Methods

Study setting

In Tanzania, the national AIDS control programme began providing free ART in public sector referral hospitals, and subsequently rolled out treatment to district hospitals and health centres [19]. HIV testing is provided by more than 500 static and mobile VCT clinics and through “opt-out” or provider-initiated testing and counselling, which has been implemented in health care facilities and antenatal clinics since 2007 [6,7].

Monitoring and evaluation (M&E) of the national ART rollout is coordinated by the Tanzanian Commission for AIDS, focusing on routine data collection for key indicators, including numbers of individuals initiating ART. In selected areas of the country, such as the TAZAMA study site, specialized M&E research is being undertaken to document the uptake and demographic impact of ART in the context of a long-term HIV cohort study [20].

This study, located in Kisesa ward in the rural north-west part of the country, collects longitudinal demographic and serological surveillance data, providing a rich background against which M&E of HIV service uptake can occur. An integral aspect of the project’s activities is to pilot data collection tools that can be adopted nationally for monitoring access to HIV services.
Development of the referral system

The process of designing the referral system was led by researchers from the Tanzanian National Institute of Medical Research, and included consultations with stakeholders involved in referring diagnosed HIV-infected clients from VCT to the HIV treatment clinic. These included VCT counsellors, HIV clinic staff, representatives from a local HBC programme (the Lutheran Church-run “Tumaini”, which supported HIV-infected persons in the area), and the national AIDS control programme. Other local projects referring HIV-positive persons to the HIV treatment clinic (such as a microbicide development study) also participated to avoid duplicating efforts and developing parallel systems which might increase the workload of clinic staff.

The aims of the referral system as defined during these consultations were: (1) to ensure that HIV-positive VCT clients were formally directed and linked to HIV treatment services; and (2) to monitor the proportion of HIV-positive VCT clients who were successfully linked to the HIV clinic, in order to assess the effectiveness of the VCT clinic as a “gatekeeper” to the HIV treatment programme.

The “Tumaini” programme, now operating under the name “Tunajali” provided home-based care in the area and conducted tracing visits among those of its clients who do not take up referral appointments. During the study period, two home-based care volunteers were recruited from each village and received a small monthly stipend of approximately US$8. From early 2006, Tumaini was also able to provide volunteer community escorts to accompany newly diagnosed patients from the local VCT centre to the HIV treatment clinic. Transportation allowances (approximately US$2 for a return trip) were also provided for patients to attend the HIV treatment clinic from early 2006 and were managed by the VCT counsellors.
Development of the referral forms

The content and format of the forms that were used to facilitate and document referral rates were developed after reviewing existing referral forms from other African settings [21], and were piloted during 2005. The final version of the referral form consisted of two detachable parts, with matching, unique numbers on each side to facilitate subsequent reconciliation of the two parts.

One side included basic socio-demographic information about the referred person and the referral date, and was given to the patient to present at the HIV treatment clinic. This section was completed by registration nurses at the HIV clinic, and included the date of the patient’s registration, allowing the delay between referral and registration to be calculated. By additionally recording the unique patient identifiers assigned by the HIV clinic on this section, referral data could be subsequently linked to the data recorded in patients’ HIV clinic files.

The remaining part of the form also included the patient’s socio-demographic information, and was retained by the referring party. The unique VCT identifier allocated by the counsellors for each patient was recorded on this slip to enable referral data to be subsequently linked to the VCT data. A template of the referral forms can be downloaded from http://www.tazamaproject.org . Referral slips were regularly collected from the VCT clinic and the HIV treatment clinic, and reconciled by a clinical research officer using a data-entry programme that generates standard anonymized monthly reports. All referral slips were stored in a locked cupboard to ensure patient confidentiality.
Quantitative methods

Data were analysed using Stata 10 (StataCorp, College Station, Texas, USA). The proportion of diagnosed clients who were referred to the HIV treatment clinic and the proportion of referred patients that subsequently registered at the HIV treatment clinic were calculated, stratified by sex and time period. Delays in registering at the HIV treatment clinic following a referral were calculated by subtracting the date of registration from the date of referral. Cross tabulations and chi square tests were conducted to assess for associations between sex or time period and uptake of a referral or registration at the HIV treatment clinic.
Qualitative methods

In order to explore the acceptability of the referral system, we conducted in-depth interviews with 11 health care workers involved in referring and receiving patients at the HIV treatment clinic 18 months after introducing the referral procedures. Within the context of a wider qualitative study exploring access to HIV services [22,23], we also conducted 42 in-depth interviews and four focus group discussions with referred patients to elicit their experiences of using the referral system. The focus group discussions and in-depth interviews were recorded with prior consent from participants and then transcribed, translated into English, and entered into NVIVO7 for analysis. A coding scheme was derived from the data by assigning codes to major concepts mentioned by the participants that were related to the referral scheme.
Ethical approval

Ethical approval for the study was obtained from the Medical Research Coordinating Committee (Tanzania) and London School of Hygiene and Tropical Medicine (UK).