Objectives: Difficulties with transition from inpatient to outpatient care can lead to suboptimal outcomes for patients. We implemented a protocol for systematic follow-up phone calls to families of pediatric patients after discharge, primarily to improve care transition. We also hypothesized that the phone calls would decrease readmissions and emergency department (ED) visits after discharge and improve patient satisfaction.
Methods: We conducted a quasi-experimental study examining the impact of routinely making follow-up phone calls, compared with historical control discharges. We implemented standardized attending physician phone calls to families of all patients discharged from a general pediatric hospitalist service. Calls were made within 72 hours of discharge to assess problems with transition. Charts were reviewed for documentation of difficulty with the care transition, 14-day and 30-day readmissions, ED visits, and Press-Ganey satisfaction scores. All results in the 12 months after the intervention were compared with the preceding12 months.
Results: We reached 78% of all patients’ families by phone after discharge. Of the families reached, 19.9% needed an issue addressed, half of which were medication related. There were improvements in 14-day and 30-day readmissions and 14-day ED visit rates, as well as improvement in patient satisfaction scores, but none of these results reached statistical significance.
Conclusions: Standardized, physician-performed, postdischarge phone calls identified frequent patient care issues related to difficulties with inpatient to outpatient transition, many of which were medication related. However, our study was underpowered to detect a statistically significant correlation with changes in readmission rates, ED visits, or patient satisfaction.
- postdischarge interventions
- postdischarge phone calls
- pediatric population
Transition of care from the inpatient to outpatient setting can prove challenging for pediatric patients and their families. Misunderstanding of discharge instructions, difficulty obtaining new prescriptions, and a variety of other barriers contribute to the possibility of suboptimal patient outcomes. These issues may result in patient harm, readmission to the hospital, or return to the emergency department (ED). A study of 400 consecutive hospitalized adult patients reported a 19% incidence of adverse events after discharge.1 The role of the inpatient team in coordinating and organizing the discharge process is increasingly recognized in the adult and pediatric medical literature.2,3
The rehospitalization of pediatric patients after inpatient treatment has recently gained considerable attention in the medical literature.4–6 This is in large part due to the implementation of provisions in the Affordable Health Care Act penalizing Medicare payments to institutions with high readmission rates for specific diagnoses. Although these policies do not yet apply to the care of children, they have raised the profile for hospital readmission as an indicator of health care quality. Additionally, a recent large study of children’s hospitals showed a wide variation in rate of readmission across institutions,6 further raising the question of the relationship between quality of hospital care and readmission rate. The validity of rehospitalization as a marker of health care quality, as well as the degree to which readmissions are preventable, have been matters of significant debate.7–10
Although previous pediatric studies have examined the discharge process11 and predictors of readmission,4,5,12 previous literature has not directly examined interventions for general pediatric patients in the transition period after discharge from the hospital but before follow-up with their primary care provider. There have been a number of reports of initiatives to decrease readmissions for specific diseases, such as sickle cell disease,13 asthma,14 and bronchiolitis,15 but not for a general pediatric population. A recent systematic review of pediatric hospital discharge interventions included 14 studies, 9 of which studied patients with asthma, 5 studied premature infants, and 1 studied oncology patients.16 In adult patients, a recent systematic review of studies of discharge interventions included 16 randomized controlled trials, 5 of which showed a significant reduction in readmissions.17 All of these studies had multiple interventions, but 4 of the 5 effective randomized controlled trials included both patient-centered discharge instructions and postdischarge telephone calls.
Ideally, interventions would be narrowed to the minimal effort that is effective, and therefore examining an intervention that solely includes a structured postdischarge phone call is important. A 2006 Cochrane review18 concluded that there was insufficient evidence to recommend for or against phone calls after discharge. However, the 33 studies included in the review were performed across a wide degree of disciplines, and only 3 studied a pediatric population. Two of these 3 studies were after ED visits,19,20 and 1 was specifically focused on the use of home apnea monitors.21 None of these addressed the potential impact of postdischarge calls on general pediatric inpatients.
In this article, we describe the implementation and impact of structured attending physician telephone calls to all patient families within 72 hours of discharge from a pediatric inpatient service. We hypothesized that follow-up phone calls would identify a significant number of unresolved issues after discharge, and that by addressing those issues, we would decrease ED use and hospital readmission within 1 month of discharge. Finally, because reaching out to families during the postdischarge period to address any unresolved issues is likely to be perceived positively, we hypothesized that uniformly doing so would improve patient satisfaction with care.
This study was conducted at the University of North Carolina Children’s Hospital, a 150-bed tertiary care hospital, and was approved by the University of North Carolina institutional review board (IRB# 12-1385). Our pediatric hospitalist service is an attending-only service with a mix of general and specialty patients, many of whom travel from across the state for their specialty care. The service has a comanagement relationship with pediatric neurology, directly caring for this group of patients with neurology consultation. Approximately 40% of our patients are admitted with a neurologic diagnosis.
Design and Implementation
We conducted a quasi-experimental study with an intervention and comparison with historical control. On January 1, 2012, our pediatric hospitalist group implemented a policy of calling the families of all patients discharged from our service to assess the transition out of the hospital. Some physicians in the group were already making these calls voluntarily, although the calls were not structured or systematically recorded before the intervention. A sample of 104 discharges before implementation, chosen with a random number generator, showed discharge calls were documented in 50% of patient discharges in the 12 months leading up to study initiation (95% confidence interval [CI] 41%–59%).
Our intervention was to standardize the phone calls with a script, to call every discharged patient, and to document these calls systematically. Our standardized script was created at a monthly pediatric hospitalist meeting with input and consensus from the group. The “script” was a series of prompts for the physician that included questions about the condition of the patient, medication compliance and dosages, whether durable medical equipment was obtained and seemed adequate, and when follow-up appointments were scheduled (Fig 1). Physicians were instructed to use the phone calls as a means to identify any problems, questions, or concerns that had arisen since discharge, and to document thoroughly any issues identified and how they were resolved. The documentation in the medical record system was a standardized note that included the script and the respondent’s answers to the script elements. The call was closed with a reminder that the patient would receive a mailed survey regarding his or her satisfaction with the hospitalization, and a request that it be completed and returned. Calls were intended to be made to all patients within 72 hours of discharge, with up to 3 attempts to contact families. For non–English-speaking families, a medical interpreter was used through a 3-way language line. Our physicians are typically on service for 7 days at a time, so most calls were made by the discharging physician. A running list of discharged patients was kept so that patients discharged on the last day of the service week could be called by the on-coming physician. Aspects of the discharge process, such as whether to make follow-up appointments before discharge, to fill prescription medications before discharge, and/or to make direct physician-to-physician contact with primary medical doctors, were left to the discretion of the discharging physician.
During the entire study period, University of North Carolina Hospitals used a third-party vendor, Press-Ganey Associates, Inc, to distribute patient satisfaction questionnaires to all patients after discharge. Press-Ganey is the nation’s largest vendor of patient satisfaction surveys, partnering with ∼40% of US hospitals.22 This survey includes questions divided into 10 domains, with a maximum score of 100 in each domain. Press-Ganey reports include only numeric scores, which are not linked to individual patients. The only identifying information is the service from which the patient was discharged and the hospital unit on which the patient was roomed. For the purposes of our study, we chose to analyze the domains of satisfaction with the physician and satisfaction with the discharge process, as these were hypothesized as the most likely to be affected by our intervention.
Thirty days after the conclusion of our 1-year study period (January 1, 2012–December 31, 2012), the medical record was reviewed for each patient discharged from our service in 2012 (the 2012 cohort). De-identified patient data were recorded in a spreadsheet for each discharge event (several patients had multiple discharges during the year). Recorded data included inpatient or observation status, whether postdischarge contact was documented, a notation of any problems identified in the follow-up note, and whether an ED visit or readmission at our institution occurred within 14 or 30 days of discharge. Similar de-identified data from the discharges during the previous year from January 1, 2011, through December 31, 2011 (the 2011 cohort), also were recorded for comparison. A single physician conducted the chart reviews. Demographic data, including age, gender, race, primary diagnosis, length of stay, and severity of illness (SOI) score, were collected via query of institutional data provided to the University HealthSystem Consortium, an alliance of nonprofit academic medical centers, including the University of North Carolina. The SOI score is automatically generated by the University HealthSystem Consortium for each admission by using APR DRG Software from 3M (St Paul, MN; http://solutions.3m.com/wps/portal/3M/en_US/Health-Information-Systems/HIS/Products-and-Services/Products-List-A-Z/APR-DRG-Software), with scores ranging from 1 (minor) to 4 (extreme). Finally, numeric scores from Press-Ganey surveys were obtained for the year before and after implementation of our protocol.
Readmission and ED visit rates are reported at the visit level, as some patients would have multiple admissions in the same year. If an ED visit resulted in an admission, both were recorded and counted. All readmissions were counted as new index admissions, and subsequent readmissions were counted from the date of the most recent discharge. Both were recorded and counted in their respective utilization categories. If a patient was discharged directly to inpatient psychiatry or inpatient rehabilitation, he or she was not excluded from the analysis, but the admission to the psychiatry or rehabilitation hospital was not counted as a readmission. All planned or scheduled readmissions, including scheduled infusions or procedures, were excluded from the readmissions or ED visit outcome.
Thirty-day readmission and ED visit rates in 2012 (postintervention) were compared with 2011 (preintervention). Traditional readmission data are recorded as any inpatient admission after discharge from a previous inpatient admission. However, this readmission definition excludes potentially clinically relevant readmissions or ED visits after an observation stay, and it also excludes observation readmissions. Thus, we also included a separate analysis of any unscheduled readmission (inpatient or observation status) that occurred after any inpatient or observation admission (other than for a scheduled same-day infusion monitoring). Given debate surrounding the degree of preventability of later events after pediatric discharge, we also examined 14-day readmission and ED visit data.
After we noted a trend toward lower SOI scores in 2012, a post hoc chart review was performed on the patients with SOI score of 4 to assess 30-day readmission rates in this subgroup of patients.
Patient demographics, frequency of call completion, and content of the follow-up calls are reported descriptively. Simple relative risks (without regression analysis) of readmission or presentation to the ED are reported for the 2012 cohort as compared with the 2011 cohort. We also report absolute readmission and ED visit rates, using a standard z test to compare the 2 independent proportions, with the 2-tailed P value reported. To fully display the data, 30-day readmission rates by month before and after the intervention were plotted on a Shewhart chart and interpreted by using control chart rules.23 We also performed a separate per-protocol analysis to assess for readmission or ED visits in patients who actually received a follow-up phone call in the 2012 study period. Univariate odds ratios (ORs) are reported showing the odds of having received a call in patients who were readmitted or visited the ED, with no regression analysis for potentially confounding variables. Finally, changes in average Press-Ganey scores from 2011 to 2012 are reported, with t test used to assess statistical significance of changes in scores.
In 2011, there were 538 discharges from our service, representing 463 unique patients. In addition, there were 68 discharges from same-day short stays for infusion or postoperative monitoring. In 2012, there were 583 discharges, representing 503 unique patients, and there were 78 discharges from same-day short stays. After making follow-up phone calls part of standard procedure in 2012, 513 of 661 total patients were successfully reached via postdischarge phone call (78%, compared with ∼50% in 2011 preintervention). Patient demographics for these time periods are represented in Table 1. Because designation of race in our electronic medical record is separate from ethnicity, many of those designated as “other” are of Hispanic ethnicity. The 2011 and 2012 cohorts were similar in their demographics and major diagnoses, as well as length of stay and SOI scores. There was a trend toward lower SOI scores in 2012.
Of the 513 patients reached by phone after discharge in 2012, a problem was identified in 102 (19.9%), and approximately half of those were medication related (Table 2). Additionally, 15 (14.7%) of those patients needing issues addressed reported a change in health condition after discharge. One of those patients was referred to the ED and subsequently readmitted. The category “other” consisted of issues such as social concerns, requests for refills on chronic medications, problems with a feeding regimen, or parental questions/concerns about issues that did not fit clearly into one of the other categories. Almost a third of the problems identified (29) required significant action from the physician, such as directly contacting an outpatient provider or specialist, arranging ancillary services, or calling in a new prescription (Table 3).
Despite addressing the wide variety of issues and problems after discharge through telephone calls, there were no statistically significant improvements in utilization or patient satisfaction scores. There was a 17% relative reduction in 30-day readmissions for inpatient discharges in 2012 (10.7%, down from 12.9%, risk ratio = 0.83 [0.59–1.18]). Findings were similar when including observation patients in the analysis, as well as for 14-day readmission rates. We also plotted results for 30-day readmissions on a proportion chart (Fig 2) and did not find evidence for special cause variation with standard Shewhart rules. Similarly, there was no statistically significant change in 14-day readmissions or 30-day or 14-day ED visit rates after discharge (Table 4).
We also examined these same outcomes in a “per-protocol” analysis. Of patients reached by phone after intervention, 10.4% were readmitted within 30 days, compared with 12.0% of patients who were not reached (Table 5). The odds for 30-day readmission were lowered by phone contact (OR 0.86, 95% CI 0.43–1.72) and the 14-day readmission odds were decreased further (OR 0.57, 95% CI 0.27–1.19), although neither reached statistical significance. There was a similar nonsignificant relationship between patients contacted by phone and ED visits (30-day ED visit OR 0.78, 95% CI 0.38–1.58, and 14-day ED visit OR 0.54, 95% CI 0.22–1.35).
Of the 51 patients in 2011 with SOI score of 4, 12 (23.5%) were readmitted within 30 days. This compares with 9 (23.7%) of 38 patients with SOI score of 4 in 2012.
The return rate for Press-Ganey surveys decreased during the intervention period (10.1% vs 12.3%, P = .28). Both satisfaction measures improved, although no individual change reached statistical significance (Table 6).
In this first of its kind study in the general pediatric setting, we examined the effect of postdischarge telephone calls on the transition home after hospital admission. We found that 20% of postdischarge phone calls identified a difficulty in the transition to home care and that this particular intervention allowed active problem-solving between the discharging physician and the patient. These results are similar to the 19% incidence of adverse events reported after hospitalization in adult patients.1 In our study, a large proportion of these transition issues necessitated a significant intervention, such as additional counseling, contacting other providers, arranging urgent reevaluation, or changing or discontinuing a new prescription. These data suggest that postdischarge phone calls can resolve many important issues and thus can be an integral part of improving the delivery of optimal care to the patient. Approximately 10% of patients reached by phone call had an issue related to their prescribed medications, suggesting that involving a clinical pharmacist as part of either the predischarge or postdischarge transition process also would be a high-yield intervention.
It is important to note that there were no other ongoing initiatives to reduce readmissions during the study or control period. With the exception of the change in our protocol, there were no other systematic changes that would be expected to have influenced the results. Most of the physicians involved were staffing the service during the entire 2-year period. The patient population was generally the same from one year to the next, noting the heavy neurology subset in both years. General discharge practices from an administrative and nursing perspective were systematically unchanged.
In addition to identifying numerous important issues necessitating physician action during the calls, we also noted a trend toward the decreased health care utilization after implementing the calls, including a 17% relative reduction in 30-day readmissions. The decreases in ED use and rehospitalization did not reach statistical significance. Because readmission in general is relatively low incidence, and because the readmissions that do occur may not all be preventable, it may be that a single-component intervention such as ours is simply not capable of reducing subsequent care utilization. However, if one presumes the observed decreases could be attributable to our intervention, several factors may have contributed to the lack of a statistical significance. Most importantly, with only ∼500 discharges a year, our study was underpowered to detect small changes in readmission rates. Although we did not perform an a priori power analysis and instead used discrete time periods for analysis, a study to detect a 2.5% drop from a baseline readmission rate of 13% with 95% confidence level and power of 0.80 would require ∼2500 patients per group. Enrolling that number of patients at our institution would occur over years, and introduce other unmeasured variables that could confound the outcomes and interpretation of the results. Additionally, we did not reach all discharged patients after initiating the intervention. Despite up to 3 attempts at postdischarge contact, ∼80% of patients were reached by telephone. Patients who are unable to be reached may be the very group that is most disconnected and at highest risk for adverse outcomes such as readmission. Additionally, several physicians were already making postdischarge phone calls before the intervention officially started: the preintervention rates of phone calls started at 50% and rose to 78% postintervention. Although our contact rate is higher than that achieved in a similar study in an adult population, in which 72% of patients discharged from a general medicine service were successfully contacted after discharge,24 the relatively small absolute increase may have attenuated the effects that the intervention had on utilization outcomes in our study. Repeating this study in a program that currently does not make any attending physician calls to families would better estimate the true magnitude of implementing postdischarge calls.
Our per-protocol analysis is congruent with our primary outcome analysis. As suggested previously, there may be inherent differences in the unreached patient population that also affect utilization rates, but the univariate ORs suggest that patients who were readmitted or visited the ED were less likely to have been reached by follow-up phone call. This is consistent with the theory that addressing issues by follow-up call might reduce the need for subsequent care utilization. The same power limitations (overall low numbers of readmitted or ED-visiting patients) also apply to this analysis.
Our study has some other important limitations. Although we identified a 12-month period before our intervention as a control group, the quasi-experimental design precludes the ability to prove causal relationships. In addition, our outcome measure of readmission rate and ED use included only our institution and thus did not account for any admissions or ED visits to other institutions, a common limitation in studies that use readmission rate as a quality measure of patient care. Having a single chart reviewer introduces that possibility of missing or inaccurate data without a second reviewer for confirmation. However, a single chart reviewer also allows for consistent categorization of the issues identified and documented from follow-up phone calls. Finally, the process of collecting de-identified data resulted in an inability to adjust for potentially confounding variables. Table 1 demonstrates that the 2011 and 2012 cohorts were similar in terms of baseline characteristics, but multivariate analysis may have proven insightful if we had been able to directly link readmission data to demographic and clinical variables, particularly SOI scores. Specifically, a multivariate analysis controlling for SOI may have shown a difference in readmission rates in the subgroup of patients with an SOI score of 4 between 2011 and 2012, when this subgroup was smaller.
The rate of return of the Press-Ganey patient satisfaction surveys decreased after the intervention, even though physicians specifically mentioned the survey as part of the standardized script and asked families to return it. Health care systems are increasingly attuned to patient satisfaction results and are seeking out ways to increase return rates and improve responses. Although we did not demonstrate an improvement in return rates, our findings suggest that postdischarge phone calls may either increase patient satisfaction with the physician or discharge process or shift the return rate toward a more satisfied sample.
Despite growing interest in transition of care between the hospital and outpatient settings, this is the first study to target an inpatient general pediatric population with a postdischarge telephone call. We identified a high proportion of patient families after discharge that had a difficulty during the transition, including medication problems and changes in patient condition. There was a trend toward improvement in ED visits, readmissions, and patient satisfaction scores, although none were statistically significant. These results are encouraging and suggest that pediatric follow-up phone calls need more rigorous study, ideally in a setting in which they are implemented as a completely novel intervention, alone as well as bundled with other interventions to improve care transition.
Dr Michael J. Steiner, Associate Professor of Pediatrics and Division Chief of General Pediatrics and Adolescent Medicine at the University of North Carolina, read through this manuscript and made important suggestions for revision.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- confidence interval
- emergency department
- odds ratio
- severity of illness
- Lye PS
- Feudtner C,
- Levin JE,
- Srivastava R,
- et al
- Gay JC,
- Hain PD,
- Grantham JA,
- Saville BR
- Hain PD,
- Gay JC,
- Berutti TW,
- Whitney GM,
- Wang W,
- Saville BR
- Bardach NS,
- Vittinghoff E,
- Asteria-Peñaloza R,
- et al
- Key-Solle M,
- Paulk E,
- Bradford K,
- Skinner AC,
- Lewis MC,
- Shomaker K
- Auger KA,
- Kahn RS,
- Davis MM,
- Beck AF,
- Simmons JM
- Mistiaen P,
- Poot E
- Chande VT,
- Exum V
- 22.↵Press Ganey. 2014; www.pressganey.com/ourSolutions/patient-voice/census-based-surveying/inpatient.aspx. Accessed July 28, 2014.
- Provost LP,
- Murray SK
- Copyright © 2015 by the American Academy of Pediatrics