Adopting universal testing for HIV in intensive care for patients admitted with severe pneumonia: results from our change in practice

Introduction

Despite HIV now being an eminently treatable condition, a large number of HIV-positive individuals remain unaware of their diagnosis, with late diagnosis remaining the most important factor associated with morbidity and mortality.^1,2

UK guidelines suggest HIV testing should be offered to patients presenting with pneumonia. ¹ Prevalence is likely to be higher within intensive care unit (ICU) patients compared to the general population, where early diagnosis in critically ill patients is vital, can guide appropriate investigations, allow prompt access to specialists and markedly improve outcomes. ³ Systems that improve HIV testing rates in ICUs are increasingly attractive. Relying on risk factors alone to guide testing yields poor sensitivity.

We reviewed patients admitted to our ICU with pneumonia to determine how many had an HIV test on admission or during their inpatient stay. We then incorporated routine HIV testing into our ‘ICU pneumonia screen’ and reviewed our practice to ascertain its effect.

Method

All adults admitted to our ICU between 1 November 2013 and 31 October 2014 with a diagnosis of ‘pneumonia/chest sepsis/chest infection/respiratory-tract infection’ were included. Tests requested and results were then retrospectively reviewed using our local electronic results system.

A routine HIV test was added to our automated ‘ICU pneumonia screen’ in June 2015, along with staff education. We repeated the audit process as above for admissions between 1 July 2015 and 29 February 2016.

Results

Prior to our intervention, 19 patients (28%) were tested for HIV within 24 h of admission with pneumonia with 44 (65%) not tested at any stage. After our intervention, 43 patients (73%) had an HIV test within 24 h of admission, and 47 (80%) within 48 h. See Table 1 for full results.

OPEN IN VIEWER

Table 1.

Breakdown of patients admitted to ICU with pneumonia as percentage of total pneumonia admissions during respective phase (absolute numbers in brackets).

	HIV test				No HIV test
	<24 h after ICU admission	24–48 h after ICU admission	Two weeks after ICU admission	Within two months of ICU admission	Test not requested	Test not sent	Patient died before testing
Phase 1 (n = 68)	28% (19)	0% (0)	1% (1)	3% (2)	65% (44)	3% (2)	0% (0)
Phase 2 (n = 59)	73% (43)	7% (4)	2% (1)	0% (0)	12% (7)	3% (2)	3% (2)

One patient returned a positive HIV test, and the remaining patients returned a negative result.

Discussion

Our regional HIV prevalence is 0.7/1000 population; below the 2/1000 threshold recommended by The British HIV Association (BHIVA) for implementing universal testing.^1,4 However, the neighbouring region has a prevalence of 2.07/1000 with likely patient crossover and sexual interaction. ⁴ Therefore, targeted HIV testing is vital. BHIVA recommends testing is offered in patients with pneumonia, ¹ and recent publications suggest ICUs adopt a system for universal testing in certain situations: community-acquired-pneumonia is one. ³

We found a low level of HIV testing in our unit, with only 35% of pneumonia admissions having an HIV test within two months, and only 28% within 24 h of admission, although this is higher than data from other units. ³

After departmental discussion, an HIV test was included in the electronically requested pneumonia admission profile. All patients who received the ‘ICU pneumonia screen’ automatically had an HIV test, irrespective of perceived risk factors. By sending the test universally, all subjective assessment of risk for HIV and stigma was removed.

Following the change in practice, 82% of patients received an HIV test during their admission with pneumonia; 73% within 24 h of ICU admission. Only nine patients (15%) missed having an HIV test due to failure to request and send the test, and this may in part be due to staff confusion. Ultimately, the decision to send the ‘Pneumonia screen’, and therefore an HIV test, rests with the treating clinician, and it may have been that some/all of these remaining patients did not warrant the tests clinically (e.g. perioperative aspiration pneumonia). There was a delay in the requesting/sending of the HIV test in 9% of patients, and this may indicate a need for further staff education.

This new approach provided significant benefit. Patient X presented with severe pneumonia, and deteriorated despite first-line therapy, requiring multi-organ support for septic shock and respiratory failure. She would be deemed ‘low risk’ for HIV with no obvious risk factors and would likely not have previously been tested immediately on admission. Due to our change in practice, a positive HIV result was obtained within 24 h, and she was treated for presumed (and later confirmed) PCP-pneumonia. It is likely that this early HIV diagnosis and subsequent tailored therapy were a significant factor in her survival. She was seen promptly by our HIV specialists, and rapidly commenced on antiretroviral therapy. She has since made a full recovery.

In conclusion, the addition of a routine semi-automated HIV test for all patients admitted with pneumonia has greatly increased our testing rates within ICU. This system has proven successful in identifying a patient who may otherwise have been missed, with clear benefit to their care and survival. This is in line with findings from similar testing initiatives in other ICUs. ⁵ It is thought that this change in practice will prove beneficial in reducing morbidity and mortality in patients with as yet undiagnosed HIV infections. A further discussion about testing patients admitted to ICU with other conditions (e.g. meningitis and encephalitis) may be of value. Based on our success, we suggest other centres may benefit from developing a similar system of automated HIV testing in certain presenting conditions.