Paperonium

HIV · Epidemiology · 2016

UK HIV estimate, 48,310, lies above its own range, 39,900–45,560; the paper’s chart shows about 40,000–60,400.

The summary and resultsIts chart (Figure 3A)about 48,50048,31039,900 to 45,560should be inside40,000 to 60,40040,000 men45,00050,00055,00060,000
The summaryand resultsIts chart(Figure 3A)about 48,50048,31039,900 to 45,560should be inside40,000 to 60,40040,000 men50,00060,000

Number living with HIV in 2013: each estimate and its range

The paper says

The summary and results estimate that 48,310 men who have sex with men were living with HIV in the UK in 2013, with a range of plausible values of 39,900 to 45,560, all from 742 simulated runs of the epidemic that fit UK data well.

Its own data show

The paper defines the estimate as the middle run (half the runs give lower totals, half higher) and the range as all but the lowest and highest 5%. So the estimate must lie inside; it is 2,750 above the top. The paper’s chart of the same runs (Figure 3A) shows about 40,000 to 60,400.

A range like this shows how uncertain an estimate is: how high or low the true number could plausibly be. The estimate checks out (its four groups by immune-cell count add up to exactly 48,310); going by the chart, the printed top is about 15,000 too low. Anyone quoting the summary would wrongly rule out totals above 45,560.

“The summary” means the paper’s abstract: the short summary printed at its top, which is the part most readers see.

Read the paper →Full technical write-up →

A Method to Estimate the Size and Characteristics of HIV-positive Populations Using an Individual-based Stochastic Simulation Model · Epidemiology (2016) · doi:10.1097/EDE.0000000000000423