Causes of death in the UK: results from BHIVA audit

Simon Collins, HIV i-Base

The annual BHIVA audits are an essential mechanism for highlighting issues relating to management of HIV-positive patients in the UK. Over 130 clinics (80% from outside London), participate by providing information on an aspect of care. A wide range of clinics are involved: roughly 20% of participating clinics in this audit treat <50, 50-100, 101-200, 201-500 and >500 patients.

This audit addressed causes of death, and results were detailed in an oral presentation by Sebastian Lucas at ths years Autumn BHIVA Conference.

Data was analysed from case note reviews from 90 centres that included 387 deaths among adults with HIV during the audit period of October 2004 – September 2005. Although the date of death was missing for eight patients, these were still included in the analysis.

It is positive that the annual rate of deaths in the UK patients diagnosed patients under care is relatively low. 40 centres reported no deaths among their adult HIV patients in the preceding year, including 52% of those serving 100 or fewer patients.

Demographic breakdown included: roughly three-quarters of deaths were men and one quarter women; 57% white, 33% African, 2% Caribbean, rest unknown; age: 65% between 30-50 years old and 27% were over 50; 72% occurred in a UK hospital and 22% in a community setting in the UK; 80% had no history of injection of non-prescribed drug, 9% had discontinued prior to their final illness, and 5% continued injecting drug use until their final illness.

CD4 count in the preceding 6 months was less than 50 cells/mm3 in 35% cases and over 200 cells/mm3 in around 25% cases. Viral load was >50 copies/mL in 60% patients but only high (over 100,00) in 25%, with just over 30% cases having an undetectable viral load. Deaths of patients who were effectively responding to ARV treatment largely overlapped with the 32% categorised as being unrelated to HIV. These included: 30 malignancies (7.8% of all deaths), 22 liver disease (5.7%), 17 cardiovascular disease (4.4%), 7 suicide (1.8%), 7 sepsis (1.8%), 6 accident/injury (1.6%), including one homicide, 4 (1.0%) overdose, 1 (0.3%) renal disease and 29 (7.5%) other or not stated.

Although 25 people died of cardiovascular disease, 17 of these were recorded by the centre under ‘unrelated to HIV’. Similarly, 26 deaths recorded as due to liver disease were also categorised as non-HIV related. A comment from the presenter suggested that this was an area where more careful categorisation in the future could relate these to an HIV-related complication.

The second largest category related to around 25% deaths (n=93, one-third of all HIV-related deaths) being due to late diagnosis. This accounted for >40% of deaths at clinics treating less than 200 patients, compared to 13% of the deaths at centers with >500 patients. This was reported as a minimum estimate as some deaths attributed to untreatable complications of HIV involved conditions which early treatment could have prevented. Also, there may be under-ascertainment of deaths occurring without involvement of HIV specialist services.

Late diagnosis disproportionally affected younger people (10% of these deaths vs 5% related to other causes) and non white patients (31% were white compared to 65% in other causes of death). Deaths attributed directly to late diagnosis of HIV were: PCP (28), opportunistic infections (16), TB (9), lymphoma (8), sepsis (8), multi-organ HIV (7), Kaposi’s Sarcoma (3), cardiovascular disease (3), renal (2), malignancy (1), other or multiple HIV related (6) and unknown (2). 16 cases suggested a possible clinician delay in diagnosing HIV.

Eleven deaths (3%) were due to multiple drug resistance and running out of treatment options.

Six deaths related to worsening HIV-related symptoms shortly after starting HAART, suggesting IRIS.

Five deaths were related to possible of probably adverse reactions to HIV treatment (3 lactic acidosis, 1 fulminant liver failure (isoniazid), 1 pneumonia possibly associated with non-Hodgkins lymphoma chemotherapy-related bone marrow suppression).

Patient factors included 18 deaths where people had refused treatment (only 3 of whom had used ARVs at some time). 31 deaths were also reported as being directly linked to poor adherence, mainly including bacterial infections (sepsis 12), and included 3 relating to MDR-HIV.

Only 12 patients were known to have arrived in the UK in the prior 6 months, nine of whom died as a result of late diagnosis.

This reinforces other evidence that immigration to the UK is not driven by access to HIV treatment: the majority of late diagnoses were in people who were established in the UK. However, although no deaths were reported as being related to ineligibility for NHS treatment, concern for this is very likely to be a factor for not accessing care. Importantly, deaths of UK patients who have been denied leave to remain in the UK, and have been deported to countries with very limited access to ARVs, are clearly not included in these figures.

Part of the rationale for each audit is to raise awareness and improve standards of care. This study has identified some specific issues, including: mechanisms for informing centres when patients have died in the community or at tertiary referral centres, the importance of good communication and prompt, effective referral pathways, the value of death reviews (used in less than half the deaths reported), and the need for improvement in death certification (information was missing form 50% of certificates, and HIV was also commonly not directly recorded).

Some centres reported that data gathering for this study was an instructive exercise in itself, and identified issues of communication and record-keeping that could be improved.


This audit does attempt to provide information on all deaths, but information on incidence and reporting from a wide range of clinics across the UK.

In general, it is positive that death rates are low, with one third unrelated to HIV – though both cardiovascular and renal causes may be seen as HIV-related in the future – and that only 3% were related to multiple drug resistance

A high risk associated with late diagnosis is still a serious concern. Without appropriate treatment, HIV is still a potentially fatal illness.


Lucas S. 25 years on: the causes of HIV-related death: results of BHIVA audit. BHIVA Autumn conference. 13-14 October, 2006.

Slides and supporting information relating to previous audits are available on the BHIVA website:

Links to other websites are current at date of posting but not maintained.