Dengue among immunocompromised patients: a systematic review and meta-analysis

Abstract

Background: Although there is a rising trend in both dengue cases and immunocompromised conditions, there is limited research on how common severe dengue is in immunocompromised individuals. This data is key for those advising the ever-increasing numbers of immunocompromised travellers.

Methods: We conducted a systematic review and meta-analysis of studies reporting dengue frequency or outcomes in immunocompromised populations. Non-human and review articles were excluded. Risk of bias was assessed using the ROBINS-E tool.

Results: Eighty-five studies were included; 63 had a very high risk of bias. Frequency of dengue among different immunocompromised patient cohorts varied from 0.3% to 6.3%. Of 1182 dengue cases, 664 had autoimmune diseases, 388 were post-solid organ transplant (SOT), 20 post-stem cell transplant (HSCT), 28 had haematological malignancies, 24 non-haematological malignancies and 58 were HIV-positive. Severe dengue and mortality were estimated at 0.27 [0.22-0.33] and 0.14 [0.10-0.18], decreasing to 0.16 [0.09-0.27] and 0.04 [0.03-0.05] when very high risk or small-sample studies were excluded. Twenty-three (5.6%) of post-transplant dengue patients were considered as donor-related. Mortality reached 66.7% in HSCT and 10% in SOT. Dengue RNA was detectable up to four months in blood and up to two years in urine; viable virus was isolated from urine at nine months.

Conclusions: Dengue in immunocompromised, especially HSCT, is associated with high severity and mortality. It also has the potential for prolonged viral persistence.

Keywords: Dengue; immunocompromised; severity; systematic review.

Figure 1

Meta-analysis comparing risk of severe dengue between immunocompromised and non-immunocompromised populations. Each included study is listed in the left column, along with the number of dengue cases and total sample size in the immunocompromised and the non-immunocompromised. Data are presented as risk ratios (squares) and 95% confidence intervals (error bars). The weight of each study, calculated using the random-effects model, is shown in the rightmost column. The pooled estimate is represented by a diamond, where the width reflects the 95% confidence interval.