The access-block effect: relationship between delay to reaching an inpatient bed and inpatient length of stay

This was a retrospective cohort study of all patients admitted through the ED to an inpatient bed at the Canberra Hospital, Australian Capital Territory, during 1999. The hospital is a 500-bed mixed adult and paediatric tertiary hospital serving a population of about 500 000.

During 1999, there were 48 430 presentations to the ED, with 11 906 admissions eligible for the study. Patient records were consulted to resolve discrepancies between the source databases for 334 admissions (2.8%).

Mean total ED time was 4 h 28 m. Access block occurred in 919 admissions (7.7%), and the access-block group had a mean total ED time of 10 h 30 m, compared with 3 h 58 m in the no-block group.

Mine is the first study to demonstrate a relationship between access block in the ED and a measure of patient outcome outside this department. Patients who experienced access block had a mean inpatient LOS 0.8 days longer than those who had no access block. This amounted to 19% of the average LOS in the first 10 days after ED presentation. Considering the average additional 6.5 hours that access-block patients spent in the ED, this is more than 24 hours extra in hospital from ED presentation to discharge. Altogether, the excess inpatient LOS (compared to average no-block inpatient LOS) for access-block patients amounted to over 700 bed-days per year.

The access-block effect on LOS was relatively independent of the severity of patient illness and diagnosis, but appeared to be greatest in patients who arrived on inpatient wards outside office hours.

Proving an association does not prove cause and effect. Although the available data do not allow the reasons for the access-block effect on LOS to be identified, they allow some possible explanations to be examined.

The obvious possibility that access-block patients were "sicker", requiring longer "work-up" in the ED and then longer duration of care on the ward, appears to be refuted by the results. The access-block effect on LOS was seen in all triage categories except category 1 (the sickest patients), and also at all ages over 40 years, despite considerable differences in mean LOS between subgroups. Furthermore, in the subgroup analyses, the rate of access block was generally not highest in the group with the longest LOS.

It is also possible that some feature of presentation, such as time of arrival in the ED, causes both longer ED time and longer inpatient stay (eg, because of difficulty in accessing investigations). However, the access-block effect persisted across a wide range of patient characteristics and most times of arrival in the ED. Although both access block and LOS increased in June, this coincided with a period of recognised hospital overload.

A third possibility is that when it is difficult to access inpatient beds the mix of patients admitted may differ. Patients with minor conditions requiring only a short stay may be more likely to be treated with alternatives to inpatient admission (eg, discharge, admission to the ED only, or transfer elsewhere). This would account for the relative deficiency of patients with LOS of one day in the access-block group, but not for the excess of patients with LOS over seven days. Also, the low rate of access block (7.7% of all admissions from the ED) makes it unlikely that there were many times when large numbers were managed using alternatives to admission.

Fourthly, patients who experience access block may receive different treatment from the no-block group. For example, care during a prolonged stay in the ED may differ from that in an inpatient ward, or patients may be more likely to be admitted to outlying rather than "home" wards, where staffing or organisational issues may prolong LOS. Data on patient diagnoses were limited, but the access-block effect appeared to occur across a range of diagnostic categories. This possibility requires further investigation.

There were insufficient data to be certain if the relationship between LOS and total ED time was linear with outliers, or U-shaped with a gradually increasing "tail". An initial U-shape is credible: patients with very short ED times would not have been investigated in the ED, so are likely to have been "sicker" (eg, those transferred rapidly to theatre) than those with the "normal" ED time of 3–4 hours. Whatever the exact nature of the relationship, these data support the use of an ED time of eight hours as the threshold for access block — patient groups with ED time over this level had mean LOS exceeding that of all groups with shorter ED time.

A final possibility is that the access-block effect may be a function of the inpatient rather than the ED system of care. The strong relationship between the access-block effect and time of arrival on the inpatient ward, with the greatest effect for arrival during the evening, night or weekend shift, is the best evidence for this. It suggests that ward functions that occur in office hours may be associated with shorter LOS. For example, access-block patients arriving on the ward out of office hours are unlikely to be reviewed immediately by the regular inpatient team.

This study is limited by the nature of the data available, which derive from a single centre and include limited clinical details. A prospective case–control study would allow better identification of the factors that contribute to the access-block effect and should consider whether patients are managed with alternatives to inpatient admission and whether they are admitted to home or outlier wards, as well as recognised diagnostic coding. If the access-block effect occurs within Diagnosis Related Groups, it would have ramifications for funding models as well as major implications for management of hospital patient flow.