A programme to improve quality of care for patients with chronic diseases, Kazakhstan

Abstract Objective To evaluate the effect of a disease management programme in Kazakhstan on quality indicators for patients with hypertension, diabetes and chronic heart failure. Methods A supportive, interdisciplinary, quality improvement programme was implemented between November 2014 and November 2015 at seven polyclinics in Pavlodar and Petropavlovsk. Quality improvement teams were established at each clinic and quality improvement tools were introduced, including patient flowsheets, decision support tools, patient registries, a patient recall process, support for patient self-management and patient follow-up with intensity adjusted for level of disease control. Clinic teams met for four 3-day interactive learning sessions within 1 year, with additional coaching visits. Implementation was managed by five local coordinators and consultants trained by international consultants. National and regional steering committees monitored progress. Findings Between July and October 2015, the proportion of hypertensive patients with the recommended blood pressure increased from 24% (101/424) to 56% (228/409). Among patients with diabetes, the proportion who recently underwent eye examinations increased from 26% (101/391) to 71% (308/433); the proportion who had their low-density lipoprotein cholesterol measured increased from 57% (221/391) to 85% (369/433); and the proportion who had their albumin : creatinine ratio measured increased from 11% (44/391) to 49% (212/433). The proportion of chronic heart failure patients who underwent echocardiography rose from 91% (128/140) to 99% (157/158). All patients set themselves self-management goals. Conclusion This intensive, supportive, multifaceted programme was associated with significant improvements in quality of care for patients with chronic disease. Further investment in coaching capacity is needed to extend the programme nationally.


Research
Improving care quality, Kazakhstan Benjamin TB Chan et al.
used successfully in high-income countries for chronic disease management in primary care. For example, the Health Disparities Collaboratives in the United States of America improved the quality of diabetic care among vulnerable populations. 10 This paper provides new information on how quality improvement techniques can be applied in a middle-income country with a distinct culture, governance system and primary care infrastructure.

Methods
In Kazakhstan, primary care is provided through polyclinics by specialists, therapists (i.e. internists), general practitioners, nurses, psychologists, social workers and health educators. Laboratory and diagnostic imaging services are also available on site. Polyclinics are publicly funded and provide essential services for free within their catchment areas. Urban polyclinics report to the health department of the oblast (i.e. subnational region), which in turn reports to the national health ministry. We investigated the effect of the initial design and testing phase of the disease management programme, from November 2014 to November 2015. The programme was implemented in seven large urban polyclinics in Pavlodar and Petropavlovsk (population 308 000 and 195 000, respectively). Three clinic teams worked on diabetes, three worked on hypertension and two worked on chronic heart failure. In one clinic, two disease types were tackled simultaneously. This phase did not include private clinics or public clinics in rural areas, which offer a limited range of services.

Programme design
To assess quality, countries of the former Soviet Union traditionally relied on clinical protocols, which specified standards for medical practice against which physicians were audited and sanctioned if found noncompliant. 11 This approach assumed that poor care quality was due to a lack of effort that could be remedied by punishment and ignored the fact that poor quality was often due to systemic obstacles. In contrast, the disease management programme adopted a supportive, teambased, multifaceted approach to quality improvement that aimed to help clinic teams address the root causes of poor care in an environment that emphasized learning, analysis and improving work processes. The programme used the Chronic Care Model as a blueprint for designing a primary health-care system to manage chronic diseases and included the following components: (i) decision support tools for clinicians; (ii) an information system; (iii) care delivery system design; and (iv) patient selfmanagement. 12 Decision support tools are intended to remind clinicians of the actions to be taken in different situations. They address the problem that guidelines are often complex and easy to forget and that some health-care providers may not be aware of their contents. 13 The main tool was a flowsheet -a one-page document included in each patient's chart to remind staff which tasks should be performed and documented at each clinic visit. The document also recorded clinical data, such as blood pressure, laboratory measurements and healthrelated behaviours. A flowsheet was developed for each targeted condition based on international examples. Other tools included simple, one-page algorithms for diagnosis or selecting treatment and checklists for the tests required. These tools were user-friendly alternatives to clinical protocols, which can be lengthy, legalistic and dense. All tools were approved by a clinical advisory group. The clinical information system comprised a patient registry, which addressed the problem that healthcare providers may be unaware of gaps in care that need attention. At each patient encounter, clinic staff entered data required by the flowsheet into an Excel database (Microsoft Corporation, Redmond, USA), which automatically calculated values for quality indicators. Staff could then review areas of weakness monthly and target them for improvement. The registry also reported changes in indicators over time, which helped in monitoring the programme's impact.
A care deliver y system was designed to ensure key processes were performed consistently. The system addressed the problem that the steps involved in delivering care are often poorly coordinated or implemented, or inefficient. There were three process improvements: (i) a recall process was created to ensure patients overdue for follow-up or a test returned to the clinic; (ii) patient segmentation was introduced to group patients by level of disease control; and (iii) structured visits were introduced. Box 1 describes these approaches in more detail.
The program introduced support for patient self-management, an approach which helps patients manage their condition themselves. Research shows that patients engaged in their own care who understand their condition Box 1. Improvements to optimize quality of care of chronic diseases, Kazakhstan, 2015

A recall process
This process helped to ensure that patients overdue for follow-up or a test returned to the clinic. Practice guidelines recommend patients with diabetes undergo measurement of HbA1c every 6 months and LDL cholesterol measurement every 12 months. The patient registry was designed to generate recall lists of patients overdue for follow-up or a test. Each polyclinic was required to refine its recall process. Typically, polyclinics assigned one individual to review recall lists weekly and ensure patients were phoned or otherwise encouraged to return to the clinic.

Patient segmentation
This process aimed to group patients by level of disease control. For example, diabetes patients with a blood pressure and HbA1c and LDL cholesterol levels within desired limits were deemed optimal. Those with an HbA1c level above 7% were suboptimal and an HbA1c level over 9% indicated poor control. Each clinic developed standard processes for determining how frequently and intensely each patient group should be followed up. For example, a patient with well controlled hypertension could be seen every 6 months, whereas one with a systolic and diastolic blood pressure above 160 and 100 mmHg, respectively, could be seen monthly until control was achieved. Previously in Kazakhstan, all patients were seen monthly. The aim of segmentation was to improve efficiency by reducing unnecessary visits for healthier patients and reallocate staff time to those who needed more attention.

Structured visits
Clinic teams were encouraged to identify all tasks included in follow-up assessments, to assign tasks to different team members, to consider shifting tasks between team members (e.g. from a specialist to a primary care physician) to improve efficiency and to develop a routine to avoid omitting tasks by mistake.
Improving care quality, Kazakhstan Benjamin TB Chan et al.
and know how to modify unhealthy behaviour benefit most from improved clinical care. 14 Clinic staff learned how to shift from simply providing information to patients or using scare tactics to induce change to, instead, engaging in supportive dialogue. Staff also learned to coach patients to set small, but realistic and specific goals and to help them make several small changes that could gradually strengthen their selfconfidence.
The programme was consistent with the three pillars of WHO's framework on quality in primary health care: (i) empowered people and engaged communities; (ii) multisectoral policy and action for health; and (iii) health services that prioritize the delivery of high-quality primary care. 15 The first pillar was addressed by the programme's patient self-management component. The second was addressed by a concurrent project funded by the World Bank, which aimed to expand health insurance coverage, introduce accreditation and provide financial incentives for good performance. The third was addressed by the programme's decision support tools, performance feedback and process improvements.

Implementation
We emphasized group learning over multiple encounters instead of traditional lecture-style teaching by using the Breakthrough Series Collaborative model developed for multisite quality improvement initiatives. 16 Clinic teams attended four 3-day learning sessions in Pavlodar or Petropavlovsk to receive training from international consultants on implementing quality improvements. Each city had a regional coordinator (a physician with management experience) who worked with the polyclinics and was also trained by the international consultants. Skills, such as support for patient self-management, were taught by studying clinical cases and role-playing. Before each session, teams were assigned preparatory work and sessions were used to report progress, identify obstacles and brainstorm solutions with other participating teams. Between learning sessions, the international consultants made coaching visits and participants conducted Plan-Do-Study-Act cycles to test and customize quality improvement tools from elsewhere and adapt them for local use (Fig. 1).
A formal leadership structure was established at different levels. Each polyclinic identified a clinical coordinator (i.e. team leader) and formed an interdisciplinary quality improvement team. The health ministry appointed a national coordinator and the two regional coordinators noted above. Progress across all sites was reviewed by a national steering committee and, at the regional level, by regional steering committees.
The core implementation team comprised five international consultants (two full-time equivalents) and two full-time local consultants and was active over 13 months. The programme's costs included staff remuneration, the cost of office space, room rental, printed material and translations, and travel costs for meetings within the country and for six missions by international consultants. There were substantial inkind contributions of personnel time from health ministry staff and other key stakeholders, which included time for participating in steering committees and clinical advisory groups. One full-time staff member from the health ministry was designated the programme liaison officer.

Evaluation
Our investigation employed a quasiexperimental study design, where differences in quality indicators from before to after the intervention were examined for a single study group. Clinic teams submitted data monthly from July to October 2015. During this time, teams implemented programme components, such as recall processes, patient segmentation and support for patient self-management. Differences between the two periods were assessed using a two-tailed t-test for the difference between proportions. Quality indicators were selected for diabetes, hypertension and chronic heart failure  Notes: The diagram was adapted from the Institute for Healthcare Improvement. 16 Interdisciplinary quality improvement teams from each pilot site attended quarterly learning sessions and were supported in between sessions by email, visits from consultants and phone conferences. Teams also started reported data each month midway into the collaborative.
Research Improving care quality, Kazakhstan Benjamin TB Chan et al.
by reviewing indicators used in other countries or recommended by clinical guidelines (Table 1). All indicators were approved by a national clinical advisory group. Indicators included both process measures (e.g. adoption of guideline recommendations on the use of drugs and tests, and on follow-up) and outcome measures (e.g. blood pressure, blood sugar and cholesterol levels).

Findings
Learning sessions began in January 2015, indicators and flowsheets were established by March 2015 and the patient registry became operational by June 2015. All learning sessions between January and October 2015 included training on patient self-management. Between July and October 2015, the proportion of hypertensive patients whose blood pressure was under control Improving care quality, Kazakhstan Benjamin TB Chan et al.
increased significantly (Table 1 and Fig. 2), as did the proportion whose blood pressure was checked at the last clinic visit (Fig. 3). There were also significant increases in the proportion of patients with diabetes who underwent low-density lipoprotein (LDL) cholesterol and albumin: creatinine ratio assessment and had eye examinations in the past year. However, there was no significant change in foot examinations or regular glycosylated haemoglobin (HbA1c) measurement. The proportion of patients with good control of LDL cholesterol (i.e. under 2.5 mmol/L) decreased significantly from 27% (59/221) to 17% (64/369) but there was no significant change in the proportion with good glucose control (i.e. an HbA1c level under 7%) or with a systolic and diastolic blood pressure under 140 mmHg and 90 mmHg, respectively. The proportion of patients with chronic heart failure who underwent echocardiography increased significantly from 91% (128/140) to 99% (157/158). All patients had self-management goals documented and 223 health-care providers underwent basic training on patient self-management. All seven polyclinics achieved a significant improvement in at least one quality indicator.

Discussion
Our investigation showed that the quality improvement tools for chronic disease management developed in high-income settings could be deployed effectively in Kazakhstan. Improvements were achievable despite fewer national resources and the country's history of limited primary care development. In 2009, only 17% (26 vs 156 per 100 000 population) of physicians in the country were general practitioners and their training programmes were relatively new and focused on knowledge rather than practical skills. 17 Implementation of the disease management programme was associated with substantial improvements in care quality process measures, such as ensuring patients had recently undergone recommended tests. The recall lists generated by the patient registry were critical for success because they identified patients who needed to return for missed tests. Our observations are consistent with those of the United States' Health Disparities Collaborative, which found that improvements were greatest for similar quality indicators. 18 Although quality outcome measures improved for hypertensive patients, similar outcomes did not improve for patients with diabetes over the shortterm. However, clinical algorithms were introduced relatively late in the programme and they might have had little impact during the observation period. Moreover, it may require more time to optimize decision-making for more complex treatment decisions. In the Health Disparities Collaborative, early results also showed no improvement in diabetes outcomes, 18 but some sites demonstrated reductions in HbA1c levels after 4 years of follow-up. 10 The unusual finding that the proportion of patients with diabetes and an LDL cholesterol level < 2.5 mmol/L decreased probably occurred because increased testing led to greater inclusion of people who were not compliant with the regular visit schedule and who were also probably less likely to comply with dietary recommendations. Members of the national steering committee noted that statins were not free under universal health insurance in Kazakhstan -drug policy may, therefore, need to change. Similarly, the proportion of patients with diabetes whose HbA1c level was measured did not change. Although HbA1c testing is free, some participants noted that budgetary constraints at clinics hindered access to the test. Better planning could improve access.
As the disease management programme had numerous complex components, frequent interactions between international consultants, local coordinators and participants were key to success. These interactions helped clinic teams adapt the quality improvement tools developed elsewhere for local use and assisted in problem-solving. During learning sessions and coaching visits, implementation problems were observed, such as the incomplete use of flowsheets, data entry errors, incorrect techniques in patient self-management discussions and confusion about interpreting guidelines, algorithms or indicators. The traditional learning model of attending a single lecture would probably not have resulted in similar improvements.
The programme's formal leadership structure provided an accountability mechanism that probably contributed to its success. Progress was reviewed

Fig. 3. Change in care quality process indicators, disease management programme, Kazakhstan, 2015
Patient with chronic heart failure who underwent echocardiography Patient with hypertension whose blood pressure was checked at the last polyclinic visit Patient with diabetes whose HbA1c was measured in the past 6 months Patient with diabetes who had a foot examination in the past year Patient with diabetes whose Research Improving care quality, Kazakhstan Benjamin TB Chan et al.
regularly at national and regional steering committees, where problems were identified and solutions discussed. The implementation rate of different quality improvement tools and in the improvements achieved varied between clinics. Later in the programme, clinic teams were asked to present progress reports to their peers during learning sessions and to compare results with each other. This friendly competition helped motivate teams to improve. Recent WHO recommendations for governments on improving health care emphasize the need for clear strategies on care quality to ensure success and sustainability. 19 Specifically, WHO guidelines recommend: (i) setting priorities and targets; (ii) engaging stakeholders; (iii) specifying accountability; (iv) identifying indicators; and (v) creating information systems for performance feedback and reporting. 20 As part of this project which was financed by a World Bank loan, the consulting team in Kazakhstan made recommendations on a national chronic disease strategy that were consistent with WHO's framework. Stakeholders were engaged in programme design through national and regional steering committees and clinical advisory groups. These committees served as an accountability structure. In addition, it was recommended that accountability be strengthened by extending accreditation criteria to include programme components, such as the use of a patient registry and flowsheets. The quality indicators identified and listed in Table 1 were approved nationally. Regarding information systems, it was recommended that the patient registry be incorporated into future electronic medical records. Finally, financial incentives were introduced to improve primary care performance and recommendations were made on how incentives could be better aligned with the programme's objectives.
Our quasi-experimental study design was limited by the lack of a control group. However, it is unlikely the large improvements we observed over a short time were due to any factor other than the disease management programme. Moreover, there was no major change in infrastructure, staffing, catchment population or remuneration at pilot sites during the study period. Another limitation was that, although all patients set themselves self-management goals, the quality of the self-management support provided for patients was not assessed. Future studies should include a patient survey to evaluate this support.
The generalizability of the study's findings may be limited for two reasons. First, only urban settings were included; implementation of the programme in rural settings with fewer resources may require more support. Second, although Kazakhstan has relatively few primary care physicians, the polyclinic model has strengths that may have contributed to success, such as different health disciplines working together in the same facility. In addition, data literacy was good and most clinics already had data entry staff. Implementation may be harder in settings without equivalent staffing.
Following the success of this pilot, attempts were made to extend the programme throughout Kazakhstan by establishing trainers in each oblast to support local polyclinics within the existing health-care system hierarchy. Designing a system to support clinic teams throughout the country proved challenging because the quality improvement model we employed requires high-intensity support and because the number of local coordinators and consultants trained was insufficient for rapid expansion. Currently, a new project financed by a World Bank loan is underway that will increase the number of local facilitators. Our experience confirms that investment in capacity building at the ground level is essential for ensuring sustainability.
The disease management programme in Kazakhstan included a holistic package of interventions, such as patient flowsheets, decision support tools for clinicians, process improvements, support for patient self-management, measurement of quality indicators and performance feedback through an electronic registry. Our pilot study found that significant improvements in care quality could be achieved without an increase in clinic staff. However, success depended critically on intensive coaching and regular support for local clinic teams. The priority for policy-makers who wish to apply this approach in their own countries is to invest in building capacity to provide external support for local clinic teams. Also important are strong leadership, an accountability structure, incentives and continued engagement with stakeholders within the framework of a national plan for improving health-care quality. ■ Improving care quality, Kazakhstan Benjamin TB Chan et al.

Resumen
Un programa para mejorar la calidad de la atención a los pacientes con enfermedades crónicas, Kazajstán Objetivo Evaluar el efecto de un programa de gestión de enfermedades en Kazajstán sobre los indicadores de calidad de los pacientes con hipertensión, diabetes e insuficiencia cardíaca crónica. Métodos Entre noviembre de 2014 y noviembre de 2015 se llevó a cabo un programa de apoyo, interdisciplinario y de mejora de la calidad en siete policlínicos de Pavlodar y Petropavlovsk. Se establecieron equipos de mejora de la calidad en cada clínica y se incorporaron instrumentos de mejora de la calidad, como hojas de evolución de pacientes, instrumentos de apoyo a la toma de decisiones, registros de pacientes, un proceso de llamadas para recordar citas a los pacientes, apoyo a la autogestión de los pacientes y seguimiento de los pacientes con una intensidad ajustada al nivel de control de la enfermedad. Los equipos clínicos se reunieron en cuatro sesiones de aprendizaje interactivo de tres días en el plazo de un año, con visitas adicionales de entrenamiento. Cinco coordinadores y consultores locales, formados por consultores internacionales, gestionaron la implementación. Los comités directivos nacionales y regionales supervisaron los progresos realizados.