Risk of HIV-1 Transmission for Parenteral Exposure and Blood Transfusion: A Systematic Review and Meta-Analysis

Rebecca F. Baggaley; Marie-Claude Boily; Richard G. White; Michel Alary


AIDS. 2006;20(6):805-812. 

In This Article


Summary of the Search

Sixty estimates for parenteral (medical, IDU and accidental injections) and 14 for blood product exposure were found. After evaluation, 30 articles were included for parenteral exposure,[15,16,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45] providing 26 estimates, and 10 articles were included for blood product exposure,[46,47,48,49,50,51,52,53,54,55] providing 11 estimates. The 30 included studies providing estimates of parenteral transmission consisted of three papers describing IDU risk,[15,16,18] 26 papers providing 21 estimates for accidental percutaneous injury[20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45] (where five publications were accompanied by a more recent abstract from which the estimate was taken)[27,28,34,35,37,38,39,40,41,56], and one reporting both percutaneous injury and contaminated injection risks.[19] The 11 estimates for blood product exposure included two from Ward et al.,[49] but only the estimate in which blood products were proved to be contaminated was included in summary estimates. Forest plots of infectivity estimates are presented in Figs 1 and 2. Summaries of the characteristics of included and excluded studies are available on request.

Figure 1.

Summary of studies estimating transmission probability for parenteral exposure to HIV. Bars represent 95% confidence intervals (CI) except for Gisselquist,[19] which is a range. psz: estimate weighted by sample size; pp: estimate weighted by precision (adapted inverse variance method) - see details in method. *For needlestick and other accidental sharp injury only, without excluded studies and Gisselquist 2002 estimate[19] (no sample size available). **All excluded studies except for superseded publications and those without sample sizes.

Figure 2.

Summary of studies estimating transmission probability per contaminated blood product transfusion. Bars represent 95% confidence intervals (CI). psz: estimate weighted by sample size; pp: estimate weighted by precision (adapted inverse variance method) - see details in method. *All excluded studies except for superseded publications and those without sufficient data.

Parenteral Exposure

Nineteen of the 25 estimates were derived from prospective studies (one for IDU and 18 for needlestick injury), in which study participants were typically followed up for about 9 to 12 months following one contact/exposure for needlestick or many injection events for IDUs. Studies tended to be prospective in that health care workers or laboratory workers were encouraged to report exposures, after which they were followed up, rather than enrolment in a study before exposure. Two studies of needlestick transmission were retrospective with health care workers sometimes reporting more than one contact/exposure.[20,21] One was of unspecified design,[44] while Gisselquist[19] used data from a case-control study[57] to derive an estimate. The remaining two estimates, one for IDU[18] and one for contaminated injections,[19] were also indirectly derived from mathematical models (one for each type of exposure).

Injection Drug Users. Infectivity estimates for IDU ranged from 0.63%[16] to 2.4%[18] (median = 0.8%)[15,16,18] (Fig. 1). Kaplan and Heimer favoured their lower estimate of 0.67% per exposure,[18] and Hudgens et al.[15] concluded that their 0.84% estimate was consistent with this. Given the heterogeneity of estimates between studies and subtypes, it was inappropriate to combine results.

Contaminated Injections. The only study available evaluated the estimated infectivity per contaminated injection as 1.9-6.9%, a very wide range, based on data of limited quality from nosocomial outbreaks in Russia, Romania and Libya[19] and on a large number of assumptions.

Accidental Percutaneous Injury. Infectivity estimates for accidental percutaneous injury are more reliable than those for IDU and contaminated injections, primarily because the number of exposures is usually one per person, and the infection status of the index case can be determined relatively easily. In the few instances where an individual experienced more than one exposure, these were counted as separate events in our analysis. All studies calculated transmission probabilities as the proportion of exposure events resulting in transmission, except for Gisselquist,[19] who calculated the HIV transmission probability for deep injuries only, to be used as an approximation for contaminated medical injections. The author makes a large number of assumptions, and deep injuries may involve transfer of more infective fluid than contaminated injections. As no confidence intervals were provided or could be derived, this estimate was excluded from summary estimates.

Infectivity estimates for needlestick exposure from all 22 studies ranged from 0.00%[21,22,23,24,25,26,27,28,29,30,32,33,34,35,42] to 2.38%[20] (pp = 0.23%; 95% CI = 0.00-0.46%; psz = 0.24%; 95% CI = 0.14-0.40%, n = 21). Transmission of HIV via this route is uncommon and therefore the majority of studies produced estimates of 0.00% (,[21,22,23,24,25,26,27,28,29,30,32,33,34,35,42]n = 13); the average sample size for these studies was 67 injuries, compared with 644 injuries for studies providing non-zero estimates. When including only studies reporting at least one transmission event and excluding Gisselquist,[19] estimates ranged from 0.18%[36,37] to 2.38%[20] (pp = 0.25%; 95% CI = 0.01-0.49%). Estimates of included studies with pp and psz gave similar results, but we prefer to report pp which produced wider confidence intervals ( Table 1 ). There was no evidence of heterogeneity between studies (df = 20; P = 1).

Analysis of subsets of studies is shown in Table 1 . The estimate based on studies with no other reported risk factors for HIV transmission (pp = 0.13%; 95% CI = 0.00-0.54%) was not significantly different from the overall estimate (pp = 0.23%). The infectivity of symptomatic patients (AIDS or AIDS plus AIDS-related complex patients) approximately doubles (psz) or triples (pp) the risk. However, results should be interpreted with caution because of the small sample sizes.

Blood Product Transfusion

Infectivity estimates for transmission per transfusion with infected blood products are presented in Fig. 2. Eight of the 10 estimates are based on retrospective and two on longitudinal studies following transfusion with infected blood products.[52,55] Infectivity estimates ranged from 27.0 to 100.0%, with summary estimates of pp = 82.2% (95% CI = 79.0-85.4%) and psz = 80.2% (95% CI = 76.7-83.3%). However, the 10 studies displayed considerable heterogeneity (Q = 198.36; χ9 2: P < 0.0001). The six studies which could confirm that donations were contaminated[49,50,52,53,54,55] were more homogeneous (Q = 5.99; χ6 2: P = 0.42431) and produced higher infectivity estimates (range, 88.3-100.0%; pp = 92.5%; 95% CI = 89.0-96.1%).


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: