Simvastatin to reduce pulmonary dysfunction in patients with acute respiratory distress syndrome: the HARP-2 RCT

Prior evidence

There was a large body of evidence from in vitro and animal studies suggesting that statins might be beneficial in ALI. 10 In summary, statins improved epithelial and endothelial function to reduce alveolar capillary permeability and reduced pulmonary oedema. In addition, they modulated the inflammatory cascade; regulated inflammatory cell recruitment, activation and apoptosis; and reduced cytokine and protease activity. This suggested that statins might improve outcomes, as high levels and persistence of inflammatory mediators in ALI are associated with poor outcome. 11

Lack of published randomised controlled trials of statins in acute lung injury

There was a lack of published randomised controlled trials (RCTs) of statins in ALI. We conducted a systematic review, searched registries of ongoing clinical trials and contacted national and international experts in ALI. The National Institutes of Health had conducted a Phase III multicentre trial involving statins, but this involved rosuvastatin (Crestor, AstraZeneca UK Ltd) compared with placebo for up to 28 days in patients with sepsis-induced respiratory failure in the USA. 12 Our trial examined simvastatin and investigated ALI due to all aetiologies, as well as studying the potential mechanism of action by which statins act. In addition, unlike the US trial, an economic evaluation was undertaken in this study.

Observational studies support a clinical trial of a statin in acute lung injury

Acute lung injury is the most common complication of severe sepsis. 13 In patients with sepsis, most observational studies14–17 suggest that statins are associated with better outcomes, as measured by morbidity and mortality. Similarly, most observational studies18–20 have suggested a beneficial effect of statins in patients with pneumonia, supporting a potential role for statins in modulating pulmonary inflammation.

The Irish Critical Care Trials Group (ICCTG) has undertaken a prospective observational study in patients with ALI, which found that mortality was lower in patients receiving statins during their ICU stay. After adjusting for plateau pressure, severity of illness and other relevant covariates in a multiple logistic regression model, patients receiving statins had a much lower probability of death, although this failed to reach significance [odds ratio (OR) 0.27, 95% confidence interval (CI) 0.06 to 1.21; p = 0.09]. 21 Similarly, in a retrospective study, statin usage in patients with ALI was associated with increased ventilator-free days (VFDs) and reduced mortality, although again this was not significant. 22 These observational studies were not powered to examine the effect of statins on mortality.

It was not clear if the association with better outcomes in these studies was due to statins as opposed to statins representing a surrogate marker for improved access to health care. Moreover, these studies did not demonstrate whether or not beneficial effects would occur when statins were commenced after the onset of ALI. Although these data suggested that statins may have a potentially beneficial pharmacological treatment in ALI, a trial powered for important clinical outcomes was required.

Simvastatin reduces lipopolysaccharide-induced pulmonary and systemic inflammation in humans

We had conducted a study to examine if simvastatin modulates pathogenic mechanisms important in the development of lung injury in a model of acute lung inflammation induced by inhaled lipopolysaccharide (LPS) in healthy human volunteers. 23 In this double-blind, placebo-controlled study, participants were randomised to the simvastatin group or the placebo orally for 4 days prior to LPS inhalation group. Pre treatment with simvastatin reduced mediators of early ALI in bronchoalveolar lavage fluid including tumour necrosis factor alpha (TNF-α), neutrophil myeloperoxidase and protease release as measured by neutrophil elastase and matrix metalloproteinase (MMP)-7, -8 and -9. Furthermore, there was a significant reduction in systemic inflammation as measured by plasma C-reactive protein (CRP) levels. These effects were associated with reduced nuclear factor kappa B translocation. These novel findings provided the first proof of the principle that simvastatin has important anti-inflammatory effects in vivo in humans who are challenged with aerosolised endotoxin. These mechanistic findings were supported by a randomised placebo-controlled study that found 80 mg of simvastatin for 4 days reduced systemic cytokine responses induced by low-dose intravenous LPS in healthy subjects. 24 Finally, a randomised placebo-controlled study in patients with acute bacterial infection found that simvastatin, commenced prior to the development of sepsis-induced organ dysfunction, also reduced the levels of systemic inflammatory cytokines [TNF-α and interleukin 6 (IL-6)]. 25

Proof of concept that simvastatin improves pulmonary and non-pulmonary organ dysfunction, reduces inflammation and is well tolerated in patients with acute lung injury

We had completed a single-centre, randomised, double-blind, placebo-controlled Phase II study of simvastatin (80 mg for up to 14 days) in 60 patients with ALI. By day 14 there was a trend to improvement in pulmonary dysfunction, as measured by the oxygenation index (OI), respiratory system compliance and lung injury score in the simvastatin-treated group. Non-pulmonary organ dysfunction, as measured by the sequential organ failure assessment (SOFA) score was significantly lower in the simvastatin-treated group, with improvements in cardiovascular, renal and coagulation function. There was no difference in outcome for patients with sepsis- or non-sepsis-related ALI. Importantly, 80 mg of simvastatin was well tolerated with no increase in adverse events (AEs). In addition, we found that, unlike placebo, simvastatin decreased pulmonary IL-8 2.5-fold by day 3, with a trend to a decrease in IL-6 2.9-fold. In addition, at day 14 plasma CRP was lower with a trend to reduced plasma IL-6 in the simvastatin-treated group. 26

Together these results reflected the beneficial effects seen in previous in vitro and animal studies. 10 The study described above was not designed or powered to show an effect of simvastatin on VFDs or mortality. However, pulmonary and non-pulmonary organ dysfunction, as well as high levels of inflammatory cytokines, were associated with fewer VFDs and higher ICU mortality, which suggested that simvastatin might lead to improved clinical outcomes.

The findings above were supported by two small prospective randomised controlled studies involving the acute use of statins in patients with sepsis and pneumonia. 27,28 Choi et al. 27 studied 10 mg of atorvastatin (Lipitor^®, Pfizer) given once daily in 74 patients with sepsis and pneumonia. Hospital mortality was reduced in the atorvastatin group compared with placebo, although this failed to reach significance (47% vs. 53%; p = 0.06). Similarly, Gonzalez et al. 28 conducted a study of 80 mg of simvastatin given once daily or placebo for 14 days in 40 patients with sepsis and found that simvastatin decreased the length of hospital stay.

Rationale for simvastatin 28-day duration of treatment

The decision to examine treatment for up to 28 days was based on (1) data from our proof-of-concept study that demonstrated ongoing clinical improvement to day 14,26 (2) data showing that the upper interquartile range for duration of ICU stay in patients with ALI/ARDS is 14–18 days5,7,9 and (3) observational trials that showed benefit with no reported toxicity when statins were continued throughout the ICU stay. 21,22

Rationale for 80 mg of simvastatin dosage

Although there were a large number of data suggesting that statins might be beneficial in animal models of ALI, only a single animal study compared two doses of simvastatin (5 mg/kg or 20 mg/kg given intraperitoneally 24 hours before and concomitantly with LPS to induce lung injury) and only the higher dose was effective in attenuating lung injury. 31

Importantly, a retrospective observational study of statin usage in patients with sepsis found a greater mortality benefit in patients who were receiving a higher dose of statin. 32

A dose of 80 mg of simvastatin is the only dose with proof-of-concept data and is well tolerated in ALI. Therefore, 80 mg of simvastatin compared with placebo once daily was investigated in this study.

Although it is acknowledged that the risk of adverse side effects is dose related, on the basis of available evidence 80 mg of simvastatin is safe, particularly given that the duration of treatment was only up to 28 days and these patients were intensively monitored.

There are no effective pharmacological therapies for acute lung injury

The Cochrane systematic review of pharmacological treatments that included 22 studies of 14 different drugs concluded that ‘effective pharmacotherapy for ALI is extremely limited, with insufficient evidence to support any specific intervention’. 33

The National Heart, Lung and Blood Institute working group considered the future research directions in ALI in 2002 and concluded that clinical trials underpinned by mechanistic investigations were essential to develop new therapies for ALI. 34

Primary outcome measure

The primary outcome measure was VFDs to day 28, defined as the number of days from the time of initiating unassisted breathing to day 28 after randomisation, assuming survival for at least 2 consecutive calendar days after initiating unassisted breathing and continued unassisted breathing to day 28. If a patient returned to assisted breathing and subsequently achieved unassisted breathing to day 28, VFDs were counted from the end of the last period of assisted breathing to day 28. A period of assisted breathing lasting < 24 hours and for the purpose of a surgical procedure was not counted against the VFD calculation. If a patient was receiving assisted breathing at day 27 or died prior to day 28, VFDs were counted as zero. Patients transferred to another hospital or other health-care facility were followed to day 28 to assess this end point.

In keeping with previous trials,38,39 unassisted breathing was defined as:

• extubated with supplemental oxygen or room air or

• open T-tube breathing or

• tracheostomy mask breathing or

• continuous positive airway pressure of ≤ 5 cm H₂O without pressure support.

Patients receiving pressure support via non-invasive ventilation were defined as receiving assisted ventilation.

Secondary outcome measures

The secondary outcomes for this clinical trial included clinical outcomes, safety, biological mechanisms and data for the economic evaluation.

Inclusion criteria

• Patients receiving invasive mechanical ventilation.

• Patient with ALI4 as defined by:

  • acute onset of hypoxic respiratory failure (PaO₂ : FiO₂ of ≤ 40 kPa from two arterial blood gas tests > 1 hour apart)

  • bilateral infiltrates on chest radiograph consistent with pulmonary oedema

  • no clinical evidence of left atrial hypertension or, if measured, a PAOP of ≤ 18 mmHg. If a patient has a PAOP of > 18 mmHg, then the other criteria must have persisted for > 12 hours after the PAOP had declined to < 18 mmHg, and still be within the 48-hour enrolment window.

Acute onset was defined as follows: the duration of the hypoxia criterion (1) and the chest radiograph criterion (2) must have been < 28 days at the time of randomisation.

Infiltrates considered ‘consistent with pulmonary oedema’ included any patchy or diffuse infiltrates not fully explained by mass, atelectasis, or effusion or opacities known to be chronic (> 28 days). The findings of vascular redistribution, indistinct vessels and indistinct cardiac borders were not considered ‘consistent with pulmonary oedema’.

All ALI criteria (under the second bullet point above) must have occurred within the same 24-hour period. The time of onset of ALI was when the last ALI criterion was met. Patients were enrolled within 48 hours of ALI onset.

Exclusion criteria

• Aged < 16 years.

• There had been > 48 hours since the onset of ALI.

• Patient was known to be pregnant.

• A CK level of > 10 times the upper limit of the normal range.*

• Transaminase levels of > 8 times the upper limit of the normal range.*

• Patients receiving ongoing and sustained treatment with any of the following: itraconazole, ketoconazole, human immunodeficiency virus (HIV) protease inhibitors, nefazodone (Dutonin, Bristol-Myers Squibb), ciclosporin, amiodarone, verapamil or diltiazem.

• Patients with severe renal impairment (estimated creatinine clearance of < 30 ml/minute) not receiving renal replacement therapy.

• Severe liver disease (Child–Pugh score of > 12).

• Current or recent treatment (within 2 weeks) with statins.

• Physician decision that a statin was required for proven indication.

• Contraindication to enteral drug administration (e.g. patients with mechanical bowel obstruction). Patients with high gastric aspirates due to an ileus were not excluded.

• Domiciliary mechanical ventilation except for continuous positive airway pressure/bilevel positive airway pressure used for sleep-disordered breathing.

• Known participation in other investigational medicinal product trials within 30 days.

• Consent declined.

• Treatment withdrawal imminent within 24 hours.

• Non-English-speaking patients or those who did not adequately understand verbal or written information unless an interpreter was available.

*If CK, ALT and AST values were not available as part of routine care, a blood sample was obtained after informed consent but before randomisation.

The CK, ALT and AST values could be obtained up to 72 hours prior to randomisation.

The following amendments relating to eligibility were made during the study.

• Protocol version 2.0: the exclusion criteria were amended to allow patients receiving low-dose erythromycin as a prokinetic to be included.

• Protocol version 3.0: concomitant use of clarithromycin and erythromycin and domicilliary ventilation for sleep-disordered breathing were removed as exclusion criteria.

• Protocol version 4.0: level of ALT and AST for eligibility and discontinuation of study drug was changed from five times the upper limit of normal to eight times the upper limit of normal.

Informed consent procedure for the UK

Informed consent forms approved by the Research Ethics Committee (REC) were provided to each trial site. The principal investigator (PI) was responsible for ensuring that informed consent for trial participation was given by each patient or a legal representative. This required that the informed consent form was signed and personally dated by the patient or by the patient’s legally acceptable representative. An appropriately trained doctor or nurse took consent. If no consent was given, then the patient was not randomised into the trial.

The incapacitating nature of the condition precluded obtaining prospective informed consent from participants. In this situation, informed consent was sought from a personal legal representative (PerLR) or professional legal representative (ProfLR) when a PerLR was not available.

Informed consent/assent procedure for Ireland

Informed consent/assent forms approved by the CREC were provided to each trial site. The PI was responsible for ensuring that informed consent/assent for trial participation was given by each patient or their representative respectively. This required that the informed consent/assent form be signed and personally dated by the patient or by their representative. An appropriately trained doctor or nurse took consent. If no assent was given, the patient was not randomised into the trial.

The incapacitating nature of the condition precluded obtaining prospective informed consent from participants. In this situation, informed assent was sought from the patient’s representative or from a professional representative if no suitable representative was available.

Assessment of causality

Each AE was clinically assessed for causality based on the information available (i.e. the relationship of the AE to the study drug). For the purposes of this trial, the causality was assessed using the categories presented below. Drug-related AEs were defined as those considered by the PI to have a possible, probable or definite relationship to the study drug. The PI at each site was responsible for evaluating all AEs for causality using the following guide:

• Unrelated: clinical event with an incompatible time relationship to study drug administration, and that could be explained by underlying disease or other drugs or chemicals.

• Unlikely: clinical event for which the time relationship to study drug administration makes a causal connection improbable, but that could plausibly be explained by underlying disease or other drugs or chemicals.

• Possible: clinical event with reasonable time relationship to study drug administration, but that could also be explained by concurrent disease or other drugs or chemicals.

• Probable: clinical event with a reasonable time relationship to study drug administration, and is unlikely to be attributed to concurrent disease or other drugs or chemicals.

• Definite: clinical event with plausible time relationship to study drug administration, and that cannot be explained by concurrent disease or other drugs or chemicals.

The AEs were reported and documented on the relevant pages of the case report form (CRF), in accordance with the procedures outlined below. The PI at each site was responsible for evaluating all AEs for expectedness in addition to causality and severity.

Drug-related AEs were defined as those with a possible, probable or definite relationship to the study drug. The site PI was asked to assess causality and record this as a ‘yes’ or ‘no’ in the CRF.

Adverse event reporting period

The AE reporting period for this trial began on enrolment into the trial and ended 30 days following the last administration of the study drug. All AEs assessed by the PI as possibly or probably related to the study drug and all SAEs that occurred during this time were followed until they were resolved or were clearly determined to be due to a patient’s stable or chronic condition or intercurrent illness(es).

Adverse event reporting

Because the HARP-2 trial was recruiting from a population that was already in a life-threatening situation, it was expected that many of the participants would experience AEs. Events that were expected in this population (i.e. events that were in keeping with the patient’s underlying medical condition) were not required to be reported as AEs.

An adverse reaction (AR) is an AE that is related to the administration of the study drug. Any AEs that were related to the study drug were reported on the AE form within the CRF.

The following are ARs that were expected and reported on the AE form within the CRF:

• CK level of > 10 times the upper limit of normal

• ALT/AST level of > 8 times the upper limit of normal.

An unexpected adverse reaction (UAR) is an AE that is related to the administration of the study drug and that is unexpected, in that it has not been previously reported in the current summary of product characteristics (SPC). Clinicians were instructed to report all UARs.

Serious adverse event reporting

A SAE was defined as an AE that fulfilled one or more of the criteria for severity:

• results in death

• is immediately life-threatening

• requires hospitalisation or prolongs existing hospitalisation

• results in persistent or significant disability or incapacity

• results in congenital abnormality or birth defect

• requires medical intervention to prevent one of the above, or is otherwise considered medically significant.

Because the HARP-2 trial was recruiting from a population that was already in a life-threatening situation, it was expected that many of the participants would experience SAEs. Events that were expected in this population (i.e. events that were in keeping with the patient’s underlying medical condition) and that were collected as outcomes of the trial, including death and organ failure, were not reported as SAEs.

The SAEs were evaluated by the PI for causality (i.e. their relationship to study drug) and expectedness. All other SAEs were reported using the SAE form in the patient’s CRF and were reported to the CTU within 24 hours of the clinician becoming aware of the event. The CTU was responsible for reporting SAEs to the sponsors, ethics committees, MHRA and IMB within the required timelines as per the regulatory requirements.

A serious adverse reaction (SAR) is a SAE that is related to the administration of the study drug. The following SAR was expected and clinicians were advised that it must be reported on the SAE form within the CRF:

• need for renal replacement therapy in patients with CK levels of > 10 times the upper limit of normal.

The SUSARs are SAEs that are considered to be caused by the study drug and are unexpected (i.e. their nature or severity is not consistent with the SPC). All SUSARs were the subject of expedited reporting to meet regulatory requirements.

All AEs and SAEs were classified using common terminology criteria for adverse events (CTCAE) version 4 (v4.03, 14 June 2010) apart from those AEs/SAEs that were protocol related. According to CTCAE v4.0, elevated ALT, AST and CK fall under the broad category of investigations. However, these AEs were expected ARs in this trial and, as such, they have been presented separately.

When CK levels were elevated and required renal replacement therapy, this SAR fell under the musculoskeletal category and was also reported separately.

Hospital data

The Acute Physiology and Chronic Health Evaluation (APACHE) II scores were used as part of the description of the trial population. For centres that participated in the Intensive Care National Audit and Research Centre (ICNARC) Case Mix Programme (CMP), the APACHE II scores were obtained from ICNARC; therefore, these centres supplied only the CMP number for HARP-2 trial participants. Centres that did not participate in the CMP collected all of the data to allow calculation of the APACHE II score.

At enrolment, the patients’ demographic characteristics, ventilatory and physiological variables and admission APACHE II scores were recorded. The cause of ARDS was identified by the treating clinician. For each day in the ICU, ventilatory and physiological variables as well as data on organ support based on the UK’s critical care minimum data set40 were recorded. Vital status at 28 days was collected but cause of death was not recorded.

Data were collected and recorded on a two-part carbonised CRF by the site research team from the time the patient was considered for entry into the trial through to their discharge from hospital. In the event that a patient was transferred to another hospital, the site research team liaised with the receiving hospital to ensure complete data collection. On completion of the data collection period, the PI signed off the CRF; the top copy of each CRF was returned to the CTU and the bottom copy was retained in the CRF booklet at the recruiting site. Submitted data were reviewed for completeness on receipt at the CTU and entered onto a secure, backed-up custom database. Entries on the CRF that were ambiguous, unintelligible or incomplete were queried with the hospital research staff who completed the CRF.

Discharge and follow-up questionnaires

To provide an economic evaluation, a HRQoL questionnaire was measured using the EuroQol-5 Dimensions, three-level version (EQ-5D-3L) questionnaire administered at hospital discharge by site staff. The CTU followed up surviving patients with a further EQ-5D questionnaire at 3, 6 and 12 months post randomisation. Resource utilisation data were also collected at 6 and 12 months.

To minimise the risk of causing distress by contacting relatives of patients who had since died, the CTU used a NHS central register and/or contacted the patient’s general practitioner (GP) to ascertain the patient’s survival status prior to any contact being made.

If questionnaires were not returned, a maximum of two telephone contacts were made to the patient to check that the questionnaire had been received and that the patient was happy to complete it. In the event that the patient stated non-receipt, a second copy of the questionnaire was sent out. If the second questionnaire was not returned, the patient was contacted again by telephone to follow up. To increase the percentage of questionnaire returns an amendment was submitted to the ethics committee to obtain permission to change our procedures to send out a £5 (€5 for sites in Ireland) gift voucher with the first questionnaire as a thank you gesture for patients taking part in the study.

Methods for assays

The MMP-8, IL-6, Ang 2 and RAGE were measured using Duoset enzyme-linked immunosorbent assay kits from R&D Systems Europe (Oxford, UK). According to the manufacturer’s instructions, samples were run neat or in reagent diluent (phosphate-buffered saline/1% bovine serum albumin) to obtain values within the detection range of the standard curve. When neat values were less than the lowest standard, then they were assigned the value of the lowest standard. The range of detection for the analyses was as follows.

• MMP-8: 31–4000 pg/ml

• IL-6: 9–600 pg/ml

• Ang2: 94–6000 pg/ml

• RAGE: 62.5–4000 pg/ml.

The CRP was measured by immunoturbidimetric assay performed by Randox Laboratories (Crumlin, Northern Ireland). The range of detection was 0.3–402 mg/l.

Plasma 25-hydroxyvitamin D was measured by liquid chromatography mass spectrometry by colleagues in the laboratory of Barbara Obermayer-Pietsch, Heidelberg, Germany. Values at or below the lower limit of detection were assigned a value of 6.99 ng/ml.

Sample size calculation

Sample size assumptions were based on previously published data. Assuming a mean number of VFDs of 12.7 days [standard deviation (SD) 10.6 days],41 we estimated that a sample of 524 patients would need to be enrolled in order for the study to have 80% power, at a two-tailed significance level of 0.05, to detect a mean between-group difference of 2.6 VFDs. On the basis of data from the Pulmonary Artery Catheters in Management of Patients in Intensive Care (PAC-Man) trial,42 a dropout rate of 3% was estimated and, therefore, a total of 540 patients (270 in each group) was required.

When the primary outcome measure of VFDs was available for 270 patients, a sample size review was undertaken by the DMEC independent statistician. The purpose of this was to check that the within-groups variance was not substantially underestimated, which would mean that the sample size had been underestimated.

No other data were analysed. The group allocation of the patients was not revealed and this review did not compare the two groups to examine treatment effects. In keeping with recommendations on interim sample size review,43 the review would not lead to a reduction of the sample size. The review led to a recommendation that the sample size remained unchanged.

Amendment one (main changes)

Protocol v1.0_24.06.10 was submitted to ORECNI in the original application for ethics approval. ORECNI requested some changes that necessitated amending the protocol to v2.0_01.09.10. The major changes included:

• amending the exclusion criteria to exclude non-English-speaking patients or those who did not adequately understand verbal or written information unless an interpreter was available

• amending the exclusion criteria to include the wording ‘currently’ and ‘sustained’ in relation to the use of listed concomitant medications. Amiodarone was added to the list of concomitant medications

• giving clarification that the 80-mg dose was given as two 40-mg tablets

• adding SOFA score to the schedule of assessments in day 14 and day 28.

Amendment two (main changes)

Protocol v2.0_01.09.10 was amended to v3.0_16.05.11. ORECNI and MHRA approved the following changes.

• Change of address of the chief investigator.

• Non-pulmonary organ failure-free days added to secondary outcomes.

• An additional blood sample was added at day 21 to measure CK and liver function.

• The exclusion criteria were amended to allow for patients receiving erythromycin as a prokinetic to be included in the study.

• Change of address of study drug supplier.

• Other changes: additional sites added.

Amendment three (main changes)

Protocol v3.0_16.05.11 was amended to v4.0_18.07.11 and was approved by ORECNI and MHRA to include the following changes.

• The exclusion criteria were amended to remove clarithromycin and erythromycin.

• Clarification was given that domiciliary ventilation used for sleep-disordered breathing would not be included as mechanical ventilation.

• Scheduling for research samples submission was changed to allow that research samples due on bank holidays or weekends could be collected up to 2 days after the due date (with the exception of day 1).

Amendment four

There was no change to the protocol. Additional sites were added.

Amendment five (main changes)

Protocol v4.0_18.07.11 was amended to v5.0_13.01.12. This amendment was approved by ORECNI and MHRA to include the following change.

• The exclusion criteria were amended to change the upper limit of normal for ALT and AST from more than five times the upper limit of normal to more than eight times the upper limit of normal.

Amendment six (main changes)

Protocol v5.0_13.01.12 was amended to v6.0_09.01.13. This amendment was approved by ORECNI and MHRA to include the following changes.

• The protocol was amended to allow NICTU to use a NHS central register and/or contact the patient’s GP to ascertain the patient survival status prior to any contact being made.

• The PIS and consent form were amended to inform patients and patient representatives of this change.

• A GP letter was created to advise the patient’s GP of future NICTU contact in relation to patient survival status.

Amendment seven (main changes)

Additional sites added and change of PI at three sites.

Minor amendments included:

• the inclusion of a £5 or €5 thank-you voucher

• change of CTU (NICTU) address from Education and Research Centre to Elliott Dynes Building.

Data collection and procedures

To ensure that accurate, complete and reliable data were collected, the CTU provided training to site staff in the format of investigator meetings and/or site initiation visits. The CTU provided the PI and research staff with training on good clinical practice, the study protocol, completion of the CRF and trial procedures including standard operating procedures.

Primary outcome

The number of VFDs did not differ significantly between the two study groups (Table 4) [12.6 days (SD 9.9 days) with simvastatin and 11.5 days (SD 10.4 days) with placebo; mean difference 1.1 days, 95% CI −0.6 to 2.8 days; p = 0.21]. 37 There was also no significant between-group difference in the number of VFDs after adjustment for the baseline PaO₂ : FiO₂ ratio (mean difference 1.4 days, 95% CI −0.3 to 3.2 days; p = 0.10).

Short-term secondary outcomes

The change from baseline to day 28 in the OI (Figure 3 and see Table 4) did not differ significantly between the two groups, nor did the SOFA score (Figure 4 and see Table 4). 37 There were no significant differences in the number of days free of non-pulmonary organ failure or in mortality at 28 days (see Table 4). 37 Mortality at ICU discharge or hospital discharge (see Table 4) was also not significantly different between the two groups. 37 Among survivors, the mean duration of the ICU stay was 13.9 days (SD 14.4 days) in the simvastatin group and 14.4 days (SD 13.3 days) in the placebo group (mean difference −0.5 days, 95% CI −3.2 to 2.2 days; p = 0.71). The mean duration of the hospital stay was 37.7 days (SD 64.5 days) and 35.4 days (SD 31.1 days) for the simvastatin group and the placebo group, respectively (mean difference 2.3 days, 95% CI −8.0 to 12.6 days; p = 0.66). From randomisation to day 28, there were no significant differences between the two groups in the probability of breathing without assistance or the probability of survival (Figure 5). 37

FIGURE 3.

The OI and 95% CI at days 1, 3, 7, 14 and 28.

FIGURE 4.

Mean SOFA score and 95% CI at days 1, 3, 7, 14 and 28.

FIGURE 5.

Kaplan–Meier plot for probabilities of survival at 28 days and breathing without assistance, from the day of randomisation (day 0) to day 28, according to whether patients received simvastatin or placebo. (a) Unassisted breathing; and (b) survival. From The New England Journal of Medicine, McAuley DF, Laffey MD, O’Kane CM, Perkins GD, Mullan B, Trinder J, et al. Simvastatin in the acute respiratory distress syndrome, vol. 371, pp. 1695–703. 37 Copyright © 2014 Massachusetts Medical Society. Reprinted with permission.

From The New England Journal of Medicine, McAuley DF, Laffey MD, O’Kane CM, Perkins GD, Mullan B, Trinder J, et al. Simvastatin in the acute respiratory distress syndrome, vol. 371, pp. 1695–703. 37 Copyright © 2014 Massachusetts Medical Society. Reprinted with permission.

Bootstrapped t-test for non-pulmonary OFFDs was not statistically significant (mean difference 1.6 days, 95% CI –0.3 to 3.5 days; p = 0.10). To adjust for the multiple testing for the change in OI and total SOFA score, a p-value of 0.0125 is considered to be statistically significant.

Exploratory biomarker analysis

Safety outcomes

A total of 121 patients experienced AEs and a total of 170 AEs were reported in the study. The SAEs were reported in 25 patients (11 patients in the simvastatin group and 14 patients in the placebo group). In total, 28 SAEs were reported (12 in the simvastatin group and the 16 in the placebo group), with one patient in the simvastatin group having two SAEs and two patients in the placebo group each having two SAEs. In the simvastatin group one SAE was assessed to be related to study drug and this was not assessed to be an unexpected SAE. In the placebo group three SAEs were assessed to be related to study drug and, of these, two were thought to be unexpected SAEs.

There were two SUSARs reported during the HARP-2 study, neither of which was fatal. Both SUSARs were reported to the relevant competent authorities and ethics committee within the applicable 15-day window.

Table 26 summarises AEs, SAEs, SUSARs, ARs and SARs by treatment group. 37 Table 27 shows that the absolute values of AST and CK were significantly elevated at day 14 in the simvastatin-treated group, and that the elevation in CK persisted to day 21, reflecting the increased incidence of AEs related to CK and elevated transaminases. Table 27 presents the mean (SD) for highest ALT (units per litre), AST (units per litre) and CK (units per litre) collected over the course of the trial.

Aim and perspective

The aim of the economic evaluation was to assess the cost-effectiveness of simvastatin compared with placebo at 12 months. The study was a cost–utility analysis undertaken alongside the main trial. Incremental cost-effectiveness ratios (ICERs) were used to present cost per quality-adjusted life-year (QALY). This ratio represents the difference in mean health service cost of the simvastatin and placebo groups divided by the difference in mean QALYs between groups to establish the cost per QALY gained, in keeping with the recommendations of National Institute for Health and Care Excellence (NICE). 48 The perspective of the analysis was the NHS and Personal Social Services. In addition, data of private and informal care were collected but these were not included in the cost-effectiveness analysis.

Health and social care service use and costs

Data on the health and social care services used by all patients were collected 12 months post randomisation. Data relating to patients’ primary hospital admission were collected prospectively via the CRF until primary hospital discharge or death. Data on levels of care and organ support were obtained from the daily data collection section of the CRF. These data (on levels of care and organ support) were only available for a maximum of 28 days or until discharge to a separate HDU or ward. It was assumed that patients were discharged to a rehabilitation ward for respiratory disorders. Other ICU days included the remaining time spent in the ICU after 28 days (if applicable) and any other stay in the ICU during a patient’s primary admission, but level of care was not recorded for these ICU days. It was anticipated that some patients would move multiple times between ICU, HDU and the ward during their primary admission, thus, the CRF tracked this movement so that length of stay could be calculated for each. Patients’ use of medications other than the study drug was not included in the economic analysis.

Use of health and social care services after hospital discharge was collected retrospectively via questionnaires posted out to surviving patients at 6 and 12 months post randomisation. Mortality status was established prior to posting via contact with the research sites, engagement with the Health and Social Care Information Service (now known as NHS Digital), and contact with GPs. The questionnaire asked patients to report their use of primary and secondary care services, and private and informal care.

Individual-level resource use was combined with unit costs to estimate total costs of health and social care use for each participant in the trial. Unit costs were obtained from publicly available sources and set at 2013–14 prices.

These were NHS Reference Costs49 for hospital resources, the Unit Costs of Health and Social Care 201450 for general practice services and the NHS Electronic Drug Tariff51 for the cost of simvastatin (Table 28).

Health outcomes

The economic analysis used the QALY as the measurement of health outcome. The QALY reflects the impact of an intervention on both the quality and quantity of a patient’s life and is recommended by NICE in economic evaluations. 48 QALYs were derived using the EQ-5D-3L. 53 This generic HRQoL instrument, which has been used frequently in the critically ill,54–56 provides a description of health using five dimensions (mobility, self-care, usual activities, pain/discomfort and anxiety/depression), each with three levels of severity. Patients also place their health on a visual analogue scale (VAS), for which 0 represents the worst imaginable health state and 100 the best imaginable health state. The UK social preference weights for EQ-5D-3L health states57 were used to obtain single utility values from the responses.

The area under the curve method was used to estimate patient-specific QALYs accrued over the 12-month period. This method involves multiplying the utility of the patients’ health state at a certain time point by the duration of the health state and then summing these over the study period. It was assumed that changes in utility over the time points were linear.

The EQ-5D-3L was administered at discharge, and at 3, 6 and 12 months post randomisation. All questionnaires were posted out to survivors, except for at discharge when it was administered face to face. As patients were unconscious and receiving invasive mechanical ventilation at the time of randomisation, baseline utility could not be measured. Instead we used the utility value for an unconscious state (–0.402) from the tariff. 57 This approach has been used previously in economic evaluations of therapies for patients with ARDS. 54,55

Over the course of the trial it became apparent that many patients were still in hospital at 3 months and were then not administered the EQ-5D-3L at this time point. Furthermore, the timing of the discharge questionnaire varied from patient to patient. For these reasons QALYs were calculated using only baseline, 6- and 12-month utility scores in the first instance as these captured the complete study period. Discharge and 3 months were used in the QALY calculation in a sensitivity analysis.

To maximise the return rate of postal questionnaires, the following evidence-based strategies were introduced.

• Franked, addressed return envelopes were provided with each 3-, 6- and 12-month questionnaire.

• A ‘thank-you’ voucher to the value of £5 (or €5 for Ireland) was enclosed with the first questionnaire sent out.

• A covering letter from the trial manager was included with each questionnaire.

• A follow-up telephone call was made if the questionnaire was not returned with the offer to resend a copy of the questionnaire if needed.

• A second and final follow-up telephone contact was made as a final reminder.

• To minimise the risk of causing distress to the relation of a patient who died after leaving the hospital, the CTU contacted the patient’s GP and NHS Digital to ascertain the patient’s survival status prior to sending out any questionnaires.

Analysis and reporting

We undertook the following:

• analysis of service use and costs

• analysis of health outcomes

• cost–utility analysis.

Analysis of service use and costs

Patients’ use of services during their primary admission to hospital, including simvastatin, is presented in Table 29. As level of care was not recorded for days spent in ICU after 28 days, these data are presented separately. No differences were statistically significant.

Patients’ use of other hospital services (hospital inpatient stay, outpatient attendances and accident and emergency visits) following discharge is presented in Table 30. Patients in the simvastatin group reported fewer hospital inpatient days but more outpatient and accident and emergency visits than the placebo group at both 6 and 12 months; however, these differences were not statistically significant between groups.

Patients’ use of community health services from baseline until 12 months is presented in Table 31. There were no statistically significant differences between the groups.

Patients’ use of care services is presented in Tables 32 and 33. In the period from discharge to 6 months, mean imputation was used for three patients who used a home help, three patients who used a carer paid for by the health service and one patient who used delivered meals. For the patient who used delivered meals, the mean value from the 6- to 12-month period was imputed as only one patient reported using the service from discharge to 6 months. In the period from 6 to 12 months, mean imputation was used for four patients who used a carer paid for by the health service. Patients in the placebo group received statistically significantly more carer visits paid for by the health service between 6 and 12 months than those in the simvastatin group. All other differences were not significant.

Patients’ use of private and informal carers is presented in Table 33. Private paid carers represent a cost to the patient rather than the health or social care provider and although they are not included in the subsequent ICER calculation, as they are outside the health-care provider perspective, they are presented for completeness. At 6 months, there was a statistically significant association between treatment and the use of unpaid carers, with a larger proportion of patients in the placebo group using unpaid carers.

Costs for individual resource components were grouped together to estimate mean costs for the primary hospital admission, other hospital services, community services and care-related services over the study period (Table 34). These total costs were calculated only for patients with complete data on each of the cost components. Lower mean costs were observed in the simvastatin group for all service types, with the exception of the primary hospital admission. The total mean cost difference at 12 months was statistically significant (–£4987, 95% CI –£10,060 to –£390). No other difference was statistically significant.

Health outcomes analysis

Mean EQ-5D-3L utilities and VAS scores at discharge, 3, 6 and 12 months, and QALYs at 12 months are presented in Table 35. Patients in the simvastatin group had higher HRQoL at each time point than those in the placebo group, with the most notable difference at 6 months. No differences between the groups were statistically significant. Both groups experienced an overall decline between discharge and 12 months.

Cost–utility analysis

Results from the cost–utility analysis are presented in Table 36. Simvastatin is both less costly and more effective than placebo; thus, it is the dominant strategy with a mean difference in costs of –£3601 and a mean difference in QALYs of 0.064. In this situation, the ICER is not calculated as its magnitude does not convey any meaning. 58 Sampling uncertainty is represented by the joint distribution of the bootstrapped differences in cost and QALY on the cost-effectiveness plane for the primary analysis (Figure 13). The majority of the points lie below the x-axis (indicating that simvastatin is cost saving) and to the right of the y-axis (indicating simvastatin produces more QALYs than placebo). The small number of points lying outside of this area indicate a small degree of variability surrounding the presence and magnitude of cost savings and effectiveness.

FIGURE 13.

Cost-effectiveness plane for the primary cost-effectiveness analysis (showing bootstrapped replications of mean incremental costs and QALY gain and the WTP threshold of £20,000 per QALY). 44

The CEAC for the primary analysis presented in Figure 14 summarises this uncertainty for the decision-maker and presents the probability of simvastatin being cost-effective compared with placebo at different thresholds of WTP per QALY gain for the primary and sensitivity analyses. The CEAC for the primary analysis indicates that, at a WTP threshold of £20,000 per QALY gain, the probability of simvastatin being cost-effective is 99%.

FIGURE 14.

Cost-effectiveness acceptability curve (showing the probability of simvastatin being cost-effective compared with placebo for the primary and sensitivity analyses). Amended from Agus et al. 44 This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Amended from Agus et al. 44 This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Sensitivity analyses were conducted to determine the impact of changing particular assumptions on the cost-effectiveness (see Table 36). Although there are some notable effects on differential mean costs and effects, the CEACs for the sensitivity analyses (see Figure 14) indicate that cost-effectiveness of simvastatin was robust to these changes in the assumptions, with the probability of it being cost-effective at £20,000 per QALY never dropping below 90%.

Trial Management Group

Daniel McAuley (Chief Investigator, UK), John Laffey (Chief Investigator, Ireland), Christine McNally (Trial Manager), Clíona McDowell (Statistician), Ashley Agus (Health Economist), Colette Jackson, Jim O’ Neill (Trial Co-ordinator), Angela Toner (Data Manager), Susan McMullan (Data Manager), Patricia Rafferty (Senior Monitor), Roisin Boyle (Monitor), Emma Deenihan, Michael Scully and Veronica McInerney (HRB Galway Clinical Research Facility, Galway, Ireland).

Trial Steering Committee

Duncan Young (Independent Chairperson), John Radcliffe Hospital, Oxford, UK.

Rupert Pearse, Barts and the London School of Medicine and Dentistry, Royal London Hospital, London, UK.

Kathy Rowan, ICNARC, Tavistock House, London, UK.

Barry Williams, Critical Care Patient Liaison Committee (CritPaL), Dorset, UK.

Daniel F McAuley, Wellcome–Wolfson Institute for Experimental Medicine, Queen’s University of Belfast, Belfast, UK.

John Laffey, Department of Anaesthesia and Intensive Care, Clinical Sciences Institute, National University of Ireland Galway, Galway, Ireland.

Lynn Murphy, NICTU, The Royal Hospitals, Belfast, UK.

Data Monitoring and Ethics Committee

Geoff Bellingan (Chairperson), University College London, London, UK.

David Harrison, ICNARC, Tavistock House, London, UK.

Anthony Gordon, Imperial College/Charing Cross Hospital, London, UK.

Clinical collaborators (* denotes principal investigator)

Addenbrooke’s Hospital, Cambridge: Andrew J Johnston,* Archana Paikray, Cat Yates, Petra Polgarova, Esther Price, Amy McInerney and Katarzyna Zamoscik.

Aintree University Hospital, Liverpool: Ged Dempsey* and Colette Seasman.

Altnagelvin Area Hospital, Londonderry: Lynn Gilfeather,* Noel Hemmings and Sinead O’Kane.

Antrim Area Hospital, Antrim: Paul Johnston,* Lukas Pokorny, Chris Nutt and Orla O’Neill.

Arrowe Park Hospital, Wirral: Prashast Prashast,* Chris Smalley and Reni Jacob.

Beaumont Hospital, Dublin: James O’Rourke,* Syed Farjad Sultan and Carole Schilling.

Birmingham Heartlands Hospital, Birmingham: Gavin D Perkins,* Teresa Melody, Keith Couper, Ron Daniels, Fang Gao and Julian Hull.

Bristol Royal Infirmary, Bristol: Timothy Gould,* Matthew Thomas and Katie Sweet.

Cork University Hospital, Cork: Dorothy Breen* and Emer Neau.

Dumfries and Galloway Royal Infirmary, Dumfries: Willis J Peel,* Catherine Jardine and Paul Jefferson.

Freeman Hospital, Newcastle upon Tyne: Stephen E Wright* and Kayla Harris.

Frenchay Hospital, Bristol: Matthew Thomas* and Sarah Hierons.

Galway University Hospital, Galway: John Laffey* and Veronica McInerney.

Good Hope Hospital, Birmingham: Gavin Perkins.*

Guy’s and St Thomas’ Hospital, London: Luigi Camporota* and Katie Lei.

Harefield Hospital, Harefield: Sundeep Kaul* and Molly Chivburi.

Hull Royal Infirmary, Hull: Andrew Gratix,* Rachael Bennett, Victoria Martinson, Lisa Sleight and Neil Smith.

King’s College Hospital, London: Phillip A Hopkins,* Daniel Hadfield, Sarah Casboult, Fiona Wade-Smith, Julie Dawson, Clare Mellis, Clair Harris, Georgina Parsons and Sinead Helyar.

Leeds Teaching Hospitals NHS Trust, Leeds: Andrew R Bodenham,* Stuart Elliott, Zoe Beardow and Sian Birch.

Mater Misericordiae University Hospital, Dublin: Brian Marsh* and Teresa Martin.

Norfolk and Norwich University Hospitals, Norwich: Akesh Dhrampal* and Melissa Rosbergen.

Papworth Hospital, Cambridge: Stephen Webb* and Fiona Bottrill.

Poole Hospital, Poole: Henrik Reschreiter,* Helena Barcraft-Barnes and Julie Camsooksai.

Queen Elizabeth Hospital Birmingham, Birmingham: Andrew Johnston,* Aisling Clarkson, Conor Bentley, Lauren Cooper, Yongyan Qui, Natalie Mitchell, Ronald Carrera and Arlo Whitehouse.

Royal Berkshire Hospital, Reading: Christopher M Danbury,* Nicola Jacques and Abby Brown.

Royal Derby Hospital, Derby: David Rogerson* and Craig Morris.

Royal Infirmary of Edinburgh, Edinburgh: Timothy Walsh,* Mike Gillies, Grant Price, Kallirroi Kefala, Neil Young, David Hope, Corrienne McCulloch, Jean Antonelli, Pam Ramsay, Kirsty Everingham, Louise Boardman, Heidi Dawson, Fiona Pollock and Joanne Thompson.

Royal Liverpool University Hospital, Liverpool: Ingeborg D Welters,* Lee Poole, Peter Hampshire, Alison Hall, Karen Williams, Anna Walker, Laura Youds, Samantha Hendry, Victoria Waugh, Julie Patrick-Heselton and David Shaw.

Royal Preston Hospital, Preston: Irfan Chaudry* and Jacqueline Baldwin.

Royal Sussex County Hospital, Brighton: Stephen Drage* and Laura Ortiz-Ruiz de Gordoa.

Royal Victoria Hospital, Belfast: Daniel McAuley,* Leona Bannon, Vanessa Quinn, Lia McNamee and Griania White.

St George’s Hospital, London: Maurizio Cecconi* and Johannes Mellinghoff.

St Vincent’s University Hospital, Dublin: Donal Ryan* and Alistair Nichol.

The Royal Free Hospital, London: Banwari Agarwal,* Paula Meale, Sarah James, Kulwant Dhadwal, Daniel Martin, Agnieszka Walecka and Stephen Ward.

Ulster Hospital, Dundonald: Thomas J Trinder,* Samantha Hagan, Janice Montgomery, Catherine Leonard, Elizabeth Lemon and Tom Trinick.

University Hospitals Coventry and Warwickshire, Coventry: Murthy Buddhavarapu,* Geraldine Ward and Christopher Bassford.

Victoria Infirmary, Glasgow: Alan Davidson,* Kate McGuigan, Anissa Benchiheub and Naomi Hickey.

Western Infirmary, Glasgow: Alexander Binning* and Steven Henderson.

Whiston Hospital, Liverpool: Julie A Wood.*

Worcester Royal Hospital, Worcester: Andrew J Burtenshaw,* Dawn Kelly, Terry Martin, Jessica Thrush, Julie Wollaston, Stephen Graystone, Gavin Nicol and Gareth Sellors.