Neonatal Acute Kidney Injury: Recognition, Diagnosis, Prevention & Follow-up

What to Hold in Mind at the Bedside

A newborn's creatinine must be read as a curve, not a point. In the first days of life it partly reflects the mother's concentration through placental equilibration; its expected fall varies with gestational age; low muscle mass blunts its sensitivity; and a positive fluid balance can dilute it. Apply neonatal modified KDIGO (Kidney Disease: Improving Global Outcomes) using the lowest previous creatinine and the urine-output criteria — but check data quality and clinical context before you trust the stage.

When AKI is suspected, search in parallel for the mechanisms that produce it: low perfusion, fluid accumulation, sepsis or inflammation, obstruction or vascular disease, congenital kidney and urinary-tract disease, and nephrotoxin exposure. Management is supportive and cause-directed. Resist the two reflexes that do the most harm in the newborn — the automatic fluid bolus given without a phenotype that predicts fluid responsiveness, and the automatic diuretic given as if it treated the injury rather than the symptom. Finally, document the AKI episode and build the kidney-health follow-up plan before discharge, because the risk outlives the hospitalization.

Throughout this guide, statements carry an evidence label and a calibrated verb — Recommend (guideline/consensus or established standard), Consider (reasonable in a defined context; evidence incomplete), Do not routinely use (insufficient evidence or potential harm), and Investigational (research-only or locally validated). Effect estimates are given with their population and, where they drive a conclusion, their confidence interval. Association is not restated as causation.

Why the Neonatal Kidney Is Different

Start with the endowment. Nephrogenesis — the formation of new nephrons — begins early in gestation and, in a normally grown fetus, continues to roughly 34–36 weeks. There is no meaningful nephron formation after term. A neonate born extremely preterm therefore arrives with an interrupted nephron count: some postnatal nephrogenesis may continue briefly, but it proceeds under extrauterine stress and can be abnormal, and it does not fully recover the trajectory that intrauterine life would have provided. Low nephron endowment is best understood as a life-course susceptibility factor — a smaller filtration reserve to draw down over decades — and not as a synonym for current chronic kidney disease (CKD).

Now the postnatal transition. At birth, renal blood flow and glomerular filtration rate (GFR) are low; both climb steeply over the first weeks as renal vascular resistance falls and cardiac output redistributes toward the kidney. Tubular function is immature in parallel: concentrating ability, sodium handling, bicarbonate reclamation, and pressure-flow autoregulation are all still maturing. The practical consequence is a kidney with little margin. When hypoxemia, hypotension, systemic inflammation, vasoactive drug exposure, and nephrotoxins act on an organ whose reserve is already small and whose autoregulation is blunted, the same insult that a mature kidney would absorb can push a neonatal kidney into functional failure. This is the physiologic reason risk-based surveillance, rather than reactive creatinine-watching, is the right posture in the highest-acuity infants.

What Neonatal AKI Is — and What Current Tests Actually Measure

Define AKI as an abrupt disturbance in kidney function with systemic consequences. That definition is deliberately broad, because the useful clinical model separates six things that the word "AKI" tends to blur together. Susceptibility is the immature or reduced renal reserve the infant brought to the bedside. Exposure and stress are the insults acting on it — hypoperfusion, hypoxia, inflammation, toxic exposure, obstruction. Injury is cellular and tissue damage, which may precede any conventional change in filtration. Dysfunction is the measurable change in filtration and/or urine output that our criteria capture. Consequences are the downstream electrolyte, acid-base, volume, nutritional, drug-clearance, and multisystem effects. Reading these as a sequence — susceptibility → exposure → injury → dysfunction → consequences — is what lets you act during the window in which injury has begun but dysfunction has not yet declared itself.

Be precise about what each test in this sequence measures. Serum creatinine and urine output are functional markers: they report filtration and, indirectly, perfusion and tubular handling — late in the causal chain. Urinary neutrophil gelatinase-associated lipocalin (uNGAL) is an injury-associated marker that can rise earlier, when tubular cells are stressed. Cystatin C is predominantly a filtration marker, like creatinine, but read through a different lens because it is far less dependent on muscle mass. None of these is a biopsy, and none names an etiology. A number tells you that filtration has changed or that tubules are stressed; it does not tell you why, and it does not substitute for the parallel search for mechanism described below.

How Common, and Why It Matters

Neonatal AKI occurs in roughly one-third of critically ill neonates, but that pooled figure hides enormous variation by gestation, illness, surgery, and — importantly — by the definition applied. The anchor is AWAKEN (the Assessment of Worldwide Acute Kidney Injury Epidemiology in Neonates study), a multicentre, multinational, observational cohort across 24 NICUs, in which 605 of 2,022 neonates (29.9%) developed AKI. After adjustment for potential confounders, AKI was independently associated with higher mortality — adjusted odds ratio 4.6 (95% confidence interval 2.5–8.3) — and with longer hospitalization.³ Read that association carefully: it is robust and it survives adjustment, but a cohort of this design cannot prove that every AKI episode is directly causal of death rather than a marker of sicker physiology. Both readings can be true at once, and the clinical implication — take neonatal AKI seriously as a prognostic signal — does not depend on resolving the causal question.

The outcomes worth discussing extend well beyond mortality: prolonged mechanical ventilation, bronchopulmonary dysplasia, length of stay, recurrent AKI within the admission, and — over the longer term — hypertension and CKD. Here the causal language must stay disciplined. Much of the association between neonatal AKI and later kidney or cardiovascular disease is entangled with prematurity and critical illness themselves, and disentangling the independent contribution of the AKI episode is precisely one of the field's open questions (see Uncertainties). What is actionable now is not a claim about mechanism but a claim about surveillance: infants who had neonatal AKI belong in a kidney-health follow-up pathway.

The Creatinine Problem — Read the Curve, Not One Number

More clinical error in neonatal AKI comes from misreading creatinine than from any other single source. Six features make the newborn value harder to interpret than an adult's, and each has a direct bedside implication.

Maternal signal

At birth, neonatal serum creatinine partly reflects the maternal concentration across an equilibrated placenta. A first value is therefore not a stable neonatal baseline — it is closer to a shared value that must fall as the newborn's own kidneys take over. Staging a ratio off a day-of-life-one creatinine risks anchoring to the mother's physiology, not the baby's.

Expected postnatal decline

In a healthy newborn, creatinine falls over the first days to weeks; the timing and slope of that fall vary with gestational age and illness, and more premature infants tend to reach their nadir later, sometimes after a transient early rise. This variability is exactly why this guide does not present a "normal fall" calculator. Do not routinely use a universal expected-decline curve to declare a value normal or abnormal: a valid tool would require a gestation- and assay-specific reference dataset, and publishing a synthetic curve invites false reassurance in the infants least able to afford it.

Low creatinine generation

Creatinine is generated from muscle. A neonate — particularly a growth-restricted or extremely preterm one — has little muscle mass, so a clinically meaningful fall in GFR may move the creatinine only slightly. The number's insensitivity is the trap: a "reassuring" creatinine in a tiny infant can coexist with a real reduction in filtration.

Fluid dilution

A positive cumulative fluid balance expands the volume of distribution and can dilute creatinine, delaying recognition of a rise. Fluid-adjusted (fluid-corrected) creatinine is an adjunctive research calculation — the PENUT (Preterm Erythropoietin Neuroprotection Trial) secondary analysis showed that correcting creatinine for fluid balance reclassifies some premature neonates as having AKI who were missed by the uncorrected value⁷ — but it is Investigational for staging: it is not part of the neonatal modified KDIGO consensus definition and should not be used to assign a stage.

Failure to decline

In HIE cohorts, an alternative phenotype has emerged: a creatinine that simply fails to fall as expected, without necessarily meeting a conventional rise-based criterion, identifies infants at risk who standard staging can miss.⁸ Consider failure-to-decline as a clinically meaningful signal that should heighten surveillance — but present it as an emerging phenotype, not as a new formal KDIGO stage, because it has not been incorporated into consensus staging and its thresholds are not standardized.

Assay and sampling limitations

Finally, the measurement itself is a variable. Jaffé-based assays are subject to interference (bilirubin, chromogens) that enzymatic assays largely avoid, and the two are not interchangeable when you compute a small absolute change of 0.3 mg/dL. Small blood volumes constrain sampling frequency, and every draw contributes to iatrogenic anemia — a real balancing harm of intensive surveillance. Point-of-care values should be trended within the known performance of the specific platform, not mixed indiscriminately with central-laboratory values.

The Current Diagnostic Standard — Neonatal Modified KDIGO

Neonatal modified KDIGO is the current standard for staging, and it should be used — with its limitations held in view. The two most consequential modifications from adult and pediatric KDIGO are that the comparator is the lowest previous serum creatinine (not a fixed "baseline"), reflecting that the healthy neonatal trajectory is downward, and that the urine-output thresholds are neonatal-specific.

*Urine-output criteria are frequently assessed over 24 hours in neonatal studies, because reliable hour-by-hour collection is often unavailable. The optimal neonatal threshold and duration remain unsettled — preserve the exact local protocol and the source definition whenever a shorter or longer window is used.

Several operating rules keep the staging honest. Use the lowest previous serum creatinine as the comparator under the neonatal modification. Stage by the most severe criterion met — if creatinine says stage 1 and urine output says stage 3, the infant is stage 3. Validate units and timestamps before computing any ratio, and do not convert an isolated value into a ratio when no eligible prior value exists within 7 days. Recognize that stage 0 does not mean the absence of kidney stress or injury — it means no KDIGO criterion was met from the available data, which is a different and weaker claim. KRT started for AKI automatically meets stage 3. And note the origin of the absolute threshold: a creatinine ≥2.5 mg/dL was chosen because it corresponds approximately to an estimated GFR (eGFR) below 10 mL/min/1.73 m² — but eGFR is not required, and should not be used, to assign an AKI stage.

Data Quality Before Staging

A stage computed from bad data is worse than no stage, because it carries false authority. Walk this checklist before you commit a neonatal modified KDIGO stage to the chart or a research database.

• Confirm gestational age, postnatal age, birth weight, current weight, and the weight used for urine normalization — the denominator determines the urine-output stage.
• Confirm the creatinine units and the assay/platform (Jaffé vs enzymatic; point-of-care vs central lab). Do not compute a 0.3 mg/dL change across assays.
• Exclude duplicate or implausible timestamps; a mis-stamped sample can invent or erase a rise.
• Establish whether a lower prior value exists within 7 days — the eligible comparator for the ratio criteria.
• Assess maternal creatinine relevance in the first postnatal days before anchoring a ratio to a birth value.
• Verify urine-collection method: catheter vs diaper weighing, leakage, stool contamination, catheter obstruction, and undocumented voids all corrupt the rate.
• Review cumulative intake/output and weight change for a fluid-dilution effect on creatinine.
• Determine whether diuretics, vasopressors, KRT, ECMO, or recent surgery alter interpretation of either marker.

Risk-First Recognition

Because creatinine is a late marker, recognition should begin with risk, not with a number. The point of enumerating risk is not to compute a probability — this guide deliberately avoids a universal additive risk score, which would need external validation it does not yet have — but to trigger surveillance in the infants who most need it. Group the risk domains so nothing is missed on rounds.

The operating principle is to let any high-risk context move the infant onto a surveillance pathway — scheduled creatinine, deliberate urine-output measurement, and a running tally of nephrotoxin days — rather than waiting for a threshold to be breached and then reacting.

Diagnostic Approach When AKI Is Suspected

When a signal appears — a rising or failing-to-fall creatinine, low urine output, or simply a high-risk exposure — work through a phenotype-based sequence rather than reaching for a reflex intervention.

Hold a differential across five categories — pre-renal/hemodynamic, intrinsic, postrenal, vascular, and congenital/genetic — and resist a common shorthand: labelling an episode "pre-renal" does not make it benign. Sustained hypoperfusion can coexist with, and progress to, structural tubular injury; "pre-renal" describes a mechanism, not a reassurance.

Biomarkers and Measured Function

The appeal of a biomarker in the newborn is obvious: creatinine is late, muscle-dependent, and dilutable, so a marker that rises earlier or reads filtration without the muscle-mass confound would be genuinely useful. The honest state of the evidence is that these markers add information but do not yet come with the validated, universal thresholds that would let them stand alone.

Serum cystatin C

Cystatin C is filtered by the glomerulus, is far less dependent on muscle mass than creatinine, and does not substantially cross the placenta — three properties that make it attractive in neonates. A large multicenter cohort proposed cystatin C–based neonatal AKI criteria and reported that they detected substantially more cases than the modified creatinine criteria, in infants who were themselves at elevated risk of in-hospital mortality.⁹ The caveats are real: values remain age- and context-dependent, and neonatal reference intervals and AKI definitions are not fully standardized. Do not routinely use a cystatin C change to assign a neonatal KDIGO stage — the staging system is creatinine- and urine-output-based, and substituting a different analyte into it is not validated.

Urinary NGAL

Urinary NGAL (neutrophil gelatinase-associated lipocalin) can rise earlier than creatinine after tubular injury and has shown promising diagnostic performance in preterm neonates. But its baseline varies widely by gestational age, birth weight, assay, inflammation, and clinical context, and a 2025 systematic review and meta-analysis of preterm neonates found a wide non-AKI range.¹¹ The direct implication: Do not publish or adopt a single universal diagnostic cutoff. A local program may reasonably Consider assay- and population-validated thresholds as an adjunct to surveillance — and uNGAL has been used specifically to screen for nephrotoxic-medication-associated AKI in neonates¹⁴ — but as a supplement to, not a replacement for, creatinine and urine-output monitoring.

Other biomarkers

Kidney injury molecule-1 (KIM-1), interleukin-18 (IL-18), liver-type fatty acid-binding protein (L-FABP), the cell-cycle-arrest pair tissue inhibitor of metalloproteinases-2 and insulin-like growth factor-binding protein 7 (TIMP-2·IGFBP7), osteopontin, epidermal growth factor (EGF), and uromodulin have all been studied, and a broader systematic review has examined serum and urinary biomarkers for predicting AKI in premature infants.¹⁰ Across this literature the pattern repeats: promising signal, insufficient standardization, and no actionable universal neonatal threshold. Treat them as research tools for now.

eGFR and measured GFR

Estimating equations for GFR are limited in neonates by rapidly changing physiology, uncertain steady state, and dependence on the biomarker and assay. Do not diagnose or stage AKI from an eGFR decline alone. Measured GFR using an exogenous filtration marker is a specialized tool for defined questions, not a routine acute bedside test.

Prevention and Nephrotoxin Stewardship

Prevention is where the most convincing actionable evidence in neonatal AKI sits, because a large share of neonatal AKI is nephrotoxin-associated and therefore, in principle, modifiable. The unit-level work is unglamorous and effective: name AKI clinical champions across neonatology, nursing, pharmacy, and nephrology; adopt a written neonatal AKI definition and documentation standard; identify high-risk infants prospectively; standardize urine measurement and creatinine surveillance; implement nephrotoxin alerts (or a manual list where the electronic record cannot support them); carry AKI into handoffs, rounds, and discharge summaries; and audit exposure days, AKI events, screening adherence, and balancing measures such as blood sampling and transfusion.

The Baby NINJA program

Baby NINJA (Nephrotoxic Injury Negated by Just-in-time Action) is the concrete exemplar. In a single-center, level-IV NICU quality-improvement program, infants were screened for high-risk exposure — ≥3 nephrotoxic medications within 24 hours, or ≥4 calendar days of intravenous aminoglycoside — and those who met the trigger had a daily serum creatinine until 2 days after exposure ended or AKI resolved, whichever came later. Over the program, high-risk exposure fell from 16.4 to 9.6 per 1,000 patient-days, the proportion of nephrotoxin-associated AKI fell from 30.9% to 11.0%, and AKI intensity (days in AKI) also fell.¹² Implementation strategies for spreading the bundle have since been described.¹³

State the limitations as plainly as the result: this is a nonrandomized, single-center quality-improvement study, its effect is a bundle effect rather than an isolated intervention, and its medication definitions and implementation are local. It is compelling implementation evidence — not a multicenter randomized trial — and that distinction should survive editing. One design note carries over to any adopting unit: Do not hard-code a universal nephrotoxin list into clinical logic without pharmacy ownership. Store the list in versioned content that local teams control, so it can be updated as formularies and evidence change, and so an unavoidable but necessary therapy is never miscoded as an error.

Management After AKI Is Recognized

Organize management by physiologic objective, not by a rigid sequence. The four objectives below run in parallel, and the judgment is in balancing them against each other in a specific infant.

Restore effective perfusion without causing fluid harm

Treat the cause of shock and impaired oxygen delivery. A fluid bolus is a therapeutic trial, justified only when the clinical phenotype suggests the infant might be fluid-responsive — not a reflex response to oliguria or to a rising creatinine. Reassess after every intervention, and remember that edema or a positive cumulative balance does not prove adequate intravascular volume. Use hemodynamic and echocardiographic context wherever it is available. Do not give repeated boluses to an anuric, interstitially overloaded infant on the theory that "the kidney needs volume."

Manage fluid accumulation

Track weight, cumulative intake/output, insensible-loss assumptions, sodium exposure, respiratory status, and the degree to which fluid is displacing nutrition. Diuretics may help manage established fluid overload in a diuretic-responsive infant, but they neither reverse kidney injury nor, by producing urine, prove that the kidney has recovered. Consider early escalation when fluid constraints begin to prevent ventilation, nutrition, medication delivery, or correction of metabolic complications — that functional threshold, not a creatinine value, is often the real trigger for KRT.

Electrolytes and acid-base

Hyperkalemia, hyponatremia and hypernatremia, metabolic acidosis, and disturbances of calcium, magnesium, and phosphate are neonatal emergencies that require cause-specific, institution-approved protocols. This guide deliberately publishes no treatment doses — neonatal electrolyte correction is dose-sensitive and context-dependent, and belongs in your unit's emergency protocols and neonatal drug references, not in a general reference page. Recommend that each of these link to a local, currently approved protocol.

Medication stewardship and nutrition

Stop nonessential nephrotoxins. Recalculate doses and intervals using neonatal pharmacology resources and TDM, and do not assume that a creatinine-based eGFR accurately predicts clearance for every drug — during rapidly changing function, and during KRT, dosing must be reassessed frequently. On nutrition, the goal is to preserve growth and protein-energy delivery: Do not reflexively restrict protein simply to lower urea, because catabolism and impaired growth carry their own harms in a neonate. Account for electrolyte burden, fluid concentration, catabolism, and losses during KRT, and involve a neonatal dietitian and pharmacist; numerical nutrition targets are context-dependent and should follow the neonatal renal-nutrition literature and local protocol¹⁹ rather than a fixed prescription embedded here.

Methylxanthines — Established Signal, Narrow Indications, Open Questions

Theophylline in perinatal asphyxia / HIE

The mechanistic rationale is coherent: during asphyxia, adenosine accumulates and, acting at renal A1 receptors, causes afferent arteriolar vasoconstriction that lowers GFR. Theophylline, a nonselective adenosine-receptor antagonist, is proposed to blunt that vasoconstriction. Systematic-review evidence supports a real kidney signal — a prophylactic single dose of theophylline (or aminophylline) can reduce severe renal dysfunction and AKI in term neonates with severe birth asphyxia.¹⁵¹⁶ Two disciplines of interpretation matter. First, separate the kidney outcome from survival and neurodevelopmental outcomes — the renal benefit does not automatically translate into better long-term brain outcomes, and the trials were not powered to settle that. Second, applicability in the current therapeutic-hypothermia era, along with safety, drug interactions, and dosing, depends on your local neonatal protocol.

Caffeine

Observational data in preterm infants suggest a lower AKI incidence with early caffeine exposure, but confounding by indication and survivor/treatment bias remain plausible explanations, so this is an association, not evidence of a protective effect. Any newer systematic-review evidence should be weighed only after its full peer-reviewed publication details and included studies can be verified — a search-engine preview is not a citable source. Do not start caffeine solely for kidney protection outside a trial or an approved neonatal indication (such as apnea of prematurity, for which caffeine is separately indicated).

Kidney Replacement Therapy

KRT in the neonate is a high-level escalation decision; the technical prescription (dwell volumes, clearance dose, anticoagulation, blood priming, ultrafiltration targets) belongs to the treating team and a dedicated technical resource, not to a general guide. Two things are worth fixing firmly here: the indications are clinical and trajectory-based, and modality choice is driven by the infant's physiology and the unit's capability.

Indications are clinical, not creatinine-only

Escalate for refractory hyperkalemia or another dangerous electrolyte disturbance; refractory metabolic acidosis; symptomatic or progressive fluid overload compromising organ support; uremic complications; oliguria or anuria with an inability to deliver necessary nutrition, blood products, or medications; and selected intoxications or inborn metabolic disorders. Notice that none of these is a creatinine number — the functional inability to run the rest of the infant's care is usually the real trigger, which is why a creatinine threshold should never be the gate that delays KRT.

The 2020 update of the International Society for Peritoneal Dialysis (ISPD) guidelines for peritoneal dialysis in AKI covers both adults and pediatrics and is the reference point for PD practice.¹⁷ On the device side, the I-KID (Infant Kidney Dialysis and Ultrafiltration) stepped-wedge cluster-randomized study compared PD, continuous venovenous hemofiltration, and a novel infant-dedicated hemodialysis device — part of a broader move toward equipment designed for neonatal circuit volumes rather than adapted from adult hardware.¹⁸ Do not take dwell volume, clearance dose, or ultrafiltration targets from a general guide; those are set at the bedside with nephrology.

Recovery, Recurrent AKI, and Discharge Documentation

Define recovery carefully. Recovery is a return toward a gestation- and postnatal-age-appropriate trajectory, resolution of the metabolic and volume complications, and stable fluid status — not merely a creatinine that has landed in a "normal" range. A creatinine that looks normal in a low-muscle-mass infant can still sit on an abnormal trajectory, and recurrent AKI within the admission is common in the most premature infants. The transition out of the NICU is exactly where kidney risk tends to be dropped, so the defense is a structured, portable record that follows the infant.

Post-NICU Kidney Health Monitoring

The follow-up architecture below reproduces the 2024 modified-Delphi consensus statement on kidney health monitoring in NICU graduates.²⁰ Its evidence level should be stated as the consensus itself reports it — predominantly expert consensus (largely level 3 or 5), not derived from randomized trials — and a related expanded discussion elaborates the reasoning for critically ill term and late-preterm infants after AKI.²¹ The longitudinal signal that motivates it is real if imperfectly causal: preterm birth is associated with a higher risk of CKD from childhood into mid-adulthood,²² and the FANCY cohort found more renal dysfunction in early childhood among very-low-birth-weight children who had neonatal AKI than in those who did not.²³

Universal discharge kidney evaluation for at-risk infants

For preterm infants born before 34 weeks, critically ill infants who had AKI, and infants with critical cardiac disease, the consensus recommends, at discharge: a properly obtained blood pressure, a serum creatinine, and kidney-health education with explicit follow-up ownership. Evidence of kidney disease at discharge — a creatinine ≥0.5 mg/dL, a blood pressure above the 95th percentile, treated hypertension, nephrocalcinosis, or CAKUT — should trigger pediatric nephrology follow-up according to local guidance.

Consensus follow-up router

Where two categories apply, follow the earlier and more intensive recommendation and record every qualifying reason; the high-risk critical-cardiac pathway supersedes the ordinary at-risk cardiac pathway. A "comprehensive kidney-health assessment," per the consensus, comprises interval history and exposures, growth, a properly measured blood pressure, a serum creatinine with context-appropriate assessment of filtration, urinalysis and urine protein/albumin assessment where feasible and indicated, and kidney imaging when clinically indicated. Where the consensus deliberately leaves components flexible, do not invent a fixed universal panel — adapt to the local standard.

Implementation in Lower-Resource NICUs

Competent neonatal AKI care does not require advanced biomarkers or infant-dedicated machines, and it should not be framed as though it did — a framing that would let a well-equipped unit feel exempt from the cheap, high-yield work and leave a resource-limited unit feeling it cannot begin. For the Philippines and similar settings, the standard holds while the tooling adapts.

Use paper or spreadsheet triggers where no electronic alert exists. Standardize diaper weighing, and apply the convention that 1 g of weight gain approximates 1 mL of urine only where the scale precision and contamination limits are understood. Prioritize serial creatinine for the highest-risk exposure windows when blood volume is constrained, rather than sampling everyone equally. Maintain a pharmacist-owned local nephrotoxin list. Establish referral or teleconsult criteria with the nearest pediatric nephrology service before you need them urgently. Ensure discharge summaries carry the peak stage, the last creatinine, the blood pressure, the major exposures, and the follow-up date. And do not imply that urinary NGAL, cystatin C, infant-specific CKRT systems, or measured GFR are prerequisites for good care — they are refinements, not entry requirements.

Uncertainties and Research Agenda

An honest guide names what it does not know, because manufactured certainty removes the clinician's ability to reason about the infant who sits between the trials. The following remain genuinely unsettled.

Twelve Questions From the Bedside