Exploration of the Hospital Recovery Scale as an endpoint to assess clinical improvement in hospitalized influenza patients
The best way to evaluate the effectiveness of a new therapy for hospitalized patients with influenza is not known. The researchers are exploring a new measure, called the Hospital Recovery Scale (HRS), that might be used in clinical research to show the effectiveness of a new influenza treatment in patients who are hospitalized for influenza. The HRS captures the clinical status at a specific timepoint and consists of six categories that assess the patient status from best to worse: (1) Not Hospitalized [Home](2) Non-Intensive Care (ICU) Hospitalization, Not Requiring Supplemental Oxygen [Ward](3) Non-ICU Hospitalization, Requiring Supplemental Oxygen [Oxygen](4) Admitted to the ICU, Not Requiring Invasive Mechanical Ventilation [ICU] (5) Requiring Invasive Mechanical Ventilation [Vent.](6) Death [Death]The HRS describes the hospitalization status of an influenza patient each day. Measured daily, the HRS provides insight in the progression of an influenza patient's hospitalization status over time. The patient's status can be analyzed and compared between (sub)groups of patients. Similarly designed ordinal scale endpoints have been recently used in other research in severe influenza patients requiring hospitalization. The ZORO study (GSK-NAI114373; NCT01231620) has been selected because it is a large study that includes adolescent and adult patients hospitalized with influenza with the expectation that it contains the data to be able to characterize patients appropriately on the HRS and other endpoints. All patients who are infected with influenza and were treated will be included in this exploration, with primary interest in patients infected with Influenza A. The objectives of this research are:
• To explore the HRS as a new endpoint to assess clinical improvement in hospitalized influenza patients
• To explore the correspondence of the HRS outcome with other clinical endpoints in a hospitalized influenza patient setting
• To explore the correspondence of the effect of the HRS outcome with other established clinical outcomes in a hospitalized influenza patient setting between (sub)groups The other clinical outcomes which will be explored include:
• what proportion of patients experience complications of influenza
• how long patients need to stay in the hospital
• what proportion of patients need intensive care
• what proportion of patients need to use a breathing machine (mechanical ventilation)
• what proportion of patients die
• how long it takes before patients return to pre-influenza activity and function
• how long it takes to recover from influenza symptoms
• how long it takes for breathing to improve to usual levels
• how long patients need extra supplemental oxygen.
The results of this analysis will be made public and will be submitted for presentation at a scientific conference and will also be submitted for potential publication in scientific journals.
[{ "PostingID": 4740, "Title": "GSK-NAI114373", "Description": "A Phase III international, randomized, double-blind, double-dummy study to evaluate the efficacy and safety of 300 mg or 600 mg of intravenous zanamivir twice daily compared to 75 mg of oral oseltamivir twice daily in the treatment of hospitalized adults and adolescents with influenza" }]
Groups of patients will be identified, using treatment and/or baseline characteristics. This will be evaluated by assessing differences in effect between the (sub)groups on the HRS at study day 6, expressed as a common odds ratio from proportional odds modeling.The endpoints listed below are of interest and are assumed to be available or derivable from the data set, based on the anonymized ZORO study report. The planned analysis methods indicated will be used to compare the identified (sub)groups in the data set. Hospital Recovery Scale (HRS)• DefinitionCategories are of increasing severity and mutual exclusive:1. Not hospitalized2. Non-ICU hospitalization, not requiring supplemental oxygen3. Non-ICU hospitalization, requiring supplemental oxygen4. Admitted to the ICU, not requiring invasive mechanical ventilation5. Requiring invasive mechanical ventilation6. Death• The HRS will be analyzed with the proportional odds model, adjusted for baseline HRS category, providing a common odds ratio to assess differences in effect between (sub)groups.Incidence of complications associated with influenza • DefinitionIncidence of complications associated with influenza after the start of study drug administration (Y/N)Complication categories: - Any complication - Pulmonary complications, including respiratory failure, primary viral pneumonia, secondary bacterial pneumonia, exacerbations of chronic underlying pulmonary diseases, bronchitis - Extrapulmonary complications, including cardiovascular and cerebrovascular disease (eg, myocardial infarction, congestive heart failure, arrhythmia, stroke), muscular disorders, central nervous system involvement, acute exacerbation of chronic kidney disease, severe dehydration, decompensation of previously controlled diabetes mellitus, sinusitis, otitis• The incidence of complications associated with influenza will be analyzed using logistic regression, providing an odds ratio to assess differences in effect between (sub)groupsTime to hospital discharge• DefinitionTime from (start of) first study drug administration to hospital discharge• CensoringSubjects who complete or withdraw from the study while still hospitalized will be censored at the date of completion or withdrawal, respectively. In case of death, censoring will be done using a time-to equal to the maximum time across all subjects• Time to hospital discharge will be analyzed using an Accelerated Failure Time (AFT) model and the Kaplan Meier method, providing an AFT ratio and KM estimates of median time to event to assess differences in effect between (sub)groups, and the Gehan Wilcoxon test to asses significance of any differences.Incidence of post-baseline ICU admission• DefinitionOccurrence of ICU admission after (start of) first study drug administration (Y/N)Subjects with multiple ICU admissions are counted once only.In case a patient is in the ICU at baseline and (after discharge from the ICU) the patient is re-admitted to the ICU post-baseline, this subject will be counted as 'Admitted to ICU post-baseline'• The incidence of post-baseline ICU admission will be analyzed using logistic regression, providing an odds ratio to assess differences in effect between (sub)groupsIncidence of post-baseline start of mechanical ventilation• DefinitionOccurrence of mechanical ventilation after (start of) first study drug administration (Y/N). Subjects with multiple periods of mechanical ventilation are counted once only.In case mechanical ventilation is applied at baseline and a second episode of mechanical ventilation is initiated post-baseline, this subject will be counted as ‘Started mechanical ventilation post-baseline'• The incidence of post-baseline start of mechanical ventilation will be analyzed using logistic regression, providing an odds ratio to assess differences in effect between (sub)groupsIncidence of all-cause mortality • DefinitionDeath after (start of) first study drug administration (Y/N). • The incidence of post-baseline start of all-cause mortality will be analyzed using logistic regression, providing an odds ratio to assess differences in effect between (sub)groupsTime to return to pre-morbid functional status• DefinitionReturn to pre-morbid functional status is defined as the first time point a patient first reaches 6 points on the Katz Activities of Daily Living total score. • Censoring - None of the available records indicate return to pre-morbid functional status: censoring will be done on the day of the last observation. - In case of death, censoring will be done using a time-to equal to the maximum time across all subjects• Time to return to pre-morbid functional status will be analyzed using an AFT model and the Kaplan Meier method, providing an AFT ratio and KM estimates of median time to event to assess differences in effect between (sub)groups, and the Gehan Wilcoxon test to asses significance of any differences. Time to return to pre-morbid level of activity• DefinitionReturn to pre-morbid level of activity is based on the Three-Point Activity Scale [bed rest, limited ambulation, unrestricted]. The first time point a patient first scores the outcome ‘unrestricted' the patient is considered to be returned to the pre-morbid level of activity. In case the pre-morbid level of activity is available then the patient is considered to be returned to the pre-morbid level of activity if the post-baseline level equals the pre-morbid level.• Censoring - Last available record does not indicate return to pre-morbid level of activity: censoring will be done on the day of the last observation. - In case of death, censoring will be done using a time-to equal to the maximum time across all subjects• Time to return to pre-morbid level of activity will be analyzed using an AFT model and the Kaplan Meier method, providing an AFT ratio and KM estimates of median time to event to assess differences in effect between (sub)groups, and the Gehan Wilcoxon test to asses significance of any differences.Time to resolution of influenza symptoms• DefinitionResolution of influenza symptoms is defined as the first of a successive series over 24 hours (2 recordings) during which all symptom scores for each of the assessments are 0 or 1 for all 7 primary influenza symptoms (cough, sore throat, headache, nasal symptoms [rhinorrhea, congestion], feverishness, myalgias, fatigue). The 2 successive recording should be done over 3 scheduled successive analysis timepoints, with 1 missing timepoint in between being allowed. • Censoring 1) Last available record(s) indicate resolution of influenza symptoms (but insufficient recordings to meet the time to resolution): censoring will be done at the first record of resolution from the last series of recordings of resolution of influenza symptoms 2) Last record does not indicate resolution of influenza symptoms: censoring will be done at the next planned timepoint after the last observation 3) In case of death, censoring will be done using a time-to equal to the maximum time across all subjects• Time to resolution of influenza symptoms will be analyzed using an AFT model and the Kaplan Meier method, providing an AFT ratio and KM estimates of median time to event to assess differences in effect between (sub)groups, and the Gehan Wilcoxon test to asses significance of any differences.Time from start of study drug to clinical response • DefinitionThe time (in hours) from first study drug intake until the first assessment of a successive series over at least 24 hours during which at least 4 of 5 symptoms (temperature, blood oxygen saturation, heart rate, SBP, respiration rate) are normalized (see table), with at least normalization of temperature and blood oxygen saturation. The successive recordings may include 1 missing timepoint in between if resolution for at least 24 hours still can be concluded.Normalization Criteria - Temperature: Oral temperature ≤ 37.0 °C (if data available: without the use of antipyretics within 8 hours) - Oxygen saturation: a) ≥ 94% on room air without supplemental oxygen b) In case pre-influenza infection oxygen saturation <94%: - 3% increase compare to baseline oxygen saturation - Return to pre-influenza oxygen saturation- Respiration rate: ≤ 24/min or return to pre-influenza infection oxygen requirement in patients with chronic oxygen use- Heart rate: ≤ 100/min- Systolic blood pressure: ≥ 90 mmHg (if data available: without inotropic support given within 2 hours of assessment)•CensoringOnly data during hospitalization will be used.In case clinical response is not met, data will be censored. Censoring will be done as follows 1) Last record(s) during hospitalization indicate normalization (but insufficient recordings to meet the time to clinical response criteria): censoring will be done at the first record of normalization 2) Last record during hospitalization does not indicate normalization: censoring will be done at hospital discharge • Time from start of study drug to clinical response will be analyzed using an AFT model and the Kaplan Meier method, providing an AFT ratio and KM estimates of median time to event to assess differences in effect between (sub)groups, and the Gehan Wilcoxon test to asses significance of any differences.Time from start of study drug to respiratory response• DefinitionThe time (in hours) from first study drug intake until the first assessment of a successive series during at least 24 hours during which where both blood oxygen saturation and respiration rate are normalized (see table) The successive recordings may include 1 missing timepoint in between if resolution for at least 24 hours still can be concluded.• CensoringAnalogous to Time to clinical response.• Time from start of study drug to respiratory response will be analyzed using an AFT model and the Kaplan Meier method, providing an AFT ratio and KM estimates of median time to event to assess differences in effect between (sub)groups, and the Gehan Wilcoxon test to asses significance of any differences.Time from start of study drug to end of supplemental O2• Definition The time (in hours) from first study drug intake until the first assessment of a successive series for 24 hours (4 recordings) during which O2 saturation on room air is >94%. In case pre-influenza O2 saturation on room air is available, the O2 saturation on room air should be above or equal to the pre-influenza O2 saturation on room air.The 4 successive recording should be done over 5 scheduled successive analysis timepoints, with 1 missing timepoint in between is allowed• CensoringAnalogous to Time to clinical response.• Time from start of study drug to end of supplemental O2 will be analyzed using an AFT model and the Kaplan Meier method, providing an AFT ratio and KM estimates of median time to event to assess differences in effect between (sub)groups, and the Gehan Wilcoxon test to asses significance of any differences.The direction and magnitude of the effect (expressed in terms of a Z-value) will be compared descriptively across the endpoints. Determine relationships between a patient's HRS category (on Day 6) with the patient's result on the listed endpoints, by means of analysis of variance, AFT model, fisher's exact test, logistic regression, proportional odds model or other appropriate method depending on the type of endpoint with the HRS category on Day 6 as explanatory variable / classification variable. Changes to the analysis plan may be needed if expected data is not available in the study data base.
W. van Duijnhoven, I. Van Dromme, S. Haesendonckx, J. Witek, L. Leopold The Hospital Recovery Scale: A clinically useful endpoint in patients hospitalized with influenza Contemporary Clinical Trials vol 123 (2022)
https://doi.org/10.1016/j.cct.2022.106952