What Is the Sarnat Exam and How Is It Scored?

The Sarnat exam is a bedside neurological assessment used to evaluate newborns suspected of having brain injury from oxygen deprivation around the time of birth, a condition called hypoxic-ischemic encephalopathy (HIE). First developed in 1976, it grades the severity of encephalopathy into three stages based on observable signs like alertness, muscle tone, reflexes, and autonomic function. The severity grade directly shapes treatment decisions, particularly whether a newborn qualifies for therapeutic hypothermia (cooling therapy), and it remains one of the most widely used clinical tools in neonatal neurology.

What the Exam Actually Measures

The modified version of the Sarnat exam, which is the form most hospitals use today, evaluates six categories of neurological function. Each category is scored on a four-point scale from normal (scored as 0) through mild (1), moderate (2), and severe (3). The six categories are level of consciousness, spontaneous activity, muscle tone, posture, primitive reflexes, and autonomic system activity.¹

The primitive reflexes score uses the worst result between the Moro reflex (the startle response when a baby feels like it’s falling) and the sucking reflex. Similarly, the autonomic score takes the worst result among heart rate, pupil responses, and breathing pattern. This “worst of” approach means the exam captures the most concerning sign in each domain rather than averaging them out.

In practice, the examiner observes the baby and interacts with it in specific ways: checking whether the infant tracks light or sound, assessing how the arms and legs resist passive movement, triggering the Moro reflex, watching for abnormal posturing, and noting whether breathing and heart rate are stable. The whole process happens at the bedside without any equipment beyond a stethoscope and a penlight.

The Three Stages of Severity

The original 1976 classification by Sarnat and Sarnat described three stages based on clinical observation and EEG patterns in newborns who had experienced fetal distress.² Those three stages remain the backbone of the exam today, though the details of how each stage is defined have been refined over the decades.³

Stage 1 (mild): The baby is hyperalert or jittery, with normal or slightly increased muscle tone. Reflexes may be exaggerated, and the infant might have dilated pupils. Seizures are absent, and the EEG is usually normal. Most infants in this stage recover without lasting problems.
Stage 2 (moderate): The baby appears lethargic or obtunded, with reduced muscle tone and strong flexion posturing in the hands and feet. Seizures are common, and the EEG often shows abnormal periodic patterns. The original work found that infants who stayed in stage 2 for fewer than five days generally appeared normal in later infancy, while those who remained in stage 2 for more than seven days, or whose EEG did not normalize, were at risk for lasting neurological problems or death.²
Stage 3 (severe): The baby is stuporous or comatose, with flaccid tone and absent reflexes. Pupils may be fixed, and the infant often requires mechanical ventilation. The EEG shows severe suppression or a burst-suppression pattern. Outcomes in stage 3 are the most serious, with a high risk of death or major disability.

Stage 2 is particularly important in clinical decision-making because it represents the threshold at which therapeutic hypothermia is offered. Cooling therapy has been shown to improve outcomes in moderate and severe encephalopathy, so distinguishing mild from moderate encephalopathy quickly and accurately is one of the exam’s most consequential jobs.

Why Repeat Exams Matter More Than a Single Snapshot

Encephalopathy is not a static condition. A baby’s neurological state can change substantially in the first hours and days after birth, and a single exam at one time point doesn’t capture the full trajectory. This is why clinicians perform the Sarnat exam repeatedly, often at standard intervals during and after cooling therapy.

A secondary analysis of the HEAL trial, which enrolled infants with moderate or severe HIE, looked at both the qualifying Sarnat exam (performed early, to decide whether to start cooling) and a rewarming Sarnat exam (performed after cooling was completed). Both exams predicted outcomes at age two, but the pattern of change between them was especially informative. Infants whose encephalopathy remained at the same severity or worsened by the rewarming exam had substantially higher odds of death or disability than infants whose exam improved, even when the initial qualifying exam was severe.⁴

To put that in concrete terms: infants who still had moderate encephalopathy at the rewarming exam had roughly six times the odds of a poor outcome at two years compared to those who improved, and infants with persistent severe encephalopathy had about 50 times the odds.⁴ The trajectory, in other words, tells you more than the starting point alone. A baby who begins at severe but improves to moderate may do better than a baby who begins at moderate and stays there.

How Reliable Is It Between Different Examiners?

Any clinical assessment that depends on human judgment raises questions about consistency. If two clinicians examine the same baby, will they arrive at the same stage? The answer is mostly yes, but with some important caveats.

When experienced examiners performed the modified Sarnat exam on preterm infants born between 32 and 36 weeks, agreement was very good to excellent across most of the six categories. The exceptions were the Moro reflex and, to a lesser degree, muscle tone, where examiners were more likely to disagree. The Moro reflex was particularly unreliable in babies born before 35 weeks, likely because the reflex itself is less mature and harder to interpret at younger gestational ages.⁵

Training matters here, and recent work has explored whether frontline providers such as neonatal nurses can be taught to perform the exam reliably. A pilot trial randomized nurses with no prior hands-on Sarnat exam experience into two training approaches: practice on a mannequin versus practice on a healthy term newborn. Both groups first received the same didactic teaching. Their exams were then video-recorded and scored by blinded neonatologists.⁶ The study reflects a broader push to extend the exam beyond neonatologists and fellows to nurses and other staff who are often the first to evaluate a baby, especially during nights and weekends.

The Thompson Score and How It Compares

The Sarnat exam is not the only scoring system for neonatal encephalopathy. The Thompson score is another widely used tool, and hospitals sometimes prefer it for its speed and simplicity. A head-to-head comparison of the two found a strong correlation between them, but meaningful differences in how they classified individual babies, especially early on.

In the first hour after birth, the modified Sarnat staging and the Thompson score disagreed most often in mild-to-moderate cases: the Thompson score classified about 40% of infants as mild, while the modified Sarnat classified only about 9% as mild and placed far more into the moderate category. Severe classifications were more consistent between the two systems from the start. As hours passed, agreement steadily improved, reaching strong concordance by the time of discharge.⁷

One practical difference: the Thompson score was consistently faster to complete, taking about 55 seconds less at every time point.⁷ That might not sound like much, but in a delivery room where multiple things are happening at once and a decision about cooling needs to be made within the first six hours of life, a quicker exam has real appeal. The tradeoff is that the Sarnat’s more granular categories give richer clinical information for tracking changes over time.

The early disagreement between the two systems in mild-to-moderate cases is particularly relevant because that is exactly the boundary where treatment eligibility is decided. A baby who scores as mild on the Thompson score but moderate on the modified Sarnat staging would qualify for cooling under Sarnat criteria but not under Thompson criteria. Hospitals that rely on one system over the other may not realize they are applying a slightly different threshold for intervention.

What Brain Imaging Adds to Clinical Staging

The Sarnat exam tells you how the brain is functioning at the moment of the exam, but it does not show what the injury actually looks like inside the brain. MRI fills that gap, and the two tools are complementary rather than redundant. Brain imaging typically occurs a few days after birth, once the baby is stable enough for the scan, and it reveals both the location and severity of structural damage.

In a large cohort of newborns with encephalopathy, just over half had visible brain injuries on MRI. The most commonly affected areas were the deep gray matter (including the thalamus and basal ganglia), white matter, and cortex.⁸ The thalamus was the single most frequently injured structure within the deep gray matter group, found in about 70% of those with basal ganglia or thalamic involvement.

What may surprise some clinicians is that MRI abnormalities are not limited to moderate and severe cases. Among infants classified as having mild HIE on clinical exam, a study found that about 61% still had some injury visible on MRI. Watershed injury, deep gray matter injury, and punctate white matter lesions were the most common patterns in the mild group.⁹ A separate smaller study similarly found MRI abnormalities in 87% of infants with mild encephalopathy, though only a portion of those injuries were specifically attributable to HIE.¹⁰

The implication is that the Sarnat exam alone can underestimate the extent of brain injury, especially at the mild end of the spectrum. A baby who looks clinically well enough to be classified as stage 1 may still carry structural damage that affects development later. This is part of why some centers now advocate for brain MRI in all infants with any degree of encephalopathy, not just those in stage 2 or 3.

When Prematurity Complicates the Picture

The Sarnat exam was designed for full-term newborns, and applying it to preterm babies introduces a layer of ambiguity. Normal neurological findings in a 33-week-old baby look different from those in a 40-week-old, because reflexes, tone, and alertness are all still maturing. A result that looks mildly abnormal on the Sarnat scale might simply reflect immaturity rather than injury.

Research on late preterm infants (born between 34 and 36 weeks) with encephalopathy found that they scored worse on the Moro reflex and the respiratory component of the autonomic assessment compared to their full-term counterparts with similar overall injury. However, their total Sarnat scores were not significantly different.¹¹ The conclusion was reassuring in one sense: prematurity alone is unlikely to push a baby into a higher severity category on the overall score. But the individual components can be misleading, and an examiner unfamiliar with how prematurity affects the Moro reflex might overinterpret a normal developmental finding as a sign of brain injury.

This is also where the inter-rater reliability data mentioned earlier becomes practically important. If the Moro reflex is already the most unreliable component among experienced examiners, and prematurity makes it even harder to score accurately, that particular item deserves careful interpretation in any baby born before 37 weeks. Some researchers have suggested that reflex components might need age-adjusted norms, though no widely adopted version of the exam has incorporated that yet.

Blood Markers as a Complement to Clinical Staging

Clinicians have long looked for blood tests that could objectively confirm or add precision to the clinical staging. One candidate is serum creatinine, which reflects kidney function and rises when the kidneys are damaged by the same oxygen deprivation that injures the brain. A study measuring serum creatinine in newborns with HIE found that levels rose progressively with each increase in Sarnat stage, measured on both day one and day seven of life.¹²

This kind of correlation is useful but should be understood in context. A blood test that moves in the same direction as clinical staging doesn’t replace the exam; it provides an independent data point. A baby with moderate encephalopathy on the Sarnat exam and markedly elevated creatinine may be at higher overall risk than one with the same Sarnat score but normal kidney function, because the creatinine suggests the oxygen deprivation was severe enough to damage multiple organs, not just the brain. Conversely, a high Sarnat score with normal creatinine might suggest that the encephalopathy has a cause other than global oxygen deprivation, prompting clinicians to look for metabolic or genetic explanations.

How the Exam Has Evolved Since 1976

The version of the exam in use today is not quite the same one Sarnat and Sarnat published nearly fifty years ago. The original was a descriptive framework that paired clinical observation with EEG findings over time. It was designed to characterize the natural history of encephalopathy, not necessarily to make rapid treatment decisions. Multiple adaptations followed, and the exam is now commonly referred to as the “modified Sarnat exam” to distinguish it from the original.³

The major shift came with the advent of therapeutic hypothermia in the 2000s. Suddenly the exam was no longer just prognostic; it became a gatekeeper for treatment. The modified versions were streamlined to make rapid, early classification feasible. Some versions dropped EEG as a required component for staging (since EEG may not be available in the first hours), while others quantified the scoring numerically so that each category produced a number that could be summed or tracked over time. The six-category, 0-to-3 scoring approach described earlier is one such quantified adaptation.¹

This numerical approach makes it easier to track a baby’s trajectory over time and to compare results across different examiners and hospitals, but it introduces its own tension. Distilling a complex neurological picture into a single number inevitably loses some nuance. Two babies with the same total score might have very different patterns of impairment: one might have severely depressed consciousness but relatively preserved tone, while the other shows the reverse. The total score treats those as equivalent, even though the underlying injury patterns and prognoses may differ. Clinicians who use the Sarnat exam well tend to look at both the total score and the individual category scores, using the former for quick communication and the latter for nuanced decision-making.

Sedation and Other Confounders

One practical challenge that rarely gets mentioned in textbook descriptions is the effect of medications on the exam. Babies with encephalopathy often receive anticonvulsant drugs to control seizures, and many of these medications suppress consciousness, reduce tone, and dampen reflexes. A baby who appears to be in stage 3 may actually have a brain that would function at stage 2 if the sedating effects of medication were removed. The reverse problem also exists: a baby whose seizures are not fully controlled may look more encephalopathic than the underlying injury warrants, because ongoing seizure activity itself depresses neurological function.

There is no clean solution to this. Clinicians generally try to time their exams to moments when medication effects are likely at their trough, and they document what drugs the baby is receiving alongside the Sarnat score. But the inherent uncertainty means that a single Sarnat exam in a heavily medicated baby should be interpreted cautiously, and serial exams become even more important for separating drug effects from true neurological status.

Temperature is another confounder worth knowing about. Therapeutic hypothermia itself can alter neurological signs: a cooled baby tends to be less active, with somewhat lower tone and heart rate, simply because of the lower body temperature. Exams performed during active cooling may slightly overestimate severity compared to the same baby examined at normal body temperature. This is one reason the rewarming exam carries particular prognostic weight, as it reflects neurological function after the temperature confounder has been removed.