Interstitial Lung Disease Associated Acute Respiratory Failure Requiring Invasive Mechanical Ventilation: A Retrospective Analysis

Background: Interstitial Lung Disease [ILD] patients requiring Invasive Mechanical Ventilation [IMV] for Acute Respiratory Failure [ARF] are known to have a poor prognosis. Few studies have investigated determinants of outcomes and the utility of trialing Non-Invasive Positive Pressure Ventilation [NIPPV] prior to IMV to see if there are any effect[s] on mortality or morbidity. Methods: A retrospective study was designed using patients at four different intensive care units within one health care system. The primary objective was to determine if there are differences in outcomes for in-hospital and one-year mortality between patients who undergo NIPPV prior to IMV and those who receive only IMV. A secondary objective was to identify potential determinants of outcomes. Results: Out of 54 ILD patients with ARF treated with IMV, 20 (37.0%) survived until hospital discharge and 10 (18.5%) were alive at one-year. There was no significant mortality difference between patients trialed on NIPPV prior to IMV and those receiving only IMV. Several key determinants of outcomes were identified with higher mortality, including higher ventilatory support, idiopathic pulmonary fibrosis (IPF) subtype, high dose steroids, use of vasopressors, supraventricular tachycardias (SVTs), and higher body mass index. Conclusion: Considering that patients trialed on NIPPV prior to IMV were associated with no mortality disadvantage to patients treated with only IMV, trialing patients on NIPPV may identify responders and avoid complications associated with IMV. Increased ventilator support, need of vasopressors, SVTs, and high dose steroids reflect higher mortality and palliative care involvement should be considered as early as possible if a lung transplant is not an option.


Background
Interstitial lung disease (ILD) patients requiring invasive mechanical ventilation (IMV) for acute respiratory failure (ARF) are known to have a poor prognosis. Few studies have investigated determinants of outcomes and the utility of trialing non-invasive positive pressure ventilation (NIPPV) prior to IMV to see if there are any effect(s) on mortality or morbidity.

Methods
We designed a retrospective study using patients at four different intensive care units within one health care system. Our primary objective was to determine if there are differences in outcomes for in-hospital and one-year mortality between patients who undergo NIPPV prior to IMV and those who receive only IMV. A secondary objective was to identify potential determinants of outcomes.

Results
Of 54 ILD patients with ARF treated with IMV, 20 (37.0%) survived to hospital discharge and 10 (18.5%) were alive at one-year. There was no significant mortality difference between patients trialed on NIPPV prior to IMV and those receiving only IMV. Several key determinants of outcomes were identified with higher mortality, including: higher ventilatory support, idiopathic pulmonary fibrosis (IPF) subtype, high dose steroids, use of vasopressors, supraventricular tachycardias (SVTs), and higher body mass index.

Conclusions
Considering that patients trialed on NIPPV prior to IMV was associated with no mortality disadvantage to patients treated with only IMV, trialing patients on NIPPV may identify responders and avoid complications associated with IMV. Increased ventilator support, need of vasopressors, SVTs, and high dose steroids reflect higher mortality and palliative care involvement should be considered as early as possible if lung transplant is not an option.

Background
Interstitial lung disease (ILD) patients can be subject to episodes of acute respiratory failure (ARF) and rapid decline during their disease course. Patients with ILD experiencing ARF have known poor outcomes once invasive mechanical ventilation (IMV) is initiated. Retrospective studies have reported in-hospital mortality ranging from 51-100% in ILD patients with ARF requiring ICU level of care (1)(2)(3)(4)(5)(6).
IMV may be required in ILD associated ARF (ILD-ARF), although its benefit is unclear unless used as a bridge to lung transplantation (LTx).
ARF from an ILD exacerbation is defined by: a subjective worsening of dyspnea within the month prior to presentation; new ground glass opacities or consolidation by chest imaging; hypoxemia with > 10 mmHg decline in PaO2; and no evidence of PE, CHF, lung infection, or pneumothorax (7,8). This definition has been modified for the acute exacerbation of idiopathic pulmonary fibrosis (IPF) to now include pulmonary infection as an etiology, but this modification has not been incorporated into the other ILD subtypes (9).
IMV can initiate and exacerbate lung injury, termed ventilator-induced lung injury (VILI), increasing mortality and morbidity (6). Noninvasive positive pressure ventilation (NIPPV) may offer some benefits of IMV, by improving oxygenation and reducing the work of breathing, and minimize risk for VILI. The effectiveness of NIPPV in avoiding IMV in patients with ILD-ARF has not been well studied. Whether or not trialing NIPPV, and potentially delaying IMV, increases mortality is also unclear.
There are no guidelines on how to select ILD-ARF patients to place on IMV. Gungor et al (2013) proposed that physicians should be guarded about the use of IMV in ILD-ARF patients that are not suitable for lung transplantation (LTx), especially in patients requiring continuous NIPPV (5). There are several studies looking at IPF patients who require IMV, revealing high in-hospital mortality. Although IPF is the most common idiopathic interstitial pneumonia, it represents a fraction of patient with ILD who have ARF (7). The outcomes of ILD patients as a whole has been less frequently reported. While there are individuals who survive IMV, there is limited data to differentiate these individuals from those who do not. As a result, this limits the ability to have informed goals of care discussions for critically ill patients.
The primary objective of this study was to investigate mortality outcomes between two cohorts of ILD-ARF patients: (1) those who are trialed on NIPPV prior to receiving IMV and (2) those receiving only IMV. The secondary objective was to identify the determinants of outcomes within the entire study 4 population and within each stated cohort.

Study population
From January 2014 to October 2018, 54 patients with ILD-ARF underwent IMV in 4 different hospitals within the University of Pennsylvania Health System. Patients were identified by using the International Classification of Diseases (ICD) codes to search within our institution's electronic medical record (EMR). Institutional review board (IRB) approval was obtained prior to the review of medical records.
Patient's records were reviewed if they were previously diagnosed with ILD and if ARF was experienced during their admission. Only the first presentation of ARF requiring IMV was used in this study in cases of patients with repeat admissions for ARF. Their ARF had to require any form of positive pressure ventilation (PPV) initially and ultimately required IMV for 24 hours (6).
ILD diagnosis based on American Thoracic Society criteria was limited in assessment based on EMR review. To improve the accuracy of the ILD diagnosis we created major and minor criteria for patient eligibility. Major criteria included: an available CT scan read by a radiologist that was suggestive of ILD and either co-existing restrictive PFTs or pulmonologist documentation that confirmed ILD as a clinical diagnosis; available tissue biopsy confirming diagnosis; or documented proof of lung transplant screening due to an ILD. Minor criteria included: either stated ILD in prior and/or current physician encounter, a CT scan referencing ILD, a pulmonary function test (PFT) with restrictive profile, an ICD code for ILD in the EMR, and a pulmonary provider documentation noting ILD in current encounter. A patient required either one major criteria or three minor criteria to be included in our study.
Patients were excluded if they: had a chronic tracheostomy, were surgical patients experiencing ARF within 48 hours after being extubated for a surgical procedure, had no ARF during their encounter, were intubated for reasons that were unrelated to ARF, or had a history of having undergone LTx.
We defined ARF based on the British Medical Journal best practice guidelines (10). Once a patient met criteria of ARF, we required evidence of supplemental oxygenation use that ultimately needed to be escalated to PPV for respiratory support. For patients already on home oxygen, we defined ARF as an increased oxygen requirement from their home requirement. All patients included in this study eventually required IMV support for greater than 24 hours for ARF.
PH diagnosis was based on prior ICD coding or defined in patients with recent echocardiography report within 6 months of encounter or during encounter with an sPAP ≥ 45 mmHg (11).
Pulse dose steroids were defined as 1 g of methylprednisolone for a minimum of 3 days (12). Stress dose steroids were defined as a documented administration at no more than 300 mg of hydrocortisone (or equivalent) in a 24 hour time period (13,14) for the duration of shock physiology or death. Observed stress dose steroid regimens were 50 mg of hydrocortisone every 6 hours or 100 mg of hydrocortisone every 8 hours.

Data Collection
Primary study outcomes were survival to hospital discharge and at one-year in patients trialed on NIPPV prior to IMV and in patients only treated with IMV. As a secondary outcome, we investigated possible determinants of the primary outcomes including: demographic data, baseline patient characteristics, ILD subtype, ventilator settings, echocardiography and ICU level interventions. We also stratified patients based on the presence or absence of pulmonary hypertension to see if this had any impact on primary or secondary outcomes. Cardiothoracic imaging, PFTs, lung histopathology, lab values, ventilator settings, and patient history were collected from EMR. Mechanical ventilation settings were recorded at the end of a 24-hour time interval from its first initiation during their hospital admission. For patients transferred from another institution, mechanical ventilator settings were recorded from: (1) their documented ventilator settings prior to transfer that was closest to the 24 hour time interval from its first initiation, or (2) ventilator settings on arrival to our institution provided they were on IMV for greater than 24 hours and had no ventilator settings available prior to transfer.

Statistical analysis
Results are presented as median and interquartile range (IQR) for quantitative variables and frequencies and percentages for qualitative variables. Fisher's exact tests were used for binominal variables and variables with more than two labels as appropriate. One-year survival was determined from the date of onset from ARF. The Kaplan-Meier survival curves were plotted for survival data with statistical evaluation through Mantel-Cox log-rank statistics. Threshold for statistical significance was p < 0.05.

Results
We identified 106 potential ILD patients who were admitted to our health system between January 2014 and October 2018, with ARF and who required IMV. Fifty-two patients were excluded: 33 patients did not have enough data to support the ILD diagnosis; 16 patients were previously LTx recipients; 2 patients had chronic tracheostomies; and 1 patient had experienced a surgical related ARF. The remaining 54 patients fulfilled inclusion criteria and were included in the analysis (Fig. 1).

General Baseline Characteristics
Tables 1 and 2 provides a detailed description of patient characteristics. The mean age of patients was 65 years and males represented 55.6% of the total cohort. An ILD diagnosis for greater than 1 year was seen in 55.6% of patients. Connective tissue disease (CTD) was the most commonly identified cause of the ILD diagnosis, representing 31.5% of the patients and idiopathic pulmonary fibrosis (IPF) was the second most common diagnosis, representing 14.8% of the patients. Chronic steroid use was found in 56% of the patients and 40% of the patients were on other chronic immunosuppressants. 55.6% of patients were transferred from another institution. Home oxygen supplementation was used in 46.3% of patients.  Assist control/volume control ventilation was the most commonly used ventilation mode.
Norepinephrine was the most common first vasopressor utilized in patients with shock physiology.
Tracheostomy was performed in 18.5% of patients, which was equally balanced in both cohorts. LTx was documented to be considered in 11 (22%) patients, 2 of whom received LTx and were both alive at one year. 6 patients were supported with ECMO: two died, two were weaned off and ultimately discharged, and two were the recipients of LTx. Palliative care consultation was performed in 48.8% of patients.

Mortality and determinants
Of the 54 ILD patients with ARF treated with IMV, 20 (37%) survived to hospital discharge and 10 (18.5%) were alive at one-year (Table 3). No mortality difference was observed in patients who were trialed on NIPPV prior to IMV versus those only treated with IMV. Significant increases of in-hospital mortality were found in patients with a higher BMI, the IPF subtype, vasopressor use, and stress or pulse steroids administration. Patients already on supplemental oxygenation at home and with known ILD diagnosis of greater than one year were found to have lower inpatient mortality. Analysis of mechanical ventilation parameters impacting in-hospital mortality is demonstrated in Table 4.
Ventilator parameters with increased PEEP and FiO2 settings and higher documented average airway pressures were associated with higher in-hospital mortality. Figure 2 demonstrates the significant mortality differences between these ventilator parameters using day of intubation as the starting point.  Significantly increased one-year mortality was found in patients with: the IPF subtype, no past medical history of CTD, presence of SVT, and vasopressor use (Table 5). In patients in the NIPPV prior to IMV cohort, an age greater than 65 years was associated with increased one-year mortality in comparison to those treated with only IMV. In the NIPPV prior to IMV cohort, those without pulmonary hypertension had higher one-year mortality. When splitting the patient cohort based on the presence or absence of PH, there was no significant mortality difference (see supplemental data- Table 6).

Discussion
This retrospective study describes the clinical course, ICU management, and outcome of 54 ILD patients requiring IMV for ARF at a tertiary-referral institution. Both in-hospital and one -year mortality was high, and no difference was found in the primary outcome. Higher ventilator support requirements, vasopressor use, high dose steroids, and the IPF subtype were associated with worse in hospital mortality. The IPF subtype, ILD exacerbation as cause of ARF, no past history of CTDs, presence of SVT, vasopressor use were independent predictors of one-year mortality.
One prior study has described the in-hospital and one-year mortality of ILD patients experiencing ARF, requiring IMV, as 53% and 59%, respectively (6). Our study demonstrated a 63% in-hospital mortality and 82.5% one-year mortality. Differences in mortality rates are likely in part to the fact that our study had a large percentage of patients that were transferred from an outside hospital; which may reflect patients with worse prognosis having failed to respond to initial medical care prior to transfer.
NIPPV trialed prior to IMV was found to have higher mortality in the study by Fernandez-Perez(6).
Others have suggested that patients receiving an NIPPV trial prior to IMV have worse clinical outcomes (15). Patients trialed on NIPPV prior to IMV is not a standardized intervention and there is no recognized definition of what time duration of NIPPV trial is needed before it can be considered as failure. Our study found no significant difference for in-hospital mortality and one-year mortality in general between the two cohorts. With the exception of sarcoidosis patients with ILD-ARF, in-patient mortality was high and more studies are needed to validate this finding regarding sarcoid patients experiencing ARF. An age of 65 years or older had a worse one-year mortality in patients trialed on NIPPV prior to IMV in comparison to those undergoing only IMV. This proposes an ethical dilemma for critical care physicians because, in general, mortality is high for ILD-ARF patients who fail a NIPPV trial and it is unclear if IMV is even worthwhile after NIPPV failure. IMV is often an intervention that has been considered futile if transplantation is not an option for patients with advanced underlying ILD; this has most commonly been discussed in IPF patients (6,(16)(17)(18). However, NIPPV trial prior to IMV initiation may identify a subset of patients who are responders and avoid need for IMV altogether.
Fernandez-Perez described that the ventilator settings which correlated with increased mortality were higher PEEP, lower tidal volume, and higher FiO2 (6). They also found higher documented mean, plateau, and peak airway pressures and lower PaO2/FiO2 ratios were associated with increased mortality. Our study supported that higher PEEP settings and higher FiO2 were associated with increased in-hospital mortality. These findings were also associated with decreased one-year survival starting from day of intubation. This suggests that ARF-ILD patients requiring IMV are subject to increased risks of barotrauma, volutrauma, and cellular injury related to hyperoxia/free radical damage. Statistical differences in tidal volume settings, recorded peak airway pressures, and recorded PaO2/FiO2 ratios between our study and the Fernandez-Perez study is likely related to power. Additionally, lower tidal volume requirements to maintain plateau pressures suggests higher disease burden and poor compliance that is prognostically concerning. It is reasonable for physicians caring for ILD-ARF patients requiring IMV to adopt mechanical ventilator strategies from ARDS protocols that incorporate using the lowest possible tidal volumes and PEEP settings (19)(20)(21). FiO2 should also be aggressively titrated to the lowest possible setting to avoid hyperoxic acute lung injury (22).
PH and its effect on the primary study outcome was unclear. Our study did not support that PH impacts mortality or morbidity in ILD-ARF. Saydain et al also found no difference in systolic pulmonary artery pressures between survivors and non-survivors (17). Zafrani et al did find that PH was a determinant for in-hospital and one year mortality (7). IMV in patients with severe PH should be avoided if possible due to potential complications of hemodynamic instability (23 SVT, aside from sinus tachycardia, was associated with higher in-hospital (p = 0.06) and one-year mortality (p = 0.037). There is limited data on the epidemiology of AAs in ILD patients. Studies in IPF have demonstrated that SVTs are common (24,25). SVT can be difficult to manage in ILD-ARF because: (1) there is no standardized approach to management in this cohort and (2) it is difficult to assess if SVT is the primary cause of ARF or if the SVT is secondary to the underlying etiology of ARF.

Limitations
Our study has several limitations that could have influenced our findings. Our study is limited in its retrospective study design and by low power, thus necessitating the need for further investigations to validate our discovered associations. Due to reliance of the diagnosis of ILD and ARF from ICD coding and EMR investigation, it is possible that we missed patients who were not coded properly. of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis, including and especially any adverse effects.

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.