Article Text
Abstract
Objective Performance indicators are used to evaluate the quality of healthcare services. The majority of these, however, are derived solely from administrative data and rarely incorporate feedback from patients who receive services. Recently, our research team developed person-centred quality indicators (PC-QIs), which were co-created with patients. It is unknown whether these PC-QIs are associated with unplanned healthcare use following discharge from hospital.
Design A retrospective, cross-sectional study.
Methods Survey responses were obtained from April 2014 to September 2020 using the Canadian Patient Experiences Survey - Inpatient Care instrument. Logistic regression models were used to predict the link between eight PC-QIs and two outcomes; unplanned readmissions within 30 days and emergency department visits within 7 days.
Results A total of 114 129 surveys were included for analysis. 6.0% of respondents (n=6854) were readmitted within 30 days, and 9.9% (n=11 287) visited an emergency department within 7 days of their index discharge. In adjusted models, ‘top box’ responses for communication between patients and physicians (adjusted OR (aOR)=0.82, 95% CI: 0.77 to 0.88), receiving information about taking medication (aOR=0.86, 95% CI: 0.80 to 0.92) and transition planning at hospital discharge (aOR=0.79, 95% CI: 0.73 to 0.85) were associated with lower odds of emergency department visit.
Likewise, ‘top box’ responses for overall experience (aOR=0.87, 95% CI: 0.82 to 0.93), communication between patients and physicians (aOR=0.73, 95% CI: 0.67 to 0.80) and receiving information about taking medication (aOR=0.90, 95% CI: 0.83 to 0.98), were associated with lower odds of readmission.
Conclusions This study demonstrates that patient reports of their in-hospital experiences may have value in predicting future healthcare use. In developing the PC-QIs, patients indicated which elements of their hospital care matter most to them, and our results show agreement between subjective and objective measures of care quality. Future research may explore how current readmission prediction models may be augmented by person-reported experiences.
- Surveys
- Quality measurement
- Patient satisfaction
- Patient-centred care
- Hospital medicine
Data availability statement
Data may be obtained from a third party and are not publicly available.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Patient-centredness is one of the Institute of Medicine’s six key domains of healthcare quality. By proving their feedback, via surveys or other means, patients can provide valuable information to improve the quality of healthcare services. Person-centred quality indicators (PC-QIs) have been developed for this purpose.
WHAT THIS STUDY ADDS
Certain PC-QIs, particularly those which focus on communication between patients and providers, medication information and discharge planning are associated with subsequent unplanned acute care use following discharge. Even when controlling for a variety of clinical and demographic characteristics, poorer scores on these PC-QIs were associated with increased odds of an emergency department visit within 7 days, as well as an unplanned readmission within 30 days of discharge.
HOW THIS STUDY MAY AFFECT RESEARCH, PRACTICE OR POLICY
Our findings suggest that efforts to increase in-hospital communication between patients and providers, and to improve education around medications and discharge processes may reduce unplanned acute care use after discharge. This may provide benefits to the patient and the healthcare system, alike.
Introduction
There are a multitude of accepted measures to evaluate the quality of healthcare services. As per the Donabedian model, these may be categorised into three main groups; structural (eg, number of hospital units using an electronic medical record system), process (eg, per cent of persons with diabetes who received A1C screening) and outcome (eg, per cent of patients with a hospital-acquired infection).1 Many of these measures of quality, however, have been developed in the absence of patients, and thus, may fail to incorporate the viewpoints of the primary recipients of healthcare services. In 2001, the Institute of Medicine presented one of the most widely used frameworks to assess the quality of healthcare services. The framework’s six key domains were that healthcare services should be safe, effective, timely, efficient, equitable and patient centred, with patient-centred defined as ‘providing care that is respectful of, and responsive to individual patient preferences, needs and values and ensuring that patient values guide all clinical decisions’.2
There is now wide recognition that patients can play a key role in evaluating the quality of healthcare services. Patient feedback about their satisfaction and experiences with healthcare services are captured in many settings, using surveys which ask the patient to evaluate multiple aspects of their care. For example, the Consumer Assessment of Healthcare Providers and Systems compliment of surveys are used to evaluate care received in settings which include hospitals, emergency departments and outpatient clinics.3 Although these surveys were developed in extensive consultations with patients, they can be quite lengthy and the items within may be of varying importance to patients.
Recently, our research team developed a comprehensive set of person-centred quality indicators (PC-QIs) using a multiphased approach with health system experts, patients and family members.4 Recognising the need to include items which matter most to patients in the evaluation of healthcare quality, the outcome of this work was a core group of 26 PC-QIs (7 structure, 16 process, 2 outcome and 1 global PC-QI) which aligned with the Donabedian model.4 The PC-QIs can be derived from validated surveys, in this work we used the Canadian Patient Experiences Survey - Inpatient Care (CPES-IC), table 1 includes the PC-QIs and corresponding domains in the CPES-IC.
To date, these PC-QIs have not been evaluated alongside other potential measures of healthcare quality. Therefore, our study aim was to examine the association between the PC-QIs with unplanned healthcare use, specifically emergency department visits and unplanned readmissions following discharge from hospital.
Methods
Data sources and study design
The study employed a retrospective, cross-sectional design, linking patient survey data with emergency department and inpatient hospital records.
Survey data were obtained using a modified version of the CPES-IC—encompassing hospital discharges from April 2014 to September 2020. The CPES-IC is a validated tool, which was developed by Canadian Institute for Health Information (CIHI), in partnership with pan-Canadian stakeholders.5 The CPES-IC is based on the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey, with additional questions developed for the Canadian context. In Alberta, the survey is completed by telephone with adult patients from 2 to 42 days after hospital discharge. The CPES-IC is provincial in scope and is administered for discharges at 93 hospitals in the province using a standard script and responses to frequently asked questions. A random sample of 10% of eligible discharges is obtained from each hospital, with approximately 25 000 surveys completed annually. The survey contains 56 questions which ask respondents to assess many aspects of their care including communication with doctors and nurses, medications, patient/family involvement in care, pain control, the physical hospital environment, coordination of care and discharge planning. Responses to each process-related question are Likert ones (eg, always, usually, sometimes, never), while overall rating questions are on a scale from 0 (worst) to 10 (best).
For the purposes of this study, the research team assigned questions from the CPES-IC survey to the PC-QIs created by Santana et al4 table 1 presents this alignment between the PC-QI framework and associated questions on the CPES-IC. A total of 21 survey questions were mapped onto eight of the existing PC-QIs.
Completed surveys were linked with corresponding inpatient records, from the Discharge Abstract Database (DAD).6 In Alberta, the DAD captures all discharges from hospitals in the province and is coded according to standards established by CIHI. Surveys were also linked with emergency department records, as available from the National Ambulatory Care Reporting System data set.7 Surveys were linked to inpatient and emergency department records using the personal health number and visit dates (eg, admission, discharge).
Variables
The primary outcomes of the study were emergency department visit within 7 days, and unplanned readmission within 30 days of hospital discharge. Each was constructed as a binary outcome (yes/no) for each index hospital discharge. A variety of demographic and clinical variables were included as covariates in our analyses. These included age at time of hospital discharge (less than 55 years, 55–69, 70 years and older), sex, level of education (less than high-school, some high school, high school or equivalent, college certificate/diploma, university, advanced or professional degrees), self-rated mental and physical health (each reported as excellent, very good, good, fair, poor), length of hospital stay (less than 3 days, 3–7, greater than 7 days), admission category (urgent, elective), number of Elixhauser comorbidities8 9 and the month and year of hospital discharge.
Statistical analyses
Survey responses to each question were calculated as percent in ‘top box’, where ‘top box’ corresponded to the most positive answer choice.10 In cases where multiple survey questions aligned with a given PC-QI, the aggregate score for the survey questions was calculated in similar fashion to calculation of HCAHPS domain scores.11
Descriptive statistics were generated for each of the CPES-IC questions examined, as well as age, sex, education and length of hospital stay. Logistic, mixed effect (multilevel) regression models were used to test the association between the PC-QIs and each of the two outcomes (emergency department visit within 7 days, unplanned readmission within 30 days). In all regression analyses, the group of respondents who did not report a ‘top box’ response comprised the reference group. Hospitals with less than 100 total observations across the study period were excluded from analyses. Data linkage and descriptive analyses were performed using SAS V.9.4. Additional data cleaning was performed using R V.4.2.2, while the regression analyses were done using the ‘melogit’ command in Stata MP.
Results
Over the study period (April 2014 to September 2020), a total of 155 094 completed surveys were collected. After applying the inclusion criteria and removing case-incomplete records (those with missing covariates or outcomes), 114 129 cases remained (73.6% of original sample) to form the analytical sample. Among this sample, a total of 11 287 (9.9%) of patients visited the emergency department within 7 days of discharge, while 6854 (6.0%) of patients were readmitted within 30 days of discharge. Descriptive statistics are presented in table 2, according to both outcomes.
Table 3 presents the associations between ‘top box’ responses to the PC-QIs and odds of readmission and emergency department visit. Both crude and adjusted odds are provided, with those not reporting ‘top box’ response as the reference group. In the crude models, ‘top box’ responses were associated with decreased odds of readmission within 30 days of discharge for every PC-QI studied. In the adjusted models, ‘top box’ responses for overall experience (adjusted OR (aOR)=0.87, 95% CI: 0.82 to 0.93), communication between patients and physicians (aOR=0.73, 95% CI: 0.67 to 0.80) and receiving information about taking medication (aOR=0.90, 95% CI: 0.83 to 0.98), were associated with decreased odds of readmission within 30 days of discharge. Like with readmissions, in the crude models, ‘top-box’ responses were associated with decreased odds of visiting the emergency department within 7 days of discharge for every PC-QI studied. In the adjusted models, ‘top box’ responses for communication between patients and physicians (aOR=0.82, 95% CI: 0.77 to 0.88), receiving information about taking medication (aOR=0.86, 95% CI: 0.80 to 0.92) and transition planning at hospital discharge (aOR=0.79, 95% CI: 0.73 to 0.85) were associated with lower odds of visiting the emergency department within 7 days of discharge.
Discussion
Main findings
In this paper, patients indicated which elements of their hospital care matter most to them and our results show agreement between subjective and objective measures of care quality. Higher PC-QI scores were associated with lower odds of a future visit to an emergency department as well as an unplanned hospital readmission. In isolation, all eight PC-QIs examined as part of this study were associated with lower odds for both outcomes. When adjusting for a variety of clinical and demographic factors which have been previously associated with variation in patient experience scores and for accounting for the effects of the full set of indicators, three of the PC-QIs showed lower odds of hospital readmission (overall experience, communication between patients and physicians, receiving information about taking medication), as well as emergency department visit (communication between patients and physicians, receiving information about taking medication, transition planning at hospital discharge).
Our study results are encouraging, and they support the findings of previous studies conducted by our own research team as well as others. In our prior works, we have observed that elements of patient experience from the CPES-IC survey are associated with patient safety indicators12 as well as unplanned hospital readmission.13–15 In similar fashion, Doyle et al conducted one of the largest systematic reviews to examine the potential links between person-reported experiences and other objective measures of healthcare quality.16 In this work, which included a total of 55 studies across various clinical settings, patient groups and disease areas, the authors consistently noted a positive association between subjective ratings of patient experience and objectively-measured health outcomes.16 These outcomes included medication adherence, screening and immunisation uptake and like with our study, measures of healthcare utilisation (eg, hospitalisation, length of stay, primary care visits).16 Although the authors, like us, stress that these associations do not always imply causation, they go on to state that the ‘weight of evidence across different areas of healthcare indicates that patient experience is clinically important’.16 Doyle et al also advocated that the three dimensions of quality (as presented in the ‘Triple Aim’; experience of care, population health, per capita cost)17 should be looked at as a group and not in isolation; a statement that we also strongly endorse from the standpoint of the current study, as well as from our previous work.16
Gilmore et al in a recent narrative review summarised the use of patient surveys from 28 full-texts, highlighting that the majority of documented uses of patient experience surveys were at the micro-level (eg, focus on improving front-line clinical care). Interestingly, few macro-level uses were documented.18 Recently, the 26 PC-QIs used in our study were the subject of an implementation plan within primary care. In this work, which involved 11 primary care providers, patients and quality improvement staff, 5 PC-QIs were prioritised along with 5 corresponding implementation strategies.19 This model of collaboration may be used in other healthcare areas to ensure successful implementation of PC-QIs, leading to improvements which matter to patients, providers and health system partners. Figure 1 provides a visual of a process for using PC-QIs to improve person-centred care.
Strengths and limitations
This study has many notable strengths. The CPES-IC is a validated survey for capturing patient experiences. In addition to the survey itself, Alberta Health Services (AHS) uses a standard script, prompts and answers to frequently asked questions. Interviewers receive training and feedback to ensure that the survey is conducted in a consistent fashion. As the sole provider of hospital care for Alberta’s 4.4 million residents, AHS has a provincial data warehouse which maintains all data sets which were used in the study. This data infrastructure allows for comprehensive, province-wide data linkages which are not possible in other jurisdictions, where care is delivered by multiple providers. The ability to link each survey with its respective inpatient hospital record allowed us to control for many variables at the individual level. The PC-QIs development followed a rigorous process with engagement of stakeholders, community partners and patients.4 This work informs further validation of the PC-QIs.
With respect to limitations, in our sample we observed a 30-day readmission rate of 6.0%. When compared with the overall 30-day readmission rate seen over the past 5 years in our province (9.2–9.5%),20 this suggests that our sample may be in better overall health, when compared with the larger inpatient population. A previous publication by our team has highlighted this, using data from 2011 to 2014.21 Second, although readmissions are portrayed in a negative light in the present work, we acknowledge that not all readmissions may be preventable.22 23 Although we controlled for a variety of clinical and demographic variables, we did not have access to data pertaining to family doctor visits, calls to 811 Health Link (a free health information service available in our province),24 and the nature/primary reason for readmission. Future work could incorporate these aspects. As with any survey, recall bias may be present, as respondents with unplanned health services utilisation may have differentially recalled the experiences of their previous hospitalisation than those who did not. With respect to the survey sampling protocol, the CPES-IC survey is administered across our province in English only, does not sample patients who were hospitalised due to a mental health concern and does not allow for proxy respondents. Given that our findings may not be generalisable to these groups, future research is needed. Our team had completed work using text mining approaches to use patient narratives as a means to address this gap.25 Other limitations that may impact the generalisability of our findings is that some patients may have chosen to not complete the survey due to its length (56 questions), and that Alberta’s population is quite an affluent one with high levels of literacy.26
Conclusions
Our study demonstrates that patient reported experience data/PC-QIs may be able to predict future unplanned healthcare use. The study findings could be used by healthcare systems to inform policies to improve overall experience with care received by hospitalised patients, enhance communication between patients and physicians, bolster information sharing with patients related to their medication regimen and to better support patients in their transition from hospital to home/community living. Addressing these areas could result in tangible reductions in subsequent emergency department and inpatient use; benefitting both patients and healthcare systems. Opportunities to report our results back to stakeholders, including patients, would be greatly appreciated by our research team and may also serve as an educational tool/prompt for patients who find themselves hospitalised in future.
Supplemental material
Data availability statement
Data may be obtained from a third party and are not publicly available.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was approved by University of Calgary Conjoint Health Research Ethics Board (CHREB) - file number REB15-1838. Given the retrospective nature of this secondary data analysis, it was impractical to obtain consent from patients. As such, our research ethics board granted a waiver of consent.
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors All authors have contributed to the conceptualisation of this study. They have reviewed the manuscript and have approved its submission. KK is responsible for the overall content as guarantor. As such, KK accepts full responsibility for the finished work and/or the conduct of the study, had access to the data, and controlled the decision to publish.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.