ABSTRACT
While previous research examined coronavirus disease 2019 (COVID-19) antiviral-vaccine interactions through exploratory subgroup analysis, none specifically designed for examining this interaction or its impact on post-acute outcomes. This study examined the interaction between nirmatrelvir–ritonavir and complete COVID-19 vaccination on reducing the risk of post-acute outcomes among COVID-19 patients. We followed COVID-19 patients hospitalized between 11 March 2022 and 10 October 2023, until 31 October 2023 in Hong Kong. Exposure groups were based on nirmatrelvir–ritonavir usage and vaccination status (fully or not fully vaccinated). Post-acute death and all-cause rehospitalization were the study outcomes. Propensity score weighting was applied to balance covariates among exposure groups, including age, sex, Charlson Comorbidity Index, and concomitant treatments. Multiplicative and additive interactions between nirmatrelvir–ritonavir and vaccination status were assessed. A total of 50,438 COVID-19 patients were included in this study and arranged into four exposure groups. Significant additive interaction on post-acute rehospitalization was observed (relative excess risk, 0.10; 95% CI, 0.02–0.19; p-value, 0.018; attributable proportion, 0.07; 95% CI, 0.01–0.12; p-value, 0.017; synergy index, 1.26; 95% CI, 1.02–1.55; p-value, 0.032). The interaction on post-acute mortality was marginally significant. In the subgroup analysis, the interaction effect is more pronounced in older adults, female, and CoronaVac recipients. In conclusion, our study demonstrated an additive interaction between nirmatrelvir–ritonavir and complete vaccination on post-acute outcomes, suggesting greater long-term benefits of the antiviral for fully vaccinated individuals compared to not fully vaccinated patients.
KEYWORDS: Paxlovid, long covid, interaction, post-covid, antiviral
Background
Long COVID-19, also known as post-COVID-19 condition or post-COVID sequelae, encompasses a wide range of new, recurring, or persistent health problems following SARS-CoV-2 infection. Long COVID is generally characterized by ongoing or emerging signs, symptoms, and conditions that occur after the acute phase of COVID-19 infection. According to a nationwide study in Scotland [1], the prevalence of long COVID-19 was 6.6%, 6.5%, and 10.4% at 6, 12, and 18 months after the initial infection.
Nirmatrelvir combined with the cytochrome P450 inhibitor ritonavir (nirmatrelvir–ritonavir) has been authorized by the United States Food and Drug Administration as antiviral treatments for COVID-19 [2]. While the literature has well demonstrated the short-term efficacy and effectiveness of nirmatrelvir–ritonavir in high-risk patients [3–5], the findings on the association between nirmatrelvir–ritonavir and long-term post-acute outcomes of COVID-19 were inconsistent among both non-hospitalized and hospitalized patients [6–9]. Apart from that, none of these studies were specifically designed for examining the interaction of nirmatrelvir–ritonavir with COVID-19 vaccines, where mixed evidence existed for the effectiveness of vaccination on reducing the risk of long-COVID-19 [10, 11].
Vaccine-acquired immunity in vaccinated individuals may interact with the antiviral action of nirmatrelvir/ritonavir and modify the effectiveness of the antiviral. This aligns with the findings of Ioannou et al. [8] who observed a higher reduced risk of several post-COVID-19 conditions among nirmatrelvir–ritonavir recipients who were also vaccinated, compared to those without any COVID-19 vaccination. Based on this, we hypothesize that there is an interaction effect between nirmatrelvir–ritonavir and complete vaccination on reducing the risk of post-acute mortality and rehospitalization of COVID-19 in an infection-naïve population. We used real-world data to examine the interaction in Hong Kong, where the presence of natural immunity from SARS-CoV-2 infection was limited prior to the Omicron epidemic.
Methods
Study design and data sources
This retrospective cohort study used two primary data sources: inpatient records from the Hong Kong Hospital Authority and vaccination records from the Hong Kong Department of Health. The Hospital Authority in Hong Kong is a statutory body responsible for managing all public hospitals and institutes, catering to over 7·3 million residents. It maintains a centralized electronic database that encompasses various healthcare utilization records, including inpatient, outpatient, and accident and emergency services. The electronic health records were linked to the COVID-19 vaccination registry held by the Department of Health. Diagnoses and procedures were coded according to the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). Data on infections, clinical outcomes, and vaccinations of all hospitalized COVID-19 patients were obtained from these sources, ensuring the representativeness of the general population in Hong Kong. The population-based cohort has also been used in other analyses [5, 7].
Participants
Inclusion and exclusion criteria
Patients included in this study met the following inclusion criteria:
First tested positive for SARS-CoV-2 by RT–PCR between 11 March 2022 (five days before nirmatrelvir–ritonavir became available in Hong Kong [12]) and 10th October 2023 (to ensure at least 21 days of follow-up before the end of data availability, which was 31st October 2023). In this study period, the RT–PCR tests were predominantly used among the inpatient population (i.e. 80% over all inpatients), and this helps to ensure a more accurate documentation for the prescription of nirmatrelvir–ritonavir. The first infection record was used to avoid potential confounding due to prior natural immunity when evaluating the interaction between nirmatrelvir–ritonavir and vaccination status. In Hong Kong, residents were required to report their test positivity results to health officials, ensuring the completeness of the infection data, together with the minimal number of cases before the Omicron outbreak due to the strict control measures in place [13, 14]. Throughout the study, a 21-day period was utilized to delineate the transition from the acute phase to the post-acute phase of SARS-CoV-2 infection, aligning with previous studies for consistency [15, 16].
Hospitalized with COVID-19, with the date of admission falling within ±3 days of the date of the positive SARS-CoV-2 test (index date) [5], accounting for potential delays in case confirmation during periods of high patient influx.
The exclusion criteria were:
Deceased within the first 21 days after the COVID-19 diagnosis.
Aged less than 18 years.
- Contraindications to nirmatrelvir–ritonavir [5–7, 17], including:
- Use of specific drugs contraindicated with nirmatrelvir–ritonavir in the 90 days before the positive RT–PCR date, such as amiodarone, lumacaftor–ivacaftor, rifampicin, apalutamide, phenobarbital, rifapentine, carbamazepine, phenytoin, ivosidenib, and primidone, and the herbal drug St John’s Wort (hypericum perforatum).
- Severe renal impairment, defined as an estimated glomerular filtration rate <30 mL/min per 1.73 m² body surface area, requiring dialysis, or having undergone renal transplantation.
- Severe liver impairment, including cirrhosis, hepatocellular carcinoma, or having undergone liver transplantation.
Received vaccinations other vaccines than the locally available vaccines (i.e. CoronaVac and Comirnaty).
Exposed and unexposed groups
In this study, there were two exposures of interest: the use of nirmatrelvir–ritonavir and vaccination status.
Regarding the use of nirmatrelvir–ritonavir, the exposed group consisted of patients who were prescribed with oral nirmatrelvir–ritonavir between three days prior to and five days after symptom onset [2, 18], and who did not receive molnupiravir between three days prior to and 21 days after symptom onset. The date of positive RT–PCR test was used as a proxy for the date of symptom onset [17], with an additional three-day buffer to accommodate for potential delays in case confirmation during periods of high SARS-CoV-2 infection rates. A sensitivity analysis was carried out to assess the result robustness when changing the three-day buffer to five days [19]. The unexposed group comprised patients who did not receive nirmatrelvir–ritonavir or molnupiravir between three days prior to and 21 days after the positive RT–PCR date.
Regarding vaccination status, patients were categorized as fully vaccinated or not fully vaccinated (including those who never received any vaccine) based on the type and number of vaccine doses received [20]. For those who received CoronaVac alone, individuals who received ≥3 doses were considered fully vaccinated, while those who received <3 doses were categorized as not fully vaccinated. For recipients of Comirnaty alone, individuals who received ≥2 doses were classified as fully vaccinated, while those who received <2 doses were considered not fully vaccinated. A vaccine dose was only counted if it was completed 14 days prior to the positive RT–PCR date because randomized trials indicated no vaccine protective effect within the first 2 weeks of vaccination, during which the body built up the vaccine-induced immunity [21]. For individuals who received combination of vaccine types, their vaccination status was determined on a case-by-case basis.
Covariates
The covariates included in the weights comprised age, sex, Charlson Comorbidity Index (CCI) calculated using diagnoses before the index date, and concomitant pharmacological and nonpharmacological treatments initiated between three days before the index date and the end of the acute phase (index date plus 21 days). Concomitant medications included dexamethasone, methylprednisolone, prednisolone, interferon-β-1b, baricitinib, tocilizumab, and remdesivir [18]. Nonpharmacological treatments included ICU admission and the use of ventilatory support, including intubation, mechanical ventilation, and oxygen supplementation [15]. The corresponding ICD-9 codes for non-pharmacological treatments and BNF codes for concomitant medications were listed in Supplement Table 1.
Outcomes
The outcomes included post-acute inpatient death and all-cause rehospitalization occurring 21 days after the index date. It is noteworthy that in Hong Kong, over 90% of all deaths are either presented at or occur in public hospitals [22]. Therefore, inpatient death in our study is representative of the overall mortality in the population. Both outcomes were assessed starting from 21 days after the index date, indicating events occurring during the post-acute phase of infection. Patients were followed up from the index date until inpatient death, rehospitalization, one year after discharge, or until the end of data availability (31st October 2023), whichever came first.
Statistical analyses
Propensity score weighting was applied to balance covariates among exposure groups. Propensity scores were estimated using multinomial logistic regression, with the dependent variable consisting of four groups: (1) no use of nirmatrelvir–ritonavir + not fully vaccinated, (2) no use of nirmatrelvir–ritonavir + fully vaccinated, (3) use of nirmatrelvir–ritonavir + not fully vaccinated, (4) use of nirmatrelvir–ritonavir + fully vaccinated. The estimate of interest was the average treatment effect on the treated, with the focal group defined as the use of nirmatrelvir–ritonavir + fully vaccinated group. Covariate balance between groups was assessed before and after weighting using the standardized mean difference (SMD), with SMD <0.1 indicating sufficient balance [23]. To assess the robustness of the confounding control method, we also employed an external-cause re-hospitalization as a negative control outcome [24], defined as a hospitalization due to injury and poisoning (ICD-9 800-999 and E000-E999).
To investigate the interaction between nirmatrelvir–ritonavir and vaccination status on outcomes, Cox proportional hazards model was used. The proportional hazards assumption was assessed through the scaled Schoenfeld residuals plotted against time. We followed the suggestion from Mansournia & Nazemipour to include both multiplicative and additive interactions in our analysis [25]. Multiplicative interaction between nirmatrelvir–ritonavir and vaccination status was evaluated by examining the coefficient of the product term in the Cox model. Additive interaction was assessed using the relative excess risk due to interaction (RERI), where RERI ≠ 0 indicates the presence of additive interaction. The attributable proportion due to interaction (AP) was also assessed, with AP ≠ 0 indicating the presence of additive interaction. Additionally, the synergy index (S) was computed, where S ≠ 1 suggests the presence of additive interaction [26]. When calculating these measures, which were developed for assessing risk factors rather than preventive factors, the use of nirmatrelvir–ritonavir and vaccination status were recoded into risk factors. Specifically, the reference category was defined as the group comprising individuals who used nirmatrelvir–ritonavir and were fully vaccinated, representing the stratum with the lowest risk when both factors are considered jointly [27]. Therefore, a presence of additive interaction indicates an additive risk of a post-acute outcome in patients who were not fully vaccinated and were not prescribed nirmatrelvir–ritonavir. The calculation formulas of RERI, AP, and S were included in the Supplement Information.
Subgroup analyses were conducted according to age (<65 years and ≥65 years), sex, vaccine type (CoronaVac and Comirnaty), and time since last dose of vaccination (≤6 months and >6 months). A sensitivity analysis was performed to adjust for covariates related to prior healthcare utilizations, including records of outpatient clinic visits, inpatient care, and Accident & Emergency visits during 2018–2019, as a proxy for healthcare seeking behaviour. A separate sensitivity analysis was conducted to assess whether the observed interaction was primarily influenced by the booster dose of Comirnaty by excluding patients who had received it.
All analyses were conducted in R statistical software (version 4.3.2) (R Program for Statistical Computing).
Ethics approval and consent to participate
Ethics approval was obtained from the Joint CUHK-NTEC Clinical Research Ethics Committee (Ref No. 2023.006). As this study was a retrospective analysis using secondary data without any personal information, the requirement for obtaining informed consent was waived.
Results
Of the 88,643 patients having their first positive RT–PCR test results between 11th March 2022 and 10th October 2023, 50,438 patients were eligible to be included in this study (Figure 1). Of the included subjects, 10,976, 10,092, 4982, and 11,665 subjects had no use of nirmatrelvir–ritonavir + not fully vaccinated, no use of nirmatrelvir–ritonavir + fully vaccinated, use of nirmatrelvir–ritonavir + not fully vaccinated, and use of nirmatrelvir–ritonavir + fully vaccinated, respectively.
Figure 1.
Flowchart of patient inclusion and exclusion.
The patient characteristics before weighting are provided in Table 1. Across all the four groups, the proportions of female (range, 47.3% to 51.3%) and mean age (range, 63–75 years) were similar. Dexamethasone (range, 9.6% to 35.5%) and remdesivir (range, 6.3% to 23.9%) were commonly prescribed as concomitant pharmacological treatments during the acute phase of COVID-19. CoronaVac is the most common COVID-19 vaccine type (range, 50.0% to 96.6%) in each of the four exposure groups. After propensity score weighting, the SMDs of covariates were all < 0.1, suggesting a good covariate balance between groups after weighting (Supplement Figure 1).
Table 1.
Characteristics of the study population by vaccination status and use of nirmatrelvir–ritonavir before weighting.
| No use of nirmatrelvir–ritonavir (n = 21068) | Use of nirmatrelvir–ritonavir (n = 16647) | |||
|---|---|---|---|---|
| Not fully vaccinated (n = 10976) | Fully vaccinateda (n = 10092) | Not fully vaccinated (n = 4982) | Fully vaccinated (n = 11665) | |
| Age, years | 74 ± 19 | 63 ± 21 | 75 ± 15 | 73 ± 15 |
| Sex | ||||
| Female | 5589 (50.9) | 5174 (51.3) | 2545 (51.1) | 5514 (47.3) |
| Male | 5387 (49.1) | 4918 (48.7) | 2437 (48.9) | 6151 (52.7) |
| Charlson Comorbidity Index | 0 (0–1) | 0 (0–0) | 0 (0–1) | 0 (0–0) |
| Concomitant pharmacological treatments | ||||
| Dexamethasone | 3895 (35.5) | 2519 (25.0) | 770 (15.5) | 1117 (9.6) |
| Methylprednisolone | 15 (0.1) | 16 (0.2) | 3 (0.1) | 4 (0.0) |
| Prednisolone | 815 (7.4) | 624 (6.2) | 235 (4.7) | 428 (3.7) |
| Interferon-beta-1b | 84 (0.8) | 7 (0.1) | 8 (0.2) | 1 (0.0) |
| Baricitinib | 225 (2.0) | 189 (1.9) | 69 (1.4) | 85 (0.7) |
| Tocilizumab | 120 (1.1) | 96 (1.0) | 16 (0.3) | 23 (0.2) |
| Remdesivir | 2206 (20.1) | 2414 (23.9) | 487 (9.8) | 738 (6.3) |
| Concomitant nonpharmacological treatments | ||||
| Intensive care unit admission | 354 (3.2) | 399 (4.0) | 63 (1.3) | 139 (1.2) |
| Use of ventilatory support | 276 (2.5) | 218 (2.2) | 43 (0.9) | 79 (0.7) |
| Vaccine type | ||||
| CoronaVac | 4869 (44.4) | 5047 (50.0) | 2484 (49.9) | 7543 (64.7) |
| Comirnaty | 358 (3.3) | 4339 (43.0) | 77 (1.5) | 3258 (27.9) |
| CoronaVac + Comirnaty | 9 (0.1) | 706 (7.0) | 10 (0.2) | 864 (7.4) |
| No vaccineb | 5740 (52.3) | 0 (0.0) | 2411 (48.4) | 0 (0.0) |
| Time since the last dose of vaccine | ||||
| ≤6 months | 4378 (39.9) | 6380 (63.2) | 2116 (42.5) | 7831 (67.1) |
| >6 months | 6598 (60.1) | 3712 (36.8) | 2866 (57.5) | 3834 (32.9) |
For those who received CoronaVac alone, individuals who received ≥3 doses were considered fully vaccinated, while those who received <3 doses were categorized as not fully vaccinated. For recipients of Comirnaty alone, individuals who received ≥2 doses were classified as fully vaccinated, while those who received <2 doses were considered not fully vaccinated.
No vaccine: Individuals who had received 0 doses of vaccination at least 14 days before their positive PCR date or had no vaccination records.
Data are mean ± SD, n (%), or median (first quartile – third quartile).
There was no violation of the proportional hazards assumption (Supplement Figure 2). Cox regression models stratified by vaccination status indicated that, compared with patients who were not prescribed nirmatrelvir–ritonavir, those who received nirmatrelvir–ritonavir had significantly reduced risks of post-acute death in both fully vaccinated (hazard ratio [HR], 0.51; 95% confidence intervals [CI], 0.45–0.57; p-value, < 0.0001) and not fully vaccinated individuals (HR, 0.62; 95% CI, 0.56–0.69; p-value, < 0.0001) (Table 2). Additionally, Cox regression models stratified by nirmatrelvir–ritonavir use showed that complete vaccination had significant protective effect on post-acute death in both groups, regardless of whether nirmatrelvir–ritonavir was used.
Table 2.
Effect of nirmatrelvir–ritonavir use and vaccination status on post-acute death and rehospitalization.
| Fully vaccinated | Use of nirmatrelvir–ritonavir | Effect of nirmatrelvir–ritonavir a | ||||||
|---|---|---|---|---|---|---|---|---|
| No | Yes | |||||||
| Event/At risk (%) | HR (95% CI) | p | Event/At risk (%) | HR (95% CI) | p | HR (95% CI) | p | |
| Death | ||||||||
| No | 2087/10976 (19.0) | Reference | – | 519/4982 (10.4) | 0.59 (0.53, 0.66) | <0.0001 | 0.62 (0.56, 0.69) | <0.0001 |
| Yes | 874/10092 (8.7) | 0.67 (0.60, 0.75) | <0.0001 | 614/11665 (5.3) | 0.36 (0.33, 0.40) | <0.0001 | 0.51 (0.45, 0.57) | <0.0001 |
| Effect of vaccination b | 0.72 (0.66, 0.79) | <0.0001 | 0.63 (0.56, 0.71) | <0.0001 | ||||
| Rehospitalization | ||||||||
| No | 6197/10976 (56.5) | Reference | – | 2350/4982 (47.2) | 0.72 (0.69, 0.76) | <0.0001 | 0.74 (0.71, 0.78) | <0.0001 |
| Yes | 4574/10092 (45.3) | 0.87 (0.83, 0.92) | <0.0001 | 4859/11665 (41.7) | 0.67 (0.64, 0.69) | <0.0001 | 0.74 (0.71, 0.78) | <0.0001 |
| Effect of vaccination b | 0.94 (0.89, 0.98) | 0.0046 | 0.93 (0.88, 0.98) | 0.0050 | ||||
The effect of nirmatrelvir–ritonavir within different vaccination strata, which was assessed using newly calculated weights specific to each stratum. The new propensity scores were calculated using the use of nirmatrelvir–ritonavir as the independent variable.
The effect of vaccination within different nirmatrelvir–ritonavir strata, which was assessed using newly calculated weights specific to each stratum. The new propensity scores were calculated using vaccination status as the independent variable.
HR: hazard ratio.
Compared with patients who neither used nirmatrelvir–ritonavir nor were fully vaccinated (i.e. reference group), patients who used nirmatrelvir–ritonavir and were fully vaccinated had a significantly lower risk of post-acute mortality (HR, 0.36; 95% CI, 0.33–0.40; p-value, < 0.0001). This risk reduction was greater than that observed in patients who were fully vaccinated but did not use nirmatrelvir–ritonavir (HR, 0.67; 95% CI, 0.60–0.75; p-value, < 0.0001), and in those who used nirmatrelvir–ritonavir but were not fully vaccinated (HR, 0.59; 95% CI, 0.53–0.66; p-value, < 0.0001) (Table 2 and Figure 2). The additive interaction statistics between nirmatrelvir–ritonavir and vaccination were marginally significant (RERI, 0.28; 95% CI, −0.0010–0.55; p-value, 0.051; AP, 0.10; 95% CI, 0.0007–0.20; p-value, 0.048; S, 1.19; 95% CI, 0.99–1.43; p-value, 0.070).
Figure 2.
Cumulative incidence curve. A for post-acute inpatient mortality; B for post-acute rehospitalization. The shaded area indicates the 95% CI. Multiplicative interaction between nirmatrelvir–ritonavir and vaccination status was evaluated by examining the coefficient of the product term in the Cox model, represented as the HR in the plot. Additive interaction was assessed using RERI, AP and S. These indices were analysed by reversing the values of both the vaccination status and the nirmatrelvir–ritonavir, due to the observed protective effect in the original hazard ratios. Therefore, it’s important to interpret these results with caution. N/R: nirmatrelvir–ritonavir
Similar interaction effects on post-acute rehospitalization were observed (Table 2 and Figure 2). Patients with nirmatrelvir–ritonavir + fully vaccinated had a significantly lower risk of post-acute rehospitalization (HR, 0.67; 95% CI, 0.64–0.69; p-value, < 0.0001) compared with patients who did not use nirmatrelvir–ritonavir and were not fully vaccinated. This risk was also lower than that in the groups who were fully vaccinated but did not use nirmatrelvir–ritonavir (HR, 0.87; 95% CI, 0.83–0.92; p-value, < 0.0001), and those that used nirmatrelvir–ritonavir but were not fully vaccinated (HR, 0.72; 95% CI, 0.69–0.76; p-value, < 0.0001). Significant additive interaction on post-acute rehospitalization was observed (RERI, 0.10; 95% CI, 0.02–0.19; p-value, 0.018; AP, 0.07; 95% CI, 0.01–0.12; p-value, 0.017; S, 1.26; 95% CI, 1.02–1.55; p-value, 0.032).
In most of the subgroups, lower HRs were observed in patients receiving nirmatrelvir–ritonavir plus full vaccination compared to those not receiving both interventions (Supplement Tables 2–5 and Supplement Figures 3–6). Among the subgroups, the additive interaction on rehospitalization was more pronounced in patients aged > 65 years (RERI, 0.13; 95% CI, 0.02–0.23; p-value, 0.018; AP, 0.08; 95% CI, 0.01–0.15; p-value, 0.016; S, 1.28; 95% CI, 1.02–1.60; p-value, 0.031); females (RERI, 0.18; 95% CI, 0.06–0.29; p-value, 0.0033; AP, 0.12; 95% CI, 0.04–0.20; p-value, 0.0030; S, 1.59; 95% CI, 1.08–2.35; p-value, 0.020), and CoronaVac recipients (RERI, 0.15; 95% CI, 0.03–0.27; p-value, 0.012; AP, 0.10; 95% CI, 0.02–0.18; p-value, 0.010; S, 1.43; 95% CI, 1.04–1.97; p-value, 0.029). The results of the sensitivity analysis confirmed the robustness of our primary findings (Supplement Figure 7–9). No significant interaction on the negative control outcome was found, reassuring the robustness of the control of biases (Supplement Figure 10).
Discussion
While previous studies have often explored the interaction effect between antivirals and vaccination on COVID-19 outcomes through subgroup analysis, none of them were specifically designed to examine the interaction. This retrospective cohort study examined the interaction between complete vaccination for COVID-19 and the prescription of nirmatrelvir–ritonavir during an acute phase of SARS-CoV-2 infection on post-acute mortality and rehospitalization in Hong Kong, where almost all the infected cases experienced their first infection during the Omicron Outbreak. Our results primarily demonstrated a combined effect between nirmatrelvir–ritonavir and vaccination in relation to post-acute rehospitalization. With a positive additive interaction, patients who were not fully vaccinated and were not prescribed with nirmatrelvir–ritonavir had a synergistic increase in risk. The interaction analysis likely supports the idea that individuals without sufficient protection from vaccination may experience an increased risk of COVID-19 severity during the acute phase – as a result, they may not be able to receive the fully mitigating effect of nirmatrelvir–ritonavir, considering the relationship between acute disease severity and the onset of post-acute sequelae [28]. We acknowledge that the additive interaction on post-acute mortality was marginally significant. However, we speculate that the clinical benefit is warranted, based on the observed effect size and the consistency with the results of post-acute rehospitalization.
Among the studies evaluating the effect of nirmatrelvir–ritonavir on post COVID-19 conditions, only Xie et al [6]. and Ioannou et al [8]. have explored the modification effect of COVID-19 vaccine using subgroup analysis. According to the findings of Xie et. al., the reduced risk of post-COVID-19 conditions in COVID-19 patients was similar across different vaccination statuses, including those who were unvaccinated, vaccinated with one or two doses, and vaccinated with a boosted dose. On the contrary, our study findings were consistent to Ioannou et al. [8] identifying a reduced risk of several post-COVID-19 conditions (e.g. thromboembolic events) among recipients of nirmatrelvir–ritonavir who were vaccinated, compared to those who were unvaccinated. It should be noted that we used patients admitted to the hospital with COVID-19 as the study population, who generally had moderate-to-severe COVID-19. In comparison to the mild non-hospitalized populations studied by Xie et al. and Ioannou et al., our study population, particularly those who did not receive nirmatrelvir–ritonavir and vaccination, may be more prone to experience post-COVID-19 conditions due to the association with acute severity [28]. The higher event rate in the hospitalized population provided us more statistical power to detect a significant combined effect on post-acute outcomes.
Our subgroup analysis showed that the interaction effect is more pronounced in older adults, female, and CoronaVac recipients While females were suggested to have stronger immune responses as well as a potentially higher vaccine effectiveness [29], Bai et al. indicated that female was more likely to develop long COVID syndrome [30]. Similarly, older adults were suggested having worse COVID-19 outcomes than younger individuals after an acute infection [31]. The finding may thus support females and elderly were more prone to exhibit a synergistic increase in risk when they were not fully vaccinated and not prescribed with nirmatrelvir–ritonavir, compared to males. Similarly, CoronaVac has been reported to have lower effectiveness against infection and a higher waning rate compared to Comirnaty [32]. Nevertheless, the effectiveness of CoronaVac against acute severe outcome was reported as promising, although the finding was not as consistently reported in the literature. For example, it was shown that CoronaVac recipients had a higher risk of hospitalization or COVID death than Comirnaty after at least 6 months post-vaccination [33]. As our subgroup analyses divided the samples into smaller groups, the power to detect statistical significance in certain subgroups decreased, particularly in subgroups with fewer events. Therefore, careful interpretation of these findings is warranted.
Despite different study outcomes, we observed an insignificant multiplicative interaction, consistent with the findings of Cheung et. al [20]. identifying insignificant interaction between oral antiviral drugs and vaccinations against acute COVID-19 progression in a similar setting. It is worth noting that studies may often lack sufficient statistical power to detect the significance of a multiplicative interaction between two exposures. In our study, following the suggestion of Mansournia & Nazemipour [25], we assessed the additive interactions, which may be more relevant for determining the complementary or overlapping effects between the two pharmaceutical interventions [34]. In fact, epidemiological studies have widely agreed that assessing interactions on an additive scale is a more appropriate method for evaluating the public health significance of interactions [26]. Including both interaction analyses not only enhances the robustness of the analysis but also improves the understanding of how the exposures interact and affect the outcomes.
The primary strength of this investigation is that all study subjects experienced their first infection during the Omicron epidemic, minimizing the possibility of the vaccination effect being influenced by protection from previous natural infections. Due to stringent control measures in place in Hong Kong, less than 1% of the population had been infected prior to the onset of the Omicron outbreak [13, 14]. With employing this infection-naïve population, the effect size of vaccination is unlikely to be distorted, which may also explain why we observed a significant interaction effect between the vaccine and nirmatrelvir–ritonavir. Another strength of our study is the representativeness of hospitalization data, that were collected from all public hospitals in Hong Kong, accounting for approximately 90% of the hospitalization services in Hong Kong. The clinical records of patients were all digitally documented in each of the hospitals, reassuring the validity of disease diagnosis. In addition, nirmatrelvir–ritonavir was prescribed to inpatients with close monitoring of medication uptake, ensuring compliance with the prescribed treatment regimen.
Our study has several limitations. Firstly, although our study predominantly included infections caused by Omicron sub-lineages BA.2 and BA.5, a small proportion of patients infected with XBB was also included due to a shift of dominant sub-lineage shifted to XBB after April, 2023. Secondly, as this is an observational study, we acknowledge the presence of residual confounding due to limited data. Some unmeasured clinical and behaviour confounders, such as smoking behaviour and COVID-19 symptoms during acute phase of infection, were not accounted for in the analysis. Thirdly, we categorized the vaccination status into two groups: fully vaccinated and non-fully vaccinated, instead of considering the number of doses, in order to mitigate sparse data bias. However, this binary grouping does not allow the analysis of the dose–response relationship in terms of the number of doses. Fourthly, we restricted the infection-naïve population to those with their first recorded infection of SARS-CoV-2 during Omicron epidemic. While there were very limited number of infected cases reported before the Omicron outbreak, we acknowledged that the surveillance system may miss a small number of unidentified infections [13]. Lastly, we acknowledge the presence of immortal bias in the observational design, especially without a targeted trial emulation. In accordance with the COVID-19 patient management guidelines in Hong Kong, nirmatrelvir–ritonavir was strongly recommended to be prescribed within five days of symptom onset. Among the patients prescribed nirmatrelvir–ritonavir in our study, the majority initiated antiviral treatment on or before hospital admission, with the median number of days between nirmatrelvir–ritonavir prescription and the positive PCR date being 0 days (interquartile range, 0–1 d). Therefore, we believe that the immortal time bias resulting from events occurring within a short period prior to treatment initiation may not significantly impact our main findings.
In conclusion, our study demonstrated an additive interaction between nirmatrelvir–ritonavir and complete vaccination on post-acute rehospitalization, suggesting that nirmatrelvir–ritonavir would provide greater long-term benefits to fully vaccinated individuals compared to those who are not fully vaccinated. Further experimental investigations are warranted to assess the efficacy of the combined pharmaceutical interventions in order to inform the development of optimized therapeutic strategies.
Contributions
Study design and conceptualization: HW, GL, YWei, CB, KCC. Data collection and pre-processing: HW, YWei, CHKY, TYC, ZG, EKY. Data analysis and interpretation: HW, GL, YWei, KCC. Writing – Original Draft: HW, GL, CTH, KMJ, XJ, CL, SZ, CKPM, DSCH, KCC. Writing – Review and Editing: CB, CHKY, TYC, KL, AY, EKY. EKY and KCC have accessed and verified all the data. All authors critically reviewed the manuscript and gave final approval for publication.
Supplementary Material
Funding Statement
This research was funded by Health and Medical Research Fund [grant numbers COVID190105, COVID19F03, INF-CUHK-1, COVID1903003], RGC Collaborative Research Fund [grant numberC6036-21GF], and RGC theme-based research schemes [grant numberT11-705/21-N]. The funders of the study had no role in study design, data collection, data analysis, data interpretation, writing of the manuscript, or the decision to submit for publication.
Data availability statement
The cases’ surveillance data and medication records were extracted from electronic records in the system managed by the Hong Kong Hospital Authority. The vaccine history was extracted from the COVID-19 surveillance database provided by the Department of Health in Hong Kong. Restrictions apply to the availability of these data.
Declarations of interest
We declare that we have no conflicts of interest.
Ethics approval statement
Ethics approval was obtained from the Joint CUHK-NTEC Clinical Research Ethics Committee (Ref No. 2023.006). As this study was a retrospective analysis using secondary data without any personal information, the requirement for obtaining informed consent was waived.
Acknowledgements
We thank Hospital Authority and Department of Health, Hong Kong Government for providing the data for this study. The Centre for Health Systems and Policy Research funded by the Tung Foundation is acknowledged for the support throughout the conduct of this study.
Disclosure statement
No potential conflict of interest was reported by the author(s).
References
- 1.Hastie CE, Lowe DJ, McAuley A, et al. True prevalence of long-COVID in a nationwide, population cohort study. Nat Commun. 2023;14(1):7892. doi: 10.1038/s41467-023-43661-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.US FDA . Fact sheet for healthcare providers: emergency use authorization for paxlovid. [cited 2023 Dec 28] Available from: https://www.fda.gov/media/155050/download.
- 3.Hammond J, Leister-Tebbe H, Gardner A, et al. Oral Nirmatrelvir for high-risk, nonhospitalized adults with Covid-19. N Engl J Med. 2022;386(15):1397–1408. doi: 10.1056/NEJMoa2118542 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Arbel R, Wolff Sagy Y, Hoshen M, et al. Nirmatrelvir use and severe Covid-19 outcomes during the omicron surge. N Engl J Med. 2022;387(9):790–798. doi: 10.1056/NEJMoa2204919 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Wong CKH, Au ICH, Lau KTK, et al. Real-world effectiveness of early molnupiravir or nirmatrelvir-ritonavir in hospitalised patients with COVID-19 without supplemental oxygen requirement on admission during Hong Kong's omicron BA.2 wave: a retrospective cohort study. Lancet Infect Dis. 2022;22(12):1681–1693. doi: 10.1016/S1473-3099(22)00507-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Xie Y, Choi T, Al-Aly Z.. Association of treatment with nirmatrelvir and the risk of post-COVID-19 Condition. JAMA Intern Med. 2023;183(6):554–564. doi: 10.1001/jamainternmed.2023.0743 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Wang H, Wei Y, Hung CT, et al. Association of nirmatrelvir-ritonavir with post-acute sequelae and mortality in patients admitted to hospital with COVID-19: a retrospective cohort study. Lancet Infect Dis. 2024 Oct;24(10):1130–1140. doi: 10.1016/S1473-3099(24)00217-2 [DOI] [PubMed] [Google Scholar]
- 8.Ioannou GN, Berry K, Rajeevan N, et al. Effectiveness of nirmatrelvir-ritonavir against the development of Post-COVID-19 conditions among U.S. veterans: a target trial emulation. Ann Intern Med. 2023;176(11):1486–1497. doi: 10.7326/M23-1394 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Durstenfeld MS, Peluso MJ, Lin F, et al. Association of nirmatrelvir for acute SARS-CoV-2 infection with subsequent Long COVID symptoms in an observational cohort study. J Med Virol. 2024;96(1):e29333. doi: 10.1002/jmv.29333 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Huh K, Kim YE, Bae GH, et al. Vaccination and the risk of post-acute sequelae after COVID-19 in the omicron-predominant period. Clin Microbiol Infect. 2024;30(5):666–673. doi: 10.1016/j.cmi.2024.01.028 [DOI] [PubMed] [Google Scholar]
- 11.Wynberg E, Han AX, Boyd A, et al. The effect of SARS-CoV-2 vaccination on post-acute sequelae of COVID-19 (PASC): a prospective cohort study. Vaccine. 2022;40(32):4424–4431. doi: 10.1016/j.vaccine.2022.05.090 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.First shipment of COVID-19 oral drug Paxlovid distributed to HA for application. [cited 2023 Dec 8] Available from: https://www.info.gov.hk/gia/general/202203/15/P2022031500280.htm.
- 13.Boon SS, Wong MCS, Ng RWY, et al. Seroprevalence of unidentified SARS-CoV-2 infection in Hong Kong during 3 pandemic waves. JAMA Netw Open. 2021;4(11):e2132923. doi: 10.1001/jamanetworkopen.2021.32923 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Chong KC, Chan PK, Hung CT, et al. Changes in all-cause and cause-specific excess mortality before and after the Omicron outbreak of COVID-19 in Hong Kong. J Glob Health. 2023;13:06017. doi: 10.7189/jogh.13.06017 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Lam ICH, Wong CKH, Zhang R, et al. Long-term post-acute sequelae of COVID-19 infection: a retrospective, multi-database cohort study in Hong Kong and the UK. EClinicalMedicine. 2023;60:102000. doi: 10.1016/j.eclinm.2023.102000 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Daugherty SE, Guo Y, Heath K, et al. Risk of clinical sequelae after the acute phase of SARS-CoV-2 infection: retrospective cohort study. Br Med J. 2021;373:n1098. doi: 10.1136/bmj.n1098 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Wan EYF, Yan VKC, Mok AHY, et al. Effectiveness of molnupiravir and nirmatrelvir-ritonavir in hospitalized patients with COVID-19: a target trial emulation study. Ann Intern Med. 2023;176(4):505–514. doi: 10.7326/M22-3057 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.HKHA . Interim recommendation on clinical management of adult cases with coronavirus Disease 2019 (COVID-19). [cited 2023 Dec 28] Available from: https://icidportal.ha.org.hk/Home/File?path = /Training%20Calendar/149/Interim%20Recommendation%20on%20Clinical%20Management%20of%20Adult%20Cases%20with%20COVID-19.pdf.
- 19.Yang B, Lin Y, Xiong W, et al. Comparison of control and transmission of COVID-19 across epidemic waves in Hong Kong: an observational study. Lancet Reg Health West Pac. 2023 Nov 22;43:100969. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Cheung YYH, Lau EHY, Yin G, et al. Effectiveness of vaccines and antiviral drugs in preventing severe and fatal COVID-19, Hong Kong. Emerg Infect Dis. 2024;30(1):70–78. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Català M, Burn E, Rathod-Mistry T, et al. Observational methods for COVID-19 vaccine effectiveness research: an empirical evaluation and target trial emulation. Int J Epidemiol. 2024 Feb 1;53(1):dyad138. doi: 10.1093/ije/dyad138 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.HKHA . Strategic service framework for palliative care. [cited 2024 Feb 21] Available from: https://www.ha.org.hk/haho/ho/ap/PCSSF_1.pdf.
- 23.Austin PC. Using the standardized difference to compare the prevalence of a binary variable between two groups in observational research. Commun Stat - Simul Comput. 2009;38(6):1228–1234. doi: 10.1080/03610910902859574 [DOI] [Google Scholar]
- 24.Lipsitch M, Tchetgen Tchetgen E, Cohen T.. Negative controls: a tool for detecting confounding and bias in observational studies. Epidemiology. 2010 May;21(3):383–388. doi: 10.1097/EDE.0b013e3181d61eeb [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Mansournia MA, Nazemipour M.. Recommendations for accurate reporting in medical research statistics. Lancet. 2024;403(10427):611–612. doi: 10.1016/S0140-6736(24)00139-9 [DOI] [PubMed] [Google Scholar]
- 26.Rothman KJ, Greenland S, Lash TL. Modern epidemiology: Wolters Kluwer Health/Lippincott Williams & Wilkins Philadelphia; 2008.
- 27.Knol MJ, VanderWeele TJ, Groenwold RH, et al. Estimating measures of interaction on an additive scale for preventive exposures. Eur J Epidemiol. 2011;26(6):433–438. doi: 10.1007/s10654-011-9554-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Sebők S, Gyires K.. Long COVID and possible preventive options. Inflammopharmacology. 2023;31(6):2807–2817. doi: 10.1007/s10787-023-01204-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Vijayasingham L, Bischof E, Wolfe J.. Sex-disaggregated data in COVID-19 vaccine trials. Lancet. 2021;397(10278):966–967. doi: 10.1016/S0140-6736(21)00384-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Bai F, Tomasoni D, Falcinella C, et al. Female gender is associated with long COVID syndrome: a prospective cohort study. Clin Microbiol Infect. 2022;28(4):611, e9-.e16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Cohen K, Ren S, Heath K, et al. Risk of persistent and new clinical sequelae among adults aged 65 years and older during the post-acute phase of SARS-CoV-2 infection: retrospective cohort study. Br Med J. 2022;376:e068414. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Lau JJ, Cheng SMS, Leung K, et al. Real-world COVID-19 vaccine effectiveness against the Omicron BA.2 variant in a SARS-CoV-2 infection-naive population. Nat Med. 2023;29(2):348–357. doi: 10.1038/s41591-023-02219-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Wei Y, Jia KM, Zhao S, et al. Estimation of vaccine effectiveness of CoronaVac and BNT162b2 against severe outcomes over time among patients with SARS-CoV-2 Omicron. JAMA Netw Open. 2023;6(2):e2254777. doi: 10.1001/jamanetworkopen.2022.54777 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Knol MJ, VanderWeele TJ.. Recommendations for presenting analyses of effect modification and interaction. Int J Epidemiol. 2012;41(2):514–520. doi: 10.1093/ije/dyr218 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The cases’ surveillance data and medication records were extracted from electronic records in the system managed by the Hong Kong Hospital Authority. The vaccine history was extracted from the COVID-19 surveillance database provided by the Department of Health in Hong Kong. Restrictions apply to the availability of these data.