Effectiveness of Synchronous Postdischarge Contacts on Health Care Use and Patient Satisfaction : A Systematic Review and Meta-analysis

Abstract

Background:

Post-discharge contacts (PDC) after hospitalization are common practice, but their effectiveness in reducing acute care utilization after discharge remains unclear.

Purpose:

To assess the effects of PDC on 30-day emergency department (ED) visits, 30-day hospital readmissions, and patient satisfaction.

Data Sources:

MEDLINE, EMBASE, and CINAHL searched from 2012 to May 25, 2023.

Study Selection:

Randomized and nonrandomized trials of PDC within seven days.

Data Extraction:

Two investigators independently screened articles and assessed risk of bias (ROB). Single reviewers extracted data, with verification by second investigators. We conducted random-effects meta-analyses on outcomes shared by at least three studies and assessed certainty of evidence using Grading of Recommendations, Assessment, Development, and Evaluations framework.

Data Synthesis:

Of 13 included studies (11 randomized trials [RTs]), 12 delivered PDC via telephone. Three of 11 RTs were rated low ROB, with one rated high ROB. Most PDC interventions (n=10) consisted of single telephone contacts, often within three days. Eight studies focused on patients identified as higher risk by the authors. PDC showed no differences in 30-day ED utilization (5 RTs; 3,054 patients; RD 0.00, 95% CI [−0.02 to 0.03]; moderate certainty) or 30-day hospital readmissions (7 RTs; 7,075 patients; RD −0.00, 95% CI [−0.02 to 0.02]; moderate certainty).

Limitations:

Adherence and fidelity to PDC interventions were poorly described; only one study investigated non-telephone PDC.

Conclusions:

PDC within seven days of discharge were not associated with reductions in 30-day ED utilization or readmissions when compared with usual care. Health systems should reconsider the utility of universal PDC, as multifaceted interventions targeting higher-risk patients may be necessary to reduce acute care utilization post-discharge.

Primary Funding Source:

Department of Veterans Affairs (PROSPERO: CRD42023465675)

INTRODUCTION

The time following hospital discharge is recognized as a vulnerable period for patients and is associated with unplanned health care utilization (1–3). Estimated rates of all-cause readmissions within 30 days range from 14% to 23%, with many readmissions noted to occur within the first week and many thought to be potentially avoidable (4–7). Emergency department (ED) visits also are a common occurrence post-hospitalization, with estimates of 20–40% of patients using the ED in the 30 days following discharge (8,9).

Over the past two decades, such post-discharge acute care utilization has prompted an increased focus on transitional care from hospital to home. The Affordable Care Act led to the establishment of the Hospital Readmissions Reduction Program from the Centers for Medicare and Medicaid Services (CMS), which created penalties for hospitals with higher 30-day readmission rates for multiple conditions (10). In 2013, CMS subsequently expanded outpatient billing opportunities with new transitional care management (TCM) billing codes to promote timely outpatient follow-up with primary care, with an established goal to improve post-discharge utilization outcomes. Criteria for TCM billing include communication via direct contact, telephone, or electronic means with patients and/or their caregiver within 2 business days of hospital discharge along with a face-to-face visit within 7–14 days of discharge (11).

Various multifaceted care models have been developed to improve pre- and post-discharge transitional care, such as the discharge process toolkit based on Project RED (Re-Engineered Discharge) by the Agency for Healthcare Research and the Transitional Care Model (12–17). Although some models have included post-discharge components, there is limited information available to assess the direct impact of specific post-discharge contacts (PDCs) on key patient and health system outcomes. PDC efforts require a method of identifying discharged patients, personnel to contact patients or monitor asynchronous messaging, and strategies to resolve issues identified during PDC. One systematic review from 2019 identified eight quantitative studies of nurse telephone PDCs, finding no effect on hospital readmissions. However, five studies were of low methodological quality, none published after 2016, and none explored non-telephone modalities (2). We therefore performed this systematic review to assess the impact of PDCs within seven days on emergency care use, hospital readmission rates, and patient satisfaction. We chose seven days after discharge to align with the theory of TCM billing while permitting flexibility to capture studies of longer PDC periods, as long as the majority of contact was in the first week after discharge.

METHODS

This work is part of a Veterans Health Administration (VHA)–funded report (www.hsrd.research.va.gov/publications/esp). This review addresses the question, “What are the effects of synchronous PDCs on hospital readmission, emergency care use, and patient satisfaction?” We followed a published protocol (PROSPERO: CRD42023465675) and the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) reporting guidelines (18).

Data Sources and Searches

In collaboration with a reference librarian, we searched MEDLINE (via Ovid), EMBASE (via Elsevier), and CINAHL Complete (via EBSCO) from 2012 through May 25, 2023, using a combination of keywords and database-specific subject headings for patient discharge, phone or video, follow-up, readmissions, and ED use (Supplementary Table 1). The searches were independently peer-reviewed by a second librarian using the PRESS checklist (19). Editorials, case reports, letters, comments, and conference abstracts were excluded.

Study Selection

Our prespecified inclusion and exclusion criteria are listed in Supplementary Table 2. Major inclusion criteria were studies assessing the impact of bidirectional PDC interventions from a non-specialist clinical provider to an adult that occurred up to seven days from a hospital discharge and measuring 30-day hospital readmission, 30-day ED use, or patient satisfaction. Interventions could continue beyond seven days, as long at least 50% of the contact was within the initial week (prior to an anticipated TCM clinical visit). We included synchronous communication modalities to align with current and planned practices across VHA. All citations that were classified for possible inclusion based on title and abstract independently by two investigators underwent full-text review. All articles reviewed at the full-text level also were evaluated independently by two investigators, and all articles meeting eligibility criteria at full-text review were included for data extraction. Disagreement was resolved via group consensus or by a senior investigator with content or methodological expertise.

Data Extraction and Quality Assessment

Data elements including descriptors of the study populations, quality elements, interventions, and outcome details were extracted into a Covidence (Veritas Health Innovation, Melbourne, Australia) database by a single investigator and reviewed for accuracy by a second investigator. To better characterize interventions, and in keeping with emerging standards in systematic reviews with intervention complexity, we mapped each included study to a common set of core functions (ie, purpose of the change process) of post-discharge interventions: medication review; symptom monitoring; and coordination of social or health services. Study risk of bias (ROB) was assessed by the revised Cochrane risk of bias for randomized trials and cluster-randomized trials (RoB2) and the Risk of Bias in Nonrandomized Studies of Interventions (ROBINS-I) for nonrandomized studies (20,21). Quality assessment was completed in duplicate by two investigators. Disagreements were resolved by consensus between those two investigators or, as needed, with arbitration by a third.

Data Synthesis and Analysis

We summarized key study characteristics of the included studies, including participant descriptors, intervention characteristics (eg, timing, frequency of planned contacts, content, interventionist), comparator, and outcomes. We considered the feasibility of completing quantitative synthesis with meta-analysis to estimate summary effects given the volume of relevant literature, conceptual homogeneity of the studies, and completeness of results reporting. Factors determining conceptual homogeneity included whether the majority of eventual contact occurred within one week of discharge, whether the content of the contact included at least one of our identified core functions of PDC (i.e., service coordination, monitoring, medication review, psychoeducation about discharge orders, see Supplementary Table 2), and whether outcomes assessed were measuring the same underlying concept (e.g., group mean vs. group proportions) at a similar timepoint. We aggregated outcomes when there were at least 3 studies with the same outcome. Dichotomous outcomes were combined using rate differences and random-effects models as appropriate. Estimates were generated using restricted maximum likelihood estimation, and the Knapp-Hartung approach with ad hoc variance correction was used to adjust the standard errors for more conservative inference (22,23). We evaluated for statistical heterogeneity using visual inspection and used 95% prediction intervals (PIs). Meta-analyses were conducted using the metafor19 package for R (R Foundation for Statistical Computing, Vienna, Austria). We considered subgroup analysis or meta-regression to explore quantitative or qualitative interactions of pre-specified potential effect modifiers deemed important by VA stakeholders (eg, clinical staff initiating the contact, intervention content, timing of intervention). For these, pooled statistics were calculated where two or more studies were summarized. For outcome and intervention categories for which meta-analysis was not feasible, we synthesized data narratively by focusing on identifying patterns in efficacy across included studies, emphasizing evidence from higher quality studies with lower ROB.

The certainty of evidence (COE) was assessed using the approach described by the Grading of Recommendations Assessment, Development, and Evaluation working group (24). A qualitative summary rating was assigned after discussion between four investigators (JCB, SS, AMG, JMG) with either methodologic or content expertise and rated as high, moderate, low, or very low COE.

Role of the Funding Source

The U.S. Department of Veterans Affairs was not involved in the design or conduct of the study, the collection, analysis, or interpretation of the data, or approval of this manuscript.

RESULTS

We identified 104 potentially relevant articles after deduplication and title and abstract screening (Figure 1). Of these, 13 primary studies met eligibility criteria (25–37). Characteristics of included studies are shown in Table 1. One study was a cluster-randomized trial, ten were randomized trials (RTs), one was a nonrandomized trial, and one was an interrupted time-series. One RT was rated as high ROB (32) and three were rated as low ROB (Supplementary Figure 1 and Supplementary Figure 2) (25,28,37). Both nonrandomized studies were deemed serious ROB (33,34). Six studies were conducted in the USA, five in Europe, one in New Zealand, and one in Canada. Eleven studies reported hospitalization outcomes, seven reported ED utilization, four reported composite outcomes of unplanned health care use, and four reported patient satisfaction. The median sample size of included studies was 311 (range: 25–3,054). Eight studies reported they focused on patient populations at elevated medical risk, defined often by comorbid or admission diagnoses and/or age (25,27–31,33,36).

Figure 1. Literature Flow Diagram.

Table 1.

Characteristics of included studies

Study Country Sample Size Design	Population -Percent Female -Percent White	Pre- and Post-Discharge Intervention Description -Comparator	Outcome(s) (Time Point) -Risk of Bias Rating
Bell, 2016 (23) United States 851 Randomized trial	Patients 18 years or older hospitalized for acute coronary syndromes (ACS) and/or acute decompensated heart failure (ADHF), as determined by medical record review conducted by a physician using standard criteria -Percentage not reported -Percentage not reported	Intervention group: Pharmacists reconciled preadmission medications and discharge medications with patients and reported inconsistencies to the medical team prior to discharge. Pharmacists provided tailored counseling, including assessing patient understanding of the medication regimen, barriers to adherence, and troubleshooting barriers during the hospitalization. At discharge, pharmacists provided additional counseling, an illustrated medication schedule with the discharge regimen, and a pillbox, which the patient practiced filling. Pharmacists employed teach-back techniques to ensure understanding. Within 4 days of hospital discharge, study coordinators contacted patients and inquired about general health, symptoms, and any medication-related problems. If issues were detected, the study coordinator contacted pharmacists to provide reinforcement education and to resolve problems. -Usual care/routine discharge	Emergency Department visit (30 days) Hospital readmission (30 days) -ROB rating: Low
Clari, 2015 (24) Italy 219 Randomized trial	Patients between 18 and 80 years old hospitalized for “low- or medium-intensity orthopaedic surgery” (ASA score < 3) -46% female -Percentage not reported	Intervention group: Routine care and instruction for discharge. A follow-up telephone call carried out by a nurse occurred 24 – 96 hours after discharge and was designed to give the nurse the opportunity to assess the overall health of the patient. During this call, the nurse followed standardized questions and recorded whether or not an educational intervention or reinforcement technique was carried out. -Usual care/routine discharge	Patient satisfaction -ROB rating: Some concerns
Danielsen, 2020 (25) Norway 282 Randomized trial	Patients 18 years or older assigned to the following aortic valve replacement (AVR) treatments: First-time isolated AVR, AVR with concomitant coronary artery bypass grafting (CABG), or AVR with concomitant supra-coronary tube graft (SCG). -Percentage not reported -Percentage not reported	Intervention group: Prior to discharge, this group received standard discharge care, including scheduled consultation with the treating surgeon. A project coordinator called intervention patients on days 2 and 9 after discharge. These telephone calls comprised advice on the importance of physical activity in the early rehabilitation phase after AVR, and reminding participants about the availability of 24/7-telephone support to answer questions about their present health condition or instructions. Patients also had access to a 24/7-phone hotline staffed by dedicated and experienced advanced nurse practitioners during the first 30 days after discharge. -Usual discharge care	Hospital readmissions (30 days) -ROB rating: Some concerns
Farris,2014 (26) United States 630 Randomized trial	Patients 18 years or older admitted with diagnosis of hypertension, hyperlipidemia, heart failure, coronary artery disease, myocardial infarction, stroke, transient ischemic attack, asthma, chronic obstructive pulmonary disease or receiving oral anticoagulation. -Percentage not reported -91.4% White	Minimal Intervention Group: - Pharmacist case manager (PCM) verified admission medications with community pharmacy, in addition to medication review by unit pharmacist. - PCM made recommendations to inpatient medical team and educated patient during hospitalization, provided discharge medication counseling and wallet card medication list. Strategies were reviewed to enhance self-management. - No call. Enhanced Intervention Group: In addition to PCM activities described for Minimal Intervention Group, the Enhanced group also received the following after unblinding to PCM at discharge. 1) PCM created discharge care plan and faxed to community physician and pharmacy. 2) PCM phoned patient 3–5 days post-discharge to evaluate adherence and new side effects and answer questions. Report was faxed to community physician and pharmacist if problem noted. Control/Usual Care Group: - Unit pharmacist performed medication review. - Unit nurse provided discharge summary and medication list.	Hospital readmission (30 days and 90 days) Emergency Department visit (30 days and 90 days) -ROB rating: Low
Haag,2016 (27) United States 25 Randomized trial	Previously independently living adults, ages 60 or older, enrolled in a care transitions program (CTP) who were at high risk for an emergency department visit or hospital readmission. -Percentage not reported -Percentage not reported	Intervention group: Pharmacist-led medication therapy management consultation between 3 to 7 business days after discharge. Pharmacists reviewed EMR to complete a comprehensive review of all prescriptions, nonprescriptions, and herbal meds. This review included the identification, resolution, and prevention of drug-related problems including adverse events or the use of inappropriate medication(s). -Usual care group: a nurse practioner home visit within 3 days after discharge to review medication.	Emergency Department visits (30 days) Hospital readmissions (30 days) Composite of both Emergency Department visits and hospital readmissions (30 days) -ROB rating: Some concerns
Lee,2020 (28) United States 2372 Randomized trial	Patients aged 21 years or older hospitalized between January 15, 2017, and March 31, 2018, within Kaiser Permanente Northern California with heart failure (HF); identified by diagnosis codes for HF as the primary hospital problem, or diagnosis code for a HF-related sign or symptom as the primary hospital problem in combination with a HF-specific diagnosis code as a secondary problem -44% female -60.4% White	Intervention group: Telephone appointment - patients were called by a nurse or pharmacist who were previously trained and experienced using a structure HF protocol, including directions for titrating diuretics and other HF-related medications and ordering lab tests based on reported symptoms and vitals. Nurse/pharmacist also had immediate access to supervising physician and could arrange for expedited appointments as necessary. -Usual care group: In-person clinic appointment - scheduled primarily with their primary care physician	Hospital readmission for heart failure (30 days) Hospital readmission for ay cause (30 days) -ROB rating: Some concerns
Lindpaintner, 2013 (29) Switzerland 60 Randomized trial	Patients at “high risk” for adverse events after discharge and with either oral anticoagulation, newly ordered insulin, polypharmacy (defined as > eight regularly used medicines at the time of admission), or new diagnosis requiring four or more long-term medicines. Patients also either lived alone, received home nursing care prior to admission, or required complex wound care. -50% female -Percentage not reported	Intervention group: Prior to discharge the nurse care manager (NCM) conducted a comprehensive structured assessment of symptom burden, prior medication adherence, family caregiving, functional status using the Barthel Index, cognition using a German adaptation of the Mini Mental Test and the Clock-drawing Test, and comorbidity, using the Charlson Comorbidity Index (CCI). The NCM then conferred with ward teams about discharge planning and joined the team for rounds. The NCM called patients with a structured assessment within 24 hours of discharge, evaluating self-efficacy and giving reminders about self-management strategies and follow-up appointments; was available by pager 24/7 for 5 days following discharge; ended the intervention with a home visit and a letter to the primary care physician. The NCM used proprietary case management software (e-case) adapted for the project to collect data and generate correspondence. The individualized interventions emphasized adherence, self-management skills, and extending the network of support available to patient and family caregivers. -Usual care/routine discharge	Hospital readmission (30 days) -ROB rating: Some concerns
Lundby,2020 (30) Denmark 64 Randomized trial	Patients 18 years or older discharged from the gastrointestinal unit with gastrointestinal diseases (inflammatory bowel disease, cancer within the gastrointestinal system, or complex fistulas) -48% female -Percentage not reported	Intervention group: Prepared patient information and performed discharge counseling, medication review, discussion with physician, patient counseling at discharge, medication report to primary care physician, and phone follow-up to the patient 3 days after discharge. The discharge counseling included an updated medication list and a written summary of the counseling for the patient to take home, including a direct phone number to the pharmaconomist performing the counseling. -Usual care	Patient satisfaction -ROB rating: High
Robinson,2015 (31) New Zealand 20682 Interrupted time-series	Patients ages 65 years or older with an acute medical admission that were identified as being “high risk” patients -Percentage not reported -Percentage not reported	Intervention group: Pre-discharge components included nutrition screening and possible referral to a dietitian; allied health review; and discharge medication reconciliation and patient education by a pharmacist. Post-discharge component: a telephone assessment, education, and support by a team of experienced community nurses on the first and third days post-discharge. -No comparator	Hospital readmission (28 days) Emergency Department visit (28 days) -ROB rating: Serious
Sarangarm,2013 (32) United States 279 Nonrandomized trial	All English- or Spanish-speaking patients discharged from an internal medicine team between 8 am and 5 pm Monday through Friday based on pharmacist availability to perform discharge counseling -44% female -89% White	Intervention group: Patients received discharge counseling from a pharmacist including education about medication administration, side effects, and disease state. Pharmacists also completed medication review; identified duplicative, unnecessary, or incomplete therapy; checked for drug interactions; verified formulary drug coverage; and ensured prescription completeness using a standardized checklist to minimize interpharmacist variability during counseling and standardized patient education leaflets. -Usual discharge care	Composite hospital readmissions and Emergency Department visits (30 days) -ROB rating: Serious
Soong,2014 (33) Canada 334 Cluster randomized trial	General medical patients aged 18 and older discharged home after hospitalization -Percentage not reported -Percentage not reported	Intervention group: Patients received a copy of the electronic discharge summary and patient-specific instructions. In addition, the provider reviewed written discharge instructions with the patient and/or caregiver. A non-clinicical patient navigator called patients or caregivers within 3 days following discharge with a minimum of 5 attempts conducted. A standardized intervention phone script soliciting information on general health status, comprehension of discharge instructions, and reinforcing these instructions was read to the patient. The caller utilized a modified teach-back method to educate the patient on discharge instructions, medications, and follow up recommendations. -Usual care	Emergency Department visit (30 days) Hospital readmission (30 days) -ROB Rating: Some concerns
Sorknaes,2013 (34) Denmark 266 Randomized trial	Patients who were at least 40 years old and 1) diagnosed with COPD verified by spirometry 2) admitted with acute exacerbation of COPD (AECOPD) “defined by increased need or medicine and increased dyspnoea, increased expectorate volume or increased coughing” -Percentage not reported -Percentage not reported	Intervention group: Prior to discharge, assessment of inhalation techniques was performed and a decision was made concerning discharge therapies. Daily teleconsultations (initiated within 24 hours of discharge) were conducted by a nurse via video for an average of 7 days post-dsicharge. Telemedicine equipment loaned to patient at discharge allowed patient to measure pulse, oxygen saturation, and spirometry and transmit the information to the nursing providing the intervention. The week after teleconsultations were finished, a telephone follow-up call was made. -Conventional treatment	Hospital readmissions (within 26 weeks) -ROB rating: Some concerns
Yiadom,2020 (35) United States 3054 Randomized trial	All inpatients discharged from a general medicine service -Percentage not reported -Percentage not reported	Intervention group: A semi-structured script assessed patient knowledge of discharge diagnosis and plan, with attention to medication changes, follow-up appointments, and anticipated discharge support services (medication procurement, visiting health assistance, and needed equipment). Patients were asked to “teach back” their discharge plan. Gaps in knowledge or planned care transition supports were identified and addressed as needed. -Usual care	Hospital readmission (30 days) -ROB rating: Low

The predominant modality of delivery for these interventions was telephone (N = 12) (25–35,37); one study employed videoconferencing (36). Studies varied in timing of PDC (range: 24 hours to 7 days post-discharge) with most (N = 9) initiating contact in the first 3 days post-discharge. Personnel involved in the PDC interventions included pharmacists (28–30,32,34), nurses (26,30,31,33,36,37), and non-clinical staff (ie, study coordinators (25,27), patient navigators (35)). Four studies used more than one type of personnel (25,27,28,30). Most PDC interventions (N = 10) (25–27,29–31,33,35,37) consisted of a single telephone contact, with two studies allowing additional patient-driven contact with a provided hotline (27,31). One study had two direct telephone contacts on the first and third day post-discharge (33), and one study used daily videoconference contacts for a range of 5 to 9 days (36). Many studies also had extensive pre-discharge components consisting of enhanced interactions with a pharmacist for medication counseling or discharge planning counseling with hospital providers. Eight interventions reported using a structured protocol with a mix of assessments conducted during the contact (25,27,29–31,35–37). Core functional components of the contacts varied; the most common component across interventions was some type of medication review process (N = 10) (25,26,28–34,37). The second most common component was coordination of services (N = 9) (28,30–37), while seven performed symptom monitoring (25,26,28,30,33,35,36). Only three interventions stated that the contacts included all three core functional components (28,30,33). All interventions used usual care as the comparator, with one study operationalizing usual care as an in-person appointment with a patient’s usual primary care provider (30). Additional details of these interventions are in Supplementary Table 3.

Eleven studies measured all-cause hospital readmissions at approximately 30 days post-discharge (range: 28–30 days) (25,27–31,33–37). Individually, none of these interventions led to significant reductions in 30-day readmission rates relative to usual care. Although PDC interventions and personnel involved varied, seven of the nine RTs were deemed to have sufficient conceptual homogeneity and provided enough information to perform a meta-analysis (25,27–30,35–37). Pooled analysis of 7,075patients demonstrated no significant impact of PDC on 30-day hospital readmissions (RD= −0.00, 95% CI [−0.02, 0.02]). As shown in Figure 2, effect estimates were generally consistent across studies (95% prediction interval [PI] [−0.02, 0.02]). These results were corroborated by the results of the nonrandomized trial and interrupted time-series studies that were not included in the meta-analysis and also found no significant impact on 30-day readmissions (33,34). Similarly, two studies that also looked at disease-specific readmissions did not identify a reduction in readmissions for their studied conditions (30,36). One RT excluded from the pooled analysis reported group means for number of hospital readmissions, in contrast to the other summarized studies reporting group proportions, and found no significant differences (36). The other, small RT (N = 57) excluded from the pooled analysis reported greater 30-day hospitalizations in the PDC group compared to usual care (p = 0.026), though a point estimate and number of readmissions in each group were not provided (31). Only one study performed sub-analyses and found no significant associations in subgroups of sex or age (30).

Figure 2. Effects of post-discharge contact interventions in first seven days on 30-day hospital readmission^a.

^a Abbreviations:

Core functions: MR = medication review; SM = symptom monitoring; SC = service coordination. Y = yes for presence of that function.

IV = inverse variance

Seven studies measured all-cause ED use at approximately 30 days post-discharge (25,28,29,33–35,37). Individually, no included study showed a significant reduction in 30-day ED use relative to usual care. Pooled analysis of the five RTs encompassing 3,054 patients showed no significant difference of PDC on 30-day ED utilization (RD = 0.00, 95% CI [−0.02, 0.03]; 95% PI [−0.02, 0.03]) (Figure 3) (25,28,29,35,37). There was no evidence of statistical heterogeneity across these studies. The interrupted time-series trial and the nonrandomized trial also did not show a significant difference in 30-day ED utilization with PDC interventions (33,34).

Figure 3. Effects of post-discharge contact interventions in first seven days on 30-day emergency department utilization^a.

^a Abbreviations:

Core functions: MR = medication review; SM = symptom monitoring; SC = service coordination. Y = yes for presence of that function.

IV = inverse variance

Four studies measured a composite outcome of 30-day unplanned health care utilizations. Three of these four studies combined hospital readmissions with ED utilization for their composite outcome (25,29,34), while the fourth also included unscheduled clinic visits (28). Results were consistent with the other 30-day utilization outcomes; individually, these studies showed no impact of PDC on a reduction in 30-day unplanned health care use relative to usual care control. Based on the meta-analysis of the three RTs encompassing 1,456 patients, there was no significant difference in the odds of 30-day unplanned utilizations in relation to PDC (RD = 0.01, 95% CI [−0.09, 0.10]; 95% PI [−0.09, 0.10]) (Supplementary Figure 3) (25,28,29).

Four RTs encompassing 3,397 patients measured aspects of patient satisfaction (26,31,32,37), including clarity of information communicated during the PDC (26), overall satisfaction with the post-discharge process (31,32), and patient experiences with the hospital (37). Given the heterogeneity in measurement of patient satisfaction studied, meta-analysis of these results was not performed. Overall patient satisfaction was high across control and PDC groups, with only one small study (N = 60) reporting a significantly higher patient satisfaction in the PDC group (31).

Subgroup meta-analyses performed on intervention characteristics for both 30-day readmission and 30-day ED use are shown in Supplementary Figures 4–6. We did not observe any association of PDC with 30-day readmissions, 30-day ED use, nor 30-day unplanned healthcare utilization by any of the following characteristics: professional background of the person initiating the PDC, timing of the PDC within either three days or seven days, use of a structured assessment in the PDC, and any combination of core functions included in the PDC.

The COE was deemed moderate for findings based on included randomized studies and very low for nonrandomized studies. Based on nine RTs, our findings were graded as moderate COE for no effect of PDC on 30-day hospital readmission (25,27–31,35–37). This category was downgraded only for imprecision related to the overall number of studied patients. The results from the two observational studies that reported outcomes on hospitalization were downgraded to very low certainty for very serious ROB, given serious indirectness as well as imprecision (33,34). Similarly, our results from five randomized trials were graded as moderate COE for no effect of PDC on 30-day ED use, having been downgraded for imprecision, as above (25,28,29,35,37). The results from the two observational studies reporting ED use were rated as very low certainty (33,34). We did not conduct a GRADE evaluation for the composite outcomes of unplanned health care utilization, though the overall patterns of outcomes were similar to those for hospitalization and ED use.

DISCUSSION

Post-discharge patient contacts are frequently incorporated into hospital transitions of care and are a required component of transitional care payment structures under CMS. The promise of these structures is to encourage seamless transfers of information and care responsibilities to outpatient clinicians while potentially avoiding costly subsequent healthcare utilization in acute care settings. Despite the wide deployment of PDC strategies, there remain many unanswered questions about their impact on key health care outcomes. We found no evidence of an effect of PDC within the first seven days on 30-day hospitalization (moderate COE from nine RTs), 30-day ED utilization (moderate COE from five RTs), or composite 30-day unplanned healthcare use in separate meta-analyses. Additionally, we found PDC likely has no effect on patient satisfaction with care. Furthermore, no study meeting eligibility criteria investigated multiple delivery modalities, nor any single modality other than telephone or videoconferencing, making it impossible to address the impact of newer modalities or their combinations.

There are several considerations for interpreting our findings on the lack of impact of PDC interventions. First, pre-discharge planning during inpatient care is a routine procedure in most health systems and has become more robust over time (38,39). These discharge planning procedures vary but generally include medication review and counseling, patient and/or caregiver education, and coordinating care with community healthcare providers (39). In the 13 studies we included, about half described some type of pre-discharge planning protocol. It is likely that similar pre-discharge procedures occurred in additional studies, as this is now considered standard of care and highlighted in the AHRQ Project RED toolkit (15,40). In fact, the vast majority of discharge planning steps in the Project RED toolkit are designated as pre-discharge tasks. Thus, the addition of a single post-discharge contact would be a minor component of a broader discharge planning intervention with little potential to have an isolated impact on hospital readmission or ED utilization.

Second, while most studies reported having a standard protocol for PDC, virtually none rigorously assessed factors influencing intervention effectiveness such as adherence (whether patients actually received the PDC intervention) or fidelity (whether the intervention a patient received followed the intended protocol). Only one large, low ROB randomized study included in this review reported such implementation assessments (37). This study conducted a post-hoc analysis of patients who were reached versus not reached for their telephone-delivered PDC, observing higher rates of hospital readmissions among the patients who were not reached. Another study found adherence to an office visit in the first seven days after discharge was lower than adherence to telephone-delivered PDC (79% vs 92%, respectively) but did not investigate differences in outcomes based on adherence (30).

Next, most PDC interventions included in this review were delivered by telephone. Telephone may be an effective modality for some important post-discharge functions, such as patient education or verification of follow-up appointments. However, it may be less effective for core functions like medication review that may be improved by visualizing medication labels, or symptom monitoring enhanced by visual examination. Only one study compared telephone to in-person PDC, finding no difference in 30-day hospital readmissions, although fewer patients followed-up with an in-person visit (30). This suggests telehealth modalities may be preferred by some patients at some times. While many patients may be likely to engage over telephone, other patients might prefer alternative modalities like text messaging, email, chatbots, or electronic health record smartphone applications. Still others may not have reliable access to a telephone or have contact information that changes frequently, reducing adherence to telephone PDC. Last, most studies included in this review focused on patients identified by the authors as at higher risk for our outcomes based on factors such as age and medical comorbidities such as heart failure. While we are not able to comment on specific subgroups of patients based on our findings, it is likely there are patients who may benefit from more intensive approaches in the transition from hospital to home than can be delivered effectively via a single-contact approach. In fact, there is evidence from studies published prior to our search dates that more intensive transition care interventions with multiple contacts before and after hospital discharge have been effective in reducing 30-day rehospitalization (16,41). As a result, a multimodal contact strategy with frequent interactions may be optimal for navigating this highly vulnerable transition period and would benefit from being studied relative to current standards of care. For example, chatbots or text messaging tools may be able to follow symptoms of interest from even before the time of hospital discharge and communicate concerns to a patient’s outpatient clinical team, while a phone or video PDC with a clinical pharmacist could perform medication reconciliation. Future studies with expanded collection of patient-level characteristics might also be able to harness heterogeneity of treatment effect analysis to predict subgroups of patients most likely to benefit from these efforts (42).

Limitations

It is important to note limitations of both the identified literature and our approach to conducting this review. In addition to those limitations described above around reported adherence and fidelity to the study interventions, many studies were small (median sample size of 311), ‘usual care’ was poorly described, and only one study performed subgroup analyses. Furthermore, social drivers of health factor substantially into unplanned healthcare utilization, and none of our studies explored these topics. Additional research enrolling sufficient numbers of patients from important subgroups (such as by age, social support status, or health literacy) could clarify whether there are patients likely to benefit from single-contact approaches versus more intensive post-discharge approaches. Next, we identified no study meeting our inclusion criteria investigating PDC modalities other than telephone and video, although formats such as email and text messaging were of interest and are being used by health systems. Without high quality studies of such modalities, we are unable to comment on whether these more recent PDC strategies might enhance existing efforts. Our definition of PDC did not include interventions that were centered on electronic symptom or telemonitoring alone, although many of our studies included these strategies. Although both VHA and civilian health systems use such monitoring strategies frequently, such standalone interventions typically do not have the majority of their contact within the first week and may only become bidirectional when a monitoring concern is escalated for clinical review. Thus, we may have missed interventions focused on remote monitoring that might serve a complementary role in a multimodal approach as noted above. Moreover, we were unable to assess the impact of face-to-face visits required for TCM billing or the combination of PDC and these visits. It is possible these strategies, in tandem, may have an effect on our studied outcomes or others of interest to health systems, such as reducing medication errors. Additionally, even though we observed no intervention effects across our studies, even small changes in readmission rates likely would be of value to health systems and may justify the incremental costs of deploying these early interventions or those with more frequent or longer-term contact. Finally, our use of modified Knapp-Hartung estimation of standard errors may have led to overly broad confidence intervals. This approach was chosen to offset the application of random effects modeling with relatively few studies. Given the uniformly small effects observed, we do not believe that our conclusions are overly conservative.

We constrained our review to studies published after 2012 to align with national policy shifts in the use of PDC in the United States. In date-limiting our search, we may have missed some relevant earlier studies; however, changes in standards of care in the discharge process over time may have rendered any preceding positive studies less generalizable. We also limited our eligibility criteria to randomized and EPOC nonrandomized design standards, excluding observational studies. Our findings are consistent with the recent prior systematic review of nurse-led telephone PDC and an earlier study from 2014 on telephone PDC (2,43). These reviews included studies with weaker methodological quality from observational and qualitative designs and those prior to 2011, generally finding no consistent impact of PDC (2,43). Our systematic review extends and strengthens these findings by only including more recent studies with higher-quality designs based on EPOC standards and an exploration of treatment effectiveness based on key intervention characteristics. Although there were too few studies in each subgroup to allow for firm conclusions, the consistency of effects across groups suggests that these study characteristics have little influence on the effects of PDC.

Conclusion

Brief post-discharge follow-up calls are widely used in the United States and elsewhere, yet our review demonstrated little supporting evidence that such brief, often single telephone call follow-ups affect the key health care outcomes of hospital readmissions and ED use at 30 days and patient satisfaction with care. While our review did not find evidence of significant impacts of brief PDC approaches, health care systems should consider the cost effectiveness of these relatively light-touch approaches on costly outcomes such as hospital readmissions. Such considerations of widespread universal brief post-discharge contacts should be balanced with the potential to target investments in more intensive post-discharge approaches focused on patients most likely to benefit from these interventions.