We updated two earlier systematic reviews that dealt with telemonitoring5 6
by searching 15 electronic databases using search methods recommended by the Cochrane Heart Review Group.9
All randomised trials evaluating remote monitoring programmes published between 1 January 2002 and 6 May 2006 were included. Databases searched included the Cochrane library and the Cochrane CENTRAL register of controlled trials, Medline (1 January 2002 to 6 May 2006), Embase, CINAHL (1 January 2002 to 6 May 2006), AMED, ISI web of knowledge, HSTAT, Ingenta, Zectoc, LILACS, and science citation index expanded (to search forward to detect studies citing the original reviews), DARE, national research register, Psych Info, and web of science. We also hand searched the reference lists in 21 published systematic reviews of disease management programmes in chronic heart failure,3 4 5 6 7 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
149 review articles on telephone support programmes in chronic disease, and those studies identified in our electronic searches that met the inclusion criteria. Unpublished conference proceedings were reviewed and published abstracts were included if the authors replied to our request and sufficient details and outcomes of studies were retrieved. Finally, we communicated with the principal investigators of the identified trials and with national and international experts in the specialty to identify any studies we had potentially missed. We did not restrict study inclusion by language but did limit our review to only randomised controlled trials.
We applied the highly sensitive search strategy from the Cochrane Collaboration.9
Keywords for searches of the database included heart failure (exp), cardiac failure (exp),
telemedicine (exp), telecare (exp), telemonitoring (exp), teleconsultation (exp), teleconference (exp), telecommunications (exp), case management (exp), comprehensive health care (exp), disease management (exp), health services research (exp), home care services (exp), clinical protocols (exp), patient care planning (exp), nurse led clinics and special clinics (exp), randomised controlled trial(s), controlled clinical trial, random allocation, double blind method, single blind method, clinical trial(s), research design, comparative study, follow-up study, and prospective study.
Search strategies were written for each database and double checked by the second reviewer, under the direction and supervision of a medical librarian.
Types of interventions
Remote monitoring programmes started by a health professional (medical, nursing, social work, pharmacists) for patients with chronic heart failure living at home were eligible for inclusion if the monitoring was carried out at least once in the first month after hospital discharge, was targeted towards the patient (that is, the patient had to be the person on the telephone), was structured (as opposed to offering telephone follow-up on an “as needed” basis), and was to be delivered as the only aftercare intervention without home visits or more than usual clinic follow-up. We excluded studies in which the remote monitoring was intended primarily to deal with the problems of care givers rather than of patients. We a priori classified programmes as being structured telephone support if they consisted of standardised telephone contact of patients with chronic heart failure and relied on reporting of symptoms alone, or telemonitoring if they consisted of telephone contact for eliciting symptoms and transmission of physiological data.
Our primary outcomes were all cause mortality, all cause rate of admission to hospital (proportion of patients readmitted to hospital at least once during follow-up), and rate of admission to hospital as a result of chronic heart failure (proportion of patients readmitted to hospital at least once during follow-up). Our secondary outcomes were health related quality of life, cost, and acceptability.
Validity assessment and data abstraction
Two investigators (RAC, SCI) independently reviewed the results of the searches for study inclusion and extracted data. We excluded any studies in which additional home or clinic visits (more than usual care) were offered to patients in the intervention or control arms. Study quality (particularly method), randomisation, and intervention, were judged using accepted criteria and compared with the review protocol.9
Disagreements between the two reviewers were resolved by a third reviewer (SS, FAMcA, or JGFC). Data abstraction was carried out independently and blinded by RAC and SCI, with FAMcA checking extracted data. Overall the inter-rater reliability on key inclusion criteria (randomisation and intervention) was strong (κ score 0.73, 95% confidence interval 0.54 to 0.92).27
Study characteristics and data synthesis
Owing to the expected differences in patient populations, programme characteristics, and length of follow-up, we carried out our primary analyses using the DerSimonian and Laird random effects model. Analyses were carried out using RevMan 4.2 (Nordic Cochrane Centre).9
As the outcomes of interest were relatively common we calculated risk ratios and 95% confidence intervals. The risk difference (difference between observed proportion of the event in the treatment and usual care groups9
) was calculated by subtracting the risk of the event in the usual care group from that of the treatment group. These data are presented with 95% confidence intervals.
We carried out intention to treat analyses—that is, all patients and their outcomes were analysed in the groups to which they were allocated, regardless of whether they received the treatment. We examined for statistical heterogeneity in each outcome of interest using Cochran's Q test and I2 statistic. Secondary outcomes (expected to be reported less often) were described and tabulated.