Costs of bleeds relating to acute coronary syndrome therapies

Abstract

Objective: To estimate the costs to manage selected types of bleeding complications in patients with acute coronary syndrome (ACS) treated with antithrombotics, including antiplatelet, and fibrinolytic therapies: decrease in haemoglobin >3 g/dL, puncture site, and bleeding requiring transfusion of blood products.

Methods: Retrospective chart reviews of ACS patients experiencing decrease in Hb, puncture site, or bleeding requiring transfusion were conducted in hospitals in France, Germany, Italy, Spain and Sweden to determine the total length of stay (LOS), stay post-bleeding, procedures to identify the extent of bleeding, number and types of transfusion products, patient demographics, reason for hospital admission and disposition upon discharge. Country-specific costs were applied to estimate the average cost per patient by type of bleeding event.

Results: Records of 158 ACS patients with the bleeding complications were examined (26.9% decrease in Hb, 35.5% puncture site, and 37.8% transfusion). The average LOS was 10.6 days for decreasing Hb, 7.7 days for puncture site bleeding, and 11.3 days for patients receiving transfusions due to these bleeding events. The average costs per patient ranged between €3,986 and €10,252. The LOS ranged widely both within and across countries.

Limitations: The study has a small sample size and costs are confounded by co-morbidities.

Conclusions: The estimated costs for decreasing Hb, puncture site bleeding and bleeding resulting in transfusions provide a tool for researchers to conduct economic analyses of bleeding events associated with therapies for ACS.

• acute coronary syndrome
• bleeds
• costs

Introduction

Patients presenting to hospital with acute coronary syndrome (ACS) including unstable angina (UA), non-ST-segment elevation myocardial infarction (NSTEMI), or with ST-segment elevation MI (STEMI) are at high risk of recurrent ischemic events both in the short term and the long term. Aspirin is a mainstay of therapy and reduces the risk of stroke, myocardial infarction (MI), or vascular death by about 25%[1], but additional therapies such as antithrombotics, including antiplatelet, and fibrinolytic therapies are added to improve patient outcomes by reducing mortality and recurrent MIs[1–7].

These antithrombolytic treatments share, to varying degrees, bleeding as a common adverse event. The rates of bleeding events are reported in clinical trials for antithrombolytic therapies, and the most commonly occurring (for example, see Yusuf et al, 2006²) are decrease in Hb >3 g/dL (decrease in Hb), puncture site bleeding and need for transfusions. Other bleeding sites such as intracranial haemorrhage, retroperitoneal haemorrhage, and fatal haemorrhage are rare and are not included here.

The healthcare resource use and costs due to antithrombolytic-related bleeding events reported in clinical trials have not been clearly described. Relevant bleeding costs are needed to inform economic analyses for the antithrombolytic drugs used in managing ACS patients. The purpose of this paper is to estimate the costs associated with resources to manage in-hospital bleeding events related to the antithrombolytic therapies used in ACS patients across selected countries and institutions using retrospective chart review. The findings can be used to estimate the cost impact to hospitals for managing the bleeding events and to estimate the cost offsets of avoiding such events.

Methods

Hospitals representative of those participating in ACS drug trials were invited to participate in a data collection exercise to identify the resources used to manage bleeding events in ACS patients (UA, STEMI, or NSTEMI).

Each site identified records for ACS patients who were treated in hospital with antithrombotics and had a bleeding event during the hospitalisation. Bleeding events included decrease in Hb > 3 g/dL (amount based upon the OASIS trial bleeding event definition, data on file GSK), puncture site bleeding, and transfusion of blood products, but without a major haemorrhage. The sample and sample size were based on convenience. The sites started with the most recent qualifying patient records and went back in time to identify more patients. Participating hospitals were selected based upon their interest and willingness to collect the relevant data. Each site searched for any type of bleed among ACS patients, but data are presented here only for the three more common bleeding events.

Information collected included admission and discharge dates, age, gender, reason for hospital admission, disposition at discharge, ward type, date and type of bleed event, procedures and laboratory tests relating to the event and number/type of products transfused. The study was considered a patient audit in most sites, and ethical approval was not necessary. No patient identifiers were recorded or sent out of the hospital.

The total length of hospital stay (LOS) was calculated from the admission/discharge dates, and the LOS post-bleeding event was calculated from the date the bleed was recorded until discharge. Only the LOS at the participating hospital was captured.

In some patients more than one bleeding type was identified, but the analysis was conducted on a per-patient basis. Patients with transfusions were classified as transfusion patients regardless of occurrence of puncture site or decreasing Hb bleeding. Patients with both decreasing Hb and puncture site bleeding were classified as decreasing Hb patients for the cost analysis.

Costs for hospital days were obtained from the latest available national tariffs and inflated to 2007–8 euros, however, French hospital costs (€735.81) were obtained from a publication[8]. German hospital per diem was a weighted average (€538.41) from DRG F52A and F52B for STEMI and NSTEMI and F58A and F58B for unstable angina using DRG browser[9]. Hospital daily costs were €653 for Spain (obtained from a privately held database[10]), €747.30 for Italy[11] and €607.43 for Sweden once converted to euros[12].

The patient data and resource use were reported by country and pooled across all sites to estimate the LOS according to bleeding types. Country unit costs were applied to country-specific resources and to pooled resources to estimate the cost per patient categorised by bleeding types.

Results

Twelve centres located in France, Germany, Italy, Spain, and Sweden agreed to participate in the study, and together they abstracted 158 patient records for this study. Workload constraints limited the abstraction of additional records.

Patient demographics were not different across countries. The average patient age was 73 (range 44–99) years, and 56.8% were male. The majority of patients had been admitted to hospital due to a STEMI (59%), followed by NSTEMI (32%), UA (9%), and MIs (1%) not otherwise specified.

The types and distribution of bleed complications by country (Table 1) indicate that of the 158 complete patient records, 27.8% were classified as decreasing Hb or decreasing Hb plus puncture site bleeding, 34.8% were puncture site only bleedings, and 37.4% received transfusions. There were 43 patient records with two or more bleeding classifications identified. The numbers are presented in Table 1 and the cost analyses are conducted on the 44 decreasing Hb plus puncture site patients together, the puncture site alone patients (55) and all patients with transfusions grouped together (59) due to the small sample sizes for the individual groups.

Table 1.  Patient bleeding event types by country.

Type of bleed	France	Germany	Italy	Spain	Sweden	Total	%
Decreasing Hb	14	2	3	13	0	32	20.3
Decreasing Hb + puncture site	1	10	1	0	0	12	7.6
Puncture site only	4	30	5	0	16	55	34.8
Transfusion only	11	0	5	10	2	28	17.7
Transfusion + decreasing Hb	4	1	4	0	2	11	7.0
Transfusion + puncture site	2	0	9	0	2	13	8.2
Transfusion + decreasing Hb + puncture site	3	2	2	0	0	7	4.4
Total patients	39	45	29	23	22	158	100

The majority of patients were treated in cardiology-related units. In France, the majority were managed in cardiology intensive care units. In other sites patients were only rarely treated in or transferred to intensive care units. The data indicate that antithrombotics and antiplatelets were the most commonly used therapies. Few fibrinolytic therapies were reported. Data are not analysed based upon the type of therapy as we are interested only in the resources to manage the bleeding regardless of the drug which may be related to its cause.

The procedures to diagnose the extent or location of bleeding reported for the total sample of bleeds were MRI, CT scans, endoscopy, and ultrasound. There were 103 ultrasound scans, 20 CT scans, one MRI, and five endocospy procedures reported across the 156 patient records. The average number of procedures per patient from the pooled patient data is presented (Table 2), but these procedures are not costed in addition to the daily hospital price, since they would most likely be included in that tariff.

Table 2.  Average number of procedures and transfusions per patient by bleeding type.

Procedure/ transfusion	Decreasing Hb + decreasing Hb + puncture site	Puncture site only	Any bleed requiring transfusion
CT scan	0.1	0.1	0.2
MRI	0	0	0.02
Ultrasound	0.5	0.8	0.5
Endoscopy	0.1	0	0.004
Transfused RBC	0	0	2.63
Transfused FFS	0	0	0.25
Transfused platelets	0	0	0.15

Transfusions were usually with packed red blood cells (RBC); however, fresh frozen plasma (FFP) and platelets were provided to some patients. The average number of units per transfused patient was 2.63 but it varied across countries. In Germany and Sweden, the average number of units (including RBC, FFP, and platelets) was 5.7 per transfused patient. The average was 2.4 for Italy and 3.0 in France. Spain had 1 unit per patient transfused. These three countries only reported transfusing RBCs.

The average total LOS and range (Table 3) is shown by the three categories of bleeding by country and across all countries. The last two columns in Table 3 show the average number of days from the reported start of the bleeding until discharge as an estimate of the days relating to the bleed. The table shows the average per-patient LOS and range by country and the overall average across all reported patients with the specific bleed type. The days in hospital after the start of the bleed ranged from 0 days (patient discharged regardless of reported bleed) to 39 days. An analysis of the four patients with 0 days in hospital post-bleeding date found that one was discharged to another hospital, two were discharged home and the disposition of the fourth is unknown. A proportion (28%) had bleeding events on the first day of hospital admission, but the range of hospital days after the bleed date among these patients was 1–23. Of patients (three) with only 1 day hospital stay and a bleeding event, two were discharged to home and the disposition of one is unknown. Across all the patients, 58% were discharged to home, 30% to another hospital, 10% unknown and 2% died in hospital. There were no trends identified by type of bleeding, except that three patients who died in hospital had transfusions.

Table 3.  Average total LOS and LOS post-bleed in days by bleed type.

Country	Patients (n)	Average total LOS (median)	Total LOS range	Average LOS post-bleed date (median)	Bleed LOS range
Decreasing Hb + decreasing Hb with puncture site bleeding
France	15	9.1 (7)	2–30	4.5 (3)	0–16
Germany	12	7.5 (8)	2–16	6.3 (7)	2–11
Italy	4	9.8 (9.5)	6–14	6.0 (4)	3–13
Spain	13	15.4 (16)	3–41	12.8 (14)	1–40
Sweden	0
Overall	44	10.6 (8)	2–41	6.9 (6)	0–40
Puncture site only
France	4	7.5 (8)	3–11	3.5 (2)	1–9
Germany	30	8.1 (7)	1–26	6.4 (6)	0–13
Italy	5	9.4 (7)	5–21	5.2 (5)	4–7
Spain	0
Sweden	16	6.6 (6)	3–16	4.6 (4)	0–16
Overall	55	7.7 (7)	1–26	5.6 (5)	0–16
Any bleeding requiring transfusions
France	20	9.6 (8.5)	2–23	7.0 (5.5)	1–22
Germany	3	16.7 (23)	4–23	13.0 (14)	4–21
Italy	20	9.5 (9)	3–18	7.5 (6)	3–17
Spain	10	14.8 (11)	1–44	12.3 (8.5)	1–39
Sweden	6	11.5 (9)	4–23	10.3 (7)	3–23
Overall	59	11.3 (9)	1–44	8.4 (6)	1–39

LOS, length of stay, excludes patients with missing admission, discharge or bleeding dates.

Costs per patient for the three bleeding categories were estimated for each country using its LOS cost data. The costs attributable to the total stay and the costs for the stay after the bleed start date (in brackets) are shown (Table 4). Due to differences in country-specific hospital costs, comparisons across countries are difficult and not recommended. Comparisons within country indicate generally that patients with transfusions tended to have the highest costs of those analysed, excepting in France and Italy. The main finding was that patients with these frequently reported bleeding categories are costly to manage. The average cost per patient with any of the three common bleed types, using within-country resources, was €6,071 France, €7,112 Italy, €4,570 Germany, €10,075 Spain, and €4,804 Sweden.

Table 4.  Total cost (€) per patient by type of bleeding and cost post-bleeding by country and pooled data.

Country	DH + DH + P		PS only		Any transfusion
	Country data	Pooled data*	Country data	Pooled data*	Country data	Pooled data*
France	6,720	7,586	5,519	5,666	5,629	8,315
	(2,894)		(2,575)		(4,121)
Germany	4,038	5,551	4,343	4,146	8,974	6,084
	(3,365)		(3,446)		(6,999)
Italy	7,286	7,704	7,025	5,754	7,099	8,444
	(4,484)		(3,886)		(5,567)
Spain	9,914	6,732	–	4,146	10,252	7,379
	(8,370)				(8,032)
Sweden	–	6,262	3,986	4,677	6,985	6,864
			(2,809)		(6,277)

LOS, length of stay.

DH = decrease in Hb ≥3 g/dL; PS = puncture site bleed.

Costs after bleed date in brackets () for country specific resource use.

All costs are country-specific; *using latest cost data available (2007–2008).

*Resource use across all patients in sample with defined type of bleed and costed with country-specific data.

The pooled resource use across all countries by type of bleed was costed using the individual country costs (Table 4). The pooled data provided a larger sample of patients by bleeding category from which to estimate the costs. As noted, the within-country variation in LOS was as large as the across-country variation. Treatment patterns for the bleeding events appeared to be patient-specific, not hospital-specific, thus analysing across all the patient records regardless of country was undertaken to provide a larger number of observations upon which to base the cost estimate.

Discussion

This descriptive study was conducted solely to estimate the resource use and related costs associated with the frequently reported bleeding events in ACS trials: decreasing Hb, puncture site, and bleeding requiring transfusions. Hospitals from five European countries participated in the data collection of the convenience sample of ACS patients treated with one of the antithrombotics, including antiplatelet, or fibrinolytic therapies. Resources collected were LOS, procedures, and transfusions. Only the LOS was used to estimate the per hospital patient costs associated with these events to avoid double counting of resources.

The LOS varied widely both within and across countries (see the LOS ranges in Table 3). Procedures were not frequently performed and transfusions were mostly of RBCs and averaged 2.6 units per patient transfused.

The authors separated the total hospital stay into the days before the bleed date and the bleed day plus those days following the bleed date. The days after the reported bleed might be attributable to the bleed and considered to be in excess if the bleed were avoided, but this may overestimate the excess days (and costs) associated with the bleed. The patients included in this record review study likely had comorbidities contributing to their LOS. Although the majority were discharged home, approximately a third were discharged to other hospitals, indicating the complexity of the cases. The resource cost estimate is restricted to the LOS at the hospital where the bleeding event occurred.

The reported LOSs ranged widely between and within countries, and the upper ranges were frequently double digits. The existing DRGs for ACS patients (generally those hospitalised with MIs) for France, Germany and Italy were examined to determine whether the data observed in this study were similar to an expected LOS. DRG 05M03Z (acute MI with complications) for France had a range of 2–154 days and an average of 11.4 days. In Germany, DRG F60A (MI with severe complications) had a range of three to 25 days, averaging 12.6 days. For Italy, DRG 108 for both Lombardy (cardiothoracic intervention) and at a national level had an average of 13.15 days. Thus, the DRGs indicate that patients with complications, which likely include bleeding events, have relatively long hospital stays. The average LOS in our sample of patients ranged between 7.7 and 10.6 days for puncture site and decreasing Hb bleedings. The LOS for transfusion patients was higher, averaging 11.3 days. The findings differ from the country averages within their DRG systems, but are not out of range.

Although the sample size is small and is based upon a convenience sample, the findings provide bleeding event cost estimates where none existed previously. The data provided can help estimate the hospital costs relating to the adverse events stemming from the use of the therapies in ACS patients hospitalised for STEMI, NSTEMI, and UA events.

Conclusion

Details on the resource use by country and across the pooled patient records are provided to enable individual hospitals or researchers to estimate costs of bleeding events. Assigning costs to the estimated resources provide data for conducting budget impact and cost-effectiveness studies to compare the values of various therapies available in managing ACS patients beyond the net clinical benefit reported in controlled randomised trials to include economic findings.

Transparency

Declaration of funding: This study was funded by an unrestricted grant from GlaxoSmithKline.

Declaration of financial/ other relationships: A.B., F.E., L.D., J.G.S., D.P.G. and M.J. have disclosed that they have no relevant financial relationships. L.M. has disclosed that she is a consultant/advisor to Ely Lilly, Daichi Sankio Bayer and Astra Zeneca. R.E.B. has disclosed that she is employed by United Biosource, a company that received sponsorship from GlaxoSmithKline to conduct this study.

Acknowledgements: The authors would like to gratefully acknowledge the data extractors at each of the clinical sites for their perseverance and diligence.