Oseltamivir drug‐disease modelling in infants with influenza virus infection Authors: Kun Wang, David W Kimberlin, Richard J Whitley, Edward P Acosta, For the National Institute of Allergy and Infectious Diseases Collaborative Antiviral Study Group Published: 2026 ABSTRACT AbstractAimsOseltamivir is a commonly used neuraminidase inhibitor for treatment of influenza infection, but little data exist regarding concentration‐response relationships.MethodsPharmacokinetic (PK) and viral dynamic data from an earlier NIAID Collaborative Antiviral Study Group trial in paediatric patients with influenza from 0–23 months of age were used to create a PK model linked to an influenza viral dynamic model. Population PK parameters were used to perform broad dose‐ranging simulations to determine the oseltamivir carboxylate exposure required to achieve 95% of maximum viral suppression (EC95). PK parameters of interest included the oseltamivir carboxylate steady‐state 12‐h area‐under‐the‐curve (AUC12) and 12‐h trough concentration (C12h). AUC12 and C12h results were related to the time‐averaged area under the log10 viral load‐time curve from Day 0 to Day 5 minus the baseline viral load (AAUCMB) using a maximum effect model.ResultsA total of 373 plasma samples for oseltamivir and 382 plasma samples for oseltamivir carboxylate from 78 infants were included. The median age and weight at the time of PK sampling were 8 months and 8 kg, respectively. Sixty‐two babies with measured influenza viral loads over the course of treatment (12 days, n = 239) were included in the viral dynamic modelling. The modelled EC95 for AUC12 and C12h were 2710 ngxh/mL and 179 ng/mL, respectively.ConclusionsPopulation pharmacokinetic modelling can be directly linked to viral dynamic outcomes and through simulations, dose‐ranging studies performed to identify target drug exposures. This approach could be used as a surrogate when concentration‐response data do not exist.Keywords: influenza, modelling, oseltamivir, pediatrics, pharmacodynamics, pharmacokinetics What is already known about this subject Standard dosing for influenza treatment generally attains necessary oseltamivir carboxylate exposures in adults. Standard dosing for influenza treatment generally attains necessary oseltamivir carboxylate exposures in adults. Oseltamivir carboxylate concentration‐response relationships have not been identified, particularly in the paediatric population. Oseltamivir carboxylate concentration‐response relationships have not been identified, particularly in the paediatric population. What this study adds We defined oseltamivir carboxylate concentrations required to suppress influenza virus in infants using pharmacokinetic and viral dynamic modelling and simulation methods. We defined oseltamivir carboxylate concentrations required to suppress influenza virus in infants using pharmacokinetic and viral dynamic modelling and simulation methods. Once daily oseltamivir dosing for prevention does not achieve adequate exposure for asymptomatic patients with existing viral titres. Once daily oseltamivir dosing for prevention does not achieve adequate exposure for asymptomatic patients with existing viral titres. Drug‐disease modelling can be used to identify target drug exposures. Drug‐disease modelling can be used to identify target drug exposures. INTRODUCTION Oseltamivir and zanamivir are neuraminidase inhibitor antiviral drugs recommended for treatment and prophylaxis of influenza infection in children and adults. 1 , 2 Oseltamivir phosphate is an orally administered prodrug that is converted to its active metabolite oseltamivir carboxylate predominantly via hepatic esterases. 3 Oseltamivir carboxylate inhibits viral neuraminidase, thus blocking release of progeny virions from infected cells and preventing viral entry into uninfected cells. 4 Phase II dose‐ranging trials are a critical component during drug development. Typically, multiple ascending doses are examined in the target population in order to establish short‐term safety, steady‐state pharmacokinetics and concentration‐response relationships, all of which assist in selection of dose(s) for subsequent Phase III trials. Multiple oseltamivir pharmacokinetic studies have been conducted by the manufacturer, 5 but only one was a dose‐ranging Phase II treatment study utilizing a healthy volunteer inoculation design and examining doses ranging from 20 to 200 mg twice daily (bid). 6 Only trough concentrations were collected for pharmacokinetics, and no dose or concentration‐response relationships were found due to the relatively narrow range of doses studied. Consequently, the pharmacokinetic parameter(s) related to treatment response and the target drug exposure for oseltamivir carboxylate to treat susceptible virus remain incompletely defined. Data from this and another Phase II study 7 have been combined in a larger pharmacokinetic‐pharmacodynamic analysis. 8 This study concluded that oseltamivir carboxylate exposure‐response relationships exist, particularly between the 24‐h area‐under‐the‐curve (AUC24) and a composite symptom score, and oseltamivir carboxylate exposures beyond those achieved with the approved regimen of 75‐mg bid may provide enhanced efficacy. Dose‐finding studies of influenza antiviral drugs are challenging due to the difficult nature of establishing concentration (or toxicity)‐response relationships, the short duration of antiviral therapy, and variability in both pharmacokinetics and the multitude of potential determinants of response. This is even more challenging in paediatrics due to their continually developing metabolic profile and the challenges of enrolling in paediatric interventional studies. Advances in modelling and simulation techniques have aided in addressing some of these challenges. Our goal in this analysis was to define the oseltamivir carboxylate exposure required for maximal virus suppression in paediatric patients using an integrated pharmacokinetic and viral dynamic drug‐disease model. Data from the NIAID Collaborative Antiviral Study Group (CASG) investigation of oseltamivir in infants (CASG 114) were utilized. 9 METHODS Population pharmacokinetic and pharmacodynamic modelling The overall approach to accomplish our objective was (1) to apply a population pharmacokinetic model to the oseltamivir and oseltamivir carboxylate concentration‐time data, (2) to develop a viral dynamic model to describe changes in influenza virus clearance upon initiation of therapy and (3) to link these models together using population parameter estimates and their variability in order to perform a large dose‐ranging simulation study to determine the optimal oseltamivir carboxylate exposure required to achieve 95% of the maximum effect (EC95). Study design, data collection, laboratory methods and participant demographics have been previously reported. 9 Briefly, CASG 114 was an oseltamivir carboxylate area‐under‐the‐curve (AUC) targeted pharmacokinetic study design in infants and toddlers from birth through 23 months of age. Five dosing cohorts were included, all of which received antiviral therapy for 5 days: Cohort IA (ages 1–2 years) received the approved oseltamivir fixed dose of 30‐mg bid (n = 10) for that age range; Cohort IB (Ages 1–2 years) received 3.5‐mg/kg bid; Cohorts IIA (n = 6) and IIB (n = 7) (both 9 through 11 months of age) received 3.0 and 3.5‐mg/kg bid, respectively; and Cohorts III (n = 22) (6 through 8 months), IV (n = 10) (3 through 5 months) and V (n = 19) (birth through 2 months) all received 3.0‐mg/kg bid. Oseltamivir and oseltamivir carboxylate concentrations for pharmacokinetic analysis were obtained at steady state on treatment Day 3 at 0 h (predose), 1 h and between 2–3, 5–7 and 10–12 h post dose. Nasopharyngeal swabs for influenza virus quantitation were scheduled to be obtained on study Days 1, 3, 5 and 10. Oseltamivir and oseltamivir carboxylate pharmacokinetic model The overall integrated pharmacokinetic‐pharmacodynamic model is depicted in Figure 1. Pharmacokinetic modelling was accomplished using a nonlinear mixed‐effects approach to establish a combined parent‐metabolite model. 10 A two‐compartment model for oseltamivir concentrations was applied and linked to a one compartment model for oseltamivir carboxylate with two conversion delays. Absorption, metabolite formation and clearance processes were assumed to be linear. Differential equations for the oseltamivir and oseltamivir carboxylate model are listed below: Structure of the final integrated pharmacokinetic‐pharmacodynamic model defined in Equations (1)–(10). A1—A3 are the drug amount of oseltamivir in the measured (central), absorption compartment and peripheral compartment, respectively. A4–A6 represent the amounts of oseltamivir carboxylate in two delay compartments and the measured oseltamivir carboxylate compartment, respectively. K a is the oseltamivir absorption rate constant. K cp and K pc are the intercompartmental rate constants between the central and peripheral compartments. K e is the oseltamivir elimination rate constant. V c is the central compartment of measured oseltamivir. τ is the transit rate constant. CL m and V m are the clearance and distribution volume of oseltamivir carboxylate, respectively. D 1 and D 2 are the signals in the delay compartments. T 1 is the transit time between delay compartments. ε is the delayed drug effect. v is the viral load, and λ is the rate of increase in virion production. where A 1–A 3 are the amount of oseltamivir in the central, absorption and peripheral compartments, respectively. A 4–A 6 represent the amounts of oseltamivir carboxylate in two sequential conversion delay compartments and the measured oseltamivir carboxylate compartment, respectively. K a is the oseltamivir absorption rate constant, K cp and K pc are the intercompartmental rate constants, and K e is the elimination rate constant. Oseltamivir concentration C p = A 1/V c, where V c is the central compartment of measured oseltamivir. Tau (τ) is the transit rate constant. CL m and V m are the clearance and distribution volume for oseltamivir carboxylate, respectively. Throughout the model, V c, CL m and V m terms represent the oseltamivir central compartment distribution volume (V c/F), and the apparent clearance (CL m/F m) and distribution volume (V m/F m) for the carboxylate metabolite. Concentration‐time data were modelled using ADAPT 5 (MLEM module). 11 , 12 , 13 A log‐normal distribution was used for describing interindividual variability. The proportional, additive and proportional combined additive random effects models were also investigated to describe the residual variability. Any concentration results below the limit of quantitation (BLQ) were estimated using maximum likelihood. Evaluation of the impact of covariates (age, weight and body surface area) was performed using standard forward selection followed by backward elimination processes. Models were compared using goodness of fit plots and the change in −2 log‐likelihood (−2LL). A decrease of more than 6.63 corresponds to a significance level of 0.01. Models were also evaluated by visual predictive check (VPC) methods. One thousand Monte Carlo simulated oseltamivir and oseltamivir carboxylate concentration‐time profiles, using the final model population means and covariance matrix obtained from the ADAPT 5 output, were reproduced in NONMEM (Version 7.2; GloboMax LLC, Hanover, MD) (1) with Perl speaks NONMEM 3.4.2. VPCs were created using the Xpose 4 package of R and all the figures were plotted using R 2.15.3 (http://www.r-project.org/). The 95% confidence interval (CI) of the 2.5th and 97.5th percentiles of the simulated data for oseltamivir and oseltamivir carboxylate were compared with the observed data and used to assess the reliability of the final model. Influenza viral dynamic modelling Four differential equations were used to describe the change in influenza viral load concentrations (Equations (7)–(10)). where D 1 and D 2 are the signals in the delay compartments and C p is the concentration of oseltamivir carboxylate. Imax is the maximum drug effect of oseltamivir carboxylate, whereas IC 50 is the concentration at one‐half Imax. T 1 is the transit time between delay compartments, ε is the delayed drug effect, v is the influenza viral load, λ is the increase rate of virion production λ=c1·1−v/vmax, v max is the maximum viral load, c 1 is the increase rate of virion production over natural death rate, and c is the maximum virion death rate. All viral load data (from baseline through Day 10) were used for fitting the viral dynamic model. Exposure‐response simulations Exposure‐response simulations were performed in ADAPT 5 using the typical values of the final pharmacokinetic‐pharmacodynamic model. For these simulations, a wide dose‐finding approach was implemented that ranged from 0.6 to 500‐mg bid (164 different doses). The primary pharmacokinetic parameters of interest were the oseltamivir carboxylate steady‐state AUC12 and 12‐h trough concentration (C12h). Therefore, at each dose simulated, a mean AUC12 and C12h was calculated. The AUC12 and C12h were then separately related to the pharmacodynamic response variable. The pharmacodynamic response variable was the time‐averaged area under the log10 viral load‐time curve from Day 0 to day 5 minus the baseline viral load (AAUCMB). To achieve this, modelled viral load responses were simulated from Day 0 through Day 5, and the AUC was determined using noncompartmental methods. The exposure‐response relationship of AUC12, C12h and decrease in AAUCMB was fitted by the following Emax model: where X represents the steady state AUC12 or C12h, E max is the maximum decrease in AAUCMB and EC 50 is the steady‐state AUC12 or C12h at one‐half Emax. RESULTS A total of 373 plasma samples for oseltamivir and 382 plasma samples for oseltamivir carboxylate from 78 infants receiving oseltamivir for influenza treatment were included. The median age and weight at the time of pharmacokinetic sampling were 8 months and 8 kg, respectively. Sixty‐two babies with measured influenza viral load over the course of treatment (12 days, n = 239) were included in the viral dynamic modelling. Final pharmacokinetic model parameters are depicted in Table 1. The two‐compartment parent model linked to a one compartment metabolite model using sequential delays adequately described the oseltamivir and oseltamivir carboxylate concentrations (Figures 2 and 3). A proportional model with fixed additive error was used to describe the residual variance. For oseltamivir carboxylate, body weight was a significant covariate on clearance and distribution volume, and age was identified as significant covariate on the absorption rate constant (ka). The −2LL of the base model was −662.7. After adding weight on CLm, the −2LL decreased to −739.2 (Δ‐2LL = 76.5) and after adding weight on Vm, the −2LL further decreased to −803.0 (Δ‐2LL = 63.8). With age on τ, the −2LL decreased to −817.1 (Δ‐2LL = 13.1). The final model incorporating these three covariates is depicted in Equations (12)–(14): TABLE 1Population estimation results for the final combined oseltamivir and oseltamivir carboxylate pharmacokinetic and viral dynamic model.ParameterMean (%RSE)IIV CV% (%RSE) K e (1/h)0.653 (21.7)11.9 (95.9) V c (L)86.8 (22.5)52.1 (22.7) K a (1/h)0.878 (26.5)46.3 (48.3) K cp (1/h)0.226 (29.3)93.6 (30.5) K pc (1/h)0.0805 (51.6)22.9 (278) CL m (L/h)‐‐38.4 (8.49) CL m1 5.93 (6.94)‐‐ CL m2 0.151 (6.85)‐‐ V m (L)‐‐45.5(22.2) V m1 68.2 (11.7)‐‐ V m2 0.182 (5.51)‐‐ τ (1/h)‐‐5.25 (337) τ1 6.15 (22.0)‐‐ τ2 0.0949 (21.1)‐‐ IC 50 (ng/mL)7.02 (89.0)63.5 (138) T 1 (d)0.274 (23.3)120 (30.7) c (mL/d)3.11 (17.5)66.2 (23.5) c 1 (mL/d)0.515 (42.0)88.5 (37.8) V max (copies/mL)109 Fix Note: K a is the absorption rate constant of oseltamivir. K cp and K pc are the rate constants for the central plasma compartment to the peripheral compartment and the peripheral compartment to the central plasma compartment, respectively. K e is the elimination rate constant. V c is the central compartment of measured oseltamivir. τ is the transit rate constant. CL m and V m are the clearance and volume distribution of measured oseltamivir carboxylate compartment, respectively. T 1 is the transit time between delay compartments. ε is the delayed drug effect. v is the viral load. c and c 1 are the maximum death rate and increase rate of virion production over the natural death rate. CL m1 and V m1 are typical carboxylate clearance and distribution volume at a median weight of 8 kg. CL m2 and V m2 are the slope parameters of clearance and distribution volume. τ1 is the typical value of rate constant of delay compartment at age of 8 months. τ2 are the slope parameters of clearance and volume distribution. IC50 is the concentration of carboxylate at 50% inhibit rate. V max is the maximum virion produced. Population estimation results for the final combined oseltamivir and oseltamivir carboxylate pharmacokinetic and viral dynamic model. Note: K a is the absorption rate constant of oseltamivir. K cp and K pc are the rate constants for the central plasma compartment to the peripheral compartment and the peripheral compartment to the central plasma compartment, respectively. K e is the elimination rate constant. V c is the central compartment of measured oseltamivir. τ is the transit rate constant. CL m and V m are the clearance and volume distribution of measured oseltamivir carboxylate compartment, respectively. T 1 is the transit time between delay compartments. ε is the delayed drug effect. v is the viral load. c and c 1 are the maximum death rate and increase rate of virion production over the natural death rate. CL m1 and V m1 are typical carboxylate clearance and distribution volume at a median weight of 8 kg. CL m2 and V m2 are the slope parameters of clearance and distribution volume. τ1 is the typical value of rate constant of delay compartment at age of 8 months. τ2 are the slope parameters of clearance and volume distribution. IC50 is the concentration of carboxylate at 50% inhibit rate. V max is the maximum virion produced. (A) Goodness of fit of individual prediction plots for oseltamivir concentrations (left panel) and oseltamivir carboxylate (right panel). The symbols are the observed data, the dashed lines are the linear regression fit, and the black solid lines are the line of unity. (B) VPC of oseltamivir (left panel) and oseltamivir carboxylate (right panel). The middle solid line is the median, the dashed lines are the 2.5th and 97.5th percentiles of the observed data, and the shaded areas are the corresponding 95% confidence intervals of the simulated data. Viral dynamics of influenza in CASG 114. Open circles represent raw viral load results (n = 239), the solid line is the simulated average decay curve, and the dashed lines is the 95% confidence interval. where CL m1 and V m1 are typical clearance and distribution volume of oseltamivir carboxylate at the median weight of 8 kg. CL m2 and V m2 are the respective slope parameters for clearance and distribution volume. τ1 is the typical value of the delay rate constant at the median age of 8 months and τ 2 is the slope parameter. Weight was also a significant covariate on τ, but the −2LL decreased by 11.2 as opposed to 13.1 with age; therefore, the latter covariate was chosen. The influenza viral dynamic model parameters are also listed in Table 1. The I max and V max were fixed to 1 and 109 copies/mL, respectively. The best influenza viral load model fits are depicted in Figure 3. Simulated concentration‐response relationships between the viral load AAUCMB and oseltamivir carboxylate AUC12 and C12h are shown in Figure 4. From these relationships, the EC50, EC90 and EC95 for AUC12 were 140, 1280 and 2710 ngxh/mL, respectively. The same values for C12h were 9.42, 84.7 and 179 ng/mL, respectively. The exposure‐response relationship of AUC12 (left), trough concentration at 12 h (C12h) (right panel) and decrease in the time‐averaged area under the log10 viral load‐time curve from 0 to 5 days minus baseline (AAUCMB). Open circles are the simulated AAUCMB following multiple doses ranging from 0.6 to 500 mg twice daily (164 different doses). Solid line is the fitted Emax curve. Vertical dashed lines are the EC90 and EC95 as indicated. Box plots are the noncompartmental results for the five paediatric cohorts in CASG 114 with shaded circles representing individual AUC12 or C12h results. Solid line within the box plot is the median parameter for that cohort and the dashed lines represent the range. DISCUSSION The goal of these analyses was to determine the optimal oseltamivir carboxylate exposure required to suppress influenza viral replication, in essence providing a target exposure for future clinical trials or clinical use of the drug to treat susceptible strains. This was accomplished by creating a drug‐disease model, linking a parent‐metabolite population pharmacokinetic model to an influenza viral dynamic model and using the population parameter estimates to perform wide dose‐ranging simulations. Although it is not known which oseltamivir carboxylate pharmacokinetic parameter is most closely linked to influenza viral suppression or clinical outcomes, we chose to explore AUC12 and C12h as they are reasonably easily obtained metrics. Modelling and simulation results indicate that an AUC12 of 2710 ngxh/mL achieves the oseltamivir carboxylate exposure required for 95% of maximum viral suppression over 5 days of treatment. This EC95 reflects a near‐maximal antiviral effect at the population level, within the context of viral decay in which immune‐mediated clearance is implicitly incorporated. Although large interindividual variability was observed for some PK parameters (e.g., Kcp), this variability has minimal impact on clinically relevant exposure metrics such as AUC and trough concentrations. Paediatric patients in the CASG 114 study 9 receiving a weight‐based dose of 3.0‐mg/kg bid or higher all achieved median AUC12 results above the model‐derived EC95 (Figure 5). However, the oldest group of children (12 through 23 months of age) who received a unit dose of 30‐mg/dose did not achieve this exposure; their median (range) AUC12 from noncompartmental analyses was 2354 (1073–4949) ngxh/mL. At the outset of the CASG 114 study, the oseltamivir carboxylate target AUC12 in CASG 114 was 3800 ngxh/mL, with a range around this target of 2660–7700 ngxh/mL. This exposure target was selected based upon the development of antiviral resistance among 1–2‐year‐old children receiving 2 mg/kg of oseltamivir (producing an AUC12 of 2800 ngxh/mL). 14 , 15 Conversely, no resistance was observed in 1–2 year old children receiving 30‐mg/dose bid, 16 which is approximately 3 mg/kg for a 1 year old with an ideal bodyweight of 10 kg and which produces a modelled AUC12 of 3800 ngxh/mL. Serendipitously, the lower target range in CASG 114 was quite similar to the modelled EC95 in the current report, and in retrospect, the 3800‐ngxh/mL target exposure would appear to be reasonable as it covers the typical observed variability in AUC12 (~30% coefficient of variation). Applying this 30% variability to our model‐derived EC95 (2710 ngxh/mL) results in a new AUC12 target of 3523 ngxh/mL, which is very close to our original 3800‐ngxh/mL target. Figure 5 depicts simulated weight‐based oseltamivir doses relative to the modelled EC90 and EC95 for toddlers in Cohort I (12–23 month olds). During the conduct of the CASG 114 study, the dose for Cohort I was modified per protocol to the weight‐based dose of 3.5‐mg/kg bid. Unfortunately, only three subjects 12–23 months of age enrolled at this dose prior to study closure, but the median (range) AUC12 was 3073 (2149–4577) ngxh/mL for these subjects. Taken together, we conclude that the oseltamivir 30‐mg bid fixed dose produces subtherapeutic oseltamivir carboxylate exposure and that a weight‐based dose of at least 3.5‐mg/kg bid is likely necessary (Figures 4 and 5) to achieve and maintain adequate concentrations for influenza treatment and to prevent the development of resistance; this is consistent with previous dosing recommendations for children 9–11 months of age. 9 Further studies on alternative oseltamivir dosing in the 1–2 year old population are warranted. Simulated weight‐based oseltamivir doses and the resulting AUC12 values. Box plots represent the median AUC12 result per dose group and the 25th to 75th percentile range. Circles represent individual simulated AUC12 results. The oseltamivir carboxylate AUC12 in adults receiving 75 mg orally bid is 2719 ngxh/mL, and the C12h is 138 ng/mL. 3 A recent study compared this dose with 150‐mg bid in adults with influenza infection and did not find a difference in Day 5 viral RNA, culture negativity or RNA decline rate in patients with influenza A. 17 However, subanalysis of influenza B patients showed faster RNA decline rate and clearance with higher‐dose treatment, so it is possible different exposures may be required for different strains. Our simulation results suggest that this lack of difference between low and high doses for influenza A is not surprising, as immune‐mediated viral clearance, although not modelled explicitly, is reflected in the effective viral clearance term governing viral decay in this reduced‐form framework. 18 Both the AUC12 and C12h from 75 mg twice daily are already at or very close to the predicted EC95, thus higher doses (and higher oseltamivir carboxylate concentrations) are unlikely to produce significant differences in response, especially given the short 5‐day duration of therapy. Other studies also have shown no benefit to higher oseltamivir doses. 4 , 19 Oseltamivir carboxylate concentrations exhibit modest variability, with a typical coefficient of variation on AUC12 ranging from approximately 20% to 30%. 3 , 9 This variability tends to be considerably higher in young infants. As a result, there is an increased probability of the standard adult dose (75‐mg bid) producing subtherapeutic exposures in a portion of the population relative to higher doses. One reason this has not manifest clinically in the low‐ vs. high‐dose studies may be that the short 5‐day duration of therapy is not long enough to capture outcome differences as a result of the variability. Another reason may be the differences in doses are not large enough, especially since the 75‐mg bid dose produces average exposures near the EC95 already. Lastly, any clinical trial seeking to establish dosing differences would require a very large sample size, as clinical outcome differences would primarily be due to pharmacokinetic variability. These modelling and simulation experiments were designed to establish target oseltamivir carboxylate exposures for treatment of susceptible virus, but not for prophylaxis. Oseltamivir is approved for influenza prophylaxis at 75 mg once daily (qd), which is half the daily dose for treatment. Although prophylactic trials have demonstrated clinical benefit of this dose, 16 , 20 the risk of developing resistance in a patient during the early stages of infection by intentionally administering a subtherapeutic dose is likely increased. A modelling study explored the efficacy of oseltamivir in reducing both symptoms and viral load. 21 Lower doses and prophylaxis regimens led to lower efficacies and increased risk of resistance emergence. The authors concluded that prophylaxis initiated during the incubation period was the primary variable leading to the development of resistance. Although oseltamivir is approved for qd prophylactic dosing, cases of oseltamivir resistance during prophylactic therapy have been previously reported. 22 , 23 It is possible that a different (lower) oseltamivir carboxylate exposure would be required to prevent infection in exposed persons with very low or no viral titres. However, if virus is present in increasing concentrations before or at the onset of symptoms, the probability will be higher that selective pressure caused by subtherapeutic drug exposure will cause the emergence of resistant viral subpopulations. Most importantly, no dose‐finding trials for the prevention of influenza acquisition have been conducted for the approved neuraminidase inhibitors. Collectively, resistance to neuraminidase inhibitors remains generally low, 24 but if an individual is at risk of contracting or has already unknowingly contracted the illness due to close contact, dosing oseltamivir to achieve a targeted carboxylate exposure might be a more prudent course of action. Modelling and simulations in this report were derived exclusively from children aged 0 up to 2 year of age. Although the rate of viral decline in this population should theoretically be similar to adults dosed to achieve comparable drug exposures against susceptible virus, direct comparisons have not been made. As opposed to previous pharmacodynamic analyses, 8 we utilized actual viral load response data from patients infected with wild‐type influenza disease within the constraints of a large clinical trial. Because viral load sampling was sparse at the individual level, a population‐based viral dynamics approach was used to estimate key kinetic parameters and the population‐level EC95, rather than to precisely characterize individual viral load trajectories. External validation of the integrated PK–viral dynamics model was not possible due to the paucity of independent paediatric data sets with concurrent viral load measurements in this age group. As such, our results may not be translatable to other influenza laboratory or wild‐type viral subpopulations. In conclusion, population pharmacokinetic modelling can be directly linked to viral dynamic outcomes. Further, using simulations of population parameter estimates and their variability, wide dose‐ranging studies can be performed to identify target drug exposures. Oftentimes, critical concentration‐response relationships are difficult to establish because initial dose‐ranging trials do not encompass an extensive range of doses. Although additional study is needed, our approach might be used as a post hoc surrogate for early‐phase dose‐ranging studies when these data do not exist. AUTHOR CONTRIBUTIONS KW: Modelling, manuscript preparation. DWK: Study design, funding, manuscript preparation. RJW: Study design, funding, manuscript preparation. EPA: Study design, funding, modelling, manuscript preparation. All authors reviewed and approved the final version. CONFLICTS OF INTEREST STATEMENT R.J.W. served on the Board of Directors of Gilead (retired 2019), which licensed oseltamivir to F. Hoffmann‐La Roche Ltd. His involvement in this study was overseen by the University of Alabama at Birmingham Conflict of Interest Review Board. No other authors have potential conflicts to report. ACKNOWLEDGEMENTS We are indebted to the patients, their families and caregivers of the participants enrolled in this study. We also thank the personnel at the clinical sites for their professionalism in conducting this trial and the expertise of CASG Central Unit staff. Wang K, Kimberlin DW, Whitley RJ, Acosta EP, For the National Institute of Allergy and Infectious Diseases Collaborative Antiviral Study Group . Oseltamivir drug‐disease modelling in infants with influenza virus infection. Br J Clin Pharmacol. 2026;92(9):3214‐3222. doi: 10.1002/bcp.70623 Funding information This work was supported under contract with the Division of Microbiology and Infectious Diseases of the National Institute of Allergy and Infectious Diseases (NIAID) (N01‐AI‐30025). Assays of oseltamivir and oseltamivir carboxylate concentrations in plasma were provided by F. Hoffman‐La Roche Limited. KW was a researcher at the University of Alabama at Birmingham during the study period. Current address: Shanghai Qiangshi Information Technology Co., Ltd. Shanghai 201120, China. DATA AVAILABILITY STATEMENT Data supporting the findings of this study are available within this manuscript and any Supporting Information. REFERENCES Associated Data Data Availability Statement Data supporting the findings of this study are available within this manuscript and any Supporting Information.