Published ahead of print on June 7, 2004, doi:10.1164/rccm.200307-1025OC
© 2004 American Thoracic Society
Peak Flow Monitoring for Guided Self-management in Childhood AsthmaA Randomized Controlled TrialDepartment of Child Health and Institute for Lung Health, University of Leicester, Leicester, United Kingdom Correspondence and requests for reprints should be addressed to Mike Silverman, M.D., F.R.C.P.C.H., Department of Child Health, Robert Kilpatrick Clinical Sciences Building, Leicester Royal Infirmary, P.O. Box 65, Leicester LE2 7LX, UK. E-mail: ms70{at}le.ac.uk
We asked whether the addition of PEF recordings to a symptom-based self-management plan improved outcome in school children with asthma. In an open-randomized, parallel-group, controlled trial, we studied children aged 7–14 years with moderate asthma. After a 4-week run-in, 90 children were randomized to receive either PEF plus symptom-based management or symptom-based management alone for 12 weeks. Thresholds for action based on PEF were 70% of best (for increasing inhaled steroids) and 50% of best (for commencing prednisolone). Children were asked to perform twice-daily spirometry at home (using an electronic recording spirometer that revealed only PEF to the study group alone) and to record a symptom diary. The mean daily symptom score was the main outcome. There were no differences between groups in mean symptom score or in spirometric lung function, PEF, quality of life score, or reported use of health services over 12 weeks. During acute episodes, children responded to changes in symptoms by increasing their inhaled steroids at a mean value of PEF of greater than 70% of best so that overall PEF did not contribute to this important self-management decision. Knowledge of PEF did not enhance self-management even during acute exacerbations.
Key Words: health service use • quality of life • spirometry Guided self-management is a cornerstone of asthma care for all age groups (1). When compared with "usual care" in a recent systematic review, self-management training backed by a written action plan reduced hospitalizations, unplanned doctor visits, emergency room attendance, work absence, and nocturnal asthma in adults (2). The introduction of self-management training programs reduces direct and indirect health-related costs (3). Studies involving children are few in number but are consistent with the adult evidence (4–6). A systematic review noted that self-management educational interventions improve many (but not all) outcomes in children but that the individual components of such programs should be investigated (7). Self-management decisions are generally based on symptoms. The role of PEF monitoring in the self-management process is unclear. Most studies show no difference in outcomes in either the hospital (8) or the community (9, 10), although one concluded that PEF-based plans were more effective than plans based on symptoms alone (11). For acute exacerbations, both PEF and symptoms have been shown to be effective in adults (12). A recent review found no overall evidence of a difference in outcome for self-management base on PEF compared with symptoms in adults (13). Limited data are available for children, and the conclusions from research in adults may not be applicable. First, the pattern of asthma seems to differ in childhood, with a greater episodic element particularly in young children, such that interval symptoms may be few and peak-flow variability minimal (14). Second, PEF measurement may be unreliable under conditions of severe airway obstruction (15). Nevertheless, PEF could be valuable as an early marker of deterioration in acute episodes. Again, however, there is controversy. Two studies claimed benefits from routine PEF monitoring, although the numbers of children were limited (11, 16), whereas another showed no benefit (17). This study was set up to test the hypothesis that the health of school children with asthma would benefit from the addition of routine PEF monitoring to symptom-based guided self-management. In addition, we sought to compare the relative sensitivity of PEF and spirometric measures of lung function during acute episodes at home.
Study Design This was an open, prospective, randomized, parallel-group trial conducted in children's homes, with recruitment in primary care and secondary care. The main outcome measure was the mean daily symptom score. Secondary measures included PEF, FEV1, quality of life (QoL), and reported use of health services. Approval was given by the Leicestershire Research Ethics Committee; signed and informed consent was obtained.
Study Protocol At recruitment, demographic data were collected, and the QoL questionnaires were completed (19, 20). Children performed twice daily spirometry (Vitalograph DSS; Vitalograph, Buckinghamshire, UK) to American Thoracic Society (1987) criteria (21). Spirometric data were electronically stored and blind to the participants. Children were randomly allocated into one of two self-management groups: group PF based on symptoms plus PEF or group S based on symptoms alone. The spirometers of those children randomized to the PF group were reprogrammed so that the PEF value for any maneuver (but not other spirometric values) was visible to them at any time. The S group did not have access to any lung function results throughout the study. The child and the main caregiver were taught self-management at a training session, which also included training in spirometry and symptom recording and which lasted 30–90 minutes according to need. A printed plan incorporating the child's own medication regime was color coded: green, PEF more than 70%, few symptoms (carry on as usual); yellow, PEF 50–70% after ß2 agonist (double-inhaled corticosteroid as well as taking additional ß2-agonist therapy); and red, PEF less than 50% after taking additional inhaled ß2 agonist, severe symptoms (commence oral prednisolone and/or seek medical help). The PEF levels for action were based on the child's best previous PEF. A written symptom diary was completed each morning, and spirometry was performed twice daily. Children were visited approximately monthly to download spirometric data, exchange the written diary for a new one, and complete QoL and Use of Health Services Questionnaires. Spirometric performance was checked, questions answered, and treatment changes incorporated into the plan. At each 4-week visit, the following were recorded: emergency general practitioner attendance, hospital visits or admissions, treatment changes, and school absence. An acute episode was behaviorally defined as increased inhaled corticosteroid for more than 2 consecutive days or the use of oral prednisolone.
Analysis
Trial Profile Attrition reduced 511 children to 89 eligible children who completed the randomized trial (Figure 1). Incomplete QoL data were available for two. Full data for analysis were available on 87 children. The 27 children who were rejected at the end of the run-in period were slightly younger than the randomized group and produced data of lower validity during the run-in period (Table 1). The chance occurrence of more boys in the peak flow (PF) group was adjusted for in the analysis. The two treatment groups were otherwise closely matched. Their mean daily inhaled corticosteroid dose at recruitment was S 317 (SEM, 26.1) and PF 323 (SEM, 29.7) mcg per day (beclomethasone equivalent).
Symptom Scores Compliance with written diary data deteriorated over time in both groups from 90% in Month 1 to 79% in Month 3. The minimum monthly compliance with the written diary was over 74% for the duration of the study, with no difference between groups. Using repeated-measures analysis of variance, there was no significant difference between the PF and S groups in the mean daily symptom score during the trial period (Figure 2), the mean daily symptom score for symptomatic days (Figure 3), or the proportion of symptom-free days (overall 70.1% and 59.6%, respectively).
Lung Function Tests The mean of the "best" values obtained during the run-in was 95% of reference values for FEV1 and 105% for PEF (22). During the study, the overall mean FEV1 was 87% of the reference value, an indication of the overall severity of children's asthma. No significant differences were found between the two groups in any lung function parameter for any study period (data for PEF shown in Figure 4). For the PF and S groups, the means (SEM) of all values of FEV1 (expressed as a proportion of the best value) during the trial period were 87.3% (0.20%) and 86.9% (0.23%), respectively, and for PEF were 83.4% (0.21%) and 80.6% (0.26%), respectively.
QoL Scores There were no between-group differences in QoL score for parents or children during the trial. This was true for overall QoL scores and for the separate activity, symptom, and emotion domains within the Pediatric Asthma Quality of Life Questionnaire (PAQLQ). In the PF group, an unexplained increase in overall mean QoL score of 0.73 was seen during the run-in period. This was greater than the 0.5 mean change, which the developers report as clinically significant. This clinically significant change was seen in all domains when considered separately: emotional domain (mean change, 0.61), activity domain (mean change, 1.1), and symptom domain (mean change, 0.7) and matched an (insignificant) improvement in lung function over the same run-in period that was not seen in the control (S) group. Caregivers of children in the PF group recorded a higher mean QoL score throughout the run-in period and the trial in both domains. None of the between-group differences was statistically significant.
Measures of Morbidity
Episodes Forty-two children increased inhaled preventer therapy in response to 59 episodes, irrespective of the level of PEF (Figure 5). The distribution of children with episodes between the two groups was as follows: S 26 (58%) and PF 16 (36%) (p = 0.06). There was a marked decline in PEF in both groups approximately 1 day before increasing treatment (Figure 5). Children began taking more reliever medication on average 1 day before increasing preventer, although there was sporadic extra reliever use 4–6 days earlier. The PF group doubled inhaled steroids before mean PEF had declined to the level for action (70%). After commencing treatment, there was an improvement in lung function, and the group mean symptom score declined rapidly with the former taking approximately 5 days to recover and the latter 2 days. Group S seems to have responded at a slightly (not significantly) lower symptom score. They used less bronchodilator, and their response in terms of lung function and symptoms was slower, taking 9 and 7 days, respectively, although they continued to take increased preventer therapy for approximately 2 weeks longer. At the onset of an episode (Day –1 or Day 0), group mean PEF was again higher than the threshold of 70%. No child reached the threshold of 70% on Day –1 or Day 0. Group S showed very little change in FEV1 even when PEF fell. In the PF group, FEV1 followed the pattern of PEF more closely, but on Day –1 when mean PEF had fallen to 75%, the mean FEV1 had only fallen to 84% of best. Values of FEF25–75% showed wide scatter and showed an even weaker relationship to exacerbations.
Ten children took oral steroids during the trial period for 14 episodes. Two children in the PF group each recorded taking three courses of oral prednisolone. There were too few such episodes for meaningful statistical analysis, but overall mean PEF on the day on which oral corticosteroids commenced was 74% of best.
This study found no differences in daily symptom score, lung function, QoL, or health service use between those children who used a self-management plan based on symptoms alone (S) and those who used peak flow plus symptoms (PF). The number of acute episodes did not differ significantly between the groups. Children responded to symptoms by stepping up their treatment before their PEF had fallen to a value of 70% of their previous best, the threshold that we chose for action, showing that overall, symptoms were more sensitive than PEF. Our results are consistent with the systematic review of data from clinical trials in adults (13). PEF measurement is widely accepted as a means of monitoring asthma, but in terms of guided self-management, the potential benefits may be limited to decision making around acute episodes, specifically in patients with severe disease (12, 16). In children, self-management studies have demonstrated varying degrees of success, and only three studies directly compared symptom and peak-flow–based management, with inconclusive results (11, 16), at best during acute exacerbations (17). The question of the role of PEF is an important one because currently it is the only cheap, widely available means of gaining lung function information. Because this trial produced a negative result, it is important to address its potential limitations. (1) After randomization, the trial could not be blinded because assignment determined management. Moreover clinicians (general practitioners and hospital consultants) responsible for patient care may have become aware of group assignment and were aware of the instructions and thresholds incorporated into the plan. If anything, the open nature of the trial might have been expected to bias the results in favor of the PF group. (2) By chance, significantly more boys were randomized to the PF group. Stratification by sex would have prevented this, but analysis took into account the maldistribution. (3) Both groups received management plans so that this study did not answer the question "is guided self-management effective?" as this had already been demonstrated by a number of studies (2, 4, 7, 8, 16, 23). (4) This study may not have had sufficient power to detect a difference in symptom scores between the groups. The mean (back transformed) symptom scores during the trial for each group were 0.59 for group PF and 0.82 for the S group, with a possible range of 0–9. A priori predictions suggest that the trial is underpowered. In retrospect, the groups were so similar that for any difference during any part of the trial to reach statistical significance, approximately 800 children would have to be studied. (5) The self-management education that we provided was thorough and might thereby have enhanced the skills of the symptoms-only group, reducing any potential benefit of PEF measurement. During asthma management trials, morbidity often improves with time (24–26) independently of therapy. This was not seen in this study after the run-in period, perhaps because the investigators were not involved in clinical management, and any health problems were dealt with by the primary care practitioner or hospital consultant. We did not recruit patients at a low point during or immediately after a hospitalization (13) or try to optimize treatment at the start (16). Most of the children (94%) were stable subjects with asthma recruited in primary-care settings and with a relatively low attack rate. Nevertheless, 49% had a significant acute episode during the 12-week trial period. For children who experience more frequent or more severe attacks, paradoxically, this kind of in-depth study demanding large numbers of measurements may be more inappropriate (27). A simpler design with fewer, more specific outcomes, including the need for emergency treatments, would then be more appropriate. The trial period of 12 weeks was considered long enough for half the children to experience an episode but short enough to promote compliance. Trial demands were high and far in excess of any current expectations of patients in a clinical setting. Children were expected to perform lung function and complete diaries daily and to make time for home visits every 4 weeks. This could have contributed to poor compliance and possibly the outcome of the study. Over a 4-week period (equivalent to the run-in period), 77% of children provided valid PEF data. This figure is likely to be far less in clinical practice where poor control associated with poor compliance is often the reason for requesting this type of monitoring. Compliance with PEF recording in children declined rapidly over the period of 12 weeks, even in this relatively cooperative group (28). Others have found PEF diaries to be unreliable in children (29). Nevertheless, there was no trend toward a relevant difference between the two groups in any outcome measure. The group mean compliance with diary recording during the 4-week run-in period in the nonrandomized group was only 56% (Table 1), questioning the value of this sort of record in clinical trials, even in volunteer families. In spite of this, clinical trials in asthma often use outcome measures recorded in written diaries (25, 30), including PEF (8) and symptoms (6). Information recorded in diaries may be important (31), and the discriminatory properties of diaries have been demonstrated in children (32). However, information obtained is valuable only if it is reliable and ideally verifiable. We were able to use QoL questionnaire data as a check on symptoms over the final week of each month of the trial. Neither parent nor child QoL records discriminated between the two groups, lending support to the negative outcome of diary data. The participants performed the tests at home, and maneuvers could not be assessed to see whether they were acceptable according to full American Thoracic Society criteria (18), although the spirometer recorded within-session validity (the sum of FVC + FEV1 was within 5% for two best blows), providing one objective assessment of performance. As was the case for the symptom diary, reduced compliance toward the end of the study reduced the amount of data available for analysis and therefore trial power. Pelkonen and colleagues (33) demonstrated higher compliance among newly diagnosed subjects with asthma who were probably enthusiastic and performed home spirometry for shorter periods of time. If children are expected to respond to changes in lung function, the choice of thresholds is critical. This study used thresholds of 70% and 50% of the recent best PEF, recorded during the 4-week run-in period. In both groups, the group mean PEF fell to below 80% as the children increasing their inhaled corticosteroids (Figure 5), but not as low as 70%. Higher thresholds, such as those suggested by Charlton and colleagues (10) may be more appropriate in this population. However, the risks of overtreatment highlighted by Douma and colleagues (34) are emphasized by the fact that the overall mean PEF for the children in this study was only 83% of best. This suggests that a number of children, if they had responded to PEF alone, would have used additional therapy at a time when they were free of significant symptoms. All of the evidence suggests that overall, small changes in lung function (both PEF and FEV1) accompany clinically significant increases in symptoms. This simply reinforces the evidence that families responded mainly to symptoms at the time of an exacerbation. It is likely that symptoms represent the sum of a host of asthma-related sensations and perceptions over the course of the night and day and during rest and exercise, whereas lung function represents one static instant. Moreover, lung function may be affected by recent bronchodilator use and may be less reliable under unsupervised conditions. The choice of the highest PEF from a technically acceptable maneuver, performed unsupervised during the run-in to determine the "best" or optimum value to calculate thresholds for management could be unreliable. The group mean value of PEF during the trial was only approximately 83% of best value obtained during the run-in period in spite of the apparently well-controlled nature of the children's asthma. The "best" value should have been accurate even if children had used a bronchodilator before performing the maneuver because the machine rejected errors such as those produced by coughing or spitting. The low value of lung function during the trial suggests that in picking the "best" value as the reference value, one may place too much reliance on a single value. This could have distorted the threshold for management in individual subjects. In defense, the mean value of the best PEF in the study groups was 105% of the reference value (21). The use of a single maximum forced expiratory maneuver to derive both PEF and FEV1 is unlikely to have produced bias (35). The median values were similar for the S group (107%) and the PF group (102%); therefore, this factor did not introduce any bias into the study. We found no evidence that FEV1 would have been more sensitive. Studies comparing the relative sensitivity of FEV1 and PEF often compare values obtained from different equipment or maneuvers (36, 37). Values for PEF and FEV1 were recorded during each spirometric maneuver, and the resulting values from the single maneuvers were directly compared. The unsupervised nature of these maneuvers may be responsible for the increased sensitivity of PEF when compared with FEV1. Reduced effort, incomplete or slow inhalation, or pause at total lung capacity would be expected to impact differently on PEF and FEV1 (38, 39). The QoL questionnaire was completed at each visit so that its reliable recording was assured. The PAQLQ has been extensively validated (19) and used in research protocols (40). Children completed the questionnaire in the presence of D.W., a trained interviewer. The fact that D.W. was involved with recruitment, training and collecting data from these children were potential sources of bias. To reduce the potential for bias, no prompting was permitted when using the interviewer-administered version of the questionnaire. Because results obtained as a result of the interview were numeric, analysis was unlikely to be influenced by this factor. The use of health services and school absence were recorded retrospectively at each visit for the previous 4 weeks. This period was relatively short and the data therefore likely to be accurately recalled. A direct comparison of adverse events between group for the run-in period and mean monthly events during the trial showed no difference. This trial fills a perceived gap in the evidence base for guided self-management by children (41). It does not support the hypothesis that the routine incorporation of PEF monitoring into guided self-management protocols for school children with asthma improves the outcome. This relatively stable group of school children with asthma made major treatment changes based on symptoms before their PEF had fallen to 70% of its best value, confirming studies in adults (10, 30). FEV1 was less sensitive. Selective use of PEF in self-management for children with more severe asthma may be more appropriate, but its efficacy has yet to be demonstrated. Raising the threshold to 80% of best PEF would lead to overtreatment of many children whose clinical condition was stable and who were relatively symptom free. Until it is clear whether the target for therapy is symptoms, lung function, airway inflammation, bronchial responsiveness, or long-term airway remodeling, it will not be easy to provide clear advice on guided self-management to families of children with asthma.
Dr. Nicola Wilson was involved in the preliminary discussions, which led to the project. Professor John Thomson provided expert advice on statistical analysis. The authors thank local general practitioners for referring patients for this study and thank the children and their parents for their cooperation.
Supported by grants from the United Kingdom National Asthma Campaign and from Glaxo SmithKline, United Kingdom. This article has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Conflict of Interest Statement: D.W. has been reimbursed by GlaxoSmithKline for attending conferences; M.S. has been reimbursed by Merck Sharpe and Dohme, AstraZeneca, and GlaxoSmithKline for attending conferences and for lecturing in the United Kingdom and overseas and received approximately £300,000 for various research projects from GlaxoSmithKline and £30,000 from Merck Sharpe and Dohme and served on the Merck Sharpe and Dohme United Kingdom Pediatric Advisory Board as chairman. Received in original form July 25, 2003; accepted in final form June 3, 2004
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
