Commentary on Dobrzanska L & Newell R (2006) Readmissions: a primary care examination of reasons for readmission of older people and possible readmission risk factors. Journal of Clinical Nursing 15, 599–606

On reading the title of this paper (Dobrzanska & Newell 2006) I felt that this was a timely re-examination of the phenomenon of readmissions of older people. Work in this area suffers from three problems: it tends to focus on clinical characteristics measured during the original admission period, so may not be useful in prevention of readmissions in primary care; it is frequently US based, a vastly different health care system to that in the UK and many other European countries and existing UK studies now appear rather dated given the radical changes in the UK health care system in recent years. The aims of this paper were, therefore, laudable; to focus on primary care identified risk factors has considerable potential value to the management of older people following hospital admission.

The authors carried out an observational study of admissions of older people occurring within 28 days of hospital discharge. They analysed the characteristics of 109 re-admitted patients, aged over 77 years, during the course of one year. There are number of concerns with the approach taken to data collection and analysis, which will have an impact on the interpretation of the findings. One problem is that the authors chose to look at admissions occurring within 28 days, presumably because reduction in this figure forms a component of government targets for the health service and data are therefore maintained on these readmissions. The literature on readmissions of older people is very variable in relation to the time period covered, ranging for example from 42 days (Jackson 1989) to one year (Rubenstein et al. 1984), but with many studies using 90 days (Smith et al. 1985, Victor & Vetter 1985, Phillips et al. 1987). Unfortunately, whilst this means that the relevance of this study to current policy can be clearly determined, it is less easy to relate to existing literature. The use of an age cut-off of 77 is also problematic given that most literature in this area uses a cut-off of 65.

A more important issue is that the limited description of the approach taken to statistical analysis hinders interpretation of the findings. The reporting of pilot study results can add little given the very small sample size for this part of the study. A large number of variables are identified, but not all are included in the analyses presented, presumably because correlations were initially used to identify variables of interest, although this is not clear. The analysis is rather complex, with multiple statistical tests used without correction factors. The authors themselves note this problem, but unfortunately did not address the issue further. These data could have been analysed and presented in a number of alternative ways that might have improved understanding of the findings. For example, the authors could have presented both Bonferroni-adjusted and unadjusted results, leaving the reader able to judge the possible impact of multiple significance testing. Alternatively, the analysis could have been carried out using a multivariate technique that would allow greater understanding of the relative importance of the different risk factors. These limitations would suggest that the findings of this study should be treated with some caution.

Nevertheless, this study does identify some potentially interesting associations. The authors report that the majority of readmissions were due to deterioration in the original presenting condition, a finding which is entirely consistent with previous studies (Victor & Vetter 1985). The cause of readmission is not analysed further, although it is noticeable that the other reasons listed are consistent with complications of hospitalization or immobility. Other findings, for example that more patients were admitted in late afternoon and early evening and in winter months, are also consistent with previous literature. The study found an interesting association between longer index hospitalization stays and a longer time to readmission. A complex analysis of the influence of living arrangements indicated that patients living alone had a longer time to readmission and that living in care was associated with the shortest time to readmission. Length of readmission stay was longer for people discharged to their own home and those with increased social services input following the index admission.

The approach to the analysis has resulted in findings that do not fulfil the original aims of the study. Little is learnt about the reasons for readmission, other than that the majority were due to deterioration in existing medical conditions, a fact already known from the literature. Of the other reasons cited, most were consistent with complications of hospitalization or immobility. Unfortunately, this study does not explore the effect of original length of stay on the likelihood of deterioration in existing condition, which might be considered more of a risk if the original hospital stay was short. Conversely, it does not explore the impact of longer stay, or greater disease severity, on readmissions for complications related to immobility. Given increasingly short hospital stays it might be necessary for practitioners in primary care to focus their energies on prevention of deterioration and complications that were once in the realm of acute care nurses. Analysis of these factors might have pointed to potential practice changes.

The majority of the statistical analysis focuses on the identification of readmission risk factors. However, if the primary intention was to investigate the risk factors for readmissions it might be expected that the comparisons would explore the influence of patients’ characteristics on the risk of readmission. As this paper does not provide data from a comparison group, i.e. those who were not readmitted, the ability to explore risk factors is limited. Although this paper identifies possible risk factors for future investigation the extent to which, these features might predict readmission is not clear. The paper does indicate some potentially important issues, particularly that people with higher levels of support, whether at home or in care settings appeared to be readmitted more rapidly and to have longer readmission stays. This only serves to highlight a well-documented criticism of using readmission rates as a performance indicator; there is no indication that these admissions were inappropriate. It is possible that, as hospital length of stay shortens, clinical problems that would previously have occurred in the acute hospital setting now occur in the community. Where a patient lives alone or lacks other support such problems may go unnoticed; it is questionable whether a longer interval to readmission in this context is desirable. Perhaps more interesting is the finding that, when complex care packages are agreed to in the original admission, patients’ readmission length of stay is longer. This may well indicate the difficulties in maintaining a care package in a ‘state of readiness’ for discharge, and it offers a potential avenue for further investigation.

In conclusion, this study, despite a number of methodological limitations, has provided a timely update to our knowledge in relation to readmissions of older people. However, I could not help but feel that in some ways this paper represents a lost opportunity to address some important issues.

Whilst many findings support those of previous studies there are some interesting indications of areas for future investigation. Of particular importance is the potential association between a number of social factors including care package provision and the time to, or length of, readmission. These may prove useful in identifying at-risk patients and in modifying services to prevent readmissions, although social care issues are likely to be beyond the scope of influence of most health care practitioners. In carrying out future work in this area, it should be remembered that readmissions may not necessarily indicate a poor outcome, but they may be an indicator of a service that is meeting patients’ needs appropriately. Further work is needed to understand fully the implications of these findings, and that work should address the appropriateness of readmissions in addition to their causes.