Results

Results

Our search yielded 273 abstracts of randomized controlled trials, of which 190 were rejected after initial screening using very broad eligibility criteria (i.e., all comparisons of intervention for post-discharge care that reported at the least readmission and/or mortality outcomes among patients with HF). The review of reference lists of related systematic reviews, selected narrative reviews, and primary articles yielded an additional 13 citations. A total of 96 articles were retrieved for full text examination. Of these, 14 articles were duplicate reports, 28 articles were conducted among mixed population, three studies were not randomized trials, and two studies that could not be retrieved were excluded. Forty-nine unique randomized controlled trials qualified for analysis in this report.^{8,13-18,30-71} Figure 1 summarizes the search and selection of articles. Appendix B summarizes the data of the included studies.

The characteristics of included studies are reported according to the settings where interventions were initiated among patients either during or after an index hospitalization. Index hospitalization is the time period indicating the beginning of study interventions or the recruitment of patients for the study. The study characteristics description is followed by a review of the key question, and the additional outcomes of interest.

All studies restricted patients' recruitment to those who were discharged to their homes or returned to their community. Studies excluded populations for whom coexisting illnesses were likely to reduce life expectancy and/or living in a nursing home setting. Studies also excluded patients with dementia or any psychiatric illnesses, poorly compensated HF, chronic pulmonary diseases, unstable angina, and acute myocardial infarction. The most commonly reported etiology of HF among patients in the included studies was due to ischemic heart disease. Hypertension was the most frequent coexisting medical disease in these patients. The studies most frequently reported LV systolic dysfunction measured as left ventricular ejection fraction (LVEF).

The severity of HF and the distribution of NYHA class among included patients varied across studies. The studies reported that the therapeutic management of HF patients was optimized as guided by the clinical practice guidelines available at the time of the trials. Among studies, overall at baseline there were no significant differences in demographics, medication use, and co-existing medical illnesses between the intervention and usual care group.

In general, patients in the intervention group had their followup in a HF clinic managed by a study or a specialist nurse under the supervision of a cardiologist. The primary care physicians, or sometimes cardiologists followed HF patients in the usual care group. Generally in the usual care group, studies failed to report care coordination of patients during their transition from an inpatient setting to the post-discharge period or outpatient clinic.

Figure 1. Flow diagram summarizes the search and selection of articles

Key question 1: In HF patients, what is the effectiveness of interventions to support post-discharge care compared with the usual care to prevent readmission?

Description of studies with all cause readmission data

Overall 37 studies reported data on all cause readmissions after an index hospitalization. Of which, 20 studies reported interventions beginning in the inpatient setting, and 17 studies that recruited patients beginning in the immediate post-discharge period (Table 2a-3a.).^{8,14-16,18,30-32,34,35,37-39,41-46,48-62,64-66} In 34 of 37 studies there were a total of 2,054 readmissions in 3,147 patients in the intervention group and 2,589 readmissions in 3,134 patients in the usual care group. Followup duration of these studies ranged from 3 months to 16 months. The remaining three studies reported data on the number of readmissions per patient,¹⁵ on the mean difference of readmissions⁵⁷ or on readmissions per patient.⁶⁰

Thirty studies provided quantitative data on the number of readmitted patients in the intervention and control group.^{8,14,16,17,30-33,36,37,39-42,44-50,52,54-57,59,60,62,65} These studies were categorized according to the type of primary interventions reported in each individual study and combined in meta-analyses. The interventions in these studies included: only telephone followup (13 studies),^{14,31,39,40,42,45,46,48,50,54,57,59,62} increased visits to a cardiology clinic (5 studies),^{30,33,37,55,56} home visits (4 studies),^17,36,49,52 home visits and increased clinic visits (1 study),⁶⁵ home telemonitoring (1 study),¹⁶ multidisciplinary care (4 studies),^8,32,44,47 self-care instruction only by a care provider or through software (2 studies).^41,60

Effectiveness of interventions to support post-discharge care compared with the usual care to prevent readmission:

Only Telephone support

Thirteen studies that included a total of 2,167 HF patients utilized only telephone support intervention, compared to usual care, and reported readmissions.^{14,31,39,40,42,45,46,48,50,54,57,59,62} One study that reported no readmission events in both arms was excluded from this meta-analysis.³¹ Twelve studies were combined in a meta-analysis (Figure 2).^{14,39,40,42,45,46,48,50,54,57,59,62} Nine studies were conducted in the US,^{14,31,39,46,48,50,54,59,62} and the remaining four studies were conducted in Europe.^40,42,45,57 Only two studies were graded poor quality;^14,46 the remaining studies were good to fair quality. Only one study recruited patients whose mean age were 80 years.⁵⁷ The severity of HF varied among the studies. Two studies utilized software assisted telephone followup by the study nurse or a specialist nurse.^14,62 Only one study reported data on ACE-I titration over the telephone followup.⁵⁰

Our meta-analysis indicated a slight reduction in readmission rates in the telephone followup group compared to the control group, but this failed to reach statistical significance (Figure 2).

Clinic visits

Five studies that included a total of 1,155 HF patients reported quantitative data on all cause readmissions among those who had increased clinic visits to a HF or a specialty clinic.^{30,33,37,55,56} In addition to increased clinic visits, all studies except one utilized telephone followup as an intervention component. One study was conducted in the US,³⁷ and the remaining studies were conducted in Canada and Europe.^30,33,55,56 One study was graded good quality,³⁰ two studies were graded fair quality,^37,56 and two studies were graded poor quality.^33,55 Only one study reported data on the care coordination during the transition from inpatient setting to post-discharge followup.⁵⁵ This study reported no difference in readmissions between the increased clinic visits group and the control group. The remaining four studies reported reduced readmissions in the intervention group compared with control group, although statistical significance was reached in only two studies. Our meta-analysis identified that the increased clinic visits group had a statistically significant decreased risk for readmission compared with the usual care group (overall RR 0.78; 95%CI 0.64 — 0.95), with considerable between-study heterogeneity.

One poor quality multi-center study conducted in the UK combined increased clinic visits with at least one home visit by the study nurse and compared with the usual care group.⁶⁵ This study reported an almost 50 percent reduction in the readmissions compared with the usual care group.

Home visits

Four studies that included a total of 633 HF patients compared readmissions in the home visits intervention group with the usual care group.^17,36,49,52 Individually, each of the four studies reported non-significant, lower rates of readmission in the intervention groups. All four were single-center studies conducted outside of the US. Three studies were graded fair quality,^17,49,52 and one study was graded poor quality.³⁶ The meta-analysis of these studies showed a statistically significant reduction in the readmissions in the intervention group compared with the usual care group (pooled RR 0.82; 95%CI 0.69 — 0.97). There was no significant between-study heterogeneity.

Home Telemonitor

Only one study of 426 HF patients evaluated readmissions in the home telemonitor group compared with the usual care group, and it reported no significant reduction in the readmissions between the groups.¹⁶

Multidisciplinary care (MDC)

Four studies that included a total of 1,279 HF patients evaluated readmissions in the multidisciplinary care group compared with the usual care group.^8,32,44,47 Three studies were conducted in the US,^8,44,47 and one study was conducted in Italy.³² All four studies were graded fair quality. All four studies noted decreased readmissions with multidisciplinary care intervention, but statistically significant results were noted in only two studies.^8,32 Two studies followed patients for 3 months,^32,44 and the remaining two studies followed patient for 1-year. ^8,47 The combined estimate in our meta-analysis indicated statistically significantly reduced readmissions in the multidisciplinary care group compared with the usual care group (pooled RR 0.63; 95%CI 0.44 — 0.90). There was significant heterogeneity between studies.

Self-care

Two studies that included a total of 438 HF patients evaluated readmissions with interventions that included increased emphasis on HF care compared with usual care.^41,60 Increased emphases on HF care were provided to patients either during their regularly scheduled visits or through educational software. Both studies were conducted in Sweden; one study was graded fair and the other poor quality. The combined estimate for reduction in readmissions was not significantly different between the groups (pooled RR 0.97; 95%CI 0.83 — 1.14).

Key question 1a: What is the relationship of the study and clinical parameters to the outcome readmission?

When data were available, we analyzed the impact of characteristics such as internal and external validity of the studies on readmission rates. Factors considered for analyses included length of followup, concurrent discharge planning in disease management programs, place of delivery of discharge planning (inpatient, outpatient), components of discharge planning and whether components were individually tailored or generalized, intensity of discharge planning, number and frequency of interventions, and patient characteristics on the outcome of all cause readmissions (Table 1). Among these factors only intermediate- to long-term followup (>6 month or >12 month versus <6 month), interventions initiated in the inpatient setting, and the age of the patient (>75 years) had a statistically significant impact on all cause readmission.

Summary of evidence

• In the meta-analyses, interventions of home visits, increased clinic visits, and multidisciplinary care along with a combination of components of education reinforcement and telephone followup reduced the risk of all cause readmission significantly compared with the usual care group.
• The intermediate- to long-term (>6 month and >12 month) followup interventions initiated in the inpatient setting, and the age of the patient (>75 years) had a statistically significant impact on all cause readmission.

Figure 2. Meta-analyses of the effect of post-discharge care interventions compared with the usual care on readmissions

Table 1. Meta-analyses by subgroups comparing interventions for post-discharge care with the usual care in HF patients

Within each subgroup, categories that are statistically significantly different appear in bold.

Interventions beginning inpatient

Study characteristics

We identified 25 RCTs with a total of 4,795 patients; these trials assessed the effectiveness of non-pharmacological interventions for post-discharge care in HF patients that began in the inpatient setting during an index hospitalization (Table 2.).^{8,15-17,30-50} The majority of the trials compared interventions with usual care.^{8,15,16,30,32-34,36,39-45,47-50} One trial compared 3 months intervention with an extended 6 months intervention.⁴⁰ One study was a 3-arm trial that compared nurse telephone followup, home telemonitor, and the usual care.¹⁶ Sixteen trials included interventions led by nurses only.^{8,15-17,31,33,35-40,43,44,47,48} All or most of the studies included interventions to educate patients about HF symptoms and disease management. However, the studies were unclear if the usual care group received educational information on HF symptoms and disease management similar to the intervention group. Only 10 of 25 studies emphasized exercise education and educational reinforcement programs as part of the education information on HF.^{17,31,32,37,39,40,43,45,46,50} All or most of the studies utilized telephone followup as one of the components in the intervention group. The main care provider either followed patients actively at frequent intervals or the provider was readily available to be contacted during business hours. Additional interventions to follow HF patients included home telemonitor (two studies), home visits (six studies), increased clinic visits (four studies), and multidisciplinary care (four studies). The description of usual care varied among the studies (Appendix B). The duration of interventions ranged from 1 week to 12 months.

The studies followed patients from 3 months to 16 months. Sixteen studies were single-center, and nine studies were multi-center. The studies mostly included patients with a mean age of 65 years and above. Eleven trials were conducted in the US,^{8,15,31,35,37-39,44,46-48} 11 trials in European countries,^{16,30,32-34,36,40-43,45} two trials in Canada,^17,50 and one in Australia.⁴⁹ There were three good (A),^30,34,38 16 fair (B),^{8,15-17, 32, 37, 39,40,42-45,47-50} and six poor (C) studies.^{31,33,35,36,41,46} The six studies were graded poor (C) quality due to various methodological reasons including lack of clear reporting of randomization methods, lack of reporting on the number of patients who met the eligibility criteria were enrolled and randomized, lack of reporting of baseline data, and errors in reporting. Data on available clinical outcomes included readmissions in 25 studies (Figure 3), mortality in 20 studies,^{15,16,30,32-47,49} length of hospital stay in nine studies, cost incurred in 12 studies, and QOL changes in 14 studies. Of the 25 included studies, 11 studies evaluated composite end point of mortality or readmissions as the primary outcome.

Mortality

Twenty studies that began an intervention in the inpatient setting during index hospitalization reported that mortality rates in the intervention group ranged from 2 percent to 31 percent over the study duration of 3 to 16 months (Table 2a).^{15,16,30,32-47,49} The mortality rates in the control or the usual care group ranged from 1 percent to 49 percent over the study duration of 3 to 16 months. Only four studies reported statistically significant decreased relative risk of mortality (0.16 to 0.62) in the intervention group when compared with control group (Table 2a).^15,30,32,42 These studies were conducted in European countries and the US, recruited more than 100 patients, and had a considerable heterogeneity in the duration of followup, interventions utilized (active telephone followup, increased clinic visits, and multidisciplinary care) and care providers (nurses, pharmacists, and cardiologist) (Figure 4). The intervention duration ranged from 6 months to 1-year and the followup duration ranged from 6 months to 1.4 year.

An additional 10 studies reported statistically non-significant decreased mortality rates in the intervention group compared with control group (Figure 4).^{16,34,37-39,43,44,46,47,49} Data with regard to intervention components were assessed across all studies that reported a decrease in mortality. There was no distinct combination of intervention components that were associated with decreased rates of mortality except for one intervention component — active telephone followup — which was utilized along with increased clinic visits, home visits, home telemonitoring or multidisciplinary care. Of note, across studies there was considerable heterogeneity with regard to individual components of intervention, duration of intervention, length of followup, and description of usual care.

Six studies reported a statistically non-significant increased relative risk of mortality (1.07 to 3.5) in the intervention group when compared with control group.^{33,35,36,40,41,45} Five of these six studies were conducted in single-centers; these five studies utilized nurses as the main provider who could be contacted by telephone when needed.^{33,35,36,40,45} Utilizing educational software only as the intervention tool increased the risk of mortality to almost four fold in the intervention group compared with the control group.⁴¹

Length of stay during readmissions

Nine studies reported data on the length of stay as number of days of stay in the hospital per patient in the intervention group compared with the usual care group (Table 2a.).^{8,17,33,35,39,42,45,47,50} Only three of nine studies reported statistically significant decreases in the intervention group compared with the control group.^33,47,50 In these studies, the study quality ranged from fair (B) to poor (C) quality, the recruitment was less than 100 to more than 200 patients, and the length of stay ranged from 3.9 to 6.4 days in the intervention group compared with 6.2 to 11.6 days in the control group. The studies utilizing specialist nurses led one or more of the following interventions: active telephone followup, home visits, increased clinic visits, and multidisciplinary care.

Quality of life

Fourteen studies provided data on patient-perceived health status using one or more validated QOL instruments (e.g., MLHF, SF-36) (Table 2a.).^{8,15,17,30,31,34-38,42,43,45,48} Although the majority of studies reported improved scores in the intervention group during followup compared with baseline scores, only three of 14 studies reported statistically significantly improved QOL scores in the intervention group compared with the control group at followup.^8,17,37 Two studies were single-center, and one was multi-center. All recruited less than 100 patients per group and received a fair (B) quality grade. These three studies utilized one or more of nurse- or multidisciplinary team-led interventions that included active telephone followup, home visits, increased clinic visits, or multidisciplinary care and 3 to 6 months of followup. There was no discernible combination of intervention components when studies reporting significant results were compared with those reporting non-significant results.

Costs

Twelve studies reported quantitative data on the total costs incurred in the intervention group compared with the control group (Appendix C).^{8,30-34,37-39,43,44,49} Two of 12 studies reported statistically significant lower total costs in the intervention group compared with the usual care.^38,44 Studies were graded good (A) orfair (B), followed more than 200 patients from 6 to 12 months, and were conducted in US. The studies utilized one or more nurse-led interventions that included home visits and multidisciplinary care, and education. The remaining studies generally reported statistically non-significant but lower total costs in the intervention group compared with the usual care.^{8,30-34,37,39,43,49} There was no distinct combination of intervention components that was associated with decreased total costs.

Composite outcome of mortality or readmission

Ten studies reported data on the combined endpoint of mortality or readmission. Eight of these reported this endpoint as their primary outcome (Appendix C).^{30, 33-35,37,38,40,41,45,49} Six studies of good (A)^30,34,38 to fair (B)^{37, 40,49} quality reported statistically significant decreased rates for composite outcome of mortality or readmission in the intervention group compared with the control group. All six studies evaluated composite endpoint as their primary outcome of interest. However these six studies varied in the intervention evaluated, were conducted in different countries (US, Europe, and Australia), and randomized almost 50 to 200 patients per group. There was no distinct combination of intervention components that was associated with improved outcome.

Summary of evidence

• Almost three-quarters of the studies with interventions beginning as inpatient were of good (A) to fair (B) methodological quality and wide to moderate applicability to the population of interest.
• The studies compared different combinations of intervention components with usual care group.
• The majority of the studies utilized an education intervention component and active telephone followup.
• Less than one-quarter of the studies utilized intervention components that increase access to care providers. Most of the studies utilized active telephone followup as one of their components of intervention.
• Three-quarters of the studies reported data on mortality. Studies were inadequately powered to ascertain meaningful differences between the interventions and usual care groups for mortality outcomes.
• Less than one-half of the studies reported data on length of stay and quality of life. No studies reported on adverse effects.
• There was considerable heterogeneity across studies with regard to individual components of intervention, duration of intervention, length of followup, and description of usual care.
• Five of 20 studies that initiated interventions in the inpatient setting noted statistically significantly decreased risk of mortality in HF patients who had a broad array of interventions (home visits, increased clinic visits, and multidisciplinary care) compared with the usual care group. All of these interventions utilized telephone followup as one of their components of intervention.
• Interventions of home visits, increased clinic visits, and multidisciplinary care also decreased length of stay, and improved quality of life compared with usual care in several studies.

Table 2. Study characteristics and intervention components of randomized trials beginning inpatient in HF patients (part 1 of 2)

Table 2. Study characteristics and intervention components of randomized trials beginning inpatient in HF patients (part 2 of 2)

Mo, months of followup; N, Total number; Nurse (S), heart failure specialist nurse; wk, week; y, year.
1. When specified nurse; these studies were unclear to whether the nurse had any training in HF care.

Table 2a. Clinical outcomes in HF patients with interventions beginning inpatient versus usual care (part 1 of 2)

Table 2a. Clinical outcomes in HF patients with interventions beginning inpatient versus usual care (part 2 of 2)

CI, confidence interval; HFQ, Chronic Heart Failure Questionnaire; Ctrl, control; d, days; diff, difference; EQOL, Euroqol; Int, intervention; LOS, length of stay; mo, month; MLHF, Minnesota Living with Heart Failure; m, mental score; mo, months of followup; N, number; nd, not documented; ns, not significant; p, physical score; QOL, quality of life; RR, rate ratio; SE, standard error; wk, week; y, year; ↓ Lower score indicates improved function; ↑ Higher score indicates better function

Table 2b. Clinical outcomes in HF patients with interventions beginning inpatient versus usual care (part 1 of 2)

Table 2b. Clinical outcomes in HF patients with interventions beginning inpatient versus usual care (part 2 of 2)

CI, confidence interval; HFQ, Chronic Heart Failure Questionnaire; Ctrl, control; d, days; diff, difference; EQOL, Euroqol; Int, intervention; LOS, length of stay; mo, month; MLHF, Minnesota Living with Heart Failure; m, mental score; N, number; nd, not documented; ns, not significant; p, physical score; QOL, quality of life; RR, rate ratio; SE, standard error; wk, week; y, year; ↓ Lower score indicates improved function; ↑ Higher score indicates better function

Figure 3. Forest plot of readmission risk, intervention components sorted by study quality in interventions beginning in the inpatient compared with the usual care group.

Ed, one or more of the educational component; HTM, home telemonitor; MDC, multidisciplinary care; Active Tel, active telephone followup; Tel cont, Telephone contact.

Figure 4. Forest plot of mortality risk and intervention components sorted by study quality for interventions beginning in the inpatient compared with the usual care group.

Ed, one or more of the educational component; HTM, home telemonitor; MDC, multidisciplinary care; Active Tel, active telephone followup; Tel cont, Telephone contact.

Interventions beginning post-discharge

Study characteristics

We identified 18 RCTs with a total of 3,123 patients that assessed the effectiveness of non-pharmacological interventions for post-discharge care immediately after an index hospitalization in HF patients (Table 3.).^14,18,51-66 Patients were recruited or consented to the trial during their index hospitalization, but interventions began 1 to 3 weeks after their discharge from the hospital. Sixteen trials compared interventions versus usual care.^{14, 18,52-57,59-66} One trial compared home visits intervention with a nurse telesupport group.⁵¹ One 3-arm trial compared nurse telephone followup, home telecare, and usual care.⁵⁸ Twelve trials included interventions led by nurses only.^{51-54,57-61,63-65} Interventions included active telephone followup (eight studies); availability of provider telephone contact (two studies); home telemonitoring (two studies), multiple home visits (six studies), increased clinic visits (two studies), and multidisciplinary care (one study). The description of usual care varied among the studies. The duration of interventions ranged from 2 weeks to 12 months.

The studies followed patients for 6 months to 12 months. In general, a clinic visit for the patients in the intervention group was scheduled at 2 to 3 weeks after discharge. In some studies, the study nurse or pharmacist visited the patients' home following their discharge to provide the first educational intervention. Twelve studies were single-center, and six studies were multi-center. The studies mostly included patients with a mean age of 70 years and above. Six trials were conducted in the US; nine trials were from European countries; one trial was from Canada; and one each from Australia and New Zealand. There were four good (A),^18,62-64 seven fair (B),^{51,52,54,56,57,59,60} and seven poor (C) ^{14,53,55,58,61,65,66} studies. The seven studies that were graded poor (C) quality had various methodological deficiencies, including reporting errors and a failure to clearly report randomization methods, eligibility criteria, number enrolled and randomized, or baseline data.

Data on available clinical outcomes included mortality in 13 studies,^{18,52-54,56-60,63-66} readmissions in 17 studies (Figure 5),^{14,18,51-62,64-66} length of hospital stay in 10 studies,^{14,51,52,57-60,62,64,66} QOL changes in nine studies,^{18,51,56,60-63,65,66} and cost incurred in seven studies. Of the 18 included studies, five studies evaluated composite end point of mortality or readmissions as the primary outcome.^55,59,63-65

Mortality

Thirteen studies that began an intervention in the 1 to 3 week period after discharge from an index hospitalization reported mortality rates in the intervention group ranged from 1 percent to 40 percent (Table 3a.).^{18,52-54,56-60,63-66} The mortality rates in the control or the usual care group ranged from 0 percent to 37 percent over the study duration of 3 to 16 months. Only one study reported statistically significant decreased relative risk of mortality (0.36), and the remainder of the studies reported non-significant decreased risk in the intervention group when compared with usual care group.^52,64 This study was conducted in Europe, recruited less than 100 patients per group, had 1-year of followup, utilized increased clinic visits during their followup, and was graded good (A) quality. The study also utilized nurse-led interventions and reported the primary endpoint as the composite clinical outcome of mortality or readmission.

Two studies reported statistically non-significantly increased relative risk of mortality (1.20, 1.21) in the intervention group when compared with usual care group.^18,60 Of note, these two studies utilized only educational interventions and medication review, and did not utilize interventions of increased access to providers (Figure 6).

Length of stay during readmission

Nine studies reported data on the length of stay during readmission as the number of days of stay in the hospital per patient in the intervention group compared with the usual care group (Table 3a.).^{14,52,57-60,62,64,66} The tenth study compared nurse telemanagement with home visits and provided data on the total number of hospitalization days.⁵¹

Only three studies reported statistically significant decreases in the length of stay in the intervention group compared with the control group.^51,64,66 A good (A) quality study from Sweden with 106 patients reported a significant decrease in the length of stay per patient in the intervention group compared with the usual care group (1.4 versus 3.9 days) during a 1-year followup.⁶⁴ The patients in the intervention group had the nurse-led education along with increased clinic visits during followup. A fair (B) quality study by Benatar that compared nurse telemanagement with home visits reported a significant decrease in the total number of hospitalization days in the nurse telemanagement group during 1-year followup.⁵¹ The third study, graded as poor (C) quality, utilized home visits by study nurses along with multidisciplinary care intervention and reported a significant decrease in the length of stay per patient (9.3 versus 12.5 days) compared with the usual care group.⁶⁶ The three studies utilized different interventions to increase access to care providers, and there was no distinct combination of intervention components that was associated with improved outcomes.

Quality of life

Nine studies provided data on the patient-perceived health status using one or more validated QOL instruments — MLHF, SF-36, and Nottingham health profile (Table 3a.).^{18,51,56,60-63,65,66} Two single-center studies reported significantly improved QOL scores in the intervention group at followup.^56,61 One study was graded fair (B) and the other poor (C) quality. Both studies utilized nurse or multidisciplinary team-led education with medication review for compliance, and followed patients for 6 months. In the intervention group, Ducharme reported significantly improved self assessed scores of MLHF from baseline,⁵⁶ while Morcillo reported significantly higher on the physical and mental health SF-36 scale from baseline.⁶¹ For comparisons between the intervention and usual care group there were no differences between those two groups.

Costs

Seven studies reported data on the total costs incurred in the intervention group compared with the control group (Appendix C).^{14,51,58,59,61-63} Two studies that utilized active telephone followup or education only in the intervention reported statistically significant lower total costs compared with the usual care.^59,61 One study was graded fair (B) and the other poor (C) quality; they followed less than 100 patients for 6 to 12 month, and was conducted in US and Spain. Two other studies that evaluated home visits reported higher total costs in the intervention group, but were not statistically significant compared with usual care or nurse telemanagement.^51,58 The remainder of the studies reported lower costs but were not statistically significant in the intervention group compared with the control group.^14,62,63

Composite outcome of mortality or readmission

Five studies reported the combined endpoint of mortality or readmission as their primary endpoint (Appendix C).^55,59,63-65 Three studies of good (A) or fair (B) quality reported statistically significant decreased risk for composite outcome of mortality or readmission. All three studies varied in the intervention evaluated, were conducted in different countries (US, Europe, and Australia) and randomized less than 100 to 200 patients in each. Across studies, there was no distinct combination of intervention components that was associated with improved outcomes.

Summary of evidence

• Almost two-thirds of the studies with interventions beginning post-discharge were of good (A) or fair (B) methodological quality and wide to moderate applicability to the population of interest.
• The studies compared different combinations of intervention components with usual care.
• The majority of the studies utilized educational intervention components and active telephone followup.
• Less than one-quarter of the studies utilized intervention components that increased access to care providers. Telephone followup was either used alone or in combination with other interventions in most of the studies.
• Across studies there was considerable heterogeneity with regard to individual components of intervention, duration of intervention, length of followup, and description of usual care.
• In only one study, increased clinic visits along with telephone support that was initiated after an index hospitalization statistically significantly reduced mortality in the intervention group compared with usual care group.

Table 3. Study characteristics and intervention components of randomized trials beginning post-discharge in HF patients (part 1 of 2)

Table 3. Study characteristics and intervention components of randomized trials beginning post-discharge in HF patients (part 2 of 2)

Mo, Months ; N, Total number; Nurse (S), heart failure specialist nurse; wk, week; y, year.
1. When specified nurse; these studies were unclear to whether the nurse had any training in HF care.

Table 3a. Clinical outcomes in HF patients with interventions beginning after discharge versus usual care (part 1 of 2)

Table 3a. Clinical outcomes in HF patients with interventions beginning after discharge versus usual care (part 2 of 2)

CI, confidence interval; HFQ, Chronic Heart Failure Questionnaire; Ctrl, control; d, days; diff, difference; Int, intervention; LOS, length of stay; mo, month; MLHF, Minnesota Living with Heart Failure; m, mental score; mo, Months; N, number; nd, not documented; NHP, Nottingham health profile; ns, not significant; p, physical score; QOL, quality of life; RR, rate ratio; SE, standard error; SF, Short form; wk, week; y, year; ↓ Lower score indicates improved function; ↑ Higher score indicates better function.

Figure 5. Forest plot of readmission risk, intervention components sorted by subgroup of study quality in interventions after discharge compared with the usual care group.

Ed, one or more of the educational component; HTM, home telemonitor; MDC, multidisciplinary care; Active Tel, active telephone followup; Tel cont, Telephone contact; * Home visits versus nurse telesupport.

Figure 6. Forest plot of mortality risk and intervention components sorted by study quality in interventions after discharge compared with the usual care group.

Ed, one or more of the educational component; HTM, home telemonitor; MDC, multidisciplinary care; Active Tel, active telephone followup; Tel cont, Telephone contact.

Interventions in the outpatient clinics

We identified six RCTs with a total of 2,654 patients that assessed the effectiveness of interventions among HF patients in the outpatient clinics (Table 4.).^13,67-71 Patients were recruited during their visits to the outpatient clinics, including specialist HF clinics, registered in the national multi-center HF registry, general medicine or geriatric clinics, and academic primary care group practice. In addition, a proportion of patients were recruited from their inpatient setting in two trials.^67,71 Five trials compared interventions versus usual care.^67-71 One trial also utilized some components from the usual care in the intervention group.¹³ Four trials included interventions led by a pharmacist,^67-69,71 and two trials were led by nurses.^13,70 Five studies included interventions to educate patients about HF symptoms and disease management,^13,68-71 and one study utilized only medication review as an intervention.⁶⁷ The studies also evaluated interventions that included education about diet and sodium restriction, self-care behaviors, exercise, and daily weight monitoring. Three studies emphasized interventions on active telephone followup. The studies did not utilize additional interventions to followup HF patients that included home telemonitor, multiple home visits, increased clinic visits, and multidisciplinary care. The description of usual care varied among the studies. The durations of interventions ranged from 6 to 12 months in four studies and were not documented in two studies.

The studies followed patients from 6 to 16 months. The patients in the intervention group had their followup in a clinic managed by a nurse under the supervision of a cardiologist or a physician or in outpatient pharmacies. The primary care physicians and/or physician assistant or nurse practitioner managed the patients in the usual care group. Two studies were single-center, and four were multi-center. The studies mostly included patients with a mean age of 60 years and above. Three trials were conducted in the US, two trials in European countries, and one trial in Argentina. There were three good (A),^13,68,70 two fair (B),^67,69 and one poor (C)⁷¹ quality studies. The poor quality study had errors in reporting, and lacked clear reporting of randomization methods and baseline data.

The severity of HF, LVEF and distribution of NYHA class among included patients varied across studies. All studies restricted the recruitment of patients to those who were ambulant. The most commonly reported etiology of HF and/or coexisting medical disease was ischemic heart disease.

Data on available clinical outcomes included mortality in five studies,^{13,67,68,70,71} number of readmissions in four studies,^67,69-71 cost incurred in one study, QOL changes in four studies,^13,67,70,71 and composite end point of mortality or readmissions in two studies.^13,67 No study reported length of hospital stay during readmissions.

Mortality

Five studies that began an intervention in the outpatient clinics reported mortality rates in the intervention group ranging from 3 percent to 17 percent (Table 4a.).^{13,67,68,70,71} The mortality rates in the control or the usual care group ranged from 5 percent to 21 percent over the study duration. The studies reported a non-significantly decreased relative risk of mortality in the intervention group when compared with usual care group (Figure 7).

Readmissions

Four studies reported data on the number of all cause readmissions that included the number of all cause readmissions and/or patients readmitted (i.e., with at least one readmission) (Table 4a.).^67,69-71 Three studies that followed patients for 6 months to 1-year reported a total of 189 readmissions in 319 patients in the intervention group, compared with 249 readmissions in 322 patients in the usual care group.^67,70,71 One other study reported data on mean readmissions per patient.⁶⁹ Only one study reported a significantly decreased number of readmissions in the intervention group compared with the usual care group.⁷¹ This single-center study was conducted in Ireland, recruited less than 50 patients per group, had 1-year of followup, and was graded poor (C) quality. A pharmacist led the followup interventions in this study that emphasized disease symptoms and management, self-care behaviors, weight monitoring, and medication review.⁷¹

Two studies reported non-significantly decreased rates of readmitted patients in the intervention group (31 percent and 34 percent) compared with usual care group (36 percent and 39 percent) (Figure 8).^13,70

Quality of life

Four studies provided data on the patient perceived health status using one or more validated QOL instruments — MLHF and SF-36 (Table 4a.).^13,67,70,71 Two of the four studies reported significantly improved QOL scores in the intervention group at followup.^13,70 Both were multi-center studies, graded good (A) quality, utilized nurse education with active telephone followup, and followed patients for 1 to 1.4 years. Both studies reported statistically significantly improved QOL in the intervention group compared with the usual care group. Sisk reported significantly improved self-assessed scores of MLHF (38.6 versus 47.3, P<0.05), and higher SF-12 physical scores (39.9 versus 36.3, P<0.05).⁷⁰ The DIAL trial reported significantly improved scores of MLHF during followup in the intervention group compared with the usual care group (30.6 versus 35, P=0.001).¹³

Costs

One study reported quantitative data on the total costs incurred in the intervention group compared with the control group (Appendix C).⁶⁹ This pharmacist-led medication review intervention study reported lower total costs in the intervention group (—$2960 per patient) compared with the usual care.⁶⁹ The multi-center study conducted in the US was graded fair (B) quality and followed more than 300 patients for up to 1-year.

Composite outcome of mortality or readmission

Two studies reported data on the combined endpoint of mortality or readmission (Appendix C).^13,67 Of these, the DIAL trial reported combined endpoint of mortality or readmission as their primary outcome and noted statistically significantly decreased relative risk in the intervention group compared with the usual care group. The trial was conducted in Argentina and randomized 1,518 outpatients with stable HF to an active telephone followup or usual care. The study reported a significant relative risk reduction of 20% (95% confidence interval 3 to 34).¹³

Summary of evidence

• A limited number of studies evaluated interventions beginning in the outpatient clinics.
• Studies compared different combinations of intervention components with usual care.
• The majority of the studies utilized a pharmacist-led intervention that mostly included medication review.
• Across studies, there was considerable heterogeneity with regard to individual components of intervention, duration of intervention.
• The data does not support any firm conclusions with regard to superiority of any particular intervention strategy.

Table 4. Study characteristics and interventions components beginning in the outpatient clinics among HF patients (part 1 of 2)

Table 4. Study characteristics and interventions components beginning in the outpatient clinics among HF patients (part 2 of 2)

N, Total number of patients

Table 4a. Clinical outcomes in HF patients with interventions beginning in the outpatient clinics versus usual care

CI, confidence interval; Ctrl, control; d, days; diff, difference; Int, intervention; LOS, length of stay; mo, month; MLHF, Minnesota Living with Heart Failure; m, mental score; N, number; nd, not documented; ns, not significant; p, physical score; QOL, quality of life; RR, rate ratio; SE, standard error; SF, Short form; wk, week; y, year; ↓ Lower score indicates improved function; ↑ Higher score indicates better function

Figure 7. Forest plot of the mortality risk in studies with interventions in outpatient clinics compared with the usual care group.

Ed, one or more of the education component; Active Tel, Active telephone followup; y, year

Figure 8. Forest plot of readmission risk in studies with interventions in outpatient clinics compared with the usual care group.

Review of recent published systematic review

At least seven systematic reviews and/or meta-analyses have been published since 2004 that reported comprehensive data relevant to the present review.^12,20-25 Because of the span of publication years and variations in the eligibility criteria, the number of studies included in each systematic review ranged from 16 studies involving 1,627 patients (Taylor 2005¹²), to 30 studies involving 8,158 patients (Holland 2005²²). In majority of these systematic reviews, with the exception of Gonseth 2004²¹ and Roccaforte 2005,²⁵ the results were stratified according to either the intervention type or categories of organizational type of post-discharge support.

McAlister 2004 found that the strategies that incorporated followup by a specialized multidisciplinary team (either in a clinic or non-clinic setting) reduced mortality, HF hospitalizations, and all-cause hospitalizations.²³ Interventions that focused on enhancing self-care activities reduced hospitalizations but had no effect on mortality. Telephone followup that advised patients to seek care by their primary care physician in the event of deterioration reduced hospitalization for HF but did not reduce mortality.

Gonseth 2004 reported that strategies within disease management programs, regardless of the type of organizational delivery, whether they were home-based or clinic-based, reduced readmissions for HF and all cause readmissions.²¹ This review included both randomized and non-randomized studies and reported that the disease management programs also reduced the frequency of the combined endpoint of re-admission or death among HF patients.

Phillips 2004 studied only strategies with comprehensive discharge planning that included post-discharge support.²⁴ Each type of support resulted in significantly fewer readmissions except for the strategy of increased clinic visits and frequent telephone contact. However, overall, Phillips found that such strategies for older patients with HF resulted in 25% relative reduction in the risk of readmission, a trend toward 12% relative reduction in all-cause mortality and for a smaller subset of studies, improvement in QOL scores, and no increase in the cost of medical care. In a later meta-regression analysis, Philips 2005 found that "complex programs" that included hospital discharge planning and no delay in post-discharge clinical followup showed a trend toward 70% reduction in risk for first readmission, two fewer days utilized per patient per readmission, and a 70% reduction in risk of HF readmission compared to usual care.¹¹ Less complex programs without hospital discharge planning resulted in less than half the effect in reduction of risk of HF readmission.

A Cochrane systematic review published in 2005 (Taylor 2005) concluded that there was weak evidence that case management interventions are associated with a reduction in admissions for HF, and that it was unclear which components of case management interventions are effective.¹²

In another systematic review published in 2005, Roccaforte reported that mortality and all-cause and HF-related hospitalizations were significantly reduced by interventions to manage HF.²⁵ This review found that high quality studies and programs lasting 3 to 6 months were those most consistently associated with a significant reduction in all outcomes considered. However, a subsequently published RCT, which compared interventions for 3 month with extended 6 month, found no measured clinical advantage in terms of death and/or HF readmission in extending a structured hospital-based disease management program beyond 3 months of discharge.⁴⁰

In a systematic review on remote telemonitoring programs for HF patients, Clark 2007 found that remote monitoring programs for patients with HF reduced admissions to the hospital and all cause mortality by nearly 20 percent while improving health-related quality of life, but had no significant effect on all cause admission to the hospital. Clark 2007 determined that telemonitoring may be more effective at shortening hospital stay than in reducing admissions, since it is likely to produce false alarms and preemptive admissions in patients who are deteriorating but not yet in crisis.²⁰ Moreover, telemonitoring may also lead to early discharge because the patient has a high level of monitoring at home. Clark builds on earlier systematic reviews by McAlister 2004 and Phillips 2004 on multidisciplinary interventions by examining uncertainties relating to the specific effect of telephone-based programs.

The systematic review by Holland 2005 aimed to determine the impact of select intervention components delivered at specific sites on the outcomes.²² The investigators were interested in the relative merits of the site of care, whether the care was delivered in the home (home visits), by telemonitoring, by telephone, in the clinic, or in the primary care physician's office. Meta-analysis showed a significant reduction in all cause readmission, though significant heterogeneity was present. Subgroup analysis showed that home visits reduced all cause readmission to the hospital, but specialty clinic-based interventions had no effect on readmission. Meta-analysis showed a significant reduction in HF readmission, which was notable and similar for home- and telephone-type interventions. Meta-analysis also showed a significant decrease in all cause mortality, especially in the telemonitoring and telephone followup interventions.