is one of the strongest predictors of ED use. Brousseau et al

[3] demonstrated on a national scale that 18- to 30-year-olds have the highest rate of ED use for pain and that mortality in this age group is ac- tually rising. Some of this has been attributed to failed transition from pediatric to adult care. With respect to pain score at triage, our analyses identified that any score above zero was associated with higher rates of revisit. An internal review of records with triage pain of zero revealed that most of those visits were patients with mild sickle cell pain who had run out of medication and presented for refill. Exclusion of these visits and removal of pain score from the model did not substantially change the performance of the score, suggesting that the other factors in the score have greater predictive value for revisits. Emergency de- partment opioid dose was also directly proportional to ED revisit rate. It is possible that individuals with higher opioid tolerance have more difficulty managing pain with Oral medication at home and thus were more likely to return to the ED. It is also telling to note the variables that fell out of the predictive model. Indicators of clinical severity such as vital signs, respiratory symptoms [22,26,27], laboratory results, and changes in hematologic indices from baseline were strong predictors of the likelihood of admission; however, they were not predictive of re- visit rate. As the derivation only included discharged patients, this sug- gests that in both cohorts, clinician discharge of patients who were “too sick to go home” did not contribute substantially to revisit rates.

Although this risk score is the first to report factors associated with ED revisits, unfortunately, our analyses did not identify an intervention to reduce revisits. After adjusting for clinical severity and other patient attributes, admission to the hospital for pain control was not associated with lower rates of ED revisit for pain. In contrast to prior literature [13], successful outpatient follow-up within 14 days was not associated with lower revisit rates. This likely reflects the fact that there was no dedicat- ed SCD clinic and that access to private hematologists was limited in the

100.00%

PERCENT WITH REVISIT WITHIN 30-DAYS

90.00%

80.00%

70.00%

60.00%

50.00%

40.00%

30.00%

20.00%

10.00%

0.00%

Derivation Cohort Validation Cohort

0 1 2 3 4 5 6

RISK SCORE

Fig. 2. Risk of 30-day ED revisit for SCD pain. The x-axis shows the range of possible risk scores (0-6), and y-axis shows the percentage of patients with 30-day ED revisits for pain in each group. Dark bars indicate derivation cohort; light bars indicate the validation cohort. In both cohorts, a cutoff of 4 or greater identifies most of 30-day ED revisits.

validation cohort. Prior Quality improvement interventions have shown that improving quality of pain management in the ED has little effect on ED use; however, increased SCD clinic use [16] and SCD-day hospitals [14,15] substantially affect ED use patterns. One potential use for this risk score would be to allocate additional resources to higher-risk visits. The choice of score cutoff would also depend on the nature of the inter- vention. For example, visits with a score of 6 represented only a fraction of total visits (9% of visits in the derivation cohort and 11% in the valida- tion cohort) but were at extraordinarily high risk for return and accounted for a quarter of all revisits. High-intensity interventions could be focused on these situations. Potential implementation methods include incorporation of the score into the electronic medical record system, or calculation with a tablet or phone-based app.

This study has a number of limitations. Although there were rigorous data collection methods in place, this was a retrospective study and data were limited to what was documented during the course of Standard care. It is possible that Prospective data collection could yield a risk

score with better calibration as our AUC was approximately 0.75. Al- though this AUC falls into the “fair” range, the score is still valuable as it can identify most of 30-day revisits while excluding most nonrevisits. Higher levels of calibration would be required in risk scores for life- threatening diagnoses such as pulmonary embolism, cervical spine frac- ture, or syncope. Clinical data regarding inpatient care were not available in this data set, so the score could not include admitted patients. In addi- tion, the clinical value of this score is not certain as the performance of the score was not compared with clinician gestalt. A prospective comparison of risk score vs clinician gestalt would indicate whether the score is appro- priate for widespread use beyond using it to target high-intensity inter- ventions. Finally, although the score was derived and validated in 2 centers with different patients and different providers, additional external validations would strengthen the generalizability of this score.

This study is an important contribution to our knowledge regarding ED use for SCD pain. The risk score is simple and has potential to be im- plemented in the context of an intervention to prevent revisits in high-

AUC* = 0.746

95% CI 0.708 – 0.784

AUC* = 0.753

95% CI 0.645 – 0.860

Fig. 3. receiver operating characteristics of the 30-day ED revisit risk score. *AUC, area under the curve.

risk situations. Future research is necessary to identify interventions that can modify revisit patterns for high-risk ED visits.

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ajem.2015.07.015.

References

  1. Hassell KL. Population estimates of sickle cell disease in the U.S. Am J Prev Med 2010;38(4 Suppl.):S512-21.
  2. Aisiku IP, Smith WR, McClish DK, et al. Comparisons of high versus low emergency department utilizers in sickle cell disease. Ann Emerg Med 2009;53(5):587-93.
  3. Brousseau DC, Owens PL, Mosso AL, Panepinto JA, Steiner CA. Acute care utilization and rehospitalizations for sickle cell disease. JAMA 2010;303(13):1288-94.
  4. Yang YM, Shah AK, Watson M, Mankad VN. Comparison of costs to the health sector of comprehensive and episodic health care for sickle cell disease patients. Public Health Rep 1995;110(1):80-6.
  5. Woods K, Karrison T, Koshy M, Patel A, Friedmann P, Cassel C. Hospital utilization patterns and costs for adult sickle cell patients in Illinois. Public Health Rep 1997; 112(1):44-51.
  6. Lanzkron S, Carroll CP, Haywood Jr C. The burden of emergency department use for sickle-cell disease: an analysis of the national emergency department sample data- base. Am J Hematol 2010;85(10):797-9.
  7. Raphael JL, Dietrich CL, Whitmire D, Mahoney DH, Mueller BU, Giardino AP. healthcare utilization and expenditures for low income children with sickle cell dis- ease. Pediatr Blood Cancer 2009;52(2):263-7.
  8. Centers for Medicare and Medicaid Services. Quality measures compendium: Medic- aid and SCHIP quality improvement. Report; 2007.
  9. Aljuburi G, Laverty AA, Green SA, Phekoo KJ, Bell D, Majeed A. Socio-economic dep- rivation and risk of emergency readmission and inpatient mortality in people with sickle cell disease in England: observational study. J Public Health (Oxf) 2013; 35(4):510-7.
  10. Sobota A, Graham DA, Neufeld EJ, Heeney MM. Thirty-day readmission rates follow- ing hospitalization for pediatric sickle cell crisis at freestanding children’s hospitals: risk factors and hospital variation. Pediatr Blood Cancer 2012;58(1):61-5.
  11. Jan S, Slap G, Smith-Whitley K, Dai D, Keren R, Rubin DM. Association of hospital and provider types on sickle cell Disease outcomes. Pediatrics 2013;132(5):854-61.
  12. Shah N, Rollins M, Landi D, Shah R, Bae J, De Castro LM. Differences in pain management between hematologists and hospitalists caring for patients with sickle cell disease hospitalized for vasoocclusive crisis. Clin J Pain 2014;30(3):266-8.
  13. Leschke J, Panepinto JA, Nimmer M, Hoffmann RG, Yan K, Brousseau DC. Outpatient follow-up and rehospitalizations for sickle cell disease patients. Pediatr Blood Cancer 2012;58(3):406-9.
  14. Adewoye AH, Nolan V, McMahon L, Ma Q, Steinberg MH. Effectiveness of a dedicated day hospital for management of acute sickle cell pain. Haematologica 2007;92(6):854-5.
  15. Benjamin LJ, Swinson GI, Nagel RL. Sickle cell anemia day hospital: an approach for the management of uncomplicated painful crises. Blood 2000;95(4):1130-6.
  16. Givens M, Rutherford C, Joshi G, Delaney K. Impact of an emergency department pain management protocol on the pattern of visits by patients with sickle cell dis- ease. J Emerg Med 2007;32(3):239-43.
  17. Krishnamurti L, Smith-Packard B, Gupta A, Campbell M, Gunawardena S, Saladino R. Impact of individualized pain plan on the emergency management of children with sickle cell disease. Pediatr Blood Cancer 2014;61(10):1747-53.
  18. Powers RD. Management protocol for sickle-cell disease patients with acute pain: impact on emergency department and Narcotic use. Am J Emerg Med 1986;4(3):267-8.
  19. Solomon LR. Pain management in adults with sickle cell disease in a medical center emergency department. J Natl Med Assoc 2010;102(11):1025-32.
  20. Tanabe P, Hafner JW, Martinovich Z, Artz N. Adult emergency department patients with sickle cell pain crisis: results from a quality improvement learning collaborative model to improve analgesic management. Acad Emerg Med 2012;19(4):430-8.
  21. Gilbert EH, Lowenstein SR, Koziol-McLain J, Barta DC, Steiner J. Chart reviews in emergency medicine research: where are the methods? Ann Emerg Med 1996; 27(3):305-8.
  22. Glassberg JA, Chow A, Wisnivesky J, Hoffman R, Debaun MR, Richardson LD. Wheez- ing and asthma are independent risk factors for increased sickle cell disease morbid- ity. Br J Haematol 2012;159(4):472-9.
  23. Goodman LS, Brunton LL, Chabner B, Knollmann BC. Goodman & Gilman’s the phar- macological basis of therapeutics. 12th ed. New York: McGraw-Hill; 2011.
  24. Ballas SK. More definitions in sickle cell disease: steady state v base line data. Am J Hematol 2012;87(3):338.
  25. Lanehart, R. (2012, March 14). Propensity score analysis and assessment of propen- sity score approaches using SAS(R) procedures. Retrieved August 7, 2015, from http:// support.sas.com/resources/papers/proceedings12/314-2012.pdf
  26. Glassberg J, Spivey JF, Strunk R, Boslaugh S, DeBaun MR. Painful episodes in children with sickle cell disease and asthma are temporally associated with respiratory symptoms. J Pediatr Hematol Oncol 2006;28(8):481-5.
  27. Glassberg JA, Spivey JF, Strunk R, Boslaugh S, DeBaun MR. Painful episodes in chil- dren with sickle cell disease and asthma are temporally associated with respiratory symptoms. Blood 2005;106(11) [28b-28b].

Leave a Reply

Your email address will not be published. Required fields are marked *