Initial PFAPA Registry
A total of 176 PFAPA
Registry forms were returned to the authors. Of this number, 94 patients were
identified with well-documented, regularly recurring periodic febrile episodes;
77 patients were excluded because the registry information did not clearly
document the periodicity of the fever with asymptomatic intervening intervals.
Five patients were excluded because they had diagnoses of other causes for
periodic fevers including cyclic neutropenia (3 patients), Behcet syndrome (1
patient), and Familial Mediterranean Fever (1 patient).
information, clinical characteristics, laboratory results, treatments used, and
clinical outcomes in these 94 patients were reviewed. The sex distribution was
52 male patients and 42 female patients. Patients resided in 22 different states
and 3 foreign countries (Sweden, Italy, and Saudi Arabia). A proportionally
larger number of patients were enrolled from Connecticut and Tennessee,
reflecting the referral population of the authors. Most children were white,
with a diverse representation of ethnic backgrounds that included German,
English, Irish, Italian, French, Jewish, and American Indian. Two children were
black, 1 Hispanic, and 1 Middle Eastern. No consistent complications of
pregnancy or delivery were seen; 6 patients were born prematurely (mean of 6.9
weeks before term). No trends in parental occupation were apparent. Twenty-eight
children were from single-parent families.
The clinical characteristics
of the febrile episodes in the 94 children are described in Table I.
Table I. Characteristics of the febrile
episodes in children with PFAPA at onset and
Registry (95% CI)
|Onset of PFAPA (y)
|Duration of episode
= .0003 for comparison of intervals in original and follow-up
Fevers began at a mean of 2.8 years of age, lasted a mean of 4.8 days, and
recurred at a mean of 28.2 days. Most parents described the sudden appearance of
high fever at the onset of each episode. Prodromes of aphthous stomatitis,
malaise, fatigue, irritability, or headaches occurring 24 hours or less before
the onset of fever were reported in 78%. During episodes the fevers would remain
in the 38.9°C to 41.1°C range for a mean of 3.8 days. In half of the patients
the fevers ended abruptly, whereas in the remainder they resolved over a 24- to
48-hour period. Additional symptoms are reported in Table II.
Table II. Symptoms reported by parents in
children with PFAPA, expressed as percent
data on 94 children were available from original Registry forms, many
parents did not complete this section. The most commonly cited reason was
that children were too young to describe their symptoms to their
†Although follow-up was obtained on 83, only 82 remarked on
‡Refers to the presence of at least 1 of the following in
each child: aphthous stomatitis, pharyngitis, or
Abdominal pain associated with the fever was usually mild and never required
surgical consultation. No patient had documented recurrent pyogenic infections.
In fact, between febrile episodes parents stated that these children had fewer
common infections than their siblings. No patient was given the diagnosis of
arthritis, pleuritis, myositis, or meningitis. Symptoms between episodes were
rare. Five of the 94 children had cervical lymphadenopathy, 4 had aphthous
stomatitis, and 1 reported low-grade fever (<38.3°C) between episodes.
Telephone follow-up was attempted for each of the
94 patients and was successful for 83 patients. All families contacted agreed to
participate in the structured follow-up telephone interview. All 83 subjects
were healthy and exceeded the 5th percentile for height and weight. Four had
neurobehavioral problems: 1 child had attention-deficit hyperactivity disorder,
obsessive compulsive disorder, and Tourette syndrome; 1 had a well-controlled
seizure disorder; 1 had autism; and 1 had attention-deficit hyperactivity
disorder with an oppositional defiant disorder. Parents of 2 children described
difficulties with peer relations and an “introverted” nature in their child that
they attributed to the periodic fever syndrome. None of the 94 patients in the
Registry or the 83 contacted for follow-up had siblings with periodic fever.
Five parents stated they had periodic fevers as young children. No specific
diagnosis was ever made, and their fevers resolved spontaneously in <2 years.
Records from these parents were not available. Fifteen parents had recurrent
aphthous stomatitis, 1 parent had a history of myasthenia gravis, and 11 others
reported uncharacterized episodes of arthritis in themselves or grandparents.
The characteristics of febrile episodes reported in the follow-up survey
are also presented in Table I
. The interval between submission of the PFAPA
Registry form and the follow-up survey ranged from <1 month to 9.4 years,
with a mean of 3.3 years. The mean age of patients with PFAPA at follow-up was 8.9 years. No change in the pattern
of onset or resolution of febrile episodes between the initial and follow-up
reports was noted. Symptoms accompanying febrile episodes in the follow-up
survey are shown in Table II
Thirty-four of 83 children contacted by telephone had not had a
febrile episode for 1 or more years. The mean duration of illness before
resolution of the symptoms in these patients was 4.5 years. Episodes occurring
just before resolution were less frequent (interval of 42.2 days, frequency of
10.4 episodes per year) but were otherwise typical.
Forty-nine of 83
children in the follow-up survey had had febrile episodes within the past year.
Episodes of fever in this group occurred at a mean interval of 40.2 days (95%
confidence interval 29.9 to 50.5 days). The interval between PFAPA episodes in these children had a bimodal distribution
as shown in the Figure.
|Figure. Histogram display
of intervals between PFAPA episodes in 44*
children with fevers within year before follow-up
Click on Image to view full
Thirty-eight children had episodes lasting a mean of 4.2 days with a mean
interval of 26.4 days and a mean of 11.8 episodes yearly. The characteristics of
these attacks were identical to those in the original Registry. The remaining 11
children with persistent febrile episodes had been having episodes for 6.8
years. In these children episodes occurred at a mean interval of 117.5 days.
However, the duration of attacks and the clinical characteristics were similar
to those reported initially in their PFAPA registry
Laboratory Test Results (Table
Table III. Laboratory findings during
evaluation of febrile episodes in children enrolled in PFAPA Registry
ESR, Erythrocyte sedimentation rate; ANA,
antinuclear antibody; RF, rheumatoid factor;
*Group A streptococcus isolated.
Other than leukocytosis (mean leukocyte count 13,000 with 62% polymorphonuclear
leukocytes) and an elevated erythrocyte sedimentation rate (mean 41) during the
episodes, no laboratory study was consistently abnormal. Immunologic and
serologic studies were uniformly nondiagnostic. Distributions of T-lymphocyte
subsets were normal in all 12 patients studied. One patient aged 18 months had
low values for immunoglobulin G (390 mg/dL) and IgA (<7 mg/dL) with normal
IgM. This pattern and his history are most consistent with transient
hypogammaglobulinemia of infancy. Two patients had mildly increased
concentrations of IgM, IgG, and IgA. All other values were within normal range
for age. IgD levels were normal in all 15 patients in which they were measured.
IgE levels were elevated in 8 of 16 patients, with values ranging from 31 to
999. Imaging studies included chest films, sinus films, gastrointestinal series,
computed tomography scans of the head and abdomen, gallium scans, and bone
scans, all of which were negative.
Therapeutic Efforts and Responses (Table
Acetaminophen temporarily reduced temperature in 6% of
patients, whereas ibuprofen reduced temperature in 33%.
Table IV. Efficacy of various agents used in
the management of PFAPA*
evaluated by parents.
Once the pharmacologic effects of these drugs had abated, the fevers recurred
with the same intensity and duration. Most patients had received multiple
courses of various oral antibiotics given either at the onset of fever or
prophylactically. Most parents related that antibiotics had little effect on the
course of each febrile episode, but 6 of the parents stated that episodes were
less severe if antibiotics were given at the onset of fever. The antibiotics
prescribed included penicillins, cephalosporins, macrolides, and sulfonamides. A
few children were treated with aspirin, acyclovir, or colchicine; all of these
were deemed ineffective.
Most of the patients given 1 or 2 doses of
corticosteroids (1 to 2 mg/kg/d prednisone or prednisolone) reported a dramatic
resolution of fever. In addition, most of the associated symptoms resolved, with
aphthous stomatitis being the slowest symptom to respond. Although
corticosteroid therapy did not prevent subsequent episodes of fever, patients
continued to respond on subsequent cycles. However, 9 families reported that the
cycles of fever became more closely spaced after successful treatment with
glucocorticoids. In fact, the reason that only 76% of the parents rated the
steroid therapy as very effective was that symptoms other than fever
occasionally persisted after therapy and that an increased frequency of episodes
was noted in some patients.
Two therapies rated as effective in some of
the patients were cimetidine and tonsillectomy with or without adenoidectomy.
Cimetidine differed from other pharmacologic treatments in that it appeared to
induce remission. Eight of the 28 respondents who used cimetidine (usually 150
mg twice daily for 6 months) reported no further febrile episodes. Tonsillectomy
had been previously associated with resolution of PFAPA recurrences.6
Among 11 patients who had tonsillectomy, febrile episodes completely resolved in
7 (64%). In addition, parents of 2 patients reported a decrease in frequency of
PFAPA episodes after tonsillectomy, whereas 2
families reported no response at all. Adenoidectomy alone was not associated
with resolution of febrile episodes.
The earliest reports of periodic fevers were presented
in the 1940s by Reimann7
and Reimann and deBarardini.8
Using the term periodic to mean episodes recurring at regular intervals,
Reimann described a group of patients with periodic fever beginning in infancy,
persisting for years to decades, having cycles of similar duration, and a
generally benign course.9-11
In 1974 he described an otherwise healthy man with high fever of 5 days’
duration, recurring at intervals of 20 to 23 days for 8 years. He summarized,
“when better known diseases are excluded, recognition of periodic fever obviates
continued unnecessary diagnostic effort and expense. The cause is unknown, and
the treatment is ineffective.”11
The similarity of Reimann’s patients to those with PFAPA is striking, particularly with respect to the
periodic nature of the fever.
A number of other syndromes characterized
by recurrent fever have been reported and are compared with PFAPA in Table V.
Table V. Characteristic of PFAPA versus other intermittent fever syndromes
juvenile rheumatoid arthritis
|Onset at <5 years of
|Length of fever
|Interval between fever
||18 to 24
pain diarrhea splenomegaly rashes
lymphadenopathy, hepatosplenomegaly, arthritis
stomatitis, rare bacterial systemic infections
|Special laboratory test
||Elevated serum IgD
with tooth loss, perforation of abdominal viscus
Familial Mediterranean Fever is characterized by brief recurrent attacks of
painful peritonitis, pleuritis, and fever, but the episodes are not
The hyper-IgD syndrome has recurrent fever, rash, abdominal symptoms, and
elevated serum IgD levels.13
However, with the use of computer analysis to assess periodicity of fever, only
1 of the 50 patients in the largest reported series of this syndrome was
demonstrated to have a periodic febrile pattern.14
Systemic onset juvenile rheumatoid arthritis has hectic spiking fevers,
generalized adenopathy, hepatosplenomegaly, and arthritis. Fever generally
persists for months without remission.15
Cyclic neutropenia is characterized by episodes of fever, aphthous ulcers,
pharyngitis, lymphadenopathy, pyogenic infections, and neutropenia occurring at
intervals of 21 to 24 days.16,17
Other than complicating infections and neutropenia, the clinical manifestations
of cyclic neutropenia and PFAPA are remarkably
similar. Hibernian fever is associated with rash, myalgias, and arthritis but is
Our experience suggests that PFAPA is a
rather common diagnosis among patients referred to our clinics for evaluation of
recurrent fevers. Miller et al19
described the clinical characteristics and outcome of 40 children with prolonged
fever seen in a large pediatric rheumatology clinic; 29 of the 40 patients had
febrile episodes occurring every 21.6 days for an average of 4.6 days, with
maximum temperatures in the range of 40°C; 18% had pharyngitis, 14% oral ulcers,
and 7% lymphadenopathy. These patients differed from ours in having a lower
frequency of pharyngitis, aphthous ulcers, and lymphadenopathy and a much higher
frequency of arthralgia. The authors chose not to assign the diagnosis of PFAPA, considering the syndrome insufficiently defined.
However, the periodic nature of the fever pattern in their patients is
strikingly similar to that seen in patients with PFAPA.
The clinical criteria proposed for PFAPA in 1989 are confirmed by this follow-up survey and
particularly the stability of the manifestations of the syndrome over time.
Parents of these children have no difficulty in distinguishing PFAPA attacks from other common infectious causes of fever
in young children and correctly anticipate when the episodes will occur. Several
mothers have commented that episodes recur as regularly as menstrual cycles, and
families often make plans based on anticipated dates of febrile episodes. The
periodic fever is the most characteristic feature, whereas the associated
symptoms are slightly more variable. Of the 94 patients evaluated in this
report, pharyngitis, aphthous stomatitis, and cervical adenopathy were seen in
70% to 88% of the children. At least 1 of these cardinal signs was present in
all patients evaluated at follow-up. The increase in the number of children with
all 3 symptoms between enrollment and follow-up may reveal a propensity for
symptoms to manifest more completely with time or may be attributable to more
complete examinations by physicians with a high degree of clinical suspicion. We
have modified the laboratory component of the 1989 diagnostic criteria5
to include evaluation for cyclic neutropenia and exclude an elevated erythrocyte
sedimentation rate and leukocytosis (Table VI).
Table VI. Diagnostic criteria used for PFAPA
fevers with an early age of onset (<5 years of age)
in the absence of upper respiratory infection with at least 1 of the
following clinical signs:
||Exclusion of cyclic
interval between episodes
||Normal growth and
These latter results are frequently present in febrile children and do not
provide increased specificity for the diagnosis.
Our survey indicates
that PFAPA episodes persist for several years with
unchanged symptoms and periodicity. Remission appears to be preceded by a period
of time during which attacks occur with decreasing frequency. Thirty-four of the
83 patients in the follow-up group had stopped having episodes after a mean of
4.5 years of illness. Two patients continue to have episodes after more than 17
years, but symptoms, particularly fever, are milder, and episodes are much less
frequent. All children were described as healthy; none had subsequently been
diagnosed with malignancy, autoimmune disorders, or chronic infectious diseases.
Neurobehavioral disorders were present in 4 patients, a prevalence probably not
different from that in the general population of children.
evaluations of the efficacy of various drugs are consistent with observations in
our 1987 report.1
Acetaminophen, antibiotics, and nonsteroidal anti-inflammatory drugs were
reported to have modest therapeutic benefit. Corticosteroids aborted symptoms
remarkably well in most patients. Although a single dose of 1 to 2 mg/kg
prednisone or equivalent was often sufficient to halt the episode, some patients
defervesced only after a longer course. Cimetidine was suggested as a
prophylactic treatment for PFAPA in 1992.4
Although the overall reported efficacy of this drug in our patients at follow-up
was only 29%, it appeared to be highly effective in preventing attacks in some
patients. Several mothers reported that a cimetidine-induced remission was
reversed on discontinuation of the drug.
Of special interest was the
observation of cessation of PFAPA episodes after
tonsillectomy (with or without adenoidectomy) in 7 of 11 patients undergoing the
procedure. The febrile episodes decreased, whereas recurrent aphthae persisted
in 2. In 1989 Abramson et al6
reported 4 children with recurrent fevers for 6 to 12 months with pharyngitis
and adenitis poorly responsive to antibiotics who had complete resolution after
they underwent tonsillectomy.6
We do not believe these limited data warrant recommending tonsillectomy for
management of PFAPA unless there are other
indications for tonsillectomy or we gain a better understanding of its mechanism
of action. Intermittent, very short courses of glucocorticoids are well
tolerated and effective in relieving the distressing symptoms of PFAPA attacks. We recommend a dose of 1 mg/kg prednisone or
prednisolone at the beginning of an attack, the same dose on the next morning,
and one half of that dose on days 3 and 4 as a starting point. Doses on days 3
and 4 may be omitted in some patients, as determined by trial during subsequent
The cause of PFAPA is unknown. One
clue may be the remarkable similarity of uncomplicated episodes of cyclic
neutropenia and febrile attacks in PFAPA. Both are
characterized by periodic fever, pharyngitis, mouth ulcers, and
Cyclic neutropenia is postulated to be caused by an unidentified defect in
hematopoietic precursor cells20
or by alterations in the regulation of cytokines.21
Perhaps PFAPA and cyclic neutropenia share common
pathways of immune disregulation. The ability of a single dose of corticosteroid
to abort attacks of PFAPA suggests that the symptoms
may be caused by inflammatory cytokines rather than by infection. Preliminary
studies of cytokines in patients with PFAPA indicate
that several cytokines are elevated during febrile episodes, most notably -interferon, tumor
necrosis factor, and interleukin-6 (Wilson C et al, unpublished data).
In summary, PFAPA is a not uncommon syndrome
of periodic fevers in children. The syndrome is easily recognized by the
predictable recurrence of episodes at intervals of 3 to 6 weeks and the
associated symptoms of pharyngitis, aphthous ulcers, and
cervical/lymphadenopathy. Cyclic neutropenia is excluded by normal to elevated
leukocyte counts at the beginning of the fevers. In most cases other causes of
recurrent fever can be excluded easily by the absence of periodicity and the
presence of a different set of associated symptoms. The PFAPA syndrome may persist for several years but has no
detrimental long-term health consequences.
We are grateful for the assistance received from the
multiple referring physicians and parents of the patients described.
1. Marshall GS, Edwards KM, Butler
J, Lawton AR. Syndrome of periodic fever, pharyngitis, and aphthous
stomatitis. J Pediatr 1987;110:43-6.
2. Feder HM Jr, Bialecki CA. Periodic
fever associated with aphthous stomatitis, pharyngitis and cervical adenitis.
Ped Infect Dis J 1989;8:186-7.
3. Rubin LG, Kamani N. Syndrome of
periodic fever and pharyngitis (letter). J Pediatr 1987;110:307.
4. Feder HM Jr. Cimetidine treatment for
periodic fever associated with aphthous stomatitis, pharyngitis and cervical
adenitis. Ped Infect Dis J 1992;11: 318-21.
5. Marshall GS, Edwards KM. PFAPA
syndrome (letter). Ped Infect Dis J 1989;8:658-9.
6. Abramson JS, Givner LB, Thompson JN.
Possible role of tonsillectomy and adenoidectomy in children with recurrent
fever and tonsillopharyngitis. Ped Infect Dis J 1989;8:119-20.
7. Reimann HA. Periodic disease: a
probable syndrome including periodic fever, benign paroxysmal peritonitis,
cyclic neutropenia and intermittent arthralgia. JAMA
8. Reimann HA,
deBerardinis CT. Periodic (cyclic) neutropenia, an entity. Blood
Periodic disease. Medicine 1951;30:219-45.
10. Reimann HA. Periodic fever, an
entity: a collection of 52 cases. Am J Med Sci 1962;243:162-74.
11. Reimann HA, McCloskey RV. Periodic
fever: diagnostic and therapeutic problems. JAMA 1974;228:1662-4.
12. Wolff SM. Familial Mediterranean
Fever (Familial Paroxysmal Polyserositis). In: Wilson JD, et al, editors.
Harrison’s principles of internal medicine. 12th ed. New York: McGraw Hill;
13. van der
Meer JWM, Vossen JM, Radl J, Van Nieuwkoop JA, Meijer CJLM, et al.
Hyperimmunoglobulinema D and periodic fever: a new syndrome. Lancet
14. Drenth JP, Haagsma CT, van der Meer
JW. Hyperimmunoglobulinemia D and periodic fever syndrome. The clinical
spectrum in a series of 50 patients. International Hyper-IgD Study Group.
15. Condemi JJ. The autoimmune diseases.
16. Wright DG, Dale DC, Fauci AS, Wolff
SM. Human cyclic neutropenia: clinical review and long-term follow-up of
patients. Medicine 1981;60:1-13.
17. Arav-Boger R, Spirer Z. Periodic
syndromes of childhood. Adv Peds 1997; 44:389-428.
18. Williamson LM, Hull D, Mehta R,
Reeves WG, Robinson BHB, Toghill PJ. Familial hibemian fever. Quart J Med
19. Miller LC,
Sisson BA, Tucker LB, Schaller JG. Prolonged fevers of unknown origin in
children: patterns of presentation and outcome. J Pediatr 1996;129:419-23.
20. Dale DC, Hammond WP. Cyclic
neutropenia: a clinical review. Blood Rev 1988;2:178-85.
21. Engervall P, Andersson B, Bjorkholm
M. Clinical significance of serum cytokine patterns during start of fever in
patients with neutropenia. Br J Haem 1995;91:838-45.
- From the School of Medicine, Vanderbilt
University; the Department of Pediatrics, University of Connecticut,
Farmington, Connecticut; the Department of Pediatrics, the Division of
Immunology and Rheumatology, and the Division of Infectious Diseases,
Vanderbilt University, Nashville, Tennessee.
- Submitted for publication Nov 20,
- Revised Jan 26, 1999.
- Accepted Feb 16, 1999.
- Reprints not available from
Copyright © 1999 by Mosby, Inc.
- 0022-3476/99/$8.00 +