Suspected reinfections of SARS-COV-2 in Khyber Pakhtunkhwa, Pakistan - analysis of province-wide testing database.

Dear editor,

SARS-CoV-2 reinfection is a relatively new phenomenon in the ongoing coronavirus disease (COVID-19) pandemic. Although, the first confirmed case of reinfection was identified in August 2020, studies reporting reinfection have been scarce, and limited only to single-case studies, or smaller case-series and cohorts. Only few studies have determined incidence of reinfection in large, population-based settings based on epidemiological evidence.2, 3, 4, 5, 6, 7, 8 If genotyping is unavailable, Centres for Disease Control (CDC) recently set an investigative criteria, defining SARS-CoV-2 reinfection as a second positive real time reverse transcription-polymerase chain reaction (RT-PCR) test, taken ≥90 days after the primary result. However, not all published work follows a uniformly set criteria for what constitutes confirmed reinfection.

Studies which follow the CDC criteria indicate high level of protection following recovery from infection, with reinfection suspected in less than 1.0% of those previously infected. , 5, 6, 7, 8 However, further detailed prospective studies are needed to assess frequency of recurring infection. There is a possibility that a large proportion of reinfections are undetected, which could explain the low risk of SARS-CoV-2 reinfection reported in literature. This may especially be true in lower-middle-income settings, such as ours, where testing for asymptomatic infection is rare, and carrying out mass-scale genomic sequencing is not always logistically feasible.

Pakistan is currently experiencing a fourth epidemic wave of SARS-CoV-2 transmission. This provided us an opportunity to retrospectively determine incidence of suspected reinfections after three global waves in the pandemic. Using epidemiological criteria set by CDC, we undertook this study to identify suspected cases of SARS-CoV-2 reinfection, in the province of Khyber Pakhtunkhwa (KP), using the provincial government's contact tracing database.

Methods

This descriptive cross-sectional study was approved by the Khyber Medical University (KMU) Ethical Committee. Due to retrospective nature of our analysis, we were granted a waiver from obtaining informed consent.

In KP, testing has been extended to symptomatic and suspected cases, those presenting at hospitals, people identified through contact tracing, contacts of confirmed cases, people with travel history, or anyone using the free COVID-19 response hotline. Samples are sent to one of the 28 public and private sector diagnostic laboratories, serving all regions of KP province, which has a population of 35.53 million people. This network of laboratories is regulated by the provincial health department, and all SARS-CoV-2 viral RNA RT-PCR tests are conducted under guidelines from National Institute of Health (NIH), Islamabad. To ensure uniformity in sample processing and data analysis, all laboratories in KP are periodically assessed through an External Quality Assurance program mandated by NIH. Individual-level test results are logged onto, and maintained on a centralised contact tracing database.

We extracted retrospective data for all SARS-CoV-2 RT-PCR tests carried out from March, 2020, to July, 2021, in KP. For purpose of this study, primary infection was defined as a positive RT-PCR test for SARS-CoV-2. Suspected reinfection was defined as a second positive test taken for SARS-CoV-2, ≥90 days after the primary positive test result. Analyses were performed using MS Excel. Comparison between demographic variables of primary infections and suspected reinfections were made using Chi-square test.

Results

Between March 2020 and July 2021, a total of 2065,611 SARS-CoV-2 RT-PCR tests were conducted in KP. We found 142,787 individuals had tested positive for SARS-CoV-2 once, out of which 58.2% of primary infections were hospital referrals, while 41.8% were identified through KP's comprehensive contact tracing program. A subset of 29,617 individuals were identified, who had received repeat RT-PCR testing for SARS-CoV-2. Results of the primary infection were used as start of the 90 day interval, and multiple positive test results obtained within 90 days of each other were discounted. Overall, we observed 317 (0.22% of total positive cases) suspected cases of SARS-CoV-2 reinfections (Fig. 1 ).

Fig. 1

Schematic diagram showing selection of suspected reinfection cases from total conducted tests.

Mean age of individuals with suspected reinfection was slightly lower compared to those infected once (36.94 ± 14.34, vs 38.41 ± 18.59, p-value 0.004). Suspected reinfection was found to be more common in males (74.8%, p-value < 0.001), and in young and middle-aged adults, between 19 and 45 years old (54.1%, p-value < 0.001). Our analysis shows that 73.1% (231/317) of reinfections were detected through hospital referrals, while 26.9% (85/317) reinfections were identified through contact tracing (p-value < 0.001). Higher suspected reinfections were observed in urban centres where more testing is conducted (supplementary Fig. 1), with Peshawar division reporting highest proportion (189, 59.6%) of suspected reinfection cases (p-value <0.001) (Table 1 ).

Table 1

Demographic characteristics of SARS-COV-2 primary and suspected reinfection from Khyber Pakhtunkhwa province.

Characteristics		Infection status n (%)		P-Value
		Primary infection	Reinfection
Gender	female	49,416 (34.7)	80 (25.2)	<0.001
	male	93,054 (65.3)	237 (74.8)
Age, Years	Mean ± SD	38.41 (±18.59)	36.04 (±14.340)	*0.004
Age categories	≤ 18 years	15,815 (11.4)	17 (5.4)	<0.001
	19 - 30 years	35,066 (25.3)	123 (38.2)
	31 - 45 years	39,903 (28.8)	109 (34.4)
	46 - 60 years	30,040 (21.7)	46 (14.5)
	>60 years	17,775 (12.8)	22 (6.9)
Sample type	Contact tracing	59,592 (41.8)	85 (26.9)	<0.001
	Hospital referral	82,878 (58.2)	231 (73.1)
Region	Malakand	29,478 (20.8)	58 (18.3)	<0.001
	Hazara	16,596 (11.7)	29 (9.1)
	Mardan	14,270 (10.1)	18 (5.7)
	Peshawar	61,706 (43.5)	189 (59.6)
	Kohat	12,036 (8.5)	16 (5.0)
	Bannu	4307 (3.0)	3 (0.9)
	DI Khan	3550 (2.5)	4 (1.3)

calculated using independent samples t- test.

Discussion

These findings suggest that SARS-CoV-2 reinfection could be common in age groups often presumed less likely to suffer from severe, or critical COVID-19. This is in contradiction to evidence which shows symptomatic COVID-19 results in enhanced, long-lasting protective immunity following recovery from primary infection. Our results reveal that more than two thirds (73.1%) of recurring infections were picked up through hospital referrals, which could indicate that people with severe symptoms are more likely to seek medical attention.

Studies have demonstrated that a large proportion of recurring infections are asymptomatic, and reinfection is more common in individuals whose primary infections were also asymptomatic. , In Pakistan, the testing strategy has largely relied on targeting of symptomatic and suspected cases. Therefore, it is impossible to accurately predict the true number of SARS-CoV-2 reinfections, which may be significantly larger than what has been reported in literature.

There were several limitations to this work. Since data was gathered retrospectively, we were unable to obtain medical history, clinical demographics, or personal details, regarding individuals whose test results were utilised in our analysis. Thus, we could not correlate role of any pre-existing health conditions with suspected reinfection. Similarly, the proportion of individuals employed in high-risk professions, such as healthcare workers, was unknown. Our future work aims to address these limitations. To best of our knowledge, this is the largest multi-centre analysis (n = 142,787) of its kind carried out to date, reporting one of the highest event-rates (317 suspected reinfections) in published literature. Our results fall within range of reinfections reported in literature, despite limited testing being carried out in Pakistan for identification of asymptomatic cases. This analysis suggests that recurring infection with SARS-CoV-2 may indeed be a widespread phenomenon, although the risk is known to be low.

Declaration of Competing Interest

The authors declare no conflict of interest.