Table 2 Design and characteristics of the included studies

1. Carroll et al. [35] USA.

1/-/2098 (children aged 3-11 years) (I = 1082, C = 1016)

RCT, Community based. Duration: 21 months

Parents received no screening questions, and physicians received no prompts.

Primary outcome: Physicians’ diagnosis of childhood asthma based on prompts by the CDSS.

(+) effect. The number of children diagnosed with asthma in the intervention group was significantly more compared to the control group (8.6% vs. 5.8%, P <0.02). Effect size Cohen’s d = 0.24, 95% C.I = (0.04-0.43)

Not clear if physician training was provided.

2. Hashimoto et al. [36] Netherlands

6/-/95 (adults with adults diagnosed with severe asthma) (I = 51, C = 38)

Pragmatic Multicentre RCT, Academic and community setting Duration: 6 months

Conventional asthma treatment by pulmonologists

Primary outcome: Cumulative sparing of oral corticosteroids, asthma control using Asthma control Questionnaire (ACQ), asthma-related quality of life (AQLQ), Secondary outcomes: FEV1 (using Piko-1 device), exacerbations, hospitalizations and satisfaction (Global satisfaction scale)

(+) effect. Median cumulative sparing of prednisone was 205 mg in the internet group compared with 0 mg in the conventional group. (P = 0.02) Asthma control, AQLQ, FEV1, exacerbations, hospitalizations and satisfaction with the strategy were not different between groups. Effect size for CCS sparing effect: Cohen’s d = 0.46. 95% C.I = (0.08- 0.93)

Patients were trained in using the electronic self -management support system, recording symptoms, measuring lung function and fraction of exhaled nitric oxide (FENO)

3. Van der Meer et al. [37], Netherlands

37 general practices and 1 academic outpatient department/69/200 (adults with asthma) (I = 101, C = 99)

Multi-centre, RCT, Community and Academic setting. Duration = 12 months

Usual physician-provided asthma care

Process outcomes: (asthma knowledge, inhaler technique and self-reported medication adherence), health care provider contacts for asthma, use of internet based asthma monitoring tool, and medication changes. Clinical outcomes: Primary: Asthma-related quality of life, (32-item Asthma Quality of Life Questionnaire). Secondary: Asthma control (ACQ), symptom-free days, pre-bronchodilator FEV1 (Piko-1), daily inhaled corticosteroid dose, and exacerbations

(+ but modest effect) Improvement in asthma knowledge, inhaler technique and slightly fewer physician visits in the internet group. Treatment changes occurred more often in the internet group. Modest improvement in asthma control and lung function with the Internet intervention, but no reduction in exacerbations. Improvement in asthma-related quality of life was slightly less than clinically significant. (P ≥ 0.5). Effect size for the primary clinical outcome (AQLQ increase by >0.5): Cohen’s d = 0.6. 95% C.I = (0.3-0.9)

Education and training provided to the participants. Non-blinded nature of the study may have affected the results

4. Van der Meer et al. [38] Netherlands

37/69/200 (adults with partly controlled or uncontrolled asthma) (I = 101, C = 99)

Prospective RCT, Community and academic. Duration: 1 year

Usual care by the general practitioner according to the Dutch GP guidelines based on GINA guidelines

Primary: Asthma control using (ACQ), spirometry and ATAQ (asthma therapy assessment questionnaire).

(+) effect. Significant improvements in ACQ score after 12 months in the internet based self-monitoring group. Daily inhaled corticosteroid dose significantly increased in the Internet group compared to usual care in the first 3 months in patients with uncontrolled asthma, but not in patients with well or partly controlled asthma. After one year there were no differences in daily inhaled corticosteroid use or long-acting β2-agonists between the Internet group and usual care.

Patients were trained to measure (FEV1) daily with a hand-held electronic spirometer (PiKo1). Supervision provided by a nurse specialist. So there is heavy initial investment. Study outcomes were self-reported by the patients which may overestimate effect. The effect size for change in asthma control is quite large, esp. in the uncontrolled group making the intervention promising.

Secondary: Mean daily dose of inhaled corticosteroid (ICS), and the proportion of participant’s using long-acting β2-agonists (LABA) or leukotriene receptor antagonists (LTRA).

Effect size primary outcome i.e. change in ACQ in partly controlled asthma group*: Cohen’s d = 0.81 95% C.I = (0.33- 1.35)

Effect size for change in ACQ in uncontrolled asthma group*: Cohen’s d = 0.94. 95% C.I = (0.38-1.5). *Assuming t-test was performed.

5. Taylor et al. [39] Australia.

3/50/1 (simulated patient) (I = 27, C = 23 ED doctors)

RCT, Community setting, Duration: 4 months

Acute asthma management using paper -based clinical records, treatment order sheets and discharge documentation.

Work load & efficiency outcomes: Primary: Quality of asthma documentation–measured using 10 documentation variables (clinical parameters and discharge documentation). Secondary: consultation time

(+) effect. Significantly higher rates of documentation in 7 out of 10 variables, including provision of written short-term asthma management plans. No significant difference in consultation times. Effect size for documentation of asthma management plan provision: Cohen’s d = 0.78. 95% C.I = (0.18-1.37)

Relevant to the ED setting. A 2 minute introduction to the system, including basic functions of the program provided to physicians, which may not be enough. One simulated patient case may not reflect a spectrum of scenarios faced in the ED setting

6. Fiks et al. [40] USA.

20/-/11919 (children with asthma between 5-19 years of age) (I = 6110, C = 5809)

Cluster-RCT, Academic. Duration: 5 months.

Routine care

Health care process outcomes- rates of captured opportunities for influenza vaccination (visit-level analysis) and up-to-date influenza vaccination status among patients with asthma

(+) but modest effect. Standardized influenza vaccination rates improved 3.4% more at intervention sites than at control sites. Effect was not significant, Cohen’s d not calculated)

Primary care sites were linked to a teaching hospital. Information on the comparator was unclear-implied usual practice. Training provided to the physicians was quite thorough.

7. Bell et al. [41] USA.

12/-/19450 (children) (I = 6, C = 6) Children with persistent asthma identified by using the pediatric asthma control test (PACT))

RCT, Academic setting. Duration: 1 year

Passive asthma management tools available in the electronic health record (EHR).

Health care process outcomes: Proportion of children with at least 1 prescription for controller medication, an up-to-date asthma care plan, and documentation of performed office-based spirometry.

(+) effect. Significant improvement in adherence to NAEPP guidelines. 6% increase in the number of prescriptions for controller medications, (P = 0.006) and 3% increase for spirometry (P = 0.04) in the intervention urban practices. Filing an up-to date asthma care plan improved 14% (P = 0.03) and spirometry improved 6% (P = 0.003) in the suburban practices with the intervention. The effect size could not be calculated, as data provided was insufficient to calculate d values).

Medical practices within the Children’s Hospital of Philadelphia (CHOP) Pediatric Research Consortium-may not be generalizable. Physicians were trained to use the CDSS. The actual number of providers involved in the study is unclear.

8. Rasmussen et al. [42] Denmark.

-/-/300 (adults with asthma)

Multi-centre RCT with three parallel groups. Community setting Duration: 6 months

Two other usual care groups: specialist group, where treatment was provided by an asthma specialist in an outpatient clinic; and a general practitioner (GP) group, where treatment was provided by GPs in primary care.

Clinical outcomes Asthma symptoms: electronic diary. Asthma quality of life: AQLQ) questionnaire. Lung function: Spirometry Airway responsiveness: Methacholine challenge test.

(+) effect. Significant improvement in the Internet group compared to the other 2 groups regarding asthma symptoms, quality of life, lung function, airway responsiveness. Significant improvement in the use of inhaled corticosteroids in the internet and specialist group. Effect size comparing the Internet vs. Specialist group for asthma symptom reduction was Cohen’s d =0.53. 95% C.I = (0.19-0.87). Cohen’s d comparing the Internet vs. GP group for asthma symptom reduction was 0.64. 95% C.I = (0.29-0.99).

The number of practitioners and the number of centers participating were unclear. No training provided to the participants or the GP’s but the laboratory assistants providing spirometry and methacholine test were trained in the required protocol

9. Dexheimer et al. [43] USA.

1/-/704 (Children 2-18 years of age), (I = 358, C = 346)

RCT, Community setting, Duration: 3 months

Usual care, i.e., no reminders or automatic printout was provided.

Primary outcome: Time from ED triage to disposition (discharge or hospital admission) decision.

No effect. No difference in time to disposition. Length of ED stay and the rate of hospital admission were similar between the two groups. (Effect was not significant, Cohen’s d not calculated)

The number of physicians, respiratory therapists and nurses involved in the study is unclear.

Secondary outcomes: Guideline adherence measures including asthma education ordered, protocol found on chart, any asthma scoring performed.

10. Smith et al. [44] UK.

29/-/911 (patients 5+ years of age with severe asthma) (I = 457, C = 454)

Cluster RCT, Community setting. Duration: 2.5 years

Control practices continued usual care.

Primary outcome: number of patients experiencing a moderate-severe exacerbation

No significant difference between groups in number of people experiencing exacerbations. Relative reductions in people experiencing hospitalizations, accident and emergency, out-of hours contacts and increase in prednisolone prescriptions for exacerbations without increasing costs. (Effect was not significant, Cohen’s d not calculated)

Training on using electronic alerts provided to at least one representative from each staff group (GP, nurse, receptionist, manager/administrator, dispenser) of the intervention practices.

Secondary outcome: outpatient attendances for asthma, primary care contacts, ‘did not attends’ (DNAs) at consultations, asthma medications and cost analyses

11. Kattan et al. [45] USA.

-/435/937 (5-11year old with moderate to severe asthma) (I = 471, C = 468)

RCT, Community setting. Duration: 1 year

No letters sent to the providers of the children in the control group

Health care process outcomes: scheduled visits and changes in medications. Patient outcomes: maximum number of symptom days, ED visits and hospitalizations for asthma, and school days missed because of asthma.

(+) effect Significant increase in scheduled visits, (17.1% vs12.3%, P = 0.005). Significant increase in medication step up (46% vs 35.6%, P = 0.03). Significantly fewer ED visits in the intervention group compared with controls (0.87 vs 1.14 per year, p = 0.013). No difference in the maximum number of symptom days and number of school days missed. Effect size for % of scheduled visits resulting in step-up of medication: Cohen’s d = 0.23. 95% C.I = (0.02-0.43)

Intervention practitioners were trained. Effect size was low for medication change related outcomes. Key issues also included the design where not all children whose medication change was warranted visited the physician.

12. Tierney et al. [46] USA.

4/266/706 (246 physicians and 20 outpatient pharmacists)

RCT, Academic setting, Duration: 3 years

Four groups: physician intervention only, pharmacist intervention only, both pharmacist and physician interventions, and no intervention (controls).

Primary: Adherence to guideline based care suggestions.

No effect. No differences between groups in adherence to the care suggestions, quality of life, patients satisfaction with physicians’ or pharmacists, medication compliance, emergency department visits, or hospitalizations. Physicians receiving the intervention had significantly higher total health care costs. Physician attitudes toward guidelines were mixed. (Effect was not significant, Cohen’s d not calculated)

Hospital based academic practices. Providers included internal medicine physicians, residents and pharmacists. Training was provided to the providers. Questionnaires were administered via telephone.

Secondary: Quality of life-McMaster Chronic Respiratory Disease Questionnaire (CRQ) for COPD patients or the McMaster Asthma Quality-of-Life. Questionnaire (AQLQ). Patient satisfaction: American Board of Internal Medicine’s patient satisfaction questionnaire. Medication adherence: Inui and Morisky surveys and pharmacy dispensing records

13. Martens et al. [47] Netherlands.

-/53/-

Clustered RCT, Community setting Duration: 1 year

28 GPs received reactive computer reminders (CRS) to improve prescribing of cholesterol-lowering drugs

Primary outcome: prescription according to the guidelines as a percentage of total prescriptions of a certain drug. Secondary outcome: user friendliness.

(+) but not significant effect. CRS with reactive reminders improved drug prescribing behavior. Preliminary results also indicate reduction in the number of prescriptions according to the advice of the computerized guidelines not to present certain drugs. It was perceived stable and user friendly. (Effect was not significant, Cohen’s d not calculated)

Preliminary study. Both groups served as control to one another. Not specific to asthma/COPD.

14. Martens et al. [48] Netherlands. Follow-up of the above study

14/53/-

Clustered RCT, Community setting. Duration: 1 year

28 GPs received CRS reminders on cholesterol prescriptions

Guideline appropriate prescriptions as a percentage of total prescriptions (of the drug category involved) for the same diagnosis on the individual GP level. Absolute number of prescriptions for a specific diagnosis per GP per 1000 enlisted patients.

No effect. No favorable effects were found for CRS with the message to prescribe certain drugs. On the other hand, CRS with the message not to prescribe certain drugs sometimes positively influenced the prescribing behavior of GPs. (Effect was not significant, Cohen’s d not calculated)

Not specific to asthma/COPD. Both groups served as a control group to one another. Authors report the study to be underpowered due to high inter doctor variation in prescribing behavior (Cluster effect). Training was provided.

15. Martens et al. [49] Netherlands. Follow-up of the above study.

20/48/-

Clustered RCT, Community setting. Duration: 1 year

28 GPs received (reactive) reminders on cholesterol prescriptions

Number of GPs (competent and willing) with CRS still functioning after 1 year. Number of GPs having technical problems or are unwilling. Number of reminders/GP/month/1000 enlisted patients. GP user satisfaction (satisfaction questionnaire). GP experience (content and extensiveness of CRS). Barriers and facilitators to implementation and use of CRS

(+) learning effect from the CRS. 9% of GPs dropped out after 1 year. A significant learning curve was found (P = 0.03) for the reminders on antibiotics, asthma and COPD. GPs were satisfied with the user-friendliness and the content of the different types of reminders, but less satisfied with certain specific technical performance issues of the system. Cohen’s d = N/A (Effect size not calculable due to insufficient data provided in report)

Not specific to asthma/COPD. Both groups served as a control to one another. GP’s were trained.

16. Kuilboer et al. [50] Netherlands.

32/40/156,772 (study patients (children and adults) either had chronic bronchitis, emphysema, other chronic pulmonary diseases, asthma or COPD) (I = 20, C = 20 General practitioners)

RCT, Community and academic Duration: 10 months

Usual care

Average number of contacts, FEV1 (force expiratory volume) and peak flow measurements per asthma/COPD patient per practice, and the average number of antihistamine, cromogylate, deptropine, and oral bronchodilator prescriptions per asthma/COPD patient per practice.

(+) effect. Statistically significant increase in contact frequency with the patient (P = 0.034), peak flow measurement, FEV₁ measurements in 12-39 years age group (P = 0.02). Significant decrease in cromogylate prescriptions in the age group of 12-39 years, (P = 0.03). Non-significant decrease in deptropine, antihistamines, oral bronchodilators. (Effect size not calculable due to insufficient data provided in report).

The study focused on change in physicians’ behavior. Training was provided to the general practitioners.

17. Poels et al. [51] Netherlands

1 medical centre, several private practices/78/774 paper case descriptions. (10 case descriptions per GP).

Simulated cluster- RCT Community. Duration = 10 months

GPs in the control group simply received the spirometry test results, and the flow–volume and volume–time curves.

Primary: Difference between the percentage agreement of the cases’ diagnoses between GPs and expert panel judgment before and after interpretation of spirometry

No Effect. There were no differences between the computerized expert support and control groups in the agreement between GPs and expert panel on diagnosis of COPD, asthma and absence of respiratory disease. A higher rate of additional diagnostic tests was observed in the expert support group. (Effect was not significant, Cohen’s d not calculated)

This was a simulated study- no real patients involved. Training was provided.

Secondary: Impact of the expert system intervention on the GPs decision-making processes through six measures: additional diagnostic test rates; width of differential diagnosis; certainty of diagnosis; estimated severity of disease; referral rate; and medication or non-medication changes.

18. Poels et al. [52] The Netherlands

44/-/2098 (I = 15. C = 15. Chest physician = 14)

Cluster-RCT Duration: not mentioned.

Usual care had standard spirometry software (i.e. no additional interpretation support).

Primary: A change of diagnosis after spirometry interpretation support. Secondary: referral rate, additional diagnostic tests, and disease management changes.

No effect. Differences in proportion of changed diagnoses were not statistically significant. There were no differences in secondary outcomes. (Effect was not significant, Cohen’s d not calculated)

Training was provided.

19. Frickton et al. [53], USA.

15/102/59,147. (Patients with medically complex conditions like xerostomia, diabetes mellitus, COPD, congestive heart failure).

RCT with three arms (provider activation, patient activation and control group), Community setting. Duration = 2 years

Patients in the control group received usual care. Neither the patients nor the provider’s, received alerts about a patient’s medical status or personalized care guidelines.

Primary: Total use-the overall frequency with which providers accessed the guidelines web site via the EDR for any patient. Targeted use—the proportion of providers who accessed the care guidelines in general and for targeted patients at the point of care. Ongoing use—the proportion of providers who continued to access the web-based guidelines through-out the study period.

(+) effect. Participants in the provider and patient activation groups increased their use of the system during the first six months. Provider activation was more effective than was patient activation. (P < 0.05). However, it was not sustainable, and by the end of the study, the rate of use had returned to baseline levels despite participants’ continued receipt of electronic alerts. (Effect expressed as Odds Ratio for web use for provider group in first six months = 4.4 (95% C.I = 1.6-12.1) and 6-12 months after implementation compared to controls = 1.7 (95% C.I = 0.1-2.9). For provider + patient activation group, effect expressed as Odds Ratio for web use in first six months =2.1 (95% CI, 0-9-4.8) and 6-12 months after implementation compared to controls = 1.4 (95% CI 0.5-3.5).

Dental clinic based. Study was not specific to asthma/COPD patients.