Risk of venous thromboembolism in outpatient parenteral antimicrobial therapy (OPAT): a systematic review and meta-analysis
Our latest research was just published in the prestigious International Journal of Antimicrobial Agents (International Society of Antimicrobial Chemotherapy), 6 July 2023.
Risk of venous thromboembolism in outpatient parenteral antimicrobial therapy (OPAT): a systematic review and meta-analysis
by
Access the full paper via:
- ScienceDirect
- PubMed
- Free access Share Link (valid until September 27, 2023)
Highlights
- A systematic review to investigate the incidence of venous thromboembolism (VTE) in outpatient parenteral antimicrobial therapy (OPAT) settings.
- We found a low incidence of catheter-related and non-catheter-related VTE in OPAT patients.
- Our findings do not support universal thromboprophylaxis nor routine use of inpatient VTE risk assessment model in the OPAT setting.
- High index of suspicion should be maintained, especially for patients with known risk factors for VTE.
- The study adds to the growing evidence that OPAT is a safe alternative to inpatient care.
ABSTRACT
The risk of venous thromboembolism (VTE) in outpatient parenteral antimicrobial therapy (OPAT) is not fully understood and the optimal strategy for thromboprophylaxis remains unclear. This systematic review investigated the incidence of VTE in OPAT settings (PROSPERO CRD42022381523). MEDLINE, CINAHL, EMCARE, EMBASE, Cochrane Library and grey literature were searched from earliest records to 18 January 2023. Eligible were primary studies reporting non-catheter-related or catheter-related thromboembolic (CRT) events in adults who received parenteral antibiotics in home or outpatient settings. In all, 43 studies involving 23,432 patient-episodes were reviewed. Four studies reported non-catheter related VTE while 39 included CRT. Based on generalised linear mixed-effects models, pooled risk estimates of non-catheter-related VTE and CRT were 0.2% (95% confidence interval [CI], 0.0 – 0.7%) and 1.1% (95% CI, 0.8 – 1.5%; prediction interval [PI], 0.2 – 5.4%), respectively. Heterogeneity was largely attributed to risk of bias by meta-regression (R^2 = 21%). Excluding high-risk studies, CRT risk was 0.8% (95% CI, 0.5 - 1.2%; PI, 0.1 - 4.5%). From 25 studies, pooled CRT rate per 1,000 catheter-days was 0.37 (95% CI, 0.25 – 0.55; PI, 0.08 – 1.64). Our findings do not support universal thromboprophylaxis nor routine use of inpatient VTE risk assessment model in the OPAT setting. However, high index of suspicion should be maintained, especially for patients with known risk factors for VTE. An optimised protocol of OPAT-specific VTE risk assessment should be sought.
KEYWORDS Complications; deep vein thrombosis; outpatient parenteral antimicrobial therapy; risk assessment; systematic review; thromboembolism; vascular access device
Statistical methods
The primary study outcome was incidence of catheter-related thromboembolism (CRT) and non-catheter-related venous thromboembolism (VTE). We estimated population-averaged incidence proportions, pooled across studies, using a random-intercept logistic regression model with maximum-likelihood estimation. The model assumed a Binomial distribution for the observed number of VTE cases in each study and a Normal distribution for the random effects, with a logit transformation. This approach correctly incorporates studies reporting zero cases and maintains confidence limits of pooled proportions within the zero to one range. The resulting confidence interval (CI) estimates the expected (average) VTE risk of all possible studies.
Higgin-Thompson's I² statistic was used as a summary index of the variability in VTE incidence across studies that cannot be attributed to sampling error. Because I² is usually high and may not be discriminative for prevalence or incidence data, between-study variance (τ2) and its respective 95% prediction interval (PI) were additionally reported. The PI describes the range of VTE risks that can be expected in new studies. A forest plot was constructed to illustrate the distribution of VTE incidence across the studies along with 95% CIs calculated by Wilson's score method. To examine potential sources of variation in VTE incidence among the studies, multivariable meta-regression analysis was conducted with the Binomial-Normal mixed-effects model. Adjusted odds ratios (aORs) with respective 95% CΙ summarised the strength and direction of associations between study-level covariates and VTE incidence. For each covariate, a covariate-specific R2 was calculated as the portion of between-study variance that was reduced after the inclusion of that covariate in the model (in the presence of all other variables). Moreover, for each covariate level, pooled estimates of VTE incidence were calculated based on univariate subgroup analysis. Candidate covariates for the regression analysis were decided a priori in our study protocol. The following variables were examined: publication year; [World Health Organization (WHO) region]; study design; and risk of bias (classified as either low-to-moderate or high).
Leave-one-out sensitivity analysis was carried out to assess the robustness of pooled estimates of VTE incidence against excessively influential studies. To address time-dependent confounding from studies that recorded VTE incidence over different risk periods, we sought studies reporting OPAT duration statistics, calculated incidence density rates (expressed as events per 1,000 catheter-days), and estimated population-averaged incidence rates using a Poisson-Normal mixed-effects model. All analyses were carried out in STATA (Version 17; Statcorp, College Station, TX, USA).
Figure: Forest plot of the results of the random-effects meta-analysis (Poisson-Normal mixed-effects model) of the incidence density rate of catheter-related venous thromboembolism in outpatient parenteral antimicrobial therapy. n/N denotes the number of CRT cases over the total number of OPAT/IV catheter-days in each study. The diamond's centre is the population-averaged CRT incidence rate. The diamond's length and the respective grey vertical area indicate the 95% confidence interval of the pooled average estimate. The extended blue line continuing through the confidence interval and the respective bluish-grey vertical area indicate the 95% prediction interval of CRT incidence expected in new studies. Abbreviations: CI, confidence interval; CRT, catheter-related venous thromboembolism