Kwashiorkor in the United States: Fad Diets, Perceived and True Milk Allergy, and Nutritional Ignorance

Background  Kwashiorkor is the edematous form of protein-energy malnutrition. It is associated with extreme poverty in developing countries and with chronic malabsorptive conditions such as cystic fibrosis in developed countries. Rare cases of kwashiorkor in affluent countries unrelated to chronic illness have been reported. We present 12 cases of kwashiorkor unrelated to chronic illness seen over 9 years by pediatric dermatologists throughout the United States, and discuss common causative themes in this easily preventable condition.

Observations  Twelve children were diagnosed as having kwashiorkor in 7 tertiary referral centers throughout the United States. The diagnoses were based on the characteristic rash and the overall clinical presentation. The rash consisted of an erosive, crusting, desquamating dermatitis sometimes with classic "pasted-on" scale—the so-called flaky paint sign. Most cases were due to nutritional ignorance, perceived milk intolerance, or food faddism. Half of the cases were the result of a deliberate deviation to a protein-deficient diet because of a perceived intolerance of formula or milk. Financial and social stresses were a factor in only 2 cases, and in both cases social chaos was more of a factor than an absolute lack of financial resources. Misleading dietary histories and the presence of edema masking growth failure obscured the clinical picture in some cases.

Conclusions  Physicians should consider the diagnosis of kwashiorkor in children with perceived milk allergies resulting in frequent dietary manipulations, in children following fad or unorthodox diets, or in children living in homes with significant social chaos. The presence of edema and "flaky paint" dermatitis should prompt a careful dietary investigation.

KWASHIORKOR is the edematous form of protein-energy malnutrition and is distinguished from marasmus by the presence of edema, hypoalbuminemia, and dermatosis.^1,2 It is endemic in Africa and in developing countries and is associated with a diet that is protein deficient.^1,3-5 Cases in affluent countries have usually been associated with chronic malabsorptive conditions such as cystic fibrosis.⁶ Several cases of kwashiorkor in developed countries have been reported previously^7-10; these cases were caused by nutritional ignorance, food faddism, or food allergen avoidance.

We report 12 cases of kwashiorkor in infancy and early childhood seen over 9 years by pediatric dermatologists throughout the United States, and discuss common causative themes.

The medical records of 12 patients with kwashiorkor were reviewed. The patients were seen over a 9-year period. All cases were diagnosed by pediatric dermatologists at major medical centers based on the characteristic rash and the overall clinical presentation, including elements of the medical history, laboratory values, other physical examination findings, skin biopsy results, and response to therapy.

Patient 5 was a 5-month-old white girl who was referred to the Dermatology Clinic, University of California, San Francisco, with a 3-day history of swelling of the lower extremities and vomiting and a 3-week history of a worsening rash on the face, trunk, and diaper area. The rash had been unsuccessfully treated with empirical systemic and topical corticosteroids, topical antifungal agents, antihistamines, and emollients.

Her medical history was significant for chronic constipation. She had been breastfed for 6 weeks and had been switched to a vegan formula of brown rice emulsion, black strap molasses, chlorophyll, acidophilus extract, flaxseed oil, and vitamins. Her family was vegan. Her 3-year-old brother apparently thrived on the same formula but was breastfed longer.

On physical examination, she was an irritable edematous infant. Her height, head circumference, and weight were significantly decreased for her age. She had diffuse, fine, reddish brown scale over her abdomen in a reticulated pattern (Figure 1). Crusting and superficial erosions were noted in the diaper area (Figure 2). Her lower extremities had extensive desquamation and edema (Figure 3).

The results of the laboratory evaluation were unremarkable except for microcytic anemia, low albumin and total protein levels, and mildly elevated liver transaminase levels. A skin biopsy was performed, and the results showed confluent parakeratosis and pallor in the upper third of the epidermis with a sparse superficial perivascular infiltrate (Figure 4). These findings were consistent with the diagnosis of kwashiorkor.

This patient was briefly treated in the intensive care unit with antibiotics (for an Escherichia coli urinary tract infection), albumin, diuretics, and parenteral and nasogastric nutrition. Her rash and edema improved rapidly. She was switched to formula by mouth and was discharged after 7 days of hospitalization.

The medical histories of all 12 patients are summarized in Table 1. Patients ranged in age from 1 to 22 months (mean age, 7½ months). There were 8 boys and 4 girls. Six patients were Hispanic, 2 were white, 2 were black, 1 was Native American, and 1 was East Indian. None of the patients had any serious chronic illnesses. Two patients had a history of atopic dermatitis. One patient was born prematurely at 27 weeks and had a history of asthma and developmental delay.

Half of the cases (patients 5-10) were the result of a deliberate deviation from a standard diet because of a perceived intolerance of formula or milk. The subsequent replacement diets were all deficient in protein. Two patients (patients 1 and 2) had significant financial and social stresses directly contributing to the development of kwashiorkor. Three patients (patients 3, 4, and 12) developed kwashiorkor as a result of parental nutritional ignorance. Other contributing factors included poor feeding skills, secondary ileitis, and suspected milk protein allergies. The duration of symptoms ranged from 2 weeks to 5 months.

The clinical findings and key laboratory values of the 12 cases are summarized in Table 2. Of the 12 patients, 9 were at or less than the 25th percentile for weight and 11 were less than the 25th percentile for height. The most common physical examination findings were rash, irritability, and edema. The rash consisted of a diffuse, erythematous, crusting, erosive, desquamative dermatitis most prominent in the flexures, face, perioral area, and diaper area. Some patients had the classic "pasted-on" scale that has been called "flaky paint" dermatitis (Figure 5 and Figure 6). Four patients had noticeably sparse hair, and 1 had mild depigmentation of his hair for the 3 months before presentation. The classic "flag sign"—alternating light and dark bands of pigmentation along a single strand of hair—was not seen in any of our patients.

Laboratory abnormalities included decreased total protein and albumin levels, elevated hepatic transaminase levels, and evidence of associated iron and zinc deficiency. Of the 5 patients with a low serum zinc level, patients 1, 5, and 8 improved without zinc supplementation, but patients 6 and 10 did receive zinc supplementation. Patient 2 had increased IgA, IgM, and IgG levels, but these results were not believed to be contributory to the cause of the patient's condition, as the tests were drawn to rule out an immunodeficiency. Patient 6 had a positive radioallergosorbent test result to milk, soy, and barley. Patient 10 had an increased stool α₁-antitrypsin level, suggesting a malabsorptive state, but had a negative sweat chloride test result for cystic fibrosis. Other laboratory investigations that had negative or normal results were as follows: DNA analysis for mutations for cystic fibrosis (patient 12); tests for sweat chloride (patients 2, 3, 5, 10, and 11), stool α₁-antitrypsin (patients 5 and 6), thyroid function (patients 1, 2, 5, and 6), serum amino acids (patients 5, 11, and 12), urine organic acids (patients 1, 5, 11, and 12), biotinidase (patients 1 and 12), immunoglobulins (patients 6, 7, and 9), human immunodeficiency virus antibody (patient 11), rapid plasma reagin (patient 11), D-xylose (patient 9), and T-lymphocyte subsets (patient 7); and mitogen studies (patient 2). Skin biopsy specimens were obtained in 7 of the 12 patients and revealed findings consistent with kwashiorkor with either confluent parakeratosis with pallor of the outer third of the epidermis or confluent parakeratosis and psoriasiform hyperplasia.

Initial diagnostic impressions by the referring physicians included atopic dermatitis, viral exanthem, staphylococcal scalded skin syndrome, zinc deficiency, scabies, tinea corporis, Langerhans cell histiocytosis, epidermolysis bullosa, and hypothyroidism. Therapeutic trials of empirical oral and topical therapy had failed in several patients before referral to pediatric dermatologists, which led to delays of up to several weeks in diagnosis.

Eleven patients had a dramatic improvement of their symptoms and signs with caloric and protein supplementation. Patient 12 died of heart failure in the pediatric intensive care unit. The results of an autopsy revealed an enlarged fatty liver, pneumocystis infection of the lung, herpes esophagitis, and bacterial and candidal superinfection of the skin and esophagus. Patient 5 was treated for an E coli urinary tract infection and had a dramatic response to antibiotics and caloric and protein supplementation. Patient 3 had a disseminated cutaneous human herpesvirus infection on presentation and recovered with appropriate therapy.

Williams¹ formally characterized kwashiorkor in 1953 from more than 20 years of observations of children in Africa. In African children, clinical features include edema, fatty liver changes, apathy, irritability, growth failure, depigmentation of the hair, a dermatosis, a low-protein diet (typically associated with weaning), and the onset of symptoms in the second or third year of life. In contrast, our patients and many previously described by North American researchers differ in several respects from those described in Africa. They generally had an age of onset in the first year, had a rapid onset of edema, and lacked hepatomegaly.^7-14

The differences in presentation may be due to differing clinical settings. Although early case reports in the United States were associated with conditions of extreme poverty in rural Louisiana¹⁵ or Kentucky¹⁶ or on Navajo reservations,^17,18 the most recent cases of kwashiorkor in the United States result from intentionally unorthodox diets as a result of food faddism, nutritional ignorance, or food allergen avoidance.^7-9 Elements of all of these causes are found in our case series. Despite these differences, most researchers use the term kwashiorkor to refer to the edematous form of protein-energy malnutrition in general and do not limit the term to the specific presentation in African children.

Only 2 of our cases were associated with poverty. Both families lived in financially impoverished situations, but neither lived in the extreme conditions described in earlier studies.^15-18 In both cases, chaotic social circumstances seemed to have a greater effect on infant feeding practices than did an absolute lack of financial resources.

Nutritional ignorance is an important cause of kwashiorkor in the United States. All 3 of our cases with this cause occurred in families with limited or intermittent access to medical care. In all 3 cases, milk or formula was restricted simply because of the parent's or child's preference, and none of the families knew that their child needed milk or formula. Previously published cases of kwashiorkor due to nutritional ignorance were caused by excessive dilution of formula,⁷ substituting formula with nondairy creamer,⁸ a prolonged clear liquid diet after a hospitalization for gastroenteritis,¹¹ and parents eliminating milk from their child's diet because of a perceived preference on the part of the child.¹²

Half of our cases (patients 5-10) were due to a combination of presumed food allergy, with avoidance of certain foods, nutritional ignorance, and food faddism. All 6 patients had a chronic symptom, such as vomiting, diarrhea, constipation, atopic dermatitis, pulmonary congestion, or colic, that was perceived to be a manifestation of milk or formula intolerance. These symptoms led to dietary manipulations by the parents. The subsequent replacement diets were deficient in protein and ranged from fad diets (Rice Dream milk, vegan brown rice emulsion, and goat's milk) to cultural diets (plantains and lentils, rice flour water, and atholé [a liquid emulsion of barley, water, and cinnamon]). Similar clinical circumstances have been published in the literature. Previous studies^13,14 have described patients who were diagnosed as having a milk allergy by health care providers and who subsequently developed kwashiorkor from substitute diets deficient in protein. Other previous studies^7,17 mention cultural replacement diets, such as banana flakes and water and cannella (a mixture of sugar and water with purported medicinal values in the treatment of diarrhea). Four previous cases^7,19 were attributed to a vegetarian diet. Perhaps the most concerning story in our series is patient 10, who was given Rice Dream, which has 1 g of protein per serving, instead of formula as a treatment for colic at the suggestion of a popular woman's magazine. With the trend toward holistic remedies and food faddism, this scenario is likely to become even more common.

Several patients (patients 1, 2, 11, and 12) had seemingly adequate diets on cursory inspection, which lowered the initial suspicion for kwashiorkor. The patients were subsequently found to have significant social chaos at home, a secondary milk protein allergy, or poor feeding skills. In some of these cases, the dietary history may have been unreliable because of social chaos and language or cultural barriers. Another misleading factor is that several infants (patients 6 and 12) were weaned to solid foods at an early age. Even if the solid foods were high in protein content, it is unlikely that these infants could have met their entire protein requirements with table foods only. Several of the diets had a relative but not absolute protein deficiency. One could speculate that some patients have a decreased threshold for developing kwashiorkor, perhaps because of decreased protein reserves or an underlying malabsorptive condition.

The clinical findings of our 12 cases were consistent with those found in previous reports. The most common physical examination sign was a rash, although this may reflect a referral bias in our series since all of these cases were diagnosed by pediatric dermatologists. The dermatoses ranged from erythematous or hyperpigmented desquamating plaques, most prominent in the flexures and diaper area, to the classic flaky paint dermatitis that, when present, is pathognomonic for kwashiorkor. Edema was almost universally present and sometimes obscured the lack of muscle wasting, giving patients the appearance of "sugar babies."⁹ The edema was noted to have a rapid onset in several of our patients. The absence of the classic flag sign (alternating light and dark bands of pigmentation along a single strand of hair) is probably because our patients did not experience an intermittent nutritional rescue as would be experienced in developing countries. The flag sign has been reported in only 1 patient in the United States.¹⁶

Skin biopsies were performed in 7 of the patients, and the results pointed to the correct diagnosis in all 7. The distinctive histopathologic findings include a superficial perivascular infiltrate of lymphocytes, pallor of keratinocytes in a band across the upper part of the epidermis, and confluent parakeratosis.²⁰ The most specific finding is the pallor, ballooning, and necrosis of keratinocytes in a band across the upper part of the epidermis and is considered nearly pathognomonic for a dermatitis due to a nutritional deficiency.²¹ An alternative pattern that has been described lacks the epidermal pallor and has only psoriasiform epidermal hyperplasia. These histologic patterns, however, are not pathognomonic for kwashiorkor and may also be found in other uncommon conditions that seem to have a nutritional deficiency as a common causative factor. They include acrodermatitis enteropathica, necrolytic migratory erythema, pellagra, methylmalonic acidemia, essential fatty acid deficiency, cystic fibrosis, and multiple carboxylase deficiency.

The pathophysiological features of kwashiorkor have been debated for the past 3 decades. Evidence has been presented implicating aflatoxins,^22,23 free oxygen radicals,^24,25 leukotrienes,^26,27 zinc deficiency,^28,29 and essential fatty acid deficiency^30,31; the traditionally held theory is that of relative protein deficiency.^32,33 The crux of the debate lies in the wide range of presentations and circumstances of this disease and in the wide range of metabolic derangements evident in patients with this disease. Our patients, for instance, were clearly subject to different environmental forces than were patients in India and the Gold Coast of Africa. Kwashiorkor is likely a multifactorial disease for which multiple circumstances, such as dietary imbalance or deficiency, toxin exposure, infection, emotional trauma, and perhaps genetic predisposition, accumulate to form a final common pathway.

Kwashiorkor can be a difficult diagnosis to make because of a low level of clinical suspicion, generalized edema masking the decreased muscle mass, and the lack of familiarity with the typical physical findings. An unreliable dietary history may further obscure the diagnosis. Initial misdiagnosis can lead to delays in corrective therapy and increased morbidity and mortality. The differential diagnosis includes other nutritional deficiencies, such as zinc, free fatty acid, or multiple carboxylase deficiency; immunodeficiencies; metabolic disorders; malabsorption syndromes; cystic fibrosis; and Langerhans cell histiocytosis.

The differentiation of kwashiorkor and zinc deficiency deserves special comment. Five of our patients had low zinc levels, and the argument can be made that their skin lesions could have represented acrodermatitis enteropathica rather than kwashiorkor. Indeed, the 2 conditions have similarities in skin lesions (albeit with differences in distribution) and virtually identical skin biopsy findings. But edema and hypoalbuminemia, which are not features of acrodermatitis enteropathica, were prominent findings in all 5 patients with low zinc levels. Of these 5 patients, 3 (patients 1, 5, and 8) improved rapidly with refeeding without zinc supplementation, and in the 2 others (patients 6 and 10), zinc supplementation was given in the course of their recovery. In these latter 2 patients, it may be impossible to exclude zinc deficiency as a contributing factor, but the presence of edema and hypoalbuminemia make it likely that protein malnutrition was the primary problem.

Patients with this clinical presentation should undergo a laboratory evaluation, including a complete blood cell count and the determination of total protein, albumin, zinc, sweat chloride, and biotinidase levels and of human immunodeficiency virus serologic features. Consideration should be given to obtaining the levels of stool α₁-antitrypsin, urine organic acids, serum amino acids, and immunoglobulins. A biopsy may be helpful in differentiating kwashiorkor from conditions such as atopic dermatitis and Langerhans cell histiocytosis, but the characteristic findings are not specific and can also be found in other nutritional deficiencies.

In conclusion, kwashiorkor does occur in developed countries. Physicians should consider this diagnosis, particularly in patients with frequent dietary manipulations, in those following unorthodox diets, or in those living in homes with significant social chaos.

Accepted for publication August 20, 2000.

Corresponding author: Ilona J. Frieden, MD, Department of Dermatology, Campus Box 0316, University of California, San Francisco, 1701 Divisadero Ave, San Francisco, CA 94143-0316 (e-mail: ijfrieden@orca.ucsf.edu).