A UK study using The Health Improvement Network (THIN) database to identify symptoms and clinical diagnoses associated with HIV infection has assessed their predictive value in diagnosing HIV in primary care.

The THIN database contains anonymised patient records from 386 UK general practices; these include sociodemographic data, diagnoses, treatments, clinical measurements, laboratory results, secondary care referrals and hospital diagnoses.

The study looked at patients aged 18 years and over who were diagnosed with HIV or AIDS at least a year after registration with the practice. This cohort was age- and sex-matched with five-year cohorts (ratio 1:3) who were not known to be HIV positive, registered with the same practice in the year the case was diagnosed.

National HIV Testing Guidelines published in 2008 bythe British HIV Association (BHIVA) have listed a number of HIV indicator conditions.1 However, some are not specific to HIV, e.g. any STI and any lymphadenopathy of unknown cause.

The THIN database was used to determine the incidence of each of the Read coded HIV indicator conditions during a 12-month period in cases and controls. A logistic regression model was used to determine the association between the indicator condition and HIV infection.

A total of 3,515 patients were eligible which included 939 cases and 2,576 controls. The most common indicator conditions in the year before HIV diagnosis were: diarrhoea (67 cases had at least one episode; 7.1%), oral candidiasis (n = 35, 2.7%), herpes zoster (n = 32, 3.4%), lymphadenopathy (n = 28, 3.0%) and weight loss (n = 27, 2.9%). However, three quarters (74.2%) of the cases had none of the HIV indicator conditions and 20.8% had just one (n = 195).

The diagnoses most strongly associated with HIV infection were bacterial pneumonia, oral candidiasis and herpes zoster, P < 0.001, see table 1.

Clinical signs and symptoms with a strong association with HIV included:

• weight loss OR 13.4 (95% CI:5.0-36.0)

• pyrexia of unknown origin OR 7.2 (95% CI:2.8-18.7)

• diarrhoea (one consultation OR 3.7 (95% CI:2.3-6.0), two consultations OR 4.4 (95% CI:1.6-12.1)

According to audits of new HIV diagnoses from BHIVA, many patients who were diagnosed with HIV may have had several consultations in primary care with HIV-related symptoms within a year before diagnosis.

This study would appear to provide a clear strategy for GPs to offer targeted HIV testing to patients with conditions that are much more likely to predict subsequent HIV diagnosis.

However, the authors comment that further research is needed to validate this model in general practice.

 Dr Richard Ma