University of California, San Francisco Logo

University of California, San Francisco | About UCSF | Search UCSF | UCSF Medical Center

Home > Global Health Literature Digest > A Cluster-Randomized Trial
A cluster-randomized trial of provider-initiated (opt-out) HIV counseling and testing of tuberculosis patients in South Africa
Global Health Sciences Literature Digest
Published December 16, 2008
Journal Article

Pope D, DeLuca N, Kali P, et al. A cluster-randomized trial of provider-initiated (opt-out) HIV counseling and testing of tuberculosis patients in South Africa. JAIDS 2008 Jun;48(2):190-95.

Objective

To determine whether the implementation of provider-initiated HIV counseling would increase the proportion of tuberculosis (TB) patients who receive HIV counseling and testing

Study Design

A cluster-randomized intervention trial of TB clinics with a 1:1 ratio of allocation to the intervention or to standard HIV counseling and testing (control clinics)

Setting

Twenty medium-sized primary care TB clinics in Port Elizabeth, South Africa. The clinics were selected based on the presence of a designated TB nurse and new registrations of at least 3 TB patients per month.

Participants

A total of 754 adult (18 years and older) TB patients who newly registered at one of the 20 selected clinics between August 12, 2005 and November 10, 2005 were included in the analysis. Patients were eligible for inclusion only if they were at least 18 years old and remained in care for at least 14 days after date of registration to allow them time to receive HIV counseling and testing.

Intervention

Randomization was constrained to balance the number of patients seen each month, the number of TB cases registered each month, the number of TB patients reported to have received HIV counseling and testing in the months before the intervention, and a summary score of the extent of TB/HIV collaboration, as determined by study staff. Clinics randomized to the intervention initiated a program of provider-initiated HIV counseling, which shifts the decision-making burden from the patient to the provider. Nurses at all 20 selected clinics received HIV counseling and testing training and were presented treatment protocols according to the national guidelines. Nurses at clinics randomized to the intervention were invited to attend a 2-day training course to learn the strategy of routine provider-initiated counseling and testing. No new personnel were provided for the clinics.

Primary Outcomes

The main outcome was the percentage of newly registered TB patients who received HIV counseling and testing. Secondary outcomes were the percentage of patients that tested HIV positive and the percentage that received cotrimoxazole for prophylaxis of opportunistic infections and referral for HIV care. A registry form was specifically designed for the study to document the dates of HIV counseling and testing and the results.

Results

Between August 12, 2005 and November 10, 2005, 1,072 adult TB patients registered at 20 selected clinics; 754 (70%) met the study inclusion criteria and were included in the analysis. The majority of patients had sputum smear tests that were positive for pulmonary disease, and a significant proportion had a history of prior TB treatment. Overall, 13.8% (104/753) of patients received HIV counseling and testing. A significantly higher proportion of patients in the intervention clinics received HIV counseling, 20.7% (73/352), compared to those in the control clinics, 7.7% (31/402)(P = 0.011), and HIV testing, 20.2% (n=71) versus 6.5% (n=26) (P=0.009). Of those counseled, 97% of patients in the intervention clinics accepted testing versus 79% in the control clinics (P=0.12). A total of 8.5% of patients in the intervention clinics were found to be HIV-infected, whereas 2.5% were found to be HIV-infected in the control clinics (P=0.044). In both study arms, fewer than 40% of patients identified as HIV-infected were prescribed cotrimoxazole (6/31 and 4/11 in intervention and control clinics, respectively).

Conclusion

The authors concluded that the use of an opt-out strategy was associated with significantly higher HIV counseling and testing rates and that the time to testing tended to be faster in the opt-out arm than in the control arm, but that the overall proportion of those counseled and tested in the intervention arm remained unacceptably low (21%).

Quality Rating

This was a randomized clinical trial of good quality. Randomization was constrained so that the 2 study arms were nearly balanced in terms of the total number of clinic patients seen each month, the number of TB cases registered each month, the number of TB patients reported to have received HIV counseling and testing in the months before the intervention, and a summary score of the extent of TB/HIV collaboration as determined by study staff. A customized SAS macro was used to realize the desired constrained randomization. Outcome measures were captured by a standardized form in clinics in both arms, although there is no report of quality control over compliance in filling out the form by clinic personnel. Sample size calculations and analysis of data correctly accounted for the clustered study design. Generalizability of the findings may have been limited by including only clinics with three or more new TB registrations per month, although it is not clear what bias, if any, this might introduce.

In Context

Although an opt-out provider-initiated strategy should increase the uptake of HIV testing by TB patients by shifting the burden of decision making from the patient to the health care provider, the data supporting this approach in African settings has not been systematically evaluated. A cluster-randomized trial that trained TB nurses in The Free State Province of South Africa to use an algorithm for the diagnosis and management of respiratory disease, including TB, had voluntary counseling and testing rates of only 9.7% in the intervention group and 7.3% in the control group.(1) That said, recently TB programs in Rwanda, Malawi, and Guyana have reported substantial increases in HIV testing of TB patients when national initiatives that provide additional training and resources were coupled with the use of the opt-out counseling and testing strategy.(2,3)

Programmatic Implications

As the authors mention, this study was conducted in primary care setting that was already overburdened with a large clinical workload. Although HIV testing for TB patients is recommended by South African and international guidelines, implementation depends on the capacity of local health facilities to additional clinical tasks into already overstretched workloads. The opt-out strategy increased the proportion of patients who were tested for HIV, but the overall low uptake of 21% may reflect a weakness of the opt-out strategy or just a lower priority placed on a research project by clinic personnel. The authors recommended that additional interventions to optimize HIV testing for TB patients need to be evaluated.

References

  1. Fairall LR, Zwarenstein M, Bateman ED, et al. Effect of educational outreach to nurses on tuberculosis case detection and primary care of respiratory illness: pragmatic cluster randomised controlled trial. BMJ. 2005;331:750-754.
  2. WHO. Global Tuberculosis Control: Surveillance, Planning and Financing. World Health Organization, Geneva, Switzerland. 2007.
  3. Persaud S, Mohanlall J, Bateganya M, et al. HIV counseling, testing, and care of tuberculosis patients at chest clinics-Guyana, 2005-2006. MMWR Morb Mortal Wkly Rep. 2006;55:849-851.