Pain management in children and adolescents with sickle cell disease


  • Jeanette M. Jerrell,

    Corresponding author
    1. Department of Neuropsychiatry and Behavioral Science, University of South Carolina School of Medicine, South Carolina
    • Department of Neuropsychiatry and Behavioral Science, University of South Carolina School of Medicine, 3555 Harden Street Ext., Suite 301, Columbia, SC 29203
    Search for more papers by this author
  • Avnish Tripathi,

    1. Department of Epidemiology and Biostatistics, University of South Carolina Arnold School of Public Health, South Carolina
    Search for more papers by this author
  • James R. Stallworth

    1. Department of Pediatrics, University of South Carolina School of Medicine, South Carolina
    Search for more papers by this author

  • Conflict of interest: The views expressed do not necessarily represent those of the funding agency or official findings of the South Carolina, Department of Health and Human Services (Medicaid).

Pain Management in Children and Adolescents with Sickle Cell Disease

In a cohort of 2,194 children with sickle cell disease (SCD) treated in community-based services, we explored the types of medications used to treat vaso-occlusive (VOC) pain episodes, and the relative effectiveness of nonsteroidal anti-inflammatory drugs (NSAIDs), opioids, and adjunctive antidepressants or anticonvulsant medications on reducing acute VOC pain visits over time. Pharmacologic treatments for VOC pain consisted mainly of NSAIDs and weak opioids. Significantly more patients with more than 3 inpatient or ER VOC pain visits during their first year of SCD treatment were prescribed stronger opioids, SSRIs, SNRI/heterocyclics, and anticonvulsants. Prescription of both stronger opioids and SSRI antidepressants or anticonvulsants was significantly associated with lower cumulative rates of acute VOC pain visits over time. Using an observational study design and existing clinical data, these findings are intended to illustrate the potential clinical advantages of combining adjunctive antidepressants or anticonvulsants with primary pain medications for relief of acute VOC pain over time.

In children with SCD, erythrocytes become deoxygenated, dehydrated, and crescent-shaped, and tend to aggregate or stick to blood vessel walls, blocking blood flow in bones and organs, and causing recurring, VOC pain episodes [1]. With recurring acute VOC pain episodes, severe SCD represents a chronic pain condition [2]. For effective long-term management, pharmacological interventions for acute VOC pain [3] could be combined with adjunctive interventions for chronic pain [4].

The chronology of pain in children with SCD includes eight phases [3]. The first three phases involve specific pain sites (e.g., dactylitis), prodromal signs of VOC pain, and an early, mild form of VOC pain. A mild oral analgesic may be given (e.g., acetaminophen), and NSAIDs may be effective in reducing bone and joint pain. Phase 4 is characterized by pain accelerating to a moderate level, interfering with daily activities, and requiring stronger oral analgesics (e.g., weak opioids [codeine, hydrocodone]). In Phase 5, VOC pain increases to a severe level and plateaus for an extended time period, requiring hospitalization and intravenous (IV) opioids (e.g., morphine, oxycodone). The acute phase of VOC pain episodes may last 3–5 days. The VOC pain starts to decrease in Phase 6 and in Phase 7 decreases rapidly, as treatment with of IV analgesics is gradually decreased and the use of sustained-release oral opioids is initiated. In Phase 8, VOC pain is resolved or decreased to a “tolerable level,” treated at home, using mild to moderate strength analgesics. Some patients may require “breakthrough” oral opiates (e.g., morphine elixir).

Hydroxyurea (HU) has been shown to reduce episodes of VOC pain in observational and randomized controlled trials [5–10], and regular blood transfusions are used for prevention of recurrent pain in patients who have not responded to HU [2, 4]. Pharmacotherapies for chronic pain include agents from multiple drug classes (simple analgesics, NSAIDs, opioids, anticonvulsants, and antidepressants) [4]. The tricyclic antidepressants [11–13] as well as the serotonin norepinephrine reuptake inhibitors or heterocyclic (SNRI/ heterocyclic) antidepressants venlafaxine, bupropion, and duloxetine [14–17] are efficacious in the management of chronic pain. The SNRI/ heterocyclics produce fewer side effects in children and adolescents and, when used as adjunctive analgesics, they effectively improve patient overall quality of life [11]. When used solely for pain management, the selective serotonin reuptake inhibitors (SSRIs) have either been less robust (i.e., paroxetine, citalopram) or lacked any efficacy at all (i.e., fluoxetine) [11, 18]. Two anticonvulsants, gabapentin and pregabalin, are backed by the strongest evidence as analgesics for multiple types of pain [18] but others may be used. No systematic studies of the combined effects of these classes of agents on VOC pain were found in the literature.

This pediatric SCD cohort was 60% African American, 53% male, entered into the Medicaid data set at 5–6 years of age, and remained in the data set for about 7 years. Most were prescribed NSAIDs or weak opioids (Table I). The majority of patients treated with NSAIDs were taking ibuprofen, motrin, and naproxen. The weak opioids prescribed were codeine/acetominophen and hydrocodone. The stronger opioids prescribed were morphine and oxycodone. Those prescribed SSRIs were taking citalopram, escitalopram, paroxetine, fluoxetine, or sertraline. Prescribed SNRI/heterocyclic agents were mainly venlafaxine, mirtazapine, bupropion, or duloxetine. The primary anticonvulsants prescribed were carbamazepine, valproic acid, gabapentin, or phenytoin, with fewer patients receiving pregabalin (Table I). Ten percent of the pediatric SCD cohort had more than three acute VOC pain visits during their first year of SCD treatment, and significantly more of these were prescribed stronger opioids, SSRIs, SNRI/heterocyclics, and anticonvulsants (Table II). As shown in Table III, only those children prescribed both stronger opioids and SSRI antidepressants or anticonvulsants had significantly lower cumulative rates of acute VOC pain visits over time, controlling for other interventions, which could impact VOC pain.

Table I. Descriptive Analyses of 2,194 SCD Cohort
Independent variableSCD Cohort
  • a

    Hispanic, Middle Eastern, Asian Indian, or mixed.

 African American1305 (59.5%)
Non African Americana889 (40.5%)
 Female1041 (47.4%)
 Male1153 (52.6%)
Mean age at SCD diagnosis as documented in Medicaid5.7 (SD = 5.3)
Years in Medicaid7.2 (SD = 3.1)
Percent receiving care at specialty clinic13.4 (SD = 14.9)
SCD VOC pain treatments
Mean blood transfusions per year1.6 (SD = 1.9)
Prescribed hydroxyurea: Yes209 (9.5%)
Adenotonsillectomy performed: Yes256 (11.7%)
Prescribed NSAID: Yes740 (33.7%)
Prescribed weak opioid: Yes1477 (67.3%)
Prescribed stronger opioid: Yes494 (22.5%)
Prescribed SSRI antidepressant: Yes122 (12.0%)
Prescribed SNRI/heterocyclic antidepressant: Yes111 (10.9%)
Prescribed anticonvulsant: Yes74 (3.4%)
Table II. Vaso-Occlusive Pain Severity During First Year of SCD Treatment
Medication PrescribedaVOC Pain First Year Severity–Low (N = 1972; 89.9%)VOC Pain First Year Severity–High (N = 222; 10.1%)P-value
  • a

    Patients could be prescribed more than 1 medication.

  • Low = 0–3 pain visits; High = >3 pain visits and Pain Treatments Received.

NSAIDs572 (29.0%)168 (75.7%)<0.0001
Weak Opioids1258 (63.8%)219 (98.7%)<0.0001
Stronger Opioids342 (17.3%)152 (68.5%)<0.0001
SSRIs93 (4.7%)29 (13.1%)<0.0001
SNRI/heterocylics91 (4.6%)20 (9.0%)0.005
Anticonvulsants48 (2.4%)26 (11.7%)<0.0001
Table III. Effectiveness of Pain Treatments in Reducing VOC Pain Visits Over Time
PredictorEstimated Coefficient95% Confidence IntervalP-value
Adenotonsillectomy−0.21(−0.42, −0.01)<0.0001
Transfusions per year0.17(0.07, 0.26)0.0003
Percent receiving specialty care0.01(0.009, 0.02)<0.0001
Amount NSAIDs prescribed per year0.01(0.004, 0.02)0.007
Amount weak opioids prescribed per year0.08(0.07, 0.10)<0.0001
Prescribed stronger opioids and SSRIs−0.02(−0.04, −0.01)0.0009
Prescribed stronger opioids and anticonvulsants−0.0002(−0.0003, −0.0001)<0.0001

Most of these psychotropic medications have infrequent but potentially important hematologic side effects or may interact with the anticoagulants used in medically ill patients [19]. The SSRIs citalopram, paroxetine, fluoxetine, and sertraline as well as some SNRIs may inhibit platelet function and are associated with an increased risk of bleeding complications (e.g., gastrointestinal bleeding, ecchymoses, epistaxis, hematomas, hemorrhage) or bruising, especially with the concomitant use of aspirin or NSAIDs [19]. Citalopram is associated with leukocytosis and leukopenia, whereas sertraline is associated with thrombocytopenia; both are associated with the development of anemia [19]. The SNRI/heterocyclic agents, venlafaxine, mirtazapine, and bupropion, are associated with leukopenia. Furthermore, venlafaxine is associated with the development of anemia and leukocytosis, and mirtazapine is associated with the development of anemia, eosinophilia, agranulocytosis, pancytopenia, and thrombocytopenia, whereas duloxetine is only associated with bruising and bleeding [19]. The anticonvulsants, carbamazepine, and phenytoin, carry an increased risk of agranulocytosis, leukopenia, and thrombocytopenia, carbamazepine is also associated with the development of anemia, eosinophilia, and leukocytosis, whereas valproic acid is associated with pure red cell aplasia (as is carbamazepine) and thrombocytopenia. Gabapentin is associated only with leukopenia and neutropenia [19].

While this SCD cohort represents a large, heterogeneous group of children and adolescents, and the long-term observational study provides information regarding important clinical interventions and their impact on pediatric VOC pain, the data were not gathered using a prospective, controlled design and structured clinical research interviews were not employed to confirm any of the assigned medical conditions. Previous studies have shown that although Medicaid databases provide much less detailed information on individuals and care than primary data collection, physician diagnoses and utilization data correspond to clinical medical record reviews in the majority of the cases [20, 21]. Furthermore, these results report associations and, as a result, directions of causality cannot be inferred. Pediatric SCD patients who dropped out of treatment or were periodically ineligible for Medicaid coverage are not represented in this data set and their outcomes may differ from those patients who remained in Medicaid over time.

Practitioners will need to evaluate the individual benefit-risk ratio of combining analgesics with antidepressants or anticonvulsants in SCD children and adolescents with severe pain, rather than solely increasing the opioid dose [22], and realize that none of these psychotropic agents is FDA-approved for pain management in children and adolescents. Controlled trials regarding the adjunctive use of psychotropics for VOC pain relief are needed to determine which agent combinations are safe, effective, improve quality of life, and reduce the personal and payer burden associated with multiple acute VOC pain episodes per year [23–25].

Materials and Methods

Medical and pharmacy claims for the calendar years January 1, 1998 through December 31, 2006 were used to identify a cohort of child and adolescent patients (ages 17 and under) enrolled in and eligible for Medicaid for a minimum of 9 months in each calendar year included in this analysis, who had a service encounter with a diagnosis of SCD (thalassemia and SC trait were not included). The selection date was the date of the first encounter in which the SCD diagnosis was documented in the Medicaid system. No information was available on their treatments before this start date. This procedure yielded a total N = 2194 pediatric SCD cases during the 11-year period. Medicaid medical claims were used to identify a service encounter, date of service, and the International Classification of Diseases, 9th Clinical Modification diagnosis codes related to that visit. Pharmacy claims identified the medication dispensed, and the date the prescription was filled. This study was approved by the University of South Carolina Institutional Review Board as exempt from human subject research guidelines (45 Code of Federal Regulations part 46).

Primary or secondary diagnosis codes for VOC inpatient or ER pain visits and for blood transfusions, an adenotonsillectomy, and the percent of services received from an SCD specialty clinic were obtained from the Visits file. Prescriptions of HU, NSAIDs, weak and stronger opioids, anticonvulsants, and antidepressants were coded from the Pharmacy file. The analgesics examined were prescribed acetaminophen, NSAIDS (naproxen, ibuprofen, motrin, etc.), weak opioids (codeine or hydrocodone, with or without acetaminophen), or stronger opioids (morphine or oxycodone, with or without acetaminophen). Antidepressants were categorized as SSRIs for citalopram, escitalopram, fluoxetine, fluvoxamine, paroxetine, and sertraline, or SNRIs/ heterocyclics for bupropion, duloxetine, maprotiline, mirtazapine, nefazodone, trazodone, or venlafaxine. Anticonvulsants were coded for gabapentin, pregabalin, carbamazepine, valproic acid, and phenytoin. The SSRI and SNRI/heterocyclic antidepressants were prescribed for diagnosed depression, and the anticonvulsants were prescribed for seizures, not for the treatment of VOC pain.

To examine the relative impact of pain medication categories over time on frequency of acute pain episodes, a negative binomial regression model (for non-normally distributed count data; PROC GENMOD facility in SAS version 9.1; SAS Institute, Cary, NC) was employed to calculate a ratio of the log rate of VOC acute pain visits per total years in the Medicaid data set with prescribed NSAIDs, opioids, antidepressants, or anticonvulsants as the independent variables. Individual risk factors (age, gender, race), receipt of HU or an adenotonsillectomy (dichotomously coded as yes/no), which might be associated with hypoxic episodes correlating with the onset of pain crises [26, 27], mean number of blood transfusions per year, and the percentage of SCD services received from a specialty clinic for SCD were used as control variables to explain differences in total pain episodes. Interaction terms between the strong opioids and antidepressants or anticonvulsants were used to examine the combined effect of these medications. Resulting model estimates, 95% confidence intervals, and P-values are reported for statistical significance.


Data analysis was supported by the University of South Carolina School of Medicine Departments of Neuropsychiatry, Pediatrics, and Internal Medicine.

Jeanette M. Jerrell*, Avnish Tripathi†, James R. Stallworth‡, * Department of Neuropsychiatry and Behavioral Science, University of South Carolina School of Medicine, South Carolina, † Department of Epidemiology and Biostatistics, University of South Carolina Arnold School of Public Health, South Carolina, ‡ Department of Pediatrics, University of South Carolina School of Medicine, South Carolina.