Re: Readmission rates
Joseph P Drozda Jr’s editorial1 synthesises the findings of two linked studies on readmissions to hospital, and concludes that in spite of valuable studies such as these ‘our understanding of the causes of readmissions …… remains rudimentary’.
Both studies were carried out in the US, but this is not a uniquely American problem and it is important that we address these questions in the UK. Financial penalties for 30-day readmissions were proposed in the UK in 2010 and existing models that aimed to predict readmission within a year were revised to predict readmission within 30 days.2 The UK population is ageing and it is essential that the drivers of hospital admission and readmission are understood. This will only be achieved if further research on hospital admissions focuses on patients and their individual level risk factors: Drozda reminds us that ‘patients are readmitted and not diagnoses’.
The NHS routinely collects Hospital Episode Statistics (HES) and it is clear that admission and readmission rates both rise with age3;4. However these routine data do not allow investigation of what happens to individual patients - needed if the causes of admission and readmission are to be fully understood. The UK does however have a rich resource of cohort studies involving older people with detailed characterisation. We have therefore exploited data linkage technology to bring together data from the Hertfordshire Cohort Study (HCS)5 with HES data to identify the causes of admissions and readmissions among older people in the UK
The data comprise a comprehensive baseline (1999-2004) summary of the social, lifestyle and clinical status of 2997 community-dwelling men (n=1579) and women (n=1418) aged 59–73, routinely collected information about each hospital admission they experienced during the following decade and the date and cause of death for the 275 people who died.6
Our study has many strengths. First, admissions histories can be dovetailed with mortality records; Donzé et al7 do not appear to have considered deaths. Secondly, our database includes men and women who did not experience admission during follow-up; they are an important control group for comparison with individuals who did. Thirdly, a wide range of baseline data was collected prospectively by a team of research nurses and doctors according to strict research measurement protocols. Such data are not available from routine sources.
In response to Drozda Jr’s editorial we have conducted a preliminary analysis of readmissions among the men and women. We defined three binary outcome variables to classify whether or not a study participant was: ‘ever admitted or died’; ‘ever readmitted within 30 days or died’ and ‘ever readmitted as an emergency within 30 days or died’. We describe these variables and their associations with burden of co-morbidity at baseline as indicated by the number of systems medicated (identified by coding all prescription and over-the-counter medications taken by study participants according to the British National Formulary).
The cohort accumulated 8741 admissions during the follow-up period, although 829 individuals experienced none. The table shows that 1198 (75.9%) men and 989 (69.7%) women were ‘ever admitted or died’. 458 (29.0%) men and 289 (20.4%) women were ‘ever readmitted within 30 days or died’, of whom 313 men and 180 women were readmitted as emergencies.
The table shows that greater baseline burden of co-morbidity was strongly associated with higher proportions of men and women experiencing hospital admission; readmission within 30 days; and emergency readmission within 30 days, or death. For example, 48% of men and 45% of women who had four or more systems medicated experienced a readmission within 30 days or died; only 21% of men and 13% of women who had no systems medicated went on to experience a readmission within 30 days or death.
These simple analyses confirm the substantial burden of hospital admission and readmission even among young-old community-dwelling men and women. Moreover, they demonstrate the potential of the HCS database to identify important patient-level characteristics that are predictive of admission and readmission and to suggest avenues for intervention to reduce hospital admissions and improve care.
Drozda Jr has called for additional insights into the drivers of hospital readmission; using the wealth of data available in the HCS study we intend to provide them.
(1) Drozda Jr JP. Readmission rates. BMJ 2013;347:f7478.
(2) Billings J, Blunt I, Steventon A, Georghiou T, Lewis G, Bardsley M. Development of a predictive model to identify inpatients at risk of re-admission within 30 days of discharge. BMJ Open 2012;00:e001667.
(3) Office for National Statistics. General Lifestyle Survey - Health tables 2009 Table 7.28 Trends in inpatient stays in the 12 months before interview by sex and age, 1982 to 2009. 2011.
(4) Robinson P. Hospital readmissions and the 30 day threshold. 2010. CHKS.
(5) Syddall HE, Aihie Sayer A, Dennison EM, Martin HJ, Barker DJP, Cooper C. Cohort Profile: The Hertfordshire Cohort Study. Int J Epidemiol 2005:34,1234-1242.
(6) Simmonds SJ, Syddall HE, Walsh B, Evandrou M, Dennison EM, Cooper C et al. Understanding NHS hospital admissions in England: linkage of Hospital Episode Statistics to the Hertfordshire Cohort Study. Age Ageing. In press 2014.
(7) Donze J, Lipsitz S, Bates DW, Schnipper JL. Causes and patterns of readmissions in patients with common comorbidities: retrospective cohort study. BMJ 2013;347:f7171.
Competing interests: No competing interests