Chronic Cough

The Spectrum and Frequency of Causes, Key Components of the

Diagnostic Evaluation, and Outcome of Specific Therapy1-3

RICHARD S. IRWIN, FREDERICK J. CURLEY, and CYNTHIA L. FRENCH

Introduction
In 1981 (1), a systematic manner of eval-
uating patients with chronic cough was
presented. Because it was based upon
evaluating the locations of the afferent
limb of the cough reflex, it was designat-
ed an anatomic, diagnostic protocol. The
protocol was designed to encourage cli-
nicians to consider extrapulmonary as
well as pulmonary conditions as poten-
tial causes of chronic cough.
Utilizing the anatomic, diagnostic pro-
tocol, it was reported that it was possible
to consistently determine the cause(s) of
chronic cough. In so doing, other impor-
tant findings emerged (1, 2). First, al-
though cough was most commonly due
to a singlecause (82010), it also could have
dual causes (18010). Second, cough most
commonly (96010) was due to four dis-
orders, postnatal drip syndrome, asth-
ma, chronic bronchitis, and gastroesoph-
ageal reflux (GER). Third, although most
smokers have a cough, they were not the
group of patients who most commonly
sought medical attention complaining of
cough. Fourth, when history, physicalex-
amination, and chest roentgenograms did
not suggest a diagnosis, methacholine in-
halational challenge (MIC) was deter-
mined to be the first additionallabora-
tory test that should be routinely ordered.
It was the only means of identifying 57010
of the asthmatics. Fifth, unless the chest
roentgenogram was abnormal, and this
was an uncommon occurrence (4010),
fiberoptic bronchoscopy had a very low
diagnostic yield (4010). Sixth, if the
cause(s) of cough could be determined,
SUMMARY A suCCHSfuI, aystem8tlc, anatomic, diagnostic protocol for evaluating petlenta with
chronic cough waspresented In 1981.Todetermine whether It was stili VIIlld,we prospectively evalu-
ated, over a 22-month Interval, 102 consecutive and unselected Immunocompetent pldlenta com-
plaining of cough an average of 53 ± fIT months (range, 3 wk to 50 yr). Utilizing the anatomic,
diagnostic protocol modified to Include prolonged eeophageel pH monitoring (EPM), the cau_
of cough weredetermined In 101of 102 (99~) petlenta, ludlng to specific therapy that ... succeu-
fulln 98~. Cough was due to one condition In 73~, two In 23~, and th.... In 3~. Postnaul drip
syndrome was a cause 41tH. of the time, asthma 24tH., gastroe8ophageall8ftux (GER)21tH., chronic
bronchitis 5~, bronchiectasis 4tH., and miscellaneous conditions 5tH.. Cough wasthe sole present-
Ing manifestation of asthma and GER 28 and 43~ of the time, respectively. While history, physical
examination, methacholine Inhalatlonsl challenge (MIC), and EPM yielded the most fntquent true
positive results, MIC ... falsely positive 22~ of the time In predicting that asthma ... the cause
of cough. Laboratory testing was partlcularty useful In ruling out suspected possibilities. We con-
clude that the anatomic diagnostic protocol Is stili VIIIId and that It has well-defIned strengths
and limitations. All REV RESPIR DIS 1990; 141:640-647
ent physicians to manage chronic cough ing been told by their physician that they
varied. Even though 140/0 of the patients would have to learn to live with their cough
had been correctly diagnosed by their because the cause could not be determined.
referring physician, they had been in- The 108patients represented 380/0 of the new
patients seen by this author during the time
effectively treated. of the study. The entrance criterion of a du-
Since 1981, we have sensed that the pa-
ration of at least 3 wk was part of our 1981
tients referred to us with chronic cough study (1); it was again chosen in an attempt
of unknown etiology have become more to eliminate patients who had the usually self-
complicated and difficult to diagnose. To limited and transient cough associated with
substantiate our impression, reassess our the common cold. Because six patients failed
success rate, and determine whether or to return for follow-up visits, our study group
not our diagnostic protocol was still val- consisted of 102 fully evaluated patients (43
id, we decided to prospectively evaluate men and 59 women 6 to 83 yr of age, with
another series of patients with chronic a mean age of 51 ± 18).They had complained
of cough for an average of 53 ± 97 months
cough of unknown etiology. Because we (range, 3 wk to 50 yr).
had become aware that chronic cough
may be the sole presenting manifestation
of GER (3), we incorporated prolonged
esophageal pH monitoring into our di- (Received in original form April 13, 1989 and in
agnostic protocol. revised form August 21, 1989)
Methods 1 From the Pulmonary and Critical Care Medi-

specific therapy had an excellent chance
of being successful. Specifictherapy, that
is, therapy directed at the etiology or
operant pathophysiologic mechanism,
was eventually successful in eliminating
cough as a complaint in 900/0 of the pa-
tients overall and 970/0 of the patients
who followed their treatment plan as
prescribed. Seventh, the ability of differ-
Patients
From July 9, 1985to May 15, 1987, 108con-
secutive and unselected, immunocompetent
patients with chronic cough of unknown eti-
ology with a duration of at least 3 wk, previ-
ously seen byat least one other physician,were
seen by RSI in our pulmonary outpatient clin-
ic. Seventy-seven percent were referred by a
physician; 23% were self-referred after hav-
cine Division, Department of Medicine, Univer-
sity of Massachusetts Medical School, Worcester,
Massachusetts.
1 Presented in part at the Annual Meeting of the
American Thoracic Society, Las Vegas, Nevada,
May 11, 1988 (Am Rev Respir Dis 1988; 137:330)
J Correspondence and requests for reprints should
be addressed to Dr. Richard Irwin, Pulmonary and
Critical CareMedicine Division, University of Mas-
sachusetts Medical Center, 55 Lake Avenue North,
Worcester, MA 01655.

640
 CHRONIC COUGH
Diagnostic Protocol
The diagnostic protocol used in this study was
a refinement of the 1981 approach and is as
follows.
(1) For all patients, a history was taken and
a physical examination was performed, con-
centrating on anatomic locations of the af-
ferent limb of the cough reflex. With rare ex-
ception, a chest roentgenogram was ordered
in all.
(2) For current smokers and patients ex-
posed to environmental irritants, no addition-
al studies wereordered for 4 wk until response
to elimination of the irritant was assessed.
(This only applied to six patients in this study;
they composed our chronic bronchitis group.)
(3) Depending on the results of the initial,
routine evaluations and elimination of irri-
tants, none, some, or all of a range of studies
were done. If the history, physical examina-
tion, and chest roentgenogram did not sug-
gest an etiology, the first additional study or-
dered was pulmonary function testing (spi-
rometry before and after administration of
bronchodilator or methacholine). On the oth-
er hand, if the history and physical examina-
tion suggested postnasal drip syndrome, sinus
roentgenograms and an allergy evaluation
were obtained before the pulmonary function
studies, If the chest roentgenogram was ab-
normal and consistent with cancer or infec-
tion, expectorated-sputum examinations, fi-
beroptic bronchoscopy, or both were ordered.
(4) If an etiology still did not emerge, tests
for GER were ordered, even in the absence
of upper gastrointestinal symptoms. If the
barium swallow was negative, prolonged
esophageal pH monitoring was performed.
As a last resort (in the absence of any radio-
graphic and clinical findings suggesting the
cause of cough), fiberoptic bronchoscopy,
cardiac studies, or both were ordered.
(5) When the history, physical examination,
and laboratory evaluation suggested more
than one cause, the cause(s) of chronic cough
was determined by observing which specific
therapy eliminated cough as a complaint.
Pulmonary function studies consisted of
spirometry before and after inhalation of
metaproterenol or methacholine. FVC was
performed on a IO-L survey spirometer (No.
06031; Warren E. Collins, Braintree, MA).
From the best of three forced spirographic
tracings, maximal midexpiratory flow
(FEF25 - 75 " . ) and FEV I were measured. The
FEV I was then expressed as a percentage of
FVC, FEV I/FVC07o. If FEV I/FVC07o was ab-
normal (4-6), spirometry was repeated 30 min
after the inhalation of two puffs (1.3 mg) of
a metered dose of metaproterenol. If FEV 1/
FVC070 was normal, patients weregivena MIC
to detect cough-variant asthma (7). We de-
termined a MIC as the response in FEV I to
inhaled methacholine (acetyl beta-methacho-
line) diluted in physiologic saline. A positive
response, consistent with symptomatic asth-
ma was defined as a decrease in FEV I from
baseline of 20070 or greater after the inhala-
tion of 195 or less cumulative dose units of
methacholine. Our method of performing
MIC has been previously reported (8).
GER was evaluated by upper gastrointes-
tinal roentgenograms, esophagoscopy, ma-
nometry, and/or prolonged esophageal pH
. monitoring. Reflux esophagitis was endoscop-
ically interpreted according to the criteria of
Sonnenberg and coworkers (9). Esophageal
motility studies were performed using a Nar-
co Biosystems MMS-l00 apparatus and in-
terpreted according to standard methods and
criteria (10-12). Esophageal pH sensors were
positioned under radiologic control (13). Af-
ter determining by barium swallow whether
or not pharyngeal dysfunction with or with-
out aspiration, GER, dysmotility, and/or a
hiatus hernia were present and the location
of the gastroesophageal junction, two cath-
eters 3 mm in diameter with calibrated poly-
crystalline and antimony pH electrodes at
their tips (Biosearch Medical Products, Som-
erville, NJ) were passed transnasally under
fluoroscopy into the esophagus. The presence
and size of a hiatus hernia or GER were de-
termined by having the patient carry out a
Valsalva maneuver or the examiner compress
the patient's abdomen in the upright and su-
pine positions. The distal sensor was posi-
tioned at least 6 em above the gastroesopha-
geal junction. Unless it had been determined
that a hiatus hernia was present, the record-
ing tip was found to lie, as has been previous-
ly determined by' Spencer (13), at the junc-
tion of the middle and lower thirds of the
esophagus. If a hiatus hernia was present, the
distal sensor was still positioned 6 em above
the gastroesophageal junction; however, its
position in the chest was adjusted upwards
the length of the hernia. The proximal sen-
sor was positioned in the hypopharynx/prox-
imal esophagus at least 2 em above the tho-
racic inlet. Both catheters were secured at the
nose with tape and attached to Ambi-24
Digitrappers (Biosearch Medical Products)
with patient event marker buttons. Prior to
insertions, catheters and digitrappers were
calibrated using pH buffers of 1.00 and 7.00.
The reference wire had been modified to al-
Iowa single reference voltage to be used by
both recorders.
During the monitoring period, patients
filled out an event/symptom diary that had
simultaneously running columns for (1) event
marker pressed, (2) time, (3) positions (su-
pine and upright), (4) meal start, (5) meal fin-
ish, (6) reflux symptoms, (7) medication tak-
en, (8) went to bed, (9) got up, (10) snack start,
(11) snack finish, (12) cough, and (13) com-
ments. Before the patients recorded their di-
ary entries, they were instructed to first simul-
taneously activate the event marker buttons.
Although patients were not allowed to take
any medications for GER, they were allowed
unrestricted ambulation and a normal diet
except for absence of food and beverage hav-
ing a pH less than 5.
After completing the monitoring period,
the catheters were removed, and the digitrap-
pers were interfaced with an IBM XT com-
641
puter and Epson printer for displaying and
charting of collected data. Tracings and di-
aries were integrated and analyzed by hand
for the following: (1) adequacy of calibration
curves of pH 1.00and 7.00, (2) longest reflux
event in minutes in supine and upright posi-
tions, (3) percentage of 24 h that pH < 4, (4)
percentage of time that pH < 4 in supine and
upright positions, (5) number of reflux events
~ 5 min, (6) number of coughs and reflux
symptoms per 24 h, (7) number oftimes cough
and reflux events occur simultaneously, and
(8) number of reflux events per 24 h. Using
the method of Ward and coworkers (14), the
beginning of a reflux event was defined by
a fall in pH < 4, the end by a rise to pH >
5. A cough was considered to be induced by
reflux only if cough occurred simultaneous-
ly with a pH < 4 and it followed the decrease
in pH < 4. GER was considered a potential
cause of cough when any GER parameter was
abnormal (14, 15) and/or a reflux event in-
duced cough. A discussion of how to inter-
pret esophageal pH monitoring data in evalu-
ating GER as a cause of chronic cough has
been previously published (3).
Pretreatment Diagnostic Criteria
Prospective criteria were established for the
presumptive diagnoses of postnasal drip syn-
drome, asthma, GER, chronic bronchitis, and
bronchiectasis as the cause of chronic cough.
Postnasal drip syndrome was considered
when (1) patients described the sensation of
having something drip down into their
throats, nasal discharge, and/or the need to
frequently clear their throats, or (2) physical
examination of the nasopharynges and oro-
pharynges revealed mucoid or mucopurulent
secretions and/or a cobblestone appearance
of the mucosa.
Asthma was considered when (1) patients
complained of episodic wheezing, shortness
of breath plus cough, and were heard to
wheeze, or (2) reversible airflow obstruction
was demonstrated by pulmonary function
testing (FEV I increased at least 15070 from
baseline and approached normal after
metaproterenol even in the absence of
wheeze), or (3) MIC was positive in the pres-
ence of normal routine spirometry and ab-
sence of wheeze. The diagnosis of asthma was
not made in any patient who had experienced
an obvious respiratory tract infection within
2 months prior to examination and in whom
cough was transient and self-limited.
GER was considered when (1) patients
complained of heartburn and a sour taste in
their mouths, or (2) upper gastrointestinal
contrast roentgenograms demonstrated reflux
of barium, esophagoscopy with biopsy showed
esophagitis, and/or prolonged esophageal
monitoring was abnormal in the absence of
upper gastrointestinal complaints.
Chronic bronchitis was considered when
patients met the criteria set forth by the Brit-
ish Medical Research Council (16)and failed
to demonstrate reversibility when pulmonary
function studies revealed airflow obstruction
842 IRWIN, CURLEY, AND FRENCH

(FEV I remained reduced after increasing no
more than 140/0 from baseline after the inha-
lation of metaproterenol).
Bronchiectasis was considered when (I) pa-
tients complained of cough with expectora-
tion of> 2 tablespoons of discolored phlegm
in 24 h, or (2) chest roentgenogram demon-
strated changes typical of bronchiectasis (17).
Posttreatment Diagnostic Criteria
The final diagnosis of the cause of cough re-
quired fulfillment ofpretreatment criteria plus
having cough disappear with specific thera-
py. Specific therapy for postnasal drip syn-
drome depended upon the etiology; the eti-
ology was diagnosed by clinical criteria (18)
and sinus roentgenograms (19). Allergic,
perennial nonallergic, postinfeetious, environ-
mental irritant, and vasomotor rhinitis were
treated predominantly with intranasal bee-
lomethasone dipropionate, occasionally with
an antihistaminic decongestant preparation
(dexbrompheniramine maleate plus d-iso-
ephedrine) and, when feasible, avoidance of
environmental precipitating factor(s); vaso-
motor rhinitis that failed to respond to the
above measures was treated with intranasal
ipratropium bromide; sinusitis was treated
with a combination of antibiotic, deconges-
tant nasal spray (oxymetazoline hydrochlo-
ride) and dexbrompheniramine maleate plus
d-isoephedrine. Asthma was usually treated
with bronchodilators alone or, on occasion,
with corticosteroids. Oral rather than inhala-
tional bronchodilators were prescribed when
Statistical Analyses
Probability statistics (21)were used to describe
the testing characteristics of individual com-
ponents of the diagnostic protocol in terms
of sensitivity, specificity, positive predictive
value, and negative predictive value. A test
was deemed true positive or false positive on
the basis of the response to specific therapy.
If it led to specific therapy that was success-
ful, it was determined to be a true positive
result. Differences between groups were com-
pared with Student's t test and chi-square anal-
ysis. The 0.05 level of significance (type I er-
ror rate) was used throughout.
Results
Spectrum and Frequency of the
Causes of Chronic Cough
Using posttreatment diagnostic criteria,
the cause(s) of chronic cough was deter-
mined in 101of 102 (99%) of the patients.
Although cough was due to a single con-
dition in 73% of patients (figure I), it
was caused by multiple disorders in 26070.
Three percent of the time, a patient's
cough was due to three concurrently ex-
isting conditions. The only patient whose
100
75
73~
cough remained undiagnosed after our
evaluation met pretreatment criteria for
a postnasal drip syndrome secondary to
chronic sinusitis, cough-variant asthma,
and GER; yet cough did not improve af-
ter prescribing therapy for all three pre-
sumptive diagnoses.
In the 102patients, 131 causes of cough
were identified. Their spectrum and fre-
quency are shown in figure 1. Postnasal
drip syndrome was the single most com-
mon cause of chronic cough; it was a
cause 41070 of the time. It was frequently
due to more than one of the following
upper respiratory tract conditions: sinus-
itis (39070), allergic rhinitis (23070), peren-
nial nonallergic rhinitis (37070), postin-
fectious rhinitis (6070), vasomotor rhini-
tis (2070), drug-induced (ACE inhibitor)
(2070), and environmental irritant (chlo-
rine gas) (2070). There were 54 patients
whose cough was partially or solely due
to postnasal drip syndrome. Their mean
age was 48 ± 21 yr (range, 6 to 80 yr).
They had complained of cough for an
average of 45 ± 96 months (range, 3 wk
to 600 months).

the latter were observed to provoke coughing

in clinic. GER was treated with a high pro- CAUSES
tein, low fat, antireflux diet, eating three meals OF 50
a day, not eating or drinking for 2 to 3 h prior COUGH
%
to lying down except for taking medications,
head of bed elevation, and metoclopramide
25
and/or H2 blockers. Chronic bronchitis was
treated solely with the cessation of smoking
3~
or the elimination of the irritant from the en- 1~

vironment. Angiotensin-converting enzyme 0

2 3 UNKNOWN
(ACE) inhibitor-induced cough was treated
with the cessation of the medication. Bron-
chiectasis was treated with antibiotics, chest
100
physiotherapy and postural drainage, and the-
86'\.
ophylline and/or beta-agonists. Bronchogenic
carcinoma was treated with resectional lung
surgery, sarcoidosis with corticosteroids, and 75
CAUSES
left ventricular failure with furosemide and, OF
if indicated, digoxin. COUGH
If more than one disorder was implicated 0/ 0 50
as the cause of cough, the effect of therapy
on the more clinically prominent condition
was evaluated before adding treatment for the 25
other. No patient received nonspecific, symp-
tomatic therapy (20) as long as specific ther-
apy was available. The success rate of specif-
ic therapy was assessed during follow-up clinic o
PND ASTHMA GER CHRONIC BRaN· MISC PND
visits or by telephone when patients lived out BRaNCH CHIECT ASTHMA
of state and country. Therapy was deemed + Ior GER
successful when patients no longer com- Fig . 1. The causes of chronic cough. Top panel. The cause was determined in 99% of patients; it was due to
plained of cough because it had markedly im- a single condition in 73% of patients and to multiple disorders in 26%. Bottompanel. The spectrum and frequency
proved, was controlled with treatment, or of the 131causes (PND = postnasel drip syndrome; GER =. gastroesophageal reflux ; Bronch = bronchitis; bron-
disappeared. chiect = bronchiectasis; mise = miscellaneous).
CHRONIC COUGH 643

75 70 '1>
Asthma wasthe second most common
condition diagnosed, causing cough 240/0
of the time. In 28% of these patients,
50
cough was the sole presenting manifesta-
USEFUL-
tion of their asthma (i.e., cough-variant NESS
asthma). There were 32 patients in the (%)

group. Their mean age was 48 ± 18 yr

(range, 13to 74 yr). They had complained
of cough for an average of 48 ± 71
months (range, 3 wk to 240 months). HX PE PFTs MIC UGI Esoph.
GER was the third most common di- Studies pH

agnosis made; it was a cause of cough

21 % of the time. In 43 % ofthese patients,
cough was the sole presenting clinical
manifestation of their GER (i,e., "silent"
GER). There were 28 patients in this
group; nine have previously served as sub-
jects in a published study (3). Their mean
age was 56 ± 12 yr (range, 29 to 83 yr).
They had complained of cough for an
average of 58 ± 76 months (range, 3 wk
to 240 months).
Chronic bronchitis was the fourth most
common cause of cough, occurring with
a frequency of 50/0. In 83% of these pa-
tients, chronic bronchitis was due to the
inhalation of tobacco smoke, in 17% to
the inhalation of dust in an industrial set-
ting. There weresix patients in this group.
Their mean age was 53 ± 8 yr (range,
41 to 64 yr). They had complained of
cough for an average of 99 ± 188months
(range, 3 months to 480 months).
Bronchiectasis was the fifth most com-
mon diagnosis made, causing cough 4%
of the time. There were five patients in
this group. Their mean age was 66 ± 11
yr (range, 55 to 83 yr). They had com-
plained of cough for an average of III
± 144 months (range, 6 wk to 312
months).
Miscellaneous conditions were diag-
nosed 5% of the time. Cough was due
Fig. 2. The relative usefulness (true positive result) of each component of the diagnost ic protocol in determining
the 131 causes of cough (HX = history; PE = physical examination; PFTs = pulmonary function tests; MIC =
methacholine inhalational challenge; UGI = upper gastrointestinal; esoph. pH = prolonged esophageal pH monitor-
ing ; bronch = flexible fiberoptic bronchoscopy).
other patient was placed in the postnasal Usefulness of the Diagnostic Protocol
drip diagnostic group since her medica- The relative usefulness of various com-
tion appeared to cause cough by a post- ponents of the diagnostic protocol in de-
nasal drip mechanism (22). termining the 131 causes of cough, when
Overall, chronic cough was due to post- true positive results only were taken into
nasal drip syndrome, asthma, and/or account, are shown in figure 2.
GER 86% of the time. In the 24 patients In the postnasal drip syndrome group,
whose coughs were due to two concur- 100% of patients complained of post-
rently existing conditions, the diagnoses nasal drip, throat clearing, or nasal dis-
were as follows: postnasal drip syndrome charge and/or had mucus or a cobble-
and asthma in 13, asthma and GER in stone appearance in the oropharynx. The
six, postnasal drip syndrome and GER prevalences of each of these complaints
in three, GER and bronchiectasis in one, and physical signs in this group of pa-
and GER and industrial bronchitis in tients and other diagnostic groups appear
one. In the three patients whose coughs in table 1; for statistical purposes, only
were due to three concurrently existing patients with one of the five major causes
conditions, the diagnoses were as follows: of cough were tabulated. In the absence
postnasal drip syndrome, asthma, and of any upper respiratory tract symptom
GER in two, and Zenker's diverticulum, or sign (e.g., postnasal drip, throat-
bronchiectasis, and GER in one. clearing, nasal discharge, oropharyngeal
TABLE 1
PREVALENCES OF UPPER RESPIRATORY SYMPTOMS AND SIGNS·
Diagnostic Groups

to bronchogenic carcinoma in two pa- Chronic

tients, left ventricular failure in one, sar- PNDS Asthma GER Bronch itis Bronchiectasis
coidosis in one, Zenker's diverticulum in Variables (n = 36) (n = 11) (n = 14) (n = 3) (n = 5)

one, and drug-induced (ACE inhibitor) Symptoms

in one. There were six patients in this PND 61 64 57 33 60
group. Their mean age was 60 ± 17 yr TC 72 64 79 33 80
NO 36 36 29 67 20
(range, 34 to 83 yr). They had complained PND or TC 81 73 93 33 80
of cough for an average of 4 ± 17months PND + TC 53 55 42 33 60
(range, 1 month to 12 months). PND, TC, NO 25 27 0 33 0
Overall, ACE inhibitor drug-induced PND, TC, or NO 92 73 93 67 100
cough was diagnosed in two patients. The Signs
patient categorized here did not meet Cob 67 64 64 0 40
Muc 56 46 50 67 80
presumptive criteria for postnasal drip
Cob + Muc 31 36 29 100 40
syndrome, asthma, chronic bronchitis, Cob or Muc 92 73 86 67 80
bronchiectasis, and her chest roentgeno-
Symptoms and signs
gram was normal. Although she had an PND, TC, Cob, or Muc 100 82 100 67 80
abnormal prolonged esophageal pH PND, TC, NO, Cob, or Muc 100 82 100 100 100
monitoring session in the absence of
Definitionof abbreviations: PNDS - postnasal drip syndrome; GER - gastroesophageal reflux ; PNO - postnasal drip ; TC -
GER symptoms, her cough disappeared throat.<:Jearing; NO - nasal discharge; Cob - cobblestone appearance; Muc - mucus .
with cessation of an ACE inhibitor. The • Only patients with one 01 flve major causes 01 cough have been tabulated.
644 IRWIN, CURLEY, AND FRENCH

mucus, or cobblestone appearance), the TABLE 2 had cystic fibrosis or a defect in humoral
cause was more likely to be asthma (p RESULTS OF HISTORY, BARIUM SWALLOW, immunity, one had GER and Zenker's
= 0.028). Otherwise, no single upper re- AND PROLONGED ESOPHAGEAL pH diverticulum and one had GER that may
spiratory symptom or sign or combina- MONITORING IN THE GER GROUP have initiated and/or perpetuated the
tion thereof was significantly more fre- Patient Barium Esophageal bronchiectatic condition.
quent in any diagnostic group. Chronic No. Symptoms Swallow pH Monitoring In the miscellaneous group, diagnosti-
sinusitis was diagnosed as the cause of 1 + NO NO
cally useful components of the protocol
the postnasal drip syndrome 39070 of the 2 + NO + were chest roentgenograms and bron-
time based upon sinus roentgenograms 3 + + + choscopy in the two patients with bron-
revealing > 6 mm of mucosal thicken- 4 + + + chogenic carcinoma; history, physical ex-
ing, air-fluid levels, or opacification of 5 + + + amination, and chest roentgenogram in
6 + + +
any sinus and elimination of cough with 7 0 + + the one patient with left ventricular fail-
specifictherapy. Clinically,it wasnot pos- 8 + NO NO ure; chest roentgenograms and bronchos-
sible to distinguish those patients who 9 + + NO copy with transbronchoscopic lung bi-
had their postnasal drip syndrome from 10 + + NO opsy in the one patient with sarcoido-
11 + + NO
sinusitis from those who did not. Of the 12 + + NO
sis; barium swallow in the one patient
21 patients who had sinusitis, none com- 13 0 + NO with Zenker's diverticulum; history in the
plained of head, face, or teeth pain, and 14 0 + NO one patient categorized here with ACE
only five complained of discolored na- 15 + 0 + inhibitor-induced cough.
16 + 0 +
sal discharge. On the other hand, 7 of 17
The testing characteristics of various
+ 0 +
33 patients with a postnasal drip syn- 18 + 0 + laboratory components of our diagnos-
drome that was due to a cause other than 19 + 0 + tic protocol are shown in table 3. With
sinusitis also complained of discolored 20 0 0 + the exception of barium swallow, the oth-
21 0 0 + er tabulated tests have sensitivities and
nasal discharge.
22 0 0 +
In the asthma group, pulmonary func- 23 0 0 +
negative predictive values of looft/o. Al-
tion testing was obtained in all patients 24 0 0 + though esophageal pH monitoring and
and found to be helpful. Four of the 32 25 0 0 + fiberoptic bronchoscopy were the only
patients had reversible airflow obstruc- 26 0 0 + tests to have positive predictive values>
27 0 0 +
tion demonstrated by spirometry prein- 28 0 0
63070, they were performed in a small
+
halation and postinhalation of bron- number of highly selected patients. On
chodilator and 28 had a positive MIC Definition of abbreviations: GER = gastroesophageal reflux; the basis of a specific therapy failing to
+ = GER; 0 = no GER; NO = not done.
consistent with symptomatic asthma. eliminate cough as a complaint and sub-
Nine of the positive MIC werein the set- sequently determining the cause, the
ting of cough without wheeze, dyspnea, following tests were falsely positive in
or chest discomfort, a normal physical with cessation of smoking or the avoid- predicting the cause of cough: sinus roent-
examination of the lungs, and normal ance of an environmental irritant. genograms 16ft/o of the time in predict-
baseline spirometry. In the bronchiectasis group, four of ing that sinusitis was a cause of cough,
In the GER group, 57ft/o of the 28 pa- five patients complained of cough that chest roentgenograms 21ft/o of the time
tients complained of either heartburn or was productive of> 2 tablespoons of dis- in predicting that the abnormality was
a sour taste in their mouths, whereas 43070 colored phlegm in 24 h, whereas five of associated with the patient's cough, bar-
had no such symptoms. The results of five had chest roentgenographic changes ium swallow 13ft/o of the time in predict-
history, barium swallow, and prolonged typical of bronchiectasis. During bron- ing that GER was a cause of cough,
esophageal pH monitoring are summa- choscopy, the one patient who denied spirometry prebronchodilator and post-
rized in table 2. Barium swallowrevealed productive cough had excessive mucopu- bronchodilator 33ft/o of the time in pre-
GER in 11 of 25 or 44070 of the patients rulent secretions emanating from the air- dicting that asthma wasa cause of cough,
in whom it was obtained. Prolonged waysthat corresponded to the bronchiec- MIC 22ft/o of the time in predicting that
esophageal pH monitoring was abnor- tatic changes seen on the chest roentgen- asthma was a cause of cough, and fiber-
mal and consistent with GER as the cause ogram. While no patient in this group optic bronchoscopy 4070 of the time in
of cough in 20 of 20 (looft/o) of the pa-
tients in whom it was obtained. In ten
TABLE 3
patients without GER symptoms who
underwent barium swallow and pro- TESTING CHARACTERISTICS OF COUGH PROTOCOL
longed esophageal pH monitoring, pro- Tests n TP TN FP FN Sens Spec PPV NPV
longed esophageal pH monitoring was ·
Sinus XR 98 21 61 16 0 100 79 57 100
abnormal in all, whereas barium swal- CXR 100 11 68 21 0 100 76 36 100
low showed GER in only one. Thus, BAS 54 12 22 7 13 48 76 63 63
prolonged esophageal pH monitoring Spiro + BO 12 4 4 4 0 100 50 50 100
was the only way of diagnosing GER in MIC 86 28 39 19 0 100 67 60 100
Bronch 23 8 12 1 0 100 92 89 100
32ft/o of the patients in this group. pH probe 25 23 2 0 0 100 100 100 100
In the chronic bronchitis group, the di-
agnosis was suggested by history in all Definition of abbreviations: TP • true positive; TN • true negative; FP • false positive; FN - false
negative; sans • sensitivity; Spec = specificity; PPV - positive predictive value; NPV • negative
six patients and confirmed by having predictive value; CXR = chest radiograph; BAS = barium swallow; Spiro • spirometry; BO • bron-
cough disappear or markedly improve chodilator; MIC • methacholine inhalational challenge; Bronch • bronchoscopy.
CHRONIC COUGH
predicting that the abnormality found
was associated with the cause of the pa-
tient's cough.
Outcome of Specific Therapy
Specific therapy was successful in elim-
inating chronic cough as a complaint in
100 of 102 (98070) patients. On average,
patients were seen in clinic on 3.5 ± 2.2
occasions (range, 1 to 14) over 96 days
before a specific diagnosis was made and
definitive, specific therapy had been
prescribed.
When cough was due to GER alone,
it took more time to respond to specific
therapy than when it was due to other
single conditions. The average length of
time of 179 ± 205 days for cough to re-
spond to specific therapy in the GER
group was greater than the 70 ± 115 days
in the postnasal drip syndrome group (p
= 0.08), the 67 ± 69 days in the asthma
group (p = 0.07), the 107 ± 140 days
in the bronchiectasis group (p = 0.04),
the 39 ± 18 days in the chronic bronchi-
tis group (p = 0.02), and the 51 ± 83
days in the miscellaneous group (p =
0.07).
The length of time required to success-
fully treat patients increased with the
number of causes of cough. On average,
it took 92 days for one cause, 110 days
for two causes, and 159 days for three
causes.
The only patients who continued to
cough were one patient whose cough de-
fied diagnosis and one patient with uni-
lateral interstitial fibrosis secondary to
GER with documented aspiration whose
initial improvement in cough with inten-
sive medical treatment was not sustained
and who subsequently failed surgical
therapy for GER. Although we had en-
tertained the pretreatment diagnoses of
postnasal drip syndrome that was due to
chronic sinusitis, cough-variant asthma,
and GER as the possible cause(s) of
cough in the former patient, her cough
remained unchanged after prescribing
specific therapy for all three conditions.
Discussion
From our results, four conclusions about
chronic cough emerged that deserve
comment.
First, the concept upon which the ana-
tomic, diagnostic protocol was original-
ly based is still valid. By routinely con-
sidering extrapulmonary as well as pul-
monary conditions as potential causes of
cough, we were still able to, almost with-
out exception, determine the cause(s) and
prescribe specific therapy that was sue-
cessfuI. The diagnostic and therapeutic
results achieved in this study were simi-
lar to those reported in our first study
(1),even though the patients in this study
weremore complicated and diagnosis was
more difficult. Comparative data lend
credence to this statement: (1) although
18070 of patients in our first study had
two conditions simultaneously causing
their cough (1), 23070 of patients in this
study had two causes and 3070 had three
causes; (2) GER had increased in fre-
quency as a cause of cough, from 10070
(1)to 21070 and, although it was clinically
suspected in all patients in the previous
study, cough was the sole presenting clin-
ical manifestation in 43070 of patients
with GER (i.e., "silent" GER) in this
study; (3) although a positive MIC pre-
viously had been indicative of asthma as
the cause of cough, it was falsely posi-
tive 22OJo of the time in this study; (4)
although eventual successrates with ther-
apy were similar in both studies, immedi-
ate improvement in cough occurred less
often in this study. Whereas cough im-
proved immediately in all but one patient
in our 1981 study (1), patients reported
here had to be seen in clinic an average
of 3.5 times over an average of 96 days
before a specific diagnosis was made and
specific therapy was prescribed.
In addition to our two prospective
studies, we are aware of three other pub-
lished studies (23-25) involving series of
patients complaining of chronic cough.
Each used the same anatomic, diagnos-
tic protocol as we did and reported over-
all diagnosis and specific treatment
results similar to our two studies. A dis-
cussion of the results of all five of the
above studies has been previously pub-
lished (2). Whereas we have demonstrat-
ed the utility of the anatomic, diagnostic
protocol in the adult tertiary care setting,
Holinger (23) and Poe and coworkers (24,
25) have confirmed its value in the pedi-
atric tertiary care and community set-
tings, respectively.
Second, the strengths and limitations
of our diagnostic protocol were more
clearly revealed in this study than in our
previous one (1). With the exception of
barium swallow, the laboratory testing
protocol was consistently helpful in rul-
ing out suspected possibilities, whereas
history, physical examination, or the re-
sult of any laboratory test were not al-
ways predictive of the cause of cough
without assessing the response to specif-
ic therapy. When the relative usefulness
of a true positive result of each part of
the diagnostic protocol in determining
645
the causes of cough was considered, his-
tory, physical examination, MIC, and
prolonged esophageal pH monitoring
were the most frequently helpful com-
ponents. Although 100070 of the patients
whose cough was due to a postnasal drip
syndrome had symptoms of postnasal
drip, throat clearing, or nasal discharge
and/or signs of mucus or a cobblestone
appearance in the oropharynx, these
complaints and findings werenot specific
for this disorder since they werealso com-
monly seen in other conditions. Pulmo-
nary function studies, primarily MIC,
and upper gastrointestinal studies, pri-
marily prolonged esophageal pH moni-
toring, were the most useful laboratory
studies ordered, whereaschest roentgeno-
grams and fiberoptic bronchoscopy were
much less frequently helpful. Fiberoptic
bronchoscopy was only helpful in the set-
ting of an abnormal chest roentgeno-
gram. Although a negativeMIC ruled out
asthma as a cause of cough, a positive
MIC was consistent with, but not neces-
sarily diagnostic of, asthma as the cause.
In this study, a positive MIC was the on-
ly means of identifying nine of 32 (28070)
of our asthmatics; on the other hand, a
positive MIC was falsely positive 22070
of the time. A prolonged esophageal pH
monitoring session positive for GER was
the only means of identifying 9 of 28
(32070) of our patients whose cough was
due to GER; it was not falsely positive
in any patient in this study.
Third, chronic cough was frequently
due to common disorders, and it was not
unusual for it to have more than one
cause. Chronic cough wasmost common-
ly caused by postnasal drip syndrome,
followed by asthma, GER, chronic bron-
chitis, bronchiectasis, and a few miscel-
laneous conditions. Postnasal drip syn-
drome, asthma, and/or GER were the
cause(s) of chronic cough 86070 of the
time in this study, and they were consis-
tently suggested by history, physical,
MIC, and prolonged esophageal pH
monitoring; moreover, these three con-
ditions werethe cause(s) of cough in 99010
of those patients who were not smoking
and not exposed to environmental irri-
tants and who had normal chest roent-
genograms. Although these results were
similar to those in our previous report
(1),there were differences to be noted be-
tween the studies in the spectrum and fre-
quency of conditions causing chronic
cough. GER had increased in frequency
from 10 to 21010, and in 43070 of the pa-
tients with GER in this study, cough was
the sole presenting clinical manifestation.
646
Although wewereaware that cough could
be the sole manifestation of GER (3) and
had incorporated prolonged esophageal
pH monitoring into our diagnostic pro-
tocol because of this, we had not previ-
ously known the prevalence of this GER
variant group. In addition, bronchiecta-
sis and ACE inhibitor-induced cough,
not diagnosed in the prior study, caused
cough 4 and 2010 of the time, respective-
ly. Because of the ever-increasing num-
ber of patients who are being treated with
ACE inhibitors and the modest frequen-
cy with which these drugs induce cough,
ACE inhibitor-induced cough will no
doubt occur with a greater frequency in
future studies on chronic cough. In a re-
cently reported prospective survey (26),
the prevalencesof cough in patients treat-
ed with captopril, enalapril, and cila-
zapril were 11.5,24.7, and 13.3070, respec-
tively. Even though weprospectivelystud-
ied a group of consecutive and unselected
patients, we would caution those who
would anticipate the identical spectrum
and frequency in all patients with cough.
Our data were generated in a group of
patients seen in a tertiary, pulmonary
clinic that specializes in evaluating pa-
tients with chronic cough of unknown
etiology. None of these patients were
immunocompromised.
Fourth, specific therapy was almost al-
ways successful. Definitive, successful
treatment of cough depends upon first
determining its precise cause(s) and then
initiating specific therapy for the etiolo-
gy or operant pathophysiologic mecha-
nism of the underlying disorder(s). Be-
cause an accurate diagnosis was made in
all but one of our patients, we were able
to evaluate the outcome of our specific
therapy. Specific therapy was successful
in 100of 102(98010) of all of the patients
evaluated in this study and in 100 of 101
(99070) of the patients in whom we were
able to make an accurate diagnosis. In
addition to eliminating cough as a com-
plaint, specifictherapy servesanother im-
portant function in the evaluation of pa-
tients with chronic cough; it confirms the
correctness of the presumptive diagnoses.
If the putative specific therapy does not
make the cough go away, the presump-
tive diagnosis is either wrong or therapy
is submaximal. In the one patient whose
cough defied diagnosis, we were uncer-
tain which of the above possibilities ap-
plied since the patient never strictly ad-
hered to our treatment regimens. Because
cough was successfully treated without
therapy that would be considered non-
specific (20) (e.g.,centrally acting antitus-
IRWIN, CURLEY, AND FRENCH
sives, agents that may alter mucociliary persistent cough in the adult: the spectrum and fre-
factors irritating the cough receptor), any quency of causes and successful outcome of spe-
cific therapy. Am Rev Respir Dis 1981; 123:413-7.
more than a limited role for nonspecific 2. Irwin RS, Curley FJ. Is the anatomic, diagnos-
therapy must be seriously questioned. Al- tic work-up of chronic cough not all that it is hacked
though the above treatment results were . up to be? Chest 1989; 95:711-3.
similar to those in our previous study (1), 3. Irwin RS, Zawacki JK, Curley FJ, French CL,
Hoffman PJ. Chronic cough as the sole presenting
it took a longer period of time in this manifestation of gastroesophageal reflux. Am Rev
study before cough began to improve. We Respir Dis 1989; 140:1294-300.
believe that this was due to (1) the great- 4. Morris JF, Koski A, Johnson Le. Spirometric
er time it took to identify the two and standards for healthy nonsmoking adults. Am Rev
three separate simultaneously contribut- Respir Dis 1971; 103:57-67.
5. Dickman ML, Schmidt CD, Gardner RM.
ing conditions that occurred more fre- Spirometric standards for normal children and
quently in this study, and (2) the greater adolescents (ages 5 years through 18 years). Am
frequency of GER as the cause of cough Rev Respir Dis 1971; 104:680-7.
in this study. Our present data revealed 6. Morris JF, TempleWP, Koski A. Normal values
that, of all the diagnostic groups, cough for the ratio of one-second forced expiratory vol-
ume to forced vital capacity. Am Rev Respir Dis
in the GER group took the longest time 1973; 108:1000-3.
to resolve.Although our data do not pro- 7. Corrao WM, Braman SS, Irwin RS. Chronic
vide us with any reasons for this slow cough as the sole presenting manifestation of bron-
resolution in cough caused by GER, we chial asthma. N Engl J Med 1979; 300:633-7.
8. Irwin RS, Pratter MR, Holland rs, Corwin RW,
speculate that a possible explanation is Hughes JP. Postnasal drip causes cough and is as-
that because therapy for GER decreases sociated with reversible upper airway obstruction.
but does not totally eliminate the reflux Chest 1984; 85:346-52.
of acid, the slow and gradual resolution 9. Sonnenberg A, Lepsien G, Muller-Lissner SA,
Koelz HR, Siewert JR, Blum AL. When is esopha-
of cough might be due to a slowly heal- gitis healed? Dig Dis Sci 1982; 27:297-302.
ing mucosal injury in the esophagus or 10. Dodds WJ. Instrumentation and methods for
respiratory tract that was still being ex- intraluminal esophageal manometry. Arch Intern
posed to acid, even though it was quali- . Med 1976; 136:515-23.
tatively and quantitatively less. 11. Dalton CB. Manometric study, measurements
and interpretations. In: Castell DO, Richter JE, Dal-
In conclusion, we believe that the ana- ton CD, eds. Esophageal motility testing. New York:
tomic, diagnostic protocol as originally Elsevier Science Publishing Co., 1987; 35-78.
conceived is still valid. The protocol's 12. Richter JE. Normal values for esophageal mo-
strength was and continues to be that it tility testing. In: Castell DO, Richter JE, Dalton
encourages all of us to consider extrapul- CB, eds. Esophageal motility testing. New York:
Elsevier Science Publishing Co., 1987; 79-89.
monary as wellas pulmonary conditions 13. Spencer J. Prolonged pH recording in the study
as potential causes of chronic cough. It of gastroesophageal reflux. Br J Surg 1969; 56:
has led us to the realization that extrapul- 912-4.
monary disorders such as postnasal drip 14. Ward BW, Wu we, Richter JE, Lui KW,
Castell DO. Ambulatory 24-hour esophageal pH
syndrome and GER are frequently the monitoring: technology searching for a clinical ap-
cause of chronic cough. Although the plication. J Clin Gastroenterol 1986; 8(Suppl
concept of the anatomic protocol is val- 1:59-67).
id, the protocol will periodically need to 15. Wu WC. Gastroesophageal reflux and pH test-
ing. In: Castell DO, Richter JE, Dalton CB, eds.
be updated to incorporate new or chang- Esophageal motility testing. New York: Elsevier
ing information. For instance, we would Science Publishing Co., 1987; 198-208.
not have been able to reduplicate our ini- 16. Medical Research Council. Committee report
tial successes had we not incorporated on the aetiology of chronic bronchitis. Definition
prolonged esophageal pH monitoring in- and classification of chronic bronchitis for clini-
cal and epidemiologic purposes. Lancet 1965;
to our diagnostic protocol to evaluate the 1:775-8.
possibility that chronic cough might be 17. Fraser RG, Pare JAP. Diseases of the airways.
the sole manifestation of GER. Other In: Fraser RG, Pare JAP, eds. Diagnosis of dis-
modifications made to our original pro- eases of the chest. Vol III, 2nd ed. Toronto: W.B.
tocol include the following: (1) intranasal Saunders Co., 1979; 1443-55.
18. Seebohm PM. Allergic and non-allergic rhi-
corticosteroids rather than antihistamine nitis. In: Middleton E Jr, Reed CE, Ellis EF, eds.
decongestants were routinely prescribed Allergy: principles and practise. St. Louis: CY. Mos-
for postnasal drip syndromes not caused by Co., 1978; 868-76.
by sinusitis and (2) oral rather than in- 19. Evans FO, Sydnor JD, Moore WEC, et al. Si-
nusitis of the maxillary antrum. N Eng! J Med 1975;
halational beta-agonists were prescribed 293:735-9.
for the asthmatic component of cough 20. Irwin RS, Curley FJ, Pratter MR. The effects
when the latter was observed to provoke of drugs on cough. Eur J Respir Dis 1987;71(Suppl
cough in clinic. 153:173-81).
21. Griner PF, Mayewski RJ, Mushlin AI, Green-
land P. Selection and interpretation of diagnostic
References tests and procedures: principles and applications.
1. Irwin RS, Corrao WM, Pratter MR. Chronic Ann Intern Med 1981; 94(Part 2:553-600).
CHRONIC COUGH
22. Curley FJ, Irwin RS, Pratter MR, et al. Cough
and the common cold. Am Rev Respir Dis 1988;
138:305-11.
23. Holinger LD. Chronic cough in infants and
children. Laryngoscope 1986; 96:316-22.
24. Poe RH, Israel RH, Utell MJ, Hall WJ. Chron-
ic cough: bronchoscopy or pulmonary function test-
ing? Am Rev Respir Dis 1982; 126:160-2.
25. Poe RH, Harder RV, Israel RH, Kallay MC.
Chronic persistent cough: experience in diagnosis
647
and outcome using an anatomic diagnostic pro-
tocol. Chest 1989; 95:723-8.
26. Goldszer RC, Lilly LS, Solomon HS. Preva-
lence of cough during angiotensin-converting en-
zyme inhibitor therapy. Am J Moo 1988; 85:887.