The Impact of SARS-CoV-2 Lineages (Variants) on the COVID-19 Epidemic in South Africa

Emerging SARS-CoV-2 variants have been attributed to the occurrence of secondary and tertiary COVID-19 epidemic waves and also threatening vaccine efforts due to their immune invasiveness. Since the importation of SARS-CoV-2 in South Africa, with the first reported COVID-19 case on the 5th of March 2020, South Africa has observed 3 consecutive COVID-19 epidemic waves. The evolution of SARS-CoV-2 has played a significant role in the resurgence of COVID-19 epidemic waves in South Africa and across the globe. South Africa has a unique observation of the evolution of SARS-CoV-2, with distinct SARS-CoV-2 lineages dominating certain epidemic periods. This unique observation allows for an investigation of the detected SARS-CoV-2 lineages' impact on COVID-19 transmissibility and severity through analysis of epidemiological data. In this study, inferential statistical analysis was conducted on South African COVID-19 epidemiological data to investigate the impact of SARS-CoV-2 lineages in the South African COVID-19 epidemiology. The general methodology in this study involved the collation of South African COVID-19 epidemiological data, the regression and normalisation of the epidemiological data, and inferential statistical analysis. This study shows that the evolution of SARS-CoV-2 resulted in an increase in COVID-19 transmissibility and severity in South Africa. The Delta SARS-CoV-2 VOC resulted in increased COVID-19 transmissibility in the South African population by 53.9 to 54.8 % more than the Beta SARS-CoV-2 VOC and the predominantly B.1.1.54, B.1.1.56 C.1 SA SARS-CoV-2 lineage cluster. The Beta SARS-CoV-2 VOC resulted in more severe COVID-19 in South Africa than the Delta SARS-CoV-2 VOC. While, both the Beta and Delta SARS-CoV-2 VOC resulted in more severe COVID-19 than the initial SARS-CoV-2 lineages detected in the South African first epidemic wave period. The Delta, Beta SARS-CoV-2 VOCs, and the predominantly B.1.1.54, B.1.1.56 C.1 SA SARS-CoV-2 lineage cluster were observed to cause similar COVID-19 hospital case fatality and discharge rates in South African hospitals.


Introduction
On the 11 th of March 2020, the World Health Organisation (WHO) declared the Coronavirus Disease 2019 (COVID-19) a global pandemic [1]. The COVID-19 pandemic has resulted in more than 4 809 907 deaths in the reporting period up to the 1 st of October 2021 [2]. Public health measures such as nationwide lockdowns aimed at reducing the transmission of COVID-19 have come at a great cost to the world economy [3]. While, there is global consensus on the health risk posed by COVID-19, ground-breaking vaccine developments, and a great drive towards the vaccination of the world population against COVID-19. There are however challenges that persist in controlling the Global COVID-19 transmission and severity. One challenge is the large disparity in access to vaccines between the developing and developed worlds [4]. Another challenge is the emergent of SARS-CoV-2 lineages and sub-lineages (variants) with increased transmissibility [5]. Lineages and sub-lineages are a series of entities (in this case genetic) forming a single line of direct ancestry and descent [6]. Emerging SARS-CoV-2 variants have been attributed to the occurrence of secondary and tertiary COVID-19 epidemic waves and also threatening vaccine efforts due to their immune invasiveness [7]. SARS-CoV-2 is the virus that causes COVID-19 upon infecting a human host. SARS-CoV-2's early characterisation was reported as a positive pan-Betacoronavirus from bronchoalveolar lavage samples and its whole genome sequence acquired through Illumina and nanopore sequencing. It was identified to have features typical of Coronaviruses belonging to the Betacoronavirus 2B lineage. Betacoronaviruses belong to the subfamily Othrocoronavirinae of the Coronaviridae family, order Nidovirales. A close relationship was found between SARS-CoV-2 with the BatCoV Rat G13 virus (from the bat (Rhinolophus affinis)) at 96 % identity [8]. The genetic sequence of SARS-CoV-2 was shared on the 12 th of January 2020 [9,10]. Coronaviruses are positive-stranded Ribonucleic Acid (RNA) viruses. The presence of spike glycoproteins in their envelope gives them a crown-like appearance under an electron microscope [11]. The genome of SARS-CoV-2 is a single positive-stranded RNA approximately 29 903 bases (nucleotides) pairs in length [9,10,12,13]. There are at least 50 different open reading frames (ORFs) in the SARS-CoV-2 genome which include start codon (AUG) and stop codons (UAG, UAA, UGA). Origin of transcription sequences allow for SARS-CoV-2 to encode for 50 protiens with structural, nonstructural and accessory functions [10,14,15]. Two-thirds of the SARS-CoV-2 genome encodes for the two main transcription units, ORF1a and ORF1b which are attributed in the encoding of polyproteins, PP1a and PP1ab respectively. PP1ab contains ORFs for 16 nonstructural proteins (Nsp1-Nsp16). Non-structural proteins have functions in replication, proofreading, translation, suppressing host proteins, blocking immune responses, and stabilisation [10,15]. One-third of the SARS-CoV-2 genome encodes the structure and accessory proteins. Accessory genes are positioned along with the structural genes. There are at least nine ORFs for accessory proteins in the SARS-CoV-2 genome. Accessory proteins do not play a significant role in viral replication but play a role in the interaction between the viral genome and the host particularly blocking the production of cytokines [10,14,15]. With regards to structure, there are four main structural proteins in SARS-CoV-2. These are the Spike (S), Membrane (M), Envelope (E), and the Nucleocapsid (N) proteins. The Nucleocapsid proteins coat the SARS-COV-2 RNA for genetic protection [10,[13][14][15].
During infection, the Spike glycoprotein undergoes cleavage (unique furin-like cleavage site (FCS)) into amino (N) terminal S1 and carboxyl (C)-terminal S2 subunits [10,16]. The SI subunit facilitates the incorporation of SAR-CoV-2 into the host cell while the S2 subunit is responsible for membrane fusion. The S1 subunit has a receptor binding (RBD) and N-terminal (NTD) domain. SARS-CoV-2 binds to the human angiotensin-converting enzyme 2 (ACE2) receptors which are abundant in the respiratory epithelium but are also expressed by other cells such as the upper oesophagus, enterocytes (ileum), myocardial cells, proximal tubular cells (kidney), and urothelial cells (bladder). The binding is achieved through the SARS-CoV-2 Spike Protein (S1). Once the attachment has been achieved priming of the SARS-CoV-2 Spike Protein (S2) by the host transmembrane serine protease 2 (TMPRSS2) allows for viral cell entry and replication endocytosis [10,13].
Of interest in this study is the impact of the SARS-CoV lineages and sub-lineages in the COVID-19 epidemic waves in South Africa. Since the importation of SARS-CoV-2 in South Africa, with the first reported COVID-19 case on the 5 th of March 2020, South Africa has observed 3 consecutive COVID-19 epidemic waves [2,38]. The first COVID-19 epidemic wave in South Africa was observed between the reporting periods of 5 March to 30 September 2020 with a peak of 173 587 Active COVID-19 cases and a peak date of 26 July 2020 [39,40]. The second COVID-19 epidemic wave in South Africa resurged on the 1 st of December 2020 with 212 529 peak Active COVID-19 cases and a peak date of 15 January 2021 [41]. On the 27th of April 2021, the third COVID-19 epidemic wave in South Africa resurged. The third COVID-19 epidemic wave in South Africa had two observed peaks on the 10 th of July and on the 26 th of August 2021 with 211 052 and 169 039 Active COVID-19 cases respectively lasting up to the end of September 2021 [2]. The response by the Government of South Africa to the COVID-19 epidemic was the establishment of a National Coronavirus Command Council to oversee the epidemic, the use of health policy measures including NPIs to try to mitigate the transmission of COVID-19 and the implementation of COVID-19 vaccination programmes to try to vaccinate the South African population against COVID-19 [42-47]. As of 19 September 2021, 15.8 million people have been vaccinated against COVID-19 in South Africa mainly with the Pfizer-BioNTech (Comirnaty) and the Johnson & Johnson/Janssen COVID-19 vaccines [48].
"Globally, systems have been established and are being strengthened to detect 'signals' of potential VOIs or VOCs and assess these based on the risk posed to global public health" [21]. In South Africa, the Network for Genomics Surveillance in South Africa (NGS-SA) was formed to understand the spread of SARS-CoV-2 [49]. The NGS-SA was launched in June 2020 and this consortium comprised of the National Health Laboratory Services (NHLS) . CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint  [50]. In the first COVID-19 epidemic wave in South Africa, 16 SARS-CoV-2 lineages specific to South Africa were identified from 1 365 high-quality whole genomes [49]. From these 16 lineages, three main clusters (B.1.1.54, B.1.1.56, and C.1 SARS-CoV-2 lineages) were identified to have caused approximately 42% of the SARS-CoV-2 infections in South Africa [49]. Another sub lineage specific to South Africa was the B.1.106 lineage that emerged in the Kwa-Zulu Natal province in a nosocomial outbreak during the first COVID-19 epidemic wave [49]. The prevalence of this sub-lineage decreased as a result of control measures [49,51]. The C.I lineage (first identified C lineage of SARS-CoV-2) was the most geographically spread lineage in South Africa's first COVID-19 epidemic wave [49]. Before the resurgence of the second COVID-19 epidemic wave in South Africa, the Beta (B.1.351, B1.351.2, B.1.351.3 lineages) SARS-CoV-2 VOC (formerly GR/501Y.V2) was identified in an analysis of 2 704 South African SARS-CoV-2 genotypes (samples up to 14 th of December 2020) from the GISAID database. The Beta (B.1.351 lineage) SARS-CoV-2 VOC was detected in samples collected in the months of October 2020 [52]. The Beta (B.1.351 lineage) SARS-CoV-2 VOC had 9 mutations in the Spike protein at L18F, D80A, D215G, R246I, K417N, E484K, N501Y, D614G, and A701V. Three of those mutations (K417N, E484K, N501Y)) were located in the RBD. The Beta SARS-CoV-2 lineage became the dominant lineage in South Africa's second COVID-19 epidemic wave rapidly replacing the three main clusters (B.1.1.54, B.1.1.56 and C.1 SARS-CoV-2 lineages) identified during the first COVID-19 epidemic wave [52]. A study by [53] showed that the Beta SARS-CoV-2 lineage required a halfmaximal inhibitory concentration (IC50) 6 to 200 fold higher than the lineages identified in the first wave. Indicating that the Beta SARS-CoV-2 variant may escape neutralising antibody response developed from prior infections [53]. In the resurgence of the third COVID-19 epidemic wave in South Africa, four SARS-CoV variants were identified which were the Alpha, Beta, Eta, and Delta SARS-CoV variants. Genomic data for South African samples, identified 65 % of 1147 whole genomes from May 2021 to be the Beta SARS-CoV-2 variant. The Alpha, Delta, and Eta SARS-CoV-2 variants accounted for 6 %, 16 %, and 1 % of those samples respectively. In June 2021, with 2931 genetic sequences in that period, the Delta SARS-CoV-2 variant had become the dominant variant in samples collected in South Africa at 66 % while the Beta and Alpha SARS-CoV-2 variants accounted for 16 % and 4 % respectively [54]. By the end of South Africa's third COVID-19 epidemic wave in September 2021, the Delta SARS-CoV-2 variant accounted for 96 % of the 186 whole-genome sampled in that period while the C1.2 SARS-CoV-2 lineage account for 1 % of those samples [54]. The C1.2 SARS-CoV-2 lineage was a new South Africa specific SARS-CoV-2 lineage identified in South African samples in the month of May 2021 and evolved from the C.1 SARS-CoV-2 lineage. The C.1.2 SARS-CoV-2 lineage had several mutations with multiple substitutions (R190S, D215G, N484K, N501Y, H655Y and T859N) and deletions (Y144del, L242-A243del) in the Spike protein. There was also an accumulation of additional mutation (C136F, Y449H and N679K) in the C.1.2 SARS-CoV-2 thought to likely impact neutralisation sensitivity. The C.1.2 has been detected across the majority of South African provinces and in seven other countries [55].
The evolution of SARS-CoV-2 has played a significant role in the resurgence of COVID-19 epidemic waves in South Africa and across the globe. South Africa has a unique observation of the evolution of SARS-CoV-2, with distinct SARS-CoV-2 lineages dominating certain epidemic periods. This unique observation allows for an investigation of the detected SARS-CoV-2 lineages' impact on COVID-19 transmissibility and severity through analysis of epidemiological data. In this study, inferential statistical analysis was conducted on South African COVID-19 epidemiological data to investigate the impact of SARS-CoV-2 lineages in the South African COVID-19 epidemiology.

Methodology
The general methodology in this study involved the collation of South African COVID-19 epidemiological data, the regression and normalisation of the epidemiological data, and inferential statistical analysis.

Collation of South
Where n is the number of patients and i is the reported case date. Then the COVID-19 Hospital Daily Discharge Rate (DR) and Case Fatality Rate (CFR) were then calculated using Equation 4 and Equation 5 respectively.

Equation 5
Where n is the number of patients and i is the reported case date. Excess mortality is an account of deaths from all causes relative to expected deaths based on previous trends/observations. Excess mortality/deaths allow for accounting for miscounted or underreported COVID-19 Deaths and indirect Deaths related to the COVID-19 pandemic [59]. South African Weekly Natural and Excess (Natural) Deaths were obtained from the South African Medical Research Council (SAMRC) [60] for the reporting period of 29 December 2019 to 18 September 2021. The Weekly Unreported Excess Deaths (Natural) to COVID-19 Death Ratio (ECDR) was then calculated based on the methodology in [58] using Equation 6:

Equation 6
Where i is the Weekly Reported Date.
. CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

Stratification of South African COVID-19 Epidemiological Data
To draw inferential comparisons regarding the impact of the evolution of SARS-CoV-2 in the South African COVID-19 epidemiology, the collated South African COVID-19 epidemiological data in Section 2.1.1 and 2.1.2 was stratified based on the observed COVID-19 epidemic wave period 1, 2 and 3 in South Africa. Based on a review of [49,52,54], Table 1 was generated summarising the Sotuh African (SA) SARS-CoV-2 lineage clusters observed in the genomic data in the observed COVID-19 epidemic wave periods. The South African COVID-19 epidemic wave period 1, 2 and 3 were classed as reported case data in the period of 5 March 2020 to 30 September 2020, 01 October 2020 to 26 April 2021 and 27 April 2021 to 19 September 2021 respectively. The labels of stratified variables were given a reference of "_1", "_2", "_3" for the three respectively epidemic periods. By this stratification, the cluster of lineages identified in Table 1 was assumed to be the SARS-COV-2 lineages resulting in the respective COVID-19 epidemic waves in South Africa. For Cumulative Epidemiological Data (South African Cumulative COVID-19 Admission Age Profile, Cumulative COVIID-19 Hospital Deaths Age Profile and Cumulative COVID-19 Patients Discharged Alive) the data was adjusted using Equation 7 to remove the cumulative data from the previous COVID-19 epidemic period.
Where n is the number of patients, i is the reported date and j is the last reported date of the previous COVID-19 epidemic period. The stratification of data in this study was done by splitting the data using the epidemic period variable in the IBM SPSS STATISTICS 27 Software.
. CC-BY 4.0 International license It is made available under a perpetuity.
is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

Regression of South African COVID-19 Epidemiological Data.
In an epidemic, several factors influence the outcome of the observed epidemic. This includes testing, reporting capacity and population movement for a human to human infectious diseases such as COVID-19 (disease daily effective contact rate) [61]. For inferential statistical comparative analysis to be conducted between epidemiological data from different epidemic periods, the covariance of the epidemiological data needs to be accounted for in the analysis. Covariance is the measure of the combined variability of two random variables [62]. The independent variable in the covariance is a predictor variable influencing the outcome of a dependent variable. [58] showed that there was a positive correlation between South African COVID-19 Daily Tests and Cases in the first COVID-19 epidemic period. The South African COVID-19 daily testing was also shown not to be consistent throughout the period. To understand the covariance between South African COVID-19 daily tests and cases, Descriptive Statistical Analysis, Bivariate Analysis using the Two-tailed Pearson and Spearman tests and Analysis of Variance using the Univariate General Linear Model was conducted on the COVID-19 Daily Tests and Cases (Dependent Variable) using the IBM SPSS STATISTICS 27 Software for the three COVID-19 epidemic periods. Table 2 shows that the mean COVID-19 daily tests in the first, second and third South African COVID-19 epidemic wave period were 20 575±14 062, 31 046±14 115 and 46 822±18 460 respectively. The value of these means indicates the difference in testing capacity in the three epidemic periods. The Skewness values shown in Table 2 indicate that the normal distribution of the South African COVID-19 Daily Tests and Cases were positively skewed with the COVID-19 Daily Cases being more skewed than the Daily Tests. The skewness of COVID-19 Daily Cases was expected considering that the distribution is lognormal (Galton distribution) with differential (calculus) relations to normally distributed random variables such as the COVID-19 effective daily contact rate in this probability space. The application of differentials to model COVID-19 using random variables in stochastic COVID-19 epidemiological modelling was shown to be effective in modelling the COVID-19 epidemiology in South Africa [58,63,64].  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint  Table 3  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

Linear Regression of Number of Reporting Hospitals, COVID-19 Active and Hospitalised Cases
The NICD DATCOV surveillance system in South Africa only started publishing reports on the reporting date of 24 May 2020 thus data from 5 March 2020 to 23 May 2020 in the first COVID-19 epidemic wave period is missing. Data from 9 October 2020 to 26 October 2020 in the second COVID-19 epidemic wave period was also missing. The number of hospitals reporting to the NICD DATCOV surveillance system in South Africa' first COVID-19 epidemic wave period was initially 204 facilities and the facilities increased to 551 by the end of this period. In South Africa's second and third COVID-19 epidemic wave periods, the facilities reporting by the end of these periods were 646 and 664 respectively. To understand the covariance between South African COVID-19 Active Cases, Number of Facilities Reporting to the NICD DATCoV surveillance system and COVID-19 Hospital Admitted Cases (Dependent Variable), Bivariate Analysis using the Two-tailed Pearson and Spearman tests and Analysis of Variance using the Univariate General Linear Model was conducted on these variables using the IBM SPSS STATISTICS 27 Software for the three COVID-19 epidemic periods. Table 6 and Table 7

s Correlation Coefficients and P-Values (Sig. (2-tailed)) between Number of Facilities Reporting to the NICD DATCoV System (Independent Variable), COVID-19 Active Cases (Covariant Variable) and COVID-19 Hospital Admitted Cases (Dependent Variable) in the First, Second and Third COVID-19 Epidemic Wave in South Africa (Correlations)
The values of the Standardized, Pearson and Spearman's Correlation Coefficients observed indicate a strong positive correlation between COVID-19 Active and Hospital Admitted Cases. This result was expected based on probability theory. An increase in COVID-19 active cases increases the probability of stochastic proportions of COVID-19 severity including severe and critical COVID-19 thus higher hospitalisations. This correlation was also well demonstrated by stochastic COVID-19 epidemiological models such as in [58,63,64]. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

Regression of the COVID-19 Daily Effective Contact Rate and Movement Restrictions (NPIs)
South Africa's COVID-19 health policy response to the COVID-19 epidemic waves in South Africa was implemented in the form of National Lockdown Alert Level policies. The National Lockdown Alert Level policies were largely entry and exit screening at borders, limitations of movements and gatherings, closure/limitations of institutions and business activities, ban/limiting of alcohol and tobacco industries, isolation, quarantine of potentially infected persons, contact tracing protocols, use of personal protective equipment (PPE) and hygienic protocols [58]. In South Africa's first COVID-19 epidemic wave period the National Lockdown Alert The daily effective contact rate is the average number of adequate contacts per infective per day. It is directly proportional to the reproductive number [61]. [58] showed the impact of NPIs particularly movement restriction on the COVID-19 daily effective contact rate in South Africa. In this study, a relationship between the community mobility of the South African population and the COVID-19 Daily Effective Contact Rate was established. [58] showed that an increase in 1 Alert Level in the South African National Lockdown Alert Levels was 30 % more effective in reducing population movement and resulted in a reduction in the COVID-19 Daily Effective Contact Rate by 4.13 % to 14.6 % compared to the preceding Alert Level. The study also showed that the National Lockdown Alert Levels 1 and 2 had a similar impact on the population movements in the South African communities in the first COVID-19 epidemic wave period [58]. Figure 1 shows the South African community mobility in the Retail and Recreation, Grocery and Pharmacy, Parks, Transit Stations, Workplaces and Residences locations in the three COVID-19 epidemic periods. The modulus mean movement change from baseline in South African locations in the period of 26 March to 25 May 2020 for the first COVID-19 epidemic wave period were higher than those observed in the second and third COVID-19 epidemic wave periods. This period corresponds to the implementation of the National Lockdown Alert Levels 5 and 4 in the first COVID-19 epidemic wave. In the second and third COVID-19 epidemic wave periods the highest alert levels implemented were the adjusted is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Alert Level 3 and 4 respectively. The adjustment in these Alert Levels resulted in more eased movement restrictions than its predecessors while the National Alert level 5 lockdown implemented in the first COVID-19 epidemic wave period had the hardest population movement restrictions experienced by the South African population. These differences resulted in the lower modulus mean movement change from baselines in South African locations in the second and third COVID-19 epidemic wave periods. Thus, caution must be taken when concluding the impact of SARS-CoV-2 lineages on the observed COVID-19 epidemic in South Africa as the different movement restrictions might have also had an impact.

Normalisation and Inferential Statistical Analysis of South African COVID-19 Epidemiological Data 2.3.1 Comparative Analysis on the COVID-19 Transmissibility in the South African COVID-19 Epidemic Wave Periods
Disease transmissibility is the ability for a disease to be transmitted from one individual to another through infection. The basic reproductive number (R0) is a parameter that is largely used to describe the transmissibility of a disease. The basic reproductive number is the number of secondary infections that a primary infected person would produce in a completely susceptible population [61]. If R0 is greater than one then the disease is in an endemic equilibrium which results in positive exponential growth in cases. If the R0 is less than 1 then the disease is in the disease-free equilibrium which results in negative exponential growth in cases. The peak of the COVID-19 epidemic waves represents the turning point of the effective reproductive number. [58] showed through stochastic COVID-19 epidemiological modelling that the effective reproductive number in the first COVID-19 epidemic wave in South Africa was between 1.98 to 0.40. The NICD conducted a modelling analysis to estimate the initial basic reproductive number in South Africa using the method in [67] and estimated that it was 1.29 (95%CI: 1.9601.58)) during the National Lockdown Alert Level 5 rising to 1.5 by end of April 2020, 1.05 (95%CI:1.01-1.09) during the National Lockdown Alert Level 3 between 1 June and 1 August 2020 [68].
In this study, the COVID-19 transmissibility was measured through the magnitude of mean and variance of the COVID-19 Daily and Active Cases. Considering the linear positive correlation between the COVID-19 Daily Tests and Cases, for the comparative inferential analysis of COVID-19 Daily and Active Cases between the COVID-19 epidemic wave periods a normalised parameter (COVID-19 Daily Positive Tests) was developed based on Equation 8 to normalise the variance of testing in reported cases. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

Equation 8
Where n is the number of cases/tests and i is the reported case date. For the comparative inferential statistical analysis conducted to understand the impact of the SARS-CoV-2 lineage clusters in Table 1  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint For the comparative inferential statistical analysis conducted to understand the impact of the SARS-CoV-2 lineage clusters in Table 1 Table 8 shows the descriptive statistics for the COVID-19 active, daily cases and daily positive tests for the first, second and third COVID-19 epidemic wave periods in South Africa.  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

Results and Discussions 3.1 Impact of SARS-CoV-2 Lineages on the COVID-19 Transmissibility in South Africa
The copyright holder for this this version posted October 25, 2021. ; https://doi.org/10.1101/2021.10.22.21265316 doi: medRxiv preprint detected in the third COVID-19 epidemic wave resulted in higher COVID-19 transmissibility in the South African population than the SARS-CoV-2 lineage clusters 1 and 2. The difference between the mean COVID-19 daily positive tests in South Africa's first, second and third COVID-19 epidemic wave period can also be observed in Figure 2. A paired T-test of the COVID-19 daily positive tests between the first and second COVID-19 epidemic wave period showed no significant difference at 95 % confidence interval between these COVID-19 epidemic periods with a p-value of 0.709 (shown in Table 9). While the paired T-test of the COVID-19 daily positive tests between the Pair 2 (first and third COVID-19 epidemic wave period) and Pair 3 (second and third COVID-19 epidemic wave period) showed significant difference at 95 % confidence interval between the respective COVID-19 epidemic periods with p-values of 1.82×10 -11 and 5.87×10 -05 respectively (shown in Table 9).  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

Impact of SARS-CoV-2 Lineages on the COVID-19 Hospital Admissions
The COVID-19 hospital to active cases indicates the severity of COVID-19 based on the relative amount of COVID-19 hospitalised cases to the active cases and accounts for the covariance of active cases in hospitalised cases. According to WHO, patients with severe and critical COVID-19 are most likely to be hospitalised while those with moderate COVID-19 may present to an emergency unit or primary care/outpatient department [71]. Table 10 shows the descriptive statistics for the COVID-19 hospital to active cases for the first, second and third COVID-19 epidemic wave periods in South Africa. Table 10 shows that the mean COVID-19 daily hospital to active cases in South Africa's first, second and third COVID-19 epidemic wave period were 6.8±1.82 %, 14.5±4.68 % and 10.6±2.81 % respectively. The second COVID-19 epidemic wave period in South Africa had the highest COVID-19 hospital to active cases followed by the third COVID-19 epidemic wave period. The difference of the mean COVID-19 hospital to active cases between Pair 1 (COVID-19 epidemic wave period 1 and 2), Pair 2 (COVID-19 epidemic wave period 1 and 3), Pair 3 (COVID-19 epidemic wave period 2 and 3) was 113 %, 55.8 % and 36.8 % respectively. The difference between the mean COVID-19 hospital to active cases in South Africa's first, second and third COVID-19 epidemic wave period can also be observed in Figure 3. Paired T-tests of the COVID-19 hospital to active cases between the Pair 1, Pair 2 and Pair 3 showed significant difference at 95 % confidence interval between the respective COVID-19 epidemic periods with p-values of 4.25×10 -21 , 9.10×10 -44 and 3.90×10 -06 respectively (shown in Table 11). The results in Section 3. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint   is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint  Table 12 shows the descriptive statistics for the COVID-19 hospital admission status for the first, second and third COVID-19 epidemic wave periods in South Africa.  Figure 4 shows the COVID-19 hospital admission status profile in the first, second and third COVID-19 epidemic wave periods in South Africa.  Figure 4 shows that most COVID-19 hospitalised cases in South Africa were hospitalised in the general ward (72.8-78.7 %). Figure 4 also shows that the COVID-19 patients on Oxygen were the second-largest admission is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint  1-16.7 %). The COVID-19 patients on oxygen in the second and third COVID-19 epidemic wave period were higher than those observed in the first COVID-19 epidemic wave period with those in the second COVID-19 epidemic wave period being the highest. Paired T-tests of the mean COVID-19 hospital admission status in the Pair 1, Pair 2 and Pair 3 showed significant difference at 95 % confidence interval between the respective COVID-19 epidemic periods with pvalues in the range of 2.86×10 -41 to 0.00219 respectively (shown in Table 13). Table 13 shows that the COVID-19 patients admitted in the intensive care unit in the first and third COVID-19 epidemic wave period were not significantly different at a 95 % confidence interval with a p-value of 0.514.

Figure 4: The mean COVID-19 Hospitalised Admission Status (with 95 % Confidence Interval (CI) Error Bars) in the First, Second and Third COVID-19 Epidemic Wave Period in South Africa
The results in Section 3.2.2 suggest that the SARS-CoV-2 lineage clusters observed in the first, second and third COVID-19 epidemic wave periods in South Africa resulted in a similar distribution of the COVID-19 hospital admission status profile however the mean COVID-19 admission status was significantly different at 95 % confidence interval. The results also show that the SARS-CoV-2 lineage cluster 1 (predominantly B.   is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint  Table 14 shows the descriptive statistics for the COVID-19 hospital admission age profile for the first, second and third COVID-19 epidemic wave periods in South Africa. Table 14Table 12 shows that the mean COVID-19  hospitalised cases in the first COVID-19 epidemic wave period in the ages of 0 to 9, 10 to 19, 20 to 29, 30 to 39,  40 to 49, 50 to 59, 60 Figure 5 shows the COVID-19 hospital admission age profile in the first, second and third COVID-19 epidemic wave periods in South Africa.  Figure 5 shows that most COVID-19 hospitalised cases in South Africa's first, second and third COVID-19 epidemic wave period were in the ages of 50 to59 (19.6 %), 50 to 59 (24.0 %) and 60 to 69 (15.5 %) years respectively. COVID-19 patients in the age groups of 40 to 49 were the second-largest admitted age groups in the first and third COVID-19 epidemic wave period (14.1-16.7 %) while for the second COVID-19 epidemic wave period it was the age group of 60 to 69 years (18.2 %). COVID-19 admitted patients in the age groups of 0 to 19 years were relatively low (1.8 to 4.4 %) and highest in the second COVID-19 epidemic wave period. The incidence of COVID-19 in children has been reported to be low around the world particularly severe and critical COVID-19 requiring hospitalisation [8]. Figure 5 shows that the COVID-19 hospital admitted age profile for the first and is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

Impact of SARS-CoV-2 Lineages on the COVID-19 Hospital Admission Age Profile
The copyright holder for this this version posted October 25, 2021. ; https://doi.org/10.1101/2021.10.22.21265316 doi: medRxiv preprint second COVID-19 epidemic wave period was relative normally distributed, while the distribution in the third COVID epidemic wave period was left-skewed with increased incidence in the age groups over 70 years.

Figure 5: The mean COVID-19 Hospitalised Admission Age Profile (with 95 % Confidence Interval (CI) Error Bars) in the First, Second and Third COVID-19 Epidemic Wave Period in South Africa
In general, paired T-tests of the COVID-19 hospital admission age profile in Pair 1, Pair 2 and Pair 3 showed significant differences at 95 % confidence interval between the respective COVID-19 epidemic periods with pvalues in the range of 7.66×10 -44 to 0.040 respectively (shown in Table 15). However, Table 15 also shows that the COVID-19 patients admitted in age group 0 to 9 years in first and third COVID-19 epidemic wave periods, age group 20 to 29 years in second and third COVID-19 epidemic wave periods, age group 60 to 69 years in first and, second and third COVID-19 epidemic wave periods, age group 70 to 79 years in first and second COVID- is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint   Ages of 0 to 9, 10 to 19, 20 to 29, 30 to 39, 40 to 49, 50 to 59, 60 to 69, 70 to 79,  80 to 89 years between Pair 1 (1 st and 2 Table 16 shows the descriptive statistics for the COVID-19 hospital death age profile for the first, second and third COVID-19 epidemic wave periods in South Africa. Table 16Table 12    is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint  Figure 6 shows that most COVID-19 hospitalised deaths in South Africa in the first, second and third COVID-19 epidemic wave period were in the ages of 60 to 69 (25.8 %), 50 to 59 (41.2 %) and 60 to 69 (25.6 %) years respectively. COVID-19 hospitalised deaths in the age groups of 70 to 79 were the second-largest death age groups in the first and third COVID-19 epidemic wave period (18.1-21.3 %) while for the second COVID-19 epidemic wave period it was the age group of 60 to 69 years (28.5 %). COVID-19 hospitalised deaths in the age groups of 0 to 29 years were relatively low (0.194 to 1.97 %). The incidence of COVID-19 deaths in children has been reported to be low around the world [8]. Figure 6 shows that the COVID-19 hospitalised death age profile for the first, second and third COVID-19 epidemic wave periods were similar in distribution except for the high incidence of COVID-19 hospitalised deaths in the age group of 50 to 59 in the second COVID-19 epidemic wave period. In general, paired T-tests of the mean COVID-19 hospitalised deaths age groups between the Pair 1, Pair 2 and Pair 3 showed significant difference at 95 % confidence interval between the respective COVID-19 epidemic periods with p-values in the range of 5.04×10 -40 to 0.00238 respectively (shown in Table 17). Table 17Table 13 shows that the COVID-19 hospitalised deaths in age groups of 0 to 9 years in the first, second and third COVID-19 epidemic wave period were not significantly different at a 95 % confidence interval with p-values of 0.144 to 0.177.  Table 18 shows the cumulative COVID-19 death risk ratio for South African COVID-19 hospital death age groups for the first, second and third COVID-19 epidemic wave period in South Africa regarding the age group of 0 to 9 years. The cumulative risk of death in COVID-19 hospitalised deaths increased with increasing age group groups with age groups of 50 to 69 years having the highest risk.  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint confidence interval with a p-value of 0.005. This result suggests that the mean COVID-19 hospital discharge rate in the first and second, second and third COVID-19 epidemic wave periods were statistically similar while that between the first and third COVID-19 wave periods were different. Paired T-tests of the excess natural deaths, weekly reported COVID-19 deaths and ECDR in Pair 3 showed no significant difference at 95 % confidence interval with p-values of 0.908, 0.918 and 0.760 respectively (shown in Table 20). While the paired T-test in Pair 1 and Pair 2 showed a significant difference at a 95 % confidence interval with p-values in the range of 0.001 and 3.64×10 -06 to 0.015 (shown in Table 20). These results suggest that the mean excess natural deaths weekly reported COVID-19 deaths and ECDR in the second and third COVID-19 epidemic wave periods were statistically similar while that between the first and third, first and second COVID-19 wave periods were statistically different.  is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

Conclusions
Based on the COVID-19 detection rate in South Africa, the SA SARS-CoV-2 lineage cluster 3 which was predominantly the Delta (B.1.617.2) SARS-CoV-2 VOC resulted in an increase in the COVID-19 transmissibility in the South African population by 53.9 to 54.8 %. This is relative to the SA SARS-CoV-2 lineage cluster 1 (predominantly B. The SA SARS-CoV-2 lineage clusters 1, 2 and 3 resulted in a similar distribution of the COVID-19 hospitalised death age profile. Most COVID-19 hospitalised deaths in South Africa in the first, second and third COVID-19 epidemic wave period were in the ages of 50 to 69 years. COVID-19 hospitalised deaths in the age groups of 0 to 29 years in South Africa were relatively low (0.194 to 1.97 %). The cumulative risk of death in COVID-19 hospitalised deaths increased with increasing age group groups with age groups of 50 to 69 years having the highest risk. The SA SARS-CoV-2 lineage cluster 2 (predominantly the Beta (B.1.351) SARS-CoV-2 VOC) resulted in an increase in the incidence of hospitalised deaths in age groups of 50 to 59 compared to the SARS-CoV-2 lineage clusters 1 and 3.
The mean COVID-19 hospital case fatality rate in South Africa's first, second and third COVID-19 epidemic wave periods were 2.06±1.10 %, 2.29±1.60 % and 2.08±1.16 % respectively. The number of excess natural deaths not accounted for in the COVID-19 reported deaths were the same as the COVID-19 reported deaths in the first COVID-19 epidemic wave period and twice the COVID-19 reported deaths in the second and third COVID-19 epidemic wave period. The SA SARS-CoV-2 lineage clusters 1, 2 and 3 resulted in similar COVID-19 hospital case-fatality rates in South Africa. While the SA SARS-CoV-2 lineages clusters 2 and 3 resulted in similar COVID-19 discharge rates, excess natural deaths, weekly reported COVID-19 deaths and ECDR in South Africa. The SA SARS-CoV-2 lineage cluster 1 (predominantly the B.1.1.54, B.1.1.56 C.1 SARS-CoV-2 lineages) resulted in statistically different excess natural deaths, weekly reported COVID-19 deaths and ECDR in South Africa relative to the aforementioned SARS-CoV-2 clusters.
This study shows that the evolution of SARS-CoV-2 resulted in an increase in COVID-19 transmissibility and severity in South Africa. The Delta SARS-CoV-2 VOC resulted in increased COVID-19 transmissibility in the South African population while both the Beta SARS-CoV-2 VOC and Delta SARS-CoV-2 VOC resulted in more severe COVID-19 than the initial SARS-CoV-2 lineages detected in South Africa's first epidemic wave period. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

Acknowledgements
ARI would like to thank the members of the ARI African Disease Demographic Research Group (ADDRG) and African COVID-19 Modelling Research Group (ACMRG) for their voluntary commitment to working in the ARI COVID-19 Research Project. We acknowledge the work by the National Institute for Communicable Diseases (NICD), Western Cape Department of Health Provincial Health Data Centre (PHDC), South African Medical Research Council (SAMRC) and the Network for Genomics Surveillance in South Africa (NGS-SA) in which the ARI COVID-19 Project draws a lot of its data from. Lastly, we want to salute the scientific community, governments, health care workers, essential personnel in their response to the pandemic and we pay homage to those who have lost their lives due to the COVID-19 pandemic. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint