Scientific Papers

Hospital infections and health-related quality of life after cardiac surgery: a multicenter survey | Journal of Cardiothoracic Surgery

Description of Image

Setting and study population

In this retrospective study, all adult patients who underwent elective or urgent (i.e. those who require an intervention for medical reasons within the current intake) cardiothoracic surgery between 01-01-2015 and 31-12-2019 and completed pre- and one year post-operative Short Form Health Survey 36-version 2 or Short Form Health Survey 12 quality of life questionnaires (SF 36-2/SF 12) were included. Data were extracted from the Netherlands Heart Registration (NHR), a nationwide registry of all invasive cardiac interventions, comprising data from all Dutch hospitals [13]. The NHR facilitates value-based outcome monitoring, including quality of life outcome. Participating hospitals are responsible for data collection and registration and check their own data. The NHR analyses patient data, provides online dashboards and reports relevant outcome indicators in yearly, publicly accessible reports. Each year, within the NHR, data validation and verification is performed by standardized quality controls and monitoring visits (audits). In addition, the distribution of patient-relevant outcomes between hospitals is observed to verify that no striking differences exist. In the case of a significant variation in outcomes, processes of healthcare delivery are discussed and good practices are shared [14]. Datasets consist of a mandatory standard part (with a 90% completeness requirement) and a voluntary part, including HRQoL [13,14,15]. Within this study, mainly data of the mandatory standard set was used. This study complies with the Declaration of Helsinki and Good Clinical Practice guidelines. The study protocol is available at the International Clinical Trials Registry Platform, under main ID NL9818 [16]. The study was approved by the institutional review board MEC-U (W19.270) and conducted in agreement with the principles of the Declaration of Helsinki. A waiver for informed consent for analysis with the data of the NHR data registry was obtained. Preoperative, intraoperative- and postoperative data with a one-year follow-up period were collected and stored in a pseudonymized database.

Measurement of the HRQoL and definitions

The SF-36-2 is a standardized, validated and widely used HRQoL assessment tool [17]. The SF-12 is a validated shortened version of the SF-36 questionnaire [18, 19]. Both questionnaires consists of (36 resp. 12) multiple choice questions divided over four physical health domains and four mental health domains. The individual scores of all physical health domains (physical function, role limitations due to physical problems, body pain and general health perception) are combined and expressed as physical health score (PHS). In this study we did not include the mental health score. Patients completed HRQoL before and one year after surgery. If a minimum of 50% of the questions was answered in each physical health domain of SF 36-2 and 100% of the questions of SF 12, patients were included in this study.

Based on this PHS, physical recovery was calculated by PHS 1 year after surgery minus baseline PHS. Patients with a score > 0 were included in the physical recovered group (R). Patients with a score ≤ 0 were allocated to the non-recovered group (NR).

Definition of hospital infections

Patients were categorized into two groups. One group consisted of patients who developed any hospital infection (Infection (I) group). The other group consisted of patients with no hospital infection (Non-Infection (NI) group). Hospital infections were defined as every registered peri- and postoperative infection during hospital stay, according to predefined criteria irrespective of site or severity. These registered infections were deep sternum wound infection (positive cultures and/or surgical drainage and/or antibiotic therapy), pneumonia (positive cultures), urinary tract infection (positive cultures) or arm-/leg wound infection (positive cultures and/or surgical drainage and/or antibiotic therapy [20]. Each of these variables is part of the NHR’s mandatory variable set, which has a 90% completeness requirement.

Statistical analyses

Multiple imputation

For the primary analysis, i.e. being the association between hospital infections and non-improvement of PHS adjusted for potential confounders, 2.2% of all values was missing, with the proportion of missing data per variable ranging from 0 to 27.1% (Additional file 1: Table 1). Infection status was missing for n = 67 (0.8%) of patients, and there was no missing data for the outcome variable due to the study design. Since values were assumed to be missing at random, we used multiple imputation to impute 45 datasets using chained equations with imputations drawn using predictive mean matching. For more details, see the Additional file 1. Pooled analysis based on imputed data was used for all analyses in the main paper, unless state otherwise.

Descriptive statistics

Normally distributed continuous variables are presented as mean ± standard deviation (SD) and variables with a non-normal distribution as median [interquartile range, IQR]. Categorical variables are described with numbers and percentages. The student’s t-test, Mann–Whitney U-test, and the Chi-square test were used to assess differences between the I-group and the NI-group, as appropriate. All analyses were performed using SPSS 24 for Windows® (SPSS INC, Chicago, IL, USA). A p-value < 0.05 was considered to be statistically significant. The standardized mean difference (SMD) was calculated as a measure of (im)balance in potential confounders between the NI en I group. Sufficient balance was considered achieved with an SMD < 0.1.

Primary study question

In order to obtain an unbiased estimate of the association between hospital infections (NI/I) and physical recovery (NR/R), in the primary analysis, we used binary logistic regression in which the inverse of the propensity score for infection risk was included as a weight variable. Robust sandwich variance estimators were used to deal with the artificially increased sample size by applying weights. In the propensity score model we considered variables marginally associated (p < 0.1) with non-recovery as independent variables [21]. Variables that can affect physical recovery, variables both associated with hospital infections and physical non-recovery (confounding factors) and predictive markers described in literature were also included in the model [11].

The final propensity score model with infection as outcome variable included the following independent variables: baseline PHS, unstable angina pectoris, age, gender, extra-cardiac arteriopathy, chronic lung disease, critical preoperative state, DM, NYHA Class III or IV, left ventricular ejection fraction (LVEF), recent myocardial infarction, urgency and surgery on thoracic aorta. More details on the propensity score model can be found in the Additional file 1.

Secondary study question

To assess if DM modifies the association between hospital infections and physical recovery, the primary analysis was stratified for DM status. Differential effects of hospital infections on physical recovery for patients with and without DM was formally tested by adding an interaction term (DM*hospital infections) to the main model.

Description of Image

Source link