This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
We conducted in two parts a systematic review of randomized controlled trials (RCTs) on electronic symptom reporting between patients and providers to improve health care service quality. Part 1 reviewed the typology of patient groups, health service innovations, and research targets. Four innovation categories were identified: consultation support, monitoring with clinician support, self-management with clinician support, and therapy.
To assess the methodological quality of the RCTs, and summarize effects and benefits from the methodologically best studies.
We searched Medline, EMBASE, PsycINFO, Cochrane Central Register of Controlled Trials, and IEEE Xplore for original studies presented in English-language articles between 1990 and November 2011. Risk of bias and feasibility were judged according to the Cochrane recommendation, and theoretical evidence and preclinical testing were evaluated according to the Framework for Design and Evaluation of Complex Interventions to Improve Health. Three authors assessed the risk of bias and two authors extracted the effect data independently. Disagreement regarding bias assessment, extraction, and interpretation of results were resolved by consensus discussions.
Of 642 records identified, we included 32 articles representing 29 studies. No articles fulfilled all quality requirements. All interventions were feasible to implement in a real-life setting, and theoretical evidence was provided for almost all studies. However, preclinical testing was reported in only a third of the articles. We judged three-quarters of the articles to have low risk for random sequence allocation and approximately half of the articles to have low risk for the following biases: allocation concealment, incomplete outcome data, and selective reporting. Slightly more than one fifth of the articles were judged as low risk for blinding of outcome assessment. Only 1 article had low risk of bias for blinding of participants and personnel. We excluded 12 articles showing high risk or unclear risk for both selective reporting and blinding of outcome assessment from the effect assessment. The authors’ hypothesis was confirmed for 13 (65%) of the 20 remaining articles. Articles on self-management support were of higher quality, allowing us to assess effects in a larger proportion of studies. All except one self-management interventions were equally effective to or better than the control option. The self-management articles document substantial benefits for patients, and partly also for health professionals and the health care system.
Electronic symptom reporting between patients and providers is an exciting area of development for health services. However, the research generally is of low quality. The field would benefit from increased focus on methods for conducting and reporting RCTs. It appears particularly important to improve blinding of outcome assessment and to precisely define primary outcomes to avoid selective reporting. Supporting self-management seems to be especially promising, but consultation support also shows encouraging results.
This paper presents the second part of a comprehensive review of randomized controlled trials (RCTs) focusing on electronic communication between patient and provider to improve health care service quality. This patient-provider partnership is defined by patients or parents reporting symptoms or health information electronically [
Part 1 of this review identified the following typology of the field in terms of [
Five specific patient groups mainly based on the International Classification of Primary Care (ICPC) definitions [
Four health service innovation categories: consultation support, monitoring with clinician support, self-management with clinician support, and therapy.
Research targets: consultation support studies primarily aimed to improve patient-centered care and secondarily to provide health benefits. Monitoring studies focused on health benefits, patient-centeredness outcomes, and reduced health care costs. Self-management studies mainly aimed for health benefits and secondarily patient-centered outcomes.
This part of the review looked into the methodological quality on the RCTs, and summarized effects and benefits of electronic symptom reporting of the methodologically best RCTs.
It is possible to achieve effects of electronic symptom reporting at the health care professional, health care system, and patient levels.
At the
At the
At the
However, our preliminary screening found that studies in the field typically are small in terms of number of patients involved and are best described as feasibility studies [
We wanted to limit our work to the most mature stage of a complex intervention before implementation, the RCTs [
The overall aim of this review was to systematically assemble the knowledge gained from RCTs focusing on electronic communication between patient and provider to improve health care service quality.
The objective for this second part of the review was to (1) assess the methodological quality of the RCTs identified in the first part of the review, and (2) summarize effects and benefits of electronic symptom reporting from data published in the methodologically best RCT articles. The benefits will be presented with regard to patients, health care professionals, and health care systems.
The review in general followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) recommendations [
We included original RCTs if patients or parents reported recent symptoms or health information electronically, either to clinical health care personnel or to a system, where the receiver processed and interpreted the data for health care purposes and provided feedback. Feedback did not need to be given electronically. The focus was on asynchronous systems that can be established within the health care system. If the control group reported symptoms or health information, this information was not received by the health care professional or system. For a detailed description of the inclusion and exclusion criteria, see part 1 of the review [
We searched Medline, EMBASE, PsycINFO, the Cochrane Central Register of Controlled Trials, and IEEE Xplore to retrieve RCTs about human medicine presented in the English language, published from 1990 to November 2011. For a detailed description of the search methods, see part 1 of the review [
Studies (abstract and full text) were selected independently by two authors (MAJ and EH), and all disagreements were resolved by consensus discussions. We extracted 84 variables from each included article, guided by the Cochrane data collection checklist [
Extracted variables focused on the assessment of methodological quality, including evaluation of the risk of bias in the results, and outcome measures and results relevant to electronic symptom reporting. A full presentation of the extracted variables and the citations can be found on the website of the Norwegian Centre for Integrated Care and Telemedicine [
We assessed the methodological quality of each article, including risk of bias and three additional variables reflecting feasibility, theoretical evidence, and preclinical testing.
Risk of bias was assessed according to Cochrane’s recommended domain-based evaluation, the criteria for judging risk of bias [
Theoretical evidence and preclinical testing are both recommended as part of a framework for the design and evaluation of complex interventions to improve health [
The “assessment of the overall risk of bias involves consideration of the relative importance of different [bias] domains” and the review author’s judgments “about which domains are most important in the current review” [
Incomplete outcome data refers both to attrition and to exclusion of participants through as-treated or other subgroup analyses. When assessing selective reporting, we accepted that a primary outcome variable could be represented through a group of well-defined measures, as long as authors reported all measures and time points properly and completely in the results section. On the other hand, we assessed studies as having a high risk for selective reporting if we found any incongruence between the published protocol and the reported primary results, or if the variable used to make power calculations was not part of the reported primary outcome measures. We applied the same logic if the authors’ main conclusions did not rely on previously defined outcome parameters and therefore had to be regarded as limitedly interpretable post hoc findings.
The blinding of outcome assessment risk of bias was judged as high if there was no information indicating involvement of any independent personnel for assessment of outcome other than those performing the intervention. If the patients were the outcome assessors for the primary outcome, and all the patients had access to a common online discussion page, we regarded this as having knowledge about which intervention they and other patients received.
The Cochrane criteria for the unclear risk judgment is primarily defined as “insufficient information to permit judgment of ‘low risk’ or ‘high risk’” [
Three of the authors (TS, GB, MAJ) assessed the risk of bias independently. In all cases of disagreement, a discussion took place until consensus was achieved.
We agree that it is not acceptable to “present analyses and interpretations based on all studies, ignoring flaws identified during the assessment of risk of bias” [
For all other articles, we extracted the primary outcome and present it in an effect table according to the article’s health service innovation category. Some articles defined more than one outcome variable as their primary outcome. In these cases, we chose the first variable presented in the article’s text to be included in our effect table.
Since only a few of these RCTs had a follow-up after the intervention, we chose the immediate postintervention outcome when extracting effects. We calculated within- and between-group pre- to postintervention differences and report the extracted
In addition to the primary outcome effects, other extracted results from the articles with acceptable quality are reported for each health service innovation according to who might benefit: patients, health professionals, or the health care system. The reporting makes use of the Institute of Medicine (IOM) definitions stating that health care should be safe, effective in terms of health benefits, patient centered, timely, efficient (reduced time, reduced health care costs for the health system, and resource utilization of the health professional), and equitable [
Of 642 records identified and 444 abstracts reviewed, 32 articles presenting 29 studies were included [
All except 2 studies were designed as parallel studies with random allocation of patients. Of the parallel studies, 4 had three arms [
Even if we accept that patients and personnel were not blinded, no articles met all the quality requirements, and many articles satisfied few methodological quality criteria (
All articles had interventions that could be implemented as planned in a real-life setting. Thus, this aspect is not included in
The quality assessment with regard to random sequence allocation was the bias domain with the best results, with three-quarters of the articles judged to have low risk. For only about half of the articles, the risk of bias for allocation concealment, incomplete outcome data, and selective reporting was judged to be low risk. With regard to selective reporting, 25% of the articles used several primary outcomes, and 28% had not defined or remained unclear regarding the primary outcome (Tables 4-7 in part 1, [
For barely more than one-fifth of the articles, we judged the blinding of outcome assessment bias to be low risk. High risk of bias due to inadequate concealment of the allocated intervention from participants and personnel during the study is very challenging in telemedicine and eHealth research. This was, not surprisingly, achieved for only 1 of the included articles, Yardley et al [
We excluded 12 articles assigned a high risk or unclear risk for both selective reporting and blinding of outcome assessments from the following effect report. We excluded 3 consultation support articles: Berry et al [
The reported effects (
In the monitoring category, 2 of the asthma studies confirmed their hypothesis, while we lack positive results for chronic obstructive pulmonary disease and cardiovascular monitoring. Also, the hypothesis was confirmed in 2 studies on consultation support in the cancer patient group.
Judgments of methodological quality in the reviewed randomized controlled trials of electronic symptom reportinga.
Article | Theoretical evidence/ |
Random |
Allocation |
Blinding of |
Blinding of |
Incomplete |
Selective |
Berger et al [ |
yes/yes | + | + | – | – | + | + |
Bergström et al [ |
yes/unclear | + | + | – | + | + | + |
Berry et al [ |
yes/yes | + | + | – | – | + | – |
Boyes et al [ |
yes/no | ○ | ○ | – | – | – | – |
Carrasco et al [ |
yes/no | + | – | – | – | + | + |
Chan et al [ |
yes/no | + | ○ | – | – | ○ | – |
Chan et al [ |
yes/yes | + | ○ | – | – | – | – |
DeVito Dabbs et al [ |
yes/yes | + | – | – | – | + | + |
Glasgow et al [ |
yes/yes | + | + | – | – | ○ | + |
Guendelman et al [ |
yes/yes | ○ | + | – | – | + | + |
Jan et al [ |
yes/no | ○ | + | – | – | + | – |
Kearney et al [ |
yes/yes | + | + | – | – | – | – |
Leveille et al [ |
yes/no | ○ | ○ | – | – | + | + |
Lewis et al [ |
yes/unclear | + | + | – | + | ○ | – |
Lewis et al [ |
unclear/unclear | + | + | – | – | ○ | – |
Nguyen et al [ |
yes/yes | + | + | – | – | – | + |
Nguyen et al [ |
yes/no | + | + | – | – | + | – |
Oerlemans et al [ |
yes/no | + | – | – | – | + | + |
Prabhakaran et al [ |
no/no | + | + | – | – | + | – |
Rasmussen et al [ |
yes/no | – | + | – | – | ○ | + |
Ruland et al [ |
yes/yes | ○ | ○ | – | ○ | ○ | + |
Ruland et al [ |
yes/yes | + | – | – | + | + | + |
Santamore et al [ |
no/no | ○ | ○ | – | – | ○ | – |
Schwarz et al [ |
yes/no | + | + | – | + | + | + |
Stevens et al [ |
yes/no | + | ○ | – | – | – | – |
van der Meer et al [ |
yes/unclear | + | + | – | – | + | + |
Velikova et al [ |
yes/no | + | ○ | – | + | + | + |
Vernmark et al [ |
yes/no | + | ○ | – | + | + | + |
Wagner et al [ |
yes/no | + | ○ | – | – | + | + |
Willems et al [ |
yes/no | + | – | – | – | + | + |
Williams et al [ |
no/no | ○ | ○ | – | – | ○ | ○ |
Yardley et al 2010 [ |
yes/no | + | + | + | + | – | ○ |
a Articles were identified in a comprehensive search in Medline, EMBASE, PsycINFO, Cochrane Central Register of Controlled Trials, and IEEE Xplore from 1990 to November 2011, and were published in the time period 2002–2011.
b + = low risk, ○ = unclear risk, and – = high risk.
c Blinding of participants and personnel is extremely challenging in telemedicine and eHealth innovations. We thus did not consider this bias to be crucial to the quality judgment of the articles.
d We considered a low risk of bias for selective reporting and for blinding of outcome assessment to be the best indicators for identification of studies with high methodological quality.
Effects reported in consultation support randomized controlled trials of acceptable quality of electronic symptom reporting, in alphabetic order of first authora.
Article | Primary |
Intervention |
Number at |
Within-group |
Between-group |
Hypothesis |
|||||
C | I1 | I2 | C | I1 | I2 | C-I1 | C-I2 | ||||
Leveille et al [ |
Number of patients discussing chronic condition with physician during consultation | Superior | 120 | 121 | NAd | –80.5% | –84.9% | NA | 4.4% |
NA | No |
Ruland et al [ |
Congruence (weighted) between patients’ preconsultation-reported health issues and issues discussed in consultation | Superior | 25 | 27 | NA | –12.8 | –33 | NA | 20.2 |
NA | Yes |
Ruland et al [ |
Number of patients’ symptoms and problems addressed by physicians as documented in inpatients’ records | Superior | 70 | 75 | NA | –7.9 | –10 | NA | 2.1 |
NA | Yes |
Velikova et al [ |
Health-related quality of life, functional assessment of cancer | Superior | 72 | 144 | 70 | NRe | NR | NR | –0.02 |
NR | No |
a Articles were identified in a comprehensive search in Medline, EMBASE, PsycINFO, Cochrane Central Register of Controlled Trials, and IEEE Xplore from 1990 to November 2011, and were published in the time period 2002–2011.
b C = control group, I1 = intervention group 1, I2 = intervention group 2.
c
d Not applicable.
e Not reported.
Effects reported in monitoring with clinical support randomized controlled trials of acceptable quality of electronic symptom reporting, in alphabetic order of first authora.
Article | Primary |
Intervention |
Number at |
Within-group |
Between-group |
Hypothesis confirmed? | |||||
C | I1 | I2 | C | I1 | I2 | C-I1 | C-I2 | ||||
Carrasco et al [ |
Number of patients exhibiting poor hypertension control | Superior | 143 | 142 | NAd | 64.3% | 68.3% | NA | –4.0% |
NA | No |
Guendelman et al [ |
Patients experiencing limitations in activity due to asthma last 14 days | Superior | 68 | 66 | NA | 25% | 35% | NA | –9% |
NA | Yes |
Lewis et al [ |
Number of hospitalizations | Superior | 20 | 20 | NA | –7 | –4 | NA | –3 |
NA | No |
Rasmussen et al [ |
FEV1 e | Superior | 100 | 100 | 100 | 0.004 | 0.187 | 0.035 | –0.183 |
–0.031 |
I1: Yes |
Schwarz et al [ |
Mean number of hospital readmissions in group | Superior | 51 | 51 | NA | –0.33 | –0.32 | NA | –0.01 |
NA | No |
Willems et al [ |
Asthma-specific quality of life | Superior | 54 | 55 | NA | –0.06 | –0.29 | NA | 0.23 |
NA | No |
a Articles were identified in a comprehensive search in Medline, EMBASE, PsycINFO, Cochrane Central Register of Controlled Trials, and IEEE Xplore from 1990 to November 2011, and were published in the time period 2002–2011.
b C = control group, I1 = intervention group 1, I2 = intervention group 2.
c
d Not applicable.
e Forced expiratory volume of air in the first second of expiration.
Effects reported in self-management with clinical support randomized controlled trials of acceptable quality of electronic symptom reporting, in alphabetic order of first authora.
Article | Primary |
Intervention |
Number at |
Within-group |
Between-group |
Hypothesis |
|||||
C | I1 | I2 | C | I1 | I2 | C-I1 | C-I2 | ||||
Berger et al [ |
Social Phobia Scale | Equivalence | 27 | 27 | 27 | 16.2 | 16.3 | 17.9 | –0.1 |
–1.7 |
Yes |
Bergström et al [ |
Panic Disorder Severity Scale | Equivalence | 60 | 53 | NAd | NRe | NR | NA |
|
NA | Yes |
DeVito Dabbs et al [ |
Perceived self-care agency (in follow-up after surgery) | Superior | 17 | 17 | NA | –15.1 | –19.44 | NA | 4.34 |
NA | Yes |
Glasgow et al [ |
Number of laboratory procedures completed in accordance with national diabetes guidelines | Superior | 28f/417 | 24f/469 | NA | –0.09 | –0.37 | NA | 0.28 |
NA | Yes |
Nguyen et al [ |
Patients’ self-report of dyspnea with activities of daily living, Likert scale | Equivalence | 24 | 26 | NA | –4 | –2.5 | NA | –1.5 |
NA | Yes |
Oerlemans et al [ |
Cognitive Scale for Functional Bowel disorders | Superior | 38 | 38 | NA | 1.84 | 8.99 | NA | –7.15 |
NA | No |
van der Meer et al [ |
Asthma-related quality of life | Superior | 99 | 101 | NA | –0.18 | –0.56 | NA | 0.38 |
NA | Yes |
Vernmark et al [ |
Symptom reduction: Beck Depression Inventory | Equivalence | 29 | 29 | 30 | 5.2 | 9.9 | 11.9 | –4.7 |
–6.7 |
Yes |
Yardley et al [ |
Patient enablement scores, after 4 weeks. | Superior | 346 | 368 | NA | 2 | 3 | NA | –1 |
NA | Yes |
a Articles were identified in a comprehensive search in Medline, EMBASE, PsycINFO, Cochrane Central Register of Controlled Trials, and IEEE Xplore from 1990 to November 2011, and were published in the time period 2002–2011.
b C = control group, I1 = intervention group 1, I2 = intervention group 2.
c
d Not applicable.
e Not reported.
f Cluster randomized by primary care physicians.
g 214 replied with regard to patient enablement, 95 in the intervention group, 119 in control.
Effects reported in therapy randomized controlled trials of acceptable quality of electronic symptom reportinga.
Article | Primary |
Intervention |
Number at |
Within-group |
Between-group |
Hypothesis |
|||||
C | I1 | I2 | C | I1 | I2 | C-I1 | C-I2 | ||||
Wagner et al [ |
Intrusion measured by the Impact of Event Scale | Superior | 26 | 29 | NAd | 3.77 | 11.5 | NA | –7.73 |
NA | Yese |
a Articles were identified in a comprehensive search in Medline, EMBASE, PsycINFO, Cochrane Central Register of Controlled Trials, and IEEE Xplore from 1990 to November 2011, and were published in the time period 2002–2011.
b C = control group, I1 = intervention group 1, I2 = intervention group 2.
c
d Not applicable.
e Authors considered
Main research focus and overview of confirmed (+) and not confirmed (–) hypothesis for articles included in effect review of randomized controlled trials of electronic symptom reporting, by health service innovation category and patient groupa.
Patient group | Consultation support | Monitoring with |
Self-management with |
Therapy |
Cancer | 3 articles (2 excluded): |
0 articles (1 excluded) | ||
Respiratory and lung diseases: asthma | 3 articles (4 excluded): |
1 article: |
||
Respiratory and lung diseases: COPDb | 1 article (2 excluded): |
1 article: |
||
Respiratory and lung diseases: other | 2 articles: |
|||
Cardiovascular disease | 2 articles (1 excluded): |
|||
Psychiatry | 0 articles (1 excluded) | 4 articles: |
1 article: |
|
Diabetes | 1 article (1 excluded) |
|||
Mixed | 1 article: |
|||
Number of articles excluded due to low quality | 3 | 8 | 1 | 0 |
a Articles were identified in a comprehensive search in Medline, EMBASE, PsycINFO, Cochrane Central Register of Controlled Trials, and IEEE Xplore from 1990 to November 2011, and were published in the time period 2002–2011.
b Chronic obstructive pulmonary disease.
c Hypothesis was to demonstrate equivalence between intervention and control.
The outcomes were categorized by IOM’s quality domains [
The electronic symptom reporting systems also showed positive outcomes for patient symptom distress [
Most clinicians found the summaries useful for identifying problems and providing communication [
The one trial with patient coaching [
Only 2 monitoring studies reported benefits for patients, while nearly no benefits for the health system and none for the health professionals were reported. The 2 studies identifying health benefits for the patient focused on asthma outcomes for children [
All the studies, except 1, that aimed to demonstrate reduced health care costs belong to the monitoring group. However, with one exception, no health care costs or health care system benefits were identified: there was no improvement in total number of home care services or informal social support [
Articles on self-management support were of higher quality, allowing a larger proportion of studies to be assessed with respect to effects. All self-management interventions were found equally effective to or better than the control option, with only one exception [
Patient health benefits were reported for follow-up after lung transplantation [
For health professionals, resource utilization was reported. There was some reduction in the number of physician consultations, due to increased asthma control [
At the health care system level, health care cost benefits were analyzed and reported for Internet treatment of panic disorder, which was nearly 4 times cheaper than group treatment [
Patients receiving email therapy for complicated grief improved significantly relative to participants in the waiting list condition, and were quite satisfied with the treatment [
Results are discussed with respect to methodological quality of the included RCTs and the effects and benefits gained from electronic symptom reporting between patient and health care provider.
Overall, the research field appears to be characterized by a comparably large number of low-quality articles that have serious methodological drawbacks.
In total 25% of the articles had multiple primary outcomes, and 28% had not defined or remained unclear regarding the primary outcome.
We extracted effect data only from articles with acceptable quality, which represented 62.5% of all included articles. About half of the articles in the consultation support and monitoring categories were excluded due to low quality, whereas only 1 of the 10 articles on self-management had to be excluded (1 of 2 articles describing the same study). The study hypotheses were confirmed in 13 of the 20 remaining articles. The hypotheses were confirmed in all 4 equivalence studies.
Overall, articles on self-management support were of a higher quality, allowing a larger proportion of studies to be assessed with respect to effects. All the self-management interventions are equally effective to or better than the control option, with one exception [
In the monitoring trials, health benefits were identified for asthmatic children [
The cancer studies in consultation support are encouraging, since it was found to provide patient-centered care, ensuring that patient-reported symptoms guided the clinical decisions.
According to our requirements, seven of eight quality criteria should be fulfilled for a study to be considered methodologically correct (accepting that patients and personnel are not blinded). Unfortunately, none of the included articles received positive scores on all criteria, and many articles met just few of them. This lack of adequate methodology negatively affects the overall quality of the RCTs, as pointed out in other studies and reviews [
A total of 9 studies had an unclear primary outcome description [
Of the 9 studies with unclear primary outcome and excluded due to low quality, 6 belong to the monitoring category. None of the self-management studies were excluded (only a secondary analysis article where the primary analysis article still is included), which may reflect that this area is more mature. Self-management has already proven to be quite efficient for many long-term diseases [
The heterogeneity in intervention and research targets limits the possibilities to draw reliable conclusions with respect to the effects. Furthermore, designing, conducting, and reporting high-quality RCTs in this field in general is a great challenge, as they have to deal with complex interventions. The complex interventions include several components acting both independently and interdependently [
An important strength of this review is that the methodological quality assessment was based on Cochrane’s recommended domain-based evaluation for assessment of risk of bias [
However, when conducting the bias evaluation, we could have split the bias regarding blinding of participants and personnel in two—that is, considering study participants and personnel separately. The reason for this is that some of the trials blinded patients to which intervention they received, but did not blind the involved health care personnel.
Some of the assessed articles had sources of bias outside those specified by the Cochrane Collaboration. An example is multiple end-point criteria and hypotheses without adequate adjustment for multiple statistical tests, which may cause problems with final interpretation of results [
Interpretation of evidence depends on many factors and is rarely straightforward [
Where 8 of the articles used more than one primary outcome, we decided to extract the first variable presented in the text to use in our effect measurement. All 8 had sufficient methodological quality, and 4 demonstrated a positive effect, while the other 4 did not. However, 3 of the 4 that did not demonstrate a positive effect: the self-management study by Oerlemans et al [
Despite using a very comprehensive search strategy, we have reason to believe that we did not quite succeed in covering the area of psychiatry adequately (see part 1 of the review for a more detailed discussion [
As a result of the heterogeneous outcome data in the studies, a meta-analysis was not possible.
Studies of low quality are typically associated with an overestimation of benefits [
In addition to better definition of the primary outcome to avoid selective reporting, we also recommend improving the account of how the proposed intervention should work, how the intervention links to the outcome measures, and the use of intention-to-treat analysis. As we rarely found that principles such as patient empowerment and patient-centered care were appropriately taken into consideration, this is also a recommendation for future studies.
We also encourage researchers to carefully consider whether it is necessary to demonstrate that the intervention is superior, or if it is sufficient to demonstrate that it is equivalent, as some of the studies designed as superiority studies would have had positive results if they had been designed as equivalence studies. Examples of such studies are the monitoring study of Willems at al [
More than half of the monitoring, self-management, and therapy interventions lasted 4 months or less (see part 1), which might be too brief to achieve the intended effect for long-term conditions, especially for complex interventions where both patient and provider often need some time to get used to the technology. Some of the negative studies may have had too short a time frame for an effect to materialize. For example, in Leveille et al’s study of nurse coaching, 38% of intervention participants had less than 2 weeks between completing the screening survey and their indexed appointment, so a longer intervention period might have led to better outcomes [
The self-management support trials were very successful and showed the most promising results, and should thus be an important guide for further research.
In the consultation support category, two related questions need to be investigated in future studies: (1) does completion of questionnaires, simply giving patients the opportunity to express how they feel, have a positive effect on patient well-being, regardless of whether the results are fed back to physicians? [
The therapy category comprises innovations where the whole treatment, and all communication between therapists and patients, is conducted exclusively electronically. Unfortunately, we identified only 1 therapy article, so we cannot say anything about the general effects without conducting a new search, using specific psychiatry and Internet therapy-related terminology, as suggested in part 1 of the review [
An important contribution to the field would be to identify theoretical models that link the health service innovations, and their various components, with expected effects for patients, health professionals, and the health care system in a way that may support the design of the next generation of studies.
The number of studies within each combination of patient group and health service innovation is too small to draw final conclusions. However, if we look at electronic symptom reporting with regard to health service innovations, consultation support and self-management seem to bear various potential benefits for all stakeholders, at the patient, health care provider, and health care system levels.
Symptom reporting in cancer consultation support seems to require little staff effort to empower the patient, and there is little reason to doubt the accuracy of real-time reported symptom data when compared with average or retrospective ratings [
Electronic symptom reporting for consultation support should also hold potential for other health conditions—for example, in raising sensitive issues the patient might find difficult to disclose in a face-to-face setting, such as stigma associated with sexually transmitted diseases or mental health problems [
The positive effects on patients’ self-management should encourage health care providers to promote future services based on the best practice of these innovations. Some of the self-management studies also point to improved cost effectiveness, as shown by Bergström et al with their Internet treatment for panic disorder, which was nearly 4 times cheaper than the group treatment [
This review’s positive result regarding self-management studies should be seen in light of the comprehensive Cochrane review on interactive health communication applications for people with chronic diseases [
Even in the subgroup of RCTs, the research methods in the included trials are of low quality. The field would benefit from an increased focus on methods of conducting and reporting RCTs. It appears particularly important to improve blinding of outcome assessment and to more precisely define the primary outcomes to avoid selective reporting.
Electronic communication between patients and health care providers is an exciting area of development for innovative health services, in line with current policies strengthening patient-centered service delivery models and information and communication technologies to increase efficiency and quality. Supporting self-management seems to be especially promising, but results from consultation support trials are also encouraging.
Number of patients and articles, and the average number of patients per article, in each combination of patient group and health innovation category.
International Classification of Primary Care
Institute of Medicine
Preferred Reporting Items for Systematic Reviews and Meta-analyses
Randomized Controlled Trial
This study was partly funded by grants from the Health North Authorities (TFP-661-07), Norwegian Research Council grants (1749349), and The Norwegian Centre for Integrated Care and Telemedicine (NST). We want to express our thanks to colleagues at NST for useful advice, support, and feedback.
None declared.