Table 3.
All included paper findings as reported and the decisions underpinning the harvest plots
| Citation design, sample group/size and risk of bias score | Outcome categories, FU | Reported outcomes-values for intervention (I)/control (C) aindicates the primary outcome (if stated) | Researcher’s interpretation for the harvest plot |
|---|---|---|---|
| Agrawal30 n = 60 children FU: 4 m | Clinical-unscheduled care, 4 m | Compared to controls, children in the intervention group had: | Illustrated as a consistent significant positive effect |
| Fewer acute asthma events: I: 0.50 (SD 0.71) vs. 1.0 (SD 0.61); p = 0.02 | |||
| Overall risk of bias: Unclear | Clinical-asthma control, 4 m | Compared to controls, children in the intervention group had: | Illustrated as a consistent significant positive effect |
| Improved symptom score: (from the symptom diary) I: 21.9 (SD 14.4) vs. C: 33.7 (SD 10.9); p = 0.0006 | |||
| Fewer nocturnal awakenings: I: 1.75 nights/month (SD 1.30) vs. C: 3.25 (SD 1.20); p = 0.001 | |||
| Reduced school absenteeism: I: 1.5 days/month (SD 1.4) vs. C: 2.54 (SD 1.79); p = 0.015 | |||
| Process | Not assessed | – | |
| Behavioural | Not assessed | – | |
| Behera31 CCT n = 523 adults | Clinical-unscheduled care, 1 yr | A reduction in hospital admissions is illustrated graphically (the authors state that there was a significant decrease in hospital admissions in the intervention group at FU compared to the control group) | Illustrated as a consistent significant positive effect |
| FU: 2 wks, 6 m, 1 yr Overall risk of bias: high | Clinical-asthma control, 2 wks, 6 m, 1 yr | Symptom scores decreased in both groups | Illustrated as a consistent significant positive effect |
| I: Baseline: 18.14 (SD 41.23) vs. FU 1 yr: 12.61 (SD 28.66) | |||
| C: Baseline: 18.76 (SD 42.64) vs. FU 1 yr: 10.69 (SD 24.30) | |||
| Logistic regression: compared to the control group, more intervention group patients showed a significant improvement in symptom scores at 2 w, 6 m and 1 yr (p < 0.001) | |||
| Process, 2 wks, 6 m, 1 yr | Knowledge scores increased significantly in the intervention group and fell in the control group; | Illustrated as a consistent significant positive effect | |
| I: Baseline: 13.04 (SD 4.06) vs. FU 1 yr: 28.13 (SD 15.70); p = < 0.001 C: Baseline: 11.44 (SD 4.0) vs. FU 1 yr: 9.47 (SD 2.89); p = < 0.001 | |||
| Logistic regression: Compared to the control group, more intervention group patients showed a significant increase in knowledge scores at 2 wks, 6 m and 1 yr (p < 0.001) | |||
| Behavioural, 2 wks, 6 m, 1 yr | Reported self-care in acute attacks showed no change in attitudes in either group, but significantly more patients in the intervention group adopted the recommended position (sitting, leaning forward) and practiced breathing exercises during an acute attack as compared to control patients | Illustrated as a significant positive effect but hatched to show inconsistency | |
| Ghosh32 n = 276 adult, adolescent, children/parent | Clinical-unscheduled care, 1 yr (assessed by diary in months 4, 8 and 12) | Fewer total number of ED visits, but no between group difference in proportion with ED visit | Illustrated as positive but hatched to indicate inconsistency |
| Number of ED visits in the 3-month diary: I: 11.6 (SD 16.2) vs. C: 21.8 (SD 25.0); p = 0.002 | |||
| Proportion with ED visits in the 3-month diary: I: 42.9 vs. 50.0% (p = 0.117) | |||
| Number and duration of hospitalisations were both significantly reduced | |||
| Hospital days in the three diary months: I: 5.8 (SD 10.7) vs. 12.5 (SD 19.8); p = 0.016 | |||
| Proportion hospitalised in the three diary months: I: 27.1 vs. C: 36.8%; p = 0.043 | |||
| FU: 4 m, 8 m, 1 yr Overall risk of bias: high | Clinical-asthma control, 1 yr (assessed by diary in months 4, 8 and 12) | Fewer productive days lost in the intervention group during the three diary months | Illustrated as a consistent significant positive effect |
| Day lost: 17.6 (SD = 24.2)/34.1 (SD = 38.8); p = 0.003 | |||
| PEFR was significantly improved in the intervention group relative to the control group; | |||
| Mean PEFR from diary cards I: 332 (SD 50.78) vs. 290 (SD 77.69); p = < 0.001 | |||
| Process | Not assessed | – | |
| Behavioural | Not assessed | – | |
| Shanmugam33 CCT n = 66 | Clinical-unscheduled care | Not assessed | – |
| FU: 29 days Overall risk of bias: unclear | Clinical-asthma control, 29th day | Asthma control improved in the intervention group compared to the control group | Illustrated as a consistent significant positive effect |
| Mean ACT score for each question was greater in the intervention group at FU: p < 0.05 | |||
| (Overall mean ACT scores are not reported) | |||
| Lung function showed a greater increase in the intervention group compared with control | |||
| PEFR (L/min): Baseline: I: 282 (SD 95) vs. C: 265 (SD 93); FU: I: 336 (SD 88) vs. C: 268 (SD 85); p = < 0.05 | |||
| Process | Not assessed | – | |
| Behavioural | Not assessed | – | |
| Griffiths34 n = 44 practices/324 – (South Asians I: 95 C: 69 n = 164), adults, adolescents, children | Clinical-unscheduled care, 1 yr | [Note: these data are an a priori sub-group analysis] | Illustrated as a consistent no effect |
| aTime to first unscheduled care effect on South Asians was not significant between intervention and control; South Asians HR 0.72, 0.48 to 1.09 | |||
| aProportion attending unscheduled asthma care: no between group differences in whole population. No data for South Asian sub-group, but authors state that ‘intervention effect was non-significant for other sub-group analysis’ | |||
| FU: 2 m, 9 m, 1 yr Overall risk of bias: low | Clinical-asthma control, 2 m, 1 yr | [Note: these data are an a priori sub-group analysis] | Illustrated as a consistent no effect |
| Symptoms: no between group differences in whole population. No data for South Asian sub-group, but authors state that ‘intervention effect was not significant for other sub-group analysis’ | |||
| Process | Not assessed | – | |
| Behavioural, 2 m, 1 yr | [Note: these data are an a priori sub-group analysis] | Illustrated as a consistent no effect | |
| Self-management behaviour: no between group differences in whole population. No data for South Asian sub-group, but authors state that ‘intervention effect was not significant for other sub-group analysis’ | |||
| Griffiths35 n = 84 practices/375 elders, adults, adolescents, children, primary and secondary care clinicians | Clinical-unscheduled careI: 171 days/C: 189 daysI: 72 days/ C: 339 days1 yr | Unscheduled care: there was no between group difference in healthcare use | Illustrated as a consistent no effect |
| aTime to first unscheduled contact FU: HR = 1.19 (0.92 to 1.53); p = 0.185 | |||
| Proportion without unscheduled care FU: OR = 0.72 (0.45 to 1.16); p = 0.175 | |||
| Time to first unscheduled primary care contact FU: HR = 1.20, 0.92 to 1.57 p = 0.177 | |||
| Time to first routine review in primary care FU: HR = 2.22, 1.67 to 2.95 p = < 0.001 | |||
| Corticosteroid prescriptions: There was no between group difference in steroid prescriptions | |||
| Steroids FU: I: 1.16 vs. 0.98 Adjusted incidence rate ratio: 1.14 (0.87–1.49) | |||
| FU: 3 m, 1 yr Overall risk of bias: low | Clinical-asthma control, 3 m, 1 yr | Asthma control: there was no between group difference in symptom score | Illustrated as a consistent no effect |
| Process, 3 m, 1 yr | Symptom score FU 1 yr: 9.9 (SD 5.0) vs. C: 10.1 (SD 4.2) AHR: −0.04 (−1.16 to 1.09); p = 0.949 | ||
| Self-efficacy was improved at 3 m but not at 1 yr follow-up; | Illustrated as a consistent no effect. Another bar plotted to illustrate the 3 m finding—as a consistent significant positive effect | ||
| At 3 months: I: 6.7 (2.1) vs. C: 6.3 (1.9) AHR: 0.44 (0.05 to 0.82); p = 0.027 | |||
| At 12 months: I: 6.4 (1.8) vs. C: 6.3 (1.6) AHR: 0.25 (−0.13 to 0.63); p = 0.188 | |||
| Behavioural | Not assessed | – | |
| Moudgil36 n = 689 (White Europeans 345, Indian subcontinent 344); adults, adolescents, children | Clinical-unscheduled care, not stated ISC: n = 294 (I: 151 C: 143) | [Note: these data are an a priori sub-group analysis] | Illustrated as a consistent no effect |
| Number of asthma events/episodes for South Asians: no between group differences | |||
| aProportion with an admission. I: 5.3 vs. C: 6.3% OR 0.83 (0.28 to 2.44); p = 0.9081 | |||
| Proportion with an A&E attendance. I: 1.4 vs. C: 4.0% OR 2.92 (0.52 to 21.2); p = 0.3184 | |||
| Proportion with out-of-hours primary care. I: 2.8 vs. C: 2.6% FU: OR 0.95 (0.19 to 4.60); p = 1 | |||
| Proportion with a GP consultation. I: 55 9 vs. 50.3%. OR 0.80 (0.49 to –1.30); p = 0.3971 | |||
| Proportion with a steroid course. I: 20.3 vs. 19.9%. OR 0.97 (0.53 to 1.79); p = 1 | |||
| FU: 4 m, 8 m, 1 yr Overall risk of bias: High | Clinical-asthma control, 1 yr ISC n = 280 | [Note: these data are an a priori sub-group analysis] | Illustrated as a consistent significant positive effect |
| Quality of life in South Asians was significantly better in the intervention group | |||
| Change in AQLQ FU: I: 0.11 vs. −0.15. Between group mean difference 0.26 (0.17–0.36); p = < 0.001 | |||
| Process | Not assessed | – | |
| Behavioural | Not assessed | – | |
| Poureslami37 n = 92 (47 Chinese, 45 Punjabi); Adults | Process, 3 m, 6 mPunjabi n = 43 | [Note: these data are an a priori sub-group analysis] | Insufficient data |
| aKnowledge: no comparison data for intervention and control groups | |||
| FU: 3 m, 6 m; 1 telephone survey interview Overall risk of bias: unclear | Behavioural, 3 m, 6 m | [Note: these data are an a priori sub-group analysis] | Insufficient data |
| Punjabi n = 43 | Understanding physician instructions; on amedication and proper inhaler use skills: no comparison data for intervention and control groups | ||
| Blixen42 n = 28, Adults | Clinical-unscheduled care, 3 m, 6 m | Healthcare use: no data provided, though stated as no significant between group differences | Illustrated as a consistent no effect |
| Clinical-asthma control, 3 m, 6 m | Quality of life: There was no significant between group differences | Illustrated as a consistent no effect | |
| Overall AQOL score. FU 6 m: I: 4.59 (SD 1.48) vs. C: 4.43 (SD 1.52); p = 0.12 | |||
| FU: 3 m, 6 m Overall risk of bias: high | Process | Not assessed | – |
| Behavioural, 3 m, 6 m | Self-management behaviours: no data, though stated as no-significant between group differences | Illustrated as a consistent no effect | |
| Fisher38 n = 249 Adolescents, children, parents | Clinical-unscheduled care, Quarterly for 3 yrs | aAcute care: no data given (results illustrated graphically), though authors stated no significant between group differences in acute care (hospitalisations and ED attendances p = 0.35) | Illustrated as a consistent no effect |
| Clinical-asthma control | Not assessed | – | |
| FU: 3, 6, 9, 12, 16, 20, 24, 28, 32, 36 m Overall risk of bias: unclear | Process | Not assessed | – |
| Behavioural, Every quarterly until 3 yrs | aAsthma management: no significant between group differences in the non-validated assessment of parent’s reported attitude about asthma and asthma management | Illustrated as a consistent no effect | |
| Attitudes about asthma FU: I: 2.34 vs. C: 2.24 (p = 0.35) | |||
| Appropriate thresholds for seeking help Baseline: I: 30 vs. C: 47%; FU: I: 51 vs. C: 53% p = 0.77 | |||
| Fisher46 n = 191/parents, coaches | Clinical-unscheduled care, 1 yr, 2 yr | aHospitalisation Compared to controls, the intervention group had fewer hospitalisations; | Illustrated as a consistent significant positive effect |
| Hospitalised at least once FU I: n = 35/96 (36.5%), 55 vs. C: 55/93 (59.1%); 95% CI (0.11–0.34); p = .002 | |||
| FU: 6, 12, 18, 24 m Overall risk of bias: low | Clinical-asthma control | Not assessed | – |
| Process | Not assessed | – | |
| Behavioural | Not assessed | – | |
| Ford43 n = 241 (African American = 163, Caucasian = 78) | Clinical-unscheduled care, 4 m, 8 m, 1 yr | aED visits No impact [Note: these data are an a priori sub-group analysis] | Illustrated as a consistent no effect |
| ED visits/year I: Baseline: 5.0 (SD 3.6) vs. FU 2.7 (SD 3.3); C: Baseline: 6.7 (SD 8.4) vs. FU: 4.8 (SD 6.8) | |||
| No between group comparisons reported | |||
| Clinical-asthma control, 4 m, 8 m, 1 yr | Limited days of activity No impact [Note: these data are an a priori sub-group analysis] | ||
| Days/person: I: Baseline: 20.6 (SD 25.4); FU: 18.7 (SD 36.8) C: Baseline: 27.8 (SD 33.4); FU: 27.9 (SD 55.7), no between group differences reported | |||
| FU: 4 m, 8 m, 1 yr Overall risk of bias: high | Process, 1 yr | aKnowledge and beliefs: no effect [Note: these data are an a priori sub-group analysis] | Illustrated consistently no effect |
| Mean scores I: Baseline: 14.1 (SD 2.9); FU: 14.6 (SD 3.2) C: Baseline: 14.3 (SD 2.3); FU: 14.7 (SD 2.3) | |||
| No between group differences reported | |||
| Behavioural | Not assessed | – | |
| Keslo39 n = 52 adults | Clinical-unscheduled care, 1 yr | Unscheduled care: compared to controls, the intervention reduced ED visits but not hospitalisations | Illustrated as a significant positive effect but hatched to show inconsistency |
| aChange in ED visits Baseline: I: 4.4 (SD 2.7) vs. C: 3.4 (SD 2.6); FU: I: 2.6 (SD 2.6 vs. C: 3.5 (SD 2.7) Between group difference p = < 0.01 | |||
| Change in hospitalisations Baseline: I: 1.3 (SD 1.3) vs. C: 1.0 (SD 1.2); FU: I: 0.5 (SD 0.8) vs. C: 0.5 (SD 0.9) Between group difference p = 0.37 | |||
| FU: 1 yr, telephone every 2 wks to every 6 m Overall risk of bias: unclear | Clinical-asthma control | Not assessed | – |
| Process, After intervention | No data reported for knowledge | Insufficient data | |
| No data reported for medicine treatments | |||
| Behavioural | Not assessed | – | |
| Keslo44 n = 39, adults | Clinical-unscheduled care, 1 yr, 2 yr | Unscheduled care: Intervention group had a greater reduction in hospitalisations and ED visits | Illustrated as a consistent significant positive effect |
| aChange in ED visits. Median (IQR) visits 2 years, I: 0 (0, 0) vs. C: 2 (1.5, 2); p = < 0.05 | |||
| aChange in hospitalisations. Median (IQR) hospitalisations, I: 0 (0, 0) vs. C: 0.5 (0, 1); p = < 0.05 | |||
| FU: every month then every 2–3 m Overall risk of bias: High | Clinical-asthma control 6 m, 1 yr, 18, 2 yr | No control group data reported for quality of life, asthma bother or peak flows | Insufficient data |
| Process, before and after intervention | No control group data reported for Knowledge control group | Insufficient data | |
| No control group data reported for medicine treatments control group | |||
| Behavioural | Not assessed | – | |
| Velsor-Friedrich40 CCT n = 102, children | Clinical-unscheduled care, 2 wks, 5 m, 1 yr | Unscheduled care: the intervention group had significantly more unscheduled visits at 5 m and 1 yr | Illustrated as a consistent significant negative effect |
| Medical visits at 5 m. Mean (SE) I: 0.12 (0.05) vs. C: 0.02 (0.04) | |||
| Medical visits at 1 yr. Mean (SE) I: 0.07 (0.03) vs. C: 0.00 (SD 0.02); p = 0.01 | |||
| FU: 2 wks, 5 m, 1 yr Overall risk of bias: unclear | Clinical-asthma control, 2 wks, 5 m, 1 yr | Symptom days: greater reduction in days with symptoms in intervention compared to control | Illustrated as a consistent positive effect but hatched to show inconsistency |
| Symptom days at 5 m. Mean (SE). I: 2.15 (SE 0.30) vs. C: 1.42 (SE 0.21) | |||
| Symptom days at 1 yr. Mean (SE). I: 1.26 (SE 0.33) vs. C: 1.49 (SE 0.23); p = 0.047 | |||
| PEFR: intervention group had greater increase in PEFR at both FU time-points | |||
| % increase in PEFR at 5 m. I: 2.9% (SE 2.0%) vs. C: 2.9% (SE 1.0%) | |||
| % increase in PEFR at 1 yr. I: 7.5% (2.0%) vs. C: 2.9% (SE 1.2%); p = 0.046 | |||
| School absences: no between group difference in days absent from school | |||
| Days absent at 1 yr. I: 9.03 vs. C: 14.4 days | |||
| Process, 2 wks, 5 m, 1 yr | Knowledge, self-efficacy and self-esteem/motivation: no significant between group differences | Illustrated as a consistent no effect | |
| Asthma knowledge test at 5 m: I: 14.05 (SE 0.55) vs. C: 13.35 (SE 0.38) | |||
| Asthma belief survey at 5 m. I: 4.23 (SE 0.10) vs. C: 4.15 (SE 0.08) | |||
| Self-perception inventory at 5 m. I: 2.80 (SE 0.08) vs. C: 2.85 (SE 0.05) | |||
| Behavioural, 2 wks, 5 ms | Self-practice/asthma self-care: No significant between group differences | Illustrated as a consistent no effect | |
| Denyes self-care agency instrument at 5 m: 72.03 (SE 2.46) vs. 70.57 (SE 1.68) | |||
| Asthma self-care instrument at 5 m I: 68.87 (SE 2.89) vs. C:70.41 (SE 2.00) | |||
| Velsor-Friedrich41 CCT n = 52, children | Clinical-unscheduled care, 2 wks, 5 m, 1 yr | Urgent medical visits (and medications): no significant between group differences at any time point | Illustrated as a consistent no effect |
| Urgent doctor visits at 12 m. I: n = 4 (14%) vs. C: n = 5 (20%) | |||
| No data; some data on medicine use was provided | |||
| FU: 2 wks, 5m, 1 yr, 2 yr Overall risk of bias: unclear | Clinical-asthma control, 2 wks, 5 m, 1 yr, 2 yr | Symptoms, PEFR and school absences: no significant between group differences at any time point | Illustrated as a consistent no effect |
| Proportion with > 1 day with symptoms/2 wks at 1 yr. I: 14 (50%) vs. C: 13 (54%) | |||
| % increase in PEFR from baseline at 1 yr. I: 26.21% (SD 0.22) vs. C: 27.80% (SD 0.31) | |||
| Average days absent from school. I: 9.03 vs. C: 14.4 | |||
| Process, 2 wk, 5 m, 12 m | Knowledge and self-efficacy: Intervention group had higher scores at all time-points, but neither group improved over time | Illustrated as a consistent positive effect but hatched to show inconsistency | |
| Asthma Knowledge: test at 1 yr. Adjusted mean I: 14.28 (SE 0.80) vs. C: 11.88 (SE 0.87); p = 0.03 | |||
| Asthma belief scale at 1 yr. Adjusted mean I: 4.09 (SE 0.14) vs. C: 3.82 (SE 0.15); p = 0.01 | |||
| Self-esteem: no significance between group differences at any time point | |||
| Self-perception inventory at 1 yr. Adjusted mean I: 2.71 (SE 0.08) vs. C: 2.78 (SE 0.10) | |||
| Behavioural, 2 wks, 5 m, 1 yr | Asthma self-care practice/general self-care: intervention group had higher scores at all time-points, but neither group improved over time | Illustrated as a consistent positive effect | |
| Denyes self-care agency instrument. I: 75.55 (SE 2.60) vs. 67.41 (SE 2.82); p = 0.01 | |||
| General self-care. I: adjusted mean I: 72.99 (SE 3.26) vs. C: 63.75 (SE 3.53); p = 0.2 | |||
| Velsor-Friedrich45 RCT n = 137, adolescents | Clinical-Unscheduled care, 6 m, 12 m | Hospital visits: no significance between group differences p > 0.05 (no other data provided) | Illustrated as a consistent no effect |
| FU: 2 m, 6 m, 1 yr Overall risk of bias: high | Clinical-asthma control, 6 m, 1 yr | Symptoms reduced in both groups; no significant between group differences | Symptom takes priority. Illustrated as a consistent no effect |
| PEFR: no significance between group differences | |||
| School absences reduced in both groups; no significant between group differences | |||
| Process, 6 m, 1 yr | Knowledge, self-efficacy improved in both groups; no significant between group differences | Illustrated as a consistent no effect | |
| Coping frequency/efficacy, no significance between group differences | |||
| Behavioural, 6 m, 1 yr | Self-care practice, no significance between group differences | Illustrated as a consistent no effect |
For conflicting outcomes within a category, the decision process was dependent upon priority of evidence including:
• Defined primary outcomes in an adequately powered sample/sub-group analysis (for the latter we will consider a prior sub-group analysis)
• Outcomes measured using a validated instrument (as opposed to non-validated instruments)
• Outcomes that were clinically and statistically significant (e.g., achieved significance defined minimum clinically important difference)
• If doubts remain, the author’s interpretation was considered to provide context for the final decision
Note:
• For quality of life outcomes, we will use the overall score, if no overall score is stated the outcome will not be plotted
• Asthma related quality of life scales will be given priority (e.g., AQLQ) over generic quality of life scales (e.g., ED5D)
• For the clinical-asthma control category, symptoms will be a priority over other outcomes in the same category as it is a better indicator of asthma control
Abbreviations; FU follow-up, wks weeks, m month, yr year, RCT randomised control trial, CCT clinical control trial, ED emergency department visits, I intervention, C control, CI confidence interval, AQLQ quality of life questionnaire, AQ20 the airways questionnaire 20, ACT asthma control test, F F statistics, AHR adjusted hazard ratio, HR hazard ratio, OR odds ratio, EES estimated effect size, PEFR peak expiratory flow rate, SD standard deviation, SE standard error, DF degree of freedom, p p-values