A health worker knowledge, attitudes and practices survey of SARS-CoV-2 infection prevention and control in South Africa

A cross-sectional study design was utilised with both descriptive and analytical components. The knowledge, attitudes and practices (KAP) survey was part of a rapid appraisal of the COVID-19 occupational health and safety response in four provinces in South Africa.

The study was conducted in public sector (government-run) facilities in four of the nine provinces of South Africa viz. Gauteng Province, Limpopo Province, Mpumalanga Province, and North-West Province. The public sector facilities comprised district, regional, tertiary, central, and specialised psychiatric hospitals as well as community health centres.

The study population comprised frontline health workers within the four provinces that are based in casualty (accident and emergency) departments, outpatient departments, and entrance points of health facilities. These are likely to be the initial points of contact between undiagnosed COVID-19 cases and health facility staff thereby placing workers stationed in these areas at risk of COVID-19 transmission. This included both clinical and non-clinical staff. The health facilities from which participants were drawn were conveniently sampled. At each facility, the target was to invite at least one health worker from each of six categories of staff to participate viz. medical doctors, nurses, cleaners, porters, security personnel and clerks. Clinical associates (similar cadre to physician assistants) were also invited to participate but are not employed at all facilities. Health workers were selected at each facility based on their availability at the time of the rapid appraisal. This was done to minimise disruption to normal service delivery.

A structured questionnaire with closed-ended questions in four sections (socio-demographic characteristics, knowledge items, attitude/perception items and practice items) was utilised for data collection which took place over a two-month period commencing mid-April 2020. The knowledge items of the KAP survey were based on the WHO interim guidance of 25 January 2020 [12]. The WHO interim guidance released on 19 March 2020 did not reflect any changes with respect to the knowledge items included in our questionnaire [14]. For the knowledge items, participants were provided with a statement and had to indicate if they agreed, disagreed or were unsure. The attitude and perception items used a five-point Likert response scale (strongly disagree; disagree; unsure; agree; strongly agree). The practice items measured both individual and facility practices with the WHO guidance being used as the basis for most of the items [12]. The responses from which participants could select for the practice items were ‘yes’, ‘no’, ‘not sure’, and ‘not applicable’. The ‘not applicable’ option was provided as not all practice items were applicable to all categories of health workers and ‘not applicable’ responses were not included in the data analysis. The questionnaires were self-administered with a researcher or trained research assistant being in the vicinity to address any queries. Physical distancing, wearing of medical masks and hygiene measures were employed to reduce the risk of transmission of SARS-CoV-2 during the data collection process.

Data entry was done on Microsoft Excel and imported into STATA version 16 (v15 (Statacorp; http://​www.​stata.​com) for analyses. For the knowledge items, ‘unsure’ and non-responses were treated as being incorrect and were included in the denominator when calculating the proportion of correct responses. The number of questions answered correctly were added to give a score out of 10. Participants were classified as having adequate knowledge if they answered six or more questions correctly and good knowledge if they answered 8 or more questions correctly. Logistic regression was utilised to determine factors associated with a good knowledge score. Socio-demographic and employment variables with a p-value less than 0.25 on univariate logistic regression were included in the initial multivariate logistic regression model. The final multivariate model was developed by dropping variables from the initial model using a backwards stepwise approach. For each attitude/perception item, the proportion of participants selecting each Likert-scale option was calculated. Proportions were also calculated for each of the practice items.

A total of 286 health workers from 47 health facilities participated in the survey. The health facilities comprised 4 community health centres, 22 district hospitals, 11 regional hospitals, 7 tertiary hospitals, a central hospital, and 2 specialised psychiatric hospitals. The socio-demographic and employment characteristics of the participants are displayed in Table 1. The majority of participants were aged between 30 and 49 years (62.9%) and the median age was 39 years. There was substantially more female (59.6%) than male (40.4%) participants. A large proportion of participants (46.6%) did not possess a diploma or degree. Except for clinical associates (1.8%), the different occupational categories of frontline staff were well represented. The ‘other’ category included various occupational categories such as allied health professionals, pharmacists, and pharmacy assistants. As would be expected based on the facilities selected, the largest proportion of participants (45.1%) worked at district hospitals.

The proportions of health workers who responded correctly to each of the knowledge items are shown in Table 2. More than 90% of participants provided correct answers on each of the two items related to the mechanism of transmission of SARS-CoV-2. Just over half of participants (52.1%) knew that droplet and contact precautions were required for both confirmed cases of COVID-19 and for suspected cases. The lowest proportion of correct responses (6.6%) was obtained for one of the questions related to the appropriate personal protective equipment (PPE) to be used in the routine care of COVID-19 patients. In total, there were three items where the proportion of correct responses fell below 50%. Medical doctors attained the highest or joint-highest proportion of correct responses on eight of the ten items.

The mean score on the 10 knowledge items was 6.3 (SD = 1.6) (Table 2). Medical doctors had the highest mean score of 7.9 (SD = 1.2) followed by nurses with a mean score of 7.0 (SD = 1.4). The difference between doctors and nurses was statistically significant (p = 0.002). Cleaners, clerks and security personnel all had mean scores below 6. Approximately two-thirds of participants (67.4%) answered six or more questions correctly while less than a quarter of all participants (24.1%) managed to score eight or more. The majority of participants in all but one of the health worker categories answered six or more of the questions correctly. With the exception of medical doctors, the majority of participants in each of the health worker categories failed to achieve a minimum of eight correct responses.

The final multivariate model for factors associated with a knowledge score of ≥8 is shown in Table 3. The initial model consisted of sociodemographic characteristics with p < 0.25 following univariate logistic regression viz. age, education, occupation, years of experience and type of facility where employed. Health workers aged 40 years and older had lower odds of getting 8 or more of the knowledge items correct. Medical doctors and nurses had higher odds of scoring 8 or more as did health workers employed at tertiary hospitals.

Only half of participants (50.7%) felt adequately prepared to deal with patients with COVD-19 at the time of the survey (see Table 4). A majority of participants (86.1%) indicated they would wear the required PPE even if it were uncomfortable. Almost two-thirds of participants (65.1%) strongly disagreed with the statement that they would still treat a COVID-19 patient even if the required PPE was not available with a further 21.7% disagreeing with the statement. A substantial proportion of participants (46.8%) felt they could be infected with COVID-19 at their facility regardless of the precautions they take. One-tenth of participants (10.0%) indicated they would continue to report for duty even if they had symptoms suggestive of COVID-19. More than 5% of participants (6.4%) indicated they would resign to avoid contact with COVID-19 patients.

Participant responses to questions related to their own practices as well as that of their health facility are shown in Table 5. Just over half of participants (51.3%) confirmed they had undergone occupational health and safety training. Slightly more than half of participants indicated they had received infection prevention and control training (55.6%) and training on the correct use of PPE (56.2%). Less than 90% of participants indicated they had access to medical masks (85.7%) and gloves (88.7%) with approximately half of participants (51.0%) indicating they had access to respirators in their department. The vast majority of participants indicated they practiced hand hygiene after touching a patient (97.7%) and after touching a patient’s surroundings (90.3%).

Our survey was conducted as part of a rapid appraisal of COVID-19 occupational health and safety preparedness in four provinces of South Africa. We used convenience sampling to maximise the number of facilities that could be assessed in a short period of time. While we believe that our sample provides a fair reflection of the knowledge, attitudes and practices in health facilities in the four provinces, the non-probability sampling strategy needs to be borne in mind with respect to generalisation of the findings. Our study focussed on health workers that were likely to be the initial point of contact with COVID-19 cases at the various facilities. It is possible that their knowledge, attitudes, and practices differ from health workers working in wards and intensive care units who need to provide longer term care to COVID-19 patients. As we did not have access to patient statistics at facility level, we did not compare knowledge, attitudes and practices between health workers at healthcare facilities with high COVID-19 patient volumes and those at facilities who see few COVID-19 patients. The scientific community’s understanding of COVID-19 has been rapidly evolving since the first cases were discovered. We developed our questions on the accepted understanding of COVID-19 at the time of the survey and analysed the results as such. With respect to an acceptable knowledge score, it is not clear-cut as to what an appropriate cut-off would be particularly with clinical and non-clinical cadres involved in our study. We, therefore, presented results for an adequate score of 60% and a good score of 80%. Health worker reporting of practices may not reflect actual practices and it is likely that the self-reported hand hygiene compliance is an overestimate. Substantial discrepancies between self-reported hand hygiene practices and direct observation have been reported in the literature [23, 24].