Infection with no organism found

A culture measures growth, not presence. Reading “negative” as “nothing there” confuses two statements. The first question to any silent result is therefore: under what conditions could this organism have grown at all?

A swab catches the surface. What is needed is three to five tissue samples from different compartments. Several samples also separate a true finding from contamination: one organism in one of six samples means something different from the same organism in four of six.

Usual cultures are discarded after two to five days. Slow, oxygen-shy organisms need one to two weeks. Without an express order the short run applies — and “does not grow” then says more about the laboratory’s routine than about the patient.

On screws, plates and prostheses bacteria sit in a biofilm — stuck fast and running on low, and therefore hardly found in neighbouring tissue. Ultrasound loosens them; the procedure is called sonication and is done on the removed material, not on the person.

Under running treatment almost nothing grows any more. If an answer is needed, a break of about two weeks usually comes before sampling. That is a weighing-up, not a matter of course — but without it more silent cultures follow, and the question stays where it was.

16S rRNA PCR and sequencing find genetic material even from bacteria that no longer grow, and within hours. The price: no susceptibility testing. One knows who was there, but not what it responds to — which, under running treatment, can still be the only way open.

No detection does not rule an infection out. Nor does it prove one. Grafted bone can partly dissolve without infection, material can loosen mechanically. A picture only becomes suspicious through the combination: dissolving, persistent pain, and a course that does not look like healing.

Cutibacterium acnes does not sit on the skin but in the sebaceous glands beneath it. Skin disinfection reaches the surface. Measured across 50 shoulder operations, the skin surface was colonised in 100 per cent of cases after complete preparation, the dermis beneath it in 46 to 61 per cent. Typing at 90 shoulder prostheses showed further that in seven of nine people the organisms in the depth came genetically from their own skin — despite chlorhexidine and cefazolin. And it is a question of time: in one study 4 per cent of the skin was still colonised right after disinfection, and 47 per cent at wound closure. The glands refill the surface during the operation.

The only agent for which a network meta-analysis found a clear effect: 17 randomised trials with 1350 participants, 9 of them with 775 in the network. Benzoyl peroxide lowered the odds of a positive culture to about one eighth compared with soap or placebo (odds ratio 0.12; interval 0.04 to 0.36). Chlorhexidine, hydrogen peroxide and povidone-iodine showed no assured difference there. Amount and duration decide: a single application cleared 66.7 per cent in one review, several applications 78.9 per cent. The dermatological basis gives 90 per cent reduction after 24 hours and 99 per cent after 72 hours with twice-daily use — hence the usual two to three days. A three-arm trial from 2026 with 156 participants combined five applications over three days with iodine on skin and subcutaneous tissue: in the joint 31, 21 and 10 per cent, in the muscle layer 42, 33 and 13 per cent.

All these studies measure the organism load and the number of positive cultures. None measures the infection rate as its main question. The 2026 trial reports two infections among 156 participants, one each in the first two groups and none in the third — that is two events, a figure and not a proof. After shoulder arthroscopy the clinical infection rate is 0.22 per cent in an analysis of 3586 procedures. To secure a difference there would take tens of thousands of participants; no such trial exists. What is shown: the organism load can be lowered. What is not shown: that an infection is thereby prevented — neither shown nor refuted.

Not every trial found an effect. A randomised trial in 50 men compared a rinse-off benzoyl peroxide soap with a chlorhexidine wash: in the dermis 46 against 61 per cent, the difference not assured. The trials with a clear result used a gel or cream that stays on the skin. That contact time makes the difference is the authors’ explanation, not the result of a comparison. It remains a question of amount and duration.

Iodine on the subcutaneous layer after the skin incision: across 108 procedures markedly less Cutibacterium. Hydrogen peroxide: two single trials with an effect — joint cultures 10 against 35 per cent —, the network meta-analysis found no assured difference; the picture is contradictory. Changing the blade after the skin incision: across 100 procedures 12 per cent of the first blades grew something, in men 21.6 per cent — what is measured is contamination of the blade, not the benefit of the change. Clipping instead of shaving: 0 against 3.3 per cent, very small absolute figures. And duration: the longer it is open, the more time the glands have to refill.

Here there is a guideline. The German-language S3 guideline on perioperative antibiotic prophylaxis states: for arthroscopic and open shoulder operations prophylaxis shall be given, and as a single dose for a shoulder prosthesis — both by expert consensus with 100 per cent agreement. It names cefazolin, 2 grams into the vein, 3 grams above 100 to 120 kilograms, 30 to 60 minutes before the incision. Cutibacterium acnes is hardly resistant to it — and still, at 90 shoulder prostheses under cefazolin, 26.6 per cent of cultures were positive. The gap is not one of resistance but of distribution: the organism sits in glands without blood vessels and is carried mechanically into the depth by the incision.

Measured across 173 people and 692 cultures: men carried a good three times the amount women did, and those under 40 markedly more than those over 60. Both held when all factors were calculated together. Shoulder hair was a factor in men too, but not by amount — grade 1 lay higher than grades 2 to 3. Smoking, a high body mass index and diabetes showed up at first and disappeared again in the overall calculation; in smokers the load was even lower, not higher. The density of the sebaceous glands itself was never measured and set against it directly — it is the explanation, not the measurement.

The most constant sign of the silent infection is pain that never quite goes after the operation and is not explained by loading.

Dissolving around screws and plates, a loosening, a graft barely there at a later operation. Over months this is often the loudest sign.

What otherwise makes an infection recognisable is missing here. That is precisely why it is called low-grade — quiet, not harmless.

Inflammatory markers are mostly normal or only minimally raised; the white cells are sometimes even lowered. A normal CRP does not argue against infection here.

Delayed healing, oozing that lasts longer than expected, a suture behaving differently — little on its own, a building block together with the rest.

Perhaps the most important sign is not a finding but a comparison: it goes differently from everyone else after the same procedure, and nobody can say why.

Often normal or only slightly raised in the silent infection. A normal CRP does not rule it out — it is the value that most often gives false reassurance here.

Not merely unremarkable but sometimes even lowered. Anyone expecting a rise will miss it.

Slower than CRP and raised for longer. Over weeks sometimes more telling than a single value — but it too can stay silent.