Self-reported HIV testing practice among physicians treating tuberculosis in Australia and New Zealand

INTRODUCTION

There is a strong interaction between the human immunodeficiency virus (HIV) and tuberculosis (TB) epidemics. Patients with HIV infection are more likely to develop active Mycobacterium tuberculosis (MTB) disease, develop disseminated disease, ¹ reactivate latent MTB infection, ² become re-infected with MTB ³ and present with extrapulmonary ⁴ or smear-negative disease ⁵ than people without HIV infection.

There are clear benefits for people with HIV infection to know their HIV status, with treatment-associated increased survival and reduced HIV transmission behaviours being well documented. ^6,7 Despite this, only 21% of people with a new diagnosis of TB had their HIV status reported to the Australian national TB surveillance register between 1999 and 2005, with 4.7% (74/1583) of those reported having both infections. ⁸ It is unclear as to whether the other 79% of cases were offered an HIV test. TB is an acquired immunodeficiency syndrome (AIDS)-defining illness and TB was reported as the AIDS-defining condition in 4.6% (n = 28) of AIDS cases in Australia between 2003 and 2005. ⁹ National guidelines in Australia do not currently specifically recommend a routine offer of HIV testing in cases of TB. ^10,11 In contrast, the Guidelines for Tuberculosis Control in New Zealand recommend a routine offer of HIV testing in cases of TB. ¹²

Previous studies have identified that patient factors influence physician HIV testing behaviour when treating people with TB, as does the specialization of the physician. ¹³ Significant rates of HIV infection were identified in groups that physicians perceived to be at ‘low risk’ and did not elect to offer testing to. Systematic questioning by skilled staff using a standardized instrument did not identify HIV risk factors in all HIV-infected TB patients in one study. ¹⁴

Our objective was to assess the rates of clinician self-reported HIV testing practice in patients with TB and to ascertain the factors associated with a routine offer of HIV testing by clinicians in cases of TB in Australia and New Zealand.

MATERIALS AND METHODS

The link to an anonymous, on-line questionnaire was emailed to members of the Thoracic Society of Australia and New Zealand and the Australasian Society for Infectious Diseases. The questionnaire was designed by the authors and piloted among the target population.

The questionnaire included demographic data and asked recipients to estimate the number of HIV tests they had ordered, the number of patients with HIV that had consulted them (whether TB related or not) and the number of cases of TB diagnosed or treated in the past five years. Participants were asked to indicate how often they would order an HIV test in 10 differing scenarios on a Likert scale that included the responses ‘always’, ‘mostly’, ‘sometimes’, ‘rarely’ or ‘never’. The first scenario (referred to as the ‘standard case’) was a 38-year-old married man with smear-positive pulmonary TB. One factor was altered in subsequent scenarios to ascertain patient factors that were associated with different rates of testing. The questionnaire concluded with free text fields asking about any factors that would increase or decrease the likelihood of offering HIV testing in the respondents' practice. The scenario data were dichotomized into those who universally offered an HIV test in each scenario (‘always’) and those who did not (all other responses).

McNemar's test for correlated proportions was used to compare the proportions of respondents that would always offer an HIV test in each scenario. Logistic regression using a forward conditional model (SPSS version 14) was undertaken to identify key factors predictive of a universal offer of testing. Australian jurisdictions with less than 20 respondents were pooled for this analysis. Odds ratios (OR) and 95% confidence intervals were calculated. Significance in all tests was set at P < 0.05 and tests were two tailed. The study was approved by the human research ethics committees of the University of New South Wales (NSW) and South Eastern Sydney and Illawarra Area Health Service.

RESULTS

Responses were received from 309 subjects after invitation emails were sent to 918 people on the email distribution lists of the two societies. Nineteen responses were excluded, as respondents were either not currently working in Australia or New Zealand (n = 5) or were working exclusively in paediatrics (n = 14). The characteristics of the respondents and the frequency with which the respondents self-reported that they would always offer an HIV test to the ‘standard case’ are shown in Table 1. To assess the self-reported TB caseload of respondents, we used the minimum of the range of reported number of TB cases diagnosed or treated by the responding clinicians. The minimum number of cases managed by the sample was 2088 in five years. Only three (1%) respondents had not managed a case of TB in five years.

Sixty-one percent (177/290) reported that they would always offer an HIV test to the ‘standard case’, with 52 (18%) stating that they would mostly offer a test, 12 (14%) sometimes, 16 (5%) rarely and three (1%) never offering a test in this scenario.

The proportions of respondents that would always offer an HIV test in different clinical scenarios are shown in Table 2. In comparison to the ‘standard case’, respondents were significantly more likely to report always testing cases with oral candidiasis, those from South Africa and an unmarried male case. Significantly fewer clinicians always reported testing a 78-year-old case, a smear-negative case of TB and a female case compared with the ‘standard case’. There was no difference in the frequency of a universal offer of HIV testing with extrapulmonary TB or those who were described as aboriginal or Caucasian Australian.

Key predictors of universal HIV testing in the ‘standard case’ were identified using logistic regression analysis (Table 3). Variables including clinician gender, age, place of work, being in charge of a chest clinic, number of TB cases managed and number of HIV cases consulted in the last five years were not significant in the model. Infectious diseases physicians were more likely to report always recommending an HIV test than respiratory physicians. Those who had performed 10 or fewer HIV tests in the last five years were less likely to universally offer HIV testing. There were also differences seen in the location of practice, with New Zealand clinicians being significantly more likely to universally offer HIV testing than New South Wales clinicians. Clinicians in the a priori combination of ‘other Australian jurisdictions’ were more likely to test than their NSW counterparts.

Logistic regression analysis of the other scenarios found the strongest predictor of universal HIV testing to be the specialty of the clinician and in most scenarios the number of HIV tests undertaken in five years remained significant (Table 4). Clinicians from New Zealand were more likely to always test in the case of extrapulmonary disease than their colleagues in NSW. Specialist physicians were less likely to test a female case than their trainee colleagues.

Of the 290 subjects, 246 (85%) responded to the question ‘What factors if any would make you more likely to test for HIV?’ Of these, 107 (43%) reported always offering an HIV test regardless of any perceived risk factors; 139 (48% of all respondents) reported that certain factors influenced their decision to offer an HIV test. These factors were grouped into themes and included HIV epidemiological factors (132), clinical HIV factors (65), clinical TB factors (43) and others (63) (including comments that were not easily classifiable such as ethnicity, age and history). The factors that respondents reported that would make them less likely to test mirrored these, with 76 citing a lack of epidemiological risk factors, 11 citing a lack of clinical HIV factors, 47 TB clinical features and 124 reporting other factors. These factors are summarized in Table 5.

DISCUSSION

This is the first study examining clinicians' self-reported HIV testing behaviour for patients with TB in the Australasian setting. We found 61% of clinicians self-reported that they would always HIV test a ‘standard’ TB case and 43% stated that they would always HIV test any case with TB. Significantly more respondents stated that they would test a patient with oral candidiasis, a clinical sign of HIV-related immunodeficiency. ¹⁵ However, 13% did not always offer an HIV test to a case with oral candidiasis, despite the clinical significance of this physical sign. Interestingly, not all physicians used TB factors to guide testing, as the case with extrapulmonary TB was always tested at the same frequency as the standard case, and the smear-negative TB case was less likely to be always tested. This is of some interest, as both of these factors are associated with an increased prevalence of HIV infection in the TB-infected population. ^4,5

HIV epidemiological factors appear to be in use by some clinicians as in the case of South African origin (a country with a higher HIV prevalence than Australia or New Zealand) and the unmarried male cases (a possible surrogate for homosexuality for some clinicians) were always HIV tested by more clinicians, and the older male and the female cases were tested significantly less often (possibly as the prevalence of HIV infection in Australia is significantly lower in women and the elderly). ⁹ However, significant rates of HIV infection were found in these traditional ‘low-risk’ groups in a similar population in Canada. ¹³

Infectious diseases physicians were significantly more likely to always offer an HIV test to the standard case than respiratory physicians (80% versus. 43%). This emerged as a strong independent factor and corresponds with previous reports. ¹³ The location of practice was another significant influence on the likelihood of being offered an HIV test, with those from New Zealand and clinicians in a group of regions in Australia being significantly more likely to offer a test than those practicing in NSW. Is this surprising? The TB policy for New Zealand recommends a routine offer of HIV testing of all TB cases ¹² and 56–68% uptake on this recommendation has been seen in countries where a similar policy exists. ^16,17 The group of other Australian jurisdictions includes the Northern Territory, which has a clearly stated policy to always offer HIV testing to people with TB. Clinicians that reported ordering 10 or fewer HIV tests in the past five years were less likely to test TB patients for HIV infection and this may suggest a degree of discomfort, lack of skill, lack of time or lack of resources (e.g. interpreters) to assist with the HIV testing process. A recent review of reasons that physicians do not HIV test in the United States found policy level, logistical and educational barriers. ¹⁸ Logistical barriers were mentioned by some participants in our study and they included forgetting to offer a test, patient refusal and language barriers, although our study was not designed to assess this.

A number of respondents who did not always test described risk factors and other patient factors that would increase their likelihood of testing, including HIV epidemiological factors, clinical HIV factors or TB-related factors. An Australian study looking at HIV screening in pregnancy also found risk factor assessment as the most commonly reported reason for not always antenatally HIV testing. Pre-test counselling and time factors were other issues implicated. ¹⁹

Our study was not designed to assess clinicians' ability at determining risk factors or how often they do this. Selective screening is very dependent on physicians' knowledge of risk factors and an ability to take the appropriate history. Up to 9% of people diagnosed with HIV in Australia have no risk factor for infection identified, signifying that risk factor ascertainment is not always possible. ⁹

There are limitations to this study, chiefly the nature of self-reported behaviour using a questionnaire and the response rate to our survey. Although a third of potential subjects responded to our survey, 41 (14%) of respondents reported being in charge of a chest clinic and the responding group reported that they had diagnosed or treated at least 2088 TB cases in the previous five years. This would represent at least 40% of all TB cases in Australia over five years. ⁸ Also, less than 1% of our sample reported that they had not diagnosed or treated a case of TB in five years and therefore we believe that we sampled the target population. We did not sample general physicians or other medical practitioners that may provide care for people with TB. There are little data to assess if the clinician's self-reported behaviour reflects their clinical behaviour. The data reported to the Australian TB register suggest that the rates of HIV testing in actual practice are lower (with 37% reported to the register in 2005), but this could be due to under-reporting and the register does not record whether a test was offered but declined.

At least 1% of cases of TB in Australia have HIV infection and some of us have previously recommended that all patients diagnosed with, or treated for, TB, regardless of identified risk factors, be offered an HIV test. ²⁰ We suggest that this recommendation be included in the Australian National TB Strategic Plan, the Australian National HIV Testing Policy and all state and territory TB management guidelines.

In those countries where HIV testing is recommended as part of the TB management, the HIV status is reported in up to 70% of cases, ^16,17 so policy change is only one area that needs to be addressed. Further research is needed to identify the barriers to HIV testing in the Australian and New Zealand setting where TB is diagnosed and treated. Ways to modify clinician behaviour in this setting may include: (1) targeting relevant clinicians for education and training around HIV testing (such an approach has increased the antenatal HIV testing rates in Australia); ²¹ (2) including a written recommendation for HIV testing on any diagnostic pathology results where active TB is diagnosed and (3) considering task transfer, with the responsibility for initiating and performing HIV testing being shifted to nurses or other skilled personnel, which has been found to be both efficacious and satisfactory to patients in other settings. ²² An alternative approach may be to adopt opt-out testing. ²³

A number of clinicians in our sample report that they believe that a universal offer of HIV testing should be made to all patients diagnosed with TB. We have found that other clinicians who use a selective testing approach use patient and clinical factors to assist in the decision to test, even though some do not identify cases at significant risk of HIV infection. We recommend policy change, education and practice change to increase the frequency that HIV testing is offered to patients with TB. The prevalence of HIV infection in this population is sufficient to justify a routine offer of HIV testing, even in the absence of identified HIV risk factors. ²⁰