Reducing inequity in primary care clinics treating low socioeconomic Jewish and Arab populations in Israel

Abstract

Background

An organization-wide inequity-reduction quality improvement (QI) initiative was implemented in primary care clinics serving disadvantaged Arab and Jewish populations. Using the Chronic Care Model (CCM), this study investigated the types of interventions associated with success in inequity reduction.

Methods

Semi-structured interviews were conducted with 80 staff members from 26 target clinics, and information about intervention types was coded by CCM and clinical domains (e.g. diabetes, hypertension and lipid control; performance of mammography tests). Relationships between type and number of interventions implemented and inequity reduction were assessed.

Results

Target clinics implemented 454 different interventions, on average 17.5 interventions per clinic. Interventions focused on Decision support and Community linkages were positively correlated with improvement in the composite quality score (P < 0.05). Conversely, focusing on a specific clinical domain was not correlated with a higher quality score.

Conclusions

Focusing on training team members in selected QI topics and/or tailoring interventions to meet community needs was key to the interventions' success. Such findings, especially in light of the lack of association between QI and a focus on a specific clinical domain, support other calls for adopting a systems approach to achieving wide-scale inequity reduction.

Introduction

Inequity in health and health care in Israel is pervasive on many levels. Despite the availability of national health insurance since 1995,¹ differences in health indicators and the distribution and quality of health services are apparent by ethnic, religious, socioeconomic, education and employment status and by geographic location.² For example, infant mortality is more than twice as high among Muslims as it is among Jews³; income-related inequities are reported across most of the common chronic conditions, including hypertension, heart diseases, diabetes, depression and respiratory diseases⁴; and the chances of dying from heart diseases for women with lower levels of education have risen in the past two decades from two to five times, compared with more educated women.^5–9 Inequity in the distribution of healthcare services in Israel has also been widely documented; there are significant gaps in specialty care, rehabilitation, and imaging services by geographic location and socioeconomic status (SES).^4,10

Despite evidence of the growing inequity,⁷ programs to address this phenomenon have been, until quite recently, mainly sporadic. Israel's health authorities began to comprehensively address health inequity just less than a decade ago,¹¹ mainly through promoting equitable healthcare distribution by expanding services in the north and south peripheral regions, expanding dental care to children, and adding a geographical parameter to the allocation formula that distributes resources among the health funds.^12,13

In addition to these national inequity reduction efforts, in 2008 the largest health fund in Israel, Clalit Health Services (Clalit), which covers and provides services to ∼54% of the Israeli population, launched a comprehensive inequity reduction program embedded within its primary care clinics. Clalit's strategy employed a quality improvement (QI) scheme aimed at reducing gaps in a composite measure, the Quality Indicator Disparity Scale (QUIDS), of seven health and healthcare indicators, spanning prevention and control of the common chronic conditions diabetes, hypertension and lipid control; prevention of anemia in infants; mammography screening; fecal occult blood tests and influenza vaccinations.¹⁴

Clalit's strategy succeeded in reducing inequity between clinics serving low-SES Arab and Jewish minority groups and the general Clalit population.^14,15 However, inconsistencies in implementation and outcomes between the various participating clinics have been identified. Such variation is not uncommon in QI inequity reduction efforts¹⁶; some studies favor QI for minimizing health and healthcare gaps,^15,17 whereas some show increased inequities over time.^18,19 Yet, understanding the source of such inconsistencies is challenging, because current literature describes the intervention approach only generally, without specifying the types and scope of interventions and organizational mechanisms put in place to support QI implementation.^20–22

Because chronic conditions account for most inequities in health outcomes,²³ a suggested framework for understanding how QI initiatives work is the Chronic Care Model (CCM).^24,25 According to the CCM, the essential elements for providing high-quality chronic care are Delivery System Design; provision of Patient Self-Management; Decision, Information and Health System Support; and creation of Community Linkages.²⁵ The CCM components are associated with improved delivery and quality of care in general^26,27; however, how the main elements of chronic care management are associated with QI in inequity reduction efforts is unclear.²³

Thus, the aim of this study was to identify what types of interventions (classified according to the CCM categories as well as the seven clinical domains of the QUIDS) are implemented as part of Clalit's inequity reduction strategy, and to test whether this classification can contribute to the understanding of variations in inequity reduction outcomes.

Methods

Setting

The study was conducted in Clalit, an insurer and integrated delivery system provider for over four million enrollees (53% market share). Clalit's organizational structure consists of a central management, eight geographical regional headquarters, and three to four sub-regional headquarters per region. Each region is responsible for its member population, providing all primary and specialty care, as well as imaging, labs and pharmacy services. Primary clinics are organized in teams of physicians, nurses, pharmacists and administrative personal. Clalit's integrative Electronic Health Record based information system facilitates the monitoring of quality across a wide range of clinical domains, enabling the generation of reports and feedback to individual units (such as primary care clinics) and to all managerial levels.

Within the inequity reduction initiative, Clalit identified 55 clinics with significant gaps in the quality of care, serving ∼400 000 people (∼10% of Clalit's total enrolled population) of mainly socioeconomically disadvantaged and minority populations. The 55 clinics were those with the 10% lowest scores on the composite QUIDS score. For each clinic the set target was to close the gap between the clinic's QUIDS score and that of all other comparable clinics (medium to large, serving 5000 persons or more) in its region. While this target was set by Clalit's central management, the strategies and interventions applied by the target clinics were conceived at the regional, sub-regional and clinic levels.

Study population

The current study was conducted in 26 of the 55 target clinics and their associated managerial levels (sub-regional headquarters for each 2–4 clinics and region's headquarters). The 26 target clinics were randomly selected, representing rural and urban areas, and serve patients from a variety of ethnic and cultural population groups.

In each targeted clinic, interviews were held with the inter-professional managerial team, comprising medical director, nursing director, head of administration and pharmacist (in clinics having an in-house pharmacy). We also interviewed the medical and nursing directors of each clinic's sub-regional and regional headquarters. This selection was affected by the need to gather a variety of perspectives, as individuals working in the same clinic or health sector often have different views of policy and procedures.²⁸

Procedure

Interviews were conducted between April and September 2010 by two of the authors (S.S.-S. and E.S.) with 80 team members at the 26 target clinics, 19 sub-regional managers and 10 regional headquarters directors. The interview guide, adapted from the work of Chin et al.,²⁹ focused on eliciting the perceptions of leaders and staff about what makes inequity reduction QI initiatives sustainable. The guide was reviewed by three of the authors (S.S.-S., E.S. and M.G.) to establish validity. Interview questions were aimed at identifying interventions geared toward overall improvement in work processes, as well as in each of the seven clinical areas represented by the QUIDS. Questions focused on the types of interventions implemented and on the relations among the clinics' team members and between the clinics and the sub-regional and regional teams.

Interviews were transcribed and analyzed to elicit information on the types of interventions employed. A key informant was identified for each target clinic. The information provided by the key informant was augmented by information from interviews with other clinic team members. Upon completion of the interviews, a summary was sent to each target clinic for additional comments and approval of content.

Outcome

Outcome was defined as change in the QUIDS¹⁴ between 2009 and 2011. The QUIDS is a weighted score of seven quality indicators: (i) diabetes control (percentage of all registered diabetes patients with HbA1c <9%); (ii) blood pressure control (percentage of all registered hypertensive patients with BP ≤160/100 mm Hg); (iii) lipids control (percentage of all known hyperlipidemia patients with LDL <100 mg/dl); (iv) anemia prevalence in infants (percentage of infants aged 9–18 months with hemoglobin >105 g/l); (v) performance of mammography screening (once every 2 years, for women aged 50–75 years); (vi) performance of fecal occult blood tests (once every year for persons aged 50–75 years) and (vii) influenza vaccinations in targeted populations (>65 years of age or chronically ill). The seven indicators were then weighed to form a composite score (QUIDS) from 0 (poor quality) to 100 (excellent quality). Details of the weighing scheme are provided elsewhere.¹⁴

Coding instrument and measures

We mapped the types of interventions that target clinics employed using an adaptation of a tool developed by Grossman et al.,²⁴ which classified interventions according to the CCM domains: (i) delivery system design, (ii) self-management support, (iii) decision support, (iv) information systems, (v) community linkages and (vi) health system organization.²³ We also classified interventions according to the specific clinical domain which they pertained to, or as ‘general interventions’ if they did not pertain to a specific clinical area. For example, use of a mobile mammography screening unit was classified both as a ‘mammography screening clinical indicator specific intervention’ and as an intervention pertaining to ‘Delivery System Design’. Conversely, implementing weekly multidisciplinary team meetings to devise a QI work plan was classified as a ‘general intervention’ and as pertaining to ‘Delivery System Design’.

The coding instrument was pilot-tested by two of the authors (S.S.-S. and E.S.), who independently classified data from four clinics and discussed their findings to reach consensus. Agreement between the two raters approached 100%. We used these data to calculate the overall number of interventions employed, the number of interventions according to each CCM category, and the number of interventions according to the seven clinical domains plus the ‘general interventions’ category. Table 1 provides examples of common interventions according to each CCM category or clinical domain.

Table 1.

Examples of interventions according to the chronic care model (CCM) and clinical categories

CCM	Examples of common interventions
Delivery system design	A change in the care practice roles of target clinic personnel such as the involvement of administrative staff in proactive patient follow-up Implementing weakly multidisciplinary team meetings to devise a weekly targeted work plan on quality improvement measures
Self-management support	Patient education on clinical issues such as diabetes, mammography screening or hypertension through individual meetings, workshops and lectures in the clinic
Decision support	Quality of care training workshops for clinic team members on selected topics (mostly focused on improving diabetes care) Training workshops on utilization of quality indicator reporting and tracking
Information systems	Utilizing patient electronic health records for management of care to support proactive follow-up. Generation and review of performance data on attainment of quality indicators
Community linkages	Culturally tailoring interventions to meet the community needs, such as working with religious leaders to promote screening, vaccination or healthy lifestyle, and establishing partnerships to promote community health days
Health system organization	The support provided by the organization's leadership for implementing interventions
Clinical quality indicators
Diabetes	Creating a doctor-nurse-pharmacist team meeting and consult for patients targeted for intervention Community adapted patient education interventions such as culturally adapted diabetic cooking class, lectures in religious places of worship
Hypertension	Patient education on self-monitoring by using an intervention based on home measurements
Lipid	Doctor-Nurse-Pharmacist team meeting to review and adjust medication
Mammography	Securing a mobile screening unit for conducting screening tests on site at the clinic Scheduling appointments in hospitals and securing organized group transportation for clinic patients
Fecal occult blood test	Dispensing test kits to relevant patients by all clinic personnel as well as monitoring and following-up on return of the kits to the clinic
Anemia	Proactive follow-up to adherence of iron supplement of target population
Flu vaccinations	Creating partnerships with community religious leaders to promote flu vaccinations, such as vaccinating jointly for people of Muslim faith traveling to Mecca the needed vaccinations along with flu vaccination

Analysis

We examined baseline characteristics of target and comparable clinics in terms of size, ethnicity and SES. SES levels were defined by geocoding technique linking the primary clinic's address with Israel's Central Bureau of Statistics socioeconomic indicator score rating all towns in Israel, with 1 representing towns of low SES and 10 high SES.¹⁴ We calculated the QUIDS score for each target clinic and all other comparable clinics (all medium to large, serving 5000 persons or more) in the same region at baseline (2009) and at 2-year follow-up (2011).

In an exploratory analysis, we used the Pearson coefficient to examine the association between the number of interventions employed by target clinics according to each CCM category and each clinical domain, and change in QUIDS score.

The study protocol was approved by the Committee on Human Studies, Clalit Health Services (authorization no. 172/2011).

Results

Characteristics of target clinics and comparison clinics

The total number of enrollees in the target clinics was 217 306, and in the comparable clinics was 1 310 840. In target clinics, most enrollees were of Arab ethnicity (78.1%); as opposed to a Jewish majority (57.1%) in comparable clinics. Most enrollees in the target clinics were of low SES (78.91%). In comparison, only 48.79% of the enrollees in the comparable clinics were of low SES, and enrollees were dispersed between all three clinic SES categories. Examination of the difference in the individual target clinics' QUIDS scores pre- and post-implementation of intervention (2009–11) showed an average 7.2% improvement in QUIDS scores but no change in the comparable clinics (Table 2). The gap between target and comparable clinics' QUIDS scores narrowed from 0.062 to 0.021 difference (reduction of 66.1%). Yet, this achievement was not uniform and reduction ranged between 23.55 and 100%.

Table 2.

Characteristics of target clinics and comparison clinics

	Comparison clinics N = 1 310 840 N (%)	Target clinics N = 217 306 N (%)
Region (%)
A	304 721 (90.93)	30 403 (9.07)
B	457 413 (84.01)	87 061 (15.99)
C	161 720 (83.71)	31 463 (16.29)
D	386 986 (84.98)	68 379 (15.02)
Ethnicity (%)
General Jewish population	57.07	21.14
Ultra-Orthodox Jews	0.24	0.76
Arab	42.70	78.10
Socioeconomic status (%)
Low	48.79	78.91
Medium	36.60	21.09
High	14.62	–
QUIDs Score 2009, mean (SD)	0.638 (0.017)	0.576 (0.042)
QUIDs Score 2011, mean (SD)	0.639 (0.004)	0.618 (0.035)
Percent change in QUIDs Score (2009–11)	0.1%	7.2%

Interventions implemented

Target clinics implemented a total of 454 different interventions, an average of 17.5 per clinic (Table 3). On average, clinics implemented four different interventions per CCM category. Delivery system design was the CCM strategy most used, averaging 8.7 interventions per clinic. Next was the use of the organization's information systems. Interventions focusing on Health System Organization, stemming from central management involvement, such as financial support, were implemented in just two target clinics.

Table 3.

Quality improvement interventions implemented by target clinics according to the chronic care model (CCM) and clinical quality indicators

	Total number of Interventions	Average per clinic (SD)
CCM
Delivery system design	225	8.7 (3.0)
Self-management support	40	1.5 (1.6)
Decision support	33	1.3 (1.0)
Information systems	119	4.6 (1.8)
Community linkages	34	1.3 (1.4)
Health system organization	3	0.1 (0.3)
Average across all clinics		4.4 (3.1)
Clinical quality indicators
Diabetes	90	3.5 (1.7)
Hypertension	17	0.7 (1.2)
Lipid	14	0.5 (0.9)
Mammography	81	3.1 (1.1)
Fecal occult blood test	48	1.9 (1.0)
Anemia	7	0.3 (0.8)
Flu vaccinations	21	0.8 (0.9)
General interventions	176	13.0 (2.2)
Average across all clinics		3.0 (1.2)

Among the interventions implemented by target clinics according to clinical domains, general non-clinical interventions, such as improvements to teamwork, were the most common. Of the seven clinical domains, the most common interventions were geared toward improving diabetes control (average 3.5), followed by mammography screening (average 3.1) and performance of fecal occult blood tests (average 1.9). Interventions to improve the reduction in prevalence of anemia in infants were rare (average 0.27).

Association between number of interventions and difference in QUIDS scores

Analyses of the number of interventions implemented by target clinics in the six different CCM categories showed a statistically significant positive relationship between the number implemented for decision support and QUIDS improvement (P< 0.05). Another CCM domain that positively correlated with QUIDS improvement was community linkages (P < 0.05). No other statistically significant correlations of this kind were found. Nor did we find any significant relationship between the number of interventions implemented in each of the seven clinical domains and QUIDS improvement (Table 4).

Table 4.

Relationship between target clinics quality improvement interventions implemented and difference in QUIDS Score

	Difference in QUIDs Score 2009–11*Pearson r
CCM
Number of interventions
Delivery system design	−0.125
Self-management support	−0.381
Decision support	0.451*
Information systems	−0.261
Community linkages	0.437*
Health system organization	0.008
Quality indicators
Diabetes	−0.120
Hypertension	−0.344
Lipid	−0.115
Mammography	0.227
Fecal occult blood test	−0.283
Anemia	−0.191
Flu Vaccination	0.107

*P < 0.05; **P < 0.01.

Discussion

Main findings of the study

This study provides detailed data on the number and types of interventions implemented as part of an organization-wide QI initiative directed at reducing inequities in health and health care, by targeting minority and low-SES clinics with low achievements on the selected inequity-associated quality measures. We measured the effectiveness of the intervention according to the CCM as well as the seven clinical quality indicators which were the focus of the program.

Our study shows that almost 40% of all interventions implemented were not geared toward a specific quality indicator but instead focused on general improvements that could benefit quality of care across several domains. Classification according to the CCM categories showed that target clinics were particularly active in the areas of delivery system design focusing on the reorganization of the clinic's teamwork and implementing proactive follow-ups, along with interventions centered on information systems, improving the use of patients' electronic health records for management of care as well as the generation of auditing tools using aggregated performance data.

What is already known on this topic

A focus on delivery system design and information systems was also reported by Grossman et al.,²³ who evaluated the health-disparities collaborative QI program and its effect on inequity reduction. Delivery system design and/or the fostering of information systems is considered essential in launching actions to reduce inequity.³⁰ Thorlby et al.³⁰ interviewed staff at healthcare organizations and found that information systems were viewed as essential components of QI efforts, as they allowed not only identification of existing inequities but also tailoring of interventions accordingly.

Our finding that a focus on decision support and/or the promoting of community linkages is associated with inequity reduction is supported by a recent review by Clarke et al.³¹ It indicated that because inequities are often the result of challenges embedded in the management of complex systems, it is essential to shift responsibility for change from the patient to the organization. Such a change from a mostly patient-education centered approach to a focus on an organization-wide and/or community-wide initiative has the potential to reach a broader target population and achieve sustainability over time. Similarly, Chin et al.³² report that training healthcare teams in health inequity issues may help improve care and reduce quality gaps. Clarke et al.'s review³¹ also states that although community-based interventions are relatively sparse, they have shown promise in reducing inequity. Similarly, our study found that target clinics implemented unique community-tailored interventions, such as a culturally tailored cooking classes for Arab diabetic patients; partnerships with religious community leaders to promote screening, vaccinations or a healthy lifestyle; or collaborations to promote community health days. Creating community linkages by stepping outside the clinic walls and into the community may have helped target clinics' teams to better understand local patients' needs and thus achieve greater effectiveness in overall inequity reduction.^33,34

What this study adds

Our results show that target clinics that participated in an organization-wide inequity reduction program, implemented a wide range of interventions. While the most common types of intervention focused on improving the care practice roles of clinic employees and reorganizing teamwork (delivery system design) or on using electronic health records to better manage patient care (information systems), these intervention types were not correlated with the interventions' outcomes, and might present a ‘necessary but not sufficient’ condition for success. We identified that focusing on training clinic team members in selected QI topics (decision support) and/or tailoring interventions to meet community needs (community linkages) was key to the interventions' success.

We found no significant relationship between the number of interventions implemented in the seven clinical indicators and QI. This may in part be because success depends on quality, not quantity. Such null findings provide support for the need to tailor interventions to the particular needs of patients and populations, especially those of minorities or other unique groups. The QI framework often fails to contend with a population's cultural barriers and needs.³⁴ Cooper et al.³⁵ state that interventions will not be sustainable over time if they are not adapted and made relevant to both the organization and the patient. Such findings, especially in light of the lack of association between QI and a focus on a specific clinical domain, support other calls for adopting a systems approach to achieving wide-scale inequity reduction.

Limitations of the study

Several limitations to this study should be considered. First, the sample of clinics was relatively small. However, our focus was on eliciting information about the types of interventions employed by target clinics through a rigorous comprehensive and validated assessment, including interviews with 109 employees directly involved in the program. Still, future studies should examine the relation between the types of interventions employed and success in inequity reduction across a broad range of populations and clinics. Additionally, information obtained from interviews may be subject to recall bias. Nonetheless, this was the most comprehensive source available, as no official documentation of intervention types was available. To overcome this shortfall, we identified the key informant for each target clinic and complemented the information retrieved through interviews with other clinic staff as well as managers at relevant sub-regional and regional units. Finally, other characteristics may have contributed to the outcome. Studies have shown that the organizational context,^36,37 such as perceived team effectiveness,³⁸ composition of the clinic's team³⁹ and organizational culture,⁴⁰ affects the inequity reduction process. Future studies should examine the effect of organizational context on the implementation process and outcomes of inequity reduction initiatives.

Funding

The study was supported in part by a PhD scholarship (awarded to Spitzer-Shohat) from the Israel National Institute for Health Policy Research (NIHPR). The NIHPR had no role in the design and conduct of this study, the analysis or interpretation of the data, or the preparation of the manuscript.