Psychometric Properties of the Patient Dignity Inventory in a Longitudinal Cohort of Acute Respiratory Failure Survivors

Dr. Lydia E. FedericoResident Research Spotlight

Critical Care Medicine, 2026View Publication →

Study Overview

More than half of patients who survive acute respiratory failure (ARF) go on to develop post-intensive care syndrome (PICS), a lasting mix of physical, psychologic, and cognitive impairments. In other populations living with long-term impairment, such as patients receiving palliative care, interventions centered on preserving dignity have improved anxiety, depression, existential distress, and relationships with the care team. Yet dignity has received very little formal study among ICU survivors, in part because there was no validated way to measure it over time in this group.

The Patient Dignity Inventory (PDI) is a 25-question, Likert-based instrument (scored 25 to 125, with lower scores indicating better dignity conservation) originally developed in palliative care. Dr. Lydia Federico and colleagues had previously shown that the PDI correlates strongly with quality of life and disability measures in ARF survivors. In this study, they set out to complete the validation: testing internal consistency, generalizability, and internal structure over time, establishing a minimal clinically important difference (MCID), and trimming the instrument to make it easier to use.

The team performed a secondary analysis of a prospective cohort of adults mechanically ventilated for at least 24 hours in four University of Pennsylvania hospital ICUs (June 2022 to April 2023). Patients completed the PDI within one month of ICU discharge and again at 3, 6, and 12 months. Of 967 eligible patients, 103 enrolled, and 72 (70%) completed the 12-month assessment.

Key Findings

The PDI showed excellent internal consistency at every timepoint (Cronbach's alpha 0.94 to 0.96), and removing any single question did not change this.
In the generalizability analysis, 75% of score variance was explained by differences between patients, 25% by timepoint, and essentially none by the individual questions, yielding a phi coefficient of 0.93 (excellent consistency). In other words, the PDI reliably distinguishes one survivor from another while still picking up change over time.
The sources of dignity-related distress shifted across the year after discharge. Early on, physical, psychologic, and cognitive impairments dominated (49% of variance at baseline). By 12 months, the leading factor reflected existential, interpersonal, and spiritual concerns (45% of variance).
A telling example: the item "Feeling like I am no longer who I was" loaded most strongly with physical and cognitive impairments at baseline (0.76), but by 12 months it aligned with existential and interpersonal concerns (0.73). Loss of identity starts as a consequence of the body and becomes a question of self.
The MCID for the full PDI was 6.61 points, averaging an anchor-based estimate of 9.19 and a distribution-based estimate of 4.03.
Iterative item reduction produced a 14-question version that preserved good to excellent reliability (Cronbach's alpha 0.89 to 0.93; phi 0.92) and stable factor structure, with an MCID of 3.65 points.

Clinical Significance

This is the first longitudinal assessment of the PDI among critical illness survivors, and it positions dignity as a distinct and measurable dimension of recovery after the ICU. The number of factors identified here (two to three) differed from prior PDI studies in palliative care (five) and acute critical illness (four), which suggests that ARF survivors face their own particular threats to dignity. That is a useful reminder that a validated instrument from one population should not be assumed to behave the same way in another.

The evolving factor structure also points to how care might be timed. In the weeks after discharge, dignity-centered support may be best delivered through concrete work on the sequelae of critical illness, such as physical and cognitive rehabilitation. Later in the first year, the emphasis may need to shift toward community reintegration, relationships, and spiritual support. With an MCID in hand and a shorter 14-item version available, investigators now have the tools to design and evaluate interventions that target these needs across the recovery trajectory.

For residents, the practical takeaway is that recovery from critical illness is not only about strength, breathing, and cognition. Months after discharge, survivors may be wrestling with who they are and how they fit back into their lives. Asking about those concerns in post-ICU and primary care follow-up can surface distress that standard functional assessments miss.

Limitations include 21% loss to follow-up at 12 months, possible recall bias from baseline surveys collected up to one month after ICU discharge, and a single-health-system cohort, so validation in larger and more diverse populations is still needed.

📏 Methods Highlight

Minimal Clinically Important Difference (MCID)

A change in a score can be statistically significant without mattering to the patient. The MCID is the smallest change in a score that patients would perceive as meaningful, and it is what turns a questionnaire into something you can use to judge whether an intervention actually worked.

There are two main ways to estimate it, each with its own weakness:

Anchor-based: link score changes to a change in a related outcome that already has an accepted meaningful threshold (the "anchor"). Here, the anchor was the EQ-5D utility score, using patients whose EQ-5D changed by at least 0.11. This grounds the MCID in patient experience, but it is only as good as the anchor.
Distribution-based: use the statistical spread of the scores themselves. Here, the authors averaged the standard error of the mean, 0.3 × SD, and 5% of the baseline score range. This is independent of any anchor, but it says nothing directly about clinical meaning and is sensitive to sample variability.

MCID = (anchor estimate + distribution estimate) ÷ 2
= (9.19 + 4.03) ÷ 2 = 6.61

Averaging the two approaches balances clinical and statistical perspectives, producing a more robust estimate than either alone.

Tip:When a trial reports a "significant improvement" on a patient-reported outcome, compare the effect size to the instrument's MCID. A difference smaller than the MCID may be real but not meaningful to patients.