The medical ethics review board of the Academic Medical Center at the University of Amsterdam approved the study, but waived the need for the patient’s informed consent, as they judged the study would not have any impact on the patients’ psychological or medical integrity. Hence, no written informed consent was asked from the participants for their clinical records to be used in this study. However, patient records and information were anonymised and de-identified prior to analysis.

This study was performed in a tertiary referral university teaching hospital in a period of 7 months (March till September 2011). All 14 wards belonging to the departments of Surgery (n = 8), Paediatrics (n = 5) and Neurology (n = 1) were willing to cooperate. The numbers of patients and caregivers investigated by means of the various qualitative and quantitative methods is shown in Figure 1. At the time of the study no uniform hospital-wide discharge policy existed.

The qualitative part of the study comprised a focus group, semi-structured in-depth interviews, and non-participatory observations of the daily rounds. We used multiple methods because some reasons for discharge might not surface by mere observations. The personal interviews mainly attempted to gauge organisational reasons for discharge, while the observations during patient rounds focused more on medical and patient-related reasons for discharge. Thus, these different methods allowed triangulation of the data to enhance their completeness and veracity [14]. The focus group and the interviews were carried out during the first weeks of the daily rounds observation period. The topics list for the focus group and interviews was based upon earlier research on discharge planning [5], [6], [15], [16]. During the focus group, interviews, and observations, no other persons than participants and researchers were present who could have influenced these encounters.

For the one-hour focus group session, 7 senior nurses responsible for quality assurance within their wards met to gather and discuss reasons they perceived as relevant for discharge.

For the interviews, we aimed to recruit a purposive sample of at least two doctors and one nurse for each of the 14 wards. We approached professionals with different grades of responsibilities regarding the discharge policy in order to collect all conceivable reasons for discharge. We interviewed twice as many doctors as nurses, because doctors were primarily responsible for setting the moment of discharge. We eventually interviewed a total of 27 doctors (3 professors, 15 consultants/staff specialists, 9 residents) and 13 nurses (10 head nurses and 3 registered ward nurses).

For the qualitative and quantitative observations of the daily rounds, our study sample consisted of 426 patients, admitted to the departments of surgery (N = 230 adults; 54.0%), paediatrics (N = 142 children; 33.3%), and neurology (N = 54 adults; 12.7%). The numbers per (sub)specialty are shown in Table 1. In each department, discharge data from patients were collected consecutively during a period of 2 to 4 weeks, until a sufficient number of patients was reached. “Sufficiency” was determined by data saturation (i.e., until the observations yielded no more new insights) and representativeness (i.e., typical age and gender distribution and types of disorders for that ward or specialty).

Quantitative data were obtained by counting the reasons for discharge mentioned during the daily rounds. Semi-quantitative data were collected by means of a patient questionnaire to appreciate their satisfaction and experience with the discharge process when they were about to leave the hospital (Appendix S1). We extracted relevant questions from validated, nation-wide used questionnaires appreciating general patient satisfaction during hospitalisation [10], [17]. Some of these items were altered to address the discharge rather than the hospitalisation. Each question was to be answered on a three- or five-point Likert scale (Appendix S1).

We first piloted this questionnaire for a period of 2 weeks. Feedback from patients and a local, independent expert in questionnaire research, were used to fine-tune the questionnaire. We received feedback on the size of the questionnaire, and the readability and completeness of the discharge information.

We included patients from 8 surgery, 5 paediatric and 1 neurology wards. Details about the included patients (N = 426) and their discharge are presented in Table 1. Two-thirds of the patients were adults. One quarter of the surgical patients did not undergo surgery. The technical complexity levels of the surgical procedures for adult patients performed were: 1 (in 7% of the patients), 2 (9%), 3 (6%), 4 (27%), 5 (20%), 6 (17%) and 7 (15%), in keeping with the tertiary referral level of the hospital.

In six rounds the participants of the focus group shared their reasons for discharge. These are detailed in Table 2. The four items that emerged could be categorised into medical and non-medical reasons for discharge. The medical reasons comprised the themes “Patient-related items”, “Medical devices, laboratory tests, and medical care”, and “Physical checks”. Non-medical reasons were captured in the theme “Organisation-related reasons”.

The most frequently used discharge criteria in the various departments, defined as those used in at least 50% of the patients, are shown in Table 3. Overall, the most commonly used criteria were: clinical criteria (patient’s clinical condition, body temperature, blood pressure, oxygen saturation, removed iv-drip), discharge criteria (outpatient appointment, discharge letter, medication list) and patient-related criteria (pain management, (un)answered questions, and agreement as to discharge). Items from the themes “medical devices, lab tests, and medical care” and “other” emerged less frequently as relevant discharge criteria.

Overall, 5 out of the 11 most frequently used discharge criteria were non-medical reasons; appointment made at the outpatient clinic, discharge letter supplied, medication list provided, patient agrees with discharge, and his or her questions have been answered.

Relatively few additional department-specific criteria were found to be important, i.e., the differences in discharge criteria between the specialties were small. Some criteria were particularly relevant to a certain specialty, e.g., the need for a subsequent care institution (in Neurology, Orthopaedic or Vascular Surgery), the ability to perform activities of daily living (Orthopaedics and Vascular Surgery), or signs of an infection (GI-surgery).

In 12% (53/426) of the patients who were discharged, one or more items were scored as “red”; not yet ready for discharge (Table 4). For only 8 out of these 53 patients the reason belonged to the overall top-12 of most mentioned reasons (i.e., patient not yet clinically well and discharge letter not ready). The deciding reason(s) the patients were discharged anyway could not be traced. Some of the remaining “red flags” at discharge suggested that in some cases there may have been a gap between patients’ and caregivers’ perception regarding readiness for discharge. These “red flags” mostly concerned different interpretations of the readiness of the home situation for discharge, in which case a – sometimes temporary - solution was sought. In 4 out of the 426 patients discharge criteria could not be recorded completely because the patients died during hospitalisation; 3 on the Neurology ward, and 1 on the Thoracic Surgery ward.

A total of 269 questionnaires were returned, 68% of these were from surgical patients. Many patients had already left the hospital without having completed the questionnaire. Other reasons are given in Figure 1. The smaller group of patients who filled out the questionnaires did not differ from the characteristics of the total group of 426 patients. The results are shown in Table 5. In general, patients were satisfied with their discharge process: the overall mean marks ranged between 7.0 (Vascular Surgery) and 8.6 (Urology). The lower mark seemed to coincide with the patients’ perception of a limited amount of information they received and the little influence they had on the discharge planning, albeit that only 7 vascular patients responded. On the other hand, the Vascular and Trauma Surgery wards scored high as to the information about problems that might occur after discharge, while the parents of infants were relatively less satisfied with the information supply.

The hospital stay met the expectations of the vast majority of adult patients (79–100%), while the parents of infants were somewhat less satisfied (69%). Patient satisfaction about the patients’ influence on the length of stay was quite variable among the wards, ranging from 0% to 86%. About one third of the adult patients and nearly half of the children (or their parents) felt their personal situation and assistance needed at home was insufficiently taken into account by their caregivers before discharge. Furthermore, patients were least satisfied with the information provided about what they were allowed to do or should avoid after discharge and their involvement in the planning of their discharge.

Apparently, all relevant discharge criteria could be collected by triangulation of the various methods applied. The criterion “patient clinically well” seems a rather ill-defined catch-all term, but was used regardlessly. The relatively low number of remaining ‘red flags’ at discharge suggests that the criteria found were indeed relevant. Based on the common criteria found in this study, some organisational improvements were detected to achieve a more efficient discharge procedure. Until recently, different discharge protocols were in use at various departments within our hospital. Progress was made through the introduction among surgical specialties of the SURPASS^© checklist [22], which includes several of the most frequently used items we found here. The criteria and improvements as found in the present study will foster the incorporation of a hospital-wide (early) discharge protocol.

Although we collected all reasons mentioned for discharge until saturation was reached, our sample size may have been too small to detect rare reasons for discharge. The frequency or apparent hierarchy (as shown in Table 3) of the reasons for discharge may not necessarily reflect their importance. However, by means of the mixed methods technique, we also found the most relevant reasons for discharge, which we have included in our inventory (Table 2).

About half of the patients studied were from surgical wards. This study started at an inconvenient moment for some other departments as they were in the process of reorganisations at that time. This might hamper the external validity of our results. On the other hand, the discharge criteria were quite similar among the different specialties and seemed little age- or disease-specific. Apparently, the criteria we found among many surgical patients appear to be generic and applicable to a wide range of clinical specialties.

In our study the number of patients completing the questionnaire was substantially lower than those investigated for discharge criteria. This discrepancy was mostly due to the fact that the investigators were notified late about the patients’ discharge and patients desired to leave the hospital right away.

The hospital setting could have influenced the duration of the interviews, since professionals were approached during work hours. Yet, we believed the practical advantages of higher participation levels of professionals outweighed this disadvantage.

We did not specify in our analyses the case-mix of our hospitalised patients. This may differ from other, non-university hospitals, which may in turn limit the generalisability of our results. However, the reasons for discharge found do not seem to be specific to tertiary referral hospitals only.