AUTHORS
Jason Robert Hodges, PhD, Pei-Lin Chen, MPH2, Sri Sai ANWITHA3, Omolola Oladeji4, Shreya Sirohi5, Corey Jackson, PhD6, Deepa Manwani, MD7, Guolian Other Person Kang, PhD8, Jerlym Porter, PhD, MPH9
AFFILIATIONS
1 Saint Jude Children's Research Hospital, Memphis, Tennessee, United States
2 Saint Jude Children's Research Hospital, Memphis, Tennessee, United States
3 University of Memphis, Memphis, Tennessee, United States
4 University of Memphis, Memphis, Tennessee, United States
5 University of Memphis, Memphis, Tennessee, United States
6 Saint Jude Children's Research Hospital, Memphis, Tennessee, United States
7 Saint Jude Children's Research Hospital, Memphis, Tennessee, United States
8 Saint Jude Children's Research Hospital, Memphis, Tennessee, United States
9 Saint Jude Children's Research Hospital, Memphis, Tennessee, United States
ABSTRACT
Introduction: Pica is a recognized comorbidity in sickle cell disease (SCD) with increased clinical and research focus over the last twenty years. A 2023 meta-analysis of pediatric patients with SCD in the United States estimated a pooled prevalence of pica of 39.5% (Jorgensen et al. 2023). Pica has been found to be associated with an HbSS genotype and younger age (Ivascu et al. 2001), but correlates such as treatment and environmental exposures are less clear. We characterized disease severity, treatment exposure, acute-care burden, and neighborhood-level social determinants of health (SDOH) as contributors to pica among patients with SCD in a large, well-phenotyped longitudinal cohort. Hypothesis: patients with a history of pica would be more likely to have severe disease, less likely to utilize hydroxyurea and chronic transfusion, have a higher number of hospitalizations/ED visits, and be more likely to live in disadvantaged neighborhoods than patients without a history of pica.
Methods: Participants were selected from the Sickle Cell Clinical Research and Implementation Program (SCCRIP) cohort, aged 3 to 18 years. We compared participants with pica against those without a history of pica by treatment exposure (hydroxyurea (HU) and chronic transfusion (CTXN)), ED/inpatient utilization, hematologic labs, and 12 neighborhood SDOH indices (Economic Hardship Index, Social Vulnerability Index, Area Deprivation Index, Index of Concentration at the Extremes) measures, 3 food-access flags, and lead paint exposure, i.e., pre-1960 housing). Exact-Wilcoxon rank-sum tests were used for continuous variables, chi-square for categorical variables, and Fisher's exact test when expected cell counts were small. The cumulative incidence of first pica diagnosis by age was estimated using the Kaplan-Meier method and compared by treatment exposure groups using a Cox proportional regression model, both in the overall cohort and in the HbSS/Sβ0 subgroup. The false discovery rate (FDR) was controlled using the Benjamini–Hochberg procedure.
Results: 1,316 participants were screened. Genotype distribution: HbSS/ Sβ⁰ n=814 (61.8%), HbSC n=347 (26.4%), HbSβ+ n=122 (9.3%), other n=33 (2.5%). Pica was identified in 246 (18.7%) participants. Overall cumulative incidence of first pica diagnosis by age 18 years was 21% (95% CI:18-23%). Among 814 patients with HbSS/Sβ⁰ genotype, 178 (22%) had a history of pica, compared with 68 (14%) patients with other genotypes (q<0.001). Patients with pica were younger, with a mean age of 13.2 vs 14.8 years (q<0.0001). A history of ever being on chronic transfusion was significantly associated with a markedly lower cumulative prevalence: 7% (95% CI: 3%-11%) among CTXN-exposed vs 22% (95% CI: 20%-26%) unexposed overall (7% (95% CI: 3%-11%) vs 29% (95% CI: 25%-33%) within HbSS/Sβ⁰). HU exposure showed no difference in the overall cohort (19%, 95% CI: 16%-23% HU-exposed vs 22%, 95%CI: 18%-25% unexposed), but a small difference was shown within the HbSS/ Sβ⁰ group (28%, 95%CI: 23%-34% HU-unexposed vs 21%, 95% CI: 17%-25% HU-exposed). Overall, pica was associated with more ED visits (1.65 vs 0.65/patient-year, q<0.0001) and more inpatient admissions (0.69 vs 0.34/patient-year, q<0.0001). Additionally, those with a history of pica had lower Hgb levels (9.55 vs 10.1 g/dL, q<0.001), higher HbF (17.4% vs 11.9%, q<0.0001), and higher ARC (q=0.04). None of the 12 SDOH indices differed significantly between the groups.
Conclusion: Pica is associated with disease severity (as evidenced by clinical values and genotype), treatment non-exposure (chronic transfusion), and acute care burden, but not with neighborhood socioeconomic disadvantage. The dominant confounder in the analysis was age. A surveillance/ascertainment bias is also plausible: patients on chronic transfusion have more frequent, hematology-focused encounters, which may under-document non-hematologic behavioral symptoms like pica relative to less intensively monitored patients. Pica may serve as clinical marker for disease severity, warranting closer monitoring rather than solely as a marker of socio-environmental deprivation. Future studies could utilize SCCRIP’s robust phenotypic data to generate more prospective, multivariable, age-adjusted models to assess these associations. Additionally, including individual- or family-level SDOH variables may more directly assess these associations than neighborhood-level SDOH variables.
DISCLOSURES
Jason Robert Hodges, PhD: No relevant conflicts of interest to declare.