Variation in Family Involvement on Rounds Between English-Speaking and Spanish-Speaking Families

This is a cross-sectional observational study performed by secondary database analysis of a longitudinal quality improvement (QI) initiative. The original data were collected via observational audits of family-centered rounds as part of a QI project to increase use of FCR and to standardize FCR components.

The study was conducted at a large urban (413-bed) quaternary children’s hospital located in an ethnically and linguistically diverse county (56% Hispanic, 20% non-Hispanic White, 13% Asian, 8% African American, and 3% multiracial).¹⁹  An estimated 33% of the families cared for by the hospital have a primary language other than English per the admission demographics. Of these, ∼80% speak Spanish.

Family-centered rounds has been the standard model for communication during inpatient admissions at this institution since 2013, although some variability in FCR structure and content has been noted, which was the main impetus for developing the QI project used in this study. All medical teams at our hospital participate in FCR with the exception of oncology. There are set rounding times each morning. Participants in FCR include physicians, bedside nurses, and clinical care coordinators who facilitate discharge planning. The data presented here includes 3 types of team structures: resident general pediatric teams (composed of 1 senior resident, 2 or 3 first-year residents, 1 medical student, and supervised by 1 attending physician from the Division of Hospital Medicine or Division of General Pediatrics), resident subspecialty teams (composed of 1 senior resident, 2 or 3 first-year residents, 1 medical student, and supervised by 1 subspecialty attending physician and fellow), and hospitalist-only/hospitalist comanagement teams (including an attending physician from the Division of Hospital Medicine and often involving pediatric subspecialists in a comanagement model). Subspecialties who supervise resident teams directly include hematology, allergy/immunology, endocrinology, pulmonology, neurology, nephrology, rheumatology, and gastroenterology. Subspecialty groups that have comanagement service lines with the hospitalist division include cardiology, liver transplant, intestinal rehabilitation, and pulmonology.

Multiple QI interventions related to FCR were implemented at our institution before the data collection period of this study. These include educational presentations on FCR (“Rounding Like a Ninja”) given to medical students and residents rotating through the inpatient wards, standardized scripting to encourage family and nursing participation (“Families First, Nurses Next”), and regular FCR auditing with postrounds feedback provided to the resident and hospitalist teams. A pilot program to increase availability of in-person interpreters for rounds was active on 6 of the 7 units where the resident general pediatrics, resident subspecialty, and hospitalist-only/sospitalist comanagement teams round during the study period. These interpreters prioritized rounding with resident teams but also rounded with the hospitalist-only/hospitalist comanagement teams when possible.

For families with LEP, interpretation options include telephone interpretation, video remote interpretation, and in-person interpretation provided by a staff interpreter. Telephonic and video remote interpretation were provided via third-party vendors using a tablet. The choice of interpretation type is generally dependent on which resources are available at that time. At times, members of the medical team speak with families on rounds using Spanish language skills. Per policy, they should only use languages other than English when speaking with patients and families if they have passed a telephone assessment for language proficiency, and they should only interpret for other team members if they have passed at the highest level for language proficiency and taken a course on interpretation best practices. However, data on whether a team member passed the proficiency test was not available.

This study uses data collected by observational audits of FCR using an electronic audit tool developed by expert local consensus. The project started in 2017 but the current version of the instrument, which includes family language and interpretation type, came into use in June 2019. Observers completing the audit include research assistants and faculty members who were involved in the QI project team. Auditors were not part of the care team, did not record any identifiers, and did not interact with the families or teams. Auditors received training on how to use the audit tool and were provided a code book with data definitions. New research staff conducted initial audits with a faculty leader for the present project, and any questions or discrepancies were resolved in weekly meetings. Data were collected via convenience sampling on the basis of the availability of the observers, who generally follow 1 of the medical teams for the duration of rounds for 1 day and provide feedback on FCR after the audit session.

The data used in this study includes FCR encounters from June 2019 until March 2020, when the audits were temporarily paused because of the coronavirus disease 19 pandemic (Fig 1). Inclusion criteria included family participation in rounds and the primary language of the family was English or Spanish. Family participation could be in-person or by telephone and was defined by the audit tool as being “actively engaged and asking/answering questions.” Encounters were excluded if data were not available for all of the FCR components included in the analysis. The project was reviewed by the institutional review board and determined to be an analysis of observational audit data already collected for quality improvement purposes and therefore did not requiring institutional review (IRB CHLA-19-00502).

Encounters for this study were categorized by primary language (English or Spanish, assigned by the audit observer), type of interpretation used, and medical team type (resident general pediatrics, resident subspecialty, and hospitalist-only/hospitalist comanagement).

Outcome measures were chosen from measures on the internal FCR audit tool that were related to family involvement on rounds (Supplemental Fig 2). The primary outcome of interest was whether the rounds were conducted with the patient’s family. This was chosen because the QI team felt this best reflected the principles, as defined by the Institute for Patient and Family-Centered Care’s definition of patient- and family-centered care.²⁰  These include listening to and incorporating patient and family knowledge, values, beliefs, cultural backgrounds, sharing unbiased information to participate in decision-making, and encouraging family participation in care and decision-making at the level they choose. Therefore, rounds conducted in the hallway with the family or in the patient room with family, in which information is shared and plan of care is discussed collaboratively, were considered compliant with this measure. Rounds in which information is shared with the care team only and a plan is made without family presence, even if the family was updated afterward, or table rounds, were considered not compliant. Additional components examined included whether team members were introduced at the start of rounds, whether family questions or concerns were elicited at the beginning and/or at the end of rounds, whether the plan for the day was discussed with the family, and whether anticipated date of discharge (ADOD) and discharge criteria were discussed.

We evaluated family involvement on rounds by comparing the inclusion of relevant FCR components by language type. Patient and caregiver characteristics including patient age, communication status, and family participation in rounds were compared for Spanish-speaking and English-speaking families. The Pearson’s χ² test was performed across different FCR components. The Fisher’s exact test was used when cell counts were low. A multivariable logistic regression model was performed for the primary outcome of interest, family involvement on rounds, using language, team type, and patient age were used as predictors. A 5% level of significance was used for all tests. All statistical analyses were conducted using SAS 9.4.

Additionally, we conducted an E-value analysis, a type of sensitivity analysis that quantifies unmeasured confounding to determine whether unmeasured confounding may have contributed to the observed effects. A relatively low E-value in the context of statistical adjustments made suggests that the results could easily be nullified by a confounder. Conversely, a very high E-value relative to the point estimate may imply that the observed effect is in fact plausible, because the strength and association of the unmeasured confounder with the exposure group and outcome must be very high to negate the observed effect.

Audit encounters were eligible for inclusion if the family participated in rounds (589 of 992, or 59.4% of all encounters) and if data collection for all FCR components of interest had been completed (471 of 589, 80%). Due to limited numbers of families with primary languages other than Spanish (Fig 1), analysis was restricted to Spanish-speaking (133 families) and English-speaking (261 families).

Spanish-speaking and English-speaking families were similar in terms of patient communication status (Table 1). In the Spanish-speaking group, there were more adolescents relative to infants and young children when compared to the English-speaking families, though there were no significant differences between the 2 groups based on communication status of the patient, likely due to neurodevelopmental issues in our medically complex patient population. Almost all families who participated in rounds were present in person (3 English-speaking families and 1 Spanish-speaking family participated via telephone). Both language groups were similarly represented on the different types of medical teams.

A comparison of FCR characteristics by primary language of the caregiver highlighted several key differences (Table 2). Fewer Spanish-speaking families were included in the initial discussion on rounds (64.7%) when compared to English-speaking families (76.3%, P = .017). There were also significant differences in families being asked their questions and concerns at the start of rounds (56.3% for English-speaking families and 44.4% for Spanish-speaking families, P = .025) and whether discharge criteria were discussed with the family (82.8% for English-speaking families vs 72.2% for Spanish-speaking families, P = .018). There were no statistically significant differences in team introduction, discussion of ADOD, discussion of plan for the day, or whether family questions were elicited at the end of rounds.

A separate bivariate analysis was conducted comparing the presence of FCR components by type of interpretation used including a member of care team interpretation, phone interpreter, video remote interpreter, or in-person staff interpreter (see Supplemental Table 5). There were no significant differences present between the interpretation groups for team introductions, discussion of discharge criteria or ADOD, discussion of the plan for the day, involvement of family, or whether the family was asked about questions and concerns at the end of rounds. There was a borderline significant difference in whether family questions were elicited at the start of rounds (P = .05), which occurred most often with the telephone interpretation group.

When FCR practices were compared between 3 team types (Table 3), we found that families were significantly more likely to be involved in the initial discussion on rounds for Hospitalist-only/Hospitalist comanagement services (85.7%) compared to resident general pediatrics teams (72.0%) and resident subspecialty teams (58.0%, P < .01).

We conducted a multivariate logistic regression model to assess for differences in family involvement by language, team type, and age category (Table 4). In the univariate analysis for language, Spanish-speaking families had lower odds of being involved in the initial discussion on rounds compared to English-speaking families (odds ratio [OR], 0.57; 95% CI, 0.36–0.90). There were no significant differences by age group. Families had the lowest odds of being involved in the initial discussion on rounds for subspecialty teams (OR, 0.53; 95% CI, 0.32–0.90) and the highest odds for hospitalist-only/hospitalist comanagement teams (OR, 2.32; 95% CI, 1.19–4.50). In the adjusted model which included team type and age category, the results were similar, reinforcing that Spanish-speaking families had lower odds of being involved in the initial discussion on rounds compared to English-speaking families (OR, 0.56; 95% CI, 0.34–0.90). The findings related to team type also remained consistent, with families having lower odds of being involved in the initial discussion on rounds for subspecialty teams (OR, 0.53; 95% CI, 0.30–0.92) and highest odds of being involved for hospitalist-only/hospitalist comanagement teams (OR, 2.44; 95% CI, 1.25–4.78) compared to general pediatrics teams after adjusting for language and patient age. The E-value analysis suggests moderate confounding.

The key finding of our study was decreased involvement of Spanish-speaking families compared to English-speaking families as measured by family involvement in the initial discussion on rounds. For communication on rounds to be family-centered, the family should be an integral part of the rounding dialogue with the medical team rather than be briefly updated after an internal team discussion. Other differences detected include fewer Spanish-speaking families being asked if they had questions at the start of rounds and less discussion of discharge criteria on rounds for Spanish-speaking families compared to English-speaking families. These may reflect rounds that were truncated or less detailed when conducted with interpretation when compared to rounds conducted in English and conform to trends already noted in the existing literature, such as the exclusion of the family from initial medical discussions and information filtering,¹⁵  relative lack of detail in LEP rounding encounters,²¹  challenges with family empowerment to participate on rounds and to request interpretation services when needed,¹³  and unmet family engagement expectations.²²  Existing studies done with mainly English-speaking families have highlighted potential barriers to family participation on rounds including lack of family orientation and preparation, feelings of intimidation when engaging with a large health care team, use of medical jargon, and the presence of environmental distractions.^17,18  For families whose primary language is Spanish, navigating time constraints on rounds²²  and lack of in-person interpretation availability^2,11,21  may exacerbate these challenges.

Whereas Spanish-speaking families had similar rates of experiencing team introductions, discussing the plan for the day, and of discussing ADOD, these are the components of rounds that are often the most unidirectional types of communication in which the medical team tells the family their names and titles and explains their plans for the hospitalization and expected timeline for discharge. English-speaking families were more likely to be asked if they had questions at the start of rounds than were Spanish-speaking families. This is possibly because unanticipated or challenging questions at the beginning of FCR can shift the focus of the conversation from what was previously planned, and they are often perceived as being more time-consuming. In an inpatient environment in which rounding time is already perceived as a scarce resource and use of interpretation makes rounding encounters with families with LEP take longer than rounds in English, teams may be less likely to ask questions at the beginning of FCR with Spanish-speaking families to be more efficient. This has the potential to cause unintended downstream effects, such as failing to uncover key updates from the patient’s family that should cause a reevaluation of the medical team’s plan for the day or exacerbating misunderstandings between the team and family. Ultimately, when this information is elicited early in rounds, it can save time by creating a shared understanding of the patient’s clinical status at the start of the FCR discussion. Regardless of the relative efficiency of rounds, providing the same types of bidirectional communication for Spanish-speaking families and English-speaking families is essential to provide equitable care for patients and families of all language backgrounds.

As part of a secondary analysis of the impact of team type on FCR, we found that resident subspecialty teams had decreased levels of family involvement compared to resident general pediatrics teams, whereas hospitalist-only/hospitalist comanagement teams were the most likely to involve families in the primary discussion on rounds. This difference was even more pronounced for Spanish-speaking families. To the best of our knowledge, this is the first study to examine differences in practices on FCR for Spanish-speaking families that also incorporates variation by medical team type. It is possible that resident teams, who need to present to a multidisciplinary group that includes a supervising attending, have a different set of challenges and barriers to family-centered communication, such as increased time required for conversation between members of the medical team about potential changes to the medical plan, variation in attending preferences for rounding style, and separate time on rounds devoted to clinical teaching, either in the hallway or at the bedside, that may also decrease the overall time available to communicate with families. Further investigation is needed to better understand the impact of team dynamics on family-centered rounds, as decreased family involvement in rounds may lead to decreased access to information needed for shared decision-making, less rapport-building and alignment with the medical team, and lower levels of patient and family satisfaction. Poorer quality communication could also contribute to an increased risk of adverse outcomes for patients from families with a primary language other than English.¹² 

These subspecialty teams at our institution may face significant time pressure from competing demands, such as the need to round in the ICU, perform scheduled procedures in the operating room, or see patients in the outpatient clinics, and not all subspecialty services have uniformly adopted FCR as their standard model. Certain subspecialties have a preference for engaging in subspecialty-specific education and discussion of the care plan on rounds with resident and fellow physicians that is separate from interaction with families. Further investigation is needed to examine differences in rounding practices between various team types and elucidate specific barriers to providing family-centered care.

Overall, our findings suggest decreased family involvement in FCR for Spanish-speaking families when compared to English-speaking families, but our conclusions are subject to limitations based on the data available from our QI audits. The E-value sensitivity analysis conducted to measure the sensitivity of our findings revealed moderate confounding. Constructs such as interpretation challenges, family health literacy variation, and differences in team dynamics are likely unobservable confounders in our model.

Further work is needed to delineate the specific contributing factors associated with the differences in practice highlighted by this study and determine the most effective approaches to combat health inequity for Spanish-speaking families by promoting access to family-centered care. To date, qualitative studies in the pediatric inpatient setting have found that increasing access to interpretation services, increasing the time allotted for communication on rounds, decreasing nursing ratios and adjusting resident caps for LEP families, providing culturally effective care, and increasing the number of Spanish-proficient physicians are potential areas that can improve communication for Spanish-speaking families.^13,22  Future research should investigate the effectiveness and feasibility of these strategies.

The data used for this study was initially collected as part of a quality improvement-focused audit of FCR practices at our institution, which was associated with significant heterogeneity in the collection of outcomes data as well as limited availability of patient and family demographic information. For encounters in which families participated, approximately 20% had missing data for some or all of the FCR components of interest. We reviewed the encounters with missing data, and there was no difference between auditors, team types, and no changes in the audit instrument or rounding process compared to the encounters for which data on all FCR components of interest was available (Supplemental Table 6). We are unable to assess whether there was a bias by language, because many of these encounters with missing FCR component data have missing primary language data as well.

Family demographics and language data were collected by auditors observing teams on FCR and were not drawn from the electronic medical record or by directly surveying families. This may have led to miscategorization of primary language or families who spoke English or Spanish being categorized as “other” or “unknown” and who were, therefore, excluded from this study erroneously. There were several families (n = 22) where the primary language was listed as Spanish, but no interpretation was used. We elected to include all of these families in the Spanish language group because the auditors noted the family’s primary language in the course of their observations and had no consistent way of assessing language proficiency or understanding of communication on rounds. This limited our ability to analyze differences between the groups on the basis of differences in language comprehension and may have obscured more nuanced differences in FCR practices for Spanish-speaking families with LEP. A comparison was done between these families with Spanish as a primary language in which no interpretation was used (n = 22) and the families with Spanish as a primary language where interpretation was used on rounds (n = 111) and is included as Supplemental Table 7. The only significant difference found in our bivariate analysis was that, when no interpretation was used, fewer families had discharge criteria discussed. The rest of the components were distributed in a way that was fairly similar between the 2 groups. We also attempted to conduct a subanalysis of differences by interpretation type (Supplemental Table 5) but our ability to detect associations was limited by small sample sizes. Quality of interpretation was unable to be assessed, because we were not able to verify that all medical team members who provided interpretation were certified to do so.

The QI project implemented multiple interventions during the time period analyzed by this study (see Methods), including increasing availability of in-person interpretation via a pilot program in which interpreters were assigned time slots to work with specific teams. The QI interventions were designed to decrease disparities by emphasizing a consistent approach to FCR at an institution by using techniques such as a suggested speaking order and standardized scripting; thus, the differences in this study were likely underestimated. Ongoing monitoring of existing interventions is needed to ensure disparities are not exacerbated when new variables are introduced.

Another limitation is that this study was only able to examine differences in practice on FCR where families were present and able to participate in rounds either at the bedside in-person or by telephone. There may be family, occupational, or financial circumstances that disproportionately impact Spanish-speaking families that may make it more challenging for them to be present on rounds that would be unaccounted for in our data. Furthermore, our analysis was limited to encounters where families were “actively engaged” in rounds. This data element was defined by the improvement team with the goal of ensuring that teams engaged families in the discussion and not merely held rounds in their physical presence. Although the majority of encounters excluded for this reason (75.9%) were because families were not physically present, a small number of those excluded were present but did not participate (6.9%). This could potentially lead to bias by disproportionately excluding more Spanish-speaking families if teams did not engage them in the discussion.

Finally, the study was underpowered to detect the significance of some differences for categories with small sample sizes (particularly subtypes of interpretation used, see Supplemental Table 5). This does not mean that these variations are not potentially associated with differences in FCR practices for Spanish-speaking families. Additional research focused on this area in depth will be needed to better understand its impact.

This study builds on existing work that has demonstrated significant disparities faced by Spanish-speaking families or families with LEP on family-centered rounds. Our research confirms overall decreased family involvement in FCR for Spanish-speaking families and reveals significant variability by team type, highlighting the need for further investigation into the relationship between medical team structure and communication practices on FCR especially with regard to subspecialty services. Additional investigation is needed to delineate the specific contributing factors associated with the differences in practice highlighted by this study and determine the most effective approaches to combat health inequity for Spanish-speaking families by promoting access to family-centered care. QI interventions are a promising tool for standardizing FCR within an institution through a targeted, iterative approach, but balancing metrics should be monitored closely to ensure that any changes to rounds ultimately result in improved care for non-English–speaking families and do not exacerbate inequity.

The authors gratefully acknowledge Mei Yu Yeh, MS, for her assistance with biostatistical analysis.