How I treat a rising haemoglobin on hydroxycarbamide in sickle cell disease

Introduction

A patient with homozygous sickle cell disease has been on hydroxycarbamide for two years. The crises have stopped, the fetal haemoglobin has risen, the mean cell volume is 108 fL, and the haemoglobin, which was 7.9 g/dL at the start, is now 12.4 g/dL. A colleague proposes venesection.

The proposal has an intuitive logic. Sickle cell disease is a disease of vaso-occlusion; viscosity promotes vaso-occlusion; haemoglobin concentration is a major determinant of whole-blood viscosity; therefore a high haemoglobin in this disease is dangerous and should be lowered. Each step is reasonable on its own, and the third step is where the argument becomes uncertain rather than false: haemoglobin concentration is one determinant of viscosity, and in sickle cell disease it is not the only one. Red cell deformability, the proportion of dense cells and the fraction of polymerisable sickle haemoglobin all move at the same time, and under hydroxycarbamide several of them move in what is thought to be the protective direction while the haemoglobin rises.

The four cases are illustrative composites, not individual patients. Unless stated otherwise, the discussion is about HbSS and HbSβ⁰-thalassaemia in patients who are not on a transfusion programme; HbSC disease is treated separately, because the evidence there is different.

How the literature was searched

The sources behind this article were identified by searching PubMed to October 2026 for sickle cell disease combined with each of phlebotomy, venesection, erythrocytapheresis, blood viscosity, haematocrit, haemoglobin threshold and hydroxyurea or hydroxycarbamide, with no date limit and restricted to records in English; by reading the guideline documents of the American Society of Hematology, the National Heart, Lung, and Blood Institute expert panel, the British Society for Haematology and the World Health Organization for any haemoglobin ceiling or any mention of removing blood; and by following the reference lists of the retrieved reviews. The exact search strings were not recorded prospectively, so this is a description of what was done rather than a reproducible systematic search, and every statement below of the form "no trial" or "no controlled study" should be read as "none was found by this search".

What the guidelines say

No guideline on sickle cell disease recommends venesection or any other method of lowering the haemoglobin in a patient who is not transfused. Four documents shape practice, and what they omit matters as much as what they say (Table 1).

Two observations follow, and the first needs stating more carefully than it usually is. The 10 g/dL figure does double duty. In the NHLBI panel it is a target to be reached before surgery ¹. In the ASH transfusion guideline the same region of the scale functions as a limit: the post-transfusion haemoglobin should not exceed 11 g/dL after simple pre-operative transfusion, and a patient whose haemoglobin is already above 9–10 g/dL should have red cell exchange rather than simple transfusion, precisely so that the haematocrit is not pushed higher ². So 10–11 g/dL is a genuine ceiling — on transfused blood. What does not appear in any of these documents is a ceiling on a haemoglobin the patient has made themselves. Where the guidelines name a total haemoglobin outside the transfusion context, it is a minimum rather than a maximum: the ASH cerebrovascular guideline frames secondary stroke prevention around keeping the sickle fraction below 30% while maintaining the total haemoglobin above a floor ³.

Second, the hydroxycarbamide guidelines define toxicity by the marrow. The British Society for Haematology guideline builds its dose-escalation and monitoring scheme on blood counts ⁴, as did the Multicenter Study of Hydroxyurea (MSH), in which the doses reached at the end of the study ranged from 0 to 35 mg/kg per day and treatment was described as requiring careful monitoring ⁵. A rising haemoglobin does not appear as a toxicity criterion in either. A clinician who reduces the dose or removes blood because the haemoglobin has risen is applying a stopping rule that the trials establishing the drug did not use.

Four cases

Case 1 is taken up under The asymptomatic rise, Case 2 under Question the number first, Case 3 under Altitude and Case 4 under When there are symptoms.

Case 1 — The expected rise

A 26-year-old man with HbSS has taken hydroxycarbamide for two years at 25 mg/kg per day. Before treatment his haemoglobin was 7.9 g/dL with a fetal haemoglobin of 5%. Now the haemoglobin is 12.4 g/dL, fetal haemoglobin 22%, mean cell volume 108 fL, reticulocytes 1.8%, lactate dehydrogenase normal. He has had no crisis in 18 months and feels well. A colleague proposes venesection of 450 mL. Should the haemoglobin be lowered, and if so, how?

Case 2 — The number with a second explanation

A 34-year-old woman carries a diagnosis of sickle cell disease made in childhood at another centre. She takes hydroxycarbamide 15 mg/kg per day, a starting dose that has never been escalated. Haemoglobin is 13.4 g/dL, mean cell volume 71 fL, reticulocytes 2.1%. She has had two painful episodes in her life, both after surgery. Is this a hydroxycarbamide effect at all?

Case 3 — The patient who lives at altitude

A 29-year-old man with HbSS lives in Abha, at about 2,270 m. On hydroxycarbamide his haemoglobin has reached 12.6 g/dL, fetal haemoglobin 19%. His brother, with the same genotype and on the same dose, lives in Jizan at sea level and runs a haemoglobin of 11.4 g/dL. Is 12.6 g/dL in Abha the same finding as 12.6 g/dL at sea level? The two brothers are a teaching device, not a measurement: the difference between any two individuals also carries sex, weight, adherence, iron status and chance.

Case 4 — The rise with symptoms

A 41-year-old woman with HbSS on hydroxycarbamide has a haemoglobin of 13.6 g/dL and fetal haemoglobin of 20%. Over four months she has developed morning headache, facial plethora and three painful episodes after 18 months without one. She snores heavily and her husband reports witnessed apnoeas. Now what?

Where the haemoglobin ceiling came from

The habit of capping haemoglobin in sickle cell disease is a transfusion habit. Raising the haematocrit by transfusion in a patient whose circulating red cells still contain polymerisable sickle haemoglobin increases whole-blood viscosity, and the sickle fraction is diluted rather than removed. Out of that came a working limit around a haemoglobin of 10 g/dL and a haematocrit of 30%, usually stated without a source. It is not quite sourceless: transfusion guidance has conventionally set an upper haematocrit limit of about 0.30 to 0.35 to avoid hyperviscosity, and in vitro studies of sickle and normal red cell mixtures in buffer appeared to confirm that this range is optimal for oxygen delivery ⁶. The viscosity of such mixtures rises with the total haematocrit, and the same line of work supported the concept that exchange transfusion is more beneficial than simple transfusion, suggesting that the absolute level of sickle cells may be a better guide than the ratio of sickle to total haematocrit ⁷. The clinical half of the rationale is thinner: as Ballas and colleagues set it out, the limit rests on case reports of neurological complications after transfusion to a high haemoglobin rather than on a controlled comparison ⁸. That is a reasonable basis for caution about transfused blood and a weak basis for anything else, and their argument is that the threshold should be reconsidered now that other treatments raise the haemoglobin above 100 g/L — the level above which transfusion complications had been seen — and have raised the same viscosity concern by analogy ⁸.

The question reaches beyond hydroxycarbamide, and the clearest recent example is a warning rather than a reassurance. In the phase 3 HOPE trial, voxelotor raised haemoglobin by more than 1.0 g/dL at week 24 in 51% of participants on the 1500 mg dose against 7% on placebo ⁹. In September 2024 the manufacturer withdrew it worldwide, stating that the data suggested an imbalance in vaso-occlusive crises and fatal events and that the overall benefit no longer outweighed the risk ¹⁰. The European Medicines Agency, whose committee recommended suspension on 26 September 2024, later confirmed it, citing eight deaths on voxelotor against two on placebo in a trial in children at higher risk of stroke and more vaso-occlusive crises on voxelotor than on placebo in trials, while noting that the final registry analysis did not confirm an increase in pain episodes and that the underlying mechanisms remain unclear ¹¹.

Voxelotor can therefore no longer be offered as evidence that raising the haemoglobin in this disease is safe. Nor can it be used against hydroxycarbamide: it raises haemoglobin by increasing the oxygen affinity of haemoglobin S, not by inducing fetal haemoglobin or improving deformability, and no peer-reviewed account of the mechanism of harm exists ¹¹. Gene therapy is a third case: with lovotibeglogene autotemcel the median total haemoglobin rose from 8.5 g/dL to 11 g/dL or more from 6 months onwards, with resolution of severe vaso-occlusive events in those who could be evaluated ¹². Whether hyperviscosity events were sought or reported could not be verified from the published report, so no claim is made that none occurred.

Why a drug-driven rise is not a transfusion-driven rise

The two mechanisms differ on most of the variables that determine whether blood flows (Table 2). The table sets out a mechanistic hypothesis, not a demonstrated equivalence of outcomes.

The last row is the closest thing to a direct answer. Lemonne and colleagues followed 28 adults with sickle cell anaemia before hydroxycarbamide and at 6, 12 and 24 months, measuring red cell deformability by ektacytometry at 30 Pa and blood viscosity by cone-plate viscometer ¹³. Haemoglobin concentration rose under treatment, fetal haemoglobin and mean cell volume rose, white cell and platelet counts fell, and blood viscosity did not change across the two years. Deformability was above baseline by 6 months and went on rising to the end of follow-up. The authors' interpretation was that the gain in deformability probably offsets the effect of the rising haemoglobin on viscosity, and contributes to the clinical improvement ¹³. An earlier report from the same group measured viscosity at 45, 90 and 225 s⁻¹ and found it unchanged on hydroxycarbamide at each of them ¹⁵.

That evidence has to be weighed at its real weight. Twenty-eight patients, no control arm, group means reported: a study of that design cannot exclude a rise in viscosity in individual patients, or in patients who reach a higher haemoglobin than this cohort did, and it measured nothing below 45 s⁻¹. Viscosity in sickle cell disease is shear-thinning — it falls as shear rate rises — and the hyperviscosity concern applies most to the low-shear conditions of the microcirculation and of stasis ¹⁶. Unchanged viscosity down to 45 s⁻¹ is reassuring as far as it goes, and it does not go to the bottom of the range. Anyone defending the venesection practice should press on exactly this point, and the honest answer is that the measurement has not been made.

There is a second reason not to treat the haemoglobin as the whole question. What matters physiologically is not viscosity alone but the ratio of haematocrit to viscosity, an index of how effectively red cells deliver oxygen: raising the haematocrit adds carrying capacity and adds viscosity together, so there is an optimum rather than a direction. In sickle and normal red cell mixtures in plasma, an optimal haematocrit by this index appeared only above about 50 s⁻¹; at 5 s⁻¹ any rise in haematocrit was offset by a proportionally larger rise in viscosity ⁶. The same has been measured in patients: after simple transfusion, viscosity, haematocrit and the high-shear ratio rose while the low-shear ratio fell, which the authors read as impaired oxygen transport to low-flow regions ¹⁴. The right question about a patient at 12.4 g/dL on hydroxycarbamide is therefore not "is this above 10?" but whether it is above or below the haematocrit that maximises oxygen transport for these particular cells — and in treated patients that has not been measured. Improving deformability should raise the optimum; the rising haematocrit moves the patient along the curve; where the two leave a given patient is unknown.

So the position defended here is a hypothesis with mechanistic support, not an established fact: haemoglobin concentration and blood viscosity are not the same variable, and under hydroxycarbamide they appear to come apart. A patient at 12.4 g/dL with a fetal haemoglobin of 22%, a mean cell volume of 108 fL and more deformable cells may have no higher whole-blood viscosity than the same patient had at 8 g/dL with rigid, dense cells. Reacting to the haemoglobin alone treats the number as though it carried the meaning it carries in the transfusion setting, which it may not.

The wider rheology literature is consistent with the hypothesis without confirming it. Patients with HbSS on hydroxycarbamide have more deformable red cells, in normoxia and during deoxygenation, than untreated patients with the same genotype ¹⁷. Connes's review describes the two poles of the disease: patients with the least deformable cells are the most haemolytic and most anaemic and are prone to leg ulcers, priapism, cerebral vasculopathy, glomerulopathy and pulmonary hypertension, while those whose cells are more deformable are less anaemic and may have higher blood viscosity with a greater risk of vaso-occlusive events ¹⁸. That is a constitutional phenotype, set by genotype, co-inherited α-thalassaemia, fetal haemoglobin level and cellular factors, and not a measurement of what happens when a drug moves one patient along it. Whether a treated patient acquires the risks of the less anaemic pole is the open question of this article, and the phenotype literature does not answer it either way — the more so because patients with the most deformable cells at steady state have been reported to have more frequent painful crises, those cells being also the most adherent ¹⁶.

What the evidence actually shows, on both sides

There is no interventional evidence on lowering the haemoglobin. There is observational evidence that a high steady-state haemoglobin keeps bad company, and it deserves to be stated before it is qualified.

Cohort associations in untreated patients. In the Cooperative Study of Sickle Cell Disease, 3,578 patients were followed prospectively, and among those with sickle cell anaemia a high rate of painful episodes was associated with a high haematocrit and a low fetal haemoglobin, and among adults over 20, those with high pain rates tended to die earlier ¹⁹. In the same study population, the incidence of acute chest syndrome in HbSS was lower in patients whose steady-state haemoglobin was lower and whose fetal haemoglobin was higher, an association the authors could not explain and suggested might be mediated by viscosity ²⁰. Osteonecrosis of the femoral head, studied in 2,590 patients from the same programme, was positively associated with the frequency of painful crises and with the haematocrit, and negatively associated with the mean cell volume ²¹. To these should be added a finding from outside that programme which is the most pointed of all: in 95 children with sickle cell disease and no previous stroke, none on hydroxycarbamide, a lower mean overnight oxygen saturation and a higher cerebral artery velocity each independently predicted central nervous system events, and after adjustment for both, a higher haemoglobin concentration also independently predicted them, with a hazard ratio of 1.7 per g/dL (95% CI 1.18–2.43) ²². Dips suggestive of obstructive sleep apnoea did not predict events in that cohort.

That is a real body of evidence and it should change the tone of the discussion rather than the conclusion. Four qualifications apply, each a limitation rather than a refutation. These are associations in observational cohorts, and in all of them the haematocrit is also a marker of the constitutional phenotype described above: the patient with a higher haematocrit has more deformable cells and a lower haemolytic rate, and those cellular properties travel with the number. The patients were almost all untreated, and the fetal haemoglobin that rises with hydroxycarbamide is, in these same datasets, the variable associated with fewer events — so a treated patient at 12.4 g/dL with 22% fetal haemoglobin is not the patient these cohorts describe. The haemoglobin range observed is mostly that of untreated sickle cell anaemia, so these studies say more about 9 or 10 g/dL than about 13. And none tested an intervention: an association between a high haematocrit and morbidity does not establish that lowering it reduces morbidity, any more than the association between a low haemoglobin and mortality establishes that transfusing would prevent death.

A case report, and a biomarker association. Fattori and colleagues reported a man of 28 on long-term hydroxycarbamide for sickle cell anaemia who presented with acute myocardial infarction with no pre-existing coronary lesion, having run high haematocrits during treatment; they inferred a possible contribution from raised viscosity and recommended that a haematocrit above 30% and a haemoglobin above 10.5 g/dL in HbSS patients on the drug be avoided or watched closely ²³. As far as could be determined, this single 2005 report is the principal published origin of the 10.5 g/dL figure now quoted at the bedside. It is one patient, with no comparator and no viscosity measurement. Separately, plasma thrombospondin-1 has been found to correlate positively with haemoglobin concentration and with hydroxycarbamide use and negatively with markers of haemolysis, which the investigators read as a possible marker of the viscosity and vaso-occlusion end of the disease ²⁴. That is a cross-sectional association in a disease where the viscous and haemolytic phenotypes are known to separate, and it does not show that lowering the haemoglobin lowers risk.

Uncontrolled series of phlebotomy — almost all in HbSC. The claim that nothing has been published about removing blood in sickle cell disease is wrong, and the correction is genotype-specific. In a single-centre series of 179 patients with HbSC disease, phlebotomy was used in 36% (baseline haemoglobin 11.5 g/dL) and was followed by no recurrence of acute events in 71% of those treated; the authors concluded that HbSC should be regarded as a separate disease rather than a mild form of sickle cell anaemia, with particular weight on viscosity-associated otological and ophthalmological disorders, and that wider use of phlebotomy should be studied ²⁵. A descriptive Togolese series of 27 patients — 24 HbSC and three HbSS, mean pre-treatment haemoglobin 14.3 g/dL — reported annual vaso-occlusive crises falling from 5.3 to 0.9 after a mean of 4.9 phlebotomies ²⁶. Of 14 children with HbSC started on hydroxycarbamide, 11 reached maximum tolerated dose and five of those did not improve; the four who were then phlebotomised after at least six months at that dose without improvement all improved at least partly ²⁷. There is a case report of symptomatic benefit from phlebotomy in HbSC after splenectomy raised the haematocrit ²⁸, a published treatment algorithm for adults with HbSC ²⁹, and a recent series in which iron deficiency from repetitive phlebotomy in HbSC was associated with fewer sickle-related complications ³⁰. None of these is controlled; all are small; and the dominant genotype is HbSC, in which haemoglobin C itself does not polymerise but dehydrates the cell, which promotes polymerisation of HbS ²⁷, and in which baseline haemoglobin is higher. In HbSC the practice has an uncontrolled literature behind it and is part of recognised management.

One randomised trial, for a different purpose. Phlebotomy has also been used to remove iron, and that is the setting of the only randomised trial found in which patients with HbSS on hydroxycarbamide were venesected. In SWiTCH, children with sickle cell anaemia, stroke and iron overload were randomised to continue monthly transfusion with deferasirox chelation (n = 66) or to switch to hydroxycarbamide escalated to maximum tolerated dose, followed by monthly phlebotomy (n = 67). There were no strokes on transfusion and chelation and seven (10%) on hydroxycarbamide and phlebotomy, still within the trial's non-inferiority margin for stroke; the trial was closed for futility after an interim analysis showed equivalent liver iron content, and the investigators concluded that transfusions and chelation remain a better way to manage children with sickle cell anaemia, stroke and iron overload ³¹. Phlebotomy there was used to remove iron, not to control the haemoglobin, and it was given together with the switch to hydroxycarbamide. In this article's reading, the design does not separate the effect of switching from transfusion to hydroxycarbamide from the effect of phlebotomy, so the excess of strokes cannot be assigned to either. No trial of venesection to lower the haemoglobin was found in any genotype, and the one randomised trial in which HbSS patients on hydroxycarbamide were venesected used it to remove iron, in children after stroke, with a result that favoured transfusion and chelation.

The one published rationale for removing red cells, and what it actually says

One coherent published rationale for taking red cells out in this disease exists, and it is not the hyperviscosity argument. Castro and colleagues described four adults who had sickle cell anaemia together with iron-limited erythropoiesis ³². In two the iron deficiency had arisen on its own, from gastrointestinal bleeding and low-grade haemoglobinuria. In the other two it was created on purpose by repeated red cell apheresis, under a pilot protocol whose stated aim was to reduce intracellular polymerisation of haemoglobin S by bringing down the mean corpuscular haemoglobin concentration (MCHC).

What they found is worth reading. Lowering iron did not cost these patients haemoglobin: values sat between 7.8 and 9.0 g/dL, and two of them ended less anaemic than they began. MCHC, reticulocyte count, red cell distribution width and dense cell counts all fell, bilirubin and lactate dehydrogenase fell with them, and in one patient red cell survival lengthened. The calculated intracellular polymer fraction fell from a baseline range of 0.48–0.53 to 0.33–0.48, and it fell even though iron restriction also reduced the fetal haemoglobin concentration in three of the four. Of the two patients whose iron was depleted deliberately, one went from a mean of 4.5 hospital days a month for pain over a two-year baseline to 0.5 days a month over three years of follow-up, and the other had a leg ulcer heal quickly. The authors recommended that controlled iron restriction be explored in selected patients ³².

Three things must be said about it, and the same yardstick is used here as was used above for the single case report and the biomarker study. The mechanism is MCHC reduction, not haemoglobin reduction: the physiological target is the concentration of haemoglobin inside the cell, not the concentration in the blood, and those are different variables that a venesection programme would move in different ways. It is four patients, uncontrolled, published in 1994, and no trial of the strategy was found in the search described above; the nearest recent work is the HbSC phlebotomy series in which induced iron deficiency was associated with fewer complications ³⁰. And the protocol was aimed at patients who remained symptomatic. It is not a rationale for bleeding an asymptomatic patient whose haemoglobin has risen because a drug is working.

What lowering the haemoglobin costs

The question is never whether a high haemoglobin might be harmful in isolation. It is whether the intervention's harms are smaller than the harm it averts, and here most of the quantities on both sides of that comparison are unmeasured.

The drug has randomised evidence behind it. In MSH, 299 adults with three or more painful crises a year were randomised to hydroxycarbamide or placebo. Those on the drug had a median of 2.5 crises a year against 4.5 on placebo, fewer developed acute chest syndrome (25 against 51), and fewer were transfused (48 against 73) ⁵. In the nine-year observational follow-up, in which patients could start or stop the drug by their own choice, taking hydroxycarbamide was associated with mortality about 40% lower than in those not taking it ³³. At 17.5 years, survival curves showed fewer deaths among those with long-term exposure, and 87.1% of the deaths from pulmonary complications occurred in patients who had never taken the drug or had taken it for less than five years ³⁴.

It was a low haemoglobin, not a high one, that marked the patients who died. In the nine-year MSH follow-up, the patients at increased risk of death were those whose reticulocyte count was below 250,000/mm³ and whose haemoglobin was below 9 g/dL — a low-reticulocyte, anaemic group, not simply an anaemic one. Cumulative nine-year mortality was 28% where the fetal haemoglobin was below 0.5 g/dL after the trial, against 15% where it was 0.5 g/dL or higher ³³. Two cautions govern how far that can be taken. These are associations in a self-selected observational follow-up: a low haemoglobin with a low reticulocyte count marks severity and marrow failure, and what was associated with survival was exposure to the drug and a higher fetal haemoglobin, not the height of the haemoglobin itself. And the upper tail is essentially unobserved — a cohort of adults with frequent crises in the 1990s contained very few patients at 13 g/dL, so the absence of a mortality signal at the top of the range is absence of data, not evidence of safety.

Venesection would provoke the cells least wanted. Removing blood triggers a compensatory reticulocytosis, and reticulocytes are the red cells most implicated in starting adhesion to the endothelium: they carry α4β1 integrin, which binds endothelial VCAM-1 on cytokine-activated endothelium ³⁵, and α4β1 together with glycoprotein IV (CD36) is found on circulating reticulocytes in sickle cell anaemia and not on mature red cells ³⁶. The step from there to "venesection increases vaso-occlusion" is inference, not a cited finding. No study of reticulocyte adhesion after venesection in sickle cell disease was found, and no study was found that measured red cell adhesion or rheology after venesection in HbSS.

Repeated venesection depletes iron, with effects in both directions. The Castro data suggest a falling MCHC may be favourable ³²; against that, fetal haemoglobin fell in three of their four patients, and there is the loss of oxygen-carrying capacity, the fatigue, the effect on growth in children, and the particular cost in women of reproductive age, for whom induced iron depletion carries into pregnancy. No outcome study exists either way.

Dose reduction is the other unevidenced option, and it is not a single thing. Neither venesection nor dose reduction has been tested for this indication, and both should be described that way. But the two are not symmetrical in the way the reflex assumes, and the distinction that matters is how far the dose is reduced. A reduction from maximum tolerated dose to a dose still inside the range the trials used is not the same as abandoning treatment: MSH itself concluded that maximal tolerated doses may not be necessary for a therapeutic effect, with final doses ranging from 0 to 35 mg/kg per day ⁵; BABY HUG used a fixed 20 mg/kg per day in very young children and reduced pain (177 events against 375) and dactylitis ³⁷; REACH treated children in sub-Saharan Africa from a starting dose near 17.5 mg/kg ³⁸; and in Nigerian children with a previous stroke, fixed low-dose (10 mg/kg) and moderate-dose (20 mg/kg) hydroxycarbamide gave no detectable difference in recurrent stroke or death ³⁹.

Against that, the randomised evidence on dose in children points the other way. In NOHARM-MTD, 187 children in sub-Saharan Africa were randomised to a fixed dose of about 19.2 mg/kg per day or escalation to about 29.5 mg/kg per day; the trial was stopped early because clinical events were significantly fewer with escalation. The primary outcome — a haemoglobin of 9.0 g/dL or more, or a fetal haemoglobin of 20% or more — was reached by 86% on escalation against 37% on the fixed dose, with incidence rate ratios of 0.43 for vaso-occlusive pain and 0.21 for hospitalisation, and dose-limiting toxicity no commoner on the higher dose ⁴⁰. The extended REACH follow-up, non-randomised, pointed the same way: at a mean dose of 28.2 mg/kg per day the mean haemoglobin rose from 7.3 to 8.5 g/dL and fetal haemoglobin from 10.9% to 23.3%, with incidence rate ratios comparing maximum tolerated dose to fixed dosing of 0.60 for vaso-occlusive episodes and 0.21 for acute chest syndrome, and no increase in toxicity ⁴¹. Both studies are in African children, and malaria and nutritional status shape their event rates; neither can be carried directly to a Saudi adult. But they mean that dose reduction from maximum tolerated dose cannot be presented as the neutral or conservative choice. On the best randomised evidence available, the higher dose is the one that prevented events.

There is also a mechanistic cost to dose reduction, and it is inference rather than a measured result. The haemoglobin rise under hydroxycarbamide is coupled to fetal haemoglobin induction and to improved deformability ¹³,¹⁷. A reduction large enough to bring the haemoglobin down meaningfully would be expected to reduce fetal haemoglobin and mean cell volume as well, removing the haemoglobin contribution to viscosity while degrading the rheology of the cells themselves; no trial has tested that. Hydroxycarbamide also lowers the leukocyte count, which venesection does not touch and dose reduction would reverse. In TWiTCH, children at high stroke risk who switched from transfusion to hydroxycarbamide escalated to maximum tolerated dose achieved non-inferior transcranial Doppler velocities, with no strokes in either arm ⁴². A reduction to about 20 mg/kg stays inside the range the trials used but, on the randomised paediatric data, may cost some of the benefit ⁴⁰; a reduction that visibly undoes the fetal haemoglobin response trades a benefit supported by randomised data for a harm supported by a case report and by observational associations in untreated patients.

Question the number first

In Case 2 the haemoglobin is 13.4 g/dL with a mean cell volume of 71 fL, on a starting dose never escalated, in a woman who has had two painful episodes in her life. Before deciding what to do about a number, it is worth asking whether the number means what it is assumed to mean. A haemoglobin of 13 g/dL in true HbSS is unusual even at maximum tolerated dose, and usually has a second explanation. A mean cell volume of 71 fL on a drug that raises the mean cell volume is close to a contradiction, and it is the finding that should be explained first.

The differential, in rough order of yield:

• A co-existing thalassaemia. This is the leading explanation for the microcytosis and should be stated as such. Co-inherited α-thalassaemia trait in sickle cell anaemia is classically associated with a higher haemoglobin and a lower mean cell volume, and it is common in Saudi Arabia: half of a hospital cohort of adults with sickle cell disease in the Eastern Province had co-inherited it ⁴³. Alternatively the genotype itself may be HbSβ-thalassaemia rather than HbSS, in which microcytosis is intrinsic. In either case the drug still raises the mean cell volume, but it does so from a much lower starting point, so a value of 71 fL on treatment is compatible with a good response in a patient whose untreated mean cell volume was in the 60s.
• The genotype more broadly. Is this HbSS at all, or HbSC, HbSβ⁺-thalassaemia or another compound heterozygous state, where the baseline haemoglobin is higher and the course milder? A childhood label from another centre is not a diagnosis; confirm by high-performance liquid chromatography, with α-globin genotyping. A near-normal haemoglobin with microcytosis and a trivial lifetime clinical history points this way before anything else does.
• Non-adherence. The mean cell volume is also the most useful surrogate for whether the drug is being taken. A low or falling mean cell volume in a patient supposedly on hydroxycarbamide may mean the tablets are not being swallowed, in which case neither the haemoglobin nor the mean cell volume is a drug effect at all. In Case 2 this is an additional possibility rather than a substitute for the thalassaemia, and a dose of 15 mg/kg per day that has never been escalated is in any case a starting dose, not a maximum tolerated one.
• Iron deficiency, which lowers the mean cell volume against the drug's effect and is common enough in women of reproductive age to be worth excluding before anything is inferred from the indices.
• Baseline fetal haemoglobin and haplotype. Fetal haemoglobin levels in sickle cell anaemia differ markedly between populations. In the cohorts genotyped by Sebastiani and colleagues, Saudi Arab patients from the Eastern Province and Indian patients carrying the Arab-Indian haplotype had fetal haemoglobin around 20%, a Saudi Arab cohort from the Southwestern Province around 12%, and African American patients around 7%; the paper's own finding was negative, in that the common BCL11A enhancer haplotypes had similar genetic effects in all three groups and did not account for the differences between them ⁴⁴. The paper describes a Southwestern Province cohort without naming its cities, and those patients were not on hydroxycarbamide. The practical consequence stands: a patient who starts from a higher baseline fetal haemoglobin and then gains more on the drug will reach a higher haemoglobin than figures from the African or African American literature would predict, and the haemoglobin that counts as unexpectedly high in Kano or Chicago is not necessarily the haemoglobin that counts as unexpectedly high here.
• Secondary erythrocytosis: obstructive sleep apnoea, smoking, chronic hypoxaemia from sickle lung disease, androgen use, or an erythropoietin-secreting renal lesion.
• Haemoconcentration from dehydration or diuretics at the time of sampling — the commonest and most reversible cause of a single high reading.
• Coincidental polycythaemia vera, which is rare but does occur, and which at altitude is harder to pick out: among 136 polycythaemia work-ups at moderate altitude in this region, 22 (16.7%) were JAK2-positive, and the authors' conclusion was that haemoglobin and haematocrit alone should not decide who is investigated for it ⁴⁵.

Age belongs in the background of any of these readings. In a cross-sectional study of 267 patients with sickle cell anaemia and 138 with HbSC disease, blood viscosity was higher in older patients than in younger ones, and the haematocrit fell as patients with HbSS aged ⁴⁶. The baseline against which a given haemoglobin looks high is not fixed across a lifetime.

The immediate steps are to repeat the high-performance liquid chromatography, request α-globin genotyping and iron studies, check adherence, and withhold any decision about the haemoglobin until the genotype is known. If the genotype is HbSS or HbSβ⁰-thalassaemia with α-thalassaemia trait, the argument of this article applies and there is no case for venesection. If it is HbSβ⁺-thalassaemia, a higher baseline haemoglobin is expected and the number is less surprising, although iron status and adherence still apply. If it is HbSC, the argument of this article does not transfer: the cellular lesion is different, baseline haemoglobin is higher, viscosity-associated complications are prominent, and phlebotomy is part of recognised practice with an uncontrolled literature behind it ²⁵,²⁷,²⁹,³⁰.

Altitude

Case 3 compares two brothers with the same genotype on the same dose, one in Abha and one at sea level.

The World Health Organization's 2024 guideline on haemoglobin cut-offs adjusts for elevation in bands of 500 m, and for the 2,000–2,499 m band, which includes Abha at about 2,270 m, the adjustment is 14 g/L. The guideline is explicit about how it is to be used: the figure is the amount subtracted from an individual's observed haemoglobin, or added to the cut-off that defines anaemia ⁴⁷. So individual adjustment is what WHO intends, and an earlier version of this argument was wrong to say otherwise.

Two cautions remain, and they are this article's own judgement rather than the guideline's. The adjustment was derived and validated for identifying anaemia, which is the lower tail of the distribution; WHO says nothing about the upper tail, which is what is at issue here. And locally measured increments are smaller than 14 g/L. In 57,059 complete blood counts from Jeddah at sea level, Taif at 1,879 m and Abha at 2,270 m, the mean haemoglobin at Abha was 15.40 g/dL in men and 14.71 g/dL in women against 14.81 and 13.77 g/dL at sea level — increments of 0.59 g/dL in men and 0.94 g/dL in women — and the correction proposed from those data was 0.30 g/dL per 1,000 m, explicitly smaller than the WHO and CDC figures ⁴⁸. For a patient in Abha, that puts the expected altitude contribution at roughly 0.6–0.9 g/dL rather than 1.4.

Concentration is in any case an imperfect index of what altitude has done: in healthy residents of the Aseer region at 2,250 m and of Abha, haemoglobin mass was higher than at sea level and came from a larger red cell volume, yet haemoglobin and haematocrit correlated only weakly with it ⁴⁹,⁵⁰. And in sea-level newcomers followed for six months in Abha, haemoglobin and haematocrit stayed about 6 g/L and 1.9% higher, sustained by plasma volume contraction rather than by continuing erythropoiesis, so part of a highland haemoglobin reflects a smaller plasma volume rather than more red cells ⁵¹.

Three further points keep the altitude argument from proving too much. The altitude increment is driven by hypoxia and erythropoietin, so unlike the hydroxycarbamide increment it is not coupled to fetal haemoglobin or to improved deformability; only one of the two sources of a high haemoglobin brings the rheological changes described earlier. Chronic hypoxaemia itself favours polymerisation of haemoglobin S, so the mechanism that raises the number at altitude is not benign in this disease. And whether a patient with HbSS mounts the same altitude response as a healthy highland resident has not, as far as could be determined, been measured; the local reference data above come from healthy residents. Altitude therefore explains a higher haemoglobin without establishing that the higher value is safe.

The practical consequence is the one most often missed. A haemoglobin of 12.6 g/dL in Abha is not the same finding as 12.6 g/dL in Jeddah. A clinician applying a sea-level threshold to a highland patient will conclude that a drug effect has become excessive when it may not have. Establish where the patient lives, where the baseline was measured, and whether they have moved recently. A resting oxygen saturation, with overnight oximetry where nocturnal desaturation is suspected, is more informative in a highland patient than the haemoglobin alone, because it addresses the hypoxic drive directly. Any altitude adjustment applies to the haemoglobin only, not to fetal haemoglobin, mean cell volume or ferritin. The gap between the two brothers in Case 3 is 1.2 g/dL — close to but not identical with any published increment, which is how such comparisons behave: the adjustment is a population expectation, not an entitlement to subtract a fixed amount from one patient's result.

The asymptomatic rise

Case 1 is the case the question is usually asked about: a well patient, crises abolished, haemoglobin 12.4 g/dL, fetal haemoglobin 22%, mean cell volume 108 fL, reticulocytes 1.8%, normal lactate dehydrogenase.

Every variable here except the haemoglobin indicates that the disease is better controlled than it has been. The reticulocyte count and lactate dehydrogenase indicate that haemolysis has fallen. The mean cell volume and fetal haemoglobin indicate that the drug is being taken and is working. The rheology data indicate that viscosity in this configuration did not rise with the haemoglobin over two years ¹³,¹⁵.

The recommendation here is to continue the dose, confirm the absence of hyperviscosity symptoms, record the whole picture rather than the haemoglobin alone, and watch. It is the default under uncertainty, not a demonstration of safety. The optimal haematocrit for these cells is unknown, and the cohort associations between a high haematocrit and morbidity in untreated patients ¹⁹,²⁰,²¹,²² have not been shown not to apply. What tips the balance is the asymmetry of the two sides: continuing the drug preserves a benefit established by randomised trials ⁵,³⁷,⁴⁰, while the intervention proposed against it has no outcome data for this indication and would generate a reticulocytosis of unknown consequence. The haemoglobin rise in this configuration is the therapeutic response, and the variables in the MSH follow-up that were associated with survival — exposure to hydroxycarbamide, and a higher fetal haemoglobin — are the ones this patient now has ³³,³⁴.

The monitoring question is separate from the treatment question. In a Nigerian cohort of 116 children on a mean of 18.5 mg/kg per day, haemoglobin, fetal haemoglobin and mean cell volume rose without significant toxicity despite inconsistent dose escalation and attendance for haematology tests of only 18.9%; 89.5% of parents reported satisfaction with the drug, and 61% reported dissatisfaction with daily drug use and with the frequency and cost of monitoring ⁵². That shows benefit can be obtained with intermittent monitoring, not that any interval is safe; the published schedules are those of the BSH guideline ⁴, and the thing worth monitoring is the marrow.

When there are symptoms

Case 4 is different, and it is the only one of the four in which removing blood comes into consideration. Morning headache, facial plethora and a return of painful episodes after 18 months of quiescence are not findings to set aside on the grounds that the viscosity argument is weak.

The first step is still not venesection; it is to ask what else has changed, and to exclude the dangerous causes of the symptom before the convenient one. Morning headache in an adult with sickle cell disease requires that cerebrovascular disease be considered first — central nervous system complications are among the commonest and most serious sequelae of this disease across the lifespan ³ — and brain imaging, with vascular imaging, should precede any attribution of the headache to viscosity or to sleep. That ordering is standard clinical caution rather than a cited recommendation. The same applies to a hypertensive or ophthalmological cause.

The snoring and witnessed apnoeas do point to obstructive sleep apnoea, which is commoner in children with sickle cell anaemia than in the general paediatric population ⁵³ and also common in adults with the disease ⁵⁴. It is worth diagnosing and worth treating in its own right. What cannot be claimed is a chain from sleep-disordered breathing to sickle morbidity. In an unselected prospective cohort of 140 children, neither a low mean nocturnal oxygen saturation nor a high obstructive apnoea–hypopnoea index nor a high oxygen desaturation index was associated with a clinically relevant increase in acute severe pain episodes, and the association found for mean nocturnal saturation ran in the opposite direction to the hypothesis ⁵⁵. In a small adult cohort, sleep-disordered breathing was not associated with more sickle complications ⁵⁴. The one cohort in which nocturnal hypoxaemia did predict hard outcomes predicted central nervous system events, not pain, and in that same analysis the apnoeic dips themselves did not predict events while the mean overnight saturation and the haemoglobin concentration did ²². So the reasonable position in Case 4 is that nocturnal hypoxaemia may account for the erythrocytosis through erythropoietin, that treating it is right on its own merits and may address the headache, and that the link to the painful episodes is the weakest part of the chain and should be presented to the patient as such. This reasoning is clinical inference; no study was found comparing venesection with treatment of sleep apnoea in this setting.

If such a patient has genuinely been worked up — genotype confirmed, altitude baseline established, cerebrovascular disease excluded, sleep apnoea excluded or treated, no androgen use, no renal lesion — and remains symptomatic with a high haemoglobin, then two ways of lowering it exist and they are not equivalent. The established procedure for reducing the haemoglobin and the sickle fraction together is automated red cell exchange: the ASH transfusion guideline makes exchange the mode of choice once the haemoglobin is above about 9–10 g/dL, precisely because adding cells at that level raises the haematocrit ², and the in vitro rheology supports exchange over simple transfusion for the same reason ⁷. Exchange has its own costs — venous access, alloimmunisation, cost and availability — and has not been studied for this indication either. The alternative is a single modest venesection, kept isovolaemic by replacing the volume removed with an equal volume of crystalloid, a precaution that is standard practice rather than a sourced recommendation. For a physician without apheresis, a single venesection preserves the fetal haemoglobin and the leukocyte-lowering effect that a dose reduction would give up — with the qualification from Castro's data that iron depletion itself lowered fetal haemoglobin in three of four patients ³², so repeated venesection does not leave the fetal haemoglobin untouched. Any of this remains a clinical judgement with no trial behind it, and should be described to the patient in those terms.

What a trial would need to look like

The practice persists because the question has not been asked properly, and it could be. Two studies are needed, and they differ greatly in size. A mechanistic phase II study is feasible at a modest sample: patients with HbSS on stable hydroxycarbamide above a defined haemoglobin threshold, randomised to a single venesection or to observation, with whole-blood viscosity at low as well as high shear, deformability by oxygen-gradient ektacytometry, dense cell counts and the haematocrit-to-viscosity ratio as endpoints. That would answer the physiological question this article cannot — whether these patients sit above or below their optimal haematocrit — with tens rather than hundreds of participants. A clinical-outcome study is another matter: event rates in well-controlled patients at maximum tolerated dose are low and the expected effect is small, so the sensible design is a non-inferiority trial of observation against venesection, needing a large multi-centre sample. In the meantime, target-trial emulation in registry data and local audit of how often the procedure is done, and on what grounds, are both achievable. Until then, both practices rest on physiological reasoning rather than outcome data, and should be presented to patients that way.

Conclusion

No randomised trial, controlled study or guideline supports venesection to lower the haemoglobin in HbSS sickle cell anaemia ¹–⁴. The one randomised trial in which HbSS patients on hydroxycarbamide were venesected used phlebotomy to remove iron, and its result favoured transfusion and chelation ³¹; in HbSC disease, phlebotomy has an uncontrolled literature and a place in recognised practice ²⁵,²⁷,²⁹,³⁰. The 10–11 g/dL ceiling is a real limit on transfused blood ², resting largely on case reports ⁸. The contrary evidence is observational and should not be dismissed: in untreated patients a higher haematocrit was associated with more pain ¹⁹, acute chest syndrome ²⁰ and osteonecrosis ²¹, and a higher haemoglobin predicted central nervous system events in children ²². It does not describe the treated patient, whose fetal haemoglobin has risen with the haemoglobin, and it does not show that removing blood helps. Under hydroxycarbamide, haemoglobin rose while viscosity did not ¹³,¹⁵ — in 28 uncontrolled patients, with nothing measured below 45 s⁻¹, which remains the open flank in the argument. Whether treated patients sit above their optimal haematocrit is unmeasured ⁶,¹⁴. The case for the drug is randomised ⁵, with lower mortality on long-term follow-up ³³,³⁴ and randomised paediatric evidence favouring the higher dose ⁴⁰; venesection and dose reduction are both unevidenced for this indication. A haemoglobin of 13 g/dL in HbSS usually has a second explanation: a co-existing thalassaemia, a higher baseline fetal haemoglobin ⁴⁴ and life at 2,270 m ⁴⁷,⁴⁸ are among them. It is worth telling the patient in advance that the haemoglobin is expected to rise on this drug, and that the rise is the treatment working rather than a complication.

Personal note

I do not venesect patients with sickle cell disease to lower a haemoglobin that has risen on hydroxycarbamide, and I do not reduce the dose for that reason either. I regard the practice as a transfusion-era reflex applied to a situation it was never derived from.

When I am shown a high haemoglobin in this disease, I ask three questions before I ask anything about viscosity. Is the genotype what the file says it is? Where does this patient live, and where was the baseline measured? And is there a second cause — sleep apnoea, dehydration, androgens, a renal lesion — that explains the erythrocytosis better than the drug does? In my experience the third question is answered more often than the viscosity question is ever reached.

In Abha, at about 2,270 m, I expect my patients to run higher haemoglobin concentrations than the published series from West Africa or North America, and I read their numbers accordingly. The haplotype compounds this: with a higher baseline fetal haemoglobin in this region, the haemoglobin a patient reaches on hydroxycarbamide is higher than the African literature would predict, and this is not a complication.

I treat the marrow as the thing that defines toxicity, as the trials did. A rising haemoglobin is not a reason to change the dose. A falling neutrophil or platelet count is.

If a fully worked-up patient is symptomatic with a high haemoglobin, I would consider a single modest isovolaemic venesection before I would reduce the hydroxycarbamide dose, and I would tell the patient that I am choosing between two options that have no trial evidence behind either of them. I would not make it a standing practice, and I would not repeat it without a reason beyond the number.

I would welcome a local audit of how often venesection is being done for this indication in the Kingdom, and on what grounds. My impression is that it is commoner than the absence of evidence would justify, and that it is being done to asymptomatic patients.

Disclosure and attribution

The author declares no conflicts of interest. This article was researched and drafted with the help of AI. Every reference was checked against its PubMed record or the publisher's own document, and the citations and figures were then verified a second time against those records independently of the drafting. Article identifiers were taken from retrieved records rather than from memory. The literature search behind it is described above and was not a prospectively registered systematic review, so statements that no study exists are statements about what that search found.

Where the argument rests on inference rather than on a cited result — the expected effect of dose reduction, the reticulocyte response to venesection, the ordering of investigations in Case 4, the attribution of a high haemoglobin to sleep apnoea, the isovolaemic technique, and whether a patient with HbSS mounts a normal altitude response — this is said in the text rather than letting a citation imply more than its source contains. The central rheology finding comes from one uncontrolled cohort of 28 patients, and nothing below 45 s⁻¹ has been measured.

It is written for clinicians and is not a substitute for a guideline, a specialist opinion, or the judgement of the physician treating the patient.

References

1. Yawn BP, Buchanan GR, Afenyi-Annan AN, et al. Management of sickle cell disease: summary of the 2014 evidence-based report by expert panel members. JAMA 2014; 312: 1033–1048. doi:10.1001/jama.2014.10517 PMID 25203083
2. Chou ST, Alsawas M, Fasano RM, et al. American Society of Hematology 2020 guidelines for sickle cell disease: transfusion support. Blood Adv 2020; 4: 327–355. doi:10.1182/bloodadvances.2019001143 PMID 31985807
3. DeBaun MR, Jordan LC, King AA, et al. American Society of Hematology 2020 guidelines for sickle cell disease: prevention, diagnosis, and treatment of cerebrovascular disease in children and adults. Blood Adv 2020; 4: 1554–1588. doi:10.1182/bloodadvances.2019001142 PMID 32298430
4. Qureshi A, Kaya B, Pancham S, et al. Guidelines for the use of hydroxycarbamide in children and adults with sickle cell disease: A British Society for Haematology Guideline. Br J Haematol 2018; 181: 460–475. doi:10.1111/bjh.15235 PMID 29732531
5. Charache S, Terrin ML, Moore RD, et al. Effect of hydroxyurea on the frequency of painful crises in sickle cell anemia. N Engl J Med 1995; 332: 1317–1322. doi:10.1056/NEJM199505183322001 PMID 7715639
6. Alexy T, Pais E, Armstrong JK, Meiselman HJ, Johnson CS, Fisher TC. Rheologic behavior of sickle and normal red blood cell mixtures in sickle plasma: implications for transfusion therapy. Transfusion 2006; 46: 912–918. doi:10.1111/j.1537-2995.2006.00823.x PMID 16734807
7. Schmalzer EA, Lee JO, Brown AK, Usami S, Chien S. Viscosity of mixtures of sickle and normal red cells at varying hematocrit levels. Implications for transfusion. Transfusion 1987; 27: 228–233. doi:10.1046/j.1537-2995.1987.27387235626.x PMID 3590284
8. Ballas SK, Kuypers FA, Gordeuk VR, Hankins JS, Thompson AA, Vichinsky E. Time to rethink haemoglobin threshold guidelines in sickle cell disease. Br J Haematol 2021; 195: 518–522. doi:10.1111/bjh.17578 PMID 34131897
9. Vichinsky E, Hoppe CC, Ataga KI, et al. A Phase 3 Randomized Trial of Voxelotor in Sickle Cell Disease. N Engl J Med 2019; 381: 509–519. doi:10.1056/NEJMoa1903212 PMID 31199090
10. Pfizer. Dear Healthcare Provider letter: voluntary withdrawal of Oxbryta (voxelotor). 26 September 2024. https://webfiles.pfizer.com/dear-hcp-letter-oxbryta-us-final-092524
11. European Medicines Agency. EMA confirms suspension of sickle cell disease medicine Oxbryta. 17 October 2025. https://www.ema.europa.eu/en/news/ema-confirms-suspension-sickle-cell-disease-medicine-oxbryta
12. Kanter J, Walters MC, Krishnamurti L, et al. Biologic and Clinical Efficacy of LentiGlobin for Sickle Cell Disease. N Engl J Med 2022; 386: 617–628. doi:10.1056/NEJMoa2117175 PMID 34898139
13. Lemonne N, Möckesch B, Charlot K, et al. Effects of hydroxyurea on blood rheology in sickle cell anemia: A two-years follow-up study. Clin Hemorheol Microcirc 2017; 67: 141–148. doi:10.3233/CH-170280 PMID 28759962
14. Detterich J, Alexy T, Rabai M, et al. Low-shear red blood cell oxygen transport effectiveness is adversely affected by transfusion and further worsened by deoxygenation in sickle cell disease patients on chronic transfusion therapy. Transfusion 2013; 53: 297–305. doi:10.1111/j.1537-2995.2012.03822.x PMID 22882132
15. Lemonne N, Charlot K, Waltz X, et al. Hydroxyurea treatment does not increase blood viscosity and improves red blood cell rheology in sickle cell anemia. Haematologica 2015; 100: e383–e386. doi:10.3324/haematol.2015.130435 PMID 26137960
16. Nader E, Skinner S, Romana M, et al. Blood Rheology: Key Parameters, Impact on Blood Flow, Role in Sickle Cell Disease and Effects of Exercise. Front Physiol 2019; 10: 1329. doi:10.3389/fphys.2019.01329 PMID 31749708
17. Connes P, Martin M, Boisson C, et al. Differences in oxygen-gradient ektacytometry parameters and blood viscosity between patients with sickle cell anemia and patients with sickle cell-hemoglobin C disorder. Clin Hemorheol Microcirc 2025; 90: 145–151. doi:10.1177/13860291251357796 PMID 40625188
18. Connes P. Blood rheology and vascular function in sickle cell trait and sickle cell disease: From pathophysiological mechanisms to clinical usefulness. Clin Hemorheol Microcirc 2024; 86: 9–27. doi:10.3233/CH-238122 PMID 38073384
19. Platt OS, Thorington BD, Brambilla DJ, et al. Pain in sickle cell disease. Rates and risk factors. N Engl J Med 1991; 325: 11–16. doi:10.1056/NEJM199107043250103 PMID 1710777
20. Castro O, Brambilla DJ, Thorington B, et al. The acute chest syndrome in sickle cell disease: incidence and risk factors. The Cooperative Study of Sickle Cell Disease. Blood 1994; 84: 643–649. PMID 7517723
21. Milner PF, Kraus AP, Sebes JI, et al. Sickle cell disease as a cause of osteonecrosis of the femoral head. N Engl J Med 1991; 325: 1476–1481. doi:10.1056/NEJM199111213252104 PMID 1944426
22. Kirkham FJ, Hewes DK, Prengler M, Wade A, Lane R, Evans JP. Nocturnal hypoxaemia and central-nervous-system events in sickle-cell disease. Lancet 2001; 357: 1656–1659. doi:10.1016/s0140-6736(00)04821-2 PMID 11425370
23. Fattori A, de Souza RA, Saad ST, Costa FF. Acute myocardial infarction in sickle cell disease: a possible complication of hydroxyurea treatment. Hematol J 2005; 5: 589–590. doi:10.1038/sj.thj.6200572 PMID 15692605
24. Novelli EM, Kato GJ, Ragni MV, et al. Plasma thrombospondin-1 is increased during acute sickle cell vaso-occlusive events and associated with acute chest syndrome, hydroxyurea therapy, and lower hemolytic rates. Am J Hematol 2012; 87: 326–330. doi:10.1002/ajh.22274 PMID 22318901
25. Lionnet F, Hammoudi N, Stojanovic KS, et al. Hemoglobin sickle cell disease complications: a clinical study of 179 cases. Haematologica 2012; 97: 1136–1141. doi:10.3324/haematol.2011.055202 PMID 22315500
26. Padaro E, Kueviakoe IMD, Agbétiafa K, et al. Therapeutic phlebotomy during major sickle cell disease in Togo. Med Sante Trop 2019; 29: 106–107. doi:10.1684/mst.2019.0886 PMID 31031233
27. Summarell CC, Sheehan VA. Original Research: Use of hydroxyurea and phlebotomy in pediatric patients with hemoglobin SC disease. Exp Biol Med (Maywood) 2016; 241: 737–744. doi:10.1177/1535370216639737 PMID 26993671
28. Markham MJ, Lottenberg R, Zumberg M. Role of phlebotomy in the management of hemoglobin SC disease: case report and review of the literature. Am J Hematol 2003; 73: 121–125. doi:10.1002/ajh.10328 PMID 12749014
29. Naessens V, Ward R, Kuo KHM. A proposed treatment algorithm for adults with Haemoglobin SC disease. Br J Haematol 2018; 182: 607–609. doi:10.1111/bjh.14852 PMID 28699645
30. Traets MJM, Kidane Gembremeskel AS, Eijkelenboom-Bos J, et al. Iron Deficiency in HbSC Disease Treated With Repetitive Phlebotomy Is Associated With Fewer Sickle Cell Disease-Related Complications. Am J Hematol 2025; 100: 2123–2127. doi:10.1002/ajh.70045 PMID 40905599
31. Ware RE, Helms RW; SWiTCH Investigators. Stroke With Transfusions Changing to Hydroxyurea (SWiTCH). Blood 2012; 119: 3925–3932. doi:10.1182/blood-2011-11-392340 PMID 22318199
32. Castro O, Poillon WN, Finke H, Massac E. Improvement of sickle cell anemia by iron-limited erythropoiesis. Am J Hematol 1994; 47: 74–81. doi:10.1002/ajh.2830470203 PMID 7522396
33. Steinberg MH, Barton F, Castro O, et al. Effect of hydroxyurea on mortality and morbidity in adult sickle cell anemia: risks and benefits up to 9 years of treatment. JAMA 2003; 289: 1645–1651. doi:10.1001/jama.289.13.1645 PMID 12672732
34. Steinberg MH, McCarthy WF, Castro O, et al. The risks and benefits of long-term use of hydroxyurea in sickle cell anemia: A 17.5 year follow-up. Am J Hematol 2010; 85: 403–408. doi:10.1002/ajh.21699 PMID 20513116
35. Swerlick RA, Eckman JR, Kumar A, Jeitler M, Wick TM. Alpha 4 beta 1-integrin expression on sickle reticulocytes: vascular cell adhesion molecule-1-dependent binding to endothelium. Blood 1993; 82: 1891–1899. PMID 7691241
36. Joneckis CC, Ackley RL, Orringer EP, Wayner EA, Parise LV. Integrin alpha 4 beta 1 and glycoprotein IV (CD36) are expressed on circulating reticulocytes in sickle cell anemia. Blood 1993; 82: 3548–3555. PMID 7505118
37. Wang WC, Ware RE, Miller ST, et al. Hydroxycarbamide in very young children with sickle-cell anaemia: a multicentre, randomised, controlled trial (BABY HUG). Lancet 2011; 377: 1663–1672. doi:10.1016/S0140-6736(11)60355-3 PMID 21571150
38. Tshilolo L, Tomlinson G, Williams TN, et al. Hydroxyurea for Children with Sickle Cell Anemia in Sub-Saharan Africa. N Engl J Med 2019; 380: 121–131. doi:10.1056/NEJMoa1813598 PMID 30501550
39. Abdullahi SU, Sunusi S, Abba MS, et al. Hydroxyurea for secondary stroke prevention in children with sickle cell anemia in Nigeria: a randomized controlled trial. Blood 2023; 141: 825–834. doi:10.1182/blood.2022016620 PMID 36322937
40. John CC, Opoka RO, Latham TS, et al. Hydroxyurea Dose Escalation for Sickle Cell Anemia in Sub-Saharan Africa. N Engl J Med 2020; 382: 2524–2533. doi:10.1056/NEJMoa2000146 PMID 32579813
41. Aygun B, Lane A, Smart LR, et al. Hydroxyurea dose optimisation for children with sickle cell anaemia in sub-Saharan Africa (REACH): extended follow-up of a multicentre, open-label, phase 1/2 trial. Lancet Haematol 2024; 11: e425–e435. doi:10.1016/S2352-3026(24)00078-4 PMID 38701812
42. Ware RE, Davis BR, Schultz WH, et al. Hydroxycarbamide versus chronic transfusion for maintenance of transcranial doppler flow velocities in children with sickle cell anaemia—TCD With Transfusions Changing to Hydroxyurea (TWiTCH): a multicentre, open-label, phase 3, non-inferiority trial. Lancet 2016; 387: 661–670. doi:10.1016/S0140-6736(15)01041-7 PMID 26670617
43. Alsultan A, Alabdulaali MK, Griffin PJ, et al. Sickle cell disease in Saudi Arabia: the phenotype in adults with the Arab-Indian haplotype is not benign. Br J Haematol 2014; 164: 597–604. doi:10.1111/bjh.12650 PMID 24224700
44. Sebastiani P, Farrell JJ, Alsultan A, et al. BCL11A enhancer haplotypes and fetal hemoglobin in sickle cell anemia. Blood Cells Mol Dis 2015; 54: 224–230. doi:10.1016/j.bcmd.2015.01.001 PMID 25703683
45. Alkhaldy HY, Algarni AM, Eisa N, et al. Diagnostic Evaluation of Polycythemia Vera at High Altitude. Clin Lab 2024; 70(6). doi:10.7754/Clin.Lab.2023.231150 PMID 38868887
46. Renoux C, Romana M, Joly P, et al. Effect of Age on Blood Rheology in Sickle Cell Anaemia and Sickle Cell Haemoglobin C Disease: A Cross-Sectional Study. PLoS One 2016; 11: e0158182. doi:10.1371/journal.pone.0158182 PMID 27355589
47. World Health Organization. Guideline on haemoglobin cutoffs to define anaemia in individuals and populations. Geneva: WHO, 5 March 2024. https://iris.who.int/handle/10665/376196
48. Alkhaldy HY, Awan ZA, Abouzaid AA, et al. Effect of Altitude on Hemoglobin and Red Blood Cell Indices in Adults in Different Regions of Saudi Arabia. Int J Gen Med 2022; 15: 3559–3565. doi:10.2147/IJGM.S358139 PMID 35386861
49. Alkhaldy H, Alqahtani MM, Al Amri MS, Alasmari YD, Ghazy RM, Alshehri M. Assessment of hemoglobin mass and blood volumes at moderate altitude using carbon monoxide rebreathing method. Medicine (Baltimore) 2025; 104: e42762. doi:10.1097/MD.0000000000042762 PMID 40489857
50. Alkhaldy HY, Alqahtani MM, Al Amri MS, et al. Hemoglobin Mass and Blood Volume at Moderate Altitude: Establishing Lean-Body-Mass-Adjusted Reference Values for Clinical Use. Int J Gen Med 2025; 18: 7203–7211. doi:10.2147/IJGM.S567360 PMID 41355845
51. Alshehri MA, Alkhaldy HY. Kinetics of Blood Volume and Hemoglobin Mass Adaptation in Sea-Level Newcomers Relocating to Moderate Altitude: A Six-Month Prospective Study. J Clin Med 2026; 15: 5833. doi:10.3390/jcm15155833 PMID 42589937
52. Nnebe-Agumadu U, Adebayo I, Erigbuem I, et al. Hydroxyurea in children with sickle cell disease in a resource-poor setting: Monitoring and effects of therapy. A practical perspective. Pediatr Blood Cancer 2021; 68: e28969. doi:10.1002/pbc.28969 PMID 33788390
53. Rosen CL, DeBaun MR, Strunk RC, et al. Obstructive sleep apnea and sickle cell anemia. Pediatrics 2014; 134: 273–281. doi:10.1542/peds.2013-4223 PMID 25022740
54. Sharma S, Efird JT, Knupp C, et al. Sleep disorders in adult sickle cell patients. J Clin Sleep Med 2015; 11: 219–223. doi:10.5664/jcsm.4530 PMID 25515282
55. Willen SM, Rodeghier M, Rosen CL, DeBaun MR. Sleep disordered breathing does not predict acute severe pain episodes in children with sickle cell anemia. Am J Hematol 2018; 93: 478–485. doi:10.1002/ajh.25013 PMID 29266432