A Systematic Review of Risk of HIV Transmission Through Biting or Spitting

Implications for Policy

FV Cresswell; J Ellis; J Hartley; CA Sabin; C Orkin; DR Churchill


HIV Medicine. 2018;19(8):532-540. 

In This Article


PICO (P, Patient, Problem or Population; I, Intervention; C, Comparison, Control or Comparator; O, Outcome)

The authors used the PICO framework, with the PICO "question" being formulated and answered as follows: (1) population: adults, adolescents and children; (2) intervention: bites, spitting; (3) comparator: none; (4) outcome: HIV transmission or documented absence of HIV transmission.

Search Strategy

The goal was to identify evidence relating to the risk of transmission, or lack of transmission of HIV following a biting or spitting incident. A systematic electronic search was conducted using Medline, Embase and Northern Lights databases from inception to 5 January 2018. Key natural language and controlled vocabulary search terms were used related to "HIV", "human immunodeficiency virus", "AIDS", "acquired immune deficiency syndrome" AND "bites", "bitten" OR "spit", "spat", "spitting". A second search was run using the terms relating to "HIV transmission" AND "saliva". For full search terms, see Supporting Information. We also hand searched the British HIV Association conference abstracts from 2007 onwards and Conference for Retroviruses and Opportunistic Infections abstracts from 2014 onwards, as well as the reference lists from the papers we reviewed.

Eligibility Criteria

The following inclusion criteria were applied in article selection for full-text review: (1) exposure of interest (biting, spitting or saliva) discussed and (2) outcome of interest described (by documented HIV antibody testing, with or without additional antigen testing, HIV viral load testing or phylogenetic analysis) or absence of HIV seroconversion (by documented negative HIV antibody test).

Study Selection

Two reviewers (JH and TR) independently conducted selection for full-text review by applying eligibility criteria to titles and abstracts. Two reviewers (JE and FVC) then independently assessed full-text articles for how HIV transmission had been determined and excluded articles that did not describe the exposure and outcome of interest or did not provide original case data such as narrative reviews. A list of studies for inclusion was finalized.

Assessment of Quality and Data Extraction

Reviewers designed a data extraction tool and independently applied it to each article. Data were extracted on study design, the perpetrator (HIV status, HIV viraemia, presence of blood in the mouth of the perpetrator, whether medically unwell and use of ART), the nature of the incident (whether biting or spitting, and the severity of the wound inflicted), the timing of HIV diagnosis, the nature of HIV testing and other HIV risk factors. Data were compared for consistency. No formal statistical analyses were undertaken in view of the nature of the studies identified.

No randomized controlled trials or cohort or case-control studies were identified, so a formal tool to assess risk of bias for the articles identified was not used. Instead, we discussed the plausibility of HIV transmission being attributable to the incident described based on documentation of baseline HIV status, the nature of the injury, the temporal relationship between the incident and a positive HIV test and phylogenetic analysis, where available. The plausibility of the incident being responsible for the subsequent HIV diagnosis was then classified as low, medium, high or confirmed based on pre-specified criteria (Table 1). Any disagreements were resolved by consensus or a third reviewer (JH).