Main findings of this study
This study indicates that growth in hospital activity and in costs for patients with SCD over the 10 years to 2018 has occurred at a considerably higher rate than growth in the total number of admitted SCD patients. Most of the growth has come from increasing per patient admission rates for sickle cell crises and transfusions.
When measured at constant 2017/18 Reference Costs, the cost of treating SCD patients grew from £41.9m in 2009 to £76.4m in 2018 (CAGR 6.9%). This is in line with the 0.8% per annum increase in SCD patient numbers and the 6.2% per annum increase in admissions per patient.
Our analysis indicates that the increase in SCD patients admitted to hospital over the 10 years to 2018 (9.9%, 1.0% CAGR) is consistent with the growth in the English population over the same period. Rising SCD admissions and costs are therefore principally due to rising admission rates per patient (6.2% per annum), rather than due to a disproportionate increase in numbers of patients (0.8%).
High crises patients and high transfusions patients represented a growing proportion of the SCD patient population in England. These patients are mostly high users of hospital resource and account for the majority of NHS hospital expenditure on SCD patients. In 2018, high crises and high transfusions patients represented 22% of the SCD admitted patient population, but accounted for 62% of hospital expenditures.
What is already known on this topic
A previous analysis of the HES database by Aljuburi et al reported a 58.0% increase per capita in SCD admissions (defined as admissions with a primary diagnosis of D57.0 or D57.1) in England over the period April 2001 to March 2010 (5.4% CAGR).9 This is similar to the 72.0% (6.2% CAGR) increase in admissions per SCD patient during 2009-2018 found in this study. Although there are some differences in definitions and analytical approach between Aljuburi et al and this study, it is clear that admissions for SCD have been increasing substantially over the last 20 years.
The rate of patient growth that we observed may seem low compared with the results of the NHS Sickle Cell and Thalassaemia Screening Programme which identified 3,857 newborns with results indicative of SCD in the 12 years between April 2005 and March 2017, an average of 321 per year.10 By contrast, post-2009, we observed an average of 143 children born with SCD per year admitted for the first time during the study. The Screening Programme identifies infants who, on further investigation, have clinically insignificant variants of SCD or who emigrate shortly after birth. Approximately 82% of positive screens result in an infant attending a specialist SCD out-patient clinic within 3 months of birth, rising to 98% within 6 months of birth.22 A similar picture emerges with regard to prophylactic antibiotic use, with 83% prescribed penicillin by 3 months and 97% by 6 months of age.21 It’s possible that the Screening Programme, which was implemented in England between 2003 and 2006, i.e. several years before this study period, may have reduced admissions in children with SCD, accounting for some of the gap between the Screening Programme and children born post-2009 with first admissions seen in this study.
There are no definitive data on SCD prevalence in the United Kingdom. The West Midlands Quality Review Service in their 2014-2016 peer review programme reported 12,175 SCD patients registered at centres in England.23 Dormandy et al reviewed all national SCD databases and estimated an SCD prevalence in England of 13,655 in 2017.1 Therefore, there are approximately 2,000-3,000 SCD patients who did not form part of our study because they were not admitted in the 10-year study period. This cohort likely comprises patients with clinically mild forms of SCD who may never be admitted, whilst others will likely require hospital care in the future as their disease progresses.
What this study adds
This study adds a contemporary 10-year perspective on the burden of SCD on the NHS in England, focusing on the two highest cost areas, admitted patient care and out-patient appointments. By using the ability of HES data to follow patient hospital activity over time, we have been able to select and quantify two groups of SCD patients (high crises patients and high transfusions patients) whose prevalence, resource use, and associated hospital costs have been growing rapidly and disproportionately to the overall growth of the SCD population.
Whilst HES data can inform on variability in hospital utilisation, the underlying reasons for such variability are probably mostly attributable to changes in guidelines and management strategies. The increase in sickle cell crisis admissions may be explained by a lowering of the threshold for admission for sickle cell patients caused by greater awareness and recognition of the need for acute management, and monitoring of patients with possible disease-related complications. Increases in transfusion numbers likely reflect changing clinical practice in recent years to greater utilisation of regular blood transfusions for primary and secondary stroke prevention, particularly in children, supported by the increased availability of automated exchange transfusions which have been deemed a cost-effective intervention by the NICE24. Transcranial Doppler monitoring, iron monitoring associated with regular transfusions, and regular blood counts required with use of hydroxycarbamide may all contribute to the increase in other SCD admissions (where such procedures are performed on a day case basis) and out-patient appointments.
SCD is now the most common serious inherited genetic disorder in England, however there is a distinct lack of literature on the resource utilisation implications of the condition on healthcare systems in the United Kingdom, or indeed other developed countries. Given the present burden as well as the evolving therapeutic landscape for this complex disease25, it is relevant and important to evaluate resource utilisation and cost-consequence of the present clinical paradigm on health systems and budget holders.
Limitations of this study
NHS Digital requires hospitals to audit the accuracy of their clinical coding annually and achieve 90% or more accuracy in primary diagnosis coding, and 80% or more in secondary diagnosis coding.26 Nevertheless, the quality of clinical coding in the NHS varies over time, and between trusts and disease areas. Though no specific study of coding accuracy in SCD exists, a systematic review of 32 clinical coding accuracy studies during the period 1989 to 2011 found that the median accuracy of NHS primary diagnosis coding was 80.3%, but that this had increased to 96.0% after the introduction of Payment by Results in 2002.27
We acknowledged above the existence of SCD individuals not included in this study. These missing individuals did not affect our results for hospital admission numbers and costs as they were not admitted to hospital during the study period. However, as most of these individuals were probably monitored in an outpatient setting, outpatient activity and costs may be understated.
This study has not examined the impact of cohort aging on resource usage, or of geographical variations within England. The median age of patients alive at the start of 2009 was 21.0 years. At the end of 2018, this had increased to 26.6 years. An aging cohort is likely to require more hospital care as complications of SCD manifest. Although the vast majority of SCD patients are in London, it is likely that resource usage and costs vary between hospitals.
Finally, not all costs were captured in this analysis (e.g. costs of dialysis or costs of treatment outside a hospital setting). The activity and costs of patients with a GP practice outside England who accessed hospitals in England are also excluded.