The search strategy yielded 7536 records, of which 75 were eligible for inclusion based on the title and abstract. The remaining 7461 records were excluded for various reasons, e.g. the study did not concern asthma, the study concerned the treatment of adult asthma patients or the study evaluated quality-of-life questionnaires. We screened full-text versions of the 75 eligible articles. Five publications were ultimately included. Figure 1 shows a PRISMA flowchart of the process of identification and inclusion of studies for the current review.

Fig. 1

Flowchart of the process of identification and inclusion of studies.

The five studies selected for this review had a total of 1085 participants (range 35–468). The age of the participants varied from 5 to 71 years. A total of 806 children were included in the studies. Four studies had an observational design. The fifth study, by Rank et al., was a sub-study of a randomized controlled trial. Three studies were conducted in the United Kingdom (UK)21,22,23 and two in the United States of America (USA)24,25. Two studies also included adult patients (55.3%24 and 57.1% respectively)21; however, subgroup analyses were conducted. Rank et al. conducted their study in twenty24 primary care practices and Andrews et al. in eight23 primary care practices. Halterman et al. recruited patients from three urban clinics and three suburban practices (in the discussion referred to as ‘primary care offices’)25. The participants in the study by Thomas et al. were recruited from nurse-led asthma clinics at two general practices21. Juniper et al. included children from five primary care sites and one hospital clinic22. Asthma was defined as ‘clinical diagnoses of asthma’23, ‘documented evidence of asthma’21 or ‘physician-diagnosed’ asthma24. Halterman et al. included all children who had a diagnosis of asthma and had >2 asthma-related visits in the prior 12 months25. In the study of Juniper et al., children were eligible if they had well-established and physician-diagnosed asthma, with current symptoms of asthma (ACQ score > 0.5)22. The questionnaires (VAS and NEAPP) were filled in by the parents in the study by Halterman et al.25. In the study by Thomas et al.21, the questionnaires were administered by a clinician. In the other three studies the questionnaires were filled in by the children (sometimes together with their parents)22,23,24. Tables 3 and 5 show this information and the main results of the included studies. The studies included in this review were considered to be too heterogeneous (in terms of the questionnaires used, setting and patient categories) to pool the data. Table 1 shows the characteristics of the included studies.

Table 1 Characteristics of the included studies.Measurement characteristics of the questionnaires

The five studies presented results of the comparison of two or more questionnaires for measuring asthma control21,22,23,24,25. These studies gave comparisons of the following structured and unstructured questionnaires: Asthma Control Diary (ACD), ACT, ACQ, C-ACT, National Asthma Education and Prevention Program (NAEPP) criteria, Royal College of Physicians’ ‘three questions’ (RCP3Q), Visual Analog Scale (VAS) and the Asthma APGAR system (APGAR is an acronym for Activities, Persistent, triGGers, Asthma medications and Response to therapy). The ACD is evaluated in the study by Juniper et al. 22. Since the ACD has not been validated in children, we do not describe this tool.

Comparison of questionnairesSymptoms and domains evaluated

Each questionnaire deals with a different combination of symptoms and domains. Table 2 shows the domains covered.

Table 2 Symptoms and domains covered by questionnaires to assess asthma control.Characteristics of the questionnairesAsthma Control Questionnaire

The ACQ score is the mean of seven questions and ranges between 0 (totally controlled) and 6 (severely uncontrolled). The last question of the ACQ concerns the value of FEV1 and is filled in by a clinician. The ACQ has been validated for children aged 11 years and older26,27,28. For children aged 6–10 it must be administered by a trained interviewer10. Three shortened versions of the ACQ have been validated as well, but the complete ACQ has the strongest measurement properties28.

Asthma Control Test

The ACT is a self-administered questionnaire for children aged 12 years and up. It contains five items29.

Childhood-Asthma Control Test

The C-ACT is a seven-item questionnaire that has three questions for parents and four questions for children. It has been validated in children aged 4–11 years13.

Asthma APGAR system

The Asthma APGAR system has recently been developed for use in a primary care population30. It was developed to be answered by both parents and children together. After they have completed the assessment, an algorithm based on that data guides the clinicians in their treatment strategy for the patient. The score that corresponds to inadequate asthma control is derived from the National Asthma Education and Prevention Program (NAEPP) guidelines31.

NAEPP criteria

The NAEPP guideline-based criteria to assess asthma control are part of the National Asthma Education and Prevention Program in the USA32. This expert panel organization emphasizes the importance of monitoring asthma control. The level of severity is determined by assessing both impairment and risk. Asthma control is determined per age category (0–4 years, 5–11 years and ≥12 years).

Royal College of Physicians’ ‘three questions’

The Royal College of Physicians in the UK has developed a practical clinical tool containing three questions (RCP3Q) to assess asthma control in primary care33. It is the most commonly used tool in the UK. The questionnaire was designed by primary and secondary care physicians and patient organizations. It was designed to be completed by a health-care professional and contains three questions with the answer options ‘Yes’ or ‘No’, with a score of 1 for ‘yes’ and 0 for ‘no’. The total score ranges between 0 and 3. An RCP3Q score of 0 indicates good asthma control and a score of 2 or 3 indicates poor control34. The UK Quality Outcomes Framework (QOF)35 encourages the use of the RCP3Q in patients aged 8 and older. The performance of this questionnaire has been evaluated in adults; however, there is limited evidence for the use in children.

Visual Analog Scale

The VAS is an unstructured method for assessing asthma control in patients. To determine the VAS score, patients (or parents) have to indicate the severity of symptoms by placing an ‘X’ along a 100 mm line. A score of 0 (X on the left) indicates ‘no symptoms’ and a score of 100 (X on the right) indicates ‘very bad symptoms’. The VAS score is collapsed into quartiles (0–25, 26–50, 51–75, 76–100) corresponding to an ascending level of asthma severity. No cut-off value has been described.

Table 3 shows the characteristics of the questionnaires for assessing asthma control that are included in the current review. Table 4 gives information on the development and the purpose of the questionnaires.

Table 3 Characteristics of questionnaires for assessing asthma control.Table 4 Development and purpose of the questionnaires.Quality assessment

Information on content validity could be derived from two studies22,24. No information was found on internal consistency. Criterion validity was evaluated in three studies23,24,25. Two studies evaluated construct validity21,22. The aspects of agreement, reliability and responsiveness were evaluated in one study22. Floor and ceiling effects were rated as poor in three studies21,23,25. All studies scored ‘intermediate’ on interpretability21,22,23,24,25. Table 6 in the supplementary information file gives a summary of the assessment of the measurement properties of all the questionnaires included in this review.

Agreement in the determination of asthma control

The study by Andrews et al. determined the accuracy of the RCP3Q score in predicting asthma control as defined by the ACT or C-ACT threshold score of 1923. For children aged 5–11, a kappa value of 0.43 for poorly controlled asthma was found, indicating moderate agreement. For children aged 12–16, the kappa value was 0.33, demonstrating fair agreement. Overall, RCP3Q scores correlated moderately with C-ACT and ACT data (Spearman’s rho correlation coefficient was −0.52 and −0.49 respectively). Table 5 shows the agreement between the questionnaires included in the current review. The legend in Table 5 gives the interpretation of the kappa values and correlation values. The study showed that the RCP3Q’s sensitivity for detecting uncontrolled asthma as defined by ACT of C-ACT ranged from 43% to 60% and the specificity from 80% to 82%.

Table 5 Agreement of questionnaires in the included studies.

Juniper et al. evaluated the measurement properties of the ACQ by comparing the results with the RCP3Q in 35 children22. Pearson correlation coefficients between the ACQ and the RCP3Q were determined. The value for cross-sectional construct validity was 0.52 and the value for longitudinal construct validity was 0.81.

Thomas et al. determined the correlation between the RCP3Q and the ACQ in adults and children. Fifteen children completed seven follow-up visits (over 12 weeks). The cross-sectional correlation coefficient in children was 0.41, however this moderate correlation was not statistically significant. The longitudinal correlation for children was 0.61 (p < 0.001). This study was an exploratory analysis.

Rank et al. tested the effectiveness of the Asthma APGAR system by comparing this questionnaire with the ACT and C-ACT24. A total of 209 participants in the overall study population were aged under 18 years (=44.7%). For children aged 5–11 years, the C-ACT and Asthma APGAR instruments were in agreement in 85.8% of the cases (95% CI 78.5–91.4%). The kappa value of 0.716 (95% CI: 0.060–0.84) indicated substantial agreement. In the age group 12–18 years, the two questionnaires were in agreement 81.3% of the time (95% CI 71.0–89.1%). The kappa value of 0.625 (95% CI: 0.45–0.80) indicated substantial agreement as well.

Halterman et al. compared the assessment of asthma control using NAEPP criteria with a VAS. The NAEPP severity classification was used as a gold standard. Both questionnaires were filled in by the parents. A critical error was defined as ‘if parents reported the child’s symptoms in the lower 50th percentile of severity for VAS, whereas the child had moderate or severe persistent symptoms according to the NAEPP criteria’. The results showed that 41% of the parents made this so-called ‘critical error’.

Ability to detect uncontrolled asthmaACT

The screening accuracy of the ACT was evaluated by Nathan et al. 29. The agreement between the ACT and a specialist’s rating of asthma control was determined. A cut-off point of ≤19 resulted in a sensitivity of 69.2% and specificity of 76.2%, with an area under Receiver operating characteristic (ROC) curve of 0.727.

C-ACT

The validation study of the C-ACT by Liu et al.13 compared the C-ACT scores with a specialist’s assessment. It found that a cut-off point of 19 results in a sensitivity of 68% and a specificity of 74% for the detection of uncontrolled asthma.

ACQ

The study by Juniper et al. showed that in children whose asthma control changes between clinic visits, the questionnaire was able to detect the change (p < 0.026)22. However, no specific information can be extracted on the ability of the ACQ to detect uncontrolled asthma. The previous validation study in adults did not provide this information either26.

APGAR

No detailed information about the ability to detect uncontrolled asthma of the Asthma APGAR system can be found in the study of Rank at al. 24. The authors did identify an ‘actionable item’ in more than 75% of the children with poor asthma control.

NAEPP

The NAEPP and ACQ criteria were compared in a study of 373 adolescents with asthma. The NAEPP identified 84.6% of the cases of uncontrolled asthma and the ACQ 64.6% of the cases11.

RCP3Q

To analyze the performance of the RCP3Q in detecting uncontrolled asthma, it was compared to C-ACT or ACT, whereby a score of 19 was defined as uncontrolled asthma23. Using a threshold RCP3Q score of ≥2 to predict uncontrolled asthma resulted in a sensitivity of 0.60 and a specificity of 0.82 for the age group 5–11 years, and a sensitivity of 0.51 and specificity of 0.81 for the age group 12–16 years.

VAS

Halterman compared unstructured assessments of asthma severity (VAS) with the NAEPP classification of severity. Of the children with moderate to severe symptoms according to the NAEPP classification, 41% of the parents rated their children in the lowest two quartiles of the VAS. The unstructured method seems to underestimate the severity level of asthma.

Ability to predict future events

None of the questionnaires included in this review provides information on the risk of future events as an outcome. Previous studies have identified several risk factors for asthma attacks or poor asthma-related outcomes36,37,38. These risk factors include e.g. younger age, history of hospitalization or an emergency department (ED) visit in the previous year, three days’ use of oral corticosteroids in the previous three months, a lower FEV1/FVC ratio37, higher FeNO levels and a recent history of asthma attacks38. A recent systematic review concluded that a previous asthma attack was the most strongly predictive factor36.