Failure of Physicians to Consider the Diagnosis of Pertussis in Children

Transcription

1 840 Failure of Physicians to Consider the Diagnosis of Pertussis in Children Shelley Deeks, Gaston De Serres, Nicole Boulianne, Bernard Duval, Louis Rochette, Pierre Déry, and Scott Halperin From the Field Epidemiology Training Program, Laboratory Centre for Disease Control, Ottawa, Ontario, Centre de recherche du Pavillon CHUL, Centre hospitalier universitaire de Québec, and the Faculty of Medicine, Laval University, Quebec, Quebec, and the Department of Pediatrics, Dalhousie University, and the IWK Grace Health Centre, Halifax, Nova Scotia, Canada To determine the ability of physicians to make a diagnosis of pertussis and factors associated with improved diagnosis, 8,235 children from 88 child care centers and 14 elementary schools from Quebec City, Quebec, Canada, were evaluated by using a questionnaire completed by parents and a medical record review. Children must have consulted a physician to be included in the evaluation. There were 558 children meeting the surveillance case definition and 416 meeting a modified World Health Organization case definition who consulted a physician. A diagnosis of pertussis was considered in 24% 26% of children meeting either case definition, made in 12% 14%, and reported for 6%. Pertussis diagnosis was significantly associated with having a history of pertussis exposure (P.003), four pertussis-related symptoms (P õ.001), and a cough for 5 weeks (P.05) and consulting in a hospital setting (P.03). The proportion of cases of pertussis diagnosed and reported is low even when children present with classical symptoms. Bordetella pertussis continues to be an important cause of studies have also demonstrated physician difficulty with diagnosis, morbidity in North American children [1 3]. The actual burden especially in adolescents and adults [3, 5, 6, 11]. of illness from pertussis is greater than reflected in national A study was undertaken in Quebec City, Quebec, Canada, statistics, because of underreporting associated with a passive between May and June The objectives were to determine reporting system, the relative insensitivity of laboratory tests, the following: the ability of physicians to make a diagnosis of and difficulties in physician diagnosis [2 7]. In the United pertussis, the factors associated with improved diagnosis, and States, it has been estimated that only 3% to 12% of cases of the proportion of cases reported to the provincial surveillance pertussis in young children are reported to the Centers for system. This study was conducted during a period of increased Disease Control and Prevention [2, 8]. Nasopharyngeal (NP) incidence of disease. Between 1991 and 1993, the incidence culture, the current diagnostic gold standard for pertussis, is of pertussis in the province of Quebec increased more than highly specific; however, it is relatively insensitive and labor- sixfold, reaching 59.4 reported cases per 100,000 population intensive [8 10]. Delayed specimen collection, prior immunization, in 1993 [1, 14]. This increase in incidence was associated and prior use of erythromycin therapy have been associ- with a general increase in public and professional awareness ated with decreased sensitivity of culture for B. pertussis [8, of pertussis. 9, 11]. The lack of a highly sensitive and specific diagnostic test reinforces the importance of adequate history taking on the part of physicians, as the diagnosis must often be made on Methods clinical grounds. Two retrospective cohort studies were conducted simultaneously Relatively little literature is available pertaining to physician in Quebec City between 31 May and 20 June The ability to make the diagnosis of pertussis. A distinguishing study methodology has been described in detail previously [15]. feature of the disease is the paroxysmal cough and characteristic Briefly, the first cohort study included children attending 88 whoop. Because the child may not cough in the physician s of the 91 child care centers in Quebec City, and the second presence, the diagnosis could easily be missed without a high included pupils attending kindergarten to grade 3 in 14 elemen- index of suspicion [12]. In a study of pediatric patients with tary schools in a suburb of Quebec City. Because the objectives pertussis, the disease was not considered in the initial differential of the current study were independent of the setting attended diagnosis in more than one-half of the cases [13]. Other by the children, data from the two cohort studies were merged and analyzed together. Parents completed a standardized questionnaire regarding all cough-associated illness of at least 2 weeks duration and pertussis-related symptoms (paroxysmal Received 20 May 1998; revised 6 November Reprints or correspondence: Dr. Gaston De Serres, Centre de santé cough, posttussive vomiting or apnea, and whoop) in their child publique de Québec, 2400 d Estimauville, Beauport, Québec, Canada G1E during the 11-month period between 1 July 1992 and 31 May 7G9 (gdeserres@cspq.qc.ca) Clinical Infectious Diseases 1999;28:840 6 Children were classified as meeting either of two clinical 1999 by the Infectious Diseases Society of America. All rights reserved /99/ $03.00 case definitions for pertussis. Because neither case definition

2 CID 1999;28 (April) Failure to Consider Pertussis 841 is 100% sensitive or specific, they provide estimates of the the 8,456 enrolled children participating. Ninety-five percent actual number of pertussis cases. The first, referred to as the of children attending child care centers had received at least surveillance case definition (SCD), is the clinical case definition three doses of pertussis vaccine, and 98% of children attending used for surveillance in Canada: a cough lasting 2 weeks elementary schools had received at least four doses of vaccine. with at least one pertussis-related symptom (paroxysmal cough, Of the 1,316 children who coughed for ú2 weeks and had at whoop, or posttussive vomiting or apnea) for which no other least one pertussis-related symptom, 778 consulted a physician cause is identified [16]. The second, referred to as the modified because of their illness. Of these children, 558 (72%) met the World Health Organization (WHO) case definition, is defined SCD, while the remaining 28% had another cause for their as a cough lasting 3 weeks with paroxysms. It is a modifica- cough when their medical records were reviewed (table 1). If tion of the WHO case definition used in the acellular pertussis we apply this proportion to the 534 children who did not consult vaccine clinical trials that requires a paroxysmal cough of at a physician, it may be grossly estimated that 151 children had least 3 weeks duration and a culture positive for B. pertussis, another cause for their cough and that 383 children met the SCD serological evidence of bordetella-specific infection as demon- for pertussis. Therefore, 59% (558) of 941 children meeting the strated by a significant rise in antibody titer, or a household SCD consulted a physician. contact with a bacteriologically confirmed case of B. pertussis Of the 648 children meeting the modified WHO case definiinfection occurring within 28 days before or after the onset of tion, 64% consulted a physician. The only difference between illness in the study subject [17]. children who consulted a physician and those who did not was To be eligible for inclusion in the study, children must have that children who consulted a physician had a greater mean met either or both case definitions for pertussis and consulted a number of symptoms (P õ.05). The remaining analysis is physician for their cough. Parental consent was obtained to ex- limited to the 558 children meeting the SCD and the 416 meetamine the children s medical records. The medical records were ing the modified WHO case definition who consulted a physievaluated independently by two physicians, and data were ab- cian. Three hundred thirteen children consulting physicians met stracted with use of a standardized form. The following children both case definitions. were excluded from the study: children for whom parental con- The mean interval between the onset of the child s cough sent was not obtained, children whose medical records were not and completion of the parental questionnaire was 3.5 months. located, and children with a history of pertussis before 1 July The following pertussis-related symptoms were reported for If a diagnosis of pertussis was not made by a physician, children meeting the SCD for pertussis: paroxysmal cough, the medical records were reviewed for the duration of the child s 79%; posttussive vomiting, 49%; posttussive apnea, 41%; and cough episode, to a maximum of 10 consultations. If pertussis whoop, 30%. Among children meeting the modified WHO case was diagnosed, the records were not reviewed beyond the point definition, the proportions were similar except paroxysmal of diagnosis. Information on symptomatology was obtained from cough was present in 100% of children, as this was a requirethe parental questionnaire, and information on other potential ment to meet this case definition. causes of cough was obtained from the medical record review. According to the children s medical records, a diagnosis of Physicians were classified as general practitioners (GPs), pertussis was considered in 24% of children meeting the SCD including emergency department physicians, or pediatricians. and made in 52% of those in whom it was considered (table Physicians having other qualifications and those whose quali- 2). Similarly, the diagnosis was considered in 26% of children fications were unknown were not included in the analysis permeeting the modified WHO case definition and made in 53% taining to physician types. Consultations were classified as of those in whom it was considered. Less than one half of the occurring in an office or hospital setting. Hospital visits inchildren (46% 47%) who were diagnosed with pertussis were cluded visits to the emergency department, visits to hospital reported to the provincial surveillance system. An NP swab clinics, or, rarely, hospitalization. To determine whether the was obtained from between 59% and 67% of children in whom cases were identified by the provincial surveillance system, identification variables for the study subjects were linked to the diagnosis was considered; cultures of 15% of these swabs those of cases reported to the Centre de santé publique de yielded positive results. All children for whom a culture was Québec, Beauport, Quebec. positive were diagnosed with pertussis; however, of children The data were analyzed by using the SAS system, version for whom a culture was negative, 36% meeting the SCD and 6.12 (SAS Institute, Cary, NC). Proportions and means were 42% meeting the modified WHO case definition were diagcalculated for descriptive statistics, and t tests, x 2 test, and nosed. All cases with a positive culture were reported to the logistic regression were used to calculate inferential statistics. provincial surveillance system. As the number of pertussis- Odds ratios with 95% confidence intervals were reported, and related symptoms increased, so did the proportion of children a P value of.05 was considered statistically significant. from whom an NP swab was obtained and for whom a diagnosis of pertussis was considered, made, and reported to the provin- Results cial surveillance system (table 2). Eighty-eight of 91 child care centers and all 14 elementary Children consulting exclusively GPs, pediatricians, or both schools agreed to be part of the study, with 8,235 (97%) of types of practitioners were compared (table 3). Children con-

3 842 Deeks et al. CID 1999;28 (April) Table 1. Characteristics of children meeting the SCD and the modified WHO case definition for pertussis according to their physician consultation status. Patients meeting SCD Patients meeting modified WHO case definition Consulted Not consulted Consulted Not consulted Characteristic (n Å 558) (n Å 534)* (n Å 416) (n Å 258) Mean age (y) Male-to-female ratio Mean no. of symptoms Mean duration of cough (w) No. (%) with type of child care Day care 388 (53) 348 (47) 286 (63) 166 (37) School 170 (48) 186 (52) 130 (63) 92 (37) NOTE. SCD Å surveillance case definition; WHO Å World Health Organization. * This number does not exclude subjects with other causes for their cough. P õ.05. sulting physicians with unknown or other qualifications were both types of practitioners. When these children were included excluded from the remaining analysis. The mean age of children in the analysis, pediatricians were significantly more likely to consulting pediatricians was younger than those con- diagnose pertussis than were GPs (P õ.05). sulting GPs (P õ.001). Pediatricians were more likely to Information about children s exposure to a known case of per- consider a diagnosis of pertussis than were GPs (P õ.04), and tussis was collected from parents. For the 133 children meeting they were also more likely than GPs to obtain an NP swab the SCD whose parents were aware that their child had been (P õ.001). When considering children who consulted exclu- exposed, 27% (24) of 90 GPs and 39% (12) of 31 pediatricians sively a GP or a pediatrician, the proportion of children diagnosed recorded this exposure in the child s medical records. For the 103 by pediatricians was greater than the proportion diag- children meeting the modified WHO case definition whose parents nosed by a GP, yet this difference did not reach statistical reported an exposure, 21% (15) of 70 GPs and 45% (nine) of 20 significance (patients who met the SCD, P Å.08; those who pediatricians noted this exposure. Pediatricians were more likely met the modified WHO case definition, P Å.06). However, to record this information in their medical records than were GPs; pediatricians made most of the diagnoses in children seeing however, the difference was not statistically significant. Table 2. Status of pertussis diagnosis among children consulting physicians according to the number of pertussis-related symptoms. No. of symptoms Variable One Two Three Four Total No. of patients meeting SCD for pertussis NP swab obtained 7 (11*) 12 (0) 18 (13) 37 (28) 14 (15) Diagnosis considered Diagnosis made Diagnosis reported No. of patients meeting modified WHO case definition for pertussis NP swab obtained 7 (10) 14 (0) 15 (8) 43 (26) 17 (15) Diagnosis considered Diagnosis made Diagnosis reported NOTE. Unless stated otherwise, data are percent of patients. NP Å nasopharyngeal; SCD Å surveillance case definition; WHO Å World Health Organization. * Percent of positive swabs.

4 CID 1999;28 (April) Failure to Consider Pertussis 843 Table 3. Characteristics of children consulting a general practitioner or pediatrician in a study of making the diagnosis of pertussis. SCD* Modified WHO case definition* GP Pediatrician Both GP Pediatrician Both Characteristic (n Å 380) (n Å 120) (n Å 44) (n Å 262) (n Å 90) (n Å 51) Mean age (y) Mean no. of symptoms Mean duration of cough (w) Mean no. of visits Percent of patients with Office visits NP swab obtained Diagnosis* considered Diagnosis* made Diagnosis* reported NOTE. GP Å general practitioner; NP Å nasopharyngeal; SCD Å surveillance case definition; WHO Å World Health Organization. * Fourteen children (3%) meeting the SCD for pertussis and 13 (3%) meeting the modified WHO case definition for pertussis consulted physicians with other or unknown qualifications and were excluded. P õ.05 when comparing children who consulted only GPs or pediatricians. Pertussis diagnosis varied according to a number of factors (table 4). The proportion of children in whom a diagnosis of pertussis was made was greater when children were older than 6 years of age, when parents were aware of a history of pertussis Table 4. Characteristics of children with a diagnosis of pertussis. exposure, when children had a greater number of pertussistotal No. of children with pertussis diagnosis/ related symptoms, or when children had a longer duration of no. (%) according to case cough. It was also greater when children consulted a pediatrician definition for pertussis or consulted in a hospital setting. However, multivariate Modified WHO logistic regression revealed that the variables significantly asso- Characteristic SCD case definition ciated with pertussis diagnosis in children meeting either case definition were having four pertussis-related symptoms, a his- Age Younger than 6 y 43/405 (11) 35/301 (12) tory of pertussis exposure, and a cough for 5 weeks and 6 10 y 23/139 (17) 19/102 (19) consulting in a hospital setting (table 5). Pertussis contact* No 27/411 (7) 25/300 (8) Discussion Yes 39/133 (29) 29/103 (28) No. of symptoms 1 8/247 (3) 5/137 (4) The current study was conducted during a resurgence of 2 15/136 (11) 10/104 (10) pertussis in the province of Quebec [1, 14]. Only 12% 14% 3 10/87 (11) 6/88 (7) of children meeting either clinical case definition were diag- 4 33/74 (45) 33/74 (45) nosed by physicians. Diagnosis was infrequent despite three Cough duration written communications sent by the public health unit to all õ5 w 26/368 (7) 11/210 (5) 5 w 40/176 (23) 43/193 (22) GPs and pediatricians in the study area about the increase in Type of practitioner the incidence of pertussis. The low incidence during the 15 GP 40/394 (10) 31/281 (11) years preceding the resurgence suggests that many practicing Pediatrician 26/124 (21) 23/122 (19) physicians have had little experience with pertussis and may Location of practitioner not have considered the diagnosis or may have felt uncertain Office 43/440 (10) 33/304 (11) Hospital 20/100 (20) 18/95 (19) in making the diagnosis [14]. Furthermore, as the vaccine cov- No. of visits erage among the children was very high, physicians may have 1 24/254 (9) 18/153 (12) overestimated the efficacy of pertussis vaccine and erroneously 2 20/134 (15) 14/90 (16) thought that pertussis was not possible in a vaccinated child with pertussis-related symptoms [13, 15]. 3 22/156 (14) 22/160 (14) Information for this study was obtained from questionnaires based on parental recall and medical records. Because of its NOTE. GP Å general practitioner; SCD Å surveillance case definition; WHO Å World Health Organization. * Children whose parents indicated that they had been exposed to pertussis.

5 844 Deeks et al. CID 1999;28 (April) Table 5. Factors associated with pertussis diagnosis in a multivariate logistic regression analysis of data for children. SCD Modified WHO case definition Factor OR (95% CI) P value OR (95% CI) P value Younger than 6yvs.6y or older 1.4 ( ) ( ).12 Pertussis contact* 4.5 ( ) õ ( ).003 No. of symptoms 1 (referent) ( ) ( ) ( ) ( ) ( ) õ ( ) õ.001 Cough for 5 w vs. cough for õ5 w 2.0 ( ) ( ).004 Pediatrician vs. GP 1.8 ( ) ( ).07 Hospital vs. office visit 2.2 ( ) ( ).03 No. of visits 1 (referent) ( ).66 ( ) ( ) ( ).46 NOTE. GP Å general practitioner; SCD Å surveillance case definition; WHO Å World Health Organization. * Children whose parents indicated that they had been exposed to pertussis. retrospective nature, recall may not be totally accurate. How- WHO case definition includes a longer duration of cough, it ever, for pertussis, information based on parental recall has should be more specific than the SCD, but including persons been shown to be reliable as pertussis-related symptoms are with other causes of cough would decrease the specificity of the characteristic and leave a strong impression on parents [18, modified WHO case definition. Thus, the two case definitions 19]. Comparing parental recall of symptoms after conclusion provide two perspectives to estimate the number of cases of of the cough episode to medical records of children consulting pertussis in the study population. It is interesting that the results a physician during their illness may have underestimated clini- of the study were quite similar regardless of the case definition cians ability to make the diagnosis of pertussis, because children used. may have had a different number or severity of symptoms, During an outbreak, as was occurring during the study pe- duration of cough, or knowledge about pertussis exposure during riod, the positive predictive value of a clinical case definition the period of consultation from that that was reported by increases and is likely to exceed 70% [10]. Therefore, we would parents after the conclusion of the child s illness. By the time expect that ú70% of children classified as having pertussis on that the parents completed the questionnaires, the child s illness the basis of clinical case definition would have actually had was over, and the duration, severity, and exposure history were pertussis. It should be noted, however, that some children meet- fully known. Incomplete medical records may also have af- ing either case definition may not have had pertussis, as neither fected the study results if physicians had considered a diagnosis case definition is 100% specific. Therefore, the study may have of pertussis but not recorded it in their records or if they made underestimated the proportion of physicians who made the an alternative diagnosis without recording it. Nonetheless, the diagnosis. magnitude of the underdiagnosis of pertussis demonstrated sug- In this study, having all four pertussis-related symptoms gests that these potential biases would not alter the study s (paroxysmal cough, whoop, and posttussive vomiting and ap- conclusions. nea) was the most important predictor of diagnosis. Nonetheless, There is not a single standard case definition for pertussis a diagnosis of pertussis was not made in more than one- used throughout the world, and no case definition is 100% half of these children. This finding suggests an unsatisfactory sensitive or specific. The SCD excludes persons with an apparent clinical performance by physicians, which was even more ap- etiology for their cough. Many of these causes, such as parent as the number of symptoms decreased. Diagnosis was sinusitis, bronchitis, pneumonia, and bronchospasm, may in significantly improved when the child s cough had been present fact be complications of pertussis. These exclusions would for 5 weeks, thereby suggesting that clinicians are less inclined decrease the sensitivity of the SCD. The modified WHO case to think of pertussis with a cough of shorter duration. definition does not make these exclusions, but its sensitivity The presence of a known exposure to pertussis was also an and specificity have not been evaluated. Because the modified important factor in diagnosis. In 24% 26% of cases, parents

6 CID 1999;28 (April) Failure to Consider Pertussis 845 may unnecessarily delay diagnosis and treatment. Clinical diagnosis is sufficient for most children presenting with typical pertussis-related symptoms. A missed diagnosis of pertussis can have three main conse- quences. First, if an alternative diagnosis is made, appropriate therapy with erythromycin, the antibiotic recommended for treatment of pertussis, will not likely be initiated [16]. Second, chemoprophylaxis will not be given to family contacts in spite of the risk of secondary transmission [20]. Treatment of persons with pertussis and prophylaxis for family contacts may be ef- fective in decreasing disease severity, duration, and transmission if erythromycin therapy is started early [6, 16, 20, 21]. Third, parents may lose confidence in a physician s clinical ability if they consulted the physician with a child having typical pertussis-related symptoms and the diagnosis was not made. This latter consequence was suggested by many written comments received from parents during data collection that revealed that a number of parents were extremely upset that pertussis was not considered by the numerous physicians who they saw, despite the increased awareness of pertussis in the community. Passive surveillance systems detect only a small proportion of all cases of pertussis [2 4, 11, 18]. In this study, cases of 46% 47% of children with a diagnosis of pertussis were re- ported to the surveillance system, representing only 6% of all children who consulted a physician and met a clinical case definition for pertussis. Because this 6% does not include patients who did not consult a physician, it can be assumed that, in this age group, there were at least 20 times more cases than the number reported to the provincial surveillance system. This estimate is conservative. In adolescents and adults, the diagnosis is even more frequently missed than in children; therefore, the level of underreporting for the entire population would be greater than reported in this study [5, 6, 13]. Because pertussis continues to be an important cause of morbidity in North America, it is necessary for clinicians to actively search for pertussis-related symptoms in all patients presenting with an afebrile cough and to consider obtaining specimens for cultures for B. pertussis. Pertussis in its classical form is so characteristic that physicians should be able to rely on patient symptomatology and their clinical ability for diagnosis, especially during an epidemic period. Unfortunately, this study reveals that there needs to be tremendous improvement in our ability to make the diagnosis of pertussis. were aware that children had had such exposure; however, in more than one-half of these instances, it was not recorded by physicians. When this information was written in the charts, it was not possible to know if the physicians actively elicited it or if the parents volunteered it. Likewise, when the information was not written in the chart, it was not possible to know whether the information was elicited by physicians and not written in the chart, as medical records are not always a complete account of the patient consultation. However, even if the information was elicited and not recorded, it was not used to establish the diagnosis of pertussis. Improved diagnosis in a hospital compared with an office setting might be explained by easier access to diagnostic procedures, such as obtaining NP swabs. In Quebec City, most physicians offices do not have the laboratory services that permit NP swabs to be obtained properly. These services are available in the hospital setting. In addition, children presenting to the hospital may have had more severe symptoms facilitating the diagnosis. Pertussis diagnosis was not significantly affected by the child s age, the number of consultations, or the type of physician consulted. Despite the fact that pertussis is more severe in younger children, there was no improvement in diagnosis in this age group. Pediatricians were more likely to make a diagnosis of pertussis than were GPs (21% 19% vs. 10% 11%, respectively); however, this difference did not reach statistical significance. Pediatricians were also more likely than GPs to obtain information about pertussis exposure, to obtain NP swabs, and to consider a diagnosis of pertussis, all of which are important in making the diagnosis. However, the fact that the diagnosis was made in less than one-fifth of children consulting pediatricians and one-tenth of those consulting GPs among a group of children meeting clinical case definitions is troublesome for all clinicians. Obtaining an NP swab was a predictive factor in diagnosis; however, swabs were obtained only from 14% 17% of children. These results suggest that culture for B. pertussis is underutilized as a diagnostic tool. Although children for whom a culture was positive were diagnosed with pertussis, there was an interpretive problem with negative results. Only 36% 42% of the children meeting the clinical case definitions who had a negative swab were diagnosed with pertussis, thus implying that, for many, a negative result was interpreted as confirmation that pertussis was not present. In fact, culture for B. pertussis is known to have variable sensitivity [8]. Isolation rates are highest when specimens are taken during the catarrhal and early paroxysmal stage, but these rates decrease thereafter [8]. Thus, negative cultures do not necessarily exclude pertussis as the etiology, especially in vaccinated children or when the cough has been present for a long time. Nevertheless, more extensive use of NP swabs is desirable because positive results are informative and confirm the diagnosis of pertussis in cases and may help to confirm the diagnosis in symptomatic contacts. However, waiting for culture results References 1. Health Canada Notifiable diseases annual summary. Can Commun Dis Rep 1996;22(S2): Sutter RW, Cochi SL. Pertussis hospitalizations and mortality in the United States, , evaluation of the completeness of national reporting. JAMA 1992;267: Black S. Epidemiology of pertussis. Pediatr Infect Dis J 1997;16:S Halperin SA, Bartolussi R, MacLean D, Chisholm C. Persistence of pertus- sis in an immunized population: results of the Nova Scotia Enhanced Pertussis Surveillance Program. J Pediatr 1989;115:

7 846 Deeks et al. CID 1999;28 (April) 5. Deville JG, Cherry JD, Christenson PD, et al. Frequency of unrecognized 14. De Serres G. Pertussis in Quebec: ongoing epidemic since the late 1980s. Bordetella pertussis infections in adults. Clin Infect Dis 1995;21: Can Commun Dis Rep 1995;21: Herwaldt LA. Pertussis in adults, what physicians need to know. Arch 15. De Serres G, Boulianne N, Duval B, et al. Effectiveness of a whole cell Intern Med 1991;151: pertussis vaccine in child care centers and schools. Pediatr Infect Dis J 7. Mark A, Granström M. Cumulative incidence of pertussis in an unvacci- 1996;15: nated preschool cohort based on notifications, interview and serology. 16. National Advisory Committee of Immunization. Statement on management Eur J Epidemiol 1991;7: of persons exposed to pertussis and pertussis outbreak control. Can 8. Onorato IM, Wassilak GF. Laboratory diagnosis of pertussis: the state of Commun Dis Rep 1994;20: the art. Pediatr Infect Dis J 1987;6: WHO meeting on case definition of pertussis: Geneva, January, 9. Strebel PM, Cochi SL, Farizo KM, et al. Pertussis in Missouri: evaluation Geneva: World Health Organization, 1991:4 5 (issue no MIN/ of nasopharyngeal culture, direct fluorescent antibody testing, and clini- EPI/PERT/91.1). cal case definitions in the diagnosis of pertussis. Clin Infect Dis 1993; 18. Isacson J, Trollfors B, Taranger J, Zackrisson G, Lagergard T. How com- 16: mon is whooping cough in a nonvaccinated country? Pediatr Infect Dis 10. Patriarca PA, Biellik RJ, Sanden G, et al. Sensitivity and specificity of J 1993;12: clinical case definitions for pertussis. Am J Public Health 1988;78: 19. Krantz I, Taranger J, Trollforst B. Estimating incidence of whooping cough over time: a cross-sectional recall study of four Swedish birth cohorts. 11. Keitel WA, Edwards KM. Pertussis in adolescents and adults: time to Int J Epidemiol 1989;18: reimmunize? Semin Respir Infect 1995;10: De Serres G, Boulianne N, Duval B. Field effectiveness of erythromycin 12. Jenkinson D. Natural course of 500 consecutive cases of whooping cough: prophylaxis to prevent pertussis within families. Pediatr Infect Dis J a general practice population study. BMJ 1995;310: ;14: Sotomayor J, Weiner LB, McMillan JA. Inaccurate diagnosis in infants 21. Bortolussi R, Miller B, Ledwith M, Halperin S. Clinical course of pertussis with pertussis. Am J Dis Child 1985;139: in immunized children. Pediatr Infect Dis J 1995;14:870 4.