Journal of the New Zealand Medical Association, 13-February-2009, Vol 122 No 1289
The validity of readmission rate as a marker of the quality of hospital care, and a recommendation for its definition
Juliet Rumball-Smith, Phil Hider
The need for quality assessment and improvement in the health care system has gained increasing significance both internationally and in New Zealand.1 Numerous criteria exist for evaluating quality in hospital services, using various indicators, including readmission rate. This rate reflects the impact hospital care has had on the patient’s condition up to the point of discharge, and also represents the efficiency of the service; inpatient hospital care being a primary source of expense in the health system, and repeated admission representing a potential source of wasted resources.2
Readmission rate is collected easily by hospital information systems, and can readily be combined with information from other databases to assess the potential impact of different variables, and control for confounders.3 However, there is no agreed definition of readmission rate in the literature, with researchers and states employing multiple time periods, and failing to acknowledge subjects who died following discharge.
A literature search was undertaken on the Medline (1966–2008) and Embase (1988–2006) databases. Search terms included: ‘quality of health care’, ‘quality indicators, health care’, ‘patient readmission’, ‘readmission’, ‘rehospitalisation’, ‘readmit$’, ‘new zealand’. Searches were limited to English language material and articles with abstracts available.
Quality in health care—Quality in health care may be defined as ‘the degree to which health services for individuals and populations increase the likelihood of desired health outcomes and are consistent with current professional knowledge’.4p4 Campbell maintains that quality can be viewed as having two simple domains, suggesting that quality of care for an individual can be defined as ‘whether individuals can access the health structures and processes of care which they need, and whether the care received is effective’.5
However, the emphasis placed on the quality of health care by the medical profession, the consumer, and the state, is relatively recent. Adverse events and incidents of ‘quality failure’ are now widely publicised. Consumers may have higher expectations of the health care they receive, and be less trusting of their provider to deliver error-free care. Health care providers need to account for resources and ensure value and efficiency of services. Funders also want to ensure that providers are meeting their expectations, and are performing adequately compared to others.
Readmission rate as an indicator of the quality of health care—Quality indicators are only one of the tools in the ‘quality toolbox’, but are indispensible in the assessment, monitoring, and improvement of the quality of patient care. Quality indicators explored in the literature include hospital multi-stay rate, length of inpatient stay, hospital mortality rates, and complication rates.6,7
The Maryland Hospital Association’s Quality Indicator Project use over 15 measurable, discrete performance indicators upon which value can be placed to describe quality.7 The wider international community has embraced the concept of quality monitoring; hospital accreditation programmes feature in over 36 countries using various indicators to ensure a minimum standard of quality.8
Readmission rate has been used as a quality indicator in various psychiatric, surgical and medical clinical specialties.9–11 It is popular as a quality indicator amongst researchers and management, for the following reasons:
Validity of readmission rate as an indicator of the quality of hospital care—Are those who are readmitted more likely to have experienced lower quality care? And conversely, are those who receive substandard care more likely to be readmitted?
Ashton et al. performed a significant meta-analysis of 29 comparative studies, including both hospital chart and database analyses, and in each study, process-of-care elements were examined in relation to 31-day readmission rate. The authors classified the datasets according to these elements, denoting the quality of care received as being either ‘substandard’, ‘normative’ or ‘exceptional’, when compared to accepted clinical standards of care.
After review of 13 studies, an odds ratio of readmission of 1.24 (95%CI 0.99–1.57) was calculated for those patients who received substandard as opposed to normative care. Sixteen studies were assessed to calculate a summary odds ratio comparing care of relatively low quality (‘substandard’ and ‘normative’ classification) relative to care of higher quality (studies classified as ‘exceptional’) was 1.55 (1.25–1.92), representing an increased risk of early readmission of between 25%–92% for those who experienced lower or normative quality of care.
This article represents substantial evidence for the validity of this indicator, yet is limited by the relatively few studies used to develop the odds ratios. However, this reflects the criteria for inclusion defined by the authors, such that the studies assessed were ostensibly homogeneous and provided robust results.16
Further information can be gained from cohort studies and case-control studies. Weissman et al reviewed over 1700 admissions in a case-control study, assessing the rate of 31-day readmission for patients hospitalised with pneumonia and congestive heart failure in four states of the US-trained physicians reviewed clinical records and compared the care delivered to patients subjectively, and against explicit clinical criteria. They discovered there were ‘significant but relatively small, differences in initial quality of care between patients who subsequently experienced related adverse readmissions and those who were not readmitted’.19p500, findings that were present after control for demographic and clinical variables, and hospital characteristics. Ashton also performed a case-control study of over 2000 patients who had diabetes, obstructive lung disease or heart failure, reviewing quality of inpatient care and 14-day readmission rate.
Chart review was used to ascertain quality of care, according to specified process-of-care criteria. After adjustment for demographic variables and clinical severity, they discovered a similar significant association between substandard care and subsequent unplanned readmission. This study was conducted within US Veterans Affairs Hospitals, and as such may not be able to be generalised to patients who experience financial restrictions in access to hospital care.20
However, there are few studies that have examined the second question, the issue of ‘false negatives’ in the calculation of readmission rate. This proportion represents those that receive substandard care, but due to other factors (such as death, or recovery) are not readmitted. This data is very difficult to extract, as they do not enter again into the hospital recording system, and it may be difficult to access information about the health of those in the community. Although many studies ignore the contribution of these false negatives to the validity of the readmission rate, it must be acknowledged that the validity of the rate has not been proven in this respect.21
There are other threats to the internal validity of this rate. Given that the rate is made up of a numerator (those that are readmitted to hospital within a given time period) and a denominator (total number of patients discharged alive within a reference period), intervening variables, confounders, demographic, and clinical factors may impact on both these figures.
Intervening variables can be defined as those that are ‘interposed in time in the causal sequence between the proposed independent and dependent variables’.18p1539 The most significant intervening variable for readmission rate is death in the community within the reference period, as a result of substandard care.
The impact of this variable is two-fold. Firstly, these patients are technically not eligible to be readmitted, so the overall denominator used to calculate readmission rate is artificially inflated by these absent subjects. Secondly, a patient that receives substandard care but dies in hospital or the community, or has a longer initial admission due to this quality breach, will not change the readmission rate despite substandard care.13
A confounding factor is defined as ‘a third variable that indirectly distorts the relationship between two other variables’.22 The literature discusses the following confounding factors with regards to readmission rate:
The external validity of a quality indicator refers to its generalisability; its ability to yield comparable results over time, and at homogeneous but differing institutions. The public acceptance of the validity of readmission rate is demonstrated by its broad use. Governments and private health purchasers such as US company Blue Cross Blue Shield use it as an indicator of quality.31 The New Zealand Ministry of Health has recently added readmission rate to its performance monitoring project.32 However, while these institutions may use the rate to detail trends over time and monitor progress, there is limited external validity due to the multiple definitions of readmission rate employed among institutions and researchers.
‘Readmission’: Chambers and Clarke define readmission as ‘the next subsequent admission of a patient as an immediate (that is, emergency or unplanned) admission ... within a defined interval of a previous (index) discharge taking place within a defined reference period’.33p301 Ashton and Wray (1996) recommend obtaining information on the frequency of death after discharge, and state that the preferable outcome variable is “death at home or readmission within n days”. If death is the worst possible result of poor quality care, then those subjects who die within the time period should be included as part of the readmission indicator. In practice, the number of these patients is small and may change the overall indicator by a negligible amount only. However, if they are not included in the analysis then the result may be biased and produce an indicator that is an under-estimate. Accordingly, it is advised that the indicator be revised to ‘readmission or deaths’ of subjects within the specified time period.
Population: The rate is calculated by dividing the numerator (‘readmissions’) by the ‘corresponding number of patients discharged (alive) within the reference period’.33p301
Time period: Researchers in quality have employed various time periods over which readmission may occur, with intervals of between 2 weeks and 12 months used.34 As mentioned above, this variation threatens the external validity of this indicator, and makes comparison between institutions and populations difficult.
This author recommends the use of a 30-day time period:
The general consensus of the literature seems to be that a one-month time period provides a logical balance.14 This assumption is supported by another common indicator used to address inpatient quality, 30-day mortality rate.
This literature appraisal was intended to provide background information on this indicator, as such this article does not offer the same quality of information as a systematic review. However, the review notes that ‘readmission rate’ has some demonstrated internal validity and is widely used in the measurement of the quality of hospital care. It is important to recognise the limitations of this measure, and minimise the influence of possible confounding variables by way of methodological and statistical techniques. Similarly, the adoption of a consistent definition of this rate by researchers will add to its external validity and improve its generalisability. This author recommends the use of the following definition: ‘the number of patients who experienced unintended, acute readmission or death within 30-days of discharge from the index admission, divided by the total number of patients discharged alive within the reference period’.
Competing interests: None known.
Author information: Juliet M L Rumball-Smith, Research Fellow, Department of Public Health and General Practice, University of Otago, Christchurch; Phil Hider, Senior Lecturer, Department of Public Health and General Practice, University of Otago, Christchurch
Correspondence: Dr. Juliet Rumball-Smith, Research Fellow, Department of Public Health & General Practice, University of Otago, Christchurch, PO Box 4345, Christchurch, New Zealand. Fax: +64 (0)3 3643697;
Acknowledgements: The author acknowledges the financial assistance of the Health Research Council of New Zealand and the New Zealand College of Public Health Medicine; and supervisory support from Dr. Patrick Graham, University of Otago, Christchurch.
issue | Search journal |
Archived issues | Classifieds
| Hotline (free ads)
Subscribe | Contribute | Advertise | Contact Us | Copyright | Other Journals