Multistate modelling to estimate excess length of stay and risk of death associated with organ/space infection after elective colorectal surgery

https://doi.org/10.1016/j.jhin.2018.08.010Get rights and content

Summary

Background

Accounting for time-dependency and competing events are strongly recommended to estimate excess length of stay (LOS) and risk of death associated with healthcare-associated infections.

Aim

To assess the effect of organ/space (OS) surgical site infection (SSI) on excess LOS and in-hospital mortality in patients undergoing elective colorectal surgery (ECS).

Methods

A multicentre prospective adult cohort undergoing ECS, January 2012 to December 2014, at 10 Spanish hospitals was used. SSI was considered the time-varying exposure and defined as incisional (superficial and deep) or OS. Discharge alive and death were the study endpoints. The mean excess LOS was estimated using a multistate model which provided a weighted average based on the states patients passed through. Multivariate Cox regression models were used to assess the effect of OS-SSI on risk of discharge alive or in-hospital mortality.

Findings

Of 2778 patients, 343 (12.3%) developed SSI: 194 (7%) OS-SSI and 149 (5.3%) incisional SSI. Compared to incisional SSI or no infection, OS-SSI prolonged LOS by 4.2 days (95% confidence interval (CI): 4.1–4.3) and 9 days (8.9–9.1), respectively, reduced the risk of discharge alive (adjusted hazard ratio (aHR): 0.36 (95% CI: 0.28–0.47) and aHR: 0.17 (0.14–0.21), respectively), and increased the risk of in-hospital mortality (aHR: 8.02 (1.03–62.9) and aHR: 10.7 (3.7–30.9), respectively).

Conclusion

OS-SSI substantially extended LOS and increased risk of death in patients undergoing ECS. These results reinforce OS-SSI as the SSI with the highest health burden in ECS.

Introduction

Surgical site infections (SSIs) are one of the most severe and dreaded healthcare-associated infections (HCAIs) in elective colorectal surgery (ECS). These infections increase morbidity and mortality, and prolong length of stay (LOS), thereby increasing patient and health costs [1], [2]. Among SSIs, organ/space (OS)-SSI has been associated with the worst outcomes [3], [4], [5].

Since colorectal surgery is a cornerstone of treatment for colorectal cancer – the third most common cancer diagnosed in developed countries – avoiding these HCAIs is an urgent matter. Multiple strategies have been shown to be successful in preventing SSIs; however, recent studies still show high rates of OS-SSI associated with colorectal surgery [6], [7], [8], [9], [10], [11], [12], [13].

Measuring the health cost of OS-SSI accurately can facilitate joint efforts by all stakeholders to implement targeted prevention strategies. Currently, from the hospital perspective, the cost of HCAIs is mostly due to extending patient LOS, which determines missed new hospital admissions [14], [15]. When estimating LOS due to HCAIs, applying statistical models that consider the time-dependent nature of the infection has been recommended. This approach permits a better control of time-dependent bias and avoids overestimation of excess LOS [16], [17].

To date, studies reporting the effect of SSI on LOS in colorectal surgery have not considered time-dependent bias [1], [4], [12]. The purpose of the present study is therefore to assess the health costs of OS-SSI measured in terms of excess LOS and risk of death during the hospital stay in a prospective cohort of patients undergoing ECS, taking into account timing of infection and competing events.

Section snippets

Setting and study design

This was a multicentre prospective cohort study of adult (aged ≥18 years) patients who underwent ECS from January 2012 to December 2014, at 10 hospitals in Catalonia, Spain. The hospital characteristics are shown in Table I. All these hospitals routinely report data to the regional surveillance programme for HCAIs: VINCat [5], [18]. All patients hospitalized for ECS at the different hospitals were followed up until discharge or death. Patients with pre-existing infection at the time of surgery

Results

A total of 2778 patients were included in the cohort; cancer was the main cause of surgery 2623 (94%). During the hospital stay, 343 patients (12.3%) developed SSI. Of those, 194 (7%) had OS-SSI and 149 (5.3%) incisional SSI. The incidence density of overall SSI was 15.7 per 1000 patient-days at risk; 8.9 and 6.8 per 1000 patient-days at risk for OS-SSI and incisional SSI, respectively. Infection occurred in a median time of six days after surgery for both OS-SSI and incisional SSI. The median

Discussion

This study shows that, among SSIs, OS-SSI had the greatest burden on LOS and mortality in patients undergoing ECS in a large cohort of patients. The results are consistent with those reported in the literature; however, previous studies frequently used matching designs to estimate excess LOS, a type of design that overestimates LOS, since they do not consider time-dependency of the infection [1], [4], [23], [24], [25], [26].

Excess LOS attributed to SSI varies from 4.1 to 15 days, although most

Acknowledgements

We would like to thank the other researchers and members of the VINCat colon surgery group: Hospital Bellvitge: D. Camprubí; Hospital de Viladecans: L. Martín, C. Sanz; Consorci Sanitari de l’Anoia: M Brugués; Corporació Saniària Parc Taulí: X. Serra-Aracil, L. Mora; Parc Sanitari Sant Joan de Déu de Sant Boi: V. Diaz-Brito, E. Moreno; Fundació Althaia: F. Obradors; Consorci Sanitari de Terrassa: E. Espejo, F. Aguilar, L. Pagespetit; Hospital Universitari Mútua de Terrassa: C. Nicolás, A.

References (37)

Cited by (13)

  • Surveillance of surgical site infection in a teaching hospital in Ghana: a prospective cohort study

    2020, Journal of Hospital Infection
    Citation Excerpt :

    Further studies are needed to determine whether personnel in other healthcare facilities could be trained to improve the return of wound cards for patients living far from hospital. The extra number of hospital days spent for patients who developed an SSI and the added risk of undergoing a radiological or another surgical intervention is similar to that found in other studies [1,17]. Adjusted for age and primary procedure, there was no difference in mortality between patients with or without an SSI.

View all citing articles on Scopus
1

Members of the VINCat colon surgery group are listed in the Acknowledgements section.

View full text