Predictors of Length of Stay and Time to Readmission in Adult Patients with Sickle Cell Disease: A Single Institution Experience in the UK

Sickle cell disease (SCD) is characterised by chronic hemolytic anemia and recurrent acute clinical events. The most common cause of hospital attendance is acute pain and in our patient population it accounts for 84% of all admissions. Financial pressure has led to an interest in the factors affecting length of stay (LOS) and readmission rate (RAR). The 30 day RAR has been highlighted by the UK government as a care standard. In the UK general hospital population the 30 day RAR is 6.5% of all admissions with an estimated cost to the NHS of £1.6 billion/year. A study in the US showed RAR to be 33.4% in the sickle cell population, with a lower rate in children (23%). In SCD multiple factors have been postulated to influence RAR and LOS including patient demographics (e.g. sex and socio-economic status), and hospital variables (hospital status). This retrospective study aims to assess the clinical factors which affect LOS and RAR in the SCD population at a busy London teaching hospital.

The study group consisted of 505 adult patients who were recorded on the King's College Hospital Sickle Cell Database between 1^st of January 2009 and 31^st of December 2010. The Electronic Patient Record was examined for patient ward, dates of admission and discharge, time to hematology review and time to readmission (TTR) were calculated. Patients with SCD are primarily cared for by hematology however, out of hours, patients may be initially admitted under another medical team and then have their care transferred to the hematology team. Red blood cell units transfused and time to first transfusion were recorded for each event. Data were analyzed statistically using t-tests or Mann-Whitney-U for binary variables (e.g. sex), and Spearman's rank test for continuous variables (e.g. age). TTR was analysed as a binary (≤30 days or >30 days) and continuous variable.

The cohort of 505 patients included 299 (60%) female and 206 (40%) male. Mean age was 35 years (range 18–80). 315 patients (63%) had HbSS and 9 (2%) had HbSβ⁰, 160 (32%) HbSC and 21 (4%) HbSβ+. 207 of the 505 patients had a total of 586 admissions over the study period (mean 3 admissions/ patient, range 1–19). 156 (75%) of the admitted patients had HbSS or HbSβ⁰ (SCA), 47 (23%) HbSC and 4 (2%) HbSB+. Age of the admission group ranged from 18–80 years (mean 33). LOS ranged from 0 – 116 (mean 7, median 5) days. 45% (264/586) of all admissions could be accounted for by 7% of patients and 83% (489/586) of admissions were patients with SCA. There were 279 readmissions during the study period, (100 [36%] within 30 days) by 83 patients. 72/83 (87%) of readmitted patients had SCA compared to 10/83 (4%) HbSC. Of readmissions within 30 days 95% were by patients with SCA.

Further analysis was limited to the SCA group. Female patients had a significantly longer LOS (median 5 days) than male patients (4 days) p = 0.002. There was significant correlation between LOS and TTR (R = 0.11 p=0.03). Patients admitted directly under the hematology team or who were transferred to their care had a significantly longer LOS (5 days) than those who were never admitted under hematology (2 days) p <0.0001. Patients admitted to the hematology wards also had a significantly longer LOS (5 days) compared with those on wards belonging to other specialities (4 days) p = 0.008. Delay to review by the hematology team correlated significantly with increased LOS (R = 0.12 p = 0.02) as does delay to transfusion (R = 0.45 p <0.0001). Interestingly delay to review by the hematology team appears to delay transfusion (R = 0.23 p = 0.011). The number of transfused units also correlated significantly with both LOS (R=0.39 p <0.0001) and an increase in time to readmission (R=0.136 p=0.008).

There are a disproportionate number of admissions by patients with SCA who also had a greater proportion of readmissions, reflecting the relative severity of this condition compared to HbSC. These results also highlight that there is a small cohort of patients with relatively more severe disease who have an increased LOS and increased requirement for intervention in the form of transfusion. Delay to review and therefore the decision to transfuse also appears to increase LOS, although involvement of the hematology team does not reduce it. Increased LOS and receiving transfusion treatment do appear to be associated with a delay in time to readmission.

No relevant conflicts of interest to declare.