In this retrospective cohort study using data from patients in Ontario, Canada, we compared SCD-specific outcomes to that among matched non-SCD general population controls to contextualize the total burden of excess morbidity and mortality of both the underlying disease and the exacerbation due to recurrent VOCs. These measures were also analyzed across age groups and by VOC frequency in the follow-up period. This study identified 3801 patients with claims for SCD in the ICES database, which differs from a previous epidemiologic study (Pendergrast et al.) that identified 3418 patients with SCD in Ontario, CA [10]. This discrepancy could be due to the slightly longer and more contemporaneous study period compared to the previous study (current study: 2010–2021 vs Pendergrast et al.: 2007–2017). Given the growing prevalence of SCD in Canada, a later study period could simply mean a greater number of patients with SCD in Ontario. Additionally, certain exclusion criteria had been applied to get to the final prevalence reported in Pendergrast et al. Those with a missing or invalid OHIP number were excluded from the result of 3418, while the current study had not applied any exclusion criteria to arrive at 3801.

Patients with SCD with recurrent VOCs, defined as those with ≥ 2 VOCs per year in 2 consecutive years, had increased clinical complications, mortality, and HCRU compared with matched general population controls. Clinical complications, mortality, and HCRU increased with increasing age and number of VOCs.

The observed lower income in patients with SCD with recurrent VOCs was in line with a previous US study that found that children living with SCD were more likely to be in a household below the federal poverty level [25]. The finding of a relationship between income quintile and number of VOCs is a novel finding that depicts potential societal impacts of reciprocal effect between SCD and socioeconomic status. Additional research specifically focusing on number of VOCs and income is needed to better understand the relationship between these two variables (i.e., increased VOCs leading to lower income because of reduced ability to work or lower income leading to increased number of VOCs due to lack of resources for self-care).

In this study, patients had a substantial incidence of clinical complications that were consistent with the pathophysiology of SCD [1]. These clinical complications and associated end-organ damage are driven by a variety of factors including erythrocyte morphology, haemolysis, cell adhesion, and vaso-occlusion [1]. Consistent with the progressive nature of SCD and previous published literature, this study found older patients and those with an increased number of VOCs had more complications [21, 26,27,28]. In this study, approximately half of identified patients with SCD with recurrent VOCs continued to have ≥ 2 VOCs per year in the follow-up period, which is similar to other studies of this patient population. This is likely driven by the long follow-up (~ 7.5 years) and non-interventional nature of the study design. Some patients who were having recurrent VOCs could potentially receive treatment to reduce their VOC frequency and thus would have < 2 VOCs per year in the follow-up period. In addition, specifically in Canada, VOCs may not be adequately captured in administrative databases in Ontario because of home management of VOCs or potential outpatient management of VOCs, which could lead to underestimation of rates of VOCs in this patient population. Thus, the subgroup of patients that continued to experience ≥ 2 VOCs per year in the follow-up is critical to understanding burden in patients continuing to experience recurrent VOCs. The substantial clinical burden in this more severe subgroup of patients with SCD demonstrates the need for treatment options that reduce VOCs at an early age to prevent accumulation of damage that can lead to clinical complications and end-organ damage.

The mean age of death in this study (~ 39 years old) is younger than that previously reported from another retrospective database analysis of patients with SCD in Ontario, Canada—this was likely due to the focus on a more severe subgroup of patients with SCD with recurrent VOCs compared with the overall SCD population examined in the previous analysis [17]. In addition, mean age of death in this study was consistent with those previously reported for individuals with SCD with recurrent VOCs in England and France [29]. Even with significant blinding, it was clear that the proportion of patients who died increased with increasing age, aligned with the progressive nature of disease. The finding that an increased number of VOCs led to increased mortality is consistent with other published literature indicating the number of VOCs patients experience are a key marker for survival [9, 30]. While our study details substantial mortality in these patients, this is likely underestimated because patients were required to have at least 1-year follow-up to ensure adequate estimation of HCRU, which could have led to survival bias, since only those who had survived at least 1 year following their index date were included. Future studies analyzing ICES data may benefit from shorter data follow-up requirements to explore early mortality and other time-sensitive outcomes. Additionally, as multivariable modeling to identify independent risk factors for mortality and other clinical outcomes was out of scope for this study, this remains an important area for future investigation.

Rates of HCRU were higher in older patients and those with a higher number of annualized VOCs, highlighting the progressive nature of SCD and the fact that VOCs are a key marker of disease severity and HCRU [9, 21]. Published studies in patients with SCD have found similar results, suggesting that VOCs are a driver of high HCRU, particularly ED visits and hospitalizations and economic burden [6, 7, 16, 21, 31, 32]. Rates of HCRU reported in this study are higher than those previously published in a retrospective analysis of patients with SCD in Ontario over a 10-year period. This is likely because our study focused on the more severe cohort of patients with recurrent VOCs [17]. The findings that folic acid, hydroxyurea, and opioids were most used to treat patients with SCD are consistent with current clinical practice. Hydroxyurea has been the mainstay treatment for SCD in the world [11, 33, 34], even though it is not currently approved by Health Canada in patients with SCD [35].

Even though hydroxyurea has not been approved in Canada, the proportion of patients with at least one claim for hydroxyurea was 58%, far exceeding what is noted in the literature, even when compared to countries like the US where hydroxyurea is approved for use.

Studies analyzing treatment use among SCD patients in other countries where hydroxyurea is approved and reimbursed report similar or lower rates of hydroxyurea prescription. Factors such as patient nonadherence, side effects, and dose-limiting toxicity have reduced the use of hydroxyurea among patients with SCD [36,37,38,39,40,41,42,43].

As our database captures only prescriptions reimbursed through the provincial public drug plan, these results may underestimate the actual prescription rate of hydroxyurea and cannot expound on the factors leading to the observed rate. Nevertheless, these results highlight the need for system-level strategies to improve access and adherence to prescribed treatment.

This is the first study detailing real-world opioid utilization among patients with SCD with recurrent VOCs in Canada. While the prevalence of opioid use in this study was lower than in previous studies of patients in the US where opioid use ranged from 40 to 94% [21, 44, 45], there was substantial use among those who received at least one opioid (~ 7 claims PPPY), and use increased as patients aged and had more VOCs. Increased opioid use in older individuals with SCD and in those with more VOCs is consistent with other published studies [21, 46]. Treatment use described in this study may be underestimated because of the limitations of the data source (e.g., the Ontario Drug Benefit claims database only captures prescription history of individuals who met provincial public drug plan criteria). Moreover, ICES does not contain robust data on blood transfusions, an important treatment modality in SCD. Future analysis of alternative data sources with more comprehensive treatment data is warranted to provide a full description of treatment patterns in Canada.

Notably, the data collected in this study were based on provincial medical records. The identification of individuals with SCD using diagnosis codes is limited by accuracy of billing codes in inpatient and ED encounters; utilizing only inpatient and ED codes could also lead to underestimation of complication rates and prevalence. Additional research using alternative data sources (e.g., registry, electronic health records) can help address limitations of claims regarding clinical complications, including identification of VOCs [16, 46, 47].

There is potential for ascertainment bias due to differential healthcare utilization between patients with SCD and general population controls. Patients with SCD typically have more frequent interactions with the healthcare system, increasing the likelihood of receiving a diagnosis for comorbid conditions. In addition, differences in diagnostic intensity may further contribute to detection bias. For example, conditions such as pulmonary hypertension may be more frequently investigated in patients with SCD than in controls because of clinical vigilance, resulting in a higher probability of diagnosis in the SCD group. These differences may contribute to an increased likelihood of identifying complications in patients with SCD relative to those in the non-SCD general population. However, because our analysis was limited to diagnoses captured in inpatient and emergency department settings where some chronic conditions (e.g., anxiety, depression) are less likely to be diagnosed, the prevalence of such conditions may be underestimated for both patients with SCD with recurrent VOCs and the non-SCD general population. Future studies examining the broader population of patients with SCD are warranted as this study focused on a more severe patient subpopulation and existing research on the overall SCD population in Canada has primarily addressed epidemiologic objectives.

While the study was not designed to evaluate all potential predictors of outcomes, we did conduct subgroup analyses by VOC frequency during the follow-up period. These analyses demonstrated that individuals experiencing VOCs had notably higher rates of clinical complications, healthcare utilization, and mortality, highlighting the disproportionate burden carried by patients with more severe disease. Although we did not conduct multivariable modeling to identify independent risk factors for mortality or other clinical outcomes, this remains an important area for future investigation.

VOCs were defined as a composite of sickle cell crisis, ACS, and priapism to reflect the broader clinical burden experienced by individuals with SCD. While these events can be clinically distinct, our data indicate that the composite VOC measure in the follow-up period was overwhelmingly driven by the sickle cell crisis ICD-10 codes. Specifically, all patients included in the study had a diagnosis of SCD with crisis, whereas priapism and ACS were identified in only 6% and 20% of patients, respectively. Given that most events are SCD with crisis, the use of the composite definition is not likely to substantially impact patient identification and outcomes in this study.

Given the minimum continuous enrollment requirement, individuals who died or were not continuously enrolled for at least 12 months post-index date were excluded and may have had different outcomes compared from those patients who met continuous enrollment criteria. Also not captured are refugees on the Interim Federal Health Program and patients with no OHIP who reside in Ontario (e.g., international students). These patients may make up a significant minority being treated for their SCD in Ontario. Comparison of results presented in terms of proportion of patients in the follow-up period should be interpreted with caution because of the variable length of follow-up between patients and matched controls.

The study period spanned from January 1, 2010, to December 31, 2021, and thus includes a portion of time during which the COVID-19 pandemic occurred. We are unable to directly assess the potential impact of COVID-19 on study outcomes. However, given the relatively limited duration of the pandemic within the overall study period, we anticipate its impact on the results was minimal.