Persistence of High Health Care Costs among VA Patients

Associations Between Social Risks and Primary Care Utilization Among Medically Complex Veterans

BACKGROUND

Social risks contribute to poor health outcomes, especially for patients with complex medical needs. These same risks may impact access to primary care services.

OBJECTIVE

To study associations between social risks and primary care utilization among patients with medical complexity.

DESIGN

Prospective cohort study of respondents to a 2018 mailed survey, followed up to 2 years after survey completion.

PARTICIPANTS

Nationally representative sample of 10,000 primary care patients in the Veterans Affairs (VA) health care system, with high (≥ 75th percentile) 1-year risk of hospitalization or death.

MAIN MEASURES

Survey-based exposures were low social support, no family member/friend involved in health care, unemployment, transportation problem, food insecurity, medication insecurity, financial strain, low medical literacy, and less than high school graduate. Electronic health record-based outcomes were number of primary care provider (PCP) encounters, number of primary care team encounters (PCP, nurse, clinical pharmacist, and social worker), and having ≥ 1 social work encounter.

KEY RESULTS

Among 4680 respondents, mean age was 70.3, 93.7% were male, 71.8% White non-Hispanic, and 15.8% Black non-Hispanic. Unemployment was associated with fewer PCP and primary care team encounters (incident rate ratio 0.77, 95% CI 0.65-0.91; p = 0.002 and 0.75, 0.59-0.95; p = 0.02, respectively), and low medical literacy was associated with more primary care team encounters (1.17, 1.05-1.32; p = 0.006). Among those with one or more social risks, 18.4% had ≥ 1 social work encounter. Low medical literacy (OR 1.95, 95% CI 1.45-2.61; p < 0.001), transportation problem (1.42, 1.10-1.83; p = 0.007), and low social support (1.31, 1.06-1.63; p = 0.01) were associated with higher odds of  ≥ 1 social work encounter.

CONCLUSIONS

We found few differences in PCP and primary care team utilization among medically complex VA patients by social risk. However, social work use was low, despite its central role in addressing social risks. More work is needed to understand barriers to social work utilization.

Labor market costs for long-term family caregivers: the situation of caregivers of persons with spinal cord injury in Switzerland

BACKGROUND

Family members are key in the provision of care to persons facing disability. To undertake the role as caregivers, they face many costs, being the setback in the labor market one of the most relevant.

METHODS

We analyze comprehensive data from long-term family caregivers of persons with spinal cord injury (SCI) in Switzerland. Using information about their working situation before and after becoming caregivers, we estimated the reduction in working hours and the associated income loss.

RESULTS

On average, family caregivers reduced their working hours by about 23% (8.4 h per week), which has a monetary value of CHF 970 per month (EUR 845). Women, older caregivers, and less educated caregivers have a much higher opportunity cost in the labor market: CHF 995 (EUR 867), CHF 1,070 (EUR 932), and CHF 1,137 (EUR 990) respectively. In contrast, family members who care for a person that works have a much lower impact on their working status, CHF 651 (EUR 567). Interestingly, the reduction in their working time is only a third of the extra work they face as caregivers.

CONCLUSION

Health and social systems rely on the unpaid work of family caregivers. To guarantee their long-term involvement, family caregivers need to be recognized for their work and potentially compensated. Without family caregivers, it is very unlikely societies can cope with the increasing need for care, as professional services are limited and expensive.

Association between patient-reported social, behavioral, and health factors and emergency department visits in high-risk VA patients

RESEARCH OBJECTIVE

To identify patient-reported social risk, behavioral, and health factors associated with emergency department (ED) utilization in high-risk Veterans Affairs (VA) patients.

DATA SOURCES

Patient survey, VA, Medicare data.

STUDY DESIGN

Prospective cohort study using multivariable logistic regression to identify patient-reported factors associated with all-cause and ambulatory care sensitive condition (ACSC)-related ED visits among VA patients at high risk for hospitalization or death.

DATA EXTRACTION METHODS

Patient-reported measures derived from a 2018 survey sent to 10,000 VA patients; clinical and demographic characteristics derived from VA data; ED visits derived from VA and Medicare claims.

PRINCIPAL FINDINGS

Among 4680 survey respondents, 52.5% and 16.3% experienced an all-cause or ACSC-related ED visit in the following year, respectively. An ED visit was more likely among individuals with functional status limitations (6.0% points (Confidence Interval [CI] 0.017-0.103)) and transportation barriers (5.2% points [CI 0.005-0.099]). An ACSC-related ED visit was more likely among individuals with functional status limitations (3.2% points [CI 0.003-0.062]) and self-rated poorer health (7.4% points (CI 0.030-0.119) poor; 6.2% points (CI 0.029-0.096) fair; 4.1% points (CI 0.009-0.073) good; compared with excellent/very good).

CONCLUSIONS

Patient-reported factors not present in most electronic health records were significantly associated with future ED visits in high-risk VA patients.

Hospital variability in modifiable factors driving coronary artery bypass charges

OBJECTIVE

Coronary artery bypass grafting is associated with significant interhospital variability in charges. Drivers of hospital charge variability remain elusive. We identified modifiable factors associated with statewide interhospital variability in hospital charges for coronary artery bypass grafting.

METHODS

Charge data were used as a surrogate for cost. Society of Thoracic Surgeons data from Maryland institutions and charge data from the Maryland Health Care Commission were linked to characterize interhospital charge variability for coronary artery bypass grafting. Multivariable linear regression was used to identify perioperative factors independently related to coronary artery bypass grafting charges. Of the factors independently associated with charges, we analyzed which factors varied between hospitals.

RESULTS

A total of 10,337 patients underwent isolated coronary artery bypass grafting at 9 Maryland hospitals from 2012 to 2016, of whom 7532 patients were available for analyses. Mean normalized charges for isolated coronary artery bypass grafting varied significantly among hospitals, ranging from $30,000 to $57,000 (P < .001). Longer preoperative length of stay, operating room time, and major postoperative morbidity including stroke, renal failure, prolonged ventilation, reoperation, and deep sternal wound infection were associated with greater hospital charges. Incidence of major postoperative events, except stroke and deep sternal wound infection, was variable between hospitals. In a univariate linear regression model, patient risk profile only accounted for approximately 10% of statistical variance in charges.

CONCLUSIONS

There is significant charge variability for coronary artery bypass grafting among hospitals within the same state. By targeting variation in preoperative length of stay, operating room time, postoperative renal failure, prolonged ventilation, and reoperation, cardiac surgery programs can realize cost savings while improving quality of care for this resource-intense patient population.

Tablet distribution to veterans: an opportunity to increase patient portal adoption and use

OBJECTIVE

Examine whether distribution of tablets to patients with access barriers influences their adoption and use of patient portals.

MATERIALS AND METHODS

This retrospective cohort study included Veterans Affairs (VA) patients (n = 28 659) who received a VA-issued tablet between November 1, 2020 and April 30, 2021. Tablets included an app for VA's My HealtheVet (MHV) portal. Veterans were grouped into 3 MHV baseline user types (non-users, inactive users, and active users) based on MHV registration status and feature use pre-tablet receipt. Three multivariable models were estimated to examine the factors predicting (1) MHV registration among non-users, (2) any MHV feature use among inactive users, and (3) more MHV use among active users post-tablet receipt. Differences in feature use during the 6 months pre-/post-tablet were examined with McNemar chi-squared tests of proportions.

RESULTS

In the 6 months post-tablet, 1298 (8%) non-users registered for MHV, 525 (24%) inactive users used at least one MHV feature, and 4234 (46%) active users increased feature use. Across veteran characteristics, there were differences in registration and feature use post-tablet, particularly among older adults and those without prior use of video visits (P < .01). Among active users, use of all features increased during the 6 months post-tablet, with the greatest differences in viewing prescription refills and scheduling appointments (P < .01).

CONCLUSION

Providing patients who experience barriers to in-person care with a portal-enabled device supports engagement in health information and management tasks. Additional strategies are needed to promote registration and digital inclusion among inactive and non-users of portals.

Drivers of high-cost persistence in rural China: A population-based retrospective study

Purpose

High-cost patients account for over 70% of total health expenditures in rural China and have become a key focus of health insurers. Persistently high-cost patients constitute a substantial proportion of medical resources. Hence, exploring high-cost persistence (HCP) and what drives it is considered meaningful and necessary.

Patients and methods

A population-based retrospective study was carried out. The annual healthcare utilization data of Dangyang New Rural Cooperative Medical Scheme from 2012 to 2017 were analyzed. Patients in the top 10% of spending in a given year were considered high-cost patients. Persistence level was estimated using Markov matrices. A total of 19,405 patients categorized as high-cost patients in 2016 were divided into two groups according to whether or not they kept high-cost status in 2017. Finally, a multilevel logistic regression model was used in examining the determinants of HCP.

Results

On average, about 31.48% of high-cost patients each year still maintained high-cost status in the subsequent year from 2012 to 2017. The elderly (OR = 2.150), families with more non-labor members (OR = 2.307), families applying for subsistence allowances (OR = 1.245), and patients with blood and immune diseases (OR = 2.614) or malignant tumors (OR = 2.077) were more likely to maintain high-cost status. Hospitalization frequency was found to be a mediator.

Conclusion

About one-third of high-cost patients in a given year had persistently high cost status in the subsequent year. Health status and family support were considered the main drivers of HCP. High inpatient service utilization as a mediator was a prominent manifestation of persistently high-cost patients. The accurate identification of persistently high-cost patients is the basis for our management.

Association Between Mental Health Conditions and Outpatient Care Fragmentation: a National Study of Older High-Risk Veterans

BACKGROUND

Healthcare fragmentation may lead to adverse consequences and may be amplified among older, sicker patients with mental health (MH) conditions.

OBJECTIVE

To determine whether older Veterans with MH conditions have more fragmented outpatient non-MH care, compared with older Veterans with no MH conditions.

DESIGN

Retrospective cohort study using FY2014 Veterans Health Administration (VHA) administrative data linked to Medicare data.

PARTICIPANTS

125,481 VHA patients ≥ 65 years old who were continuously enrolled in Medicare Fee-for-Service Parts A and B and were at high risk for hospitalization.

MAIN OUTCOME AND MEASURES

The main outcome was non-MH care fragmentation as measured by (1) non-MH provider count and (2) Usual Provider of Care (UPC), the proportion of care with the most frequently seen non-MH provider. We tested the association between no vs. any MH conditions and outcomes using Poisson regression and fractional regression with logit link, respectively. We also compared Veterans with no MH condition with each MH condition and combinations of MH conditions, adjusting for sociodemographics, comorbidities, and drive-time to VHA specialty care.

KEY RESULTS

In total, 47.3% had at least one MH condition. Compared to those without MH conditions, Veterans with MH conditions had less fragmented care, with fewer non-MH providers (IRR = 0.96; 95% CI: 0.96-0.96) and more concentrated care with their usual provider (OR = 1.08 for a higher UPC; 95% CI: 1.07, 1.09) in adjusted models. Secondary analyses showed that those with individual MH conditions (e.g., depression) had fewer non-MH providers (IRR range: 0.86-0.98) and more concentrated care (OR range: 1.04-1.20). A similar pattern was observed when examining combinations of MH conditions (IRR range: 0.80-0.90; OR range: 1.16-1.30).

CONCLUSIONS

Contrary to expectations, having a MH condition was associated with less fragmented non-MH care among older, high-risk Veterans. Further research will determine if this is due to different needs, underuse, or appropriate use of healthcare.

Disparities in virtual cardiology visits among Veterans Health Administration patients during the COVID-19 pandemic

Objective

In response to the coronavirus disease 2019 (COVID-19) pandemic, the Veterans Health Administration (VA) rapidly expanded virtual care (defined as care delivered by video and phone), raising concerns about technology access disparities (ie, the digital divide). Virtual care was somewhat established in primary care and mental health care prepandemic, but video telehealth implementation was new for most subspecialties, including cardiology. We sought to identify patient characteristics of virtual and video-based care users in VA cardiology clinics nationally during the first year of the COVID-19 pandemic.

Materials and Methods

Cohort study of Veteran patients across all VA facilities with a cardiology visit January 1, 2019-March 10, 2020, with follow-up January 1, 2019-March 10, 2021. Main measures included cardiology visits by visit type and likelihood of receiving cardiology-related virtual care, calculated with a repeated event survival model.

Results

416 587 Veterans with 1 689 595 total cardiology visits were analyzed; average patient age was 69.6 years and 4.3% were female. Virtual cardiology care expanded dramatically early in the COVID-19 pandemic from 5% to 70% of encounters. Older, lower-income, and rural-dwelling Veterans and those experiencing homelessness were less likely to use video care (adjusted hazard ratio for ages 75 and older 0.80, 95% confidence interval (CI) 0.75-0.86; for highly rural residents 0.77, 95% CI 0.68-0.87; for low-income status 0.94, 95% CI 0.89-0.98; for homeless Veterans 0.85, 95% CI 0.80-0.92).

Conclusion

The pandemic worsened the digital divide for cardiology care for many vulnerable patients to the extent that video visits represent added value over phone visits. Targeted interventions may be necessary for equity in COVID-19-era access to virtual cardiology care.

The economic value of long-term family caregiving. The situation of caregivers of persons with spinal cord injury in Switzerland

Health systems rely on the unpaid work of family caregivers. Nevertheless, demographic changes suggest a shortage of caregivers in the near future, which can constrain the long-term care policy in many countries. In order to find ways to support family caregivers, a primary effort would be to estimate how much their work is worth. This paper estimates the economic value of long-term family caregivers and how these costs would be shared by the health system, the social insurances and the cared-for person in the absence of informal caregivers. We use data of 717 family caregivers of persons with spinal cord injury (SCI) in Switzerland. We implemented the proxy-good method and estimated the market value of their work if performed by professional caregivers. Our results show that family caregivers in the sample spent an average of 27 hr per week caring for a relative for almost 12 years. This work, if undertaken by professional home care, has a market value of CHF 62,732 (EUR 56,455) per year. In the absence of family caregivers, these costs should be financed by the health insurances (47%), by the cared-for person (24%) and by the social insurances (29%). It is in the best interest of the cared-for person and of the healthcare and social systems to keep a sustained supply of family caregivers. One option is finding ways to recognise and compensate them for their work and make it less cumbersome.

Improving the Prediction of Persistent High Healthcare Utilizers: Using an Ensemble Methodology

Background

A small proportion of high-need patients persistently use the bulk of health care services and incur disproportionate costs. Population health management (PHM) programs often refer to these patients as persistent high utilizers (PHUs). Accurate PHU prediction enables PHM programs to better align scarce health care resources with high-need PHUs while generally improving outcomes. While prior research in PHU prediction has shown promise, traditional regression methods used in these studies have yielded limited accuracy.

Objective

We are seeking to improve PHU predictions with an ensemble approach in a retrospective observational study design using insurance claim records.

Methods

We defined a PHU as a patient with health care costs in the top 20% of all patients for 4 consecutive 6-month periods. We used 2013 claims data to predict PHU status in next 24 months. Our study population included 165,595 patients in the Johns Hopkins Health Care plan, with 8359 (5.1%) patients identified as PHUs in 2014 and 2015. We assessed the performance of several standalone machine learning methods and then an ensemble approach combining multiple models.

Results

The candidate ensemble with complement naïve Bayes and random forest layers produced increased sensitivity and positive predictive value (PPV; 49.0% and 50.3%, respectively) compared to logistic regression (46.8% and 46.1%, respectively).

Conclusions

Our results suggest that ensemble machine learning can improve prediction of care management needs. Improved PPV implies reduced incorrect referral of low-risk patients. With the improved sensitivity/PPV balance of this approach, resources may be directed more efficiently to patients needing them most.

Anticipating VA/non-VA care coordination demand for Veterans at high risk for hospitalization

U.S. Veterans Affairs (VA) patients' multi-system use can create challenges for VA clinicians who are responsible for coordinating Veterans' use of non-VA care, including VA-purchased care ("Community Care") and Medicare.To examine the relationship between drive distance and time-key eligibility criteria for Community Care-and VA reliance (proportion of care received in VA versus Medicare and Community Care) among Veterans at high risk for hospitalization. We used prepolicy data to anticipate the impact of the 2014 Choice Act and 2018 Maintaining Internal Systems and Strengthening Integrated Outside Networks Act (MISSION Act), which expanded access to Community Care.Cross-sectional analysis using fractional logistic regressions to examine the relationship between a Veteran's reliance on VA for outpatient primary, mental health, and other specialty care and their drive distance/time to a VA facility.Thirteen thousand seven hundred three Veterans over the age of 65 years enrolled in VA and fee-for-service Medicare in federal fiscal year 2014 who were in the top 10th percentile for hospitalization risk.Key explanatory variables were patients' drive distance to VA > 40 miles (Choice Act criteria) and drive time to VA ≥ 30 minutes for primary and mental health care and ≥60 minutes for specialty care (MISSION Act criteria).Veterans at high risk for hospitalization with drive distance eligibility had increased odds of an outpatient specialty care visit taking place in VA when compared to Veterans who did not meet Choice Act eligibility criteria (odds ratio = 1.10, 95% confidence interval 1.05-1.15). However, drive time eligibility (MISSION Act criteria) was associated with significantly lower odds of an outpatient specialty care visit taking place in VA (odds ratio = 0.69, 95% confidence interval 0.67, 0.71). Neither drive distance nor drive time were associated with reliance for outpatient primary care or mental health care.VA patients who are at high risk for hospitalization may continue to rely on VA for outpatient primary care and mental health care despite access to outside services, but may increase use of outpatient specialty care in the community in the MISSION era, increasing demand for multi-system care coordination.

Unmet health care needs and inequality: A cross-country comparison of the situation of people with spinal cord injury

Objective

To estimate and compare unmet health care needs of persons with spinal cord injury (SCI) across countries, the causes of these shortfalls, and the role of income.

Data Sources

We analyzed cross‐sectional data of 20 countries from the International Spinal Cord Injury (InSCI) survey, a compendium of comparable data on the living situation of persons with SCI. Data included information on high‐, middle‐, and low‐income countries. The survey comprises information on 12,095 participants.

Study Design

We used logit regressions to estimate the probability of unmet health care needs of persons with SCI and its causes. We adjusted the results by the individuals' characteristics and countries' fixed effects. We disaggregated the results by income decile of individuals in each country.

Data Collection/Extraction Methods

The inclusion criteria for the InSCI survey were adults aged 18 years and older with SCI living in the community, who were able to respond to the survey and who provided informed consent.

Principal Findings

Unmet health care needs are significant for people with long‐term conditions like SCI, where people in low‐income groups tend to be more affected. Among the barriers to meeting health care needs, the foremost is health care cost (in 11 of the 20 countries), followed by transportation and service availability. Persons with SCI in Morocco reported the highest probability of unmet health care needs in the sample, 0.54 (CI: 047–0.59), followed well behind by South Africa, 0.27 (CI: 0.20–0.33), and Brazil, 0.26 (CI: 0.20–0.33). In contrast, persons with SCI in Spain, 0.06 (CI: 0.04–0.08), reported the lowest probability of unmet health care needs, closely followed by Norway, 0.07 (CI: 0.05–0.09), Thailand, 0.08 (CI: 0.05–0.11), France, 0.08 (CI: 0.06–0.11), and Switzerland, 0.09 (CI: 0.07–0.10).

Conclusions

SCI is a long‐term, irreversible health condition characterized by physical impairment and a series of chronic illness. This makes SCI a high‐need, high‐cost group that faces significant unmet health care needs, which are mainly explained by the costs of health services, transportation, and services availability. This situation is prevalent in low‐, middle‐, and high‐income countries, where persons in lower income groups are disproportionately affected. To improve the situation, a combination of measures from the health and social systems are required.

Describing the characteristics and healthcare use of high-cost acute care users at the end of life: a pan-Canadian population-based study

BACKGROUND

A minority of individuals use a large portion of health system resources, incurring considerable costs, especially in acute-care hospitals where a significant proportion of deaths occur. We sought to describe and contrast the characteristics, acute-care use and cost in the last year of life among high users and non-high users who died in hospitals across Canada.

METHODS

We conducted a population-based retrospective-cohort study of Canadian adults aged ≥18 who died in hospitals across Canada between fiscal years 2011/12-2014/15. High users were defined as patients within the top 10% of highest cumulative acute-care costs in each fiscal year. Patients were categorized as: persistent high users (high-cost in death year and year prior), non-persistent high users (high-cost in death year only) and non-high users (never high-cost). Discharge abstracts were used to measure characteristics and acute-care use, including number of hospitalizations, admissions to intensive-care-unit (ICU), and alternate-level-of-care (ALC).

RESULTS

We identified 191,310 decedents, among which 6% were persistent high users, 41% were non-persistent high users, and 46% were non-high users. A larger proportion of high users were male, younger, and had multimorbidity than non-high users. In the last year of life, persistent high users had multiple hospitalizations more often than other groups. Twenty-eight percent of persistent high users had ≥2 ICU admissions, compared to 8% of non-persistent high users and only 1% of non-high users. Eleven percent of persistent high users had ≥2 ALC admissions, compared to only 2% of non-persistent high users and < 1% of non-high users. High users received an in-hospital intervention more often than non-high users (36% vs. 19%). Despite representing only 47% of the cohort, persistent and non-persistent high users accounted for 83% of acute-care costs.

CONCLUSIONS

High users - persistent and non-persistent - are medically complex and use a disproportionate amount of acute-care resources at the end of life. A greater understanding of the characteristics and circumstances that lead to persistently high use of inpatient services may help inform strategies to prevent hospitalizations and off-set current healthcare costs while improving patient outcomes.

Characterising and predicting persistent high-cost utilisers in healthcare: a retrospective cohort study in Singapore

OBJECTIVE

We aim to characterise persistent high utilisers (PHUs) of healthcare services, and correspondingly, transient high utilisers (THUs) and non-high utilisers (non-HUs) for comparison, to facilitate stratifying HUs for targeted intervention. Subsequently we apply machine learning algorithms to predict which HUs will persist as PHUs, to inform future trials testing the effectiveness of interventions in reducing healthcare utilisation in PHUs.

DESIGN AND SETTING

This is a retrospective cohort study using administrative data from an Academic Medical Centre (AMC) in Singapore.

PARTICIPANTS

Patients who had at least one inpatient admission to the AMC between 2005 and 2013 were included in this study. HUs incurred Singapore Dollar 8150 or more within a year. PHUs were defined as HUs for three consecutive years, while THUs were HUs for 1 or 2 years. Non-HUs did not incur high healthcare costs at any point during the study period.

OUTCOME MEASURES

PHU status at the end of the third year was the outcome of interest. Socio-demographic profiles, clinical complexity and utilisation metrics of each group were reported. Area under curve (AUC) was used to identify the best model to predict persistence.

RESULTS

PHUs were older and had higher comorbidity and mortality. Over the three observed years, PHUs' expenditure generally increased, while THUs and non-HUs' spending and inpatient utilisation decreased. The predictive model exhibited good performance during both internal (AUC: 83.2%, 95% CI: 82.2% to 84.2%) and external validation (AUC: 79.8%, 95% CI: 78.8% to 80.8%).

CONCLUSIONS

The HU population could be stratified into PHUs and THUs, with distinctly different utilisation trajectories. We developed a model that could predict at the end of 1 year, whether a patient in our population will continue to be a HU in the next 2 years. This knowledge would allow healthcare providers to target PHUs in our health system with interventions in a cost-effective manner.

Characterization of high healthcare utilizer groups using administrative data from an electronic medical record database

Background

High utilizers (HUs) are a small group of patients who impose a disproportionately high burden on the healthcare system due to their elevated resource use. Identification of persistent HUs is pertinent as interventions have not been effective due to regression to the mean in majority of patients. This study will use cost and utilization metrics to segment a hospital-based patient population into HU groups.

Methods

The index visit for each adult patient to an Academic Medical Centre in Singapore during 2006 to 2012 was identified. Cost, length of stay (LOS) and number of specialist outpatient clinic (SOC) visits within 1 year following the index visit were extracted and aggregated. Patients were HUs if they exceeded the 90th percentile of any metric, and Non-HU otherwise. Seven different HU groups and a Non-HU group were constructed. The groups were described in terms of cost and utilization patterns, socio-demographic information, multi-morbidity scores and medical history. Logistic regression compared the groups’ persistence as a HU in any group into the subsequent year, adjusting for socio-demographic information and diagnosis history.

Results

A total of 388,162 patients above the age of 21 were included in the study. Cost-LOS-SOC HUs had the highest multi-morbidity and persistence into the second year. Common conditions among Cost-LOS and Cost-LOS-SOC HUs were cardiovascular disease, acute cerebrovascular disease and pneumonia, while most LOS and LOS-SOC HUs were diagnosed with at least one mental health condition. Regression analyses revealed that HUs across all groups were more likely to persist compared to Non-HUs, with stronger relationships seen in groups with high SOC utilization. Similar trends remained after further adjustment.

Conclusion

HUs of healthcare services are a diverse group and can be further segmented into different subgroups based on cost and utilization patterns. Segmentation by these metrics revealed differences in socio-demographic characteristics, disease profile and persistence. Most HUs did not persist in their high utilization, and high SOC users should be prioritized for further longitudinal analyses. Segmentation will enable policy makers to better identify the diverse needs of patients, detect gaps in current care and focus their efforts in delivering care relevant and tailored to each segment.

Electronic supplementary material

The online version of this article (10.1186/s12913-019-4239-2) contains supplementary material, which is available to authorized users.