How the Sickle Cell Guidelines Weigh Hydroxyurea and Transfusion

The American Society of Hematology 2020 guidelines on cerebrovascular disease in sickle cell disease do not simply tell clinicians what to do. Each recommendation carries two labels: how strong it is and how certain the underlying evidence is. Reading the document well means reading those labels. A strong recommendation built on moderate-certainty evidence, such as annual transcranial Doppler screening, sits on firmer ground than a conditional suggestion built on low or very low certainty, and the guideline text makes that difference explicit rather than hiding it.

This piece is an appraisal of how that guideline is constructed, not clinical direction. It is educational and not medical advice. The goal is to show how the panel moved from trial data to graded recommendations, and why hydroxyurea and chronic transfusion land in different places depending on the question and the setting.

The GRADE scaffolding

The ASH panel used the GRADE framework, which forces two separate judgments. The first is the strength of the recommendation, signaled by verb choice. "Recommends" marks a strong recommendation, meaning most patients in that situation would want the recommended course and it can reasonably serve as a default. "Suggests" marks a conditional recommendation, meaning the balance of benefits and harms is closer, and the right choice depends more heavily on individual values and circumstances.

The second judgment is certainty of evidence, rated high, moderate, low, or very low. Certainty is not a measure of whether an effect exists; it is a measure of how much confidence we can place in the estimate of that effect. Randomized trials generally start high and get downgraded for problems like imprecision, indirectness, or risk of bias. Observational data generally starts low. Keeping these two axes separate is the entire point. A recommendation can be strong even when certainty is moderate, and it can be conditional even when a real benefit is plausible, because strength also folds in harms, burden, cost, and how much patients' preferences vary.

Screening: where the evidence is strongest

The guideline's firmest footing is transcranial Doppler screening. TCD ultrasound measures blood flow velocity in the large cerebral arteries; abnormally high velocities identify children at markedly elevated stroke risk. For children ages 2 to 16 with HbSS or HbSβ0 thalassemia, the panel issued a strong recommendation for annual TCD screening, tied to moderate-certainty evidence.

That strength traces directly to a trial. The Stroke Prevention Trial in Sickle Cell Anemia (STOP), whose results were published in 1998, screened children with TCD and randomized those with abnormal velocities to regular transfusion or standard care. The transfusion arm showed about a 90 percent lower risk of first stroke, and the trial was stopped early for benefit. Screening only earns a strong recommendation because an effective action follows a positive result. A test that changes nothing does not deserve a strong screening recommendation, however accurate it is.

For rarer compound heterozygous genotypes with hemolysis comparable to HbSS, the screening recommendation drops to conditional with very low certainty. The biology is suggestive, but the direct trial evidence in those specific groups is thin, and the guideline says so through its labels rather than smoothing over the gap.

Transfusion and hydroxyurea: same question, different evidence

For a child with abnormal TCD velocities in a high-resource setting, the panel strongly recommends starting regular transfusion, aiming to keep HbS below a defined threshold, for at least a year. This is the direct translation of STOP: a randomized result, a large effect, moderate certainty, a strong recommendation.

The more interesting question is what happens after that first year, and here the guideline leans on a second trial. TWiTCH (Transcranial Doppler With Transfusions Changing to Hydroxyurea), published in The Lancet in 2016, enrolled children who had abnormal velocities and had already received at least a year of transfusion, but who did not have severe vasculopathy on imaging. It compared continued transfusion against transition to maximum-tolerated-dose hydroxyurea, using TCD velocity as the endpoint, and was designed as a non-inferiority trial. Hydroxyurea was non-inferior to transfusion for keeping velocities controlled in that carefully bounded population.

The guideline reflects both the promise and the boundaries of that result. It suggests, conditionally, that after at least a year of transfusion, hydroxyurea at maximum tolerated dose may replace transfusion, but only when imaging has ruled out significant vasculopathy or silent cerebral infarcts, matching TWiTCH's entry criteria. The certainty is lower and the recommendation is conditional because the trial answered a narrow question in a selected group, and extending it beyond those bounds is exactly the kind of indirectness GRADE penalizes.

Why setting changes the recommendation

The guideline does something many readers miss: it writes different recommendations for different resource settings, because a recommendation is a judgment about action, and available actions differ. Regular transfusion demands a reliable blood supply, iron-overload monitoring, chelation, and infrastructure that many regions cannot guarantee.

So for children with abnormal TCD in low- and middle-income settings without dependable access to transfusion, the panel suggests hydroxyurea at a fixed dose of at least 20 mg/kg per day or maximum tolerated dose. This is a conditional recommendation on low-certainty evidence. Notably, the comparison here is not hydroxyurea versus a fully resourced transfusion program; it is hydroxyurea versus little effective prevention at all. The panel let context reshape the recommendation rather than pretending one global default fits every health system. That is a feature of careful guideline construction, not an inconsistency.

Reading the labels honestly

The through-line is that strength and certainty carry information, and flattening them distorts the guideline. A conditional suggestion is an explicit invitation to weigh individual values, imaging findings, and access, not a weaker version of a command. A strong recommendation signals that the panel expects it to hold across most patients in that situation.

None of this settles an individual's care, which depends on genotype, prior imaging, transfusion history, and access, and belongs in a conversation with a treating hematologist. What the ASH 2020 document models well is disciplined reasoning: name the trial, grade the certainty, state the strength, and let the setting modify the action. Read that way, the guideline is less a rulebook than a transparent argument, and its labels are the argument's load-bearing parts.