The duration of acute cough in pre-school children presenting to primary care: a prospective cohort study

Transcription

1 Family Practice Vol. 20, No. 6 Oxford University Press 2003, all rights reserved. Doi: /fampra/cmg613, available online at Printed in Great Britain The duration of acute cough in pre-school children presenting to primary care: a prospective cohort study Alastair D Hay, Andrew Wilson a, Tom Fahey b and Tim J Peters Hay AD, Wilson A, Fahey T and Peters TJ. The duration of acute cough in pre-school children presenting to primary care: a prospective cohort study. Family Practice 2003; 20: Background. Professional and parental uncertainty about the natural history of cough in preschool children may in part be responsible for the high consultation and reconsultation rates and widespread antibiotic use in primary care. A recent systematic review of the natural history of cough included studies of unrepresentative, selected patients with heterogeneous measures and definitions of cough duration. Objectives. The aim of the present study was to describe the post-consultation duration of cough, compare this with clinician and parental prediction of cough duration, and to determine the clinical factors associated with prolonged cough. Methods. A prospective cohort study of children aged 0 4 years with cough 28 days without asthma presenting to eight general practices in Leicestershire, UK was carried out. Sociodemographic and clinical data were collected, and parents and clinicians were asked to predict the proportion recovering within 7, 14, 21 and 28 days. Parents used a symptom diary to record cough and five other symptoms. Survival analysis was used to describe cough duration, and multivariable Cox regression was used to identify the factors independently associated with prolonged cough. Results. Fifty percent of the children had recovered at 10 days and 90% at 25 days. Cough was associated with fever, breathlessness, disrupted sleep and reduced activity in a high proportion of children. Longer post-consultation cough was associated with longer pre-consultation cough and use of day care facilities. Clinicians overestimated how quickly children recovered from acute cough. Parents predictions were accurate within 2 weeks, but they underestimated the proportion recovering in weeks 3 and 4. Conclusions. Clinicians should be aware that from the parental perspective, acute cough is not a trivial illness and that some children remain unwell at 3 4 weeks. For health professionals negotiating the use of antibiotics, this information may enable parental self-care, reduce medicalization and displace the need to prescribe. Keywords. Child, cough, primary care, prognosis. Introduction Cough is extremely common in pre-school children and generates considerable parental anxiety. 1,2 Fifty percent Received 17 February 2003; Revised 28 May 2003; Accepted 14 July Division of Primary Health Care, University of Bristol, Cotham House, Cotham Hill, Bristol BS6 6JL, a Department of General Practice and Primary Health Care, University of Leicester, Leicester General Hospital, Gwendolen Road, Leicester LE5 4PW and b Tayside Centre for General Practice, University of Dundee, Kirsty Semple Way, Dundee DD2 4AD, UK. Correspondence to Dr AD Hay; alastair.hay@ bristol.ac.uk present to primary care, 3 where it is the most frequently managed problem in the UK. 4,5 Between 40 and 70% of children with cough are prescribed an antibiotic. 6 8 Professional and parental uncertainty regarding the duration of cough in pre-school children may in part be responsible for the high consultation and reconsultation rates and widespread antibiotic use. With increasing patient expectations of medical care, the importance of self-care, patient empowerment and the setting of realistic expectations have never been more necessary. 9 Information about the natural history of conditions is an important part of this process 10,11 and has been shown to modify help-seeking behaviour 12 and the use of antibiotics 13 in adults. 696

2 Duration of cough in pre-school children 697 A systematic review of the natural history of cough in pre-school children 14 included data synthesized from eight randomized controlled trials (RCTs) of antibiotics. The patients included in these RCTs appeared to be an unrepresentative, selected population. In one of the cohort studies, 15 recruited children had a broad, heterogeneous range of presenting symptoms and only 63% were followed-up. The studies in the review also used different measures and definitions of cough duration, making it difficult to estimate the usual duration of the illness. While cough in young children is associated with viral infection in the majority, investigations are rarely used in primary care and therefore few children receive microbiological confirmation of the cause. For this reason, there have been repeated calls for studies of diagnosis and prognosis in primary care to be based only on symptoms and signs. 19,20 Therefore, the aim of this study was to describe the post-consultation duration of acute cough, compare this with clinician and parental prediction of cough duration and determine the clinical factors associated with prolonged cough in a representative sample of pre-school children presenting to primary care. Methods Practices and participants The study was approved by the Leicestershire Research Ethics Committee. To maximize the efficiency of child recruitment, practices with list sizes 8000 were invited by letter to participate. Recruitment took place from November to April over 2 years between 1999 and 2001, at morning and evening surgeries rotated between practices. A researcher was located in the surgery during recruitment sessions to ensure all eligible children were invited to participate. These were consecutive children aged 0 4 years with a cough of 28 days duration presenting to a GP or nurse practitioner, without asthma (identified clinically or defined as the medical recommendation for preventive or regular reliever treatment) or any other chronic disease. Data collection The data collection form and symptom diary were piloted and found to be feasible and acceptable. Children with cough were identified using a letter given to the parent on arrival at reception. The letter, on practice headed paper, asked if the child was attending for cough and directed the parent to the researcher. The parent then recorded the child s socio-demographic and clinical history variables before seeing the clinician. The clinician was asked to perform a routine clinical assessment and then record the examination and treatment details. The variables collected are listed in Table 1. Parents and clinicians were asked independently to predict the time Variables Socio-demographic TABLE 1 Children s characteristics (n = 256) % (unless otherwise stated) Age (months) Median 21 (IQR ) Male 51 Ethnicity White 82 Mixed 8 Indian 7 Pakistani 1 Black Caribbean 1 Other 1 Townsend deprivation score Mean 2.18 (SD 4.04) Attends day care, nursery or school 44 Previous or current breast feeding 50 Lives with a resident smoker 47 Shares a bedroom with at least one 23 other child Lives with at least two other children 24 Clinical history GP consultations in the preceding year a (0 1) 27 (2 4) 39 ( 4) 34 Cough main reason for consultation 66 Cough duration prior to recruitment (days) Median 5 (IQR 3 14) First clinician contact for the cough 71 Usually coughs with a cold 61 Currently receiving asthma prophylaxis b 1.6 Personal history of atopy c 29 Family history of atopy d 62 Already tried an over-the-counter remedy 46 Clinician assessing child GP only 96.1 Nurse practitioner (NP) only 3.5 Both GP and NP 0.4 Clinical examination Moderate or severe illness 23 Abnormal chest signs 22 Fever 11 Tachypnoea 9 Clinical assessment Classification of cough URTI 74 LRTI 7 RTI with bronchospasm 5 Bronchiolitis 3 Asthma e 3 Other 8 a Excluding prior consultations for this cough episode. b Included in the analysis as the recruiting clinician judged the child suitable for study inclusion. c Atopy refers to asthma, eczema or hay fever. d Family history in any first-degree relative. e New diagnosis of asthma, which was not an exclusion criterion. IQR = interquartile range; URTI = upper resporatory tract infection; LTRI = lower respiratory tract infection.

3 698 taken to complete cough resolution. The parent was then asked to complete a validated symptom diary 21 modified for use with younger children. This recorded the severity (on a Likert scale from normal/not affected = 0 to as bad as it could be = 6) of cough, shortness of breath, sleeplessness, activity levels, how unwell the child was and fever, each day until the cough had resolved (two consecutive days of normality of all symptoms). Parents were telephoned weekly to support diary completion, and the diary was returned by post. Sample size The sample size was determined by the primary aim of the study, which was to estimate the various percentages of children whose cough had recovered through time. A sample size of 220 would yield a 95% confidence interval (CI) of 85 94% around an estimate of 90% for the proportion recovered at the relevant time point, using an exact binomial calculation. 22 This was considered to be an adequate level of precision for this proportion, which was viewed in advance as being the most clinically useful. Data entry and analysis Data were single entered onto an Access database. No errors were found in 14 randomly selected cases. We used Stata version 7 for all analyses. 22 To describe the duration of the six symptoms from the symptom diary, we used survival analysis. We then explored the associations between the socio-demographic, history, examination and treatment variables (listed in Table 3), and post-consultation cough duration using univariable Cox regression. Multivariable Cox regression was then used to identify the clinical variables (i.e. those available to the clinician) independently associated with cough duration, with a small set of pre-specified variables (age, sex, ethnicity and deprivation) adjusted for in all models. The variables were first considered in four separate groups (socio-demographic, clinical history, examination findings and treatment) and then a final model was constructed across the groups. As this was an exploratory, hypothesis-generating analysis and we wanted to avoid missing potentially important variables, at the univariable stage a relatively liberal threshold for P-values of 20% was employed to select explanatory variables for the multivariable modelling. Likewise, a 10% threshold was used for the within-groups models and for the final model. These results are expressed as hazard ratios where values 1 indicate an association with increased cough duration. We checked for interactions between variables in the final model. We tested the Cox proportionality assumptions visually using log log plots and statistically by tests based on the Schoenfeld residuals. 23,24 We compared parental and clinician cough duration prediction against actual cough duration in those children with uncensored data available for the first week of the illness, using kappa statistics. Family Practice an international journal Results Practices and children Twenty-nine practices were invited; nine agreed to participate. Recruitment was stopped from one practice due to low recruitment rates related to its small pre-school population. The remaining practices were located in inner city, urban and rural areas and served a total population of , of whom 6722 (6.4%) were pre-school children. Figure 1 summarizes the recruitment of children: 843 of the 883 (95%) children presenting during a recruitment session were assessed for eligibility; 256 of the 290 (88%) eligible children were recruited. Of these, 71% were recruited at their first clinician contact for the cough and follow-up data for cough duration were available in 228 (89%). Table 1 shows that most children were under 2 years, and half were female, 82% were white, their mean Townsend deprivation score was 2.18, 18% were prescribed an antibiotic and 13% a bronchodilator. Cough duration Figure 2 summarizes cough duration data using a Kaplan Meier curve. It was 25 days before 90% of the children recovered. Half the children had recovered at 10 days and 75% at 16 days. Table 2 summarizes the duration of the five other associated symptoms and shows that cough was the slowest symptom to resolve. Parents reported other symptoms in a high proportion of children, particularly that children were unwell (95%) and experienced disrupted sleep (95%). The duration of moderately bad or worse ( 3 on the Likert scale) symptoms were similar to total symptom duration for the 90% resolution point; cough 21 days, disrupted sleep 19 days, reduced activity 19 days, child being unwell 18 days and fever 15 days. Insufficient data were available to calculate breathlessness duration for the 90% resolution point. Univariable analysis Table 3 presents hazard ratios (95% CI) for the univariable relationships between the 25 explanatory variables tested and cough duration. Nine of these were significant at the 20% level, of which seven were significant at the 10% level and four at the 5% level. Regarding the latter most significant factors, attending daycare, nursery school or school, increasing pre-consultation cough duration and recruitment at second (or higher) consultation for cough were all associated with longer postconsultation cough. Multivariable analysis Table 4 shows that increasing cough duration prior to consultation and the use of daycare were the only factors associated with increased cough duration after adjustment for age, sex, deprivation, ethnicity and each

4 Duration of cough in pre-school children 699 FIGURE 1 Child recruitment and follow-up flow diagram

5 700 Family Practice an international journal FIGURE 2 Kaplan Meier curve of cough duration (bars indicate 95% CI; numbers above the curve are censored cases) TABLE 2 Duration of the six symptoms (days) calculated using survival analysis Cough Short of Sleeplessness Reduced Unwell Fever breath activity Proportion resolved Symptom duration in days a (95% CI) (5 7) 3 (3 4) 4 (3 5) 3 (3 4) 4 (3 4) 2 (2 3) (9 10) 6 (5 7) 7 (6 9) 6 (5 7) 6 (5 7) 5 (4 6) (13 19) 11 (9 14) 12 (11 15) 12 (9 15) 11 (9 14) 10 (7 12) (23 b ) 21 (17 b ) 22 (18 25) 20 (17 22) 18 (15 25) 15 (13 19) Symptom end point Symptom resolved Censored Never symptomatic Data missing Total % Ever symptomatic 100% 77.6% 94.7% 83.3% 94.6% 72.0% a Only children ever symptomatic were included in the symptom duration calculations. b Not possible to calculate upper 95% CI due to small numbers of children with symptoms lasting this long. other. There was a dose response relationship for pre-consultation cough duration, with a hazard ratio of 0.76 (95% CI , trend test P 0.001). Although prior cough duration and whether this was the child s first clinician contact for the cough were both significantly associated with cough duration on univariable analysis, first clinician contact was not significant after adjusting for prior cough duration (P = 0.18), probably because they reflect similar aspects of the cough history. Bronchodilator treatment was associated with prolonged cough on univariable analysis (P = 0.02), but this relationship was confounded by prior cough duration, daycare and the pre-specified variables (P = 0.19). Bronchodilator treatment did not confound the relationships between prior cough duration or daycare and post-consultation cough duration (P = and 0.11, respectively). As none of the other explanatory clinical variables ever approached significance at the P 0.1 level in the multivariable models, the final model is therefore as shown in Table 4.

6 Duration of cough in pre-school children 701 TABLE 3 Univariable hazard ratios (95% CI) for cough duration Variables (no. available Median cough Hazard ratio b (95% CI) P-value for analysis) a duration (days) Socio-demographic Age (228) 2 months months ( ) months ( ) 0.75 Sex (228) Male 9 1 Female ( ) 0.57 Ethnicity (228) White 9 1 Non-white ( ) Deprivation (228) Most 9 1 More ( ) Less ( ) Least ( ) 0.44 Attends daycare, nursery or school (227) Yes ( ) Previous or current breast feeding (227) Yes ( ) 0.86 Lives with a resident smoker (228) Yes ( ) 0.38 Shares a bedroom with 1 other child (228) Yes ( ) 0.92 Lives with 2 other children (227) Yes ( ) 0.92 Clinical history GP consultations in the preceding year c (221) ( ) ( ) 0.75 Cough duration prior to consultation (227) 7 days days ( ) days ( ) days ( ) Cough main reason for presentation (223) Yes ( ) First clinician contact for the cough (227) No 13 1 Yes ( ) Usually coughs with a cold (199) No 11 1 Yes ( ) 0.28 Currently receiving asthma prophylaxis (224) Yes ( ) 0.78 Personal history of atopy d (219) Yes ( ) 0.47

7 702 Family Practice an international journal TABLE 3 Continued Variables (no. available Median cough Hazard ratio b (95% CI) P-value for analysis) a duration (days) Family history of atopy e (214) Yes ( ) 0.74 Already tried an over-the-counter remedy (226) Yes ( ) 0.11 Clinical examination Illness severity (225) Mild 9 1 Moderate/severe ( ) Abnormal chest signs (218) Yes ( ) 0.92 Fever (222) Yes ( ) 0.14 Tachypnoea (223) Yes ( ) 0.59 Classification of cough (224) URTI 9 1 Other ( ) 0.41 Treatment variables Antibiotic treatment (225) Yes ( ) 0.76 Bronchodilator treatment (225) Yes ( ) 0.02 a Numbers vary due to missing values. b A hazard ratio 1 indicates shorter cough duration than reference group; 1 indicates longer cough duration. c Excluding prior consultations for this cough episode. d Atopy refers to asthma, eczema or hay fever. e Family history in any first-degree relative. TABLE 4 Univariable and multivariable hazard ratios (95% CI) for cough duration Variable Univariable (n a = 227) Multivariable b (n a = 226) Cough duration prior to recruitment (days) ( ) 0.77 ( ) ( ) 0.64 ( ) ( ) 0.36 ( ) P-value Daycare No 1 1 Yes 0.71 ( ) 0.72 ( ) P-value a Number of cases with complete data available for all variables in model. b Adjusted for age, sex, ethnicity and deprivation and other variables. Tests of proportionality and for interactions The statistical tests based on the Schoenfeld residuals showed that the models did not violate the proportionality assumptions (P = 0.46). However, because these tests can be underpowered to detect non-proportionality, we also visually examined the complementary log log plots. This showed some evidence of non-proportionality for prior cough duration, which was confined to the first 9-day period. There was no evidence of an interaction between the two variables in the final model (P = 0.68). Clinicians and parents predictions of cough duration Figure 3 shows the difference between clinicians and parents predictions and actual cough duration. Clinicians consistently overestimate the proportion of children recovering across all time periods compared with actual cough duration. Parents predictions

8 Duration of cough in pre-school children 703 FIGURE 3 Proportion of children predicted to recover by parents and clinicians versus actual recovery at 7, 14, 21 and 28 days were accurate for the first 2 weeks, but underestimated the proportion recovered between 2 and 4 weeks. Parents and clinicians stated they did not know how long the cough would last in 25 and 2% of children, respectively. Two children had cough duration censored before 7 days. Agreement between parents and clinicians predictions and actual cough duration was fair (kappa = 0.24) and poor (kappa = 0.028), respectively. 25 Discussion Summary of main results This study shows that 50% of the children had recovered at 10 days and 90% at 25 days. Cough was associated with fever, breathlessness and disrupted sleep and activity levels in a high proportion of children. There were associations found between more prolonged pre-consultation cough and use of daycare and longer post-consultation cough. Clinicians expected more children to recover than actually did across all time periods. Interpretation of results While cough is a self-limiting illness, half the children remained symptomatic at 10 days and a significant minority (10%) at 25 days. This important information will have different uses. First, to reassure parents that prolonged symptoms are common and not necessarily a marker of serious disease. Secondly, giving parents a realistic expectation of cough duration at the outset, particularly those who waited longer before consulting, may reduce the need to reconsult. Finally, for health professionals who are negotiating the use of antibiotics, specific information about the natural history may enable parental self-care, reduce medicalization and displace the need to prescribe. Where this fits in with other research This is the first study to specifically recruit children with cough and report cough duration as the main outcome. In addition, it is an improvement on previous studies because of the high proportion of eligible children recruited and followed-up. 14 Our study showed a similar median cough duration to that found in adults presenting to primary care with cough. 26 This study found that longer pre-consultation cough was associated with longer post-consultation cough. A similar result was found in a study of patients presenting to primary care with sore throat. 27 This may be due to more severe illness in patients who wait longer to consult or may reflect consistently increased vigilance in the reporting of pre- and post-consultation symptoms. Similarly, this study suggests that children attending daycare may experience longer coughs. Children in daycare previously have been shown to experience a higher incidence of respiratory tract infections and longer upper respiratory tract infections with higher rates of complications (otitis media and sinusitis). 32 These findings may be due to differences in exposure to and transmission of the microbiological organisms in daycare. Limitations The representativeness of the sample was limited by our need to operate within funding constraints and maximize efficiency of recruitment from larger, busier practices, during the months of the year when respiratory tract infections are most common. While practices

9 704 Family Practice an international journal prepared to participate in research may be atypical, children and parents are less likely to differ systematically. The proportion of 0 4 year olds to total practice population was 6.4%, similar to the 7% found in the 1991 Census, 4 but deprived and Asian children were over-represented compared with regional 33 and national 34 populations. However, we found no evidence of independent associations between deprivation, ethnicity, or antibiotic or bronchodilator treatment and cough duration. Therefore, it seems unlikely that selection bias of practices will materially alter the conclusions of this study. Proportionality is a key assumption of the Cox regression model. While the log log plot did show some evidence of non-proportionality for prior cough duration in the first 9 days, this may be due to the waning effect of pre-consultation cough duration on post-consultation cough duration, and we would suggest that this is re-examined in future research. Conclusions Clinicians should be aware that from the parental perspective, acute cough is not a trivial illness and that some children remain unwell for 3 4 weeks. This information may reduce parental expectation for antibiotics and the need to reconsult. Future research should examine the effects of giving information about the natural history of cough to parents and clinicians on help-seeking and prescribing behaviour. Acknowledgements We wish to thank the Trent Focus Collaborative Research Network, the nine Leicestershire practices and the patients who participated in this study. This study was supported by a grant from the Department of General Practice and Primary Health Care, University of Leicester. References 1 Cornford CS, Morgan M, Ridsdale L. Why do mothers consult when their children cough? Fam Pract 1993; 10: Kai J. What worries parents when their preschool children are acutely ill, and why: a qualitative study. Br Med J 1996; 313: Wyke S, Hewison J, Russell IT. Respiratory illness in children: what makes parents decide to consult? Br J Gen Pract 1990; 40: McCormick A, Fleming D, Charlton J. Morbidity Statistics from General Practice. Fourth National Study London: HMSO, Okkes M, Oskam SK, Lamberts H. The probability of specific diagnoses for patients presenting with common symptoms to Dutch family physicians. J Fam Pract 2002; 51: Stott NC. Management and outcome of winter upper respiratory tract infections in children aged 0 9 years. Br Med J 1979; 1: Vinson DC, Lutz LJ. The effect of parental expectations on treatment of children with a cough: a report from ASPN. J Fam Pract 1993; 37: Medical audit in general practice. I: effects on doctors clinical behaviour for common childhood conditions. North of England Study of Standards and Performance in General Practice. Br Med J 1992; 304: Smith R. The NHS: possibilities for the endgame. Think more about reducing expectations. Br Med J 1999; 318: Moynihan R, Smith R. Too much medicine? Br Med J 2002; 324: Pencheon D. Matching demand and supply fairly and efficiently. Br Med J 1998; 316: Macfarlane JT, Holmes WF, Macfarlane RM. Reducing reconsultations for acute lower respiratory tract illness with an information leaflet: a randomized controlled study of patients in primary care. Br J Gen Pract 1997; 47: Macfarlane J, Holmes W, Gard P et al. Reducing antibiotic use for acute bronchitis in primary care: blinded, randomised controlled trial of patient information leaflet. Br Med J 2002; 324: Hay AD, Wilson AD. The natural history of acute cough in children aged 0 to 4 years in primary care: a systematic review. Br J Gen Pract 2002; 52: Gulbrandsen P, Fugelli P, Kvarstein G, Moland L. The duration of acute respiratory tract infections in children. Scand J Prim Health Care 1989; 7: Monto AS, Cavallaro JJ. The Tecumseh study of respiratory illness. II. Patterns of occurrence of infection with respiratory pathogens, Am J Epidemiol 1971; 94: Denny FW, Clyde WA, Jr. Acute lower respiratory tract infections in nonhospitalized children. J Pediatr 1986; 108: Chapman RS, Henderson FW, Clyde WA, Jr, Collier AM, Denny FW. The epidemiology of tracheobronchitis in pediatric practice. Am J Epidemiol 1981; 114: Medical Research Council. Primary Health Care Research Review. London: Medical Research Council, Stocks N, Fahey T. Labelling of acute respiratory illness: evidence of between-practitioner variation in the UK. Fam Pract 2002; 19: Watson L, Little P, Moore M, Warner G, Williamson I. Validation study of a diary for use in acute lower respiratory tract infection. Fam Pract 2001; 18: Stata Corporation. Stata Statistical Software (7.0). College Station (TX), Grambsch P, Therneau T. Proportional hazards tests and diagnostics based on weighted residuals. Biometrika 1994; 81: Hess KR. Graphical methods for assessing violations of the proportional hazards assumption in Cox regression. Stat Med 1995; 14: Altman DG. Practical Statistics for Medical Research. London: Chapman and Hall, Holmes WF, Macfarlane JT, Macfarlane RM, Hubbard R. Symptoms, signs, and prescribing for acute lower respiratory tract illness. Br J Gen Pract 2001; 51: Little P, Gould C, Williamson I, Warner G, Gantley M, Kinmonth AL. Clinical and psychosocial predictors of illness duration from randomised controlled trial of prescribing strategies for sore throat. Br Med J 1999; 319: Denny FW, Collier AM, Henderson FW. Acute respiratory infections in day care. Rev Infect Dis 1986; 8: Holberg CJ, Wright AL, Martinez FD, Morgan WJ, Taussig LM. Child day care, smoking by caregivers, and lower respiratory tract illness in the first 3 years of life. Group Health Medical Associates. Pediatrics 1993; 91: Petersson C, Hakansson A. A retrospective study of respiratory tract infections among children in different forms of day care. Scand J Prim Health Care 1990; 8:

10 Duration of cough in pre-school children Harsten G, Prellner K, Heldrup J, Kalm O, Kornfalt R. Acute respiratory tract infections in children. A three-year follow-up from birth. Acta Paediatr Scand 1990; 79: Wald ER, Guerra N, Byers C. Upper respiratory tract infections in young children: duration of and frequency of complications. Pediatrics 1991; 87: Hippisley-Cox J, Pringle M, Coupland C, Hammersley V, Wilson A. Do single handed practices offer poorer care? Cross sectional survey of processes and outcomes. Br Med J 2001; 323: Babb, P. Social Trends. Norwich: Office for National Statistics, 2002.