Adverse outcomes may occur during hospitalisation, such as wound infections, pneumonia or pressure sores. They are associated with poorer health outcomes for patients and may increase costs. Many studies have studied the incidence and nature of these in-hospital adverse outcomes.1^–6 However, to what extent do adverse outcomes occur after discharge, or are discovered after discharge but related to the hospitalisation? Previous studies have shown that 14–20% of the identified adverse outcomes occur outside the hospital.78 These are relevant, since an even higher proportion than those occurring during hospitalisation seem to be preventable.9 However, since these studies used a retrospective record review to identify these adverse outcomes, and therefore depend on the (incomplete) information available in the medical records, this probably underestimates the adverse outcome occurrence.10

In the Netherlands, surgical adverse outcomes are prospectively reported as part of routine daily medical care.11 A previous study has shown that in-hospital adverse outcomes are effectively identified using such a routine reporting system.12 However, postdischarge adverse outcomes may be missed in this reporting system if these are not mentioned by patients during a postdischarge follow-up visit at the outpatient clinic. On the other hand, if most of these adverse outcomes require readmission and possibly additional surgery, these will be reported. Thus, postdischarge adverse outcomes are relevant to both patients and doctors to give complete information on possible outcomes after surgery. In addition, if factors during admission could be identified that determine the occurrence of postdischarge adverse outcomes, this may enable doctors to give better information to patients at discharge on how to avoid and deal with these possible postdischarge adverse outcomes.

The aim of the present study therefore was (1) to estimate the occurrence of postdischarge adverse outcomes in surgical patients and their treatment, based on patients interviews and our prospective reporting system; (2) to explore determinants during admission that may influence the occurrence of postdischarge adverse outcomes.

MATERIAL AND METHODS

Study participants and defintions

All 2145 surgical patients of the Leiden University Medical Centre hospitalised in 2003 were approached to participate in the present study. In-hospital adverse outcomes were prospectively recorded throughout admission by surgeons and surgical residents as part of routine medical care, according to the standards of format and definition set by a Dutch nationwide routine reporting programme.1112 The definition of an adverse outcome as used in this programme is:

... an unintended and unwanted event or state occurring during or following medical care, that is so harmful to a patient’s health that (adjustment of) treatment is required or that permanent damage results. The adverse outcome may be noted during hospitalisation until 30 days after discharge or transferral to another department. The intended result of treatment, the likelihood of the adverse outcome occurring, and the presence or absence of a medical error causing it, are irrelevant in identifying an adverse outcome.

This definition is more sensitive than the definition used in most other published studies,1^–4 as events linked to the natural history of the disease or comorbidity are also included. The advantage of the above definition is that it is less subject to discussions on whether or not this could be expected given the medical condition of the patient, and therefore less likely to result in inter-observer variability.

All in-hospital adverse outcomes are reported on special forms in the patients’ records. The presence and completeness of each patient’s form is checked weekly. Written information on adverse outcomes is checked and discussed 2 weeks after discharge, in a weekly meeting attended by surgeons, residents, interns and medical students. The purpose of discussion is twofold: (1) to assess whether the written information is understandable and correct, and (2) to analyse, interpret and conclude whether anything can be learnt from this particular case. After consensus has been reached on the correct information, adverse outcomes are entered into the dedicated software and classified by nature, location and possible determinants.

A previous study has shown that in-hospital adverse outcomes are effectively identified using this nationwide routine reporting system.12 However, given the above definition, postdischarge adverse outcomes should also be included but may be missed if these are not mentioned by patients during a postdischarge follow-up visit at the outpatient clinic. Patients were therefore interviewed 4 weeks after discharge, including questions on the possible occurrence of postdischarge adverse outcomes. The timing of the interview was chosen given the postdischarge period mentioned in the above definition. In addition, this length of time was considered to be long enough to experience postdischarge adverse outcomes, and short enough to enhance recall of these adverse outcomes.

Interview

Patients were informed about the study and its purpose during hospitalisation, and were approached for a telephone interview 4 weeks after discharge by one research assistant. The interview followed a structured format, and included questions about:

• the occurrence of adverse outcomes after discharge (yes/no and specification which type, for instance bleeding, infection);

• whether the adverse outcomes were treated by a doctor (yes/no and specification of general practitioner, specialist, or other (nursing home physician, physician at rehabilitation centre)) and which type of treatment was given;

• whether the adverse outcomes had required the patient to be readmitted to hospital (yes/no and specification of date of readmission);

• whether an operation had occurred during the readmission (yes/no and specification of date of operation).

The research assistant explained the definition of an adverse outcome to patients, including some examples, to be consistent with the definition used in the nationwide routine reporting programme. To check whether all adverse outcomes had been covered, she asked the patient specifically for any GP visit or consult at the outpatient clinic that had taken place after discharge and drugs that were prescribed by a physician. If such a visit or consult had taken place or drugs had been prescribed, the reason for this visit or prescription was asked to assess whether it concerned treatment of an adverse outcome, including a specification of the type of treatment which was given. In the first instance, the patients themselves were contacted to be interviewed; if this was not possible, the reason for this was registered, and a next of kin was interviewed if possible (eg, partner, parent, child or care taker). Patients were called on several occasions (at least eight to 10 attempts), at different times of the day (morning/afternoon/evening).

Statistical analysis

We calculated the number of patient admissions with postdischarge adverse outcomes, the number with postdischarge adverse outcomes resulting in readmission and/or in additional surgery. Subsequently, to identify possible determinants, we compared patients with and without postdischarge adverse outcomes on differences in age, percentage males, percentage of high-risk patients (high risk being defined as ASA-class13 3 or higher at the first operation), percentage of high technical surgical complexity (defined as complexity class 5 or higher on a scale from 1 (simple) to 7 (complex or difficult) as developed and used by the Association of Surgeons of the Netherlands14), the occurrence of in-hospital adverse outcomes and the average duration of admission. For these comparisons, t tests were used in case of continuous variables (age, duration of admission) and chi-square tests for the other categorical variables.

Logistic regression analysis was then used with postdischarge adverse outcomes as the dependent variable, to identify the independent effect of each of the variables adjusted for differences in all the other variables. As independent variables, we included age above 65 at admission, gender, high-risk patients (defined by ASA class as described above), complex surgical procedures (as described above), in-hospital adverse outcomes and admissions of 10 days or longer. In addition, we included surgery group as an independent variable, since these are known to differ in adverse outcome probability.14 We distinguished the following large surgery groups, based on the first operation of admission, as defined by the Association of Surgeons of the Netherlands: head–neck surgery, vascular surgery, digestive tract surgery, surgery on musculoskeletal system, breast surgery, procedures from other groups, procedures from multiple groups (since these mostly involve complex surgery and thus have higher occurrences of adverse outcomes). The same analysis was performed with postdischarge adverse outcomes resulting in readmission (yes/no) as the dependent variable, to look for possible determinants for readmission.

All p values ⩽0.05 were considered to be statistically significant. Both the odds ratios from the regression analyses and 95% confidence intervals were reported.

RESULTS

A high response rate was achieved: 1960 of the 2145 patients (91.4%) agreed to participate in the present study. The most common reason for non-participation was that the patient could not be reached (172 cases, 93.0%). This was due to the patients’ death after discharge (19 cases), patients not having a telephone (67 cases) and patients who could not be reached after 10 attempts (86 cases). Of the patients who were reached, reasons for non-participation were a language problem (seven cases) and no time or did not want to participate (six cases).

Table 1 shows the differences between participants and non-participants. Non-participants are slightly younger and have a longer duration of admission (on average 3 days). No differences were found in the percentage of high-risk patients, complex patients or the occurrence of in-hospital adverse outcomes.

Postdischarge adverse outcomes and their treatment

Postdischarge adverse outcomes occurred in 487 patients (24.8% reporting 554 adverse outcomes), with 427 patients (21.8%) reporting one adverse outcome and 60 patients (3.1%) reporting multiple adverse outcomes. A total of 147 (7.5%) patients were readmitted to hospital to deal with these adverse outcomes on average 17 days after discharge from the hospital. For 63 patients (3.2%), additional surgery was required during this readmission.

In total, 554 postdischarge adverse outcomes were reported, of which most (76.2%) were treated by a specialist (table 2): 165 (29.8%) during readmission and 257 (46.4%) at the outpatient clinic. Postdischarge adverse outcomes were mostly infections (39%), symptoms without diagnosis such as pain (17%), and functional disorders (12%) (table 2). These are similar to the adverse outcomes commonly occurring during hospitalisation, where infections and dysfunctions also account for about half the adverse outcomes. The exception is the category “symptoms without diagnosis” that occurred less frequently during admission (4.6%). Most patients (72.7%) claimed they had mentioned the postdischarge adverse outcomes during follow-up visits at the hospital. However, they more frequently claimed to have mentioned the adverse outcomes treated by specialists (85.1%) than those treated by GPs (30.4%), which is likely to cause under-reporting of adverse outcomes in our in-hospital routine reporting system.

Of the 422 postdischarge adverse outcomes treated by a specialist, 399 were treated in the Leiden University Medical Centre and 23 in other hospitals, with respectively 153 and 12 adverse outcomes requiring readmission. Of the 399 postdischarge adverse outcomes treated in our hospital, 40 (10.0%) were also reported in our routine reporting system. Of the 153 adverse outcomes treated with readmission to our hospital, 38 (24.8%) were reported in our routine reporting system. The 359 adverse outcomes that were missed in the reporting system were mostly treated by administering medication (144 cases, 40.1%). Of the missed adverse outcomes, 115 (32.0%) resulted in readmission, and 48 (13.4%) were treated by additional surgery.

Determinants of postdischarge adverse outcomes

Table 3 shows that patients with postdischarge adverse outcomes were older, had relatively more high-risk patients and had more complex surgical procedures. These patients more often had experienced in-hospital adverse outcomes and on average had a longer hospitalisation (table 3).

Since many of these variables are related, for instance older patients and high-risk patients have higher probabilities on in-hospital adverse outcomes, multivariable regression analyses were carried out to disentangle the independent effect of each of these variables. Table 4 shows the results of these analyses, separately for all postdischarge adverse outcomes and those postdischarge adverse outcomes that required readmission. It is shown that in-hospital adverse outcomes and complex surgical procedures increase the probability for all postdischarge adverse outcomes and readmissions. In addition, occurrences of postdischarge adverse outcomes were nearly twice as high for breast surgery and 1.5 times as high for procedures from multiple groups. These postdischarge adverse outcomes after breast surgery were mostly infections (37%), accumulation of fluids (oedema) (41%) or pain complaints (11%), which are common complications of breast surgery mostly treated at the outpatient clinic. Men more frequently experienced adverse outcomes that required readmission, which might reflect that men tend to visit a doctor when the problem is more severe (in comparison with women) (table 4). Long hospitalisations increased the risk of readmission, independently from the occurrence of in-hospital adverse outcomes, which probably reflects that postoperative recovery was difficult in these patients. Patients undergoing vascular surgery or surgery on the musculoskeletal system less often had postdischarge adverse outcomes that required readmission.

One of the explanations for the occurrence of postdischarge adverse outcomes may be that these patients were discharged too early. For instance, if patients were discharged right before or in the weekend, their own treating physician may not have been on duty, but someone less familiar with the patient. However, adding a variable “discharge in the weekend” to the model did not have a statistically significant effect on either one of the outcome measures (odds ratio 1.16 (0.85 to 1.60) for all adverse outcomes and 0.84 (0.47 to 1.50) for adverse outcomes that required readmission), nor did it change the results of the other variables in the model. The same was true for discharge on any other day of the week.

Adding interactions to the regression model—between long hospitalisations, in-hospital adverse outcome occurrence, high-risk patients, old patients and complex surgery—did not change the results and only slightly improved the fit of the model (explaining 4% of the variance of all postdischarge adverse outcomes, and 7% of those adverse outcomes requiring readmission). The only difference was that high-risk patients (instead of complex surgery) now became a significant independent predictor of the occurrence of postdischarge adverse outcomes (odds ratio 1.53 (1.02 to 2.29)). The significance of all other predictors remained the same.

DISCUSSION

The present study has shown that postdischarge adverse outcomes occurred in 25% of all surgical patient admissions, and result in readmission of the patient in one-third of these patients. Postdischarge adverse outcomes were mostly infections. In-hospital adverse outcomes and complex surgical procedures increased the probability for both postdischarge adverse outcomes and readmissions. Men more frequently experienced postdischarge adverse outcomes that required readmission, as well as patients with long hospitalisations.

The 25% postdischarge adverse outcomes estimated in the present study are slightly higher than the percentage reported from previous studies.78 Part of this difference may be explained by the fact that our patients were treated in a university hospital, with usually more complex surgical cases than in non-teaching hospitals for instance, or to the more sensitive definition of an adverse outcome that was used in the present study. Another possibility is that it is due to patients being interviewed rather than that based on the (possibly incomplete) medical record information. The same may be true for our estimates with respect to readmissions, although the relative estimate (one-third of all postdischarge adverse outcomes) may still be valid. However, it seems unlikely that this overestimation will have selectively influenced the different subgroups, for instance those with and without in-hospital adverse outcomes, and thereby our results with respect to determinants of postdischarge adverse outcomes.

In a previous study, we have investigated differences between surgery groups in adverse outcome occurrence during admission.14 Head–neck surgery, breast surgery, surgery on kidney and urinary system, and surgery on the musculoskeletal system were shown to have fewer in-hospital adverse outcomes, while digestive tract surgery and surgery on skin and subcutis had more. Comparing these results with the results from the present study, it seems that after breast surgery, fewer adverse outcomes occur in-hospital but that more occur after discharge. Part of the explanation for this may be the general trend towards shorter hospital stay and the fact that a specific outpatient clinic exists for this type of patients. Of the adverse outcomes that may occur after breast surgery (infections/abscesses, oedema or bleeding), bleeding will occur shortly after the operation during hospitalisation, but any infections are more likely to occur after discharge anyway. Therefore, these patients can be discharged early, and any adverse outcomes can be treated at the standard follow-up visit 1 week after discharge.

With the possible determinants included in our model, only 3% of the variance in the postdischarge adverse outcomes was explained. This percentage was higher, looking at the adverse outcomes requiring readmission (6%), but was still low. One of the explanations for this finding is that during hospitalisation, a large number of measures have been taken to prevent or treat problems, to establish that patients can be discharged from the hospital. Because of this, the occurrence of postdischarge adverse outcomes becomes more random, rather than being explained by factors during to the admission.

Our results showed that most of the postdischarge adverse outcomes were treated by specialists, either during readmission or at the outpatient clinic. It was also shown that only 10% of these were captured by our routine reporting system, although it was higher for adverse outcomes requiring readmission (still only 25%), which has been shown to be effective in identifying in-hospital adverse outcomes.12 Of the postdischarge adverse outcomes missed in the routine reporting system, one-third resulted in readmission, and 13% were treated by additional surgery. Most of the other adverse outcomes missed in the reporting system were treated by medication. Given these results, we have started to systematically check all readmissions within 30 days after discharge, to improve the identification of postdischarge adverse outcomes that require readmission in our reporting system. To improve the identification of the remainder of postdischarge adverse outcomes, these should be systematically recorded at the outpatient clinic if patients come there for their follow-up visit, meaning that doctors should check with patients whether any adverse outcomes have occurred since discharge. To facilitate this, we have made the registration forms more readily available at the outpatient clinic. If systematic recording is not possible or incomplete, we should at least be aware that we tend to underestimate the overall occurrence of adverse outcomes in our information to patients.

These results suggest that we have to inform patients at discharge that adverse outcomes may occur after discharge, in particular if patients had complex surgery or have experienced in-hospital adverse outcomes. Since infections were the main type of adverse outcomes occurring after discharge, specific attention should be given to wound treatment by patients after discharge and to infection prevention measures in general. In addition, information may be given on the treatment of postdischarge adverse outcomes and the course of action if they do occur (eg, contact the outpatient clinic). Finally, reporting of adverse outcomes by patients during follow-up visits at the outpatient clinic should be stimulated and specifically asked for to improve adverse outcome reporting in our routine reporting system. If prepared for the possibility, patients are likely to respond better if adverse outcomes do occur after discharge, rather than that patients experience this as a failure of hospital care. This will improve the quality of the information and is therefore likely to improve the satisfaction of patients.