Hospital Discharge Instructions: Characteristics, Accessibility, and National Guideline Adherence

This single-site, retrospective, observational study analyzed discharges from a large, freestanding tertiary children’s hospital in an urban city center from October 1, 2017 to September 30, 2018. The institutional review board for the children’s hospital approved this study. Of ∼12 000 total hospital discharges during that time period, 4% were selected for chart review through a randomized electronic medical record (EMR) query by the clinical data analytics department. This percentage of discharges for review was selected on the basis of the feasibility of each chart reviewer performing 30 reviews. Participants were included if they were admitted under inpatient or observation status to any of the units in the children’s hospital, including medical, subspecialty, surgical, psychiatry, and intensive care areas, to broadly evaluate the diversity of pediatric hospital discharge instructions. Discharges were excluded if the patient was >18 years of age, the time in the hospital was <1 night, discharge was via transfer or death, or if the standard admission health literacy screening was incomplete.

Physician investigators performed a structured chart review to collect demographic and clinical patient characteristics, in addition to discharge instruction characteristics. All reviewers participated in a group training session to standardize the use of the structured chart review tool (Supplemental Information). Demographic variables collected via EMR query included age, sex, race/ethnicity, and primary payer. Race/ethnicity was based on patient or caregiver self-report on admission.

Clinical characteristics included primary diagnosis, medical service (defined as the service caring for the patient at the time of discharge), length of stay, number of medications on discharge, and medical complexity. Medical complexity was defined by using the following categories: no past medical history, simple past medical history (single-system disease), or complex medical history (multiple single-system diseases or serious systemic disease).^2,22  EMR query identified ED visits that were not subsequently admitted and all-cause readmissions within 30 days of discharge.

Discharge instruction characteristics included the number of words, language, Flesch-Kinkaid reading grade level, the inclusion of a commercial template, and presence of SPS key content for discharge instructions. SPS content areas included signs and symptoms of clinical worsening relevant to the patient’s condition, who to contact or where to seek care if worsening, and a reconciled list of medications (see the Discharge Instructions Chart Review form in the Supplemental Information). Commercial templates were defined as those provided by KidsHealth,²³  a standardized template with information about common discharge diagnoses available in English and Spanish (hereafter referred to as templated instructions). These templates were not patient-specific or customizable by providers but allow the option to add additional free text instruction.

Health literacy status was screened by using the Brief Health Literacy Screen (BHLS) screening tool^24,25  as a part of the standard admission process with parents or guardians. It was verbally administered by the nurse who admitted the patient as part of a larger series of admission screenings, and parental responses were recorded in the EMR. Health literacy data were collected as part of the chart review. The BHLS screening tool consists of 3 questions, takes ∼1 minute to administer, and has been previously validated when administered by nurses in clinical settings.²⁵  A health literacy deficit was defined as a BHLS score of ≤9, a cut point established in previous studies for inadequate health literacy.^26,27  Language preference was reported as part of the standard admission process with parents. Limited English proficiency was defined as a preferred language other than English.

Associations between the variables of the number of words in discharge instructions and length of stay and outcomes of adherence to SPS recommendations and BHLS score were evaluated by using the nonparametric Wilcoxon rank test. Associations between categorical clinical factors of interest were analyzed by using χ² tests or Fisher’s exact tests in which >25% of cells had expected counts <5. For the association with adherence to SPS recommendations, these factors were the use of a template for discharge instructions, discharge service, medical complexity, and adherence to SPS recommendations. For the association with BHLS score, race and ethnicity were evaluated.

ED reutilization and readmissions within 30 days were analyzed as a binary indicator variable (any visits versus no visits). Unadjusted associations between categorical clinical factors of interest and any ED visit and any hospital readmission were analyzed by using χ² tests or Fisher’s exact tests in which >25% of cells had expected counts <5. A list of clinical factors of interest are reported in the relevant Tables. Adjusted associations for hospital readmissions were further explored by using multivariable logistic regression modeling. To begin the model selection process, a fully adjusted model input all predictors of interest into a single model. A final reduced model was determined by using stepwise selection. Standard forward and backward model selection techniques were evaluated in a separate sensitivity analysis with all 3 selection methods yielding the same final model. Adjusted analyses were not employed for adherence with SPS recommendations and BHLS literacy score because spurious associations due to confounding were not suspected. Adjusted analyses for having an ED visit within 30 days were not employed because unadjusted analyses did not yield associations requiring further exploration. All reported P values are 2-sided. P <.05 was considered statistically significant.

The EMR query identified 432 discharges, with a total of 240 discharges meeting the criteria for inclusion. A total of 192 discharges were excluded, with 187 excluded because of absent BHLS score on admission screening, 4 excluded because of death at the time of discharge, and 1 excluded for discharge via transfer to another institution. Patient demographics and characteristics of their hospitalization are summarized in Table 1. Participants represented diverse racial and ethnic backgrounds, and after English, Spanish was the most common preferred language. Medically complex patients were the largest group in this sample, followed by simple past medical history, and then no past medical history. A wide variety of discharge services were represented, with subspecialty and general pediatric discharge services being the 2 largest groups.

Twenty-five percent of discharge instructions were missing at least 1 of the 3 content areas recommended by SPS. Information on “signs of worsening” (51/240 [21%]) was missing most frequently, followed by “where to seek help if worsening” (34/240 [14%]), and incomplete medication reconciliation (8/240 [3%]) (Table 2). Information on what to try at home before seeking care, such as advice on postoperative pain management or trying albuterol for an asthma exacerbation, was present in 66 of 240 (28%) discharges. One-quarter of instructions did not include contact information for who to contact if worsening, and when advice on where go to seek help was present, 77 of 206 (37%) only provided instructions to either call 911 or go to the ED.

Templated discharge instructions were used in about one-third of instructions (73/240 [30%]). For general pediatric instructions, 27 of 68 (40%) used templates, for surgical instructions, 6 of 39 (15%) used templates, and for subspecialty instructions, 27 of 133 (20%) used templates. Discharge instructions had a median length of 1361 words.

A health literacy deficit was identified in 35 of 240 (15%) discharged patients. The mean Flesch-Kinkaid grade level of the discharge instructions was 10.1 (range 6.7–12.6). For templated instructions, grade levels ranged from 6.7 to 11.1. All families with health literacy deficits received instructions at higher than a sixth-grade level. Discharge instructions for patient families with health literacy deficits were associated with fewer words compared with those without a deficit (median 1235 vs 1386, P = .03).

Limited English proficiency was reported in 41 of 240 (17%) patient families, with the remainder indicating that English was their preferred language. Of the 41 (17%) of families who preferred a language other than English, 9 (22%) received instructions in their preferred language. All the families with limited English proficiency who received discharge instructions in their preferred language were Spanish speaking, and all of their discharge instructions included a templated portion. None of the non-Spanish-speaking limited English proficiency patients received instructions in their preferred language.

Among all discharged patients, 64 of 240 (27%) had either a health literacy deficit, limited English proficiency, or both. Both a health literacy deficit and limited English proficiency were identified in 11 of 240 (5%) of discharged patients. Of the 64 patients with a health literacy deficit and/or limited English proficiency, only 9 of the limited English proficiency patients received instructions in their preferred language, and all of the patients with a health literacy deficit received instructions at a higher than sixth-grade reading level. Therefore, a total of 55 of 240 (23%) received instructions that were not accessible because of a lack of accommodation for their health literacy needs, language needs, or both.

Instructions adherent to SPS recommendations were associated with a higher number of words (P <.001; Table 3). Length of stay was not significantly associated with SPS guideline adherence. Medical complexity was negatively associated with guideline adherence, with more medically complex patients receiving discharge instructions that were less likely to meet national guidelines compared with previously healthy patients (P = .04). Similarly, patients with a simple past medical history were less likely to meet national guidelines compared with previously healthy patients (P = .02).

The use of a template for discharge instructions was associated with improved SPS guideline adherence (P <.001; Table 3). Adherence to SPS recommendations was also associated with discharge service, with general pediatric services adherent with guidelines in 67 of 68 (99%) cases, surgical services adherence in 29 of 39 cases (76%), and subspecialty services adherent in 32 of 95 cases (34%; P <.0001).

None of the 3 discharge instruction content areas recommended by national SPS guidelines were individually associated with ED reutilization or readmissions (Table 4). In addition, discharge instructions that contained all 3 SPS recommended content areas were not associated with reduced ED reutilization or readmissions.

Health literacy was not associated with ED reutilization or hospital readmission (Table 5). Limited English proficiency was not associated with ED reutilization but was associated with hospital readmission. Health literacy level and grade level of discharge instructions, number of medications, and preferred language other than English or Spanish were associated with hospital readmission in unadjusted logistic regression models; however, health literacy was not associated with hospital readmission in the final adjusted model. Candidate predictors for the adjusted model are reported in Table 6. Terms for preferred language and number of medications met inclusion criteria for the final model and preferred language other than English or Spanish compared with a preferred language of English and number of medications demonstrating a significant association with hospital readmission (Table 6; Supplemental Table 7).

In this retrospective observational study of discharges from a large tertiary care children’s hospital, 25% of discharge instructions did not meet SPS content recommendations, and instructions were frequently missing key information, such as signs of worsening to watch for and where to seek help if needed. When information on where to seek help was present, one-third of the time, the only recommendations were either calling 911 or going to the ED. In addition, only one-quarter of the instructions offered information on what to try at home before seeking care.

There were several factors associated with adherence to SPS recommendations. Discharge instructions provided to general pediatric service patients more frequently met content guidelines, perhaps due to increased utilization of discharge instruction templates, which were also associated with guideline adherence. Templates can be effective tools for both standardizing and improving the content of discharge instructions. Templates can also improve the accessibility of discharge instruction content by making instructions rapidly available at an appropriate reading level in multiple languages.^28,29  Unfortunately, the templated discharge instructions available at our institution were limited by having a greater than sixth-grade reading level and the fact that they were only available in English and Spanish. The use of templated discharge instructions is also generally limited by the institutional financial resources required to purchase commercial templates and the ongoing need for staff to identify language preferences accurately and select the correct set of instructions. Additionally, discharge instruction templates may only be available for common, low-complexity diagnoses, which may explain their more frequent use in the general medicine service compared with other, more specialized services.

Future work improving the quality of templates for children with medical complexity, limited English proficiency, and health literacy deficits may benefit vulnerable populations at risk for worse clinical outcomes such as higher rates of readmission.

Preparing for discharge may be more challenging for medically complex children than for children being discharged after treatment of a single acute diagnosis.^30,31  Despite the role of discharge instructions in communicating essential information regarding medically complex children, discharge instructions for this subpopulation had lower adherence to SPS content recommendations. Increased length of stay was not protective, perhaps because the required instructions are more complex after a longer stay. It may be more difficult to predict the trajectory of hospitalization and discharge needs of children with medical complexity or with longer stays. It may therefore also be difficult to prepare in advance for the more involved task of preparing their instructions.

There was no association between discharge instructions that met SPS content recommendations and ED reutilization or readmission. Nearly a quarter of discharge summaries in this study did not accommodate the patient’s health literacy needs and/or preferred language, making it difficult to fully assess the impact of discharge instruction content and characteristics. Additionally, we found an association in adjusted logistic regression models between limited English proficiency among those preferring a language other than English or Spanish and hospital readmission. None of the patients who preferred a language other than English or Spanish received instructions in their preferred language; however, the small sample size did not allow additional assessment of whether the increased association with readmission was related to the accessibility of their discharge instructions.

Although discharge instruction content and adherence with national recommendations were not associated with a reduction in ED reutilization or hospital readmission in this study, it remains important to find ways to ensure that instructions contain essential information. Key information, including explicit mention of relevant symptoms to watch for and who to contact if needed, should be included in discharge instructions. Validated quality measures for discharge instructions can guide institutional efforts and were associated with improved communication between patient families and their physicians.³²  This information must also be at the appropriate health literacy level and in the appropriate language to improve patient care, reduce postdischarge complications, and provide equitable care. Patient families with health literacy deficits and limited English proficiency are at risk for poor clinical outcomes^{10–12,14–16}  and represent a significant portion of the urban pediatric hospital patient population. The benefits of providing vulnerable families with information regarding what to try at home before seeking care, or where to seek help other than 911 or the ED may have a more meaningful impact on reducing ED visits and readmission in larger studies or when discharge instructions are more accessible. Excessive word count coupled with an incompatible language or health literacy level may hide relevant and helpful information from the patients who need it most. Additional studies are needed to explore the impact of discharge instruction content on clinical outcomes after improvement in accessibility for patients with health literacy deficits and limited English proficiency.

This study was conducted at a single center. However, the study site is an urban academic center with a diverse patient population and >12 000 admissions per year, making it well-suited for exploring the impact of discharge instructions on patients from a variety of backgrounds. This study was also observational in nature, and the results need to be confirmed in larger future studies, particularly to evaluate clinical outcomes such as ED visits and readmissions. Almost one-half of the study population was medically complex, and the all-cause readmission rates were higher than those found in the general pediatric population. This may limit how generalizable our findings are at other institutions. Included diagnoses represented the full spectrum of disease processes at our hospital to explore the diversity of discharge instructions. Future work informed by this study should focus on more specific patient populations and their discharge instruction needs. The study is also limited by the fact that data on ED reutilization and readmissions were only collected from the study site and not from other hospitals in the region. However, our study site is the largest pediatric hospital in the region and is the pediatric transfer center for many local EDs and community hospitals. Another limitation is that we did not identify “dot phrases” or EMR tools used in discharge instructions or any templates other than the commercial KidsHealth template. Our findings may not be applicable to other types of standardized instructions or templates. This study is limited by the fact that we excluded a larger than expected group of patients who did not have health literacy screening completed during their admission. Those patients may have characteristics that precluded participation in health literacy screening, such as parental absence at the time of admission or interpreter services being unavailable for their preferred language, which may have influenced the impact of their discharge instructions on clinical outcomes. Finally, additional work is needed to more richly explore the readability- and health literacy-related characteristics of this work beyond reading grade level.

Discharge instructions are a key element of patient empowerment for the transition from hospital to home, and much remains to be done to ensure that high-quality information is accessible to all. Although the content of hospital discharge instructions in this study of predominately medically complex children needs to be improved, the impact of discharge instructions on clinical outcomes cannot be fully assessed until they better address patient language and health literacy needs. Templates are a promising intervention for ensuring robust and consistent content that can be tailored to the language and literacy needs of patients and potentially increase adherence to national guidelines.

Thank you to Dr Waheeda Samady for her contributions to data collection and conceptualization of the project.