Counting the Infections a Hospital Makes

AAn infection is counted as hospital-acquired if it first appears forty-eight hours or more after a patient is admitted, and the neatness of that rule conceals how much work it is asked to do. The cut-off is a convention rather than a discovery: it was fixed to give surveillance a usable boundary, and no organism alters its behaviour on the second morning. Around it has grown an apparatus of counting, public reporting and, since 2016, financial penalty. What the apparatus cannot settle is the question that matters to patients, which is what proportion of these infections need never have happened at all. Estimates of that preventable fraction published between 2005 and 2020 range from a fifth to two-thirds, a spread wide enough to accommodate almost any policy. The disagreement is not really about microbiology, on which there is little argument. It is about whether the figures hospitals report are measuring infection or measuring the diligence with which infection is looked for.

BTeresa Kaldas, an epidemiologist at the Vasari Institute of Clinical Epidemiology, has spent fifteen years pressing the second possibility. In 2014 the national registry she advises revised its definition of a central-line bloodstream infection, excluding cases in which an alternative source of the organism could be identified elsewhere in the body. Reported rates across the participating hospitals fell by 19 per cent over the following year. In the same period the number of blood cultures taken per thousand patient-days fell by 8 per cent, a change nobody had requested and nobody could easily explain. Kaldas does not suggest that the revision was made in bad faith; the new definition is, by any clinical standard, the better one. Her objection is narrower and harder to answer. A rate that moves when the definition moves cannot be read as a plain record of what happened to patients, and the registry printed the two figures in separate documents.

CHer second line of evidence concerns the way compliance with hand hygiene is established. In an audit her group completed in 2019, wards were watched in the conventional manner, by trained observers standing on the ward with clipboards, and at the same time by sensors carried on staff clothing which logged every approach to a dispenser. Observed compliance averaged 71 per cent; the badges, over the same shifts and the same staff, returned 34 per cent. The gap did not narrow as the observers became familiar figures, and it grew wider on the busiest shifts. Kaldas is careful about what this establishes. It does not establish that hand hygiene fails to prevent infection, a proposition she treats as settled. It establishes that the figure a hospital publishes describes behaviour in the presence of an observer, and that a programme judged by that figure may succeed on paper while the wards beneath it are unchanged.

DDominic Aherne, an intensive-care physician who directs the infection programme at the Rosbeg Trust, regards such scepticism as a luxury of people who do not have to run a ward. Between 2018 and 2021 his unit introduced a standard bundle across twenty-three intensive care units: full barrier precautions at insertion, a daily review of whether each line was still needed, and antiseptic dressings. Catheter-related bloodstream infections fell from 3.6 to 1.2 per thousand line-days. Aherne concedes two points without being pressed. The bundle arrived while a separate national campaign was persuading clinicians to use fewer lines in the first place, so part of the fall reflects fewer opportunities rather than safer practice; and the improvement was confined to intensive care, with surgical wards recording no change worth reporting. He regards the first concession as the harder of the two to argue away.

EMoney entered the argument in 2016, when several national payers stopped reimbursing hospitals for treating infections judged to be hospital-acquired. The intention was to make prevention pay for itself; the immediate effect was a surge in paperwork. The proportion of infections recorded as present on admission rose from 11 per cent to 26 per cent within two years at the hospitals Kaldas studied, with no matching change in how ill the arriving patients were. Aherne accepts those coding figures and draws a different lesson from them: a penalty that can be avoided by clerical means will be, and the answer is to audit the coding rather than to abandon the incentive. Where the two converge is on the way the economics are usually presented. An averted bloodstream infection was valued at about £7,200 in a 2019 review, but the claim that a full infection-control programme repays its cost within a year is the one assertion in this field that no controlled study has supported.

FWhat would settle the matter is a trial in which wards are allocated at random, outcomes are judged by assessors who do not know which arm a ward belongs to, and the definition of infection is frozen for the duration — a design proposed repeatedly and funded once, in a study still recruiting. The two share more ground than their exchanges imply. Neither expects the preventable fraction to reach a hundred per cent; neither will treat a published national trend as trustworthy without knowing what has happened to the definitions behind it; and both have objected to the habit of quoting a percentage reduction without stating the definition, the case mix and the period from which it came. They differ over what to do while the evidence is missing. Kaldas would fix the measurement before the programmes are expanded; Aherne argues that a bundle which costs little and harms no one should be spread now and evaluated afterwards.