TRANSACTIONS OFTHE ROYAL SOCIETY OFTROPICAL MEDICINE AND HYGIENE (1998) 92,500-502

Increasing frequency of tuberculosis among staff in a South African district hospital: impact of the HIV epidemic on the supply side of health care

David WilkinsonlJ and Charles F. Gilksz ICentre for Epidemiological Research in Southern Africa, Medical Re- search Council, Mtubatuba, South Africa; 2Liverpool School of Tropical Medicine, Liverpool, UK

Abstract To describe the changing frequency of tuberculosis among staff in a South African hospital, and to com- pare incidence in health workers with that in ancillary staff, the number and type of cases of tuberculosis among staff diagnosed between 1991 and 1996 were ascertained. The incidence rate of tuberculosis among health workers and ancillary staff was compared with the age-specific rate in the community (20-59 years old). In 1991-1992, 2 cases of tuberculosis were diagnosed among hospital staff; but in 1993-1996 there were 20 cases diagnosed (annualized incidence rates 138/100000 and 690/100000; P<O.OOOl). Of 14 cases tested (64%), 12 (86%) were infected with human immunodeficiency virus (HIV). Most cases (82%) successfully completed treatment, but 4 died (18%). The incidence of tuber- culosis amongst health workers (5581100000 person-years of observation [PYO]) and ancillary staff (4451100 000 PYO) was not significantly different (P=O.7), but it was lower than the incidence rate among 20-59 years old people in the community (1543/100000).Tuberculosis has increased amongst hospital staff, secondary to the impact of HIV. The HIV epidemic is having a substantial impact on the health of hospital staff and interventions to counter this are urgently needed.

Keywords: tuberculosis, Mycobacterium tuberculosis, acquired immune deficiency syndrome, human immunodeficien- cy virus, incidence, hospital staff, South Africa

Introduction While it is widely appreciated that the much in-

creased demand for patient care created by the human immunodeficiency virus (HIV) epidemic is a serious threat to African health systems, the importance of the epidemic as a significant cause of illness among hospital staff, and hence the ability of the health system to re- sDond to the increased demand for care, has received l&s attention (ANONYMOUS, 1995; GILK~ et al., 1996). Data on the imnact of the HIV enidemic on the health of hospital staff&in Africa are limiied.

HIV infection among hospital workers is not associat- ed with occupational exposure to HIV, and seropreva- lence is typically equivalent to that in the community from which the staff are drawn (MANN et al., 1986). In Zambia, mortality among nurses increased more than five-fold from 0.5 per year in 1980 to 2.7 per year in 1991 (BuvB et al., 1994), and this was attributed in large part to HIV infection. Absenteeism due to illness among staff, and illness among their friends and rela- tives for whom they have responsibility, contributes to the impact of the HIV epidemic on the provision of health care by reducing the number of staff available to work on any given day (Bm et al., 1994).

Studying tuberculosis among staff is one way to study the impact of HIV on health workers because it is strongly HIV-associated. Tuberculosis may be acquired at work (HARRIES et al., 1997), making tuberculosis and HIV prevention and control for health workers an even greater priority.

Methods Setting

We studied changes in the number of staff diagnosed with tuberculosis in a single district hospital in South Africa between 1991 and 1996. HIV infection became widespread in South Africa in the late 1980s and preva- lence increased rapidly in the 1990s (SOUTH AFRICA, 1996a). The hosoital studied is situated in KwaZuh- Natal,’ the provi&e with the highest HIV prevalence (SOUTH AFRICA, 1996a). It has approximately 400 beds; patients with suspected tuberculosis are managed initially on the general medical wards before being transferred to the tuberculosis ward which has 80 beds.

Address for correspondence: Dr D. Wilkinson, Centre for Epi- demiological Research in Southern Africa (CERSA), Medical Research Council, P.0. Box 187, Mtubatuba 3935, South Af- rica; fax +27 35 550 1436, e-mail wilkinsd@mrc.ac.za

In 1991 the number of cases of tuberculosis diagnosed was 321 but this increased to 1250 in 1996, It was esti- mated that 44% of cases were attributable to HIV infec- tion in 1995 (WILKINSON & DAVIES, 1997), and HIV prevalence among adults with tuberculosis was 65% in 1997 (unpublished observations).

Tuberculosis in staff Data concerning staff were extracted confidentially

from the anonymized tuberculosis control programme register; permission to do this was obtained from the lo- cal ethics board. Case ascertainment is known to be complete because tuberculosis treatment cannot be ob- tained anywhere else in the district and all staff illness episodes are recorded in personnel files. This control programme register contained data on age, sex, type of tuberculosis, and outcome. The number of hospital staff at risk for tuberculosis was determined from hospi- tal payroll and personnel records. The diagnosis of tu- berculosis is made according to standard criteria (CROFTON et al., 1992). All patients were offered vol- untarv HIV testing and counsellinn. Treatment consist- ed of-4 antituber&ldus drugs given in standard doses, in line with national guidelines (SOUTH AFRICA, 1996b), and was given either as an in-patient or by di- rectly observed therapy in the community.

For the purpose of this analysis, staff were grouped into 2 categories: health workers (nurses and doctors), having regular and close contact with patients, and an- cillary staff (all other cadres), not having equally regular and close contact with patients. We recognize that some ancillary staff have substantial contact with patients, but we were unable to categorize staff by degree of contact because patient contact is highly variable within a par- ticular staff grade. For example, one clerk may work at the patient reception desk and be heavily exposed to tu- berculosis, while another may work in an office with lit- tle or no patient contact.

The incidence rate (per 100000 person-years of ob- servation PYOl) was calculated as the number of cases divided by the pkrson-time at risk of developing active disease. We defined the neriod at risk of develoning: ac- tive disease as being the-duration of the study <‘6 years) multiplied by the number of staff employed at the start of the study (staff turnover was less than 5% per year). Community age-specific incidence rates were calculat- ed using data from the tuberculosis control programme database and the census, allowing direct comparison to

TUBERCULOSIS AMONG HOSPITAL STAFF 501

be made between incidence among staff and communi- ty. Identical diagnostic criteria were applied to tubercu- losis suspects from both staff and community.

Results In 1991 and 1992, 2 of the 725 staff at risk were di-

agnosed with tuberculosis (annualized incidence rate 138/100000), and another 20 were so diagnosed be- tween 1993 and 1996 (annualized incidence rate 6901 100000; P<O.OOOl). The proportion of all tuberculosis cases diagnosed each year which occurred among staff members did not change significantly (Table 1).

nificantly different (relative risk 1.25, 95% confidence interval 0.54-2.94; P=O.7), even when adjusted for age, sex and HIV status (data not shown); both were sub- stantially lower than the age-specific community inci- dence for 20-59 years old persons (1543/100000; P~O~OOOl) (WILKINSON & DAVIES, 1997).

Discussion We have shown that the frequency of tuberculosis

among staff in this hospital has increased substantially in recent years, and that this increase is associated with a rapidly expanding HIV epidemic in the community.

Table 1. Proportion of cases of tuberculosis diagnosed among the staff of a South African district hospital, 1991-1996

Year

1991 1992 1993 1994 1995 1996 Total

Total no. Number of cases among staff of cases of Cases

Pulmonary Extrapulmonary All tuberculosis among staff (%)a

1 0 1 312 0.32 (O-l .7) 0 I 1 624 0.16(0-0.8) 3 1 4 703 0.57 (0.2-1.5 2 4 6 827 0.73 (0.3-1.6 4 2 6 839 0.72 (0.3-1.6 3 1 4 1250 0.32(0.1-0.8

13 9 22 4555 0.48 (0.3-0.7

a95% Confidence interval in parentheses.

Table 2. Type of tuberculosis diagnosed among staff of a South African District Hospital

T’e Number

Pulmonary 13 (59%) Smear positive 8 Smear negativea 5

Extrapulmonary 9(41%) Pleural 3 Peritoneal 3 Lymph node 2 Intracranial 1

Total 22

aAl1 confirmed by culture.

Tuberculosis incidence among staff remains lower than that in the community, perhaps reflecting better health, nutrition and access to care among workers. Long-serv- ing hospital staff report that before the advent of the HIV epidemic tuberculosis among staff was very unusu- al. We observed similar tuberculosis incidence rates among health workers and administrative and support staff, suggesting that nosocomial transmission may be relatively infrequent.

Our findings suggest that the HIV epidemic is having a substantial impact on the health of hospital staff in Af- rica, largely simply because they are members of local communities. This may threaten the supply side of health care. While we found no evidence that occupa- tional exposure to patients with tuberculosis is a risk for development of tuberculosis in staff, our numbers were

Table 3. Tuberculosis incidence by occupational category in a South African district hospital, 1991-1996

Occupational No. of No. at Population-time Incidence rate No. of No. of HIV category cases risk at risk (PYO)a (per 100000) extrapulmonary cases infected cases

Health worker 13 388 2328 558 6 (46.2%) g/13(61.5%) Ancillary worker 9 337 2022 445 3 (33.3%) 419 (44%)

~(No. at risk)x(no. of years [6]); PYO=person-years of observation.

Reflecting the age and sex distribution of hospital staff, the mean age of the 22 cases was 29.6 years (SDz5.9) and 14 (64%) were women, whereas for all patients in the tuberculosis programme the mean age was 37 years and 36% were women (BO.05). Thirteen of the 22 cases (59%) were health workers, and the re- mainder were administrative and support staff, most of whom had little direct patient contact. Nine of the 22 cases (41%) had extrapulmonary disease (Table 2). Fourteen patients (64%) were tested for HIV infection and 12 (86%) were found to be infected; in the pro- gramme as a whole, 40% were tested and 59% of these were infected with HIV (BO.05). Most staff cases (18; 82%) completed treatment successfully, but the re- maining 4 (18%) died while receiving treatment; these outcomes were similar to those of all patients in the tu- berculosis programme. All 4 cases who died were HIV infected.

The incidence rates of tuberculosis (Table 3) amongst health workers and ancillary staff were not sig-

relatively small and these observations need to be both extended and supplemented by molecular studies that might define pathways of transmission. HARRIES et al. (1997) have shown that a substantial risk of nosocomial tuberculosis transmission may exist, and we have re- ported the first case of nosocomial transmission of tu- berculosis in Africa to be documented by restriction fragment length polymorphism (WILKINSON et al., 1997). It is possible that extensive person-to-person transmission is occurring within African hospitals, but that it is being masked by high community transmission rates. If multi-drug resistance becomes more prevalent the need to protect staff will be even greater (HARRIES et al., 1997).

Hospital staff in developing countries are a particular- ly precious resource. Training is expensive and lengthy, and they carry out important and ever-increasing du- ties. This resource is under serious threat from the HIV epidemic, and it needs to be protected. What is the most appropriate strategy now? There is little experience else-

502 DAVID WILKINSON AND CHARLES F. GILKS

where to draw on, despite the fact that the problem that we describe is not an isolated one. Staff are unlikely to come for HIV testing unless there is strong positive mo- tivation to do so. It is possible that preventive threapy for tuberculosis (FOSTER et al., 1997; WILKINSON, 1998) could be promoted as part of a cost-effective package of care for staff. The package could also include confidential HIV testing and counselling with continu- ing care and emotional and spiritual support. The issues of employment and pension rights and discrimination at work need to be addressed. Early access to quality health care could be ensured through a staff health clin- ic. HIV prevention activities might also gain some mo- mentum within a more positive caring context.

Initial efforts to address the problem of HIV-related diseases among hospital staff are likely to be met with fear, suspicion and denial. This is understandable, but as staff overcome these initial reactions it will be impor- tant to have a strategy in place. What we cannot afford to do is nothing.

Acknowledgement This work was funded by the South African Medical Re-

search Council and the British Government Overseas Devel- opment Administration.

References Anonymous (1995). Impact of HIV on delivery of health care

in sub-Saharan Ahica: a tale of secrecy and inertia. Lance& 345,1315-1317.

Buvt, A., Foaster, S. D., Mbwili, C., Mungo, E.,Tollenare, N. & Zeko, M. (1994). Mortality among female nurses in the face of the AIDS epidemic: a pilot study in Zambia. AIDS, 8, 396.

Crofton, J., Horne, N. & Miller, F. (1992). Clinical Tuberculosis. St Albans, UI(: TALC~IUATLD.

Foster, S., Godfrey-Fausett, I? & Porter, J. (1997). Modelling the economic benefits of tuberculosis preventive therapy for people with HIV: the example of Zambia. AIDS, 7, 919-925.

1 Announcement 1

Gilks, C. F., Haran, D. &Wilkinson, D. (1996). Coping with the impact of the HIV epidemic-the Hlabisa-Liverpool HIV link. South African MedicalJounzal, 86, 1077-1078.

Harries, A. D., Kamenya, A., Namarika, D., Msolomba, I. W., Salaniponi, F. M., Nyangulu, D. S. & Nunn, l? (1997). De- lays in diagnosis and treatment of smear-positive tuberculo- sis and the incidence of tuberculosis in hospital nurses in Blantyre, Malawi. Transactions of the Royal Society of Tropical Medicine and Hygiene, 91, 15-17.

Harries, A. D., Maher, D. & Nunn, I? (1997). Practical and af- fordable measures for the protection of health care workers from tuberculosis in low-income countries. Bulletin of the World Health Organization, 75, 477-489.

Mann, J., Francis, H., Quinn, T. C., Bila, K., Asila, l? K., Bosenge, N., Nzilambi, N., Jansegers, L., Pist, I’., Ruti, K. & Curran, J. W. (1986). HIV seroprevalence among hospital workers in Kinshasa, Zaire: lack of association with occupa- tional exposure. Journal of the American Medical Association, 256,3099-3102

South Africa [Department of Health] (1996a). Sixth national HIV survey of women attending antenatal clinics of the pub- lic health services in the Republic of South Africa, October 1995. Epidemiological Comments (South Africa), 23, 3-17.

South Africa (1996). The South African Tuberculosis Control Pro- gramme Practical Guidelines. Pretoria: Department of Health.

Wilkinson, D. (1998). Preventive therapy for tuberculosis in HIV infected persons. In: The Cochrane Database of System- atic Reviews: The Cochrane Library, 4th edition, Garner, I’., Gelband, H., Olliaro, I?, Salinas, R. &Wilkinson, D. (edi- tars). Oxford: The Cochrane Collaboration.

Wilkinson, D. & Davies, G. R. (1997). Increasing burden of tuberculosis in rural South Africa-impact of the HIV epi- demic. South African MedicalJournal, 87, 447-450.

Wilkinson, D., Crump, J., Pillay, M. & Sturm, A. W. (1997). Nosocomial transmission of tuberculosis in Africa docu- mented by restriction fragment length polymorphism. Trans- actions of the Royal Society of Tropical Medicine and Hygiene, 91,318.

Received 27 April 1998; revised 9 June 1998; accepted for publication 9 June 1998

ROYAL SOCIETY OF TROPICAL MEDICINE AND HYGIENE Prizes

Undergraduate Project Prize The Royal Society ofTropical Medicine and Hygiene offers an annual prize of A200 for an account of work

carried out of relevance to a tropical or developing country by a non-medical student of any nationality. The work (which may be laboratory-based and not necessarily carried out in a tropical or developing country) will add to the knowledge of human or veterinary health or hygiene in the broadest sense. Particular attention will be directed towards originality and quality in the award of the prize. It is anticipated that the prize will act as a stimulus for the pursuit of excellence in research carried out by undergraduates.

Medical Student Elective Prize The Royal Society ofTropical Medicine and Hygiene offers an annual prize of &200 for an account of work

carried out by a medical student of any nationality during an elective period spent in a tropical or developing country. In awarding this prize emphasis will be laid on the originality of the work and on its contribution to knowledge or understanding of tropical diseases. Particular weight will be given to projects which have been developed and carried out by the students themselves.

Rules 1. Two prizes of A200 may be awarded annually in recognition of outstanding projects which increase knowl- edge of tropical medicine and hygiene in the broadest sense. 2. Candidates shall be nominated by their head of department, supervisor or Dean, with a supporting statement of up to 500 words. 3. The closing date for receipt of project reports is 31 December. The project should have been done or com- pleted in the previous twelve months. 4. A Committee of three shall choose the prize winners. 5. The announcement of the prize winners will be made at the March meeting of the Society. 6. The prizes will be presented by the President of the Society at the Annual General Meeting in June or July.

Please note that the Society cannot provide funds to cover students’ elective travel expenses.

Application forms may be obtained from the Administrator, Royal Society of Tropical Medicine and Hygiene, Manson House, 26 Portland Place, London,WlN 4EY, UK; fax +44 (0)171 436 1389, e-mail mail@rstmh.org