Canadian Acute Respiratory Illness and Flu Scale (CARIFS)

Journal of Clinical Epidemiology 53 (2000) 793–799

Canadian Acute Respiratory Illness and Flu Scale (CARIFS): Development of a valid measure for childhood respiratory infections Benjamin Jacobsa, Nancy L. Younga,b, Paul T. Dicka,b, Moshe M. Ippa, Regina Dutkowskic, H. Dele Daviesd, Joanne M. Langleye, Saul Greenberga, Derek Stephensf, Elaine E.L. Wanga,f,* b

a Department of Paediatrics, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada Paediatric Outcomes Research Team, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada c Department of Clinical Sciences, Hoffmann-La Roche, Nutley, New Jersey, USA d Department of Paediatrics, University of Calgary, Calgary, Alberta, Canada e Department of Paediatrics, Dalhousie University, Halifax, Nova Scotia, Canada f Clinical Epidemiology Unit, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada Received 7 July 1999; accepted 22 December 1999

Abstract Although acute respiratory infection (ARI) is the most frequent clinical syndrome in childhood, there is no validated measure of its severity. Therefore a parental questionnaire was developed: the Canadian Acute Respiratory Illness Flu Scale (CARIFS). A process of item generation, item reduction, and scale construction resulted in a scale composed of 18 items covering three domains; symptoms (e.g., cough); function (e.g., play), and parental impact (e.g., clinginess). The validity of the scale was evaluated in a study of 220 children with ARI. Construct validity was assessed by comparing the CARIFS score with physician, nurse, and parental assessment of the child’s health. Data were available from 206 children (94%). The CARIFS correlated well with measures of the construct (Spearman’s correlations between 0.36 and 0.52). Responsiveness was shown, with 90% of children having a CARIFS score less than a quarter of its initial value, by the tenth day. © 2000 Elsevier Science Inc. All rights reserved. Keywords: Acute respiratory infection; Severity of illness index, Outcome assessment; Pediatrics; Outpatients; Questionnaires

1. Introduction Acute respiratory illness (ARI) is the most common illness of childhood and the most frequent reason for children’s visits to a physician. A study of 273 toddlers attending 52 day-care centers found that they suffered a “cold-like” illness 23.4% of the time. This was 10 times the prevalence of the next most frequent illness, diarrhea [1]. Three quarters of the children consulted a physician at least once for these illnesses during the 6-month study period. In the United States, the annual economic costs of influenza epidemics are estimated to exceed $12 billion [2]. In Canada, physician billing alone account for over $200 million annually, excluding treatment costs [3]. Few of the treatments have been adequately evaluated, largely due to the absence of validated outcome measures. * Corresponding author. Vice President, Clinical and Medical Affairs, Aventis Pasteur-Connaught Campus, 1755 Steeles Avenue West, Toronto, Ontario, M2R 3T4, Canada. Tel: 416–667–2476; fax: 416–667–2939. E-mail address: [email protected]

A systematic review of 12 studies of the treatment of ARI in children found none which used a validated pediatric outcome measure of disease severity [4]. Each of these studies used a new method of assessing treatment success. Seven studies used parent-recorded symptom diaries, but none gave data on the diary’s measurement properties. Five studies used outcomes recorded by the physician only, although these children were all treated as outpatients, and so the physician is likely to have limited perception of the illness. Stein has classified disease severity measures based on the scope of disease effects measured [5]. Her four levels of scale are: Biological, Clinical, Functional, and the Financial or Social Burden of Illness. From the patient’s perspective, it is the third level, the functional impact of disease, that is most relevant. This was therefore the type of measure developed in this study. There are specific challenges to developing pediatric disease severity measures. Young children, who are unable to articulate their complaints, may manifest illness with functional problems alone. Limited language abilities of young

0895-4356/00/$ – see front matter © 2000 Elsevier Science Inc. All rights reserved. PII: S0895-4356(99)00 2 3 8 - 3

794

B. Jacobs et al. / Journal of Clinical Epidemiology 53 (2000) 793–799

children preclude self-report measures and necessitate the use of parents, clinicians, or other proxies. Issues such as development, growth, and puberty are crucial in pediatrics. The child’s place in the family, and the distinct emotional peculiarities of children all need to be taken into account when assessing a child’s functional ability or disability, especially in term of quality of life. Pediatric measures have been developed to predict mortality in serious illness [6], to predict diagnosis [7], and to measure functional components of health [8,9] or quality of life [10] in chronic diseases, but none are suitable for assessing severity of outpatient acute illness. The purpose of this study was to develop a disease severity measure appropriate for ARI, including influenza, in children. 2. Methods 2.1. Scale development A MEDLINE literature search (1996 to June 1997) was performed for studies of adults or children with ARI. Twenty-five appropriate items were identified in unvalidated questionnaires used as outcome measures in clinical trials of adult [11–13] and childhood ARI treatment [14,15] and from validated pediatric general health and functional status measures. [6,8,9,16–20] These 25 items were each written on a card. The cards, along with five blank cards, were given to three general pediatricians and to parents of 23 children brought to their pediatrician’s office with ARI. Each person was asked to remove those cards that contained

items they considered unimportant. They were permitted to add up to five other important items by writing on the blank cards. They were then asked to rank the cards in order of importance. Four items that were rarely considered relevant, or were ranked low in importance, were excluded. The excluded items were: “not responding to parents,” “can’t communicate well,” “difficult behavior,” “sticky eyes or pink eye.” Three items (“moody,” “irritable,” “fussy”) were merged into one. One item frequently suggested by parents, “clinginess,” was added. Three items unsuitable for an ordinal response scale were left out of the scale as independent questions with binary responses (“needing medical attention,” “unable to attend school/daycare,” “need to take medication”). This process resulted in the Canadian Acute Respiratory Infection and “Flu Scale” (CARIFS), consisting of 18 items each answered on a 4-point ordinal scale (no problem ⫽ 0, minor problem ⫽ 1, moderate problem ⫽ 2, major problem ⫽ 3) (see Fig. 1). The CARIFS score was calculated as the sum of the items. As a number of items (“headache,” “sore throat,” and “muscle aches or pains”) were not applicable to infants, the total for scales for items marked “Don’t Know or Not Applicable” was calculated as the mean of all applicable items multiplied by the total number (18) of items. This results in a score between 0 (best possible health) and 54 (worst possible health) for all children. 2.2. Validity assessment strategy The theoretic framework proposed by Kirshner and Guyatt for validating a health measurement instrument was

Fig. 1. Canadian Acute Respiratory Infection and ‘Flu Scale (CARIFS).

B. Jacobs et al. / Journal of Clinical Epidemiology 53 (2000) 793–799

used [21]. Construct validity (correlation between the new measure and indicators of related phenomena) and responsiveness (the degree to which the instrument reflects clinical change) were the two main methods of validation. The measures used to assess construct validity were: clinician assessment (physician and nurse), parental assessment (on a global visual analog scale), and recorded body temperature. Responsiveness was assessed by examining the performance of the scale during the period of clinical recovery. It was expected that the CARIFS score would be normal by the end of the study, because uncomplicated ARI should be resolved by 14 days. Additionally, the ability of the scale to discriminate between those who did and did not have further attendance by a physician, who did and did not obtain antibiotic treatment and did or did not suffer the complication of otitis media was determined. It was hypothesised that the CARIFS score would remain high longer in children with these indicators of complications. 2.3. Study setting This study was performed over the winter of 1997–1998 in three Canadian cities; Calgary, Halifax, and Toronto. Children up to 12 years of age attending their primary care physician with an ARI were eligible for inclusion. Children were excluded if the encounter was not the first physician visit for the ARI, or if it was not within 72 hours of onset of symptoms. ARI was defined as a history of a fever (⬎38.0⬚C oral or ⬎37.5⬚C axillary), at least one respiratory symptom (nasal stuffiness, runny nose, cough, or sore throat) and at least one systemic symptom (lethargy, myalgia, headache). Children with underlying illness for which daily medication was taken were excluded. Written parental consent, and the child’s assent in those over six years of age, was obtained. The study was approved by the Research Ethics Boards of The Hospital for Sick Children in Toronto, the Alberta Children’s Hospital, and the Izaak Walton Killam Grace Health Centre for Children Women and Families in Halifax. 2.4. Measurements A parent of each child was given a diary, consisting of 16 CARIFS score sheets, and a thermometer (Digital Fever Thermometer, Model 524051: Becton Dickinson and Company, Franklin Lakes, NJ, USA). Parents were asked to complete a CARIFS score twice a day for the first seven days, once on Day 10, and once on Day 14. At each assessment a visual analog score of the parental global rating of their child’s health on a 10 centimetre linear scale was recorded (see Fig. 1), as well as the child’s axillary temperature. Independent assessments of the child’s overall illness level (mild/moderate/severe) at enrollment (Day 1) were completed by the child’s physician and a research nurse. The nurse also recorded the Yale Observation Scale for the child. The Yale scale is a six-item observational scale, assessing the child’s appearance, activity and responsiveness. It has been shown to be a valid predictor of serious illness in

795

febrile children up to three years of age attending emergency departments [7], and was adapted for this study to accommodate children up to the age of 12. The training video produced by McCarthy, the Yale scale’s designer, was viewed by all the nurses at the start of the study. A nasopharyngeal swab was obtained from each patient at enrollment and submitted to the local laboratory for viral diagnosis. In Toronto and Calgary both immunofluorescence assays and isolation were undertaken for influenza virus, parainfluenza virus, adenovirus, and respiratory syncytial virus on all specimens. In Halifax only viral isolation was performed. The research nurse visited each family at home on Day 3 to repeat her assessment and Yale score. She telephoned the family on Day 7 and Day 10 to ensure compliance with diary keeping. Finally, at the end of the data collection period (Day 14) the study nurse telephoned again to ask about the child’s health care usage during the illness, and ask for the diary to be mailed back to the study center. Child’s health care usage included subsequent physician visits, admissions to hospital, and receipt of antibiotics. This data was recorded in a standardized interview form.

3. Data analysis All the diary and demographic data collection forms were transcribed onto a personal computer using a digital image scanner and TELEform for Windows software (Cardiff Software, San Marcos, CA, USA). The SAS software package was used for statistical analysis (version 6.12: SAS Institute Inc, Cary, NC, USA). Analyses were performed for the entire study population as well as the subgroup with influenza infection. 3.1. Item analysis Each item was assessed at the first assessment for its importance (the frequency of its endorsement as a “Major problem”), its relevance (the frequency of its endorsement as “not applicable”), its correlation with the construct (Spearman correlation, with the parent’s global visual analog scale) and its consistency with the rest of the scale (item-total correlation). In a scale with good internal consistency each item has an item-total correlation of at least 0.2 [22]. A very high item-total correlation would indicate that it has little impact on the total score. Inclusion of that item in the scale may not be necessary. 3.2. Scale analysis Differences in initial CARIFS score for different age groups, gender, viral etiologies, and study centers were assessed by analysis of variance (ANOVA). Internal consistency of the CARIFS was summarized as Cronbach’s ␣ [23]. Reliability of the scale was not assessed directly. However, test-retest intra-rater reproducibility was estimated by comparing CARIFS scores performed in the morning and evening of the second study day, by the same

796

B. Jacobs et al. / Journal of Clinical Epidemiology 53 (2000) 793–799

rater (mother). The intra-class correlation coefficient was calculated using a random effects model and repeated measures ANOVA of the variation between scores [24]. Correlation between the CARIFS score and the physician assessment, nurse assessment, Yale score, the parents’ global visual analog score, and the temperature measurements was assessed at the time of enrollment [25]. Responsiveness was shown graphically by plotting the CARIFS scores over the 14 study days to show change with clinical recovery. Changes in the CARIFS between enrollment and the home visit were compared with changes in the other measures of the construct. All correlations were expressed as Spearman’s rank coefficients. It was hypothesized that the time for CARIFS scores to normalise would be longer in those who have a complicated course. However, the period over which the CARIFS decreases may vary depending on the initial CARIFS score. Therefore, the time until the CARIFS score dropped below one quarter of the enrollment CARIFS score was considered for each patient. This duration was called the DayC1/4. Using survival analysis the DayC1/4 was compared for those with subsequent visits to health care providers versus those without, those who received antibiotics versus those who did not, and those having ear infections versus those who did not. Kaplan-Meier graphs of these durations were plotted, and the significance of differences assessed with the log-rank chi-squared test [26]. A statistical significance level of ␣ ⬍ 0.05 was set throughout. Sample size calculations were based on confidence limits of the correlation coefficients used to establish construct validity. The sample size was chosen to allow assessment of the validity of the scale among the subgroup of influenza patients independently. Therefore it was planned to recruit at least 200 children, including at least 50 with influenza. A coefficient of 0.4 found in a sample of 50 children gives 95% confidence of the true correlation being at least 0.2 in the population. The same coefficient of 0.4 found in a sample of 200 children gives 95% confidence of being at least 0.3 in the population. This sample size calculation was performed with the SamplePower computer programme (version 1.00; SPSS Inc., Chicago, IL, USA) using Cohen’s method of Fisher Z approximation [25]. While this formula is designed for Pearson’s correlation, it is assumed to be sufficiently robust to estimate sample size for Spearman’s coefficient. 4. Results In total 220 patients were enrolled in the study. Data were available from the 206 (94%) who mailed back completed diaries. These children were aged 1 month to 12 years (median 3.2 years); 65 were less than two years of age, 75 were 2 to 5 years, and 66 were 5 to 12 years of age. Influenza A infection was identified in 69 (33%), another virus in 35 (17%), and no virus was identified in 102 (50%). (see Table 1). In the 14 subjects who did not return diaries, there was no

Table 1 Baseline CARIFS scores by subgroup Number of children All children Child age (years) 0–2 2–5 5–12 Gender Boys Girls Virus Influenza Other virusa No virus isolated Site Calgary Halifax Toronto a

Initial CARIFS score

P value (ANOVA)

Mean

(SD)

206

27.99

(10.26)

65 75 66

35.09 32.35 32.41

(10.7) (10.9) (10.4)

0.243

105 101

33.36 33.10

(10.0) (11.5)

0.861

69 35 102

35.19 33.51 31.82

(9.8) (12.6) (10.5)

0.130

34 24 148

33.33 29.20 33.86

(11.0) (10.4) (10.6)

0.154

Adenovirus (3), parainfluenza (3), respiratory syncytial virus (29).

significant difference in gender distribution (57% were male, P ⫽ 0.7); mean (⫾ standard deviation) number of illness days prior to enrollment (1.85 ⫾ 0.99 in excluded subjects vs. 1.68 ⫾ 0.83 in included subjects, P ⫽ 0.5); mean (SD) age 2.88 ⫾ 2.72 in excluded vs. 4.21 ⫾ 3.26 in included subjects, P ⫽ 0.14); mean (SD) visual analog score on enrollment (3.71 ⫾ 2.37 in excluded vs. 3.90 ⫾ 1.96 in included subjects, P ⫽ 0.73; or distribution by viral etiology (2 infections with RSV, 1 each with influenza, parainfluenza and adenovirus, and 9 with unknown viral etiology, P ⫽ 0.09). All items were marked “Major problem” for at least 20% of the children at enrollment, except the items “sore throat,” “muscle aches and pains,” “vomiting,” “not interested in what’s going on” and “unable to get out of bed.” Less than 5% of the parents responded “Don’t Know or Not Applicable,” except for the items “unable to get out of bed” and, in children less than four years of age, the items “headache,” “sore throat,” and “muscle aches and pains.” Each item had a moderately strong correlation with the parent’s global visual analog rating (correlation ⬎ 0.2), except the items “vomiting” and “crying.” All item-total correlations were greater than 0.2. Only one item, “not playing well,” had a high item-total correlation (0.73). There was no difference in baseline CARIFS scores according to age, viral etiology, gender, or study site: (see Table 1). The Cronbach’s ␣ for the 18 item scale at the enrollment assessment was 0.89. The intra-rater reliability for mothers, on the second day, was 0.808. The CARIFS score correlated well with both health professional and parental measures (physician’s assessment 0.36, nurse’s assessments 0.44, Yale score 0.48 and parental global visual analog assessment 0.52). There was a higher correlation between physicians’ and nurses’ assessments (0.70), but these professionals may have conferred. Similarly, the nurse’s assessment and the Yale assessment were

B. Jacobs et al. / Journal of Clinical Epidemiology 53 (2000) 793–799 Table 2 Construct validity at enrollment (Spearman coefficients)

CARIFS Physician assessment Nurse assessment Yale score Parent global assessment Temperature

Physician assessment

Nurse assessment

Yale score

Parent global assessment

Table 3 Spearman correlation coefficients of changes between enrollment and home visit Difference in CARIFS

0.36 0.44 0.48

0.70 0.55

0.71

0.52 0.29

0.30 0.28

0.28 0.38

0.38 0.34

0.27

performed by the same individual and had a high correlation (0.71). The health professionals assessments correlated less well with the parental global ratings than the CARIFS (Table 2). Temperature correlated poorly with all other measures. Responsiveness of the scale is shown graphically in Fig. 2. The scores improved over the 14 days, consistent with clinical predictions of the course of ARI. The CARIFS decreased from a mean (SD) score on Day 1 of 28.0 (10.3) to 17.1 (11.7) on Day 3, and 2.5 (5.7) on Day 14. Changes in CARIFS correlated with changes in the See Table 3. The duration of illness after enrollment, measured as the Day C1/4, had a median of 4.5 days. This level of recovery was reached by 25% of children by 3 days, 75% by 6.5 days, 90% by 10 days, and 95% by the end of the study at 14 days.

797

Difference in nurse assessment Difference in Yale score Difference in parent global assessment Difference in temperature

Difference in nurse assessment

Difference in Yale score

Difference in parent global assessment

0.38 0.42

0.59

0.68

0.36

0.37

0.32

0.21

0.21

0.29

Survival analysis showed a significant difference (P ⫽ 0.007) in the DayC1/4 between the 136 children who had no further physician visits (median ⫽ 4.0 days), the 54 children with one further visit (5.5 days), and the 15 with two or more visits or hospital attendance (6.5 days). The increase in duration was not statistically significant for the children with ear infections or antibiotics. See Table 4. Among the 69 children with proven influenza, the DayC1/4 was significantly later for all three indicators of complicated disease. The pattern of correlation with construct measures and responsiveness for the influenza subgroup were similar to those reported for the total sample of 206 children. 5. Discussion

Fig. 2. Boxplots of responsiveness of the CARIFS to clinical improvement.

Viral ARI is the only infectious disease that affects every human. Evidence of infection with respiratory syncytial virus, for example, can be found in all children who survive to their second birthday [27]. Despite this, a systematic review of 12 studies of the treatment of pediatric ARI found none which used a validated outcome measure of disease severity [4]. The CARIFS was designed to fulfil this need. It consists of problems considered important by parents and pediatricians, as shown in the preliminary study. Each of the 18 items performed satisfactorily in this large field study. The scale as a whole shows internal consistency and validity. The CARIFS correlated better across the range of the health professional and parental construct measures than any of the other measures. Although the health professionals’ assessments correlations were higher with each other than with the CARIFS, their poorer correlation with parental assessments suggest that they miss aspects of a child’s illness which are of importance to families. This is not surprising given the brief nature of their encounters and their different perspectives. The poor correlation of fever with all other measures of illness severity was unexpected. However, even at the time of enrollment, when the children were most ill, only 59% were febrile, and by the time of the home visit, which was

798

B. Jacobs et al. / Journal of Clinical Epidemiology 53 (2000) 793–799

Table 4 Survival analysis of time to drop below a quarter of initial CARIFS Score (DayC1/4) Indicator of prolonged illness All children Ear infection Absent Present Antibiotic used Absent Present Further medical visits None 1 2 or more Influenza subgroup Ear infection Absent Present Antibiotic used Absent Present Further medical visits None 1 2 or more

Number of children

DayC1/4 median (days)

95% confidence interval of DayC1/4 (days)

Degrees of freedom

Long-Rank chi-square

P-value

151 54

4.5 5.0

4.0–5.0 4.0–6.5

1

2.493

0.114

139 66

4.5 5.0

4.0–5.0 4.5–6.5

1

3.774

0.052

136 54 15

4.0 5.5 6.5

3.5–4.5 4.5–6.0 4.0–10.0

2

10.07

0.007

54 15

4.5 6.5

3.5–5.5 5.0–10.0

1

5.80

0.0161

49 20

4.0 7.0

3.5–5.5 5.0–10.0

1

13.54

0.0002

44 20 5

4.0 6.7 10.0

3.5–4.5 5.0–10.0 ⬎4.0

2

15.72

0.0004

the other nursing assessment, only 25% remained febrile. Thus fever was not such a major factor in the illness during the period of study. The selection criteria were such that all children had a history of fever and none had been ill for more than 72 hours. Nevertheless, if the children had been recruited at the very start of the illness (or before) a different pattern may have been found. The pattern of change in the CARIES over time shows that most children improved steadily from the time of enrollment. It may have been expected that they would get worse before getting better. Again this shows that children did not present to their physician immediately at the start of symptoms. The course of the illness followed the clinically predicted pattern, with complete recovery by the end of the study period. The boxplots (Fig. 2) show that by Day 14 the CARIFS scores were minimal. By this time 80% of the children had a score below 3. This means that, of the 18 items in the scale, none was reported as a major problem, and no more than three were minor problems. Absolute CARIFS scores will have limited significance to clinicians or others unfamiliar with the scale. A “normal” CARIES score has not been defined, and it is clear that there is wide variation in children’s scores at any one time. For this reason the more meaningful variable DayC1/4 was developed. This is a measure of duration of illness, clearly comprehensible as a number of days. The calculation of the DayC1/4 allows the use of survival analysis. This technique has been shown to discriminate groups with indicators of

complicated illness, and may well be the most powerful way to use this scale as an outcome measure in clinical trials. The performance of this measure was satisfactory across the wide age range and independent of viral etiology. The stability of results across sites also supports the generalizability of this measure. The CARIFS was studied in English-speaking Canadian subjects. Further evaluations of the measure in different cultures and with translated versions are necessary. Its clinical utility to help parents determine the level of their child’s illness, and guide their response, remains to be studied. Its value in assessing other acute illnesses, including lower respiratory infections also deserves assessment. In conclusion, the CARIFS has been validated as a disease severity measure for childhood ARI. As the first such measure it is likely to find many research applications including studies of ARI severity and its determinants, treatment studies and studies of the impact of this common illness on children and their families. The responsiveness of the CARIFS seen with clinical recovery is likely to make it a useful outcome measure in randomized clinical trials.

Acknowledgments We thank the children and parents who undertook this study as well as the staff of the physicians’ offices and the following study nurses: Kwong-Yim Ang, Leya Aronson, Lorelei Audas, Debra Lea, Melissa Logue, Rosalind Reyes,

B. Jacobs et al. / Journal of Clinical Epidemiology 53 (2000) 793–799

Sheila Robertson and Heather Samson. This study was partially funded by an unrestricted grant from HoffmannLa Roche Inc. Dr. Jacobs is supported by a grant from the Research Institute of the Hospital for Sick Children, Drs. Dick and Young are supported by a grant from the Hospital for Sick Children Foundation to the Paediatric Outcomes Research Team, and Dr. Davies is supported by the Alberta Heritage Foundation for Medical Research. The findings were presented to the Society for Pediatric Research, San Francisco, May 1999.

[12]

[13]

[14]

[15]

References [1] Carabin H, Gyorkos TW, Soto JC, Penrod J, Joseph L, Collet J. Estimation of direct and indirect costs because of common infections in toddlers attending day care centers. Pediatrics 1999;103(3):556–64. [2] Neuzil KM, Reed GW, Mitchel FE, Griffin MR. Influenza-associated morbidity and mortality in young and middle-aged women. JAMA 1999;281(10):901–7. [3] Vingilis E, Brown U, Koeppen R, Hennen B, Bass M, Peyton K, Downe J, Stewart M. Evaluation of a cold/flu self-care public education campaign. Health Education Res: Theory and Practice 1993;13: 33–46. [4] Fahey T, Stocks N, Thomas T. Systematic review of the treatment of upper respiratory infection. Archives of Disease in Childhood 1998; 79:225–30. [5] Stein RE, Gortmaker SL, Perrin EC, et al. Severity of illness: concepts and measurements. Lancet 1987;2(8574):1506–9. [6] Pollack MM, Patel KM, Ruttimann UE. PRISM III: an updated Pediatric Risk of Mortality score. Crit Care Med 1996;24(5):743–52. [7] McCarthy PL, Lembo RM, Baron MA, Fink HD, Cicchetti DV. Predictive value of abnormal physical examination findings in ill-appearing and well-appearing febrile children. Pediatrics 1985;76(2):167–71. [8] Stein RE, Jessop DJ. Functional status II(R): A measure of child health status. Medical Care 1990;28(11):1041–55. [9] Walker LS, Greene JW. The functional disability inventory: measuring a neglected dimension of child health status. J Ped Psych 1991; 16(1):39–58. [10] Eiser C. Children’s quality of life measures. Arch Dis Childhood 1997;77(4):350–4. [11] Hayden FG, Osterhaus AD, Treanor JJ, et al. Efficacy and safety of the neuraminidase inhibitor zanamivir in the treatment of influenza-

[16]

[17]

[18]

[19]

[20] [21] [22] [23] [24]

[25] [26] [27]

799

virus infections. GG167 Influenza Study Group. N Engl J Med 1997; 337(13):874–80. Hayden FG, Treanor JJ, Betts RF, Lobo M, Esinhart JD, Hussey EK. Safety and efficacy of the neuraminidase inhibitor GG167 in experimental human influenza. JAMA 1996;275(4):295–9. Barker WH, Borisute H, Cox C. A study of the impact of influenza on the functional status of frail older people. Arch Intern Med 1998; 158(6):645–50. Englund JA, Piedra PA, Ahn YM, Gilbert BE, Hiatt P. High-dose, short-duration ribavirin aerosol therapy compared with standard ribavirin therapy in children with suspected respiratory syncytial virus infection. J Ped 1994;125(4):635–41. Hall CB, Dolin R, Gala CL, et al. Children with influenza A infection: Treatment with rimantadine. Pediatrics 1987;80(2):275–82. Morley CJ, Thornton AJ, Cole TJ, Hewson PH, Fowler MA. Baby Check: A scoring system to grade the severity of acute systemic illness in babies under 6 months old. Archives of Disease in Childhood 1991;66(1):100–5. Landgraf JM, Abetz L, Ware JE. Child Health Questionnaire (CHQ) A User’s Manual. (1st edition). Boston, MA, USA: The Health Institute, New England Medical Center; 1996. Young NL, Yoshida KK, Williams JI, Bombardier C, Wright JG. The role of children in reporting their physical disability. Arch Phys Med Rehab 1995;76(10):913–8. Msall ME, DiGaudio K, Duffy LC, LaForest S, Braun S, Granger CV. WeeFIM. Normative sample of an instrument for tracking functional independence in children. Clin Pediatrics 1994;33(7):431–8. McCarthy PL, Sharpe MR, Spiesel SZ, et al. Observation scales to identify serious illness in febrile children. Pediatrics 1982;70(5):802–9. Kirshner B, Guyatt G. A methodological framework for assessing health indices. J Chron Dis 1985;38(1):27–36. Kline P. A Handbook of Test Construction. London, UK: Methuen, 1986. Cronbach LJ, Meehl PE. Construct validity in psychological tests. Psychological Bull 1955;52(4):281–302. Portney LG, Watkins MP. Chapter 26, Statistical Measures of Reliability. Foundation of Clinical Research Applications to Practice. Norwich, CT, USA: Appleton & Lange; 1993:509–16. Cohen J. Statistical Power Analysis for the Behavioral Sciences. Hillsdale, NJ, USA: Lawrence Erlbaum Associates, 1988. Kaplan EL, Meier P. Nonparametric observation from incomplete observations. J Am Stat Assoc 1958;53:457–81. Hall C. Respiratory syncytial virus. In: Feigin R, Cherry J, editors. Textbook of Pediatric Infectious Diseases. Philadelphia, PA, USA: WB Saunders, 1992.