We offer a monthly newsletter dealing with the various issues surrounding

infectious diseases.  To find out more click HERE.

FINAL REPORT

The Impact of HIV/AIDS on Health Service Personnel

at Two Public Hospitals in Johannesburg

Prof. Y Veriava, Principal Investigator

Dr. D Connelly Sevilla, Lead Researcher

RN A Jordan, S Roberts, and J Tsotetsi, Research Team

Asst Profs Mary Bachman and Sydney Rosen, Consultants (Boston University)

Health Economics Research Office (HERO)

Wits Health Consortium, University of the Witwatersrand

Tel: 011 276-8888 (HERO office) Email: dsevillag@yahoo.com

Technical assistance from the

Joint Economics Aids and Poverty Programme (JEAPP) 

November 1, 2005

 I. Background and Objectives

Much has been written about the burden placed on the health sector in South Africa by HIV/AIDS. Most of this work focuses on the increased demand for health services created by AIDS and the costs of providing hospital care to HIV-infected patients. Little has been published about the impact of HIV/AIDS on nurses, doctors, and the other trained professionals who are responsible for keeping increasingly over-burdened public healthcare facilities functioning. Little is known about the contribution of AIDS to the high attrition of healthcare personnel, the impact of HIV/AIDS on healthcare worker labour productivity, or the actual financial and human capital costs of the disease to the public health system.

To help fill this gap in knowledge, a research team from Helen Joseph Hospital and Coronation Hospital in Johannesburg, working through the Health Economics Research Office at the University of the Witwatersrand (HERO) and with technical assistance from the Joint Economics Aids and Poverty Programme (JEAPP) and Boston University’s Center for International Health and Development (CIHD), analyzed the impact of HIV/AIDS on health care personnel at these two sites.  

The overall objective of this study was to describe and analyze the impact of HIV on professional-level health care workers in order to assist the National Department of Health, provincial health departments, hospitals, training institutions, and other public sector agencies to develop more effective strategies for maintaining productivity and managing costs in the public health system. To achieve this general objective, we collected and analysed data in four specific areas: 1) levels of termination and replacement of nurses; 2) reasons for termination among nurses; 3) HIV prevalence in the workforce; and 4) costs of HIV/AIDS in the nursing workforce. Our analysis focused on the nursing workforce, for which we had the most complete data sets.

II. Levels of Termination and Replacement of Nurses

We obtained retrospective data from the Gauteng Shared Services Centre (GSSC) database on all healthcare professionals at our two study sites who left employment for any reason between 2001 and 2004 and all healthcare professionals who were appointed in that period. Results for professional and staff nurses are shown in Table 1. Although attrition (terminations) of nurses was fairly constant over the period studied, rates of hiring (replacement) declined substantially. As a result, the two hospitals had approximately 100 fewer nurses at the end of 2004 than at the beginning of 2001. Most nurses who left service (83%) had fewer than five years of service.  A very small proportion (5%) of those terminating service did so due to death or ill health, suggesting that nurses’ own illness due to HIV infection is not a major contributor to overall attrition.

Table 1. Turnover of nurses at Helen Joseph and Coronation Hospitals, 2001-2004

*Attrition data were truncated for 2001 (starting 3/16) and 2004 (ending 7/31). Rates for these years have been annualised.

†Hiring data were truncated for 2004 (ending 7/31). Rate for 2004 has been annualised.

III. Reasons for Termination among Nurses

To understand why attrition among nurses is relatively high (particularly among younger staff), we conducted exit interviews of all terminating staff over a 12-month period and held more than 34 focus group discussions involving approximately 100 current staff on reasons for staff termination.

Rankings of reasons for termination by both focus group participants and exit interview participants are shown in Table 2. Among focus group participants, the reasons offered for nurse resignations were largely related to compensation and administrative/managerial issues. HIV/AIDS played a role in aggravating working conditions, however. Three quarters of nurses included an HIV-related reason among the top ten reasons for nurse resignations, and nearly half placed an HIV-related reason among their top five.    

Focus group discussions inevitably involved only nurses who have not (yet) terminated service. To explore reasons for leaving among those who had decided to resign, we conducted exit interviews with approximately 90% (n=72) of nurses who handed in their resignations between March and August 2004. Those leaving service were primarily African (62%) and female (90%). A majority (63%) stated that they were the primary wage-earner in the family. More than three quarters of participants reported that they already had a new job, suggesting substantial “pull” from other employers, in addition to work dissatisfaction internally (“push”). Of those with a new job, 48% planned to work in the private sector, and 37% planned to work elsewhere in the public sector. Exit interview participants were also asked in a separate question about the influence of HIV on their decision to leave, and 30% stated that HIV had at least somewhat affected their decision to leave.

Table 2. Reasons for nurse resignations: exit interviews (EI) and focus groups (FG)

IV. HIV Prevalence in the Workforce

We conducted a voluntary, anonymous, unlinked sero-prevalence survey of all healthcare workers at our study sites. To protect the anonymity of the survey, participants were not given their test results. Instead, they were referred to a VCT and ARV treatment clinics on and off site. Significant groundwork aimed at obtaining support for the survey from workers, unions, hospital management and staff was conducted prior to the testing through meetings, focus groups, social events, and written informational posters and handouts. Participants chose whether to give a blood or an oral fluid sample, and along with either sample, we recorded the age, sex, race, and job level of each participant and ran CD4 counts on all blood samples. Participants were given a T-shirt as a gift for their participation and potential discomfort during the survey.

Results are summarized in Tables 3 and 4.

Table 3. HIV prevalence by job category and age

*Participation by medical doctors was not sufficient to generate valid results, and medical doctors are therefore not included in the job category results. Medical doctors are included in the results by gender and age, however.

Table 4. CD4 counts of HIV-positive survey participants

National Treatment Guidelines in South Africa call for starting ARV therapy at a CD4 levels below 200, and there is a clinical relationship between CD4 levels below 350 and a heightened risk for opportunistic infections. If we apply the HIV prevalence estimates shown above to the combined workforces of the two hospitals and assume that AIDS-related mortality averages 10 percent of HIV prevalence, we would expect the number of HIV-positive staff and annual AIDS-related mortality shown in Table 5. The fact that the observed number of deaths and medical boardings in 2004 is substantially smaller than the predicted number suggests that a majority of HIV-positive staff resign before becoming ill or are on antiretroviral therapy.

Table 5. Number of HIV-positive nurses and expected and observed mortality

V. Costs of HIV/AIDS in the Nursing Workforce

Finally, we estimated the impact on the two hospitals of the HIV prevalence levels reported above, again focusing on the nursing workforce. Estimates of unit costs (e.g. the cost of recruiting a replacement nurse) and benefits policies were obtained from hospital managers. Absenteeism associated with AIDS-related illness was estimated from GSSC absenteeism records. To estimate productivity losses caused by HIV/AIDS and the demands on supervisory time resulting, we also conducted brief, structured interviews with the immediate supervisors of employees who died in service in the past 12 months.

We divided the analysis between indirect costs (productivity losses) and direct costs (expenses). Indirect costs included were absenteeism, reduced productivity when at work, vacancies, and the inexperience of replacement employees. Direct costs included were death and disability benefits, medical care, recruitment, and training. After estimating the cost per employee lost to HIV/AIDS, the results of the HIV prevalence survey were used to estimate total costs for each study site and for the public health sector as a whole.   

Results are summarized in Figure 1 and Table 6.  Direct costs were negligible, as very little is spent on recruiting or training and the cost of benefits and medical aid are borne by the government or province, not the hospital.  On average, the loss of a professional or staff nurse to HIV/AIDS “costs” the hospitals the equivalent of 0.8 years of nursing time. As a result, about 1% of total nursing time available to the hospitals is lost over the course of a year. For Helen Joseph and Coronation, this is the equivalent of almost 7 full time nurses.

Figure 1: Professional nurse days lost per AIDS-related termination

Table 6. Total loss of nurse working time due to AIDS-related terminations in 2005

· We estimate that when Helen Joseph or Coronation Hospitals loses a nurse to HIV/AIDS, the equivalent loss is between a half and a full year of working time or about 1% of total nursing time.

· The prevalence survey found 11.5% of health workers are HIV+. At least 18.9% of the HIV+ health workers should be on ARV treatment already and close to half the HIV+ employees and are risk of OI’s such as TB.

· Discontent over poor salary, excessive work burden, career ceilings and managerial unresponsiveness are cited as the principal reasons for resignations.

· Attrition rates over time and low hiring levels have led to heightened staff shortages and higher workloads that may impact on the quality of health service delivery.

· Given very high existing vacancy levels, it is important to keep current staff in place.

· The health and welfare of health care workers must be maintained to be able to cope with the growing impact of HIV on public services.

VII. Recommendations

· Attrition rates over time and low hiring levels have already led to heightened staff shortages. Given the existing high vacancy levels at the two research hospitals, it is important to implement strategies that will:

1. Retain current staff

2. Attract health workers back into the public health system

3. Increase enrollment into nursing colleges through out the country

· Based on this research, strategies should address:

1. Uncompetitive salaries

2. Increased work burden related to the increased amount of HIV/AIDS patients and critical staff shortages

3. Perceptions of managerial unresponsiveness and breakdowns in communications between workers, supervisors and managers

4. Need for part-time positions, flexible work schedules and flexible work-station allocations

· Although the overall HIV prevalence among health care workers is lower than in the population at large, it remains a problem. Eleven and a half percent of health workers are HIV+ and over half of whom need ARV’s and/or are at risk of TB. Hospitals should aim for 100% VCT uptake for staff in order to reduce the risk of opportunistic infections, specifically TB, among HIV positive staff. The health and welfare of health care workers must be maintained to be able to cope with the growing impact of HIV on public services.

· HIV-positive staff should be strongly encouraged to register at an onsite or offsite HIV clinic and start ART when eligible.

· Health care staff struggle with severe emotional strain related to the HIV epidemic. In-house grief and psychological counseling is recommended to offset staff burnout and low morale.

VIII. Opportunities for Further for Research

· We were pleased to have had the opportunity to carry out focused research at a well defined health care setting that welcomed research. These institutions offer multiple new opportunities to pursue areas of critical health-economic research that would shed further light on the impact of HIV on South African society.

· We believe staff shortages are reaching a critical level. This is a result, in part, of an increased work burden that drives personnel away from public health facilities potentially resulting in decreased quality of patient care. An assessment of the impact staffing shortages has on quality and cost of health care provision, measured in terms of patient days of hospitalization and patient outcomes, is needed to strengthen the commitment by policy makers to staff welfare and better health care provision.

· We recommend further research into the HIV prevalence and CD4 distribution of key public sectors in an attempt increase HIV/AIDS awareness, reduce HIV/AIDS stigma and promote knowledge of HIV status and an early commitment to ARV therapy.

Please do not hesitate to contact Dr. Daniela Connelly at dsevillag@yahoo.com or

Prof. Sydney Rosen at sbrosen@bu.edu for further information on this research study.

A Public Sector Workforce

HIV Prevalence Survey 

A Researcher’s Toolkit

Acknowledgments and Associations

We would like to thank all health care personnel at Helen Joseph and Coronation Hospitals for their willingness to participate in our survey. Without their participation, understanding and trust, this project would not have been possible.

We would like to thank hospital management, including human resources and area managers, and workers unions for their support for this study and facilitation during our research project.

We would also like to thank Dr. Leticia Rispel and Ms. Marion Borcherds (Employees Wellness Program EWA) of the Gauteng Department of Health and the National Departments of Health and Social Development for their interest and support of our research.

The research team is comprised of Prof. Y. Veriava, Dr. D Connelly, Sue Roberts RN, Annie Jordan RN, Josephine Tsotetsi RN, Dr. M Bachman and Prof. S. Rosen. Continuous technical support for this research was provided by the Health Economics Research Office of the Wits Health Consortium and the Center for International Health and Development at Boston University’s School of Public Health. Additional technical assistance was provided by JEAPP. Funding for this research was provided by USAID.

Introduction

It is general knowledge that the HIV/AIDS epidemic is adversely affecting the South-African population. Public servants are responsible for providing services to all sectors of society. In the same way that educators are critical to the schooling of South Africa’s youth, health workers are critical to the implementation of a successful public health response to the HIV/AIDS epidemic in South Africa, and yet they may be as susceptible to this epidemic as the general population.

In the wake of health care staff shortages in the public sector and increased levels of morbidity and mortality within the general population due to AIDS. Concern over the readiness of the public health sector to deal with the increasing patient burden has prompted stakeholders to want to ascertain the impact of HIV/AIDS on public healthcare personnel in order to formulate a proper response and anticipate the future demand for this segment of the workforce. For this reason, our research team carried out an HIV sero-prevalence study among voluntary participants at Helen Joseph and Coronation Hospitals in Johannesburg, South Africa. Additionally, blood samples found to be HIV positive were further tested to measure the CD4 cell count. Knowledge of the CD4 cell distribution provides an indication of what percentage of workers are at increased risk of opportunistic infections, such as tuberculosis, and what percentage of workers have AIDS and would likely need to receive antiretroviral therapy.

This research toolkit offers a step by step description of the procedures used and lessons learned in our research. We believe the methodology described can be applied to other health care facilities and many other public sector workforce settings.

Target audience

Users of this toolkit include:

Stakeholders in public health and human resources

Governing bodies, policy makers and NGO’s

Unions and “HIV at the Workplace” committees such as the EWA

Human resources managers, administrators and supervisors

Institutional CEO’s and CFO’s

Multidisciplinary research teams in public and academic settings

Understanding the Research Question and its Implications:

What is the HIV Prevalence rate?

This toolkit will guide institutions to determine what proportion of workers are HIV positive or negative. The reliability of the survey is dependent on a high percentage of workers voluntarily participating.

Knowledge of the overall workforce HIV prevalence allows for:

preparation for future human resource requirements in the context of AIDS related attrition

enhancement of workplace safety strategies i.e. reduce exposure to opportunistic infections by HIV positive employees

planning for the clinical but also the psychological welfare/health of all workers

planning for (on site) provision of VCT and/or ART

planning for the impact of morbidity and mortality of HIV workers in terms of reduced work productivity.

What is the distribution of CD4 cell counts in a population?

This toolkit explains how to obtain data that will show, based on CD4 counts, the degree of progression of disease in HIV positive participants. Understanding that workers with CD4 cell counts under 350 are more likely to suffer from HIV related infections such as tuberculosis (TB), and that according to South African ARV therapy guidelines, people with CD4 cell counts of below 200 are additionally at risk of life threatening opportunistic infections (OIs) and eligible for antiretroviral therapy.

Knowing The Proportion Of HIV Positive Workers With A CD4 Count Below 350 Allows For:

· Planning towards making the workplace environment safer for “at risk” workers

· Policy making regarding flexible work place assignments

· Policy making regarding workplace risk responsibility (who carries the liability if an HIV positive worker contracts TB)

· Medical and psychological treatment needs-assessment

· Planning towards increased worker attrition due to morbidity and mortality

· Planning to increase the pool of student nurses or trainees

· Planning for interventions geared toward increased HIV status awareness through access to VCT services

Objectives and Outcomes

How to Use this Research Toolkit

Once you have identified the need to carry out such a survey and have found a motivated and committed research team to assist with the survey, the toolkit will lead you though each step of the survey methodology (including data collection and analysis of your results).

A detailed methodology segment, divided into the different research stages, has been provided. Each segment is followed by a narrative (boxed) of our own experience during this project. This case study format will help you identify how a research protocol translates into actions.

We have also added a chapter on how to foresee and overcome common pitfalls.

Methodology

Protocol and ethics

1) Identify the research team members:

· A committed core research team is key to a successful project.

· Additional team members and technical assistance can be brought in according to the project needs. Examples of additional personnel include data entry clerks and/or statistical analysts.

· A lead researcher within the research team should be identified, as the person responsible for coordinating activities, for project administration, and maintaining communication between team members and stakeholders.

· At least some team members should be well known and trusted within the workforce to be surveyed. It takes a far greater effort to establish rapport and trust around a sensitive issue such as HIV status and confidentiality at the workplace with out a concrete link between the researchers and the workforce.

· Identify a source of additional help for undertaking the survey (i.e. HIV counselors or hospital/institutional volunteers or students)

2) Write a research protocol:

A research protocol is a detailed description of how you plan to carry out a study.

Submit the protocol to an academic ethics committee.

3) Create a budget based on the size of the population to be surveyed:

Key ingredients in the budget include time availability of research team, availability of infrastructure, unit costs for laboratory services and complimentary gift for participants.

Secure funding. Further costs to be included are personnel costs including phlebotomists, dissemination efforts, comprehensive laboratory fees, secure transport of samples, telecommunications, and data processing.

4) Complimentary gift :

· Identify what is an appropriate complimentary gift for participants, taking into account local culture and ethics codes (a focus group discussion with potential participants can assist in determining an appropriate gift)

· The gift should not select for any specific type of individual (i.e. hand out makeup could alienate males from participating in the study, handing out meal tickets could select for less wealthy participants with a different HIV risk etc).

· Once an appropriate gift is decided upon, source the provider and ensure the price, quantity and timeliness can be assured.

5) Identify who is directly responsible for allowing the survey to take place at your chosen workplace and ask for approval from that person/s, independently of ethics committee approval.

Helen Joseph and Coronation Hospitals Case Study

Protocol and ethics:

The two study hospitals have a common senior administrative body. The CEO of the hospitals and chief supervisors respectively provided approval for the study. Although not expressly necessary but in an effort to maximize participation in the survey, worker’s unions and staff supervisors (matrons) also provided approval for the study.

The study was approved by the Human Ethics Committee of the University of the Witwatersrand.

Study Sites and Population

· Define your study site/sites

· Define which workers are to be included in the survey

· Define which workers are to be excluded from the survey

Helen Joseph and Coronation Hospitals Case Study:

Study Sites and Population:

Our study was conducted at two discrete but related public hospitals within the Gauteng Province Department of Health. One of the hospitals provides outpatient clinical care, secondary care, and tertiary care in surgery, orthopaedics, psychiatry and internal medicine. The other hospital provides outpatient clinical care and secondary care in paediatrics, gynaecology, and obstetrics to the same patient base.

The study population included 2032 professional and non professional-level employees. For purposes of the study, “professional staff” included medical doctors, nurses, assistant nurses, nursing students, and allied health workers (pharmacists, psychologists, occupational therapists, social workers and other professionals at the same level). “Non-professional” staff included maintenance, cleaning and kitchen staff known as general assistants.

Personnel excluded from our study were hospital volunteers and outsourced security personnel as their attendance could not be verified on internal human resources rosters. HIV counsellors were excluded, given that some were recruited from the HIV clinic patient pool and their results could have given rise to an overestimate the overall HIV prevalence and progression. Similarly, employees who are absent from work on all test days were excluded from the analysis. We estimated a small underestimate of the true prevalence given that some of these absences could be related to HIV morbidity.

Note: we chose not to “sample” workers for the survey because of potential sampling biases. Depending on the size of a sample, this method delivers less reliable data which in turn has to be interpreted with statistical aids such as confidence intervals and p-values. Also, given the relatively small population size, those workers asked to be part of the survey sample could consider themselves to have been “singled out” because of their HIV status.

Dissemination and “buy-in” from all stakeholders for the survey

· Identify as many stakeholders within the institution as possible, including all levels of management, National and/or Provincial Department of Health, political appointees, workers union’s shop stewards, and other respected individuals within the workforce.

· Set up meetings with each stakeholder group to propose a cross-sectional, voluntary, anonymous, unlinked survey prevalence survey. You must present the potential gains and risks of such a survey; answer questions, receive input and establish rapport.

· Create informational materials to be freely, visibly and frequently distributed to all workers to assure awareness and participation. (See Annex 1 for dissemination material)

· Set-up mechanisms to answer queries, receive criticism and feedback from workers directly. These processes although labour intensive, build up the needed trust in the researchers to carry out a successful survey.

Helen Joseph and Coronation Hospitals Case Study

Dissemination and “buy-in”:

Building on previous experience with private sector workforce surveys in South Africa, extensive efforts were made to ensure that the prospective participants 1) understood the purpose, benefits, risks, and voluntary nature of the survey; 2) were comfortable with the steps taken to ensure anonymity and confidentiality; and 3) were willing to participate. To this end, members of the study team created informational posters placed throughout the study sites. In an attempt to prompt discussion and answer queries, an informational letter in a “question and answer” format was added to the pay-slips of all employees on two consecutive occasions leading up to the survey. A series of meetings with hospital stakeholders (management, HR, ward matrons and worker’s unions) were carried out to answer questions and receive input for the survey. In an attempt to reach workers directly, informational sessions were held on site at departments on weekends and during nightshifts. Lastly, an informal informational breakfast enabled workers to further interact directly with researchers and ask questions related to the survey. One of the key messages communicated to participants was that they would not able to obtain their HIV results from this survey, since it was entirely unlinked and anonymous, but they were informed of free in house and off premises voluntary counselling and testing (VCT) sites if they wished to know their HIV status.

No monetary compensation was provided to participants, but each participant received a t-shirt as a token of appreciation for his/her time and possible discomfort. In an attempt to reduce the likelihood of excluded hospital staff from giving samples in exchange for t-shirts, we gave t-shirts to all personnel that presented to the testing sites regardless of their exclusion or inclusion criteria,

Logistics

· Get assurance from your laboratory service provider on how many samples can be processed per day.

· Establish what quality control methods the lab uses to ensure accuracy of the results.

· Establish at what temperature specimens should be optimally stored at until they reach the lab.

· Ensure you will have complete access to employee attendance records for the days of the survey. If possible from both human resources office or supervisor and each department or unit attendance records for double verification.

· Schedule your prevalence testing days with input from human resources managers, unit managers and workers in general in order to minimize lack of participation due to previously scheduled activities such as in-house training sessions or extended meetings, holiday seasons, heavy workload days or night shifts.

· Coordinate with managers the availability of an in-house survey area large enough for at least 100 participants at a time.

· Coordinate staffing needs for the testing days. Remember, the research team may need to ask for those days off from their other posts within the institution.

· Coordinate pick-up and drop-off (transport) of samples to the lab and include a backup plan for sample delivery.

· Coordinate availability of all materials needed for the survey:

· A sample “Check List” can be found below

Materials Check List

units

Printed materials:

Printed Demographic survey sheets

Bar-codes (x 3 with the same code for each participant)

Leaflet on free, confidential, accessible VCT and ARV therapy centers Informational handouts,

plastic or laboratory specimen bags to insert demographic survey sheets

staplers, pens, high-lighters, rubber bands

[2] Testing materials for oral fluid samples:

Latex and hypoallergenic gloves

Oral testing devices including swab and containers

receptacle for waste papers etc

Receptacle for processed oral samples

Timers/stop watches

Testing materials for blood samples:

Phlebotomy chairs/armrests

Latex and hypoallergenic gloves

Size 4 and 6 needles

Sterilizing swabs

Cotton balls

Tourniquets

Needle barrels

Small plasters

Vacutainers (3cc yellow and purple top, with and w/o anticoagulant)

Receptacles for waste and receptacles for blood samples

General:

Computer and internet access

Receiving table and chairs

t-shirts (multiple sizes)

t-shirt storage and dispensing table

Lockers/ secure storage for staff and supplies

Cooling facilities (refrigerator or cooler box with maximum/minimum thermometer and/or temperature control)

Transport (for up to 4 sample deliveries/survey day)

Adequate Seating for participants and research team/volunteers

The Survey

· The rooms/areas where the survey should take place should be easily accessible to all potential participants.

· The survey rooms should be clearly marked and easily found within the institution.

· Area should be clean, have good lighting and be big enough to cope with large numbers of participants.

· Ample seating should be available in case participants must wait to have their samples taken.

· As each participant approaches the receiving table they should be asked if they are on the pay-roll of that institution. This is to expose persons excluded in the research protocol from participating in the survey. If they are volunteers or otherwise involved in the functioning of the institution, they are politely ask not to give a sample and offered a t-shirt the last day of the survey.

· Eligible participants and a research team member jointly fill in the demographic data form.

· Participants are then asked to choose either type of testing method and according to their choice, given either an oral testing device or two blood vacutainers.

· Identical bar code stickers are attached to the demographic survey sheet and either the oral testing unit or each of the two vacutainers. A plastic specimen bag is supplied to each participant to later be used to gather the sample and demographic survey sheet in a common bin for transport to the laboratory.

· The participant is then directed to either a phlebotomist or oral sample collector (team member that describes the oral sample collection method to groups of up to 6 participants at each time).

· After a participant has given their sample they are then directed towards two openly visible collection bins for blood or oral samples, where the participant themselves is asked to deposit their sample. We believe this strengthens the openness of the process and participants feel their samples are not being tracked.

· Once blood samples have reached room temperature, they are placed at 10*C and transported every two to three hours to the lab for processing.

· Each participant is then directed to the t-shirt distribution table and leaves the testing site with a t-shit and a leaflet on where to find free VCT and ARV treatment clinics.

· Researchers must be alert to recognizing returning participants to prevent participants from donating more than one sample.

Helen Joseph and Coronation Hospitals Case Study:

Data collection

The survey was carried out at the study sites to minimize inconvenience to participants. Samples were collected at each site from 9am to 2pm on four consecutive days and also from 7pm to 9pm on two consecutive nights, so enabling both night shifts staff an opportunity to participate into our survey. Eligible personnel who volunteered for the survey were asked to provide either one oral fluid sample or one blood sample according to their preference.

Samples were collected using the OraSure collection device  ADDIN REFMGR.CITE <Refman><Cite><Author>Orasure Technologies</Author><Year>2005</Year><RecNum>57</RecNum><MDL Ref_Type="Patent"><Ref_Type>Patent</Ref_Type><Ref_ID>57</Ref_ID><Authors_Primary>Orasure Technologies,Inc.</Authors_Primary><Date_Primary>2005</Date_Primary><Reprint>Not in File</Reprint><ZZ_WorkformID>22</ZZ_WorkformID></MDL></Cite></Refman> or through phlebotomy. In the cases when blood was drawn, two vacutainers of 2cc each were collected per volunteer, one for ELISA testing and the other for CD4 testing. Phlebotomists, not professionally associated with the two study sites, drew the blood samples. Each participant was also asked to provide basic demographic information on a brief questionnaire. Information was limited to four demographic variables: gender (male or female); age range (18-24, 25-34, 35-44, etc.); professional level (e.g. medical officer, nurse, allied staff or general assistant); and race (black, colored, white or Asian). These racial categories are the standard census categories in South Africa. No individual linking identifiers such as employee or ID number, name, or address were collected. Samples and questionnaires were deposited in a large container at the entrance to the testing space so that participants could see that the process was genuinely anonymous. Research study staff were available to assist with sample collection and answer questions as needed. (Research staff was also alert to intercept repeat- or excluded participants.

Oral fluid specimens were stored at room temperature, and blood samples were stored between 10 to 15°C; both types of specimens were transported to the NHLS laboratory in Johannesburg at approximately 2 hour intervals, with exception of samples collected at nights, which were handed over to the laboratories with a 9 hour interval. The testing was done at the Contract Lab Services of the National Health Laboratory Services in Johannesburg, a clinical laboratory not affiliated with the hospitals or the researchers. Contract Lab Services of The National Health Laboratory Services in Johannesburg has significant experience in handling large prevalence survey sample volumes and processing these efficiently. Both oral fluid and blood samples were tested using the ELISA HIV-1 Antibody Test. Blood samples were then processed further to determine CD4 cell counts.

To determine the participation rate for the survey at each site, employee attendance records were collected for each day/night of the survey to identify employees who were not at work during the three day period. Attendance records were cross checked. Employees who were absent for any reason (scheduled vacations, maternity leave, study leave, sick leave, etc) on all test days or nights were excluded from the analysis.

Data Analysis

Sample containers and questionnaires are matched using duplicate bar codes at the lab where HIV test results and questionnaire responses are entered into an Excel database.

This Excel spread sheet is then further analyzed by the research team and qualified statistical analysts, if necessary:

Univariate analysis was used to estimate the prevalence of HIV infection in the study population as a whole.

Results of the survey are then stratified into subgroups by sex, age range, race and job level.

CD4 cell counts of HIV positive blood samples are subjected to multivariate analysis to determine demographic predictors of risk.

Both HIV prevalence and CD4 cell count data can be analyzed using SAS software.

Helen Joseph and Coronation Hospitals Case Study

Data analysis:

HIV test results and questionnaire responses were entered into an Excel database at the laboratory and sent to the lead researcher for analysis. Sample containers and questionnaires were matched using duplicate bar codes. Univariate analysis was used to estimate the prevalence of HIV infection in the study population as a whole. Results of the survey were then stratified into subgroups by sex, age range, race and job level. Chi-square tests were used to determine differences between subgroups in bivariate and multivariate analysis. CD4 cell counts of HIV positive blood samples were subjected to multivariate analysis to determine demographic predictors of risk. Both HIV prevalence and CD4 cell count data were analyzed using SAS software.

Dissemination

Plan to disseminate your results to all the stakeholder groups that participated in the research process.

This can be managed by scheduling brief presentations during regularly planned meetings. “Piggybacking” on other meetings lessens the research team’s effort to summon workshop participants and lowers costs.

It is important to disseminate the research results as quickly as possible to institutional CEOs and other officials, particularly in Government, who are likely to be approached by local media.

Have a brief media (press) release available. It is important the results be disseminated accurately and in the correct context. This is particularly important to preserve and honour the surveyed workforce, since erroneously interpreted or printed information could offend and affect the workforce as a whole.

Dissemination strategies should include a report summary and thank you notice posted throughout the surveyed institutions or similarly annexed to pay slips.

Dissemination should be focus towards policy makers, HIV at the workplace commissions and other workforce representatives.

Dissemination strategies should include academic and/or sector specific publications, workshops and conferences.

Lessons Learned

We have listed below what we found to be critical steps in ensuring high participation rates in our prevalence survey. We have also listed areas that need special attention while planning and executing the prevalence survey.

Important Steps:

Extensive and inclusive involvement of stakeholder groups is critical; without their “buy-in” to this kind of survey, workers will feel at odds with the researchers, supervisors and/or workers unions etc and fail to participate.

During the survey, we found that there was much interest by participants in receiving a complementary gift, which in our case was a t-shirt inscribed with a small “Workers Wellness” logo. As soon as participants received their t-shirts, they put them over their uniform and went back to their work stations. This served to remind co-workers that the survey was taking place, and to motivate co-workers to participate in the survey seeing that others had overcome initial reservations.

Weeks after the survey and after the survey results were disseminated, participants still wore their survey t-shirts to work. Feedback from health workers revealed that the survey had paved the way for discussions among co-workers about topics related to HIV at the workplace. The onsite HIV clinic at the research hospitals experienced a significant increase in hospital workers presenting for VCT and ARV therapy.

“Potential pitfalls”:

Managerial/Administrative Hierarchies:

Respect managerial hierarchies within the workforce to be researched. Remember to ask for permission from each level of management to approach workers, disseminate information, carry out the survey and use available infrastructure etc. Although you may have ethics committee approval to carry out such a study, managers need to be informed at all times of what occurs with-in their jurisdictions. Opposition from managers to your research can severely affect participant turn-out to the survey, affecting the validity of the results.

Sample processing:

To avoid suspicion that samples might be some how marked to later link participants and HIV results, survey participants were handed their blood or oral fluid samples so that they could place their specimens with the demographic data sheet and barcodes in the collection bin.

Temperature control:

It is imperative to discuss the measures that will be needed to maintain the biological samples at the optimal temperature. Oral fluid samples are less complex, since they can be stored at room temperature and be processed even days after the survey. CD4 cells decay quickly and the longer the time between drawing a blood sample and its processing, the lower the CD4 cell counts will be, clearly affecting the reliability of the results.

Chain of custody of samples:

It is imperative to preserve the integrity of the samples until they are processed at the lab. Designate a “samples custodian”; responsible at all times for all samples collected. This will minimize accidental losses of samples, temperature mismanagement and possible tampering with the survey specimens.

CD4 cell count analysis on HIV negative samples:

By and large, medical literature has defined normal CD4 cell ranges, the same that are used on a daily basis through out the world as reference points. But there have been reports of particular cultural and/or racial groups possessing marginally different “normal ranges”. In our study of South African health care workers, we felt that their environmental exposure to infectious diseases and racial composition warranted extracting the CD4 cell counts of HIV negative participants. This way we could determine a population specific HIV negative CD4 cell range useful for internal comparison to CD4 cell counts among HIV + samples. This is an optional exercise, but should be considered if funding allows.

Annex 1: Informational letter and or poster

Dear co-worker, we would like to invite you

to participate in our research HIV prevalence survey!

It is very important to us everyone involved should be informed about the survey and what we aim to accomplish through it.

Questions? Here are some answers…

Q: What is an HIV prevalence survey?

It is a scientific way to find out how many people in a group are HIV positive or negative. Each participant can voluntarily and anonymously provide a saliva sample or a blood sample, which is then tested for HIV antibodies. We do not assess the HIV status of any individual.

Q: Are the results confidential?

No names, ID numbers or identifiers will be collected, nor information on the department or ward worked at, the day or time the person volunteered, the clothes the person was wearing etc). This is to ensure that finding out an individual’s HIV status not possible.

For those who wish to know their HIV status we have several independent, confidential referral options for VCT and treatment. Individual survey results are confidential!

Q: Who can participate?

So that no one feels left out or specially selected to participate, we have asked ALL health care workers at both hospitals to volunteer for the survey. This will include all professional staff (Medical doctors, nursing staff, and allied staff) and non-professional staff (general assistants).

Q: Why is everyone’s participation is so important?

The more people take part in this survey, the better and more reliable the results.

Q: How do I benefit from participating in the survey?

We believe that events like these will help inform about HIV-related issues and decrease HIV-associated stigma at the work place. As part of the prevalence survey we offer information on independent VCT and Antiretroviral Treatment sites for those who wish to embrace the opportunity to find out their HIV status.

Q: What will happen on the survey days?

A survey booth will be set up over four to five days at each hospital.

At any time during your working day you are encouraged to voluntarily come and give us a blood or saliva sample and fill out a brief anonymous questionnaire. The processes will take approximately 20 minutes, for which management has given their approval.

As a token of our appreciation,

all survey participants will be given a t-shirt!

Q: What if I refuse to participate?

Since the survey is voluntary, you are free to refuse to participate with no consequence whatsoever.

The research team hopes to count on your assistance in making this

prevalence survey a success.

Together, we can help protect fellow workers from stigma

And make our workplace a healthier and friendlier place.

[1] The Toolkit will be reproduced in a spiral-bound booklet format and distributed at dissemination opportunities; we will present the funders with this product as soon as it comes into print.

[2] We estimated that approximately 2/3 of participants would opt for the oral testing option, but we in fact had approximately ½ of participants opt for this option.