READMISSIONS

Aim: Measure the quality of hospital discharges.

Cases: A readmission is considered as potentially avoidable if a patient is readmitted while this was not foreseen at the time of release. In practice, three requirements have to be met. First, the readmission must occur within the 30 days after the previous hospital discharge. Second, the readmission should be motivated by a main readmission diagnosis corresponding to at least one diagnosis already known during the previous hospital stay or to a complication. If this condition is not fulfilled, the readmission is considered unavoidable and related to a new affection. Third, the readmission should be unforeseen at the time of the previous hospital discharge. If the readmission corresponds to surgery after a first investigation stay, to a delivery after a first stay for a pregnancy monitoring, or to a chemotherapy for cancer, it will be considered as unavoidable because it was foreseen. Potentially avoidable readmission in a different hospital is included and is identified by a computerized algorithm.

Eligible population: The analysis applies to the domain of acute somatic inpatients, at risk of readmission. Psychiatric, geriatric and rehabilitations units are excluded if there are assigned to Swiss statistics codes M500, M900, M950, M990 (only if the average length of stay exceed 10 days). Stays are also excluded if the main SQLape category correspond to a psychiatric disease (depression; psychosis, hallucination and delirium; other psychiatric disorder) without somatic co-morbidities, or if they include a rehabilitation or palliative care diagnosis (ICD-10 codes: Z50, Z54, Z515 or CHOP procedure codes : 938A or 938B). Stays with patients transferred to another hospital (Swiss variable 1.5.V03 = 4, 5, 6, 44, 55, 66) or terminated by a death (variable 1.5.V02= 5) are excluded because they are not at risk of readmission. Healthy newborns and newborn with minor disorders are also excluded, since their readmission do not depend on the quality of the preparation of the discharge. Some hospitals have achieved the shift towards day surgery, with a higher proportion of severe surgery than others. To ensure the comparability of hospitals, we exclude hospital stays candidates for day surgery. Hospitalizations for sleep apnea (ICD-10 main diagnostic code: G473) were excluded for similar reasons. Patients living in other countries are excluded, because it is highly probable that an eventual readmission would not occur in Switzerland.

Output files: The results are given globally in Readmission.xlsx Excel file. Detailed results are given per hospital stay in Eligible discharges.txt file. Stays excluded from eligible discharges are provided in the Exclusions_Readmission.txt file. General information about SQLape output files can be found here.

Interpretation: A potentially avoidable readmission corresponds to a readmission, which was not foreseen at the moment of the previous discharge. In this sense, it is always a bad surprise, and undesirable event. But there are many possible causes of this readmission, some of them being often avoidable (***, see below), others sometimes avoidable (**), and other mostly unavoidable (*).

Strength of the indicator: Most of the potentially avoidable readmissions correspond to unexpected events and, thus, provide an interesting outcome indicator. It is an essential complement to cost and length of stay indicators, to detect the presence of a potential excessive financial pressure.

Limitations: The indicator has two limitations. One is related to the delay to obtain the indicator. Indeed, the indicator relies on the information about external readmissions (occurring in other hospitals), which is only available about one year later in the Swiss hospital medical statistics. However, in a first time, hospitals can estimate this external readmission rate (about 17% in average) from previous year results to analyze upward or downward trends.

The other limitation is related to the difficulty to assess the causes of readmissions. Some causes can be suggested from other indicators (premature discharge, complications, unjustified readmissions) and deduced from other text files (LS₁<LS₀ in the index hospitalization or CP₁ = 1 or UJ₁=1 in Analysis.txt or Eligible_discharges.txt files). Some other causes might be clarified by reviewing data from the medical statistics (Review). Indeed, detailed information is often necessary to control whether ambulatory care was appropriate and delivered timely. It may be tempting to explain the readmissions by the natural evolution of the pathology, but the question is whether it could have been mitigated.

False positives usually represent 4% of cases. The aim of the case review is not to avoid all potentially avoidable readmission, but to help to reduce the excess (difference between observed and expected rates). Reviewing cases when the excess is weak (less than 15-20%) might be frustrating, especially if readmission causes are scattered, which complicates the implementation of improvement measures.

Scientific validation:  Halfon P, Eggli Y, van Melle G, Chevalier J, Wasserfallen JB, Burnand B. Measuring potentially avoidable hospital readmissions. J Clin Epidemiol 2002; 55:573-587.

Halfon P, Eggli Y, Prêtre-Rohrbach I, Meylan D, Marazzi A, Burnand B. Validation of the potentially avoidable hospital readmission rate as a routine indicator of the quality of hospital care. Medical Care 2006;44(11);972-981.