Article Text

Surveillance of unplanned return to the operating theatre in neurosurgery combined with a mortality–morbidity conference: results of a pilot survey
  1. Hélène Marini1,
  2. Véronique Merle1,
  3. Stéphane Derrey2,
  4. Christine Lebaron1,
  5. Valérie Josset1,
  6. Olivier Langlois2,
  7. Marie Gilles Baray2,
  8. Noëlle Frébourg3,
  9. François Proust2,
  10. Pierre Czernichow1
  1. 1Department of Epidemiology and Public Health, Rouen University Hospital, Rouen, France
  2. 2Department of Neurosurgery, Rouen University Hospital, Rouen, France
  3. 3Department of Microbiology, Rouen University Hospital, Rouen, France
  1. Correspondence to Dr Hélène Marini, Department of Epidemiology and Public Health, Rouen University Hospital, 1 rue de Germont, 76031 Rouen-Cedex, France; helene.marini{at}


Background Unplanned return to the operating theatre (UROT) is a useful trigger tool that could be used to identify surgical adverse events (SAEs). The present study describes the feasibility of SAE surveillance in neurosurgical patients, based on UROT identification, completed with SAE analysis at a morbidity–mortality conference (MMC) meeting.

Method For consecutive patients who underwent a neurosurgical procedure between 1 November 2008 and 30 April 2009, return to the operating theatre (ROT) was identified based on the hospital information system associated to prospective payment (HISPP). ROT was classified as planned or unplanned and UROT was further classified as related to the natural history of the disease or related to an adverse event (AE-UROT). MMC meetings were organised to discuss results of UROT surveillance and to analyse AE-UROT.

Results 1006 neurosurgical procedures were included in the surveillance. HISSP identified 152 ROTs, with 73 UROTs related to an SAE (7.3% (5.7% to 9.0%)): infectious SAE (n=24, 2.4% (1.5% to 3.5%)), haemorrhagic SAE (n=23, 2.3% (1.5% to 3.4%)), other cause SAE (n=26, 2.8% (1.9% to 4.0%)), and infectious and other cause SAE (n=2, 0.2% (0.0% to 0.7%)). Identification of AE-UROT through HISSP required a 4 h/month time frame. Eight UROTs related to SAE cases were discussed during MMC meetings, leading to the identification of non-conforming care processes and practical improvement actions.

Conclusion UROT related to SAE surveillance in neurosurgical patients was considered feasible. The association of surveillance and MMCs allowed staff to concentrate on the analysis of most frequent or most severe AEs and was a practical and useful tool to stimulate improvement. The impact on healthcare quality of SAE surveillance associated with MMC warrants further research.

Statistics from


Surgery is responsible for the majority of adverse events (AEs) identified during hospital stay.1 2 Surgical AEs represent 48%–79% of all AEs.3–8 In France, 6.5 million surgical procedures are carried out per year; therefore, according to these rates, 60 000–95 000 serious AEs could occur in the peri-operative period.1

In order to improve the quality and security in healthcare, it could be important to monitor AE rates and to implement root cause analysis of the most frequent and/or severe surgical AE, although the benefit of this practice on quality of care has never, to our knowledge, been previously assessed.

One difficulty in monitoring AE rates would be the detection of surgical AEs in routine practice. It has been suggested that a proportion of unplanned return to the operating theatre (UROT) is caused by a surgical AE and therefore it may represent a quality indicator in surgical care.9–12 However, routine identification of UROTs can also be time-consuming.

We hypothesised that UROTs could be routinely tracked through the use of hospital information system associated to prospective payment (HISPP) to identify return to the operating theatre (ROT), followed by a classification of ROT as UROT or planned ROT through an analysis of computerised records.

We also hypothesised that the reporting of AE-associated UROT (AE-UROT) rates to clinical staff combined with the analysis of AE-UROT at a morbidity–mortality conference (MMC) meeting could be associated with an improvement in healthcare quality and a decrease in AE-UROT rates.

The aim of this study was to evaluate the feasibility of implementing a routine surveillance system reporting AE-UROT rates combined with an MMC on cases of AE-UROT identified by this surveillance in neurosurgical patients. The secondary aim was to assess the performance of this method for identifying AE-UROTs for various intervals between the index operation and the UROT.



This study was carried out in 2008–2009 at a 60-bed Neurosurgery Department of Rouen University Hospital, a tertiary-care centre of 1700 acute care beds. At the time of our study, no data were routinely available in this setting regarding either ROT frequency or surgeon-specific ROT frequency.


Patients who underwent a neurosurgical procedure between 1 November 2008 and 30 April 2009 were included in the study. Patients aged <18 years were excluded.


A ROT was defined as a neurosurgical procedure performed in the operating theatre during the 30 days following an index neurosurgical procedure.

A ROT was classified as a planned ROT when the procedure was originally scheduled as a two-time procedure as mentioned in the surgery records or in the hospitalisation records, or when the procedure was defined by neurosurgeons as being routinely performed twice (eg, external ventricular drain placement before colloid cyst exclusion). A ROT not meeting these criteria was classified as a UROT.

UROTs were classified into a UROT related to the disease natural history or an AE-UROT. An AE was defined as an unintended injury or complication caused by healthcare management rather than by the patient's underlying disease process.2 6 7 13 14

AEs were classified as infectious, haemorrhagic or from another cause, or due to the association of several causes. Table 1 presents cases illustrating each type of ROT.

Table 1

Planned ROT and UROT related to SAE and disease natural history: examples

Surveillance method

Patients operated on for a neurosurgical procedure and a ROT in the 30 days following an index neurosurgical procedure were identified through the HISPP for the period 1 November 2008 to 30 April 2009. Patient age, gender and type of surgical procedure (ie, brain, spinal or other procedure) were collected. No severity score was available in HISPP for neurosurgery patients.

For each ROT, the patient's charts were retrieved in order to eliminate a false ROT due to possible coding errors.

In the second stage, a ROT was classified as unplanned or planned after the analysis of the patient's computerised record.

This assessment was performed in a blind manner by two epidemiologists (Drs HM and VM). UROTs were then classified into a UROT related to the disease process or an AE-UROT (from infectious, haemorrhagic, other cause or several causes). The final classification was validated by a neurosurgeon (Dr SD).

For each 3-month period during the study period, a report describing the frequency of ROTs and UROTs, and AEs responsible for UROT, was drawn up and communicated to neurosurgeons, anaesthetists and nurses from the Neurosurgical Department.

Mortality and morbidity conferences

An MMC meeting was organised every 3 months in order to discuss AE surveillance results, and cases of AE identified by UROT surveillance. A neurosurgeon (Dr SD) was identified as the physician responsible for the organisation of the MMC.

Each meeting was scheduled as follows: first, UROT surveillance results were presented and discussed, then three to five AE-UROT cases selected based on their frequency and severity were discussed in detail. The discussion aimed at identifying processes of care in disagreement with good practices guidelines and at proposing improvement actions. For each improvement action, a professional in charge of its implementation was designated. A report stating problems and identified improvement actions was addressed to participants after each MMC meeting. Names of patients and physicians in charge of cases discussed were not mentioned in the report.


Proportions of UROTs and of AE-UROTs were calculated with their 95% CIs.

Positive predictive values for identifying a ROT and a UROT through HISPP (ie, probability that a ROT detected through HISPP is a ROT and a UROT) were calculated.

The agreement between the two epidemiologists (Drs VM and HM) in classifying ROT into UROT/planned ROT was assessed by calculating a κ coefficient.15

The sensitivity of the AE-UROT identification method was calculated for intervals of 10 and 15 days between the index procedure and the ROT, compared with the 30-day-interval considered in our study as the gold standard.

The time necessary for data collection was assessed.

Statistical analysis was performed with Microsoft Excel 2000 for Windows (Microsoft Inc., USA) and Epi-Info 6.04 (Center for Disease Control).


Between 1 November 2008 and 30 April 2009, 879 adult patients were operated on and included in the surveillance programme (mean age: 53.7 years±0.8; male to female sex ratio: 1.1). Four hundred and forty-eight patients (51%) underwent brain procedures, 422 (48%) spinal procedures and 9 (1%) other procedures.

AE-UROT identification and rates

Use of HISPP allowed for identifying 152 ROT cases. According to the analysis of computerised records, 111 ROT cases (92 after brain procedure, 19 after spinal procedure), occurring in 74 patients, were confirmed to represent ROTs (ROT rate 8.4%, 95% CI 6.7 to 10.5). Among these 111 ROTs, 93 (83.8%) were judged as UROTs. The positive predictive value for identifying a ROT through HISPP was therefore 73% (95% CI 66% to 80%), and the positive predictive value of identifying UROTs was 61% (95% CI 53% to 69%) (figure 1).

Figure 1

An overview of ROT classification. AE-UROT, adverse event-associated UROT; HISPP, hospital information system associated to prospective payment; ROT, return to operating theatre; UROT, unplanned return to the operating theatre.

Rates of ROT according to the site of surgical procedure (brain or spine) are displayed in table 2. ROT classification is shown in figure 1.

Table 2

Comparison of classification of ROT according to the procedure site

κ Coefficient for agreement between the two observers regarding the classification of ROT between and planned UROT was 0.88 (SD 0.09). Four ROTs (%, 95% CI) were not classified because their mechanism could not be determined by computerised record analysis.

Sensitivity for identifying AE-UROT in choosing intervals of 10 and 15 days between the index procedure and ROT was 69.9% (95% CI 59.3% to 80.4%) and 86.3% respectively (95% CI 78.4% to 94.2%) compared with the 30-day-interval gold standard.

Approximately 4 h a month were necessary to determine AE-UROT rates (ie, approximately 20 min per AE-UROT): 10 min to extract data from HISPP and sort procedures by brain or spinal surgery, 3 h to classify ROT and identify AE-UROT, and 40 min to validate classification with the neurosurgeon.

Morbidity–mortality conferences

Two MMC meetings took place during the study period. Neurosurgeons, epidemiologists and neurosurgery nurses participated in both meetings. Anaesthesiologists and bacteriologists participated only in the second MMC meeting.

Five AE-UROTs were addressed during each MMC meeting. Several non-conforming care processes and improvement actions were identified during MMC. They are listed in tables 3 and 4.

Table 3

Non-conforming care processes and improvement actions proposed after discussion of AE-UROT cases during MMC meetings

Table 4

Non-conforming care processes and improvement actions proposed after discussion of AE-UROT surveillance results during MMC meetings


The aim of this study was to implement a surgical AE surveillance system based on the analysis of a UROT in a neurosurgery department. Our results suggest that this surveillance is feasible in routine practice, provides AE-UROT rates, and leads to the identification of non-conforming care processes and the implementation of improvement actions.

Our study has some possible limitations. First, HISPP is a database aimed at the hospital's financing and was not conceived for epidemiologic or quality improvement purposes.16 Therefore, identification of a ROT through HISPP has several limitations: delay (in most hospitals, data are available approximately 2 months after the end of each 1-month period), possible lack of sensitivity and specificity due to coding omission or error.17 False negatives could theoretically be overlooked using our method. However, as hospital funding relies on coding, controls are routinely performed by the hospital management to ensure that all surgical procedures are accurately coded.

In our surveillance, we were not able to routinely identify surgical procedure types or patient severity. Patients and surgical procedures were only characterised by age and surgical procedure location. Therefore, AE-UROT rate fluctuations can be related to confusion factors (patient severity18, type of surgical procedures10 18) that were not taken into account in our study.

UROT identification was routinely feasible through HISPP in our patients. Predictive positive value of identifying a ROT through HISPP in the present study was only 73% but false positives were easily eliminated in a second stage by manual assessment of computerised records, and thus did not impair the validity of the results.

This result regarding feasibility is concordant with a previous survey performed in vascular surgery patients.18 However, the feasibility across various surgical specialties cannot be derived from these two successful experiences and remains to be demonstrated in other domains.

Another important result is the fact that epidemiologists were able to identify UROTs by using HISPP without the help of neurosurgeons, with a good inter-observer agreement.15 Therefore, we could outsource a part of surveillance workload to epidemiologists, and limit the workload for the neurosurgery team for the validation of UROT types classification to 40 min a month for one neurosurgeon. This outsourcing could be a way to improve surveillance acceptability by the neurosurgery team, as it avoids data collection by neurosurgeons or neurosurgery nurses.

For the epidemiology team, the workload was estimated to be approximately 4 h a month. This workload could not reasonably be reduced by using a shorter interval between ROT and the index intervention as the use of a shorter interval led to an important loss in sensitivity. Therefore, we chose to retain a 30-day interval as suggested by other authors.10–12

Our study suggests the interest of AE-UROT surveillance for quality improvement. First, it allowed communicating and following UROT and AE-UROT rates. Our UROT rate of 9.2% was similar to those observed by Ploeg et al12 in vascular surgery (11.7%). Using the same methodology, Kroon et al11 found a UROT rate of 1.7% in a general surgery department, suggesting that UROT rates would fluctuate according to surgical specialty. To our knowledge, our study is the first to report neurosurgery UROT rates.

Our surveillance was coupled with MMC meetings. During MMC meetings, we reported and discussed AE-UROT surveillance results: our hypothesis was that this would enable analysis of frequent problems, as reported by Seiler.19 In their study, mortality and morbidity indicators in neurosurgery were presented and discussed during MMC. These authors suggested that this method had the advantage of providing accurate statistics for patient information and for continuous quality control of patient management and surgical techniques. Prospective surgical error or dysfunction record has been suggested to improve quality of care in neurosurgery.20 21 However, in our ward, the number of interventions, and the lack of an automated information system regarding ROT, precluded this systematic prospective recording. Therefore, our UROT surveillance system allowed us to check out routinely AE-UROT.

Association of AE-UROT surveillance and MMC allowed us to focus on frequent and/or severe AE-UROT cases for discussion, which we believe to be a better indication of MMC efficacy. In agreement with this hypothesis, we observed that the few instances of discussion of rare events (such as a surgical site infection after metastatic decompression) did not lead to improvement actions, mainly because follow-up investigations (trend analysis of postoperative infections, comparison with benchmarks from the literature) would have been needed to determine if these cases signalled quality of care problems and if anything needs to be done.

MMC meetings were enthusiastically received by members of the department, that is, neurosurgeons, anaesthesiologists and head nurses, who all actively participated in the meetings.

To date, these MMC meetings coupled with AE-UROT surveillance continue to be proposed in the neurosurgery department. AE-UROT rates have fluctuated between 4.5% and 7.5% during this period with no tendency towards a decrease but discussions during MMC meetings are always lively and they enable to find solutions to clinical practice problems. Neurosurgeons can take advantage of the MMC to present patients other than those identified through surgical AE surveillance. We think that two favouring factors can explain the success of these MMCs. First, the staff members of this Neurosurgery Department were used, previous to the implementation of MMC, to freely discuss clinical problems and complications. Second, MMC organisation, and especially improvement action follow-up, is assured by the epidemiology team, and therefore it is time-saving for the neurosurgery department. However, we feel that nurse participation in MMC meetings could be increased in order to provide more information about nurse patient care during discussions.

MMC meetings took place every 3 months: they allowed discussing some cases of AE-UROT. To our knowledge, there are no international guidelines regarding MMC frequency and literature reports various frequencies.22 23 French guidelines demand that each setting determines its appropriate MMC frequency.24 Here, a quarterly MMC was chosen for its feasibility and acceptability.

Our surveillance programme was limited to those surgical AEs that led to UROT. We considered that they represented a more severe subgroup of AEs. We hypothesised that improvement actions decided through analysis of these surgical AE-UROTs would be beneficial for the prevention of all surgical AEs.

In conclusion, we were able to successfully implement an AE-UROT surveillance programme in the Neurosurgery Department. Our results show that the surveillance was feasible and its coupling with an MMC was considered to have been a practical and useful tool to stimulate improvement. The impact on healthcare quality warrants further research.


The authors are grateful to neurosurgery surgeons from Department of Neurosurgery, Rouen University Hospital, anaesthesiologists from Department of Anaesthesiology, Rouen University Hospital, and nurses from Department of Neurosurgery, Rouen University Hospital, S. Hurel, neurosurgery nurse, and M. Guillard. The authors are also grateful to Richard Medeiros, Rouen University Hospital Medical Editor, for editing the manuscript.

View Abstract


  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Data are available on request from the corresponding author.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.